1 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES PUBLIC HEALTH SERVICE FOOD AND DRUG ADMINISTRATION CENTER FOR BIOLOGICS EVALUATION AND RESEARCH INTERNATIONAL ASSOCIATION FOR BIOLOGICALS NATIONAL INSTITUTE OF ALLERGY AND INFECTIOUS DISEASES NATIONAL VACCINE PROGRAM OFFICE WORLD HEALTH ORGANIZATION - - - EVOLVING SCIENTIFIC AND REGULATORY PERSPECTIVES ON CELL SUBSTRATES FOR VACCINE DEVELOPMENT - - - WORKSHOP - - - WEDNESDAY, SEPTEMBER 8, 1999 - - - The workshop was held in the Plaza Ballroom, Doubletree Hotel, 1750 Rockville Pike, Rockville, Maryland 20852, at 8:30 a.m., Harry Rubin, DVM, and Martin Myers, M.D., Co-Chairs, presiding. PRESENT: HARRY RUBIN, DVM Co-Chair MARTIN MYERS, M.D., PhD Co-Chair NAOMI ROSENBERG, PhD Session Chair HENRY PITOT, M.D., PhD Speaker ALEX VAN DER EB, PhD Speaker JAMES McDOUGALL, PhD Speaker 2 PRESENT: (continued) STEPHEN BAYLIN, M.D. Speaker WALTER DOERFLER, M.D. Speaker JAMES COOK, M.D. Speaker SATVIR TEVETHIA, PhD Speaker FRANK SISTARE, PhD Speaker MICHAEL FRIED, PhD Speaker SANDRA RUSCETTI, PhD Session Chair CLIVE PATIENCE, PhD Session Chair LEONARD EVANS, PhD Session Chair DAMIAN PURCELL, PhD Session Chair PAUL JOLICOEUR, M.D., PhD Session Chair ALSO PRESENT: DAVID ONIONS, PhD 3 I N D E X Page SESSION 2: Mechanisms of Neoplastic Development and Neoplastic Cells Tumorigenicity: Implications for Cell Substrate Development Introduction: Session Co-Chairs 5 Animals of neoplastic development 5 Henry Pitot, M.D., PhD Multistep carcinogenesis, Harry Rubin, DVM 21 Transformation by DNA viral oncogenes 45 Alex van der Eb, PhD Hit and run transformation leading to 61 carcinogenesis, James McDougall, PhD DNA methylation and epigenetic mechanisms of 75 carcinogenesis, Stephen Baylin, M.D. A new concept in viral oncogenesis 89 Walter Doerfler, M.D. Role of nonspecific NK/macrophage cell host 108 responses in assessing tumorigenicity, James Cook, M.D. Role of CTL host responses and their 125 implications, Satvir Tevethia, PhD Transgenic animals that might be useful in 137 identifying unsuspected oncogenic factors, Frank Sistare, PhD SESSION 2: Panel-Audience Discussion 155 SESSION 3 (Part 1) Viral-Viral and Viral-Cellular Interactions Introduction 229 Generation of MCF retrovirus as a model of 233 viral-viral and viral-cellular interactions Sandra Ruscetti, PhD 4 INDEX (continued) Page MuLV packaging systems as models for estimating/ 248 measuring retrovirus recombination frequency, Clive Patience, PhD In vivo and in vitro effects of mixed retrovirus 264 infections, Leonard Evans, PhD Cross-species pathogenesis of replication- 277 competent retrovirus in immunosuppressed hosts, Damian Purcell, PhD Pathogenesis of defective retroviruses, 296 Paul Jolicoeur, M.D., PhD 5 1 P R O C E E D I N G S 2 Time: 8:00 a.m. 3 CO-CHAIRMAN MYERS: On behalf of the 4 National Vaccine Program, welcome. Our first speaker 5 this morning will be Henry Pitot of the University of 6 Wisconsin, and he's going to start talking about 7 animals of neoplastic development. 8 DR. PITOT: Hopefully, this thing works. 9 Does it? Both Harry and I chose to speak down here. 10 I guess the other speakers can decide whether or not 11 they can see the slides from up here or down there. 12 I think my function this morning was 13 basically to try to cover in about 15 or 18 minutes 14 most of the more commonly used animal models for 15 cancer development, and probably without further ado, 16 I'll just start with the first slide, which is, I 17 think, familiar to most of you. 18 It's probably the most widely used animal 19 model for neoplastic development. It's, obviously, 20 used by the regulatory agencies, and some people have 21 considered it what might be considered the gold 22 standard of carcinogenesis. 23 Unfortunately, it has a lot of drawbacks, 24 as those of us that work in experimental systems know. 25 But looking at the literature, in fact, probably most 6 1 of the data on whether or not a chemical is 2 carcinogenic really comes from this sort of data that 3 you see here, which really is based on studies that 4 went back many years, actually to the 1930s where 5 animals were given a carcinogen for an extended period 6 of time until tumors developed. 7 In this particular system there's 8 actually, as you can see, a fairly standardized type 9 of thing which allows one over a two-year period to 10 determine the development of neoplasm. 11 Now nice as it might look, there are a 12 number of problems, and I don't have the time this 13 morning to go into all of the difficulties. But one 14 of the more interesting facets that's come out of this 15 whole series is this table which I borrowed from Dr. 16 Ames' publication several years ago, which basically 17 demonstrates, looking at a series of some 380 18 different chemicals that were tested, the relationship 19 between whether or not they were carcinogenic or 20 whether or not they were mutagenic. 21 I think you can look at that table and 22 immediately see that perhaps one of both the advantage 23 of this chronic bioassay and the disadvantage is that 24 there are a variety of chemicals which, obviously, are 25 mutagenic and carcinogenic, which one might expect, 7 1 but there are also a large number that are perhaps not 2 mutagenic but still are carcinogenic. 3 So this type of system has led us to the 4 finding that we can do certain things with it, but 5 clearly we cannot begin to study mechanisms and 6 looking at dissecting the whole process of 7 carcinogenesis. 8 So as a result of that, of course, one 9 goes back a number of years, actually to the 1940s 10 when some of the earlier studies on what might be 11 called, what we call today at least, multi-stage 12 carcinogenesis were carried out. 13 This slide is just a classic experimental 14 slide which shows the studies that were done 15 originally by Barren, Blum and Schubik and others in 16 the 1940s which basically was done on the back of a 17 mouse, painting with a chemical, then administering a 18 material which at that time was an irritant called 19 croton oil, but today we call a promoting agent, 20 eventually ending up, if the format of the system is 21 appropriate, with neoplasm. 22 Now what I mean by appropriate is that 23 there is first an initiating agent, this material that 24 is given on the back of the skin, followed by a 25 promoting agent. If you reverse the process, it 8 1 doesn't seem to work, at least in most instances. 2 If you, for example, try just the 3 promoting agent, nothing happens; and the interesting 4 experiment that was done by Bautwell some 20 years 5 later was that, if you change the format of the 6 administration of the promoting agent, then you do not 7 get neoplasm. 8 So here was a system which allowed for 9 carcinogenesis to occur. The endpoint of this 10 particular system is seen in this slide, which is the 11 typical papilloma of the mouse. It is not a malignant 12 neoplasm. It is benign, but it is useful in this 13 particular system, which really tried to understand 14 what might be called the latent period of 15 carcinogenesis. 16 So this system was used, actually, for 17 about 30 years, during which time the chronic bioassay 18 developed. This system never became really important 19 with respect to regulatory agencies or anything of 20 that sort, but it was used primarily in academic 21 circles in trying to understand the whole process of 22 carcinogenesis. 23 It was in the 1970s that Dr. Carl Pareno 24 at the Argon laboratory actually first sort of broke 25 the ice by demonstrating that other tissues besides 9 1 mouse skin showed this exact same phenomenon -- that 2 is, of a stage phenomena in the development of 3 neoplasia. 4 Dr. Pareno's system, which is sort of 5 shown very briefly here, was to administer a 6 carcinogen -- in this case, this is actually a 7 modification of his which is done in our laboratory -- 8 with a mitotic stimulus followed by the administration 9 of some promoting agent. 10 This material, this promoting agent, which 11 did the same thing as the irritant in the skin system, 12 now had become sort of a much more well recognized of 13 what we call today promoting agent. 14 Over the years since Pareno's 15 demonstration, there's been sort of a parallel 16 development of both the mouse skin system and the rat 17 liver system, which is what Pareno studied, in 18 addition to -- and I won't have time to name them all 19 here -- probably a dozen other different tissue 20 specific multi-stage models of carcinogenesis. 21 So it's not just unique as it was a few 22 decades ago just to the mouse skin. It clearly goes 23 well beyond that. 24 So I'll spend just a few minutes talking 25 about the liver system, simply because that's what's 10 1 done in our laboratory. I think potentially it can be 2 utilized perhaps to go on to other model systems as 3 well. 4 One of the big advantages in the liver 5 system is that we think, and there's certainly 6 reasonable evidence to argue, that one may identify 7 initiated cells. That is the very first step in the 8 development of neoplasia. This is just a histologic 9 section which shows in the middle of this section a 10 single cell which now expresses a gene, the piriform 11 of glutathionase transferase, which normal hepatocytes 12 do not express. 13 We have found this to be a very useful 14 what is called a marker for identifying initiated 15 cells and, as you'll see, these cells then under the 16 influence of a promoting agent go on to clonally 17 develop into small colonies in the organ -- this is 18 all done in vivo -- expressing the same particular 19 genetic component. 20 Now this -- It turns out not only is the 21 expression of this particular gene in these cells 22 abnormal, but a whole series of others which have been 23 studied over the years by many, many different 24 laboratories, either as markers or trying to 25 understand what actually is going on during this 11 1 process of promotion. 2 As a result of that, looking to mining 3 years of work and many, many experiments, one can come 4 up with a series of conclusions for the effect of 5 promoting agents. 6 Now I'm sure in the discussion we can 7 spend an awful lot of time on various other aspects, 8 but I'm going to home just in on these two 9 characteristics which I feel are probably the most 10 critical aspects of these agents which cause a 11 selective enhancement of cell replication basically of 12 initiated cells. That is, they cause the replication 13 of these cells different from the normal cells. They 14 will also cause normal cells to replicate, but the 15 initiated cells they cause much more effectively and 16 selectively. That was shown several years ago by 17 Farber and others. 18 The other aspect that they also do has 19 been shown recently by Dr. Schulte Herman in Vienna. 20 They selectively inhibit the programmed cell death of 21 cells. So these two selective actions, actually, I 22 would propose -- and we can discuss it -- actually can 23 explain virtually all of the actions of promoting 24 agents. 25 Now, of course, one easily says what is 12 1 the molecular mechanism of this. I think at the 2 moment we really don't know. We certainly have a lot 3 of ideas about what's going on, but there's still a 4 lot of work to be done. 5 You can also see in this system, perhaps 6 unlike the mouse skin system, that it is possible to 7 quantitate the development of neoplasia. So if one, 8 for example, administers a -- initiates cells by just 9 a single dose, a very small dose of a carcinogen, you 10 can see you increase the number of these putatively 11 initiated cells by three orders of magnitude, which 12 perhaps is not unexpected. 13 Also notice that there are a certain 14 number which occur spontaneously. Then if you promote 15 and cause these to develop into these small colonies, 16 only about one percent of these develop into that; and 17 although I don't have the quantitation on there, one 18 can also go on into neoplasm and, in fact, of this 19 perhaps one-tenth of one percent of these cells 20 develop into neoplasm. 21 So it allows one in a model system to 22 quantitate the various stages and really get an idea 23 of how effective the various stages are. 24 Now to do this, there have been model 25 systems developed -- this is one of them -- where you 13 1 can actually dissect each of the three stages. One 2 can initiate with this very small dose of a 3 carcinogenic agent, an initiating agent, which by 4 itself will do nothing. 5 One can then promote, as shown with the 6 blue line here, with a promoting agent. Then 7 somewhere down the line after these focal lesions, 8 these small lesions, have occurred, one can administer 9 another agent which Bernie Weinstein coined the term 10 progressor agents, which actually then causes during 11 this period the development of malignant neoplasms. 12 So during this period from the 13 administration of the initiating agent to this point, 14 basically you have these focal lesions. The cells are 15 different phenotypically. They are undoubtedly 16 different genetically, at least from point mutations 17 and other things, but they are not different 18 cytogenetically. They have perfectly normal 19 cytogenetics. 20 It is in the stage of progression that the 21 cytogenetics of cells become abnormal, as we all know. 22 Not only do they become abnormal, but they 23 continuously get worse. That is, they continually 24 evolve in what is called evolving karyotypic 25 abnormalities. 14 1 From this, both this model and others, 2 again one can determine certain characteristics. I've 3 just listed some of the consequences of this major 4 factor of the stage of progression. Four of them 5 here, gene amplification, gene deletion, 6 rearrangement, and a very effective way, certainly 7 cells that are in the stage of progression are much 8 more effective at accepting external genetic 9 information by transection mechanisms than are normal 10 diploid cells. 11 Consequences of -- You might say the 12 functional consequences of this particular phenomenon 13 can be seen in this slide, just some of them. There 14 are probably a number of others. I've listed a few of 15 these here, and probably for the discussion of this 16 particular symposium, one of the most important ones 17 is this particular part right here. 18 It is well known that cells in the stage 19 of progression lose the expression of the MHC 20 determinant which, of course, makes them a very 21 difficult target for the immune system, because there 22 will be no interaction between the MHC system and the 23 T-cell receptors. 24 I'm sure that will be discussed in many 25 other components here, but it does allow one by using 15 1 this multi-stage phenomenon, then to dissect this 2 component into its various units and actually try to 3 determine changes which occur at each of the different 4 stages. 5 Finally, and I've taken this from an 6 article by Ray Tennant from the NTP, the more modern 7 models of carcinogenesis are transgenic models, either 8 transgenesis itself by adding genes in the standard 9 transgenic mechanism or by gene targeting of 10 knockouts. 11 What I've listed here are three -- 12 actually, two of the more common ones, the upper two 13 here, the T-53 knockout animal, both the homozygote 14 and the heterozygote, and the so called TGAC animal in 15 which the viral Harvey ras gene has been associated 16 with either a Zeta hemoglobin promoter or also, more 17 recently, with a keratin promoter, such that one can 18 actually do many of the skin tumor painting without 19 worrying about initiation, because basically one 20 already has initiated cell populations. So promoting 21 agents actually under this system become complete 22 carcinogens, in quotes at least. 23 Just briefly, looking at the upper one, 24 the P53 deletion mutation, this is taken from an 25 article that was recently published. This just shows 16 1 the tumors that develop in these animals, and you'll 2 notice that -- these are mice -- that the normal 3 animals spontaneously, at least out to this number of 4 weeks, develop very few spontaneous neoplasms. But 5 as; you might expect, in both the heterozygotes and 6 the homozygote P53 animals, the spontaneous 7 development of neoplasia is really extremely high. 8 This creates a problem perhaps when one 9 actually is trying to determine mechanisms, because it 10 is rather difficult to add to the system something to 11 perturbate it, while at the same time the system is 12 spontaneously developing a very large number of 13 neoplasms. 14 On the other hand, if you look at the 15 neoplasms that are produced in such a system, you can 16 see that there seems to be a somewhat, at least 17 quantitatively, different spectrum. Perhaps in the 18 homozygotes deletion animal, lymphomas are the 19 predominant component, whereas in the heterozygotes, 20 one can see that the lymphomas still are a significant 21 component of it, but both the soft tissue sarcomas and 22 osteosarcomas, which are really a very small part here 23 -- these soft tissue tumors become very prominent. 24 Now there are -- This is not the least in 25 this long list of modern models. This just shows you 17 1 -- Some of you in the back probably can't read this, 2 and it's really not that critical. It's not complete 3 by any means. There are a whole series of transgenic 4 models which are now developing. 5 The difficulty is, if, of course, one 6 wants to study the development of a specific neoplasm, 7 it may be extremely useless. On the other hand, 8 because the animal is programmed for certain things, 9 it becomes perhaps difficult to study the earlier 10 stages in the development of neoplasia and, rather, 11 what one will be looking at is the later stages 12 themselves. 13 So just to sort of sum up, what I've shown 14 you are three basic models: This chronic bioassay 15 which is perhaps, from the regulatory standpoint, the 16 gold standard, but from the investigatory standpoint 17 it has many problems; the multi-stage model which 18 allows one to look and dissect at the various stages 19 of neoplasia some of its mechanisms and its 20 characteristics; and finally, the newer transgenic and 21 knockout models which, certainly in specific areas, 22 can answer specific questions, but probably they have 23 to be geared and tooled to do this, and it seems that 24 they will probably be most significant in answering 25 questions in this final stage of neoplasia -- that is, 18 1 of progression. 2 So I'll stop there, and hope I kept on 3 time. 4 (APPLAUSE) 5 CO-CHAIRMAN MYERS: We have time for a few 6 questions. If you would come to the microphone and, 7 as we are transcribing the meeting, if people could 8 identify themselves, both by name and institution. 9 DR. LEOWER: Johannes Loewer. Has ever 10 DNA of neoplastic or normal cells been tested as 11 initiator or promoter in these -- If so, what was the 12 outcome? 13 DR. PITOT: I'm not sure of all the 14 experiments. I know it was tested once in mouse skin, 15 but probably not appropriately. It didn't work. But 16 I guess that, certainly, if one can use in -- as we'll 17 turn in the next discussion, in cell culture, one can 18 clearly by transection of DNA get a situation where 19 one can get -- I'm not quite sure what the stage one 20 is dealing with, but certainly a transformation. 21 DR. FRIED: I'm Mike Fried, ICRF, London. 22 You said in the beginning that certain 23 carcinogens were not mutagenic. What is the 24 mechanism, how they work, if it's not a genetic one? 25 DR. PITOT: Well, we can spend another 19 1 three hours on that one. I'll give you my opinion, 2 and I'm sure it can be discussed otherwise. 3 I showed you that in the putatively 4 initiated cells in the liver, there are a number of 5 spontaneous lesions. We well know that spontaneous 6 carcinogenesis is very characteristic of all mammals 7 probably, more so of some than others. 8 So one could make the argument that the so 9 called non-genotoxic carcinogens which are not 10 mutagenic in the chronic bioassay or in the Ames 11 system and other systems are, in fact, promoting 12 agents. What they are doing is they are causing the 13 development of spontaneously initiated lesions through 14 promotion and then spontaneously into progression. 15 That's a fairly simplistic answer, but I 16 think it follows the facts. 17 CO-CHAIR RUBIN: I want to call attention 18 to some recent work by Zarbell and Tilley with -- I 19 think it was NMU carcinogenesis in the mammary gland 20 of rats where they found that this so called mutagenic 21 agent, in fact, was causing mammary cancer, mammary 22 epithelial cancer. But it turned out that all the 23 tumors had a Harvey ras mutation in them, and that 24 Harvey ras mutation preexisted in the mammary 25 epithelium, and the agent -- you could call it then a 20 1 promoter, but in effect what it did was select with a 2 clonal expansion of those clones in the mammary gland. 3 DR. DOERFLER: My name is Walter Doerfler 4 from Cologne in Germany. 5 Isn't it a problem perhaps with the 6 transgenic model as people are now finding applying 7 the DNA array technology that, when one does a 8 knockout, perhaps quite a number of functions are 9 altered, not at all exclusively the functions that you 10 are knocking out, but other functions. 11 So the interpretation must become 12 extremely complex. 13 DR. PITOT: I certainly agree. I think 14 that the use of the gene targeted and transgenic 15 animals may be very useful, but I think it's new 16 enough that we really don't know all of the problems 17 that may occur. 18 AUDIENCE PARTICIPANT: In your multi-stage 19 approach, how specific to the species that you are 20 testing, the fact is, particularly related to the MHC 21 expression, it is clearly published -- the regulation 22 of the expression differs from species to species. 23 So, therefore, what could be changing 24 expression in your rat model may not be applied to 25 humans. You know very well that expression of MHC 21 1 genes is all over the place. It's up-regulation, 2 down-regulation. 3 So how can you apply this overall and 4 generalize your results? 5 DR. PITOT: I think that you see the sort 6 of thing that you're discussing in the human also in 7 the rodent, not only in the rat but the mouse, and the 8 same sort of system. So that you will find neoplasms 9 that will up-regulate the MHC components, others that 10 will down-regulate them. 11 Unfortunately, that it will happen in the 12 same neoplasm, that some cells will up and some will 13 down. So, therefore, you're going to be faced with a 14 problem, I think, from the vaccine question actually 15 of trying to get cells to up-regulate, and there are 16 mechanisms to do this, so that some of the vaccines 17 may well be able to work. 18 I think it is applicable to all different 19 systems, and certainly the human is perhaps the best 20 example. 21 CO-CHAIRMAN MYERS: Thank you very much. 22 Our next paper will be presented by our Co-Chair, Dr. 23 Harry Rubin, Professor of Molecular and Cell Biology 24 at Berkeley, and he's going to talk about multi-step 25 carcinogenesis. 22 1 CO-CHAIRMAN RUBIN: As most of you know 2 may know, our speaker was supposed to be Larry Loeb. 3 He was supposed to be the first speaker. Larry Loeb 4 suffered a back accident trying to climb into a cave, 5 which he shouldn't have done, with his grandson. 6 We tried to find at a late date some 7 alternate speakers. It was too late to get them to 8 come, and so as the designated chairman I volunteered 9 to speak. That will explain the rather primitive 10 state of some of my displays, which were -- Some of 11 them were made at four o'clock this morning. 12 So I would like to talk about -- 13 basically, about the transformation of cells in 14 culture, both the transformation to the neoplastic 15 state of primary cells, if you like, of normal cells 16 obtained from the animal and in some cases even from 17 humans, and then into permanent cell, the highest. 18 The first observation that was made in a 19 semi-systematic way about the transformation of normal 20 mouse cells into sarcoma cells beginning with fiber 21 blast was made, of course, by Wilton Earle of the 22 National Cancer Institute, beginning in about 1943, 23 and working together with Sanford they investigated 24 this problem for many years. 25 Of course, as many of you know, at first 23 1 what they were working with was carcinogen induction 2 of transformation in mouse cells, but they found that 3 their controls were transforming at the same rates as 4 the carcinogen treated cells were, and that started 5 the whole study of spontaneous transformation of cells 6 in culture. 7 Probably the earliest really systematic 8 investigation of the transformation of mouse cells in 9 culture, although that's not what it was called in the 10 particular paper, was by Todaro & Green in 1963, 11 Journal of Cell Biology, and a similar systematic 12 investigation was carried out by Paul Kraemar and 13 associates in New Mexico with Chinese hamster cells. 14 What I've listed here are the basic -- not 15 necessarily the sequence of changes, but an indication 16 of a variety of changes that occur during spontaneous 17 transformation of rodent cells in culture. So let me 18 just run down them. 19 You take the mouse fibroblast, in the case 20 of Todaro & Green, and while they multiply fairly well 21 when they're first put in cell culture, the rate of 22 multiplication decreases with every passage. They go 23 through a crisis. If you're transferring a small 24 number of cells -- in that case like three times 105 - 25 - you may, in fact, lose the culture, in a sense like 24 1 you lose human cells eventually. 2 If you have enough cells at each transfer, 3 every third day you go through a crisis period, which 4 means it looks like the cell culture is dying out, but 5 then there begin to appear variance in the culture, 6 and gradually you get an increase in cloning 7 efficiency. You get an increase in the rate of 8 multiplication back to the original rate and perhaps, 9 in some cases, an increased rate. 10 Then you find that the growth -- the cells 11 start to multiply to a higher saturation density. 12 Roughly about the same time, they will grow in low 13 serum concentrations or lower serum concentrations 14 than they would grow in before. 15 Then later on at a later step the cells 16 will grow in suspension in soft agar or methyl 17 cellulose. This is very often associated with an 18 ability to produce tumors in the animal, but not 19 always. It is not an invariant accompaniment of 20 ability to produce tumors. 21 Anywhere along these lines -- and we'll go 22 into this in a few minutes -- you begin to get 23 transformed colony morphology and transformed foci in 24 the culture. Some of these things such as the 25 transformed foci or production of tumors are spoken 25 1 about in a qualitative way when, in fact, they're 2 quantitative, progressive increases in the neoplastic 3 transformation of the cells. 4 You see that, I think -- In particular, 5 this was shown by the group that worked with Chinese 6 hamster cells at Los Alamos, Kraemer & Kramm, 7 etcetera. They found that their hamster cells went 8 through many of these stages, but as they were going 9 along, they were continually testing them for ability 10 to produce tumors in the isologous animals, in the 11 Chinese hamsters. 12 What they found, that even though all 13 these changes were occurring -- there were tissue 14 culture representations of changes in the behavior of 15 the cells after they had gone through their crisis 16 also -- they would not produce tumors, and that was 17 the case even where every cell in the culture had 18 shown chromosome aberrations. After the 20th passage 19 in culture, every cell in the culture was 20 chromosomally abnormal. 21 They were still not producing tumors when 22 injected back into Chinese hamsters, but about the 23 40th or 50th passage they could produce tumors, but 24 they could produce tumors only under certain 25 conditions. 26 1 The particular conditions were (a) to use 2 10 million cells -- a lot of cells -- and instead of 3 inoculating the cells directly subcutaneously into the 4 hamster, they were inoculated into gelatin sponges 5 that had been implanted subcutaneously. Under those 6 conditions, they produced tumors of the gelatin 7 sponge, which up to that time didn't worry the hamster 8 too much, but in fact they spread into the rest of the 9 hamster after that. 10 Then with further passages when chromosome 11 aberrations really became quite severe in all of the 12 cells, the tumors could be produced by direct 13 subcutaneous inoculation, and even later tumor 14 production would occur with smaller numbers of 15 inoculated cells. 16 Now the cells would also change -- this 17 was shown first, I think, in the mouse -- would change 18 after they produced the tumor in the mouse, if you 19 take the tumor out of the mouse, and the tumor 20 required the inoculation, let's say, of a million 21 cells. Once you got the tumor, you could produce a 22 tumor with 1,000 cells. 23 So that there was continuous progression. 24 What you see here is really a complexity of 25 progression, that each one of these stages that we're 27 1 talking about is, in fact, an indication of another 2 change in the population of cells. 3 Could I have the next transparency? Well, 4 this I did draw at about 4:30 this morning, and 5 they're from memory. So if someone has a better 6 memory than I do, please inform me. 7 This is taken from Todaro & Green's paper 8 in 1963, which is real classic. What's shown in these 9 top two graphs here is a plot on the horizontal axis 10 of the number of passages of the cells in culture, and 11 the increase in the cell population every three days, 12 n-zero being the number inoculated. Sometimes they 13 used n-1, that if they counted cells at one day and 14 then they counted them in three days when they passage 15 them. 16 At first, you could start out with a 17 population that would increase about eightfold in 18 those three days, doubling every day, but with each 19 successive passage the cells would multiply slower. 20 You saw a lot of pathology in the culture. 21 Then at about the tenth passage it looked 22 like the whole culture is going to die out. This is 23 in passaging three times 105 cells, which at least at 24 that time was thought to be a low inoculant of cells. 25 Then with further passages variance appeared. That 28 1 could be distinguished morphologically from the 2 original cells. 3 They continued to grow faster and faster, 4 and they reached a plateau, which was a relatively low 5 plateau. These cells would not produce tumors in 6 immunocompromised mice, for example, which is usually 7 said to be the case, but in fact if you wait three, 8 four, five months, very often a tumor will appear. 9 Again, when you get that tumor and 10 reinoculated it into mice, it will produce a tumor 11 very fast. So all that time these cells have been 12 incubating in an occult stage in the animal, and 13 finally a variant appears there, just like it appeared 14 in cell culture. 15 Now if one increases the concentration of 16 cells at every passage -- and this is a 3T12 passage 17 which means 12 times 105 or 1.2 times 106 cells were 18 passaged at each passage; three simply means they were 19 passaged every third day -- then you get a decline in 20 the rate of multiplication of the cells, but it's a 21 much shallower decline, for one thing. Although 22 there's something that resembles a crisis, it does not 23 appear to be a very serious crisis. 24 Then you start to get the upswing in the 25 rate of multiplication of the cells, and they end up 29 1 multiplying faster than these -- maybe not much faster 2 -- but what they do do is multiply to a much higher 3 saturation density. 4 These cells, when inoculated into mice, 5 will now produce a tumor fairly quickly. The only 6 difference here is that these cells have been passaged 7 at four times higher concentration than the 3T3 cells. 8 So there's something about the higher population 9 density of the cells that furthers the transformation. 10 In fact, a few years later in 1968-1969 11 Aaronson & Todaro found that if you took cells 12 directly from the mouse, start passaging them in 13 culture, and inoculating them back into mice every few 14 passages, that the higher the density of the cells 15 that you made your passages at, the quicker the tumor 16 would start to show up in the mouse. So there was 17 something either selective or inductive about high 18 population density in inducing tumors. It is 19 obviously the selective aspect that you select for 20 cells and that you grow at high density. 21 Now I want to say a word about human 22 cells. This is a paper by Smith & Hayflick in 1974, 23 and Hayflick can correct me on this. I'm sure I've 24 got some of it wrong. Again, it's from memory. 25 Basically, this was an experiment with 30 1 clones of human fibroblasts. I think it was the WY38 2 line. So what you're looking at here is the 3 proportion of cells that go through a certain number 4 of divisions in culture. 5 So I think it's widely thought that human 6 cells will go through 50 divisions and sort of fall 7 off the end of the cliff. That's based on work with 8 large populations of cells, but when you study clones 9 and their capacity to continue multiplication in 10 culture, the capacity drops off in some clones right 11 away, and right at about roughly the tenth passage or 12 so -- I can't really give you an exact figure -- about 13 half the cells have lost the ability to multiply. 14 So it looks like the loss of capacity to 15 multiply is a stochastic event. It's a random loss of 16 capacity to multiply. What you appear to end up with 17 is the last surviving clone. That's the clone that 18 goes 50 plus or minus ten divisions in culture. 19 That result was basically confirmed by 20 Peter Rabinovitch of Seattle in 1983. He used a new 21 technique for labeling the DNA of the cells, 22 gromodeoxyuridine label. He worked with whole 23 populations of cells, and he found that the percentage 24 of human cells that were cycling decreased linearly 25 with the number of divisions that the cells went 31 1 through in culture. 2 Now if you look at published literature 3 also, it looks as though human cells will not undergo 4 spontaneous transformation. There are many ways of 5 eliciting long term growth in culture, but there's 6 been some unpublished work, the senior author of which 7 was a cytogeneticist, Tamara Ignatova, was working in 8 Marguerite Vogt's lab at the Salk Institute a few 9 years ago -- unpublished work using Li-Fraumeni cells, 10 which are unstable -- relatively unstable human 11 fibroblasts. 12 Again, what she found was that, if you 13 kept the cells in a condition of confluence, which as 14 we all know results in contact inhibition of the cells 15 -- but I think it's not often appreciated, there's 16 also a good deal of cell death that occurs at 17 confluence -- then after a month or two of leaving the 18 cells intact in a confluent layer, you start to see 19 large scale chromosome rearrangements and indefinite 20 growth. 21 So these cells apparently have undergone 22 some kind of genetic changes involving chromosomal 23 changes that are fairly easy to see that results in 24 indefinite growth. That is, in effect, spontaneous 25 transformation. Unfortunately, as far as I know, this 32 1 work has never been published, and I think there's 2 some disagreement among the group that's done it about 3 aspects of it should ultimately get published. 4 Do I have another? That's it? Well, I 5 want to turn to some slides now. I want to deal now 6 for a few minutes with a cell line, but what I'm going 7 to say about this cell line, I think, can be 8 generalized to cell lines in general. 9 So what I want to talk about is the NIH 10 3T3 cell line, which is famed in modern molecular 11 genetics as the first line in which there was 12 demonstrated a transformation of an animal cell line 13 by DNA extracted from a human tumor, the EJ bladder 14 carcinoma. 15 A problem with that original finding -- 16 there are many problems with the original finding 17 which three laboratories reported back in 1981 -- was 18 the fact that again, if you take the original line of 19 NIH 3T3 cells and you let them sit at confluence for 20 a couple of weeks, that you start to see spontaneous 21 transformation. 22 In fact, probably the outstanding 23 characteristic of Jane Hill, Todaro and Aaronson's 24 original line of NIH 3T3 cells is the ease of 25 spontaneous transformation. All you have to do is 33 1 leave the cells for two weeks at confluence and 2 transfer them once, and foci start to show up 3 throughout the culture. I'll give you some examples 4 of that. 5 So again there's something about 6 surprisingly enough from conventional genetic thinking 7 where one thinks that at the highest rate of mutation 8 or genetic change is during cell division when DNA is 9 being replicated. Actually, it appears that the 10 highest rate of chromosomal change of mutational 11 change -- we're not sure which, or maybe all of them 12 together -- occurs when the cells are inhibited at 13 confluence. 14 You might ask why. Well, there's 15 something else which I already inferred, that there is 16 a considerable amount of cell death when these cells 17 or many cells are left at confluence for an extended 18 period of time. So there's considerable damage done 19 to the culture. 20 So let me show you a few slides of the 21 kind of observations that one makes. Let's go back to 22 the first one. Okay. 23 What you see here is a culture of NIH 3T3 24 cells which ordinarily would look like this, a mono- 25 layer of cells, nontransformed cells, but when left 34 1 for a couple of weeks or if left for a couple of weeks 2 and transferred, you get a high density population. 3 This is a transformed focus where the cells are no 4 longer in any regular arrangement. They're criss- 5 crossing. You can't identify individual cells, 6 because they're piled so thickly. 7 This is an unstained slide, which was one 8 of the first observations we made back in about 1988- 9 1987, which surprised us that the cells which people 10 were talking about as being used as targets for 11 transformation by the Harvey ras oncogene would 12 transform by themselves. 13 Now another feature is, when one looks at 14 independent transformations -- and now we're getting 15 to another point in transformation. So you start out 16 with a single culture, and you start splitting it, and 17 you let each one of them go to confluence various 18 numbers of times. You get independent 19 transformations. 20 If you look at the foci in each one of 21 these dishes -- these are the cuts from individual 22 dishes -- in a rough sense, the foci, which are these 23 thick aggregations of cells on a background of mainly 24 monolayers of cells -- each of the groups of foci look 25 different from one another. 35 1 So you can't simply talk about 2 transformation of cells by producing foci. These -- 3 If you collect the cells from any of these foci and 4 inoculate them into mice, they'll produce tumors, and 5 the tumors in the mice will all look like sarcomas. 6 But in culture where you can find a distinction in the 7 appearance of the cells, each one of the independent 8 transformations is different from every other one, 9 which means that the genetic changes that are 10 occurring in these cells are different in each one of 11 these cases. 12 So there are genetic changes, chromosomal 13 rearrangements, deletions, etcetera, that are going on 14 in these cultures that cause that kind of variation, 15 which is reminiscent of what pathologists were telling 16 us way back in the Thirties and Forties, that no two 17 tumors look exactly the same. 18 Well, okay, now I want to show you want 19 will happen with the NIH 3T3 cells if you culture them 20 in a certain way. So what you see here are a series 21 of cells that have been through what we call a primary 22 assay, which means two weeks at confluence, and every 23 two weeks they are transferred to become confluent 24 again, and this is a tertiary assay, and this is after 25 eight such sequences. 36 1 You start out with a culture that has no 2 foci. By the third round of confluence, even when you 3 place only 1,000 cells on a background of 4 nontransformed cells, you see these thick foci. By 5 the eighth passage at confluence, even 200 cells are 6 producing about 100 foci. So that's a high rate of 7 transformation. 8 Here is a culture derived from this same 9 initial culture, but transferred originally about 100 10 times at low density. You go through the same 11 operations, and you see much less evidence of 12 transformation. 13 So there's continuous selection and change 14 going on in these established cultures, and one cannot 15 rely on them by saying, well, they're going to remain 16 the same as long as we continue passaging them. It's 17 very dependent, just like the original transfer of 18 cells directly from the mouse -- dependent on 19 population density. Even the established lines depend 20 on population density. 21 If you look at another criterion of what 22 was happening to these cells, in these cultures if you 23 took the undiluted cell populations, you got a very 24 rapid rise in saturation density, which remained 25 roughly constant after that; whereas, in this case 37 1 even by the seventh round of confluence, there was 2 only a minimal increase in saturation density. 3 Another important aspect of these 4 observations, contrary again to what you might expect 5 -- You might think the cells which are becoming 6 transformed would certainly grow more at high density. 7 If you would grow them at low density, would they grow 8 any faster? Quite the opposite is true. 9 Actually, at low density these cells grow 10 slower and slower, which is another indication that 11 what confluence -- extended confluence is doing to 12 these cells is damaging them, and apparently damaging 13 their DNA. The problem is that occasionally in some 14 of the cells, that particular type of damage results 15 in transformation. 16 Now if you take the original line of NIH 17 3T3 cells and clone it out, what you see here is 18 three different clones. We've done this with very 19 large numbers of clones, but these are illustrative of 20 the heterogeneity you get within any single 21 population. 22 So this clone 1A actually started to show 23 even in the first round of confluence very tiny dense 24 foci. By the second round, it was showing these huge 25 foci. Then we had to dilute it out, and it continued 38 1 to produce large foci. 2 This clone of cells transformed more 3 gradually, and here you can see progressive 4 transformation within a clone. At first, light foci, 5 broad foci are produced. Then denser foci begin to 6 show up, and finally you get foci with this clone that 7 are just about the same roughly as this clone. 8 You take clone 4B here, and in spite of 9 many rounds of confluence, here you seem to get one 10 light focus, but in the sister dish which was being 11 transferred and not being fixed, apparently were no 12 foci. Even after five rounds of confluence, there's 13 only a minimal amount of transformation. 14 So what this means is every time you are 15 working with a mixed population or an uncloned 16 population, it's always a heterogeneous combination of 17 clones that behave differently from one another in 18 transformation. 19 Okay. Now I want to talk about one other 20 point. I tried to emphasize the point that an 21 important aspect in transformation, both of primary 22 fibroblast or at least explants directly from the 23 animal, and in established cell lines is high density, 24 is contact inhibition of the cells. 25 I might mention parenthetically, it's now 39 1 been clearly established in bacteria, which is always 2 our great sounding board for genetic change -- Warner 3 Arbor who won a Nobel prize working with bacteria that 4 led to very important discoveries found that, if he 5 left a culture, several cultures of bacteria sitting 6 around for 17 years so they hadn't grown at all during 7 that time except the first day that they were sitting 8 there, that there were enormous chromosomal 9 rearrangements that occurred even in a single 10 chromosome of bacterium. 11 So cells that are being starved, in 12 effect, are great candidates for genetic change, both 13 in bacteria and apparently in animal cells. 14 Now what I'm going to illustrate in this 15 last pair of slides is something that came as a great 16 surprise to me. Having worked with cells for about 50 17 years now, I thought it was pretty hard to surprise 18 me, but these cells are pretty clever. 19 It was this. We were working with this 20 subline of NIH 3T3 cells which was very difficult to 21 transform by keeping them at confluence, and we 22 decided, well, we ought to look at what happens with 23 clones of those cells to see how heterogeneous they 24 are. 25 What we found was quite a shock to us. 40 1 Here we have two clones -- they're representative 2 clones, clone 1A of this resistant line and clone 2E. 3 Then the parental population, which we're calling 4 large A prime here. 5 Again, we go through one round of 6 confluence, and then this is the fourth round of 7 confluence and the sixth round of confluence. The 8 shock was that, while the parental culture -- which 9 is, after all, made up of all of these clones -- was 10 exhibiting very little, if any, transformation or the 11 barest minimum, the clones were showing a lot of 12 transformation. 13 So what does this mean? I won't go into 14 this point here. So we quantitated that on the next 15 slide. This is the last slide. And we made a scale 16 of transformation, which is shown here. 17 So none of them transformed to the extent 18 of producing these really thick foci, but they would 19 produce distinguishable foci, and they would produce 20 foci even on a background of nontransformed cells. 21 Just look at this first panel here. We 22 don't have to go into the second one. It's sort of a 23 cumulative observation of relatively heavily 24 transformed cells -- that is, this type here -- or an 25 accumulation of those plus more moderately transformed 41 1 ones -- these here -- or more moderately and lightly 2 transformed ones. 3 So the combination of all of them after 4 five rounds of confluence had involved about 80 5 percent of the clones that we had. At that time, 6 shown on the scale at the top, the parental culture 7 had shown no transformation at all, in spite of the 8 fact that it was made up of thousands of those clones. 9 So could I have the lights, please? So 10 the right panel is a repetition of the same 11 experiment. What that says to us, that when you get 12 a heterogeneous mixture of clones from the parental 13 population, there is some kind of mutual protection 14 against transformation. 15 We later learned that back in 1981 George 16 Post had found something very similar to this with 17 melanoma cells in the mouse, that parental populations 18 retained their consistency of behavior; whereas, the 19 clones obtained from those populations underwent a 20 great deal of variation. 21 What we have to think about in that case 22 is what that means for progression in the animals, 23 because the general model that Henry Pitot was 24 presenting to you, and general model that Wallace 25 Clarke, the great melanoma specialist, has emphasized, 42 1 that in many systems in humans, including the 2 formation of melanomas, I think, of liver cancers, of 3 colorectal cancers, of skin cancers, all seem to go 4 through a sequence, let's say, -- colorectal is 5 probably the best known these days -- of the formation 6 of foci of cells developing into polyps, into 7 adenomas, all of which are of monoclonal origin. 8 It's from those monoclonal, benign tumors, 9 whether they are moles or adenomas or warts or, in the 10 case of the liver cancer models in animals, these 11 altered hepatic foci, that the next step is most 12 likely to occur. 13 So that's a new parameter that we have to 14 deal with where we get an association between the 15 changes in the cell culture reflecting, we think, the 16 effect of clonal expansion in the organism. Thank 17 you. 18 (APPLAUSE.) 19 DR. ONIONS: Harry, I think you've proved 20 that you don't need much warning to -- 21 CO-CHAIRMAN RUBIN: Oh, I didn't go over. 22 That's a first. 23 DR. ONIONS: Is there time for one or two 24 questions? 25 DR. COFFIN: John Coffin, Tufts. What is 43 1 known, if anything, about the molecular changes that 2 are associated with these different colony 3 morphologies, for example? 4 CO-CHAIRMAN RUBIN: If you're asking me 5 what have we done, nothing. 6 DR. COFFIN: Well, what is known 7 elsewhere? 8 CO-CHAIRMAN RUBIN: Well, let me tell you 9 something negative. This is work Stuart Ansen did 10 with the NIH 3T3 cells before we ever got them. 11 So he encountered this spontaneous 12 transformation as well. So he tested the spontaneous 13 transformation by extracting DNA and doing the 14 classical transfection experiment into nontransformed 15 NIH 3T3 cells. 16 He found essentially none of them were -- 17 the DNA of none of them was able to transform more 18 rapidly than as spontaneous transformation went on. 19 So there's no indicating, at least, that the Harvey 20 ras gene had mutated there. 21 What we know from classical genetics of 22 cells in culture, actually, was done with other cell 23 lines that was independent of transformation is that 24 when you get chromosome rearrangements or deletions, 25 like the deletions that you get in loss of 44 1 heterozygosity where you lose real chunks of the 2 genome, is you end up with cells that will grow slower 3 than the original cell. 4 If you get point mutations, it's very rare 5 that that slows down the rate of multiplication, at 6 least in the thymidine kinase gene, which has been 7 looked at. 8 So we think that the correlation that we 9 see of reduced growth rate at low density of the 10 transformed cells parallels the classical genetic 11 findings, and it's likely that what we're seeing is a 12 lot of chromosome rearrangements and deletions in the 13 cell, but we don't have any proof of it. 14 DR. COFFIN: If I may raise another 15 question. 16 CO-CHAIRMAN MYERS: Make it short. 17 DR. COFFIN: Yes. Can you separate the 18 slow growth property from the transformation property 19 by forcing rapid passages to transform cells, for 20 example? 21 CO-CHAIRMAN RUBIN: Yes. Well, you always 22 then select for faster growers, and it varies with 23 whatever population you use. In one population you 24 can select from the faster growers, and they turned 25 out to be less transformed, and in another population 45 1 you don't. But you can do it. 2 Not all cells that grow slower are 3 transformed. That seems to be a more general finding 4 that involves the whole population. It's only a 5 subset of those, presumably particular chromosomal 6 rearrangements, deletions, etcetera, that result in 7 transformation. But a large proportion of those 8 changes result in a slowing down of the growth of the 9 cell at low density. Okay? Thank you. 10 CO-CHAIRMAN MYERS: Thank you. Our next 11 paper is the transformation by DNA viral oncogenes by 12 Dr. Alex van der Eb from Leiden University. 13 DR. VAN DER EB: So as you have heard last 14 night and you will hear in the coming thoughts during 15 this meeting, diploid human cells have a finite life 16 span in vitro. They divide a certain number of times 17 and then they stop dividing. 18 This property limits to some extent their 19 usefulness for the production of viral vaccines or 20 production of viral factors for gene therapy. 21 Now cell lines, continuous cell lines, are 22 immortal and are, therefore, more suitable in certain 23 aspects. However, there may be certain risks 24 associated with the use of continuous cell lines. 25 Therefore, it would be helpful if we would be able to 46 1 immortalize diploid human cells without transforming 2 them. 3 So far the only means of immortalizing 4 diploid human cells is by -- reproducibly 5 immortalizing is by transforming them with a DNA 6 virus. Whereas spontaneous transformation or 7 spontaneous cancer is a multi-step process which 8 requires a large number of different gene mutations or 9 changes or alterations in genes that accumulate over 10 the years, transformation by DNA virus apparently 11 seems to be a one-step event. 12 This is due to the fact that the viral 13 transforming genes are -- or gene is a multifunctional 14 protein that alters simultaneously a number of 15 different regulatory pathways in the cells. 16 I would like to discuss briefly 17 transformation and immortalization by DNA tumor 18 viruses, and particularly focus on SV40, an adenol 19 virus, and also say a few words about HPV, but that 20 will be more extensively dealt with by Dr. McDougall 21 later during this meeting. 22 If I can have the first slide, please. 23 This slide shows the SV40 large T antigen, 24 which is the major transforming gene of the SV40 25 virus, and it also shows the three main transforming 47 1 domains in this gene. 2 On the righthand side you see the P53 3 binding sites which overlaps with ATPase binding site, 4 which is very important for transformation by SV40. 5 A second site is more to the left, and that coincides 6 with CR1, although CR2, the conserved regions 1 and 2, 7 also is important. That is a site that's responsible 8 for binding of the RB protein and the RB family 9 proteins as well as the coactivator B300 or CBP. 10 In addition -- So this is the second 11 important part for transformation, and in addition a 12 third part is the N terminus which is the DNAJ-like 13 protein -- like domain which resembles the DNAJ 14 proteins of E. coli that have an important role in 15 molecular chaperons for -- in conjunction with HSB 16 proteins. 17 How the DNAJ and terminus of SV40 18 contribute to transformation is still unclear, but it 19 is clear that P53 binding and inactivation of the RB 20 proteins or P300 really very importantly disrupts the 21 main growth control pathways in the cell. 22 Adenovirus transforms basically in a 23 rather similar way, and the next slide shows the 24 adenovirus E1A gene, which is the major transforming 25 gene of adenovirus. As you can see here, there are 48 1 two again conserved regions here. One is CR2 which 2 are important for transformation. 3 The third region, CR3, is not important 4 for transformation, but another region, the N terminus 5 which is not conserved among adenoviruses, is on the 6 other hand again essential for transformation. 7 So the N terminus, CR1 and CR2 are 8 important for transformation of primary cells, as well 9 as association to cellular proteins. The N terminus 10 and CR1 are responsible for binding to P300 and CBP as 11 well as the P400 coactivator which is similar to -- 12 more or less similar to P300, but as CR2 and CR1 are 13 important for binding to the retinoblastoma protein 14 and the related pocket proteins, so again basically in 15 a rather similar way compared to the SV40. 16 Now the adenovirus E1A gene confers such 17 a strong growth promoting effect on cells that these 18 cells apparently -- certainly, if one has primary 19 cells in which E1A is expressed to a reasonable 20 extent, that these cells respond by activating their 21 P53, and this will lead to either growth arrest or 22 apoptosis. 23 So to counteract this growth arrest or 24 apoptosis, the E1B region is necessary. So you see 25 here the E1A region with the two proteins -- with the 49 1 two RNAs of proteins that I just showed you with CR1 2 and CR2, as well as the N terminus which is located 3 here. But the E1B region calls for two proteins, a 4 large protein and a small protein, and these are 5 essential for neutralizing the apoptotic and cell 6 cycle arrest activity of the E1A region. 7 So E1A and E1B are needed in order to 8 transform cells, because of the effect of E1A on P53. 9 Therefore, it is almost impossible to 10 transform -- obtain E1A transformed cells alone, and 11 very few cells -- and those are rodent cells that are 12 obtained that are transformed by E1A alone -- express 13 E1A to very low levels. 14 HPV, the human papilloma viruses, 15 basically transform again in a similar way. Next 16 slide, please. 17 The two transforming genes or papilloma 18 viruses are E6 and E7. E6 targets P53 and causes its 19 degradation, but E7 again targets the other growth 20 regulatory proteins, PRB, the RB protein and its 21 family members as well as probably P300, but that 22 we'll hear later in more detail. So they resemble 23 again the adenoviruses which have basically the same 24 properties. 25 Now not all these cells react in the same 50 1 way to these transforming genes or viruses. Next 2 slide, please. Here I show you SV40, how SV40 3 transforms cells, human fibroblasts, diploid 4 fibroblasts or epithelial cells and keratinocytes. 5 Fibroblasts are very efficiently 6 transformed by SV40. Immortalization, however, is 7 very rare, at least 10-7 events per senescent cell, 8 and there is a pronounced extended life span. 9 Epithelial cells, on the other hand -- not 10 on the other hand, but epithelial cells more or less 11 at the same weight are responding more or less at the 12 same rate as SV40. There is morphological 13 transformation. There is also rare immortalization. 14 It's not completely clear how much their extended life 15 span is, but I believe that there is an extended life 16 span again in these epithelial cells. 17 The situation is different for adenovirus 18 E1. Forget E1B. This is an old slide where I thought 19 that I had E1A transformed human cells, but this is 20 not the case. 21 Fibroblasts or epithelial cells appear to 22 be surprisingly resistant to transformation by 23 adenovirus. Frank Rijm in 1974 has done many attempts 24 to transform human embryonic retinal cells with DNA of 25 adenovirus, and he was initially completely 51 1 unsuccessful until he found one clone, and this clone 2 gave rise to the well known 293 cell, and that is 3 really the only clone that he has seen transfected 4 with adenovirus. 5 Fibroblasts similarly do not show 6 transformation, and I mean now transformation 7 according to the focus assay. You can introduce by 8 transfection the adenovirus E1 genes into human cells 9 where they are expressed, but we have the impression 10 that expression is lost after a while. There is no 11 morphological alteration. 12 These transformed cells -- and I'm talking 13 now about 293 cells -- They immortalize probably after 14 short crisis periods. Efficiency of immortalization 15 is not so clear, because there's only, as far as I 16 know, only one example. It's not so clear whether 17 there is an extended life span. 18 So we then switch to embryonic retinal 19 cells, because we had heard from work of Phil Denimore 20 and others that neural cells, cells of neural origin, 21 could be transformed more easily also in the rate 22 system, the rodent system; and we switched to 23 embryonic -- primary embryonic retinal cells, and they 24 can be transformed quite efficiently by adenovirus E1, 25 although still at relatively low numbers compared to 52 1 transformation of rat kidney cells, for example, but 2 it is reproducible. 3 Also there is a high frequency apparently 4 of immortalization, and there is no apparent crisis, 5 which is rather surprising. 6 I will not say much about HPV E6 and E7. 7 There is little interaction with fibroblast. There 8 may be reported very rare immortalizations by E6, E7, 9 but epithelial cells at keratinocytes are immortalized 10 more frequently, and apparently there is no distinct 11 crisis. 12 So transformation of human cells by SV40 13 has been studied most extensively. So I will briefly 14 turn now -- go back now to SV40. 15 If diploids and diploid fibroblasts -- If 16 diploid fibroblasts are transformed by SV40, then a 17 number of changes occur, and these are depicted here 18 in this slide. 19 The normal diploic fibroblasts grow to a 20 certain saturation density during a number of 21 population doublings, and then they stop dividing, and 22 an irreversible arrest, growth arrest, which is called 23 senescence, and it has been mentioned before. 24 No immortalization will ever occur, as far 25 as I know, from these senescent cells. They do not 53 1 die, but they sit there just sometimes for many 2 months. 3 If you transform the culture at this stage 4 for a -- with SV40, then the saturation density 5 increases, and the transformed cells seem to ignore 6 this senescence arrest phase and just go on for a 7 number of -- for a large number of passages sometimes, 8 which causes an extended life span. However, in the 9 end the cell -- the whole protein will die and enter 10 into so called crisis where the cells -- the 11 transformed cells really die, unlike the situation in 12 senescence. 13 Only in a very few cases, very rare cases, 14 an immortalization will occur after a shorter or 15 longer period, and this will etherize to a cell which 16 is the same as the SV40 transformed cell before the 17 crisis, but now they are immortal and can grow 18 indefinitely. 19 So what is the basis of the appearance of 20 the senescence and the crisis? That has to do with 21 the telomeres. In contrast to the germ cells which 22 have telomeres active in these cells, the normal 23 somatic cells are telomeres repressed and have no 24 telomeres activity. 25 So during population doublings, t he 54 1 telomeres are lost, become smaller and smaller until 2 a certain stage is reached, which may be about two- 3 thirds of the length of the telomeres, approximately 4 two-thirds, but certainly not all telomeres have been 5 used up, and at that stage the cells start senescence. 6 This senescence is called M1 or mortality 7 phase 1. If you, in contrast, inform with SV40, again 8 the M1 senescence is ignored, and the cells continue 9 dividing, and also the telomeres become shorter and 10 shorter until they are almost completely disappeared. 11 That coincides with the crisis which is 12 also called mortality stage 2. So the M2. If there 13 is no immortalization, the cells -- all cells will die 14 here. Immortalized cells have now in most cases 15 active telomerase. 16 So this is then the extended life span 17 between senescence and crisis, and it's also this 18 period which is characterized by the onset of 19 chromosomal abnormalities. So there are -- and that 20 effect already was also seen in the normal cells. 21 When they approach crisis, there is chromosomal 22 rearrangements. There are dicentrics formed, and so 23 on. 24 That is probably due to the fact that 25 during senescence the P53 pathway, P53-P21 pathway, is 55 1 activated for some reason. So what happens during 2 senescence is that at a certain critical length of the 3 telomeres -- we don't know exactly what the signal is 4 -- will cause activation of preexisting P53. There is 5 probably not more transcription of the P53 gene, and 6 this P53 causes accumulation of active P53 which 7 causes activation of transcription of the P21 gene, 8 the WAF-1 gene, which in turn is an inhibitor of 9 cyclin/CDK and inhibits the cell cycle. So this 10 triggers apparently a certain minimum length of 11 telomeres. 12 Now catalytic subunits of the human 13 telomerase has been isolated. The question can be 14 raised would it be possible just to introduce 15 telomerase in cells and immortalize them in diploid 16 cells and immortalize them in that way? 17 Indeed, there has been reports that this 18 is indeed the case. Dr. Hayflick has already 19 mentioned an example yesterday. Also Dr. McDougall 20 will probably talk about this later, but as Botnar and 21 coworkers have also found, that fibroblasts as well as 22 retinal epithelial cells become apparently immortal 23 when the human telomerase catalytic subunit is 24 introduced into the cells. 25 Rather, Piono et al. showed that mammary 56 1 epithelial cells as well as keratinocytes do not 2 immortalize just with only introduction of the 3 telomerase, and this works that immortalization is 4 obtained when, in addition, the E7 gene is added to 5 the cells. That, I think, will be discussed later in 6 much more detail by Dr. McDougall. 7 So this means that, if you inactivate the 8 RB, then that would lead to immortalization. This may 9 be due to the fact that there is apparently a third 10 type of senescence, and that third type of senescence 11 is called N0 and occurs usually before M1. 12 It is known that, unlike fibroblasts that 13 can be found a long time, 50 to 60, 70 or sometimes 80 14 population doublings, depending on the cell strain and 15 also depending on the lab, I have the impression, that 16 many other cells when taking a tissue culture form 17 human tissues will not grow very long, and after ten 18 or maybe 20 passages they just stop dividing. This is 19 long before the telomeres have reached the size, the 20 minimum size which corresponds to the senescence, M1 21 senescence. 22 Now Weinberg has suggested that this M0 23 state is due to a kind of physiological stress and is 24 caused somehow by suboptimal growth conditions that 25 occur in vitro. So introduction of telomerase in 57 1 cells that still have to undergo an M0 may not work, 2 because M0 is cause for physiological stress and has 3 nothing to do with the size of the telomeres. On the 4 other hand, cells that have no M0 but only an M1, 5 there it may work, but this is all theory, and we'll 6 see if it is really true. 7 So the last slide then summarizes here for 8 the two senescent stages that occurred in certain 9 cells, including the mammary epithelial cells but 10 probably many more cells, and that is that M0 occurs 11 after -- in the case of mammary epithelial cells, 12 after about 20 population doublings, and may be 13 activated by physiological stress. 14 It's controlled by the RB P60 pathway, and 15 can be bypassed by HPV. It must be E7. This must be 16 E7, right? It's wrong. Jim, it's wrong. It is E7. 17 I'm sorry. 18 This was made just before I left, and 19 there are more mistakes here, as you can see. The 20 computer did a trick and did not exactly what it was 21 supposed to do. 22 Anyway, it's bypassed by HPV E7. One, on 23 the other hand, may be triggered by certain extent of 24 telomere shortening, and which is actually the mitotic 25 clock, and is controlled by the P50-CP21 pathway, and 58 1 that is bypassed by HPV E6. 2 M2 is called crisis leading to cell death 3 and is probably caused by extensive telomere 4 shortening. It has nothing to do with all these 5 things. 6 So I think I'll stop here. Thank you. 7 (APPLAUSE.) 8 CO-CHAIRMAN RUBIN: We can have a few 9 questions now. 10 CO-CHAIRMAN MYERS: Could we have the 11 lights, please. 12 CO-CHAIRMAN RUBIN: Lights, please. 13 Could we have the lights? Yes, thank you. 14 DR. HAYFLICK: Hayflick, UCSF. Unless I 15 misunderstood your introductory remarks, there are 16 indeed two other ways in which one can transform 17 normal human fibroblasts to an immortal cell 18 population. Namely, we have done this with exposure 19 to cobalt 60 radiation, producing a cell line called 20 SUS M1. 21 It's also been done with chemical 22 carcinogens, producing a cell line called KMS T6, in 23 addition to H terp which you've mentioned. I also 24 should mention that, contrary to popular belief, there 25 are several publications in the open scientific 59 1 literature reporting spontaneous transformation of 2 normal human cells. 3 For those of you who would like those 4 references, please contact me, and I'll be happy to 5 supply you with them. Thanks. 6 DR. VAN DER EB: Thank you. Yes, you can 7 immortalize -- transform or immortalize human cells 8 also by radiation or chemical carcinogens. I think 9 this is often quite difficult. You have to expose the 10 cells many times, as far as I know at least, to the 11 chemicals or to the radiation, and it is a lot more 12 easy just to take SV 40 large-T for fibroblast. But 13 you are completely right, yes. 14 DR. HAYFLICK: Well, the reason I 15 emphasize that -- 16 CO-CHAIRMAN RUBIN: Can you get to the 17 microphone? 18 DR. HAYFLICK: The reason I thought it was 19 worth emphasizing is that the thrust of this meeting 20 is the development or the consideration of immortal 21 cell populations other than those that might be 22 currently used for vaccine production. 23 Two of the obvious choices, if one wants 24 to avoid the overt introduction of viruses or their 25 fragments into cells, is by the use of radiation or 60 1 chemical carcinogens. 2 DR. VAN DER EB: I know from cells that 3 are spontaneously immortalized at least and also 4 clearly transformed that they have lost quite a lot of 5 things like P53 and also the RB pathway. So what you 6 do then is basically the same as what occurs or 7 happens in spontaneous transformation, I think. 8 CO-CHAIRMAN RUBIN: One more question. 9 DR. BERKOWER: Ira Berkower from the FDA. 10 One concept that we've been struggling 11 with is the notion of cells that are more malignant 12 and less malignant, more transforming and less 13 transforming, with the idea that if vaccines could be 14 made in less transformed cells, that will be safer. 15 Does that make any sense in terms of these 16 molecular mechanisms that you're studying, that there 17 would be a more transformed phenotype? 18 DR. VAN DER EB: What is a more 19 transformed phenotype? 20 DR. BERKOWER: Or a more malignable 21 phenotype? 22 DR. VAN DER EB: This is difficult, I 23 think. If I see it now -- If I look at the 24 literature, then I have the impression that if you 25 take immortalization as the important final step that 61 1 you want to reach, that then HPV genes may be more 2 suitable. 3 I have the impression that they are -- The 4 cells look less transformed at least than in the case 5 of SV 40, for example, but also in the case of adeno. 6 But there's very little I can say to that, I think. 7 It's so difficult. What is immortal? Which cell is 8 more transformed than the other? 9 You would have to compare things like 10 tumorigenicity. I should say that cells transformed 11 by viruses are often not yet immortalized, I showed 12 you, but also if they are immoral, they are not yet 13 immediately tumorigenic. 14 So maybe tumorigenicity would be more 15 important than just looking at the cells, how they 16 grow. 17 CO-CHAIRMAN RUBIN: Okay. We'll have to 18 move on to the next speaker, who is Dr. James 19 McDougall from the Fred Hutchinson Cancer Research 20 Center at Seattle, Washington. 21 DR. McDOUGALL: Well, as Alex has already 22 pointed out, the two things that I would rather talk 23 about are human papilloma virus and telomerase 24 immortalization. But when Andy called me -- Andy 25 Lewis called me -- he said that somebody else had 62 1 dropped out of talking about hit and run, and so I got 2 lumbered with the job. 3 It's interesting. It's interesting, but 4 it's controversial, without doubt. I'll just put my 5 first slide on. In the olden days, and there are a 6 few of us that were there in the olden days, one of 7 the most interesting studies that was going on in 8 terms of looking at how viruses might transform cells, 9 and particularly how they might contribute to human 10 cancer, was the belief that herpes viruses might very 11 well be responsible for, for example, cervical cancer. 12 There were good reasons for believing 13 this. For example, serology gave very good evidence 14 that women with cervical cancer had high levels of 15 antibodies for herpes simplex, and it became clear 16 that this was herpes simplex Type 2. 17 So as time went on, it looked reasonable 18 to try and sort out whether or not this virus had 19 fragments, had subgenomic fragments that could 20 actually transform cells. That was one of the studies 21 we concentrated on in the early 1980s. 22 The studies were carried out mostly in rat 23 cells or hamster cells, and lo and behold, we were 24 very pleased to find that we could actually transform 25 these cells and, if we grew these cells up from an 63 1 initial transformation by exposing them to herpes 2 virus DNA and to fragments of herpes virus DNA -- and 3 I'll show you the specific fragments in a minute -- 4 that those cells would then produce tumors in their 5 rat or mouse host. 6 So this looked very good, and we felt we 7 were on the track of what might be the key genes that 8 were involved in the development of an important human 9 tumor, in this case cervical carcinoma. 10 So over the years there were a lot of 11 examples of how herpes viruses might perhaps have a 12 hit and run type of effect rather than a similar 13 situation that we see with the adeno viruses and SV 40 14 and human papilloma viruses where it's clear that very 15 specific fragments persist in the cells that are 16 either immortalized or transformed by those. 17 I just put up a series of papers that all 18 refer to the hit and run phenomenon related to herpes 19 viruses. 20 So if we look at the herpes virus genome - 21 - and here is the genome along the top here, and what 22 I really want to bring your attention to are these 23 fragments here that map in this region of the genome. 24 We initially show that this one called 25 morphological transforming region 1 would carry out 64 1 that transformation of both rat and mouse cells at a 2 reasonably efficiently level, almost at the same level 3 as SV 40 would transform these cells. 4 Subsequently, there were other studies 5 which identified two other regions which also were 6 capable of transforming, in this case, rat cells -- in 7 both of these cases, rat cells. 8 Now that was somewhat surprising in that 9 there would be more than one region of this genome 10 that would actually transform or immortalize cells 11 but, of course, this is a very large genome anyway. 12 So that that might not be too unreasonable. 13 So we continued with a large series of 14 experiments to try and track down exactly what the 15 region of the genome might be that was producing this 16 transformation effect. Unfortunately, we were able to 17 make smaller and smaller fragments of this MTR1 region 18 until we got down to a region of that DNA which, in 19 fact, was too small to have its own open reading 20 frame, and immediately one has got to start worrying. 21 If there's not an open reading frame, what 22 in fact, is affecting the cell? Is there a protein 23 being produced? In this case, clearly not. So we 24 began to worry what the mechanism might be in this 25 situation. 65 1 So one of the mechanisms we thought about 2 was that there might be just random insertion of 3 sequences into the host cell genome. So we looked at 4 these MTR1 fragments, and particularly the MTR1 5 fragment of herpes simplex virus type 2, and by 6 sequencing that we were able to show there was a 7 structure in there very similar to a bacterial 8 insertion sequence in its structure, and felt that it 9 was highly likely that this was perhaps integrating 10 and then excising randomly from cells, and this might 11 in fact have a mutagenic effect upon the cell. 12 Now interestingly, we were able to take 13 HSV1 and take a similar -- the identical region from 14 the genome of HSV1 and look at the sequence and 15 structure of that. 16 What we found was that that structure, 17 although for half of this loop, is pretty well 18 identical. The other loop did not pair sufficiently 19 to make an insertion sequence-like structure. That 20 region of HSV1 will not transform cells, whereas this 21 one will transform cells at a reasonable efficiency. 22 So this seemed to be a reasonable 23 suggestion, and perhaps we should look at, in that 24 case, mutagenesis by this fragment, the MTR1 fragment. 25 Now while we were doing these experiments, 66 1 we also realized that there was another herpes virus 2 that we should look at, and that was the human 3 cytomegalovirus. The genome is shown up here. 4 Again, the region that we were 5 particularly interested in was this early region, the 6 immediate early region of the genome, which is 7 transcribed very early after infection by the virus. 8 So one of the experiments we did was to try and work 9 out whether we could actually, first of all, transform 10 cells with this region, and the answer was yes, and 11 this was repeated in two or three different labs. 12 Then we wanted to find out, if we used 13 very sensitive PCR experiments, could we in fact 14 identify regions of that CMV immediate early region 15 present in cells that were transformed by CMV. By 16 going through two rounds of PCR to pick this up, we 17 were able to in fact identify a fragment from CMV that 18 persisted in some of the cell lines, but really in a 19 very low percentage. 20 Most of the cell lines, just like the 21 herpes cell lines, had lost the DNA that we 22 transfected into them. 23 Again, when we looked at the transforming 24 region of CMV and made lots of deletion mutants of 25 that region -- so we started off with a region here 67 1 which was quite a reasonable sized piece of DNA, 2.9 2 kilobases, and just made deletions of this all the way 3 down. 4 Again, to our surprise, we could get right 5 down to here, and we were still capable of 6 transforming cells with that region. So again, this 7 suggested to us that maybe we were looking at the same 8 situation with CMV that we had seen with herpes 9 simplex type 1. 10 In fact, in that very small region of CMV 11 DNA, again you find this same structure present, and 12 in most cases that structure was lost from the cells, 13 that it had been put in by transfection, and we 14 actually carried out not only transfection with DNA 15 but made retroviral constructs and put that region in, 16 could again show that we could transform cells with 17 that region, but that it was lost from the cells on 18 passage and in tumors produced by those cells. 19 So the DNA clearly had been present there, 20 but had been lost fairly on in the process, despite 21 the fact that these cells are capable of making 22 tumors. 23 Obviously, one of the ways to look at this 24 is that maybe these viral DNAs are acting in some way 25 as mutagens, and there was certainly a history of the 68 1 fact that herpes virus infection would produce damage 2 to chromosomes. 3 I must qualify this, of course, by saying 4 that in general herpes virus is lytic to cells, and so 5 you don't expect cells to survive. However, in most 6 herpes virus populations there's a lot of defective 7 virus present, which might very well allow some cells 8 to survive in an infection. 9 We could also show experiments done in Zur 10 Hausen's lab in Germany that you could find mutations 11 in host cell genes after transfection of herpes virus 12 DNA-N. So clearly the virus is acting as a mutant, 13 and this is just a demonstration of some thioguanine 14 resistance with a number of these herpes virus 15 transforming regions, showing that there is in fact a 16 mutation frequency that is measurable in these 17 experiments. 18 So this immediate early region in CMV 19 became a region of interest to us, but we really 20 stopped those experiments because they had been 21 started out of an interest in cervical carcinoma, as 22 I initially said. We stopped those, because human 23 papilloma viruses raised their beautiful heads, and 24 obviously we now know that this is the prime 25 initiating cause of anogenital carcinomas. 69 1 So we really stopped looking at this, but 2 about ten years later it gave us a lot of pleasure 3 when Tom Shent suddenly discovered the idea of hit and 4 run, and reported in the literature that they've used 5 immediate early genes of CMV and had essentially come 6 to the same answer, that these regions would produce 7 increased foci, increased transformation of cells, but 8 in fact the DNA was then lost from the cells, and so 9 this again looked like a hit and run phenomenon. 10 I'll just go past this one for the moment. 11 And the type of experiment that came from Tom Shent's 12 lab, which were published in 1997 in PNAS, was that we 13 heard from Alex van der Eb already about the effects 14 of adenovirus E1A, and what they showed was that if 15 you combined E1A with the CMV early regions, E1 or 16 preferable the E1 and E2, you could get a much higher 17 number of foci produced in those transfected cells 18 than if you had just E1A alone, and they showed this 19 in a number of different experiments. 20 So clearly, the CMV region, which was not 21 then retained in these cells, was nevertheless 22 contributing to the transformation of the cells, the 23 same sort of picture that we've seen earlier. 24 The other picture that they also found was 25 that there was an increased mutation frequency, 70 1 looking at APRT in this case and HPRT in this case, 2 that again if you put these immediate early genes in - 3 - and you can do this on their own or in combination 4 with E1A again -- you could demonstrate that there 5 clearly was a mutation frequency resulting from the 6 transfection of this DNA into, in this case, rat cells 7 or mouse cells. 8 So, clearly, there's no question that 9 these small fragments of viral DNA that can transform 10 cells most likely act as mutagens and are lost from 11 the cells upon passage of these cells, but the cells 12 retain their transformed phenotype. That's probably 13 the most important thing to remember out of this. 14 So in conclusion for this talk, I'd just 15 like to make these points: That it's clear that 16 subgenomic fragments of herpes simplex virus and 17 cytomegalovirus can transform cells in vitro. 18 Now these experiment are not conducted in 19 human cells, but in rodent cells, as I've already 20 said. It's clear from a number of experiments that 21 have been carried out now in certainly more than one 22 or two labs that that virus DNA, which is capable of 23 transformation, may not persist over the long term in 24 the transformed cells. 25 Although you can detect early on -- If you 71 1 put a complete open reading frame in, you can detect 2 the expression of viral proteins. You only see these 3 early, and they do not persist, again suggesting that 4 there is not only a loss of that sequence but perhaps 5 a loss of sequence over time out of these transformed 6 cells. 7 Lastly, it's clear that these viral 8 transforming regions can indeed be mutagenic, shown in 9 a number of experiments both by our labs and by Tom 10 Shent's labs. So that still leaves the possibility 11 that herpes viruses can be responsible for some -- the 12 development of some tumors, and there are still 13 experiments being looked at and some that have been 14 published showing that, even in cells that are 15 immortalized by human papilloma virus, the presence of 16 some of these morphological transforming regions 17 results in a more rapid conversion of those cells to 18 tumorigenicity. 19 So hit and run may seem an old phenomenon, 20 but it's not dead. Thank you. 21 (APPLAUSE.) 22 CO-CHAIRMAN RUBIN: Questions? 23 DR. KUNG: Hsing-Jien Kung from UC-Davis. 24 I'm just wondering, do you have some 25 hypothesis whether this is due to insertion of 72 1 mutagenesis or due to the fact it knocked out P53, 2 make the genome just generally unstable; therefore, it 3 takes its own course. 4 DR. McDOUGALL: Well, I don't think P53 is 5 not found in these cells. So I still believe the 6 hypothesis, that what you're seeing is integration of 7 this and then a random excision of it is probably the 8 reason, and that the integration is random, the 9 excision is random, and probably the way to get at 10 this, if we wanted to go back to these experiments, is 11 really to do much more detailed sequencing of these 12 regions and look at the flanking areas on any DNA that 13 might still be there. 14 It is very clear that this DNA is 15 generally lost from these cells. 16 DR. KUNG: But you do have some residual, 17 right, in some of the recent experiments? 18 DR. McDOUGALL: Some of the experiments 19 with CNV there is some material that remains there, 20 and we could go back to that and ask those questions, 21 yes. 22 DR. FRIED: Did you -- 23 CO-CHAIRMAN RUBIN: Identify yourself. 24 DR. FRIED: Fried from ICRF. Do you take 25 the region of that secondary structure that you 73 1 pointed to, and did that stop the mutagenesis? 2 DR. McDOUGALL: No. What we did was just 3 to use the HSV1 region as a control. So we didn't use 4 APH -- We didn't do anything to the HSV2 or CMV 5 regions. We just used the HSV1, which is, as you see, 6 a tail. It does not produce a nice insertion 7 sequence. We used that as our control. 8 DR. FRIED: And in Shent's experiments, 9 were there any specificity to the type of mutations 10 that were formed, based changes or -- 11 DR. McDOUGALL: I confess, I don't 12 remember the answer. 13 CO-CHAIRMAN RUBIN: We go to the back now. 14 Someone has been waiting there. 15 DR. KRAUSE: Phil Krause, FDA. To your 16 knowledge, has anybody ever taken MQR regions and 17 injected them into nude mice or something like that 18 and looked for an in vivo tumorigenicity endpoint? 19 DR. McDOUGALL: Not to my knowledge, no. 20 DR. COFFIN: John Coffin. Stem loop 21 structures such as you drew must, of course, be very 22 common, you know, kind of DNAs, and actually -- I 23 mean, is this effect specific for herpes virus DNAs or 24 have they been studied so much more than almost any 25 other DNA of similar kinds or complexities, if you 74 1 take bacteria for HDNA or -- 2 DR. McDOUGALL: Well, in the bacterial 3 systems, right. I don't know of many studies in this 4 sort of viral system, and I don't know of any that 5 have been done very recently either. 6 DR. COFFIN: So, in fact, it could be that 7 all kinds of DNAs may carry regions -- 8 DR. McDOUGALL: I wouldn't be at all 9 surprised. 10 DR. COFFIN: And it's just the herpes 11 viruses that have been looked at. 12 DR. McDOUGALL: It just happens that that 13 was a good candidate. 14 CO-CHAIRMAN RUBIN: One more question, and 15 we'll have to move on. 16 DR. BROKER: Jim, has anyone taken 17 synthetic stem loop structures or is it specific to 18 the particular sequences that these various herpes 19 viruses that are produced? 20 DR. McDOUGALL: Well, in a way, that's the 21 same as John's question. I think the answer is no. 22 AT least, I don't know of any experiments like that. 23 As I say, we got hopelessly diverted by papilloma 24 virus. 25 CO-CHAIRMAN RUBIN: Could you please give 75 1 us your name? 2 DR. BROKER: Oh, Tom Broker, UAB. A good 3 choice, Jim. 4 CO-CHAIRMAN RUBIN: We'll have to go to 5 the next speaker. Thank you very much. 6 The next speaker is Stephen Baylin, Johns 7 Hopkins University School of Medicine, who will talk 8 about DNA methylation and epigenetic mechanisms of 9 carcinogenesis. 10 DR. BAYLIN: Thank you very much. I'd 11 like to thank the organizers for inviting me. As I 12 hear some of the talks going before, with respect to 13 the area our work is in and our group works on, I 14 think of another kind of change other than the classic 15 genetic changes in DNA that we've been hearing about, 16 which is also heritable and involves DNA probably 17 through chromatin structures which are variable. 18 In terms of contribution to oncogenesis 19 right now, I think one of the interesting aspects of 20 that which has emerged is that process as it regards 21 epigenetic gene silencing during all phases of tumor 22 progression. If I can have the first slide, please. 23 Now for those of you who don't work in 24 this particular area of epigenetics, let me just 25 quickly remind you that in the eukaryotic genome and 76 1 certainly in humans and rodents, essentially the only 2 place that we methylate our DNA is at the base 3 cytosine and only at those cytosines that occur 5 4 prime to a guanosine. 5 We talk about the CPG dinucleotide as the 6 substrate for this event. Now it's an event which is 7 catalyzed by what up until recently was thought to 8 just be one mammalian enzyme, certainly DNA methyl 9 transferase. 10 There are at least two others now that 11 have activity, and this is going to probably emerge as 12 a very important part of the story in tumorigenesis, 13 the interplay and the roles of these different DNA 14 methyl transferases. But they each catalyze a 15 reaction in which acidenosine methionine is used as 16 the methyl donor group, and the methyl group is placed 17 on the 5 position of the ring of cytosine. 18 The other important aspect of this for 19 tumorigenesis which is reduced to real simplicity here 20 in terms of what's being understood about methylation 21 and its role in the genome is that in the eukaryotic 22 genome the CPG dinucleotide has been drastically 23 reduced over evolution to a small fraction of its 24 predicted frequency. 25 So most of the genome has a reduced CPG 77 1 content. However, in most of the genome there is a 2 high percentage of those CPGs that actually get 3 methylated in rodents and humans. So we have heavy 4 methylation, as shown by the yellow dots, in most of 5 the genome. 6 This methylation may play various roles, 7 one of which may be to participate in chromatin that 8 is transcriptionally repressive for repeat elements in 9 areas of unwanted transcription in the bulk of the 10 genome. 11 Now the exception to that type of 12 distribution occurs in and around about half of the 13 genes in our genome where the CPG dinucleotide 14 frequency maintained at its predictive level, and 15 these are the so called by Adrian Byrd and others CPG 16 islands. 17 For most genes -- say, for inactive genes 18 on the X chromosome of the female and selected 19 silenced alleles of imprinted genes, these CPG islands 20 are maintained free of methylation, and this is 21 thought to be a permissive state. It doesn't matter 22 whether the gene is actively transcribed or not, but 23 it's a permissive state for active transcription of 24 the gene. 25 What's being more and more frequently 78 1 recognized in the cancer cell genome is an increasing 2 list of genes in which this balance is disrupted, such 3 that in the face of loss of methylation from large 4 regions of the genome other than the promoter regions, 5 CPG islands around an increasing list of genes 6 actually have become methylated. 7 Whether it's the cause or the result is 8 still being worked out. This correlates with a 9 chromatin organization around that gene which is 10 unfavorable for transcription of the gene. 11 Now the impact for this on carcinogenesis 12 and tumorigenesis, I think, can be looked at from the 13 types of genes where this has now been defined. For 14 those genes which are known tumor suppressor genes by 15 virtue of the fact that when they're mutated in the 16 germ line of families, those families have inherited 17 forms of cancer. 18 About half of those genes, as shown in the 19 yellow, in somatic forms of cancer, noninherited 20 cancer, have had some frequency now shown and well 21 worked out of this promoter region CPG 22 hypermethylation. 23 I would point to you that one of the most 24 frequent times when this is seen is in the cyclin DRB 25 pathway for the P16 gene, and now in a number of 79 1 experimental systems, including mammary epithelial 2 cells, loss of M0 that we just heard about, at least 3 experimentally and probably in several natural tumors, 4 may be in association with hypermethylation of this 5 cyclin dependent kinase inhibitor gene P16. 6 So it can be a very early event in 7 carcinogenesis and a very important one, as shown by 8 just that one gene involvement. Next slide, please. 9 Other genes other than the classic tumor 10 suppressor genes are emerging, and these are just a 11 few, and they play a role in various kinds of 12 processes, some for enzymes that guard against DNA 13 damage such as methyl transferase, 06NGMT GSP pi. 14 Recently, for the P53-like gene where 15 mutations of -- or P73 where mutations have not been 16 found, very frequent to see hypermethylation of this 17 gene in lymphomas. So an increasing list of genes 18 which can play a potentially important read in 19 oncogenesis. 20 Just to remind you, that in many of the 21 instances where this has been defined, or in several, 22 that gene might not have been recognized through 23 classic mutational changes in the coding region for a 24 role in tumors, because several of these genes such as 25 P73, in some tumors even P16 like colon cancer, don't 80 1 seem to be altered through mutational change but have 2 a high frequency of this promoter change. Next slide, 3 please. 4 Just to remind you again how frequent this 5 is, -- this is not to meant to be read -- this is a 6 survey of major human tumors where Jim Harmon in our 7 laboratory and others have been working out these 8 hypermethylation events, and just to show you that 9 virtually every kind of human neoplasm has multiple 10 genes, not every tumor in every patient but, for 11 example, in breast cancer, if you take the list of 12 genes here shown to be hypermethylated and examine a 13 series of breast cancers, you will find that every one 14 of those breast cancers has one or more of these genes 15 hypermethylated. This now includes in about a quarter 16 of them actually in the nonfamilial type of breast 17 cancer the BRCA1 gene. Next slide, please. 18 Now in terms of what this might mean for 19 therapy of cancers, the difference, of course, here 20 from coding region mutations is that theoretically, if 21 there are no base changes that inactivate the gene on 22 a permanent basis, this is a potentially reversible 23 situation, although indeed, if you look at it in 24 culture and you look at it over time, for the most 25 part it is a heritable event, the maintenance of this 81 1 kind of gene silencer. But it certainly has raised the 2 question can reactivation of these hypermethylated 3 genes serve as some sort of therapeutic target for 4 cancer. 5 This is not a new concept. There has been 6 a drug around for a long time -- next slide, please -- 7 Peter Jones and others introduced years ago when they 8 found that drugs and congeners of 5 azacytidine can 9 cause demethylation within the genome and reactive 10 genes. So this has been the drug that people have 11 concentrated on. 12 It's only more recently that, I think, 13 people are thinking about it in terms of specific gene 14 reactivation for events in cancer, like John talked 15 about. Various other series of molecules, antisense 16 again, so called DNA methyl transferase 1 and other 17 molecules directed at the major mammalian DNA methyl 18 transferase are being tried. 19 It must be remembered that these other DNA 20 methyl transferases that have recently been described 21 are probably going to also play important roles in 22 their separate genes with separate sequences, and so 23 that's going to have to be taken into account. Next 24 slide. 25 In our laboratory, it just shows you that 82 1 most of the genes where you see the CPG island 2 methylation in culture, you can achieve at least 3 partial reactivation of any or all of these genes 4 through administration of the drug 5 deoxy 5 azacytidine. Next slide. 6 And you can bring back functional events 7 in the cells by reactivating these genes. The classic 8 example has been the loss of the mismatch repair gene 9 hMLH1, which is probably the leading cause for the 10 microsatellite instability phenotype in colon cancer. 11 Most of those colon cancers have this epigenetic 12 change at the promoter of this gene. 13 If you throw different mismatches at 14 cultured cancer cells -- This is a Hela cell in Tom 15 Kunkle's lab. Actually, this is the 100 percent for 16 each of the three different bases in a Hela cell which 17 has effective repair. 18 Here's a hypermethylated colon cancer cell 19 before azacytidine and, if you give 5 azacytidine to 20 that cell, you can restore considerable mismatch 21 function -- repair function to the cell by 22 reactivating the mLH1 gene. Next slide. 23 In terms of manipulating the genome, I 24 think what's become exciting in this area is to 25 consider holistically the role of DNA methylation in 83 1 chromatin which is transcriptionally repressive, and 2 this has been an exciting arena in the past year or 3 two. 4 Adrian Byrd's laboratory and Wolff's 5 laboratory at the National Institutes of Health have 6 made a big contribution to this recently when they 7 showed that proteins such as mECP2 actually can bind 8 preferentially to methylated cytosines in the genome 9 and recruit proteins in a transcriptionally repressive 10 complex which include a major player, histone de- 11 acetylases, various histone de-acetylases. 12 So what this does is suggest -- and 13 there's been long an argument about the chicken or egg 14 participation of methylation in the science -- that 15 methylation can potentially through recruitment of 16 proteins such as mECP2 actually see the repressive 17 complexes around these methylated sites in the 18 promoters of these genes. Next slide, please. 19 There are drugs that inhibit histone de- 20 acetylase and will reactivate many genes. This slide 21 just shows you that in our hands for the 22 hypermethylated genes in cancer cells and now in 23 multiple others, that if you block histone de- 24 acetylase alone with drugs like trichostatin at any 25 doses -- and here is mLH1 in a colon cancer cell which 84 1 is hypermethylated, not expressed by rtPCR -- you 2 can't reactivate, or we can't, using trichostatin 3 alone. However -- next slide -- if one combines a low 4 amount of a drug like 5 deoxy azacytidine to achieve 5 just a little bit of demethylation -- and here's the 6 hMLH1 again. 7 So here's the azacytidine alone, and now 8 you give trichostatin. Within six hours of doing 9 that, you can effectively reactivate the gene. We've 10 seen at least four or five genes where this is the 11 case. 12 So this is suggesting to us the following 13 type of scenario for work in a working construct, I 14 think, that could prove fertile in this arena for 15 understanding. Just let me show you that this type of 16 activation also brings the protein back to the cells. 17 This is mLH1. Here's an unmethylated 18 colon cancer cell line, the green staining. Here is 19 a line that's methylated, and this is trichostatin 20 alone. Aza will bring back a few cells, and then the 21 combination brings back in this short study a fraction 22 of three or fourfold greater number of cells. Next 23 slide. 24 In thinking about this, again in cartoon 25 form is a working hypothesis, the following type of 85 1 scenario is one that is appealing to us. As many 2 people working in transcription machinery are showing, 3 around the promoter of a gene generally there are a 4 series of activating complexes, shown in the green 5 here, and repressive complexes around the chromatin of 6 this promoter that are simultaneously bound to the 7 gene. 8 The activator complexes, importantly, 9 include coactivators, some of which, like CBP, pcaf, 10 include acetylase activity. The repressor complexes 11 often include corepressors like syn 3 which are 12 tethered to hdac, and the balance of these two, 13 determined by incoming cell signals, would then 14 determine the level of expression of that gene. Next 15 slide. 16 What I think changes when you have the 17 hypermethylation introduced into that promoter is that 18 you change the nature of this repressive complex 19 around the methylated CPG, shown here in the black, so 20 that perhaps at least two things happen. 21 One is that you tightly tether a complex 22 that, through syn3 and probably other transcriptional 23 repressive complexes which will behave in this way, to 24 be shown, you tether hdac and deacetylase inhibition 25 to that promoter region. 86 1 Our studies with the TSA would also 2 suggest it's possible that you block or have less 3 favorable access to a number of these coactivator 4 complexes which have acetylase activity, and that 5 leads to a balance of the acetylase activity and 6 methylation around the gene that results in a rather 7 tight transcriptional repression. 8 What we may be doing when we begin to 9 relieve this methylation with the drug in a little way 10 is to allow return perhaps of these acetylases and 11 coactivator complexes, and now the cell cares if you 12 block the deacetylase, and it contributes to gene 13 reactivation. Next slide. 14 This needs to be worked out by showing who 15 the players and the chromatin setup around these 16 promoters really are at these genes in cells where 17 they are methylated versus nonmethylated, which many 18 groups obviously are now working on. 19 So it has raised a possibility, which 20 actually is getting some trials now in places like our 21 institution and others, that by administration of 22 several days of low dose 5 azacytadine to patients 23 which may avoid some of the toxicity that's been seen 24 with these kinds of compounds, much of which may not 25 be due to actually effects on methylation but actually 87 1 binding of the drug to the DNA, that if you could do 2 this in combination with drugs like phenylbutarate and 3 others that are coming that inhibit histone 4 deacetylase activity, one might get a favorable ratio 5 for reactivating some of these key genes like P16 and 6 others, which might have therapeutic benefit. Next 7 slide, please. That may be my last. 8 I think that is my last slide. Thank you 9 very much. 10 (APPLAUSE.) 11 DR. DOERFLER: I would like to comment on 12 the promoter models. Now to my experience, the 13 situation is slightly more complicated than you have 14 depicted in some of these slides. 15 For instance, the finding that some of the 16 promoters are rich in CPT dinucleotides would be 17 completely unmethylated, I don't think, is so clearly 18 true. We have looked at many promoters, some of which 19 are as you showed and are active or inactive. Others 20 are methylated in spurious nucleotides. However, you 21 find them only if you apply the genomic sequencing 22 technique which detects all methyls in a promoter. 23 Lastly, when one talks about the 24 mechanism, of course, the one you proposed is very 25 plausible. On the other hand, I think there are 88 1 examples where a methide group inhibits the finding of 2 a transcription factor. 3 So there are several ways of how this 4 scheme might work. 5 DR. BAYLIN: Well, I think Dr. Doerfler's 6 comments -- I hope you could hear them -- are exactly 7 right in the time frame. So let me address the last 8 one first. 9 There are certainly genes which do not 10 have CPG islands, most of them, in which occasionally 11 a methylation event right in and around a 12 transcription factor recognition site is important to 13 access of that transcription factor to the gene, and 14 can play a role in expression. No question about 15 that. 16 Probably even within CPG islands, although 17 density seems to be the most important determinant, 18 there are also regions within the island of 19 methylation that are important for where it occurs, 20 but density is extremely important. 21 I think he's again right that, if you do 22 genomic sequencing around these promoters -- every 23 dogma changes a bit -- do you find no methylation in 24 virtually all of these CPG islands? And no, you find 25 methylation. But what is often very much protective 89 1 is a region right in and around the transcription 2 start site. 3 Often the regions of methylation that you 4 do find in normal cells in these islands is distal to 5 that region right around the transcription start site, 6 for the genes that we've sequenced at least, and this 7 is true for Ecad here, and it's true for mLH1. It may 8 be that that's the regions in which their methylation 9 is seeded into these areas during tumor progression 10 that causes the increased density and begins to more 11 and more affect the expression of that gene. But 12 you're right, it is more complex than the dogma would 13 have indicated. 14 CO-CHAIRMAN RUBIN: Thank you. I think 15 we'll have to move to the next speaker, because 16 otherwise we're not going to have a break. 17 The next speaker is Dr. Walter Doerfler of 18 the University of Cologne, who will speak on a new 19 concept in viral oncogenesis. 20 DR. DOERFLER: I'd also like to thank the 21 organizers for inviting me to this very interesting 22 and constructive meeting. Could I have the first 23 slide, please. 24 For actually not 50 years ago, but for 25 more than 30 years we have used the adenovirus model 90 1 to look at problems of viral oncogenesis. Here's a 2 model of one of the adenoviruses. We have been mainly 3 using adenovirus type 12, and we've been following the 4 model established in 1962 by Trenton Yabe and Taylor 5 who showed that, when adenovirus type 12 in small 6 doses is injected into newborn hamsters, actually 7 about 70 percent of the animals that survive the 8 injection develop tumors within 30 to 50 days. 9 That model we have been following most of 10 our studies. The interesting pursuit mainly is that 11 of viral DNA integration and its possible 12 consequences, and that's what I'd like to discuss with 13 you today. 14 So here is my model and some of the 15 concepts which at least for me are new and which we -- 16 I can't say we have proven these concepts, but we'd 17 like to pursue them in future research. 18 So definitely the adenovirus is 19 integrating its DNA, and tumor cells are transformed 20 cells. There's no free viral DNA. The site of 21 insertion is nonspecific. It's different in each 22 tumor, and there are multiple copies inserted. 23 The inserted DNA, much like any foreign 24 DNA -- if you use lambda DNA, if you use plant cells 25 and put transgenes in, the foreign DNA becomes de novo 91 1 methylated, an observation we made way back in '78. 2 Recently, we have been able to show and 3 represent some of the data today that as a consequence 4 probably of this insertion, changes of DNA methylation 5 occur at places distant, remote, from the site of 6 insertion, although the immediate site of insertion 7 can also be affected. 8 Now changes in methylation, as the 9 previous speaker just discussed with you, can lead to 10 changes in transcription, and we'd like to pursue the 11 idea that these changes might play a role in viral 12 oncogenesis. 13 Now from a totally different point of view 14 -- and that project we've been pursuing only for 12 15 years; I may not have time now to present all the data 16 on that. So I've asked for some minutes on Thursday 17 night. A much more frequent occurrence, a much more 18 frequent encounter of an organism, this foreign DNA, 19 is not necessarily the viral infection but it may be 20 uptake of foreign DNA in the foods supply, and we've 21 been studying that to some extent, and I may have a 22 chance to present the data. 23 Now here is again the concept. Viral DNA 24 or other foreign DNA inserts into the host genome. 25 This DNA is de novo methylated. We'd like to think of 92 1 this de novo methylation as an ancient host defense 2 mechanism which defends against the foreign genes. 3 As a consequence of this insertion, 4 changes in the patterns of DNA methylation occur at 5 remote sites in cellular DNA. Perhaps, depending on 6 the site of foreign DNA insertion, this may also lead 7 -- it's not been shown too stringently, at least not 8 in our lab -- in the instances we have looked at, 9 there's only indications, but there is evidence from 10 other laboratories that alterations in methylation 11 accompanied by alterations in chromatin structure, and 12 these numerous changes can lead to alterations in 13 transcription. I will show you some evidence for 14 that. 15 Now let's look at the data. When we 16 infect cells, hela cells, with adenovirus Type 12 or 17 when we -- Maybe we could dim the lights a little bit 18 -- or when we look at transformed cells -- this is a 19 transformed cell line, transformed by adenovirus Type 20 12, and a stretched chromosome preparation -- 21 chromosomes have been subjected to low centrifugal 22 force, and you can see the multiple copies of 23 integrated viral DNA, or you could see them if we 24 could dim the lights a little bit perhaps. Thank you 25 very much. 93 1 A very similar picture emerges when we 2 look at the chromosomes in infected cells. For a long 3 time we're trying to show that also infected cells, 4 productively infected cells at least early after 5 infection do show some association, actually massive 6 association, of viral DNA with the host chromosome and 7 the pictures between infected cell upon stretching the 8 chromosome and the transformed cells that we have 9 definitely shown of this DNA integrated by cloning and 10 sequencing junction fragments do look very similar. 11 So we've pursued the possibility that at 12 least transiently some of the DNA becomes also 13 inserted and productively infect the cells. As long 14 as we have the lights down, this is a blow-up of the 15 transformed cell. 16 You might imagine that there's multiple 17 copies aligned in a string, pearl-like fashion, in the 18 stretched DNA, in this case hamster DNA. 19 Now just a brief summary of some of the 20 observations, summarizing many years of work. At the 21 chromosomal and nucleotide levels there is no 22 specificity of insertion. There may be rare 23 exceptions. The integration may occur preferentially 24 at transcriptionally or otherwise may be replication 25 active sites. 94 1 The consequences are, as I mentioned 2 before, de novo methylation of the integrated foreign 3 DNA, and secondly, changes in patterns of methylation 4 in cellular DNA segments. The integration is in 5 general stable, but we have found revertants which 6 have lost all or most of the viral genome, and are 7 still oncogenic. Actually, we reported that first in 8 1982 in a paper with Ingrid Kuhlman, and in this case 9 hamster tumors were induced by injecting adenovirus 10 Type 12. 11 The integrated state was demonstrated at 12 that time by Southern blotting and also by cloning 13 junction sites. Upon cultivation of these cells, 14 occasional loss of the viral DNA was observed, 15 initially by Southern blotting in the old days, and 16 these Southern blot negative cells, clonal lines, were 17 still oncogenic just as much as the original cells 18 from the tumor. 19 Now when we used more recently, with Anna 20 Pfeffer, PCR analysis, we found very similar to what 21 Jim McDougall was just telling us, tiny bits of the 22 adenovirus Type 12 DNA in several of these lines, none 23 of the lefthand end but only snippets from the 24 righthand end. 25 Each clonal line seemed to have a 95 1 different persistent pattern of these small fragments. 2 When these cells were injected again, they were as 3 oncogenic as the cells that carried the viral DNA. So 4 we had a discussion before on the hit and run 5 mechanism. It's a possibility, but of course, there 6 are other ways of interpreting these data, but it's 7 certainly interesting in this context. 8 I will skip this slide, not to spend too 9 much time, and return to the cell line T637, hamster 10 cell line, and adenovirus Type 12 transformed. It 11 carries about 15 to 20 copies of viral DNA. As you 12 see on this fish preparation, there's a green spot 13 that contains the adeno sequences. The red regions 14 are representing endogenous retroviral genomes. 15 That's about 900 copies per cell, often located on the 16 short arms of the chromosome. 17 Now in the following slide I will use this 18 cell line and others to look at changes in methylation 19 patterns in these regions, which are abundant, and 20 changes might be readily apparent. 21 In fact, when we do such a comparison, 22 work done by Hilda Heller in our laboratory, and 23 compare patterns of methylation in the IAP region of 24 the BHK hamster cell -- that's the parent line for 25 this transformed cell line that carries the adeno 12 - 96 1 - we see both for the hepa-2 and for the hh1 cleavage, 2 both sensitive to methylation in CCCG or GCGC 3 sequences, a very striking increase in DNA methylation 4 as compared to the parent line. 5 This increase persists when the TR3 line 6 is investigated, which has lost all of the adeno 12 7 copies. This is both based on Southern blotting, but 8 there may be still snippets left. We haven't analyzed 9 that extensively yet. But it's important, I think, to 10 our way of arguing that these changes in methylation 11 persist. 12 Hilda Heller has looked at another of 13 other cellular regions, for instance, like the part of 14 the major histocompatibility complex and others in 15 these tumor cell lines and in other cell lines, and 16 has found changes in some, not in others. 17 So certainly, in these transformed tumor 18 cells there are extensive changes in DNA methylation 19 patterns. Now this may not be so surprising, because 20 we are talking about transformed cells with a 21 transformed phenotype. That most likely plays a role 22 in the eliciting of changes in DNA methylation. 23 We have considered other possibilities, in 24 addition, like the function of early viral gene 25 products which are expressed, but when we infect the 97 1 same cells, in the case of adeno 12 which also 2 expresses early viral gene function under these 3 conditions, we do not see these changes. But we have 4 not pursued this possibility, although it's not 5 definitely ruled out. 6 The third possibility we have investigated 7 more extensively, namely, the notion that just the 8 integration of foreign DNA nonviral and nonviral 9 transforming could perhaps also elicit similar 10 changes. 11 So Hilda Heller and Kristina Kaemmer made 12 a number of bhk hamster cell lines that were 13 transgenic, not for adeno DNA but for bacteriophage 14 lambda DNA. In these panels you see -- You may not 15 see it well in the back, but up front you can clearly 16 see that these cell lines are all lambda DNA 17 transgenic, as determined by FISH analysis. This is 18 a control, and again the lambda DNA is located in 19 different positions and different clones, becomes de 20 novo methylated, just like adeno DNA behaves like any 21 foreign DNA in a cell line like bhk21 cells. 22 Now we've now looked at these cells for 23 changes in DNA methylation, and we've concentrated 24 again on a subsegment of the IAP 1 region, which is 25 shown here in full detail. You see all the CG 98 1 dinucleotides. These are the lollipops here. 2 The HP sides are just these two sides. So 3 you see my concern when conclusions are drawn about 4 DNA methylation using just restrictases, because with 5 upper you just catch two out of 35, and with HP you 6 just catch two out of 35, and with Hh, the dots, you 7 catch maybe six out of 35, just a subpopulation of 8 nucleotides -- dinucleotides that can be subject to 9 changes in DNA methylation. 10 So we first looked again at Hh and HP, 11 cleaved DNA from these lambda transgenics. Could I 12 have the next slide, please? Thank you. 13 You see it mainly for the Hh. For the HP 14 it's less striking, although some changes are 15 observed, but for the Hh it's very noticeable that 16 even in sub-clones, different clones of lambda 17 transgenic cell lines show these changes. 18 I should mention that one of these cell 19 lines has also now been subjected by Knut Mueller to 20 an analysis on changes in transcription, by doing a 21 differential hybridization of C-DNAs, and Knut can 22 find quite considerable changes in the transcription 23 level of these -- not only the total level but of 24 cloned segments out of these cell lines. 25 Now to convince ourselves of these 99 1 changes, and particularly the reviewers who were 2 doubtful, we used the bisulfide treatment of the DNA. 3 In brief, the bisulfide converts a cytosine to a 4 uracil, but a methyl C is resistant to this 5 conversion. 6 So then after complete conversion of the 7 DNA, we PCR out the segment we are interested in -- in 8 our case, the IEP1 region -- clone individual PCR 9 products, and sequence up to several hundred of these 10 clones. 11 Whenever we find a C in a known sequence, 12 it's a 5 methyl C after the treatment, and we see a U 13 or a T, it's a C. 14 I just give you a quick summary of the 15 data. What you see here are the positions 1 to 34 in 16 the segment I showed you on the map out of IAP. Each 17 of these numbers represents one CG dinucleotide, and 18 in hundreds of clones we plot here the percentage of 19 clones which are methyl C in a particular location. 20 So let's first look at the bhk 21 cell 21 line, which is sort of the base. This has a certain 22 pattern of methylation, as I showed you before in the 23 Southern blot. Of course, we didn't see any -- All 24 this 34, you just saw the HP and Hh side. 25 Now when we look at P637 cell line where 100 1 we find a blatant, a very obvious increase in DNA 2 methylation by HP and Hh blotting -- cleavage, you 3 find the same effect when we look at all the CGs and 4 the percentage of those CGs that were methylated. 5 Now the lambda lines are somewhere in 6 between. I've shown two examples here, and there are 7 changes in some positions, not in others. When we 8 look at clonal sublines of CPHK21 cell line, we do not 9 see these increases. The possibility existed with 10 respect to methylation IAP in the bhk population per 11 se, but by this and another method we have applied, we 12 did not see that. So we don't pursue that any 13 further. 14 We conclude that after insertion of lambda 15 DNA in these transgenic cell lines, we do see quite 16 distinct changes in levels of methylation, and in one 17 cell line investigated so far we also see changes in 18 transcription, as done by differential hybridization 19 and CDNAs. 20 Now how might this change come about? Of 21 course, I don't know. So I show you a cartoon, and 22 one possibility we consider is that, due to the 23 insertion of foreign DNA as an alteration or you might 24 call it a destabilization of DNA surrounding here, and 25 this may be -- since chromosomes have a unique 101 1 neighborhood in a nucleus, this may be somehow 2 transmitted to other parts of the genome, and the 3 parts affected naturally depend on where the foreign 4 DNA inserts. 5 This adeno is a recent immigrant of 30 6 years ago. The IAPs have been there for 5 million 7 years, and the ancient genes like MHC can also be 8 affected. So this is a way of thinking about it. Of 9 course, no proof whatsoever, but I think the 10 observation is quite clear, and we are now preparing 11 other approaches to make -- obtain additional 12 evidence. 13 Particularly, we are thinking about using 14 perhaps the primary human cell lines, because then we 15 could have much better information about the genes in 16 which the DNA methylation patterns are changing. 17 Now lastly, and this I will just summarize 18 very quickly, the more frequent encounter of an 19 organism with DNA, at least to my judgment, is the 20 daily uptake of food. So we asked the question 21 whether DNA that is orally administered to a mouse -- 22 we used M13 DNA or a green fluorescent protein cloned 23 DNA or, more lately, we used leaf from soybeans and 24 watched for one of the genes out of the light side, 25 the ribulose biphosphate carboxylase, for its 102 1 persistence in the gut and places beyond. 2 I will not have time to do just to the 3 data, which were mainly obtained by Reiner Schubert 4 and Gerte Holbeck. The concept behind this, perhaps 5 at first a strange experiment, is it's foreign DNA 6 integrated into an established genome consistently or 7 very frequently becomes de novo methylated, and I 8 consider this a defense mechanism. 9 If you postulate a defense mechanism, the 10 next question is, well, where is the attack. The 11 attack very frequently, naturally, is in the 12 gastrointestinal tract with a huge surface and a 13 tremendous immune system. 14 So we have asked the question, can 15 intestinal depotea take up the DNA. They seem to be 16 able. We find, first of all, the DNA persisting at 17 certain times after feeding to ease the control of 18 unfed animals, TE, EDTA buffer and then there is 19 observation of M13 DNA one hours to eight hours after 20 feeding, not later on, in the cecum or in the small 21 intestine, the large intestine. At about ten-to- 22 hundredfold lower level we also find persistence in 23 the blood of fragmented forms of DNA. The largest 24 fragments here are about 1700 out of 7,250 of the M13 25 DNA, and about 800 in the blood. So it's not the 103 1 total DNA that's persisting. 2 When we go to sections -- again, you will 3 have difficulty seeing that in the back. If you want 4 to see my printouts, I can show them to you. We have 5 persistence in gut epithelia, in pia patches or in 6 liver cells that are controls, and Reiner Schubert 7 looked at hundreds of sections through the gut and 8 through the liver and the spleen, and in some of these 9 section he found -- by applying the FISH, he found DNA 10 persisting over in the nuclei of these cells. 11 Now FISH is nice, and PCR is nicer, but 12 you always have nightmares, not only when you are in 13 jet lag. You wake up four o'clock in the morning 14 thinking maybe this is the wrong interpretation. 15 So what Reiner Schubert and Doris Renz did 16 was to reclone the DNA from the spleen of animals that 17 have been fed for a week daily, and in 18 hours after 18 the last feeding spleen DNA was prepared and clones 19 were -- DNA was cloned into lambda-2 vector, and 20 lambda plaques were investigated. 21 Amongst several hundred million of 22 plaques, about five to six were found positive for M13 23 and for the mouse DNA. Then what you see here are the 24 coordinates of the M13 phage DNA. I told you, it's 25 one 7,250 nucleotide long molecule, and these are the 104 1 coordinates that we find persisting in individual 2 clones. 3 Two clones were of particular interest, 4 because they had, as the nearest neighbor to the M13 5 DNA, DNA with 70 and 80 percent homology to known 6 mouse genes. So we feel that perhaps the M13 DNA in 7 rare instances has become integrated into mouse DNA in 8 the neighborhood perhaps of pseudogenes. 9 It did not escape our attention that in 10 some clones we found 100 percent homology to known E. 11 coli genes. So if you have some fantasies, you can 12 say, well, perhaps there's a steady stream of foreign 13 DNA from the gut to the spleen, and then the uptake 14 and perhaps also some of it integrated in rare 15 instances in individual cells. 16 I will not go into details now on feeding 17 experiments where we have also looked at pregnant 18 animals and transferred to the next generation, which 19 can occur, and also found DNA in some cells in the 20 fetus and in the newborn, perhaps also in an 21 integrated state, but I will not have time to discuss 22 that. 23 Let me just say in closing, when Gerte 24 Holbeck has investigated animals which have been fed 25 soybean -- this is not naked DNA like we did in these 105 1 series of experiments which are a little bit way out, 2 but we figured first let's try an extreme case. If 3 that DNA persists, then perhaps we can also expect 4 natural feeding experiments to give results. 5 Now when we look at the, as I said before, 6 ribulose biphosphate carboxylase gene after feeding 7 leafs of soybeans, and we can also find this DNA to 8 persist actually in practically unfragmented form in 9 the gut. Of course, we don't know is it still in a 10 cell or has it been liberated. But in rare instances 11 now in three cases, we could also see by PCR some of 12 the DNA which has no homology to mouse, because mouse 13 doesn't have such a gene, has no life cycle in spleen 14 and in liver. 15 So perhaps even by that route, the natural 16 route, if you wish some of the foreign DNA, masses of 17 which we ingest daily can persist in the gut. I have 18 no doubt about that, and can enter probably via 19 columnar epithelia in the gut, pia patches, thoracic 20 duct, white blood cells, spleen and places beyond. 21 Whether it has any effect, any long range 22 effects on the cells in the way I have been 23 speculating in my first slide, of course, remains to 24 be determined, and this will be a much harder task 25 than looking for a tiny needle in a haystack, like in 106 1 a mouse. 2 So in summary, we'll continue to pursue 3 this concept that the insertion of foreign DNA may 4 have much further effects, much more far reaching 5 effects on an established mammalian genome than just 6 a local perturbation, but this local perturbation may 7 be transmitted to other parts of the genome and change 8 transcriptional and certainly methylation patterns. 9 Thank you very much for interest. 10 (APPLAUSE.) 11 CO-CHAIRMAN RUBIN: Any questions? 12 DR. COFFIN: You seem to reject -- Oh, 13 John Coffin, sir. You seem to reject hypotheses that 14 relate to selection, say from a variety of expression 15 of methylation in preexisting cell clones. But at the 16 same time you suggest that methylation is -- methylase 17 is a guardian of the genome against integration 18 expression of foreign genes. 19 If that's the case, could not your data 20 also be explained, at least so far as you presented 21 it, by the idea that preexisting cells with lower 22 amounts of methylation, the rare preexisting cells are 23 much more sensitive to transformation events and, 24 therefore, selected by the transformation experiments 25 that you showed? 107 1 DR. DOERFLER: Thank you very much for 2 this question. Of course, we are still considering 3 this possibility, and it was one of the possibilities 4 mentioned on my slide under number 4. 5 Now I have two lines of thought which are 6 at least unsupported -- can't rule it out. One, when 7 we look at many clones, subclones, of the bhk cell 8 line that we use sort of as a total which could be 9 heterogeneous, and look at -- I think Ralf Remos has 10 looked at 70 or 80 different clones, again by HP and 11 Hh cleavage. Doesn't see any differences. 12 Now that may not be good enough. So we -- 13 In the CGG genomic sequencing analysis in the region 14 that I discussed, IAP-1, Ralf also did the same type 15 of genomic sequencing analysis, and again he found no 16 striking differences -- I mean, there's always some 17 variation, of course -- to the basic total cell line. 18 So although when you have looked at 70 19 clones or at five or ten clones, subclones, this is 20 not very much, when you have to look at millions, it's 21 difficult to rule it out completely. But certainly, 22 the data -- and we're continuing this sort of analysis 23 -- do not support this notion. It's difficult to rule 24 it out completely. 25 CO-CHAIRMAN RUBIN: Are there anymore 108 1 questions or is everybody so eager to get to the 2 break? Okay, thank you very much, Dr. Doerfler. 3 We're going to take a break now, and we'd 4 like to be back by eleven o'clock to start again, for 5 sure. 6 (Whereupon, the foregoing matter went off 7 the record at 10:33 a.m. and went back on the record 8 at 11:01 a.m.) 9 CO-CHAIRMAN MYERS: If everybody could 10 take their seats, we'll reconvene. Probably the most 11 difficult thing that has to occur in a meeting is 12 ending the breaks. 13 We're going to continue our morning 14 session. Our next speaker is Dr. James Cook, who is 15 Professor of Medicine at the University of Illinois, 16 who is going to talk about the role of nonspecific 17 NK/macrophage cell host responses in assessing 18 tumorigenicity. 19 DR. COOK: Well, Dr. Lewis asked me to 20 talk about the terminus of tumorigenicity from the 21 perspective of the recipient of the challenged host, 22 what kinds of things determine the ability of cells to 23 form tumors, what the limitations are of the 24 tumorigenicity assays, and how the immune response may 25 affect our perception of tumorigenicity, depending on 109 1 what's used as the animal. 2 What I'll do is give an overview of the 3 model system we've been studying. The focus in these 4 studies have been on DNA virus oncogenes. Dr. van der 5 Eb gave a nice summary this morning of how SV40-T and 6 E1A of adenovirus immortalize and transform cells, and 7 we're going to be looking at cells that are expressing 8 E1A without worrying so much about whether they're 9 immortalized, because most of these will be super 10 transfected cells that are artificially created to 11 express the adenovirus E1A oncogene, and to ask what 12 effect that has on the tumorigenicity of the cells in 13 various hosts. 14 I'm going to focus on so called innate 15 immune responses or natural killer cells, activated 16 macrophages to a lesser extent, and Dr. Tevethia will 17 talk more about specific immune responses in a 18 following talk. 19 The correlations that we'll discuss are 20 between oncogene induced susceptibility of neoplastic 21 cells to killer cell injury. So this is sort of 22 counterintuitive here in an oncogene that's supposed 23 to be causing a cell to become immortalized and become 24 virulent is actually inducing the susceptibility of 25 the cell to immortalize it to destruction, and then 110 1 the tumorigenicity in various animals with a focus on 2 the value of immunodeficient animals in assessing 3 tumorigenicity, and then later on probably in the 4 discussion session we'll talk about the 5 generalizations of some of these data to broader terms 6 for human cell evaluations. 7 The points that I'll make are that viral 8 oncogenes in certain contexts can sensitize neoplastic 9 cells to the destruction by killer cells, can thus 10 actually reduce or eliminate tumorigenicity if the 11 host is immunocompetent, and that the results of 12 tumorigenicity assays are highly dependent on the 13 immunocompetence of the host. 14 So inducing a tumor is not the same. I 15 think we all know that, but tumorigenicity is a very 16 relative phenomenon and can't be perceived as an 17 absolute, and the level of tumorigenicity among 18 different types of neoplastic cells can be widely 19 variable. So we can't assume that a single dose of a 20 cell can always tell you the same thing. 21 This all started with studies that we were 22 doing looking at an explanation for the reason that 23 adenovirus 2 or 5 transformed hamster cells were 24 unable to form tumors in immunocompetent hamster, 25 although they could form tumors in very young, two- to 111 1 three-day-old hamsters. 2 Through a long series of experiments, the 3 upshot of that is the expression of this E1A 4 immortalizing gene or oncogene of human adenovirus 5 serotypes 2 and 5 appears to sensitize cells to host 6 defenses that are absent in the immuno-immature animal 7 or in animals that lack sufficient NK activity, and 8 this is just a representation of the generality of 9 this kind of observation. 10 If E1A is transfected and caused to be 11 expressed in early passage or low passage NIH 3T3 12 cells in a rat cell line called RN 12 that actually is 13 pre-immortalized with rats in a bhk 21 subclone or in 14 a human fibrosarcoma cell, in every circumstance when 15 these cells are tested for susceptibility to killing 16 by the relevant natural killer cell population, 17 they're much more sensitive. 18 So it's not just the adenovirus 19 transformed cell. It appears to be perhaps many, if 20 not almost any, type of cell in which E1A can be 21 expressed in reasonable concentrations. 22 Now just to tell you a summary of a lot of 23 work about E1A, as I've just shown, E1A expression in 24 cells can sensitize them to natural killers from many 25 species of origin. It also sensitizes to a wide 112 1 variety of injuries. There's nothing unique about 2 this NK cell injury. In other words, it's not a 3 specific killer cell receptor interaction with the 4 cell surface. It's something to do with injury per se 5 of certain types, and those are injuries that tend to 6 stimulate apoptosis or proapoptotic injuries. 7 From an immune perspective, natural killer 8 cells, cytotoxic lymphocytes, activated macrophages or 9 the dominant cytokine that macrophages produce -- that 10 is a cytotoxic molecule tumor necrosis factor alpha -- 11 all can elicit this kind of apoptotic response in 12 cells if they express E1A, and these are cells that 13 are inherently resistant to these types of injuries. 14 It's not unique to immune injuries, and 15 the chemical injuries can do the same thing. A 16 topicide, atopoisummaries 2 inhibitor, beauvericin, 17 a calcium regulator, hydrogen peroxide and hygromycin 18 or protein synthesis inhibitor all can trigger the 19 same kind of response in E1A expressing cells. 20 So there's something inherently weak about 21 E1A expressing cells in an Achilles heel kind of 22 fashion that makes them sensitive to undergoing 23 apoptosis when injured by exogenous agents. 24 Now the immune mediated apoptosis per se 25 is somewhat unusual from other types of E1A 113 1 sensitization in that it is P53 independent. So it 2 works equally well in cells that can't express P53, 3 and it's also resistant to blockade by the E1B 19kD 4 molecule, one of the small E1B proteins that Dr. van 5 der Eb told you about that normally blocks E1A 6 induction of apoptosis during the immoralization 7 phenomenon. 8 Now what I'll do is focus more on the 9 animal model side of the story to show you something 10 about how tumorigenicity of cells that have this 11 phenotype are being sensitized to immune mediated 12 injury can vary, depending on the type of host in 13 which it's tested. 14 So in this model bhk21 hamster cells were 15 subcloned and then transfected in stably expressed E1A 16 or E1A + E1B. It turns that E1B expression is 17 irrelevant for this model. So I won't talk about it 18 further. 19 We tested the E1A expression for its 20 ability to induce sensitivity to killing, as I've 21 shown you, and then asked whether tumorigenicity in 22 different types of animal hosts, depending on their 23 cellular immune responses, varied. 24 For example, in the bhk model we used 25 adult hamsters that have, in fact, NK cell and NT cell 114 1 responses both, nude mice which are unusual in this 2 respect in that they have defective T-cell responses 3 which we all know about, but they also have something 4 wrong with their NK cells when it comes to recognizing 5 E1A oncoprotein expressing hamster or rat cells. 6 So the NK cells are defective in this 7 regard, whereas they will kill other types of NK 8 susceptible targets, and then nude rates which are 9 more like we think of as an aphonic animal. That is, 10 they have defective NK cell responses but have 11 perfectly healthy active NK cell responses -- I'm 12 sorry, they have defective T-cell responses but active 13 NK cell responses when it comes to killing E1A 14 expressing cells. 15 So the question was: If we have E1A 16 expressed in these bhk cells, we use these animals as 17 the recipients, what happens in vitro and what happens 18 in the in vivo tumor induction model? 19 So here are the in vitro data, using 20 hamster spleen natural killer cells as the source. As 21 I've shown you before, they can kill the bhk 22 expressing cells perfectly well, as long as they 23 express E1A. 24 The nude mouse cells don't do very well. 25 They have this defect for killing hamster cells 115 1 expressing E1A, in contrast to the nude rats which 2 kill them quite well unless the nude rates are 3 previously depleted of NK cells with an antibody such 4 as anti-AZ allow GM1. So there is an NK recognition 5 of E1A expressing bhk cells in these different types 6 of animals. 7 Now this is a busier slide looking at the 8 results of tumor induction studies in these different 9 animals, and what we're talking about are the NK 10 resistant bhk 21 parental cells or the E1A expressing 11 clone that was derived from that, that are inherently 12 NK susceptible or resistant, and what happens when 13 different animals are challenged, considering their NK 14 phenotype when it comes to killing the E1A positive 15 cell. 16 So hamster that have competent natural 17 killer cells, that can kill the E1A expressing cells 18 develop tumors with about 1,000 cells, and this has 19 been true for over 30 years of testing of bhk 21 20 cells, if you look back in the old literature. It 21 takes about 1,000 cells to form tumors that are 50 22 percent endpoint in an adult hamster with bhk 21. 23 If you ask what happens when those NK and 24 T-cell competent animals are challenged with the E1A 25 expressing cells, you virtually can't get tumors with 116 1 10 million cells. So there is at least four orders of 2 magnitude change in the ability of these cells to 3 induce tumors in immunocompetent adult hamsters when 4 the cells express E1A and become NK susceptible. 5 If we ask what happens in the nude mice 6 that lack both T-cell and NK cell defenses against the 7 E1A expressing bhk cells, there's not a huge change, 8 although there is some. 9 They go from again about 1,000 cells 10 making a tumor in nude mice, a little less, to about 11 10 to 80 times more cells being able to make tumors in 12 nude mice, but certainly not the magnitude of increase 13 of resistance of the animals against E1A expressing 14 cells, as was seen in the normal hamster. 15 In fact, nude mice have activated 16 macrophages which can kill E1A expressing bhk 21 17 cells, and so it may be that there are other defenses 18 other than their Nk cells that can explain this 19 smaller change in increased resistance to tumor 20 development. 21 Then there are the nude rats that are 22 highly NK competent against E1A expressing bhk cells. 23 It takes about 30,000 cells to make a tumor in a nude 24 rate, and 10 million cells is insufficient to make any 25 tumors in these animals. 117 1 So again, there's a several orders of 2 magnitude change in the resistance of nude rates 3 against the E1A sensitized bhk 21 cells unless the 4 nude rates are NK depleted in advance, in which case - 5 - Now this was just a single challenge with 10 million 6 cells, but in which case the animals become more 7 susceptible to tumor development. 8 So it appears that E1A can reduce the 9 tumorigenicity of these cells if the animals have a 10 competent NK cell response and, obviously, there are 11 other things going on in these animals as well that 12 might lead to changes in tumor susceptibility. 13 Now if we look at a completely different 14 animal model with the same basic scenario -- that is, 15 transfecting and expressing E1A and asking what 16 happens to the susceptibility of the tumor cell in 17 immunocompetent and immunodeficient animals. 18 In this case, these are studies done with 19 Jack Rudis in which the methylcholanthrene cell line 20 MCA-102 was studied. We tested them for 21 susceptibility to killing, and also tested them across 22 the range of animals, adult mice, nude mice, and in 23 this case CD3 epsilon transgenic mice that are 24 defective for both T and NK cell responses. 25 The one difference here is that nude mouse 118 1 NK cells are competent to kill these E1A expressing 2 mouse cells. So we have a range of activities of 3 intact NK and T, defective T but intact NK, and then 4 defective both in these animals that basically lack 5 the function of either of these types of killer cells. 6 This just shows you that the nude mouse in 7 a black 6 background can in fact kill E1A expressing 8 methylcholanthrene sarcoma cells in contract to the 9 E1A negative cells. 10 These are the survival curves of these 11 three different kinds of animals challenged with the 12 E1A expressing cells. So this is the adult C57 black 13 6 mouse that has both the ability to generate a T-cell 14 response and a killer NK cell response. 15 This is tumorigenicity induced by the 16 methylcholanthrene cells. So you can think of this as 17 a survival curve over the 12 week time of tumor 18 challenge, and this is the survival curve of the E1A 19 expressing cells. 20 So there's basically very little 21 detectable tumor induction even at 10 million cells in 22 these animals. So I think of this difference in the 23 50 percent endpoints for tumor production after three 24 months as reflecting the cumulative antineoplastic 25 activities of these mice against the E1A expressing 119 1 sarcoma cells, realizing that some of those are immune 2 mediated and some of those are likely to be other. 3 If you do the same experiment in nude mice 4 and follow them over time, you get a similar type of 5 curve, although you'll notice that the tumor inducing 6 capacity in these animals is lower. 7 It only takes about 100 cells to reach an 8 endpoint in the nude mice, whereas it takes a little 9 over 1,000 cells to reach an endpoint in black 6 mice. 10 So they are inherently more susceptible to the 11 methylcholanthrene cells, in the first place. 12 There still is some resistance that's 13 present in the nude mice against the E1A expressing 14 cells, but the amplitude of this difference at the end 15 of the assay is smaller. 16 Then if you look at the animals that lack 17 any detectable T-cell or NK cell defenses because of 18 the CD3 epsilon transgenic status, basically what you 19 see is that the survival curves are not much 20 different, and the simplest way that I think of these 21 is that these mice don't really have any defenses, at 22 least from the immune response point of view, against 23 E1A expressing methylcholanthrene sarcoma cells. 24 So it probably suggests that these 25 defenses are pretty important. If it were other 120 1 nonimmune defenses, you would think of these mice as 2 being able to mount them. Here the difference in 3 tumorigenicity is not significant. 4 So what I'd like to conclude is that E1A 5 can sensitize neoplastic cells to killer cells and 6 markedly reduce the tumorigenicity of those neoplastic 7 cells, again considering the fact that you test this 8 in some kind of immunocompetent animal. 9 The lack of tumor formation by one cell 10 dose in one host at one time point may not indicate 11 nontumorigenic phenotype. So what I would do is go 12 back to the previous slide and say that, if we picked 13 a dose in here and said that we're just going to 14 challenge with a single dose of maybe a million cells 15 -- So here's a million cells at a given point in time 16 -- if you looked at any point in time across this 17 spectrum in the black 6 mice, you would find no tumor 18 formation by the E1A+ methylcholanthrene sarcoma 19 cells. 20 If you looked at the nude mice, it would 21 kind of depend on when in the follow-up period you 22 picked. If you looked at two weeks, you should see no 23 tumor formation. If you looked at four weeks, you 24 would see tumor formation, and in the CD3 epsilon 25 animals you would see tumor formation at two weeks. 121 1 So it depends on the animal. It depends 2 on the cell dose, and it depends on the time at which 3 you look at these kinds of things as what result you 4 might find. 5 So I think we can't just assume something 6 is not tumorigenic based on a single dose, single 7 time, single host assay, and that the tumorigenicity 8 likely reflects a collection of neoplastic cell 9 traits, not just a single trait in terms of 10 susceptibility to one thing or another. 11 In the context of the cumulative effects 12 of the host defense, probably there are things such as 13 natural killer cells, activated macrophages, host 14 growth factors and a variety of other things that 15 determine the outcomes of these experiments. 16 So I think it's important not to think too 17 simplistically about tumor development in animals, 18 irrespective of the type of neoplastic cell that's 19 being tested. 20 Thank you. 21 (APPLAUSE.) 22 DR. RUSSO: Carlo Russo from Merck. 23 You didn't mention anything about kera -- 24 the inhibitory -- that expressed by NK cells and what 25 role this receptor may have in modulating the NK 122 1 activity, in view of the changes in MHC expression by 2 the tumor. 3 DR. COOK: So the question is: Could 4 oncoproteins alter class 1 expression on the cell 5 surface of target cells? 6 It's known that high levels of class 1 7 expression and inhibitor expression on target cells 8 can actually lead to repressed NK cell activity, in 9 contrast to CTL responses where high levels of class 10 1 expression are necessary. So actually, these are 11 probably complementary host defenses. 12 What I can tell you is that class 1 13 expression level appears to be unrelated to this 14 sensitizing phenomenon in the E1A system. So you can 15 have cells that have normal levels of class 1 compared 16 to normal fibroblasts for levels of class 1 that are 17 very suppressed, and in both cases the cells are 18 highly susceptible to NK killing. 19 So class 1 as an inhibitor or a class 1- 20 like molecules as an inhibitor for NK killing appears 21 not to be a major phenomenon in this system. 22 DR. RUSSO: Do you think that in the 23 experiment you should include the factors that may up- 24 regulate MH expression in order to rule out the 25 possibility of inhibition of NK cells, such as gamma 123 1 interferon? 2 DR. COOK: Well, you could treat -- Well, 3 those experiments have been done. You can treat the 4 cells with interferon gamma. Class 1 expression can 5 go up. It doesn't seem to change the cells' 6 resistance, as long as they express E1A. 7 Certainly, in other models that's not the 8 case. Dr. van der Eb? 9 DR. VAN DER EB: van der Eb, Leiden. 10 If I understand it correctly, you talked 11 about adeno 2 or 5 E1A expressing cells. Right? 12 DR. COOK: Right. Correct. 13 DR. VAN DER EB: Do you have any 14 comparison with adeno 12 transformed cells which are 15 oncogenic? 16 DR. COOK: We haven't done it in these 17 exact models. We've looked at adeno 12 transformed 18 cells from viral transformation events in hamster and 19 in mouse. When you contrast adeno 2 or 5 transformed 20 cells using virus to adeno 12 transformed cells, the 21 adeno 12 transformed cells are much less NK 22 susceptible, despite the fact -- and this kind of goes 23 back to the last question -- despite the fact that 24 they all express low levels of class 1. 25 So again, I think adenovirus violates 124 1 these models of natural killer cell recognition and 2 killing in that adeno 2 and 5 appear to allow class 1 3 to be expressed most of the time, and yet sensitized 4 to natural killing which normally requires low levels 5 of class 1. 6 Adeno 12 represses class 1 and doesn't 7 appear to sensitize to natural killing, despite the 8 fact that the class 1 is reduced, and you would expect 9 that to trigger a natural killer cell response. 10 CO-CHAIRMAN MYERS: One last question. 11 DR. JOLICOEUR: Jolicoeur from Montreal. 12 I was wondering if you have other examples 13 of the oncogene or viral oncogene which behave like 14 that? 15 DR. COOK: I'm sorry? 16 DR. JOLICOEUR: If you have other examples 17 of an oncogene which will behave as E1A here. 18 DR. COOK: A nonviral oncogene? 19 DR. JOLICOEUR: Nonviral or a viral, an 20 autoviral. 21 DR. COOK: Some other example? Well, in 22 mouse systems SV40 T antigen can do something like 23 this, or in rat where it can sensitize cells to 24 killing by activated macrophages, and in some models 25 natural killer cells. 125 1 There are examples of overexpressed MIC 2 that can cause sensitivity to natural killer cell 3 activity. So I don't think E1A is unique in this 4 regard, although it appears to be unique in the fact 5 that it can do it in multiple different species. 6 SV-40, for example, if you compare 7 hamsters and mice, SV-40 transformed mouse cells or 8 sensitive SV-40 transformed hamster cells aren't, even 9 though they express T antigen at equally high levels. 10 So there may be some species specificity 11 to other oncogenes' ability to convey these same kinds 12 of traits to the targets. 13 CO-CHAIRMAN MYERS: Thank you very much. 14 Continuing on in the same theme, our next 15 speaker is Dr. Tevethia from Penn State, who will talk 16 about the role of CTL responses and their implications 17 for tumorigenicity testing and the use of tumor cells 18 as vaccine substrates. 19 DR. TEVETHIA: Thank you very much. If I 20 could have the first slide. 21 Everybody had a favorite virus and a 22 favorite oncogene. Mine is SV40, and the SV40 large 23 T antigen is my favorite oncogene. This has already 24 been introduced. So I can move on to the second 25 slide. 126 1 The point here I want to make -- and I had 2 to go back to some of the work we had done in the 3 Sixties and Seventies to make this slide -- if you 4 transform cells, either hamster cells or mouse cells, 5 BALC mouse cells or C57 black 6 mouse cells, with the 6 SV40 large T antigen, and try to transplant in nude 7 mice, all of these are transplantable in nude mice. 8 But if you try to transplant in the adult animals, in 9 hamsters they are immediately transplantable. In 10 BALC they are rarely transplantable, but only after 11 prolonged custody in vitro and passaging through the 12 immunosuppressed host. 13 In C57 black 6 mice, they are never 14 transplantable. They are only transplantable if you 15 force them to lose MBC Class 21 antigen. 16 There's obviously a very simple conclusion 17 to the role of immune response and the transplantation 18 of these in vitro transformed cells. Let me just give 19 you a clue, that every time you transplant a cell line 20 transformed by oncogene, there the oncogene acts as an 21 antigen, exogenous antigen. 22 It is processed quite differently as 23 compared to the transgenic mouse. If you express the 24 same oncogene T antigen in transgenic mouse, we have 25 60 or 70 examples of this. I picked three of them, 127 1 expressed T antigen and its own promoter produces 2 tumors in choroid plexus. 3 The point I'm trying to make, no matter 4 which promoter you pick, you can induce neoplasia in 5 almost any tissue you like, and these tumors will 6 continue to grow progressively until they kill the 7 host at different times after birth. 8 So the question really that I want to 9 address is why is it hard to transplant the cell line 10 without tumorigenic in nude mice into the adult 11 immunocompetent mice, but the same tumors induced by 12 the -- the tumors induced by the same oncogene in vivo 13 as a transgene continue to grow progressively? 14 I can tell you, the difference is where 15 this oncogene becomes as a self-antigen and that is 16 processed quite differently and the way it interacts 17 with the immune response is quite different. 18 Just to give you a just a little bit of 19 introduction, and the only thing I want to point out, 20 that the purified T antigen can induce the generation 21 of cytotoxic T-cells and can induce in the 22 participation of the tumor rejection. 23 This is just a cartoon slide, just to 24 indicate how the antigen processing works. I'll go 25 very, very briefly. When the antigen comes from 128 1 outside, broken down into peptides, peptides are 2 shunted to the tap into the endoplasmic reticulum 3 where it is assembled with the molecules, and to 4 microglobulin, is then transported back to the cell 5 substrates where it is recognized by the cytotoxic T- 6 cells. 7 Just a cartoon slide to show the 3-D 8 structure of the classical molecule with a peptide 9 bound in the groove. This is the immuno terminal. 10 This is the C terminal. What I'd like to point out, 11 there are two residues in the peptide. They are known 12 as the anchor residues. They are responsible for 13 binding to the MIC molecule in a particular way. 14 So these anchor residues will determine 15 which peptide will bind to which MIC classical 16 molecule. 17 In SV40 large T antigen, there are four 18 epitopes that we have defined, 206-215, 223-31, 404- 19 411, 489-497. They are restricted either by H2 DFP 20 molecule or H21 molecule. We have the cytotoxic T- 21 cell clone for these -- especially for these epitopes, 22 and these are the epitope designation 1, 2, 3, 4 and 23 five. 24 In order to work -- It is a complicated 25 slide. What I'd like to point out is we have cloned, 129 1 expressed these epitopes individually in the common 2 vaccinia viruses in cooperation with John Yewdells and 3 Jack Bennick, and either by themselves or behind the 4 ES signal sequence of the glycoprotein E19. 5 That allows these F clones to be delivered 6 directly into the endoplasmic reticulum much more 7 efficiently. 8 Okay. So what I'd like to describe is the 9 two-transgenic system that inactivates the immune 10 response, cytotoxic T-cell immune response by 11 different mechanisms. 12 One of them is a 501 transgenic mouse 13 system. It's driven -- T antigen driven by the late 14 alpha amylase promoter and appears in salivary glands 15 and osteoplaths. It produces osteosarcoma about one 16 year of age, and metastasized to lung, liver, and T- 17 cells are positively selected. Means that T-cells are 18 present that are expressly for the large T antigen. 19 SV11 tumor system is driven by its own 20 promoter. Tumors appear in choroid plexus. T antigen 21 most importantly appears in the thymus, expressed in 22 thymus, as a result of which there is a negative 23 selection in the T-cells, and these mice essentially 24 can be considered as naive, because they don't select 25 the T-cells. 130 1 So let me just describe the SV11 system. 2 This is a choroid plexus of the 100 day old normal 3 mice, and this is histochemistry for T antigen. 4 Obviously, this is negative. 5 In the SV11 transgenic mice, by 40 days of 6 age you barely see a few T antigen positive cells over 7 here. By 60 days you see a focus formation and 8 corresponding expression of the T antigen in this 9 focus. 10 By 80 days tumor grows, and you can see, 11 it contains -- all cells contains the large T antigen, 12 and by 100 days they are described by Terri Wendyke. 13 There's a big explosion in the tumor. Means that it 14 grows extremely large in a short period of time, and 15 it will die at exactly 104 to 105 days after birth, a 16 very predictable system. 17 Now this is a very, very complicated 18 slide. The message I want to give is this, that if 19 you immunize these mice, SV11s, either with a full T 20 antigen vaccinia virus or with its epitopes, you can 21 see they don't respond. That's all you need to look 22 at. 23 The graph here shows they all respond to 24 vaccinia virus vectors. That means they are not 25 immunosuppressed to general immunosuppression. But 131 1 these are the normal mice, litter mates that don't 2 have the transgene. They respond to all of these 3 epitopes. 4 So what we have done is to -- carried out 5 studies in which we could -- We have taken the 6 transgenic mice. We irradiated them briefly with low 7 dose of the radiation 450r. Then we reconstituted 8 them with the normal T-cells from the normal animals. 9 Then we can immunize them with either the 10 full NT antigen or with immunogenes or we can leave 11 them alone. Then we look for the development of 12 cytotoxic T-cells and look for the progression of the 13 tumors. 14 Okay. If now you irradiate them, transfer 15 the spleen cell, immunize them with a full NT antigen, 16 you have the appearance of the cytotoxic T-cells only 17 to epitope 4. This is the most dominant epitope at 18 this particular point. 19 What is important is that, if you 20 irradiate them first, as I said before, reconstitute 21 them but don't immunize, they still develop high 22 levels of cytotoxic T-cells to the dominant epitope 4. 23 Now this is important, because tumors now after 24 radiation is starting to release the antigen, and this 25 particular T antigen there is the one which has been 132 1 processed and inducing the development cytotoxic T- 2 cells. 3 By irradiation or reconstitution, you can 4 start a program immune response against the tumor 5 cells. This is the tumor looks like. I showed you 6 what the tumor looks like at 40 days of age, very big 7 tumor. In this case, the tumor remains a very small 8 size, hardly recognizable, but still you will see the 9 T antigen positive cells. 10 Now if you keep these animals long enough, 11 until they die, and I'll show you when they die, that 12 they will eventually die of the tumor, the tumor we 13 grow. This is the life span of these animals. You 14 can see the transgenic animals die, which are un- 15 irradiated, by 107 days. 16 Irradiation plus the spleen cell from the 17 transgenic mice again, then they don't live very much 18 longer; but if you irradiate and transfer normal 19 spleen cells and don't immunize, they live over 320 20 days. But they eventually develop tumors and die. 21 The second one that I want to discuss -- 22 So the previous example indicates that the T-cells are 23 completely deleted in the thymus. 24 The second example where the T-cells are 25 positive selected -- this is driven by amylate late 133 1 promoter, and it has all of the epitopes 1, 2, 3, 4 2 and 5, and they come down with osteosarcoma before one 3 year of age. 4 This is what it looks like, the 5 osteosarcoma impinging on the spinal cord. This is 6 the metastasizing to the lungs, and this is the 7 metastasis to the liver. You can see that all of the 8 cells in the tumor have the T antigen by 9 histochemistry. So they have not lost the antigen. 10 What I have done is to just simply 11 summarize the immune responses to it. What's 12 interesting here is that, if you look at the four 13 months of age, then all -- no matter how you immunize, 14 they respond -- almost all -- they respond to almost 15 all of the epitopes except 2 and 3, which is a lethal 16 epitope. They respond to vaccinia virus. 17 If you test the very same animals at 12 18 months of age, then you can see they have lost the 19 response to full NT antigen all to epitope 1, and also 20 epitope 5. They do respond on occasions to epitope 4. 21 So we wanted to know what the mechanism 22 may be, and to -- we want to differentiate between 23 different possibilities, and one of the ones we zeroed 24 in, that these T-cells which are already there are 25 dying by clone deletion. 134 1 The way we have done that is to -- We have 2 made the tetrameres. The tetrameres are made by 3 folding the class 1 molecules with the peptide and the 4 beta 2 microglobulin. Now these soluble class 1 5 molecules have a signal for the biotin -- the 6 biotinylated, and then with strep evident, you can see 7 they will make tetrameres in three of them and can 8 bind to a single T-cell to different receptors. 9 You simply to FAS analysis. What one does 10 is -- this is an example for it, and I'll just 11 illustrate one of them. We have used one of our CTL 12 clone to epitope 1, for example. They have the 13 epitope 1 tetrameres, and with the antibodies, here 14 are the CD 8. With the tetrameres, you can see all of 15 the CTL clones, they line up in the corner over here. 16 It means that -- So you can count -- you can actually 17 count the T-cells that are especially for a particular 18 epitope, and you can see that the -- This is a 19 different tetramer. It does not bind to the same CTL 20 clone. 21 So we have used this approach to identify 22 the mechanism in the 501 mice, why these T-cells are 23 being deleted. If you look at the three-month-old 24 mice 501, only 3.5 percent of the entire CDA positive 25 population is directed to epitope 4. 135 1 You can see in the control mice over 14 to 2 16 percent of the T-cells here are directed to RC 3 epitope 4, once you immunize the animals with epitope 4 4. So they are much reduced in number, to begin with. 5 By the time 11 months comes around, you have only half 6 a percent left of the epitope at specific T-cells. 7 Now if you ask the question, do these 8 animals have a tumor or do not have a tumor, and we 9 assay the animals to determine if they have tumor or 10 not, once they have tumors they have lost all of the 11 T-cells, and you can see, once you have lost it, you 12 cannot even expend them in vitro for lytic activity. 13 What's interesting is that these are not 14 anergic. Even the .5 percent of the T-cells can be 15 expended with full function. 16 Okay. The last thing will take just a 17 minute for me. This is a study, not done in my 18 laboratory, done by the laboratory of Judy Trevethia, 19 and what we were interested in this study was the fact 20 that she made the transgenic mice, which is the black 21 6, that only have 1 through 128 -- actually 1 to 127 22 amino acids of T antigen. It had the RB binding site. 23 It had the intact J domain, and that's really about 24 it. 25 You can see -- What I want to emphasize 136 1 from our point of view here, that all of the epitopes 2 are outside the region of this transgene. That will 3 be, in my opinion, might not be as good a news, 4 because you cannot do anything immunologically, you 5 know, to these animals. 6 So these animals can come down with a SNR 7 tumor. These are the adeno carcinomas. They are 8 driven by the elastase promoter, and you can see this 9 is the adeno carcinoma of the pancreas. You can see 10 a lot of duct formations. You can see this is a 11 normal pancreas with islets intact, metastasizes to 12 the liver. 13 This is again liver metastasis. What is 14 interesting about this is that, if you take the 15 newborn within four hours after birth and check the 16 pancreas, you see a dysplastic pancreas at that 17 particular point within four hours after birth. As 18 the normal pancreas of the four-hour-old, you know, 19 newborn will look like this. 20 So I guess I'll pass this one. So I think 21 my message is the following. I think that 22 transplantability may or may not give you the right 23 answers, because there's a lot of variation, and 24 beside the engine that's present in those tumor cells 25 or transformed cells, then the exogenous engine 137 1 process differently as compared to the transgenic 2 mice. 3 These are some of the people that did the 4 work. These are the people we did work in 5 collaboration, Bennick Yewdell, Barbara Knowles, 6 Arnold Levine of Princeton. Thanks very much. 7 (APPLAUSE.) 8 CO-CHAIRMAN MYERS: We have time for one 9 or two questions. Thank you. 10 Our last paper this morning is by Frank 11 Sistare and was on transgenic animals that might be 12 useful in identifying unsuspected oncogenic factors in 13 tumor cell substrates, from CDER. 14 DR. SISTARE: I want to thank the 15 organizers for this opportunity to speak. May I have 16 the first slide? 17 What I want to do is, first of all, spend 18 some time on how our center -- I'm from the Center for 19 Drug Evaluation and Research -- got involved in sort 20 of changing, as we heard this morning, the gold 21 standard, standard two-year bioassay, for ascertaining 22 the carcinogenic potential of pharmaceuticals. 23 This was an initiative that was an 24 outgrowth of an ICH process that started a few years 25 ago. A guidance was developed, again with the premise 138 1 that there was a general dissatisfaction with the 2 standard two-year bioassay, and are there alternatives 3 for assessing pharmaceutical carcinogenicity. 4 So a consortium was formed as an outgrowth 5 of that by the International Life Sciences Institute. 6 This consortium has over 50 labs worldwide. There are 7 28 pharmaceutical companies. There's members from the 8 United States government regulatory agencies, Japanese 9 and European regulatory agencies as well, that are 10 part of that. 11 Why we're focusing on alternatives -- Like 12 I said, there are often ambiguities and 13 dissatisfaction with the conventional "gold standard" 14 two-year rodent bioassay. Saccharin, as an example, 15 comes to mind as the strain or species specific event 16 which is deemed irrelevant to humans, and there are a 17 host of others. 18 The feeling is that transgenics will carry 19 target genes that are involved in human and rodent 20 cancers. During the prime course of these transgenic 21 assays, there are much lower backgrounds as opposed to 22 the backgrounds associated with some of the two-year 23 bioassays. 24 It's a six-month assay which is a 25 tremendous savings in time in terms of making a 139 1 decision on whether a pharmaceutical is going to be 2 developed or not, and there's tremendous cost savings 3 just in the conduct of the assay. In often cases, the 4 endpoints are very simple. 5 The ILSI initiative has named a number of 6 specific models. As you've heard this morning, there 7 are a host of transgenic models that could be 8 explored, but several were brought into the limelight. 9 There were the initiation promotion 10 models, which is not a transgenic model, which Dr. 11 Pitot reviewed for you this morning. There's a 12 newborn rodent assay, which again is not a transgenic 13 model, but I'm going to spend some time talking about 14 that, because that is one model system that the ICH 15 consortium -- I'm sorry, the ILSI consortium has 16 decided to focus on. 17 Then several transgenic rodent models, 18 though not named by ICH, are emerging from this ILSI 19 initiative: The two knockout models, the P53 knockout 20 mouse heterozygous, and the XPA homozygous knockout 21 mouse, the TJC mouse and the TJ RAS H2 mouse. 22 So what I want to do for you now is 23 describe these test models, give a brief overview of 24 some of the pragmatic protocol design issues that are 25 going to impact on whether or not some of the models 140 1 might be practical for assessment of vaccines. 2 I want to review for you some of the test 3 compound results. Again, these are going to be based 4 on pharmaceuticals, and then give you some mechanistic 5 speculations that are based on some of the data that 6 we're seeing now. 7 First of all, the neonatal mouse model: 8 This model has been around for probably the longest 9 time of the models that the ILSI consortium is being 10 focused on. A number of compounds have been tested. 11 The ILSI consortium has formed a standard 12 protocol for assessing these. So some of these other 13 compounds that have been tested have been used have 14 been used in different schedules and different 15 strains. 16 The protocol is -- One nice virtue of it 17 is the protocol, there's only two dosing periods, one 18 on day eight and one on day 15 of the mouse. One- 19 third of the total dose of the compound is given 20 initially and then two-thirds seven days later. The 21 test agent can be given either IG or PO by gavage. 22 The doses that are decided on are based on 23 the 28 day dose range finding study, and you simply 24 dose them for that period of time and feed them, take 25 care of them for a year, and then one year in necropsy 141 1 and do some histopathology, and then assess the 2 outcome. 3 The general concepts that are forming are 4 that this model will pick up genotoxic carcinogens. 5 Nongenotoxic or indirect active carcinogens appear 6 negative and, specifically, the genotoxic carcinogens 7 that seem to be picking up are mutagens. 8 There's some question about whether it can 9 pick up deletionogens, and some of those questions are 10 being addressed presently. Conceptually, the model is 11 -- It's like an initiation or mutation just prior to 12 periods of normal high DNA replication. So it's like 13 an induced initiation of that early on, and then the 14 normal promotion processes. 15 The P53 alluded to earlier -- mouse model, 16 P53 tumor suppressive gene is the most frequently 17 mutated gene in cancers. The human Li Fraumeni 18 syndrome has a mutated P53. Those patients have 19 greater cancer risk. 20 There's a conservation of P53 across 21 species, and the function of the P53 gene product 22 involved in cell cycle regulation, involved in 23 trafficking toward apoptosis, and also involved in DNA 24 repair, makes it an attractive product in terms of a 25 target for mutation that might knock out a lot of 142 1 these functions. 2 The assay is a six-month assay with 3 continuous dosing. There is -- It's a hemizygous 4 model. So there's one functional allele in all cells. 5 What's been found is that induced tumors, chemically 6 induced tumors -- the tumors are associated with a 7 loss of the wild type P53 allele. 8 One pragmatic concern is that one thing 9 we're learning is that, when there are subcutaneous 10 injections, when that's the route by which the drug 11 may want to be delivered, tumors are sometimes 12 associated with injection site. So the control 13 animals will often show tumors at the injection site. 14 So that can be a problem. 15 The TJC mouse model: The mouse zetaglobin 16 promoter linked to v Harvey RAS gene, and assay 40 17 terminus. This is a micro injected mouse model. 18 There were thought to be three to ten copies in the 19 initial founder mice that were generated here, but it 20 looks like these animals have wanted to go on an 21 amplification of transgene, and in our own we've shown 22 that there's about 40 copies presently now of this 23 transgene. 24 The rationale for the use of v Harvey RAS 25 is that human cancers are associated -- 25 percent of 143 1 human cancers are associated with a mutated RAS gene, 2 and conceptually all cells have this mutation. All 3 cells in this mouse were preinitiated, and what this 4 model is waiting for is a strong promotion event. 5 That becomes the event in the induction of tumors. 6 It's been found that tumors that result 7 are associated with a loss of methylation, and all 8 tumors that have been examined to date for expression 9 do express the RAS -- mutated RAS gene. So it is 10 expression dependent. 11 In terms of practical aspects of its 12 utility, it is primarily a skin paint model. So the 13 test compound is painted onto the skin. Approximately 14 200 microliters is applied to the skin. As a positive 15 control for bolestra, PMA is applied three times a 16 week. 17 Issues surround how the drug is going to 18 be delivered. You have a very aqueous, soluble drug. 19 That could be a problem. It has to be soluble in a 20 solvent that can be delivered to the animal's skin and 21 will evaporate and deliver the dose. 22 There is an alternative route that is 23 being explored. Oral gavage with 24 dimethylvinylchloride has been shown as a positive 25 control to result in four stomach tumors. So again 144 1 it's almost like a local effect, and also some rare 2 leukemic tumors. Leukemias have been seen with, for 3 example, benzine administration. 4 Oddly enough, one example of systemic 5 cyclosporin A given by gavage has resulted in topical 6 skin tumors, but that's the only example that I know 7 of where a systemic drug has resulted in a skin tumor. 8 The TgrasH2 transgenic mouse model: This 9 model is reported to have five to six copies of the 10 human normal ras gene with its own promoter and its 11 own enhancer region. There are no mutations in the 12 coding region of this transgene. 13 What's been found in transgenic mice is 14 that the 221 protein seems to be up regulated. It's 15 about two to threefold higher expression in cells, and 16 that when tumors are looked at -- when one examines 17 tumors from -- chemically induced tumors, that there 18 are a lot of somatic point mutations that are found, 19 found very frequently in these tumors. This is also 20 a six-month assay. 21 Some of the data: This is some of the 22 data that preceded the whole ILSI initiative, and this 23 is the kind of thing we're looking for. Can these 24 models pick up known human carcinogens? When you look 25 at the ones that have been tested, yes, they all can 145 1 pick up the genotoxic in carcinogens. 2 We look, for example, at P53. It doesn't 3 pick up cyclosporin A, Tcdd or diethylstilbestrol; 4 whereas, Tgac and TgrasH2. So Tgac seems to pick up 5 both genotoxic and nongenotoxic. So it picks up the 6 promoters and the complete carcinogens; whereas, P53 7 is limited to the genotoxic. 8 TgrasH2 so far seems to be able to pick up 9 both categories, although you'll see there are some 10 other examples where they can't. 11 If you look only at compounds that are 12 associated with rodent tumors, the same sort of 13 pattern falls out. Tjc can pick up genotoxic and 14 nongenotoxics. For the most part there are a few 15 deficiencies here that you'll see. Ethyl acolate, for 16 example, Tjc did not pick up. 17 P53 again, for the most part, picks up the 18 genotoxic. Glysodol is an example where both P53 and 19 Tjc did not result in a positive finding, but this was 20 multi-site, two-species, both male and female 21 carcinogen, conventionally treated bioassay. The 22 reason for these negative findings are being explored 23 presently in Dr. Tennant's lab. 24 Finally, you want models that do not react 25 to everything that's given to them. So you want to 146 1 check the question of specificity. Do they not react 2 to noncarcinogens? 3 Again, for the most part, for Tgac except 4 for resorcinol here, and rotenone, the specificity is 5 good. P53 specificity is good, and rasH2 specificity 6 is good. But again, there is a lot more work being 7 done presently with these things. 8 The other question is not just things that 9 are four-cell negative, meaning negative in both male 10 and female rats and mice into your bioassay, or things 11 that are positive in male and female rats and mice to 12 your bioassay. What about those things that get us 13 into regulatory dilemmas where it's positive in the 14 mouse but it's not in the rat, or it's positive in the 15 males but not in the females, and what do you do with 16 that kind of data? 17 Some of the compounds that gave us these 18 sort of ambiguous findings where there were some mouse 19 only findings or male only findings in the mouse and 20 rat were put into Tgac and P53. For the most part, 21 these things are not viewed as like "overly" 22 sensitive. 23 There is some level of selectivity, in a 24 sense, that again these things were negative in the 25 rat and positive in the mouse, but they came out 147 1 negative in Tgac. It wasn't sure what was going to 2 happen here. 3 In terms of our charge today is to ask 4 questions about -- You know, when we're talking about 5 vaccines, we ask questions about DNA -- when these 6 transgenic models are exposed to DNA. I don't know 7 the answer to that question, how these things will 8 respond to exogenous DNA, but one abstract that I was 9 able to locate, Gina Clarke at Genentech asked the 10 question about growth factors, mouse abnormal growth 11 factor or recombinant human IGF-1. 12 It was administered subcutaneously for six 13 months to both P53 and Tgac, and it came out negative. 14 Again, like I say, Tgac is primarily a skin model. 15 This was a subcutaneous injection. This was an 16 example where they did see some subcutaneous injection 17 site tumors, but there were no increases in the ones 18 with EGF or IGF-1. 19 The XPA deficient transgenic mouse has not 20 received near as much testing as the other models, but 21 the thinking here is that there is deficient repair 22 capabilities, and it's expected that this model will 23 be able to pick up genotoxic carcinogens. 24 For the most part, the few chemicals that 25 it has been exposed to, it seems to be bearing out. 148 1 A question of whether it can respond to nongenotox is 2 wide open, and what about genotoxes don't rely on -- 3 repair. We don't know the answer to that yet. 4 There are promising prospects sort of off 5 in the wings, if you want to think of it that way. 6 There is a sort of TgrasH2 equivalent rat model. 7 Again, if you want to ask this question about across 8 species, there's another very interesting model, a K6 9 promoter, ontheondecarboxylase transgenic mouse made 10 by Tom O'Brien. 11 This mouse behaves like a pre-promoted 12 model. Apparently, all the polyamines being 13 synthesized by ODC have the skin that occurred to the 14 sites as a state of promotion. What they're waiting 15 for is an initiation of that. They've shown with some 16 chemicals that a preinitiator will activate this 17 model. Again, it's a skin paint model. So it's sort 18 of like the slip side of the coin of the TgAC mouse. 19 There are also some very creative kind of 20 things. I just put an example from Dr. Inoue's group 21 in Japan where they're taking TgrasH2 mouse and then 22 giving it IP urethane, to accelerate the whole thing. 23 So you've got an initiator here. Then you add your 24 test compound, and they can get an answer in three 25 weeks as opposed to waiting six months. 149 1 So there's some efforts like that to 2 accelerate even further from the six month assay on 3 forward. 4 So in conclusion, these short term models 5 to define transgene carcinogens are promising. There 6 are strengths and limitations that are being 7 elucidated, and mechanistic understanding of these 8 models is expanding. 9 The P53 and the neonatal mouse identifies 10 genotoxic carcinogens but really not nongenotoxic 11 carcinogens. TgAC identifies both. TgrasH2 12 identifies genotoxic and the nongenotoxic so far. 13 Looks promising. There's further evaluation going on 14 there, and with XPA we don't have a lot of test 15 results yet to go with that. 16 I'll entertain any questions. Thanks. 17 (APPLAUSE.) 18 CO-CHAIRMAN MYERS: -- cell substrates 19 been tested in any of these models? 20 DR. SISTARE: I am unaware of anything 21 like that having been done. It certainly is something 22 that could be done, and the beauty of these things is 23 it can be done pretty rapidly, because they are six- 24 month assays. The cost is not prohibitive. 25 The animals are $150-$200 a piece. Some 150 1 of these transgenic animals, $150 a piece, but you 2 only need like 15 animals per group. So these things 3 are doable, but to my knowledge they haven't been done 4 yet. 5 DR. ONIONS: David Onions, Glasgow. I 6 enjoyed the talk very much. I just wanted to ask a 7 question in terms of how the endpoint is read. Mike 8 has very kindly allowed me to show some data later, 9 and one of the points that we find is that by looking 10 at the actual kinetics of the tumor development, that 11 actually gives you much finer data than looking at a 12 single time endpoint. But by looking at the kinetics, 13 you can actually discriminate between very low 14 carcinogenic events in multiple transgenic animals. 15 I was wondering how the readouts are done 16 in your particular case. 17 DR. SISTARE: In the TgAC, for example, 18 you can look at both the numbers of animals developing 19 papillomas and also the number of papillomas per 20 animal as a function of time, and that data is 21 analyzed in that way. But, you know, how we are 22 ultimately going to decide whether something develops 23 a tumor faster than another thing, is a more potent 24 carcinogen, that's something that has to be done 25 later. 151 1 With P53, you don't know until you dissect 2 the animal at the end. So there's really not a time 3 component that goes in there, but certainly, they will 4 look at numbers of tumors per animal, number of sites, 5 you know, is it a multi-site or is it a single site, 6 that kind of thing? Is it only the males, only the 7 females? So all these kind of things. 8 DR. ONIONS: I just have one very quick 9 comment about the P53 heterozygotes. We've used a 10 slight different line, which is the Edinburgh knockout 11 mouse, but its function is the same. 12 If you use the heterozygote, what is quite 13 surprising with some carcinogenic insults -- for 14 instance, a retroviral insertion -- you would expect 15 the normal allele to be the primary target of 16 insertion in tumor development, and it's not. 17 It actually is our other genes, but 18 actually universally the heterozygote P53 normal 19 allele is rearranged. So it's, I think, very 20 important to be very careful about assuming that that 21 is always the primary target and that you need to be 22 very careful about that assumption. 23 DR. SISTARE: Yes. No, I agree with that. 24 There has been some data that's shown that there is 25 some rearrangement in that locus, and also that, you 152 1 know, one has to keep in mind that the P53 is sort of 2 -- having only one allele, having sort of a gene 3 dosage effect there, that the loss of that function of 4 that protein is going to have these other cells in a 5 state of readiness for whatever. They're going to be 6 less prone to apoptosis -- less capable of apoptosing; 7 so other injuries, yes. I agree with that. 8 DR. LEWIS: Andrew Lewis, CBER. Has there 9 been any attempt to establish tissue culture cells 10 from these various models and to see whether you can 11 replicate the findings with genotoxic versus 12 nongenotoxic carcinogens in these tissue culture 13 systems? 14 DR. SISTARE: Well, there are P53 15 deficient cells, as we know, which kind of looks a lot 16 like the P53 model, Li Fraumeni cell lines, for 17 example. But from these particular cells, the TgAC, 18 for example -- The thing about TgAC is that it appears 19 that the target cell, at least for skin tumorigenesis, 20 is a follicular cell, a cell at the base of a hair 21 follicle. So it's not just any keratinocyte there. 22 There have been attempts, none of which I 23 know have been successful. Now there have been some 24 tumors -- tumor lines based on papillomas, but those 25 aren't as useful. TgrasH2, none to my knowledge. 153 1 DR. LEWIS: And how frequent are these 2 tumors that appear at the wound site in these P53 3 mice? 4 DR. SISTARE: In P53s with the 5 subcutaneous injection sites, it's not an 6 insignificant frequency. It's something that we've got 7 to reckon with. We've got to work out things like is 8 it pH related, what is it, that kind of thing. 9 With TgAC, if there's a wound, a full 10 length wound, you will get a papilloma every time 11 there's a wound. So there are concerns about 12 fighting, wounding, that kind of thing. You usually 13 have to guard against those kinds of things. 14 CO-CHAIRMAN MYERS: After lunch Dr. Fried 15 and the morning speakers are going to try and put this 16 all -- lead us in a discussion to put this all into 17 perspective, and we'll reconvene at 1:20 promptly. 18 There's one announcement. 19 DR. PEDEN: Yes. For the late breaking 20 session on Thursday night, there's about an hour for 21 a number of presentations. So if you want to present 22 data that will appear in the publication that comes 23 out of this meeting, come and see me, and we'll 24 decide, if we have too many, which gets presented. 25 So I'm Keith Peden, and I'll be around 154 1 here. Thanks. 2 CO-CHAIRMAN MYERS: Thank you. 3 (Whereupon, the foregoing matter went off 4 the record at 12:01 p.m.) 5 - - - 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 155 1 A F T E R N O O N S E S S I O N 2 Time: 1:24 p.m. 3 DR. FRIED: Welcome to the first panel- 4 audience discussion on mechanisms of neoplastic 5 development and neoplastic cell tumorigenicity: 6 implications for cell substrate development. 7 What we're going to have first is two 8 short talks which are relevant to the information we 9 will discuss, and the first one will be given by David 10 Onions, and it's the use of transgenics to detect rare 11 oncogene events. 12 DR. ONIONS: Thank you very much, Mike. 13 It's kind of you. 14 What I'd like to just talk to you about is 15 some work that I've been doing with my colleagues, Jim 16 Neal and June Cameron at Glasgow and relates to using 17 transgenic mice with transgenes targeted to the 18 lymphoid compartment using the CD2 locus control 19 region, and also controlling the functional expression 20 of these genes using a estrogen receptor so you can 21 control the actual target into the nucleus with 22 tamoxifen. So you can either have an inactive or an 23 active transgene. 24 Next slide, please. Now this is a 25 survival curve showing 100 percent survival here and 156 1 survival in days along this axis here. What you're 2 looking at at the green line is the survival of 3 transgenic mice carrying the human CD2 myc transgene. 4 As you can see, what happens is there's a 5 background rate of about 13 percent of lymphomas in 6 these mice over a period of about a year, and that's 7 been replicated by other people. 8 On the second curve here is a quite 9 different mouse. This is the P53 knockout mouse. In 10 this case it's the homozygous knockout mouse. It's 11 the Edinburgh strain of mice, and again there is tumor 12 development, again predominantly lymphomas, over about 13 a year period. In fact, they're all dead by a year 14 from lymphoma. 15 So that's the starting point. But what 16 becomes interesting is when you make genetic crosses 17 of these mice. So if you look at this curve here, 18 this is a cross now between this mouse and this mouse, 19 and you see that there is a very rapid acceleration in 20 tumor development. This shows at least two genes are 21 collaborating in the development of lymphoma. 22 What's more interesting, I think, from the 23 point of view of this discussion here is that you can 24 even push this system further. Notice these mice are 25 all dead by 90 days. But if you superinfect these 157 1 mice with murine leukemia virus, you can actually get 2 insertional mutagenesis that results in further 3 acceleration of tumor development. 4 By cloning the sites where that virus is 5 integrated, you can identify new genes, as we've done 6 -- in this case, new progression genes. But the 7 functionally important point I want to make here, I 8 think, from the issues we've been discussing about the 9 oncogenicity of residual DNA is that you can 10 discriminate even in mice that have this very, very 11 short life span by doing kinetic studies and looking 12 at the pattern of tumor development over time. 13 You can actually pick up very rare 14 oncogenic events or very weak oncogenic events that 15 you would certainly miss, even if you put these 16 insults into mice that carry the myc gene. So I think 17 that's the first point I would like to make. 18 This might be a very useful model system, 19 for instance, for looking at certain insertional 20 vectors like AAV that have got the mutated -- or have 21 got lacorep gene or certain retroviral vectors, as 22 well as perhaps residual DNA. 23 Could I have the next slide, please. The 24 second point I wanted to make is that not all 25 oncogenes collaborate. This just shows genes that we 158 1 have identified as collaborating genes. 2 In this case this is a collaborating we 3 identify with myc which we just published called 4 CBFA1, and CBFA was identified by this retroviral 5 tagging technique. 6 When we make transgenics with that and 7 then superinfect murine leukemia virus, we can ask 8 what genes are activated, and we find, not 9 surprisingly, that myc is activated; because we know 10 that myc interacts with CBFA1. But you can also find 11 that other genes like pim and pal are activated, but 12 they tend to be exclusive. So that we know now that 13 pim and pal lethal form the same complementation 14 group. 15 Could I have the next slide? So in fact, 16 we can begin to bring a pattern of progression 17 starting in this case with myc, and then we can put 18 pim and pal which we now put in the same 19 complementation group, and we've identified further 20 progression genes. 21 So you can begin to build up a pattern of 22 tumor -- of genes necessary for tumor progression. 23 The final point I wanted to make in 24 relevance to the kind of use of these models in a 25 functional sense to screen out of the carcinogens or 159 1 the oncogenic effect of residual DNA is this 2 conjecture slide. This is not data slide. This is a 3 conjecture slide, and it's a simplification. 4 For instance, we heard from Dr. van der Eb 5 today about the complexity of transformation 6 associated with DNA viruses. They cram a lot of 7 functions into a small space. But predominantly E1A 8 has a myc-like function. 9 So that if you were to test the 10 oncogenicity of E1A regions in a myc transgenic model, 11 it's unlikely that you would detect an acceleration of 12 tumor development. 13 On the other hand, if you were to take 14 another mouse like the CBFA mouse, which we know 15 collaborates with myc, you would probably find that 16 this wouldn't collaborate, and you would detect 17 deceleration. 18 So my final point is that when picking 19 these models, it will have to be models plural, 20 because I think you need to use a range of different 21 mice with different oncogenes to cover the 22 complementation groups. 23 (APPLAUSE.) 24 DR. FRIED: Are there any questions or 25 comments? 160 1 (AUDIENCE PARTICIPANT) 2 DR. ONIONS: Yes, we do. I didn't 3 actually show that one, but those were critical 4 experiments for us. If you take -- 5 DR. FRIED; Is your microphone on? 6 DR. ONIONS: Well, the question was, if 7 you take the myc transgenic mice and superinfect those 8 with murine leukemia virus, do you see acceleration. 9 The answer is, yes, you do. We published this in 10 Oncogene a few years ago. 11 In fact, we used those mice to actually 12 identify the new oncogenes like CBFA1 as a new 13 oncogene. 14 AUDIENCE PARTICIPANT: I would like to add 15 a note of complexity to this scheme. I would like to 16 add that not only with a single oncogene you can 17 detect a complementing -- depending on the strain of 18 mouse, the MULV you are using, you will get a 19 different complementation. 20 For example, with myc we have detected 21 over 60 percent of notch activation, while you don't 22 detect that in your system. 23 DR. ONIONS: That's right. That's 24 correct, yes. That's absolutely true. I think that 25 really just highlights one of the issues here, that 161 1 all animal models are extremely complex. 2 In instance, the LTRs and the retrovirus 3 used will bias which particular genes are activated 4 and so on. So that's absolutely correct. Again, I 5 think, reinforces that, if you are to use these model 6 systems, that you need to use a variety of model 7 systems and not rely on one. 8 DR. FRIED: Does E1A and myc collaborate 9 in vitro? 10 DR. ONIONS: Not as far as I'm aware. I 11 think that goes back to early experiments of Helmut 12 Lang who did these experiments and showed that 13 predominantly E1A had a myc-like function. So that it 14 didn't seem to collaborate. 15 Of course, if you put in E1B, then you've 16 got other functions there. 17 DR. FRIED: Thank you very much. Is Jim 18 Cook here? Okay, I think we'll go on. Jim was going 19 to give us a short talk about the generalities about 20 tumor genecosities, but I'm sure we could take that 21 later if he shows up. 22 From the talks we have heard, we know that 23 to generate a fully neoplastic or tumor state is not 24 simple, and it requires multiple events and multiple 25 stages. You know, we need sort of activation of 162 1 mitogenic signals. This is usually through oncogenes 2 and the oncogenes -- we might need more than one. We 3 might need cooperation. 4 We also probably need inactivation of 5 tumor suppressor genes, at least in the P53 and the RB 6 pathways. We have to overcome senescence. We have to 7 make sure, if the tumor is going to grow in the 8 animal, that the division is greater than the death 9 rate of the apoptosis. 10 This probably changes at the cell surface. 11 You need vascularization for the tumor and, of course, 12 we've heard a lot about overcoming immune 13 surveillance. 14 It's quite clear that the formation of 15 transformed cells or immortalized cells in vitro are 16 not necessarily the same as forming a tumor in vivo. 17 So the first panel-audience discussion 18 should probably highlight features that the regulatory 19 agencies should consider when assessing neoplastic 20 cell substrates for vaccine development. 21 I would ask the audience to actually 22 participate as much as possible, because I'm sure that 23 everybody here could answer all the questions that you 24 may have questions. So it will be a blend of both the 25 panel and the audience. 163 1 Also, if you want to ask people questions 2 that you've thought of later after they gave their 3 talk, this is the time. 4 So I think one of the first questions we 5 should ask is should we use neoplastic cells for 6 vaccine substrates, and what are the positive features 7 of neoplastic cells? Does anybody want to start on 8 that? 9 I mean, we have heard that we can grow 10 them to large numbers, and we can get their 11 reproducibility, always having the same one, freezing 12 them down. We certainly might need certain types of 13 neoplastic cells for virus growth. For high titers 14 we might need to get specific cell types. 15 Also, more recently, growth of disabled 16 virus has been done in sort of P93 cells, which Alex 17 van der Eb highlighted in his talk, which actually 18 contain some adenovirus genes, and they're sort of, I 19 guess you would say, immortal but not tumor producing. 20 They don't produce tumors at all, Alex? 21 DR. VAN DER EB: They produce tumors now, 22 I think, but initially -- That has been tested many 23 years ago, I think. They were not tumorigenic in the 24 nude mouse system, but they became it later. 25 DR. FRIED: But when you say they produced 164 1 tumors, was this on passage or do you need a lot of 2 cells? 3 DR. VAN DER EB: I think it was on 4 passage. This happened in 1974-75 or something like 5 that, and so it's a long time ago, and there is not 6 much written down, not by me. 7 DR. FRIED: Because that's mentioned in 8 the document, the CBER document, but this type of cell 9 is really used, and people are injecting things, 10 because most of the adenovectors which people are 11 using for gene therapy are grown in 293 or equivalent 12 cells, because they complement the E1A gene and allow 13 you to get disabled virus, which then is produced and 14 will not replicate. 15 On the other hand, we know Onyx is using 16 a virus for sort of gene therapy to selectively kill 17 tumor cells which do not have P53, and they also grow 18 their virus in those cells because they have E1B, and 19 that's a complementation. 20 So could you think of any other cases 21 where we would need neoplastic cells? 22 DR. PITOT: Well, being very naive in 23 this, I just have one question. Since I think already 24 on the market or at least in some trials a product of 25 neoplastic cells that is basically a product of 165 1 myeloma cells have been used already in the treatment 2 of neoplasia. So it seems that question already, in 3 a sense, has been answered in that it's been done. 4 That's quite different from the sort of 5 things you've been talking about up here. But perhaps 6 if one is talking about a protein that's going to be 7 produced, I personally wouldn't see any problem with 8 the neoplastic. 9 So I think where it may get into a 10 difficulty is in the DNA vaccine production. 11 DR. FRIED: Right. Well, in the case of 12 the hybridomas you're purifying a protein. In the 13 case of growing the viruses that you want live, you're 14 actually just -- You don't purify it to the same 15 degree as you would purify the antibody molecules, and 16 then you would inject in virus; because you want it to 17 get into -- infect a lot of cells. 18 So it's slightly different, I think the 19 less critical. 20 DR. SHEETS: Might I respond to your 21 question about other reasons we might want to use 22 neoplastic cells? 23 My name is Becky Sheets, and I work in 24 vaccines at CBER. 25 A couple of other things that we have 166 1 considered in terms of why we are thinking about the 2 acceptability of these sorts of cells is (1) you 3 mentioned adenovirus as a recombinant vector, but also 4 traditional HIV vaccine -- traditional approaches to 5 HIV vaccines have not been tried in preventive 6 settings, and by traditional approach I mean live 7 attenuated and inactivated. Part of that is because 8 HIV only really grows well in human tumors, in T-cell 9 lines. 10 Another basis for using these sorts of 11 cell substrates for vaccine production is because they 12 can be grown in -- They can be adapted to growth in 13 serum free medium. Therefore, you're taking away the 14 risk of an animal derived substance that you would be 15 adding extraneously to the vaccine. So you're 16 removing some of the risk for BSE or for some of the 17 other bovine viruses. 18 So there are additional reasons for 19 considering neoplastic cells for vaccine production. 20 DR. FRIED: Right. Also, I guess, people 21 are going to use, say, adenovirus vectors to bring in 22 other antigens, maybe HIV-type antigens. 23 DR. SHEETS: We have been approached about 24 that. Some of the adenovirus vectors are replication 25 competent and can be grown in other things, but those 167 1 that are replication defective would need to be grown 2 in a cell line such as 293. Correct. 3 DR. FRIED: I think that's a strong point. 4 Also, I guess certain viruses that one would want, 5 higher titers might be achieved in neoplastic cells 6 than some of the cells that they're growing; because 7 you can get them from different tissues where might 8 need tissue specificity for the virus to really grow 9 well. 10 Okay. So that's some of the positive 11 features. Then the question is what are some of the 12 negative features for the substrate? 13 I guess the ones are that they contain 14 activated oncogenes, and the question is whether these 15 oncogenes could somehow make it over to transfer 16 through the vaccination procedure. 17 Has anybody any feelings on that? That 18 would be -- The other thing is, if there were cell 19 contamination, would cells of a neoplastic cell be 20 transferred, and could that cause a tumor? I guess 21 one could very carefully get rid of most of the cells 22 in the preparation, but how many cells do we need for 23 tumors when we transfer it into the vaccination? 24 AUDIENCE PARTICIPANT: I have to express 25 some bewilderment about some of these issues, simply 168 1 because, as I mentioned last night, the vero cell, 2 which under many conditions is neoplastic, has been 3 licensed for the production of IPV and OPV in the 4 United States, Thailand, Belgium and France, and these 5 issues apparently have been resolved, at least to the 6 extent that they apply to vero cells. 7 Is there something that I'm missing that 8 applies to cells other than vero that are immortal in 9 neoplastic? 10 DR. FRIED: Well, I think in the document 11 it was considered that -- Bob? 12 AUDIENCE PARTICIPANT: Just to clarify, in 13 the United States at least vero cells have not been 14 licensed for an unpurified vaccine, for instance like 15 OPV, to my knowledge. IPV can be produced in vero 16 cells. 17 Of course, vero cells being used for OPV 18 may be a special case, because whatever you're giving 19 is being presented orally rather than parentally. So 20 I think that these still are issues, even with vero 21 cells, that the world is struggling with. 22 AUDIENCE PARTICIPANT: Well, we have 23 enough representatives here from FDA to tell us 24 whether IPV and OPV are licensed in this country, 25 produced in vero cells. Does anyone have a definitive 169 1 answer? OPV is not? Yours is definitive? 2 Okay. So IPV is, which implies strongly 3 that vaccines that are administered orally and 4 produced in continuously propagable abnormal 5 neoplastic heteroploid cell populations should be 6 acceptable. 7 AUDIENCE PARTICIPANT: I would like to 8 add that OPV prepared in vero cells is also pre-trial. 9 AUDIENCE PARTICIPANT: I was going back to 10 try and look at the basis of our understanding of 11 whether DNA -- residual DNA is a risk. I thought 12 there were some interesting talks today that perhaps 13 highlighted some of the issues. 14 One of the issues that wasn't stressed, 15 though, was in fact the nature of the DNA that we see 16 in Crucip and Ames, which are the primary danger, it 17 seems to me, because blood products are purified. 18 Then most of the DNA is pretty badly 19 sheared, and it's sheared below a size that would be 20 normally associated with a functional gene. Now 21 that's an average size, and we have to be careful 22 because there's clearly a distribution in size. 23 It seems to me that you're left with two 24 primary worries -- well, three worries perhaps. One 25 is of promoter insertion, because there's still some 170 1 of these size of around about a promoter. 2 The second level of concern would be, I 3 think, the still intriguing data on the MTR-like 4 sequences in herpes viruses and whether there are 5 other insertion elements like that, and they truly 6 operate. 7 I suppose the third concern I would have 8 really relates to adventitious agents we don't know 9 about, like say a good example would be bovine polyoma 10 virus which we didn't really know a lot about until 11 recently. So you have an unknown polyoma virus in a 12 cell line, and as we know, as we've heard today, you 13 can get a lot of transforming functions in a small 14 domain. 15 So I think those are the three areas which 16 seem to me to be perhaps residual concern in residual 17 DNA. 18 My only comment would be that you can talk 19 about it a lot, but the only way of actually testing 20 that, apart from large scale epidemiological studies, 21 are the kinds of models that, I think, that Frank 22 Sistare and myself mentioned. 23 DR. FRIED: But, David, I mean, none of 24 those are really neoplastic cell specific. 25 AUDIENCE PARTICIPANT: No, they're not. 171 1 DR. FRIED: They're the same for both. I 2 mean it's just DNA. I was thinking more of, you know, 3 if you start with a cell that has an activated 4 oncogene or something, could that get over. 5 I think, since we need multiple genes for 6 neoplasia, the chance of getting two or three cellular 7 genes to meet and get in is very rare. 8 AUDIENCE PARTICIPANT: I think I was 9 working around the conservative approach to exactly 10 that same conclusion. In other words, I think you can 11 enumerate a number of issues, but those seem to me so 12 remote at the moment that I would not, without further 13 experimentation -- I would go in your direction, yes. 14 DR. FRIED: Right. But I think one should 15 consider the viruses. Now the viruses put -- They get 16 a lot for their money, because they have either 17 overlapping reading frames, so one piece of DNA, in 18 the case of polyoma viruses, three genes and SV40 is 19 two genes, or they have common things that are hooked 20 up to each other, are very closely linked. 21 So we don't need different parts of the 22 genome together. I think that's something one might 23 think about. 24 DR. BERKOWER: I'm from CBER. Ira 25 Berkower is my name. 172 1 I'd like to make two points and ask for 2 comments on these points. The first one is: I see 3 vaccines, particularly anything that is an assembled 4 antigen, as being entirely different from other 5 products that might -- biomedical products that might 6 be made by recombinant DNA methods and expressed in 7 even in the same cells. 8 For example, monoclonal antibodies made in 9 myeloma cells -- myeloma cells are -- In hybridoma 10 cells myeloma cells are highly malignant, but the 11 antibodies can be highly purified. Steps like protein 12 A affinity columns and various other chromatographic 13 steps can highly purify away adventitious agents, 14 harmful DNA sequences and so on, and really should be 15 considered, in my view, completely different from 16 vaccines. 17 Vaccine purification, particularly if it's 18 a particulate type of antigen, would tend to select 19 for things that are the size of viruses. They would 20 tend not to be inactivated by any harsh chemicals or 21 denaturing agents. They don't -- They may not have a 22 specific affinity step. 23 So -- and even some of the purification 24 steps might tend to concentrate viruses into your 25 fractions rather than remove them. 173 1 So I think it's a mistake to reason from 2 other products that might be made in tumor lines to 3 vaccines. Vaccines, I think, are quite distinct in 4 that regard. In many cases, they either can't be 5 purified away from harmful agents or they can't be 6 inactivated to the same extent as would ordinarily be 7 done for other products produced in the same cells. 8 So arguing that the same cells are used 9 would not apply to vaccines. That's my first point. 10 DR. FRIED: Well, but on that point, the 11 differences between neoplastic and immortalized cells 12 -- I mean diploid cells -- there's not much difference 13 there. 14 DR. BERKOWER: That's true, but I think 15 one of the arguments that these things are safe is 16 that they've been used to make other products. That 17 argument, I think, just doesn't hold water. 18 DR. FRIED: And your second point? 19 DR. BERKOWER: The second point was: 20 Seems to me that, something I asked about before, that 21 there ought to be a notion of sort of the malignant 22 potential of some of these lines that are being 23 considered, that some lines are what I would call 24 highly malignant and probably shouldn't be used to 25 produce vaccines, even though viruses can grow in 174 1 those cells. 2 The cell itself would intrinsically have 3 a higher risk, whereas other cells would be much 4 closer to normal somehow with a lower, what I would 5 call, malignant potential, and might -- Perhaps we 6 could identify these cells and, because there are 7 continuous lines, characterize them as well as 8 possible and actually show somehow that they were less 9 malignant and, therefore, a more suitable substrate 10 for producing vaccines, say by virus growing on those 11 cells. 12 I could imagine, for example, classifying 13 cells based on some kind of oncogene inventory, and I 14 wonder if that would be possible. 15 DR. FRIED: Does the panel have any 16 comments on that? 17 DR. COOK: I think that gets back to the 18 same point Mike Fried just made, which is that until 19 you have some way to assess the relevance of what 20 highly malignant is and whether the things that convey 21 those traits to the neoplastic cells that you're using 22 for the vaccine substrate -- whether those genes could 23 be reasonably expected to be transferred across in 24 some kind of residual DNA. 25 It's sort of a nightmare and a worry, but 175 1 there's not much science to it. I mean, the question 2 is how can you say that a cell that varies by as much 3 as a millionfold compared to another cell is any 4 greater risk, other than the fact that it feels bad? 5 If nothing is reasonably expected to be 6 transferred and you have four or five oncogenes in 7 that cell, what's the likelihood that those genes are 8 going to be transferred across? I mean, think of a 9 simple experiment like a cell transformed by a pair of 10 oncogenes, that you're required to get that cell to be 11 reasonably immortalized and transformed in a 12 conventional in vitro assay. 13 Shear up that DNA and try to transfect it 14 into another cell, and see if that DNA itself rather 15 than the purified genes like we all do in plasma 16 preparations could actually create a focus in anything 17 you could imagine. I think the answer is probably 18 not. It's unlikely that that would be an issue. 19 Now how you test that is another question, 20 and even more fundamental to your issue is what's 21 highly oncogenic mean? It's relative to how you 22 assess the cells, how you quantitate those sorts of 23 things, which cells you test, what kind of host you 24 put them in, and you could probably define it to meet 25 whatever criteria you wanted to. 176 1 I think it's a real dilemma and it 2 probably requires some experiments to come up with how 3 you're going to define these issues. What do the 4 terms mean, and how are you going to use these to set 5 the guidelines that people would have to do this awful 6 job will do? 7 DR. BERKOWER: I mean, one example, which 8 is not directly related -- It seemed obvious to me 9 that, if you were going to use, for example, EBV 10 transformed B cells as a substrate, at least pick the 11 nonproducer lines. Okay? That's obvious -- to grow 12 the virus in. 13 DR. COOK: I think that's a different 14 issue, and that is you're at risk of transmitting an 15 infectious agent that can amplify itself. That's 16 different than saying you have a neoplastic cell free 17 of all known adventitious agents that contains either 18 a known or an unknown gene that regulates the behavior 19 of that cell. 20 I think transmitting infectious agents 21 should be at a different level of concern, because 22 then if that could survive the purification process, 23 that's a big problem. 24 DR. FRIED; I mean, if it's a really 25 highly tumorigenic cell, and we're not quite sure what 177 1 that means, then I think one has to be careful in 2 whether cells are contaminating the vaccine. If it's 3 four or five or six oncogenes together that are making 4 it so powerful, the chance of getting them all over in 5 the vaccine is very remote, because you need six 6 pieces. 7 AUDIENCE PARTICIPANT: I find myself in 8 the reluctant position of trying to argue against what 9 you just said, because I actually don't believe that 10 oncogenic -- DNA from the lines we've been talking 11 about is a risk. But I think I'm not sure that I 12 agree with the reasoning that you've just adduced. 13 I do think there's a difference between 14 cell lines. I think the very fact that we use vero 15 cells usually at a passage level below the level at 16 which they're tumorigenic in animal models is a 17 testimony to that. 18 I think there clearly is a difference. We 19 know that, as we pass these cells, that different 20 genetic changes are occurring, and we now know that -- 21 it's a kind of paradigm of tumor biology that multiple 22 genetic changes are required. 23 So if you're going to pick a line, why 24 don't you pick one earlier on where there's likely 25 only to be one or two changes and not one where 178 1 there's seven? 2 I think your argument, Mike, was that it's 3 so unlikely to transfer four oncogenes into a single 4 cell, but surely all our transfection data in gene 5 therapy suggests entirely the opposite. 6 If you make a cocktail of a cytokine gene 7 and a couple of other genes and inject them in, you 8 can find generally the cells that take that out have 9 all of those genes in. It's rather like a cell 10 transfection. 11 DR. FRIED: But you're putting in the 12 purified genes, not the whole cell. 13 AUDIENCE PARTICIPANT: Oh, sure. I accept 14 that. That's why I still think that this is a highly, 15 highly improbable effect. 16 DR. FRIED: I was only commenting to him, 17 saying that there may be multiple genes, you know. If 18 there are four or five, it's going to be worse. 19 AUDIENCE PARTICIPANT: But this is -- I 20 would like to say I very much agree with you, that the 21 idea is to take a cell that is barely immortalized 22 rather than the cell that is soundly full of all kinds 23 of mutations and could potentially transfer more than 24 one at a time, like you just said. 25 DR. FRIED: I don't think anybody is 179 1 disagreeing with you, that you take it earliest and 2 safest. I mean, the question is what are the risks as 3 you passage it later? 4 AUDIENCE PARTICIPANT: Well, okay. To sum 5 up my question then, would there be some convenient 6 quantitative way to assess, oh, this one is earlier, 7 that one is later? 8 DR. FRIED: Does the panel have any 9 comments? Anyone? 10 DR. COOK: So you're talking about two 11 different cell lines. 12 AUDIENCE PARTICIPANT: Yes. 13 DR. COOK: Trying to assess the relative-- 14 AUDIENCE PARTICIPANT: Which one would be 15 better for vaccine -- 16 DR. COOK: -- that cell line? 17 AUDIENCE PARTICIPANT: -- for use as a 18 cell substrate, as we're trying to decide today. 19 DR. COOK: Well, you can create artificial 20 models in which you can do that, like the animal 21 models I described this morning, and at some point I 22 guess I can present these data. But you could say, 23 okay, we're going to decide that host A is the host in 24 which this is going to be defined. 25 Somebody is going to have to make that 180 1 decision, because you can always come up with a weaker 2 host, what I call a walking test tube. You know, 3 you've got an animal, but it lacks any kind of defense 4 against a tumor, and you're going to be able to change 5 the tumorigenicity of that cell, depending on what you 6 test it in. 7 So somebody is going to have to decide, 8 well, what is the animal in which we test this. The 9 data I showed this morning was that, if you have a 10 cell line that can be easily rejected by a normal 11 mouse or a nude mouse at a given concentration, it can 12 be highly tumorigenic in a CD3 epsilon transgenic 13 mouse. 14 So you've got to define the animal in 15 which this is going to be done, and then you can say 16 how can we test the cells. Well, there are 17 quantitative ways to compare cell line A with cell 18 line B, once you settle on that in vivo model. 19 Then you have to accept the fact that 20 you're making an arbitrary decision about, okay, if 21 this one has a millionfold greater ability to make 22 tumors at this inoculum over this period of time or 23 over a range of inocula, then does that meet the 24 definition of a worse cell line versus a better cell 25 line. 181 1 AUDIENCE PARTICIPANT: Yes, but I'd prefer 2 a molecular diagnosis, if you see what I mean. 3 DR. COOK: A molecular probe that will 4 tell you what's the worst cell? 5 AUDIENCE PARTICIPANT: No. Multiple 6 probes to try and define, you know, which ones have 7 accumulated more activated oncogenes and which ones 8 are running with fewer or almost none. 9 AUDIENCE PARTICIPANT: To respond to this, 10 I would suggest that there's at least the possibility 11 that focus on the malignancy of the cell is the wrong 12 problem to be concerned with, that what we're really 13 interested in is the ability of the information in 14 that cell to be transferred, not the state of the 15 information in the cell. 16 For example, primary tissues from 17 monkeys, for example, can have pathogenic organisms in 18 there that have nothing to do with transformation and 19 would not score in a test for neoplasia, but would be 20 certainly something we would want to avoid. 21 I mean, the assumption in the question is 22 that there's a correlation between the presence of 23 information of a neoplastic nature and its ability to 24 be transferred from one cell to another. And it 25 doesn't seem to me that we have any evidence to 182 1 suggest that that correlation holds. 2 Now, certainly, if you have a scenario in 3 which you have two cells that are completely identical 4 in terms of their ability to donate information, you 5 want the one with the less damage or accumulated 6 changes in it. But that's an assumption that I don't 7 think there's any basis to make. 8 In fact, I would question whether there's 9 any reason to believe that there's any correlation at 10 all between neoplastic potential and ability to donate 11 information in a vaccine problem. 12 DR. FRIED: That was one of the topics I 13 wanted to bring up, is whether the mechanism by which 14 the cells become neoplastic could contribute to their 15 relative risk as a vaccine substrate. 16 You know, certainly, if you knock out 17 tumor suppressive genes, that's a loss of a gene. 18 That might, you know, make the cell more tumorigenic, 19 but you're not going to really transfer that unless 20 you're transferring a dominant negative P53. If it's 21 just mutated and gone, it's gone, and it's not a 22 positive situation. 23 I don't know about methylation, how people 24 feel, whether that is something that's going to change 25 that. Walter? 183 1 DR. DOERFLER: Well, you know, with 2 methylation you have to be very cautious in 3 interpreting into biological problems, because we -- 4 To be quite honest, I don't think we understand in 5 full detail the implication. 6 So from my judgment at this point, I would 7 say, well, methylation may be -- the pattern across a 8 certain stretch of the genome may be just a model on 9 which chromatin structure is built, and in some 10 instances this relates to expression of genes, 11 permanent silencing of genes, and then how does that 12 relate to the tumorigenesis? 13 Well, one model I've got up in my mind can 14 pursue, that by changing a pattern like Dr. Baylin, 15 for instance, gave you several examples where in tumor 16 cells the patterns are changed, and early on we saw 17 viral transformed cells at a very different pattern 18 into different segments. 19 So, certainly, things change, but to say 20 it is an increase or a decrease, I wouldn't be able to 21 do that; because we have looked at a fair number of 22 human tumors in a number of mainly lymphomas and 23 Hodgkin in Cologne, and looked at different -- used 24 different probes. 25 For some of the probes there was an 184 1 increase; for other ones there was a decrease, and for 2 third ones there was no change at all. 3 So if I should translate that into 4 biologic significance, my feeling at this point is 5 perhaps chromatin structures are changed segment-wise 6 in different parts of the genome in a tumor cell as 7 compared to normal cell. That's almost a sort of a 8 given. 9 That's a blatant thing that chromatin 10 structure changes, but I always thought changes in 11 methylation patterns would be an interesting indicator 12 for such change, and then we could, once we have the 13 technology, try to delve deeper and see what do these 14 changes mean in terms of altered chromatin structure 15 and altered methylation pattern -- and altered 16 transcription patterns. 17 So when we saw these changes in lambda DNA 18 transgenic cells, of course, we got very excited that 19 this might be a more general phenomenon somehow 20 relating to tumorigenesis. I don't know how Dr. 21 Baylin would feel about these proposals. 22 DR. BAYLIN: Well, again, if the question 23 becomes more the one of transfer, which I think is the 24 issue, it's hard to quite envision. The biggest 25 danger -- If you're using any virus, I think, as a 185 1 vehicle for inserting or carrying something, the virus 2 itself is the biggest threat. 3 I think there's going to be a lot more 4 learned, and some of these viral proteins, in addition 5 to binding up and taking away RB function and P53 6 function, I think, can cause probably -- It's going to 7 emerge that they probably can cause methylation 8 changes in trans by binding up or changing parts of 9 what is the methylation machinery itself, the ability 10 to target the methyl transferases and all this. 11 This is very little studied, but I think 12 most viruses can set up methylation changes, and part 13 of it may be that. It's harder to envision with 14 antigens and proteins things where the background of 15 the cell itself will influence what you could 16 transfer. That would be a danger, I think. That 17 would be my view. 18 DR. DOERFLER: Perhaps I could still make 19 -- cite one example, because in general it's -- One 20 uses the sort of the correlation or inverse 21 correlation between gene activity and levels of 22 methylation very frequently and loosely. 23 So over the years I have learned the hard 24 way that one has to be sort of very cautious, because 25 it may be only certain segments of a promoter, and 186 1 unfortunately one cannot give a blueprint to say, 2 well, it has to be 35 or 38 base pairs upstream from 3 such and such a site. 4 It doesn't seem to be the case. It's more 5 complicated. Then an extreme case, which has taught 6 me a lot, is again a virus, frog virus 3, which was 7 initially worked on by Dawn Willis and Allan Graniff 8 in Memphis. 9 This is a DNA virus which replicates both 10 in fish cells, frogs probably as well, and also in 11 mammalian cells like BHK. When you look at this DNA 12 with any methylation sensitive restrictase, it seems 13 to be totally methylated. 14 Several groups, including our own, have 15 used genomic sequencing techniques to look at all the 16 CGs, and again this is a completely methylated genome. 17 And yet, a late expression is going on, and with 18 genomes that are totally methylated in all CGs. 19 So Mark Monitz in our lab a number of 20 years ago did an experiment in which he used one of 21 these late promoters which are active, fully 22 methylated state in a viral infection, and hooked it 23 up to an indicator gene like Luciferase and 24 transfected it in the fully methylated stage. Yes, 25 it was active. 187 1 However, when he just did a Hapa 2 2 methylation, a CCGT which -- there were about eight 3 sides -- it dropped its activity. So here's an 4 example which would support the notion that perhaps a 5 certain pattern has to be present for an active or an 6 inactive promoter, and this frog virus, for reasons I 7 don't think we understand, prefers a completely 8 methylated promoter to attract proteins possibly that 9 facilitate transcription. 10 So one has to look in each promoter and 11 each biologic system separately what this methylation 12 means. So my suggestion is it's a modulator of DNA 13 protein interactions and that we have to be working 14 very hard to find out for each promoter what is the 15 important pattern for the active or inactive state or 16 for building a chromatin structure that is consistent 17 with biologic activity. 18 DR. FRIED: Thank you. I want to get back 19 to -- DNA would be the things that we're really afraid 20 of, if there are some oncogenes in a cell. So I don't 21 know how efficient DNA transfer is. 22 I mean, are there people who do DND 23 vaccines? Could they say there's a sort of naked DNA 24 go over or do you have to shield it somehow? What's 25 the efficiency if you put it in the microgram, how 188 1 many cells get it? Does anybody know those answers or 2 can say anything? 3 AUDIENCE PARTICIPANT: Ron Desrosiers 4 might make a comment, because of the SIV experiments, 5 which I think are probably the pertinent ones where 6 you get an actual readout of virus production from 7 putting in intact genomes. 8 We've done some of the experiments to him 9 using SIV, and you need to put in very large amounts 10 of DNA relatively still to get a functional expression 11 of a replication combinant virus. 12 So you're still talking in tens of 13 micrograms into an animal before you get out a 14 functional virus. So it does imply that the uptake of 15 DNA to result in a functional end expression is 16 actually extraordinarily rare, and we're talking in 17 micrograms of DNA. 18 I think the SIV experiments are the same. 19 AUDIENCE PARTICIPANT: The question was 20 what was the nature of the DNA. This is linearalized 21 proviral -- intact proviral DNA going in. We have 22 actually shown that if you actually take a retrovirus 23 and you split it into gag and pol and env constructs, 24 just as in a packaging line, put those in, you can 25 also get expression in vivo and get particle 189 1 production. 2 DR. PURCELL: Yes. My name is Damian 3 Purcell from Australia. I was just going to make a 4 comment of some work that -- on SIV that I'll present 5 on Thursday night regarding infection similar to Ruth 6 Ruprecht and de Rossier's experiments with injected 7 the naked DNA as plasmid, circular plasmid, in small 8 numbers of animals. 9 We found that we can very efficiently 10 initiate infection. When we put this DNA onto, for 11 example, gold beads, we get down below ten nanograms. 12 DR. FRIED: You're putting it on beads? 13 If you just -- 14 DR. PURCELL; On beads as, yes, gene gum, 15 and we haven't titrated it, but putting it on the 16 method of transfer, the naked DNA can under some 17 circumstances such as gene gum inoculation be 18 extremely efficient. 19 DR. FRIED: But would the naked DNA be 20 efficient? I mean, if some was -- there was some DNA 21 that got out of a cell, what's the chance of that 22 getting over? 23 DR. PURCELL: Yes. We can't really 24 comment on that, because we didn't titrate the DNA, 25 but in our experiments with naked DNA injection, we 190 1 just followed a 300 microgram amount, which -- 2 DR. FRIED: When you say it was efficient 3 with the beads, what kind of numbers are we talking 4 about? 5 DR. PURCELL: In terms of numbers of 6 animals? 7 DR. FRIED; No, in terms of amount of DNA 8 in successful infection, how could you dilute it down 9 to? 10 DR. PURCELL: We've only gone as low as 11 ten nanograms, but maybe one additional piece of 12 information is that included among the DNAs are DNAs 13 where we've made deletions in the promoter region of 14 SIV in the U3, so that the expected expression might 15 be lower than wild type and still, even with ten 16 nanograms delivered by gene gum, there's efficient 17 initiation of infection under those circumstances. 18 DR. FRIED: Thank you. 19 DR. PETRICCIANI: John Petricciani. 20 There's a large body of data actually in the 21 literature that goes back a couple of decades. One of 22 the experiments -- these are in vitro experiments on 23 DNA transfer. One of the experiments that was 24 mentioned earlier today was the transformation of 3T3 25 by human tumor cell DNA. 191 1 In that particular set of experiments, the 2 amount of DNA required to get any positive readout was 3 20 micrograms, and it required also carrier DNA in 4 addition to the human tumor cell DNA. The bottom line 5 is that in the studies of which I'm aware in vitro, 6 let alone in vivo, it's extraordinarily difficult to 7 get a positive uptake in expression of DNA, whether 8 it's cell transformation or other gene expression. 9 DR. FRIED: I assume that, if somebody 10 aware of these things, they would be nucleasing up 11 their preparations before they -- I mean, they would 12 be using -- trying to break down the DNA as small as 13 possible. 14 AUDIENCE PARTICIPANT: Well, it's a 15 question of inactivation, reduction through one or 16 another mechanism. Right. 17 DR. SHIVER: John Shiver from Merck. Our 18 experience is that plasmid DNA is not efficiently 19 taken up by direct injection into most tissues into 20 animals, whether they're non-even primates or mice, 21 and that it generally requires some degree of 22 amplification of whatever response you're looking for 23 in order to detect it, such as an immune response to 24 a transgene encoded by the plasmid or actually a virus 25 virion developing capable of replication, as perhaps 192 1 in the SIV or FIV model. 2 Looking for direct gene expressions 3 systemically is generally not successful. Also, the 4 uptake of the DNA is highly dependent upon the state 5 of the DNA. It really needs to be compacted 6 supercoiled to really -- to work even as well as it 7 does, generally, with naked DNA injection. 8 DR. FRIED: So your feeling, if there was 9 oncogenes inside of a DNA and some of that was taken 10 up by recipient cell, it would be very inefficient to 11 get that expressed? 12 DR. SHIVER: Should be very inefficient, 13 especially in the linearized fragments and form that 14 you would expect that DNA to be in. 15 DR. DOERFLER: Just in response to this 16 interesting comment, in addition to feeding DNA with 17 all that we're injecting DNA, mainly the green 18 fluorescent plasmid construct, fewest cytomegalovirus 19 promoter or SV40 promoter, recently also the ras 20 sarcoma virus promoter but we haven't done many 21 experiments on that yet, and we've also mainly failed 22 to find expression except locally in the muscle where 23 we injected it into. 24 It could be traced by FISH and PCR in 25 various organs for quite a long time, and although 193 1 these data are repeated only a number of times and not 2 for years, we've also seen it in testing the contents. 3 So this may be an interesting observation 4 for people around here. It might be excreted probably 5 via the bile route. I'm not sure how it works. But 6 definitely we can find some of this DNA in the 7 intestinal contents. 8 DR. SHIVER: We would agree with that 9 experience overall. The DNA that is residual and 10 remains tends to be extrachromosomally located -- you 11 know, with some persistence time, but generally gene 12 expression only near the site of injection. 13 DR. DOERFLER: Right. Right. So you do 14 find expression at site of injection? We do, too. 15 DR. FRIED: So, I mean, I think the 16 general feeling is, if there was DNA, cellular DNA 17 with cellular oncogenes, that would be very 18 inefficiently transferred, and I think anybody 19 preparing a vaccine would take great precautions in 20 trying to destroy the DNA or make it as small as 21 possible. Is that what the panel would think? 22 So the way the cell -- the neoplastic cell 23 was formed, we don't really expect most of these 24 properties, methylation or loss of tumor suppressor 25 genes or viral oncogenes to be transferred over in a 194 1 vaccine preparation. 2 Is there anybody who disagrees with that? 3 AUDIENCE PARTICIPANT: I don't disagree, 4 but maybe I could ask you to clarify a bit the 5 statement that you made a moment ago, that you'd like 6 to see the DNA made as small as possible in order to 7 avoid these kinds of problems. And of course, from a 8 regulatory perspective, somebody comes in with a 9 product and they're doing a particular purification 10 scheme on it and it makes the DNA a certain size, and 11 so we can always say can you make it smaller. But the 12 question is how small then is small enough? 13 What if somebody comes in with a product, 14 and they haven't made the DNA smaller? Does that mean 15 we should be worried about it? Likewise then, the 16 question of if you have one cell line which is more 17 tumorigenic than another, again the question is, if we 18 truly think that one which is less tumorigenic is 19 safer, then that should be something that one can put 20 some numbers on. Then one has to, obviously, come 21 with some kind of a threshold of sort of trying to 22 define this. 23 So I know it's thorny and difficult when 24 trying to put numbers on this kind of thing, but I 25 think the kind of information that the FDA needs to 195 1 make these decisions requires some more definition. 2 DR. FRIED: I think you probably have to 3 spike your preparations with, you know, some really 4 efficient either phage DNA or something or some of 5 these really nice, you know, circular small polyoma 6 SV40, which are probably efficient in assays, and then 7 you know, at different stages of the preparation see 8 how much and where it goes. 9 AUDIENCE PARTICIPANT: That certainly 10 sounds like a reasonable suggestion, if in fact the 11 recommendation then is that one has to presume that 12 such DNA is there, in the first place, and that one 13 does have to get it to a size that's small enough to 14 prevent these kinds of outcomes. 15 DR. FRIED: I mean, I was going to get to 16 it. But the only -- Is there a comment in the back? 17 The only thing that I was not including is possibly 18 gene amplification, which people don't look at. 19 I would certainly look at cells, because 20 those are cells that usually amplify oncogenes. They 21 also amplify more than one oncogene. More and more 22 information is coming out that oncogenes get together. 23 There are large circles that we could see 24 satellogically, but there's also generated small 25 circles of DNA. So there you're sort of amplifying 196 1 up, and I certainly -- If one was using neoplastic 2 cells -- I was going to get to this later -- you would 3 certainly look for -- you know, cytologically, for 4 double minutes or HSRs, and I wouldn't stay away from 5 those, and I certainly would look for polydispersed 6 circles which have been reported in cells, small 7 circular DNA. 8 We don't know where that comes from, but 9 certainly, a lot of the gene amplifications has to do 10 with oncogenes and myc. I think all -- If anybody is 11 using cells, they should certainly do EM analysis for 12 circular DNA and for HSRs and double minutes. 13 DR. KAPPES: John Kappes from UAB. I'd 14 like to suggest an alternative mechanism for genetic 15 transfer, at least where the vaccine would be 16 retroviral derived or relying on retroviruses for 17 cancer vaccination, for example, if our hopes of gene 18 therapy ever come to fruition. 19 Data I will present tomorrow could suggest 20 transfer of that information via the messenger RNA. 21 That is, instead of relying on DNA transfer, with HIV 22 based vectors I have evidence that the poly-A signal 23 could help in some way I don't understand yet mediate 24 -- transfer that information. 25 That is, if the oncogene message was 197 1 packaged and that polyadenylation somehow modulated 2 recombination with the transfer vector, and if that 3 was integrated, that might be an alternative concern 4 to take into account for transfer that type of genetic 5 information. 6 DR. NICHOLS: Warren Nichols, Merck. 7 There's certainly been a lot of data published over 8 several years now about taking tumorigenic DNA from 9 cells that are tumorigenic and injecting very large 10 quantities into a variety of laboratory animals and 11 newborn rats, newborn hamsters, nude mice, 12 antithymocyte treated rats and mice, and when giving 13 up to 100 micrograms of these, things like HeLa cells, 14 CHO cells, the DNA from HeLa cells, CHO cells, human 15 bladder tumor cells, cloned activated onc genes, there 16 has been no evidence of any tumor formation in any of 17 them. 18 Just as a comment on distribution, this 19 isn't a comment on distribution of cellular DNA, but 20 in looking at a plasmid from DNA vaccines, when 21 injecting 1011 plasmids into a muscle and then looking 22 at various time periods after that, we can see a great 23 many plasmids still in the muscle six weeks or 12 24 weeks later. But in the other tissues, in blood and 25 brain and spleen and a variety of tissues, 13 that we 198 1 looked at, by six weeks there's never any present. 2 If we look just to try to see where it has 3 been distributed, in day one and day two we can find 4 it in most organs. By day seven, it spontaneously has 5 all gone. When we use purification methods to 6 separate plasmid from a genomic DNA even at two days 7 or three days, it all disappears. 8 We use in looking at this PCR methods that 9 are sensitive to .005 to .01 femptograms. It's about 10 one plasmid in 150,000 nuclei per microgram of DNA. 11 So that it certainly does get around, but it certainly 12 does disappear quite quickly. None of it has ever 13 been demonstrated to be integrated in that kind of 14 system. 15 I do think that the injections with the 16 DNA derived from tumor cells in huge quantities 17 compared to what is present in any vaccine is very 18 important. 19 DR. FRIED: Could you repeat the question? 20 AUDIENCE PARTICIPANT: Were those done in 21 normal animals? 22 DR. FRIED: Were which done in normal 23 animals? 24 AUDIENCE PARTICIPANT: Injected from the 25 DNA. 199 1 DR. NICHOLS: The injection of the 2 plasmid? The question is were the injections of DNA 3 done in normal animals? The injections of plasmid 4 into animals and then following and looking -- that 5 was in normal animals. 6 The injection of cellular DNA that's in 7 the literature is injecting into newborn hamsters, 8 newborn rats, newborn nude mice and rats and mice, 9 newborn rats and mice that had antithymocytes here. 10 AUDIENCE PARTICIPANT: Dr. Nichols, excuse 11 me. Have you published that data? 12 DR. NICHOLS: Which data? 13 AUDIENCE PARTICIPANT: The ones you're 14 just referring to? 15 DR. NICHOLS: The data about the plasmid 16 injections, that is published; and there are two other 17 papers that are in preparation right now, but the 18 first one is published. The other data from the 19 standpoint of the cellular DNA being injected, that's 20 all from literature that exists now, not from us but 21 from other -- many other people. 22 DR. FRIED: Thank you. 23 DR. MINOR: Philip Minor from NIBSC. 24 Warren, how do you square your lack of detecting any 25 integration with Dr. Doerfler's comments this morning 200 1 about integrating M13 after you feed it to mice? It 2 does seem to me, there's a little bit of ambiguity 3 here about what's really going on. 4 DR. NICHOLS: Well, let's say that there 5 are always other systems you can look at, but up to 6 now with that sensitivity we haven't seen anything 7 that has remained after separation techniques. 8 We were talking about the possibility with 9 Dr. Doerfler of having a ligase in the process being 10 able to hook up things that make it look like 11 integration that isn't really integration, and he has 12 told us that there are very rare events, you know. 13 There are just two different observations, and it's 14 hard to explain what -- 15 DR. FRIED: But one is cellular DNA, and 16 Walter, yours was just adenovirus. It wasn't -- I 17 mean, Walter didn't do the comparable experiment. 18 DR. DOERFLER: Well, if I just can comment 19 on that, the DNAs we have been using were mostly DFH 20 M13 DNA double stranded form, either linear or 21 circular. Most of the experiments were done with ECO 22 1 linearized DNA. 23 I don't think there is so much of a 24 discrepancy between these observations, because we 25 looked for -- In most experiments, we looked 201 1 relatively, you know, early after application. So the 2 recloning was done over 18 hours after feeding, from 3 the last feeding, from the spleen. So we can see it 4 in the nucleus and perhaps chromosomally associated or 5 even linked to cellular DNA. 6 This was a recloning experiment and, I 7 think, for the definite this is M13 DNA. Whether or 8 not it's linked to mouse DNA, well, it was suggested 9 it could be. But, of course, there are other 10 possibilities, how we might find such a linkage. 11 That was the latest we could do. That was 12 18 hours after the last feeding. If we do it later, 13 we have great difficulties finding it. So we 14 considered that a rare event that only -- persistence 15 only for a certain time, although experiments we 16 didn't discuss in feeding pregnant animals, there may 17 be a longer persistence in the offspring. 18 DR. MINOR: It wasn't the question of 19 persistence so much as the question of actually any 20 integration at all, you see. I think what Warren was 21 implying to me was that there is no integration of the 22 plasmid DNA that he injects into the mice; whereas, as 23 I understood it from your feeding experiments, there 24 was evidence of such expression. 25 DR. DOERFLER: Well, you know, we have two 202 1 lines of evidence. One is the recloning, and the 2 second is data I didn't show, chromosomal association 3 in embryonic cells after feeding pregnant animals in 4 rare cases. So one really has to look hard for it. 5 DR. FRIED: How long after did you look? 6 DR. DOERFLER: Well, in the directly fed 7 animals, it was -- The animals had received a dose of 8 50 micrograms day before one week, and then 18 hours 9 after the last feeding the animal was sacrificed, 10 spleen extubated, and the DNA was cut and cloned. 11 For the offspring of pregnant animals, we 12 looked in fetuses and in embryos. 13 DR. MINOR: I've got a related sort of 14 technical question, too, I think. As Warren said, I 15 mean, it depends very much on what system you're 16 looking at. I wondered, most of the carcinogenicities 17 or assays that we've been hearing about today are 18 really being chemical carcinogenicity type approaches. 19 I wondered how appropriate those would be 20 to something like a DNA where you're talking about a 21 particular gene going in. I mean, to what extent is 22 this system actually appropriate for what you're 23 looking for, if you're looking for DNA oncogenicity. 24 DR. FRIED: Well, that gets to the next 25 sort of topic. How do we test for oncogenicity. 203 1 Maybe Jim will want to give his talk now. 2 AUDIENCE PARTICIPANT: Can I ask a 3 question about one of the earlier comments? 4 Did you use any transgenic mice in that 5 DNA study, because it's for the very reason that was 6 just mentioned. If an activated oncogene was 7 involved, it wouldn't cause, you know, global changes. 8 It might cause changes that would just be measured in 9 an animal that only needed a promoter or only needed 10 a knocking out of -- that was heterozygous and only 11 needed a knocking out of one copy of RB or something 12 like that. 13 The other comment: I also see a paradox 14 here in saying that there's no real risk of this DNA 15 being taken up. When you consider that almost 16 everyone of these vaccines is injected right into the 17 tissue, which is the preferred site for DNA gene 18 therapy, which is the preferred site for DNA vaccines, 19 which is the best way to get expression of 20 extracellular just naked DNA -- the best place you 21 could go is right into those same muscles. 22 I'm a little surprised. I'm actually very 23 surprised to hear that you think that just injecting 24 DNA there won't be taken up. In fact, the fact is DNA 25 at that site will be taken up. That's why Dr. 204 1 Doerfler's results are what they are, and they will be 2 expressed, and they will persist for quite a long 3 time, surprisingly long. 4 I don't know about you. I was very 5 surprised to see how long DNA injected directly into 6 a muscle will be expressed there. Whether that 7 oncogene being expressed at that site will cause 8 cancer in a mouse, whether it even could cause cancer 9 in a mouse, is maybe the next question. But I think 10 you couldn't do much more to get the DNA expressed 11 than to inject it into a muscle the way it's being 12 done. 13 I mean you could. You could damage the 14 muscle. You could inject lidocaine and various other 15 things, but it would be hard to maintain that -- I'd 16 like to interpret your result, which is that if you 17 don't add enough DNA, you don't get it; and if you add 18 too much, kill the cells that would have responded. 19 AUDIENCE PARTICIPANT: Let me just say 20 there's a big difference between being taken up and 21 being integrated. The DNA is certainly taken up by 22 the cells, and it is expressed by the cells. 23 What we're saying is there's no evidence 24 that it's integrated into the genome of the recipient 25 cells that can be determined in our studies up to this 205 1 time. 2 DR. FRIED: So it's taken up, but it's not 3 necessarily integrated. Is that -- So there's no long 4 term -- The long term persistence is 5 extrachromosomally? 6 AUDIENCE PARTICIPANT: Yes, that's 7 correct. At least with the sensitivity that we can 8 get of the .005 to .01 femptograms, it all is taken 9 away by the purification procedures that we do, and 10 nothing remains. 11 AUDIENCE PARTICIPANT: And also somewhat 12 related to this, I heard you say that you hope the DNA 13 is reduced to its smallest possible size. I believe 14 it's true, the statement that not a single vaccine is 15 treated with DNAase and that any considerations of 16 reducing DNA to its smallest size have simply been 17 missed by people who make vaccines. 18 DR. RUUD: Rupert Rudd from Solfi. We are 19 using DNAs or our vaccine, and in general here I want 20 to state that we are now looking at cells, viruses 21 that are being expressed and can be transferred into 22 vaccines. 23 I think it's also fair to say what you 24 define as vaccine -- I mean, if your final product is 25 like a monoclonal antibody, we have no problems here. 206 1 If you have a purified inactive phage, it's a purer 2 product which is also a protein, but it is from virus 3 that has been grown on certain cells. Then you may 4 have a problem. 5 I think this dispairment should be 6 broadened and more defined in what we call vaccines 7 and what kind of purpose it is, and list a series of 8 the amounts that are necessary for that particular 9 line. 10 DR. FRIED: Walter? 11 DR. DOERFLER: Just one other comment to 12 the ongoing discussion on persistence and in what 13 form. Now when we look by sectioning different organs 14 and fish after feeding DNA, then we invariably see the 15 DNA inside the nucleus. That doesn't mean it's 16 integrated. 17 Now the recloning and also chromosomal 18 association would indicate to me that perhaps very, 19 very rarely it can be integrated. So the limit Warren 20 has given us are perfectly agreeable with me, because 21 it's a very, very rare event, if it occurs at all. But 22 it's frequently in the nucleus. 23 So there's no question about its uptake 24 and, once it's in the nucleus, of course, one doesn't 25 know what might happen. 207 1 In the embryo we've also seen it in the 2 nucleus, usually in more than one cell, cell type 3 justice, which might indicate that the cells still 4 divide after receiving some of the DNA, and then this 5 pattern emerges after what we consider transplacental 6 uptake. 7 DR. FRIED: Do you think you have to have 8 cells dividing for the DNA to become integrated and, 9 when they're injected into certain tissues, that the 10 nondivided cells that DNA just persists and then goes 11 away? 12 DR. DOERFLER: Possibly. 13 AUDIENCE PARTICIPANT: I want to follow 14 up. Jim McDougall presented some data that I was not 15 aware of, and that is that fairly small pieces of the 16 herpes virus genome seemed to be associated with this 17 hit and run mechanism. 18 I didn't count all the base pairs in those 19 stem loop structures he had. It looked like to me 20 somewhere between maybe 25 and 75, but I think the 21 sequences you had that was about 500 base pairs, and 22 I don't believe you looked at any smaller than that. 23 The question is whether you could chop 24 away at those things and still show that that stem 25 loop was associated with that type of event. But I 208 1 think the question that really comes to my mind is to 2 whether those stem loop type structures are present in 3 mammalian cell DNA, and if they are associated with 4 hit and run transmission of herpes viruses and, if 5 similar stem loop structures are present in mammalian 6 DNA, is that something that we would have to worry 7 about from a regulatory perspective? 8 DR. McDOUGALL: Well, I think it's not 9 difficult to find stem loop structures in almost any 10 source that you go and look at. Their presence is 11 certainly there. 12 Did we cut them down any further? No. 13 The size that I showed was the -- That was the 14 structure that we could see in a piece of DNA that 15 actually was bigger than that, but that was the 16 consistent structure that was always present and was 17 capable of transforming cells. But again, you know, 18 I continually come back to what I know for sure. 19 That is, you have to use large amounts of 20 DNA to get this sort of transfection to work, and I 21 cannot believe that in any vaccine that's produced 22 there would ever be residual DNA at that sort of 23 level. 24 Now it's true that, you know, what you may 25 be looking at is a single event, and so if you 209 1 vaccinate 10 million people, you're likely to find one 2 somewhere out on the end of the bell curve that gets 3 a reaction to that. But that's a price that you may 4 have to pay in this situation. 5 The other -- While I'm on, the other thing 6 that I do feel is that a decision whether or not to go 7 ahead with this sort of -- this type of cell line 8 that's been immortalized or transformed by a DNA 9 virus, by papilloma or SV40 or adenovirus or whatever 10 -- I still think that that sort of decision should 11 only be made on the question of what residual DNA 12 might not be in the vaccine, and not on the basis of 13 whether a cell is partially moving toward neoplasia or 14 completely moving toward it; because I think the truth 15 is that, once you make a cell genetically unstable, it 16 is going to accumulate more changes, whatever you do. 17 So the one thing I keep coming back to is 18 how much DNA is going to be in one dose of a vaccine, 19 and I cannot see that that will ever carry enough 20 material to actually be damaging. 21 DR. HOEKENOF: I'm Ray Hoekenof of 22 Intergene in Holland. I collaborate with Alex von 23 Raab who is in the forum. 24 I'd like to add one thing to the 25 discussion whether there is good cells or bad cells. 210 1 I think there was a comment earlier this afternoon 2 that there is no evidence of any correlation between 3 the neoplastic potential of a cell and its potential 4 to transfer that neoplastic potential to the vaccine. 5 I think there is one example that was not 6 mentioned yet of the potential of transfer of such 7 potential, which is the mechanism of homologous 8 recombination. 9 I think the 293 cells which are generally 10 used for gene therapy vectors but have also been 11 mentioned for production of vaccines provide a clear 12 risk that you can generate or can transfer elements 13 from the cell line to the vaccine simply by homologous 14 recombination. In 293 cells the 5 prime ITRS is 15 present. The encapsulation signal is present. 16 So, for instance, if you would produce 17 adenoviral batches on that cell line, you can very 18 quickly generate recombinant adenoviruses that have 19 taken up parts of the 293 cell line by homologous 20 recombination, a phenomenon known as RCA. 21 I think there will be a paper presented 22 later this week by Fritz Fallaux from van der Eb's lab 23 to present cell lines that eliminate that risk, and I 24 think homologous recombination between a neoplastic 25 cell line and a vaccine is something that should be 211 1 kept in mind as a risk factor for transfer of 2 neoplastic characteristics. 3 DR. FRIED: So you would think all cell 4 lines that had transformed by viruses -- I mean, but 5 293 doesn't have -- It only has a couple of genes 6 there. Is that -- 7 DR. HOEKENOF: That is correct, but -- 8 DR. FRIED: So you think when they come in 9 with the E1A minus vectors, they're going to recombine 10 with the endogenous genes? 11 DR. HOEKENOF: Oh, in 293 that is not 12 something that you expect, but it is a fact. You very 13 quickly start to generate revertant adenoviruses which 14 result from homologous recombination between the cell 15 line and the vector. 16 So the point I'm trying to make is, if you 17 want to produce a vaccine on a cell line, make sure 18 that there is no homology present between your DNA on 19 the vector and the cell line, and maybe van der Eb can 20 comment on that. 21 DR. VAN DER EB: I agree with that. 22 There's little I have to add. 23 DR. RUSSO: Carlo Russo from Merck. Are 24 you referring to recombination with viral genes, not 25 with cellular genes? 212 1 DR. HOEKENOF: Well, in this case I'm 2 referring to recombination with viral genes, of 3 course, but any homologous recombination can lead to 4 uptake of cellular DNA into your vector. So I think 5 you should avoid that. 6 DR. FRIED: But again let's not drift 7 away. We're asking in neoplastic cells, is it any 8 different than, say, diploid cells, and are these 9 phenomena that are specific to neoplastic as opposed 10 to diploid cells? 11 DR. RUSSO: I think what we need to 12 clarify is this concept of malignancy, because it 13 seems to the FDA, at least to a couple of their 14 representatives here have this concept that there are 15 different degrees of malignancy, and perhaps the cell 16 substrate should be identified, high malignancy, low 17 malignancy, medium malignancy. 18 I think we should have a consensus here 19 whether this is considered to be true or 20 independently, as somebody suggested, the only concern 21 should be on the final. What is the residual DNA that 22 we're going to have in our vaccine, independently from 23 which cell line the DNA came. 24 DR. FRIED: Do people on the panel have 25 anything to say on what's malignant, how we would 213 1 measure that? 2 DR. COOK: Well, we've covered this in 3 bits already, it seems like, in three parts. One is: 4 How do you measure this? And we've already talked 5 about quantitating the malignancy of the neoplastic 6 potential or the tumorigenicity of the cell. 7 The second is, once you have one cell line 8 of a given kind of tumorigenic potential, what's the 9 likelihood that something from that cell line could be 10 transferred through residual DNA in such a way that it 11 would be expressed functionally to lead to that in a 12 recipient of some sort? 13 I think those are probably quite different 14 questions. You can measure the first, once you define 15 the parameters, and we can talk more about that if you 16 want to at some point. But the second issue, I think, 17 requires experiments like were referred to that are 18 published in literature that I don't know about where 19 you put in large amounts of tumor cell DNA into 20 immunodeficient recipients, and you find no evidence 21 of pathology, whether it's tumor development or some 22 other kind of illness in those animals. 23 If you can't prove that transfer of that 24 information, even though it's all sheared and chopped 25 up, is going to do anything to the recipient host, 214 1 it's kind of hard to translate the information from 2 the virulence of the cell, if you will, to the risk of 3 that cell being used to convey something, if there's 4 no evidence that it's conveyed and if it's unlikely 5 that you're going to get the right combination of 6 genes put together after the sheared DNA is carried 7 over into the recipient. 8 I mean, that's where the question is. How 9 do you test that? You can test the first, but does it 10 have any relevance for the second; that is, the 11 transfer phenomenon, and what happens in the recipient 12 -- human vaccine recipient? 13 DR. FRIED: Is it worth testing it in the 14 transgenic models? I mean, which would be the best 15 that people who presented on the transgenes? 16 AUDIENCE PARTICIPANT: Well, I think the 17 question that we sort of have to get after is whether 18 it's worth pursuing and how far or how hard we should 19 pursue the need to ask the second question. 20 We can define the first and, in fact, as 21 you know, the issue of whether cells that have very 22 low TPD50 values on the order of one to a few hundred 23 cells compared to cells that have very high TPD50 24 values of, say, in the order of 105 or 106 -- the 25 question is whether we should be more worried and 215 1 pursue with some -- oh, I can't think of the word I 2 want to use right now, but with some eagerness to look 3 at that question with cells that have that. 4 Now in order to make that decision, the 5 first thing you have to have is data that says there 6 is a cell that this particular substrate of this cell 7 that's being proposed for this particular substrate 8 falls into that category of cells that are highly 9 neoplastic or highly tumorigenic, if you will. 10 Neoplastic in this situation, I think, is 11 not quite the right word, because you can't -- I don't 12 know how you measure neoplastic. You can measure 13 tumorigenicity, but the other thing is a more nebulous 14 term when it's a tissue culture cell. 15 So if you have a situation in which you 16 have a very aggressive or a highly tumorigenic cell 17 line, should we in fact be concerned about pursuing 18 whether the components of that cell represent more of 19 a risk, and then we should in fact worry about how to 20 measure whether those risks are transferable or not, 21 I think, is the heart of our question. 22 We can measure the one, but we're not 23 quite sure we know how to measure the other, and when 24 you look at the models that are available for that, 25 there aren't very many of them, especially the tissue 216 1 culture model. 2 Now I don't know, with regard to looking 3 at DNA from cells that are highly aggressive or highly 4 tumorigenic versus cells that are weakly tumorigenic, 5 whether those types of studies have actually been 6 done. 7 Dr. Nichols is talking about a lot of 8 data. Unfortunately, I'm not familiar with that data. 9 So I don't really know whether he's looked at cells 10 that have very low TPD50 values, comparing it with 11 cells that have very high TPD50 values in his models 12 or not. But I think that's one of the things that we 13 could think of doing, if we need to get at that kind 14 of data. 15 So I think the question to the panel is 16 whether we in fact need to get at that kind of data, 17 and if so, you know, is there any collective wisdom 18 among you all that says how we should go about doing 19 it? 20 DR. SISTARE: Two questions, I think. One 21 is, if we're concerned that the cell is going to be 22 surviving some processing and the cell is going to 23 remain intact in the vaccine, then we got to approach 24 it a certain way. There, the question of this 25 neoplastic potential in the cell is a very real one. 217 1 If, on the other hand, our concern is not 2 the cell -- we don't think the cell is going to 3 survive the processing of the vaccine, and it's not 4 going to be carried over, and what we're concerned 5 about is the DNA itself, that's a different question, 6 and I don't know that it's been addressed yet. 7 It strikes me that some of the experiments 8 that Dr. Doerfler referred to and Dr. Onions referred 9 to were newborn hamsters, I believe it was, and 10 newborn young animals. They were injected or gavaged 11 or whatever. 12 I don't know how the stuff was given, but 13 they were given DNA in some form, and these animals 14 came down with tumors, if I understood your statement 15 correctly. It wasn't a primary part of your 16 presentation, but you referred to it, some old data. 17 In a sense, the newborn mouse model, the 18 neonatal mouse model that we use for testing 19 pharmaceuticals is that same model. It's a single hit 20 or double hit kind of a model with a long latency, a 21 lifetime -- you know, just a one exposure, wait a 22 lifetime and then see if a tumor develops. That might 23 be the kind of model to ask the kind of question, 24 neoplastic DNA versus primary cell DNA. 25 Now another thing that Dr. McDougall 218 1 brought up -- and it's another reason why this is a 2 very interesting experiment -- If you remember 3 correctly, he indicated that these small pieces of 4 palindromic DNA seemed to be doing the work, seemed to 5 be causing the problems, and it was the more mature 6 cell line the tumor actually lost these pieces of DNA. 7 So one might guess that the early phase of 8 the tumor might be the more pathogenic, the DNA from 9 that, as opposed to the later, the more mature, more 10 "tumorigenic" DNA. 11 So you don't know how the experiment is 12 going to turn out. If this palindromic DNA is really 13 the important stuff, it's going to go in and not 14 express -- if expression is not the most important 15 thing, but disruption of normal expression of the cell 16 is the key thing, then these are all kinds of 17 questions that could be addressed, and there are 18 models that we could do this with. 19 I don't know if the data is out there 20 already for these kinds of things. I don't know. An 21 we keep talking about efficiency of transfer and 22 looking for expression. That's not what we're trying 23 to do here, I don't think, is ask the question about 24 necessarily whether oncogenes are going to be 25 transferred. 219 1 I think the question we have to ask for is 2 the rare event. If there is transfer, and we know 3 it's going to be very efficient, when that transfer 4 occurs is it going to integrate into a part of the 5 cellular genome that's critical to the function of 6 that cell? And if it's only one cell, and it gives it 7 some selective advantage, then it's a concern. 8 So it can be a rare event that we have to 9 concern ourselves with. 10 DR. FRIED: Any other comments? 11 AUDIENCE PARTICIPANT: Given what you all 12 know as experts on neoplastic cells, suppose you have 13 a cell which is neoplastic for which you don't know 14 the mechanism by which it became immortal. What is 15 the relative pretest probability as compared with 16 other types of cells that these cells will contain an 17 abortitious agent, either infectious or latent, known 18 or unknown? 19 We'd sort of just be interested in, in the 20 sense of the panel, whether it is reasonable or 21 unreasonable to be more worried about the possibility 22 of abortitious agents being in these kinds of 23 neoplastic cells. 24 DR. FRIED: I guess, if you passage 25 things more, they may be there, and it depends on what 220 1 species we're looking at. I mean, I think Harry Rubin 2 said this morning something about if they came from 3 chicken, it probably wouldn't infect humans. Then I 4 think John Coffin said maybe it would. 5 So I don't know. I mean, I guess agents - 6 - I mean, there's just the normal tests looking for 7 viruses and nuclease resistant DNA in particles and 8 PCRing up, as mentioned in the document. I don't know 9 if the panel has any other ideas. 10 AUDIENCE PARTICIPANT: I was actually 11 going to address a different issue which was mentioned 12 earlier and has been mentioned a couple of times so 13 far, which has to do with the damage due to 14 integration of DNA in disruption or activation of 15 genes. 16 In the very best models that we have for 17 that, which is infection of chickens or mice with 18 certain retroviruses, where very reproducibly every 19 single infected animal will come down with a tumor, if 20 you actually look in cases where the tumor is being 21 caused by insertional activation and not expression of 22 preexisting oncogenes, if you actually count the 23 numbers of cells that are infected as compared to the 24 ones that actually become transformed in this way, the 25 ratio is probably a million to 10 million to one or 221 1 something like that. 2 These events are extremely rare at the 3 cellular level. The introduction of foreign DNA has 4 a very, very low probability of doing any harm to a 5 cell in terms of induction of malignancy. So we have 6 to keep that in mind. We can't think that 7 introduction of DNA is per se a very damaging event to 8 a cell. 9 You have to do it a lot of times. You got 10 to do it a lot of times in an infected animal in order 11 to get a tumor to appear. 12 DR. FRIED: You think it's preselecting a 13 cell already is halfway there? 14 AUDIENCE PARTICIPANT: Well, there might 15 be some of that. There's certainly a multi-step event 16 to it, and it's certainly true that only certain cells 17 in the body are primed for this. 18 It may not be because they're abnormal. 19 It may just have to do with something about the 20 differentiation, say, of that particular cell, but you 21 can infect many cells with avian leukosis virus, for 22 example, in a young chick, but you will virtually 23 always get a B cell tumor that arises in the bird. So 24 presumably, because there are some particular features 25 about those cells that prime them, not necessarily 222 1 mutagenically, but in a differentiation specific way 2 for the kinds of events that can then lead to 3 oncogenesis. 4 DR. FRIED: The same with the P53 mice. 5 I mean, they all get tumors by 200 days, but they 6 don't get tumors on day one or day ten. Other events 7 have to happen, and there's also a story of BRCA-1. 8 There's a woman in Scotland in a little 9 village where everybody knows everybody, and she was 10 actually a knockout. She had had both BRCA-1 genes 11 knocked out, and she didn't get a breast tumor until 12 39. So again, other events have to happen. 13 AUDIENCE PARTICIPANT: The rarity of 14 these, I think, is a very important thing to keep in 15 mind. 16 DR. TEVETHIA: I'd like to get back to the 17 issue of the degree of tumorigenicity. I think it's 18 a very important point for the cell line you're going 19 to consider making in vaccines. 20 Seems to me that we seem to be using this 21 word rather more loosely, and I think the question 22 really becomes that, if you look at any of the DNA 23 virus transformed cells, they are transformed, but 24 they don't metastasize. 25 Yet you express a gene in transgenic mice 223 1 as metastasizing. So, obviously, we're looking at a 2 lot of artificial situations when you transplant a 3 cell line into a tumor. The cell may grow rapidly. 4 It may look like a tumor. 5 One really has to define actually what the 6 tumor looks like, at the blood supply, how soon the 7 blood supply is established, and other factors that 8 are produced. 9 So I think just because the cell line 107 10 cells end up producing a nodule doesn't necessarily 11 mean that is a very high cell line to make vaccine in. 12 DR. FRIED: Okay. Is there anything more? 13 DR. McDOUGALL: I'd just like to make one 14 observation on that observation. That is that human 15 papilloma viruses actually do create a tumor that's 16 capable of metastasizing, and this is an event that's 17 just initiated by a DNA virus infection. 18 So it's not -- Although in vitro and the 19 sort of lab experiments we do it's true that these 20 cells don't metastasize, in vivo they certainly do. 21 AUDIENCE PARTICIPANT: Yes. I have some 22 unpublished data on the ability of a lot of these 23 cells metastasize, and it's, in fact, quite striking. 24 If you look at SV40 transformed hamster 25 cells, it's the size of the tumor that depends on 224 1 whether the cells metastasize or not. They won't -- 2 You can't begin to see lung metastasis in animals with 3 SV40 transformed cell tumors, hamster cells, until the 4 tumor is probably 30-40 millimeters in diameter. Then 5 the lungs are studded with metastasis. 6 We found the same thing for BHK 21 cells. 7 Now the interesting thing is that adeno-12 transformed 8 cells, which have the reputation of being highly 9 tumorigenic -- we've never seen a lung metastasis in 10 animals, even though the tumor is half the size of the 11 animal. 12 So there are differences there. We've not 13 looked -- made the same observations in mouse models, 14 but the idea that DNA virus transformed cells, 15 especially SV40 and some of these types of viruses, 16 won't produce metastasis, I think, is based on the 17 fact that you really haven't looked very hard, and you 18 have to wait until the animal gets around to 19 development of lesions. They're probably microscopic, 20 and you can't see them. 21 DR. COOK: I would just agree. I think 22 that the way these experiments are done is usually to 23 inject a large number of cells that causes tumors in 24 a fairly short period of time, and it creates an 25 environment, like Tev was saying, where you develop 225 1 localized tumors that don't look like the metastasize. 2 If you put in smaller numbers of cells and 3 wait a very long period of time, like Dr. Lewis' 4 experiments, we've seen the same thing in the BHK 5 model in newborn nude rats inoculated with small 6 numbers of adeno-5 E1A expressing cells that I 7 wouldn't h ave intuitively thought would make anything 8 once they sat around in this nude mouse to the time 9 that they were an adult. But they can have lungs full 10 of metastases if you let them sit for a very long 11 time, and the cells are put in under the cover of 12 immunodeficiency. 13 So again, this relative question about how 14 tumorigenic a cell line is is interesting, but whether 15 it has anything to do with the risk of that cell, I 16 think, is another question. 17 DR. FRIED: Yes? 18 AUDIENCE PARTICIPANT: If I may suggest, 19 you still haven't addressed that question. That's 20 what we want you to address. 21 DR. COOK: Well, I want to go back to your 22 comment, and that is, if we wanted to do number 2, 23 which is to ask about the transfer of DNA and how 24 could that be tested, because we're all talking about 25 religion here -- I mean, there should be an 226 1 experiment. 2 It seems like to put together your 3 question with Mike's suggestion earlier, has anybody 4 done a simple experiment like take DNA, like polyoma 5 DNA that's infectious, run it through a vaccine 6 production schedule of some sort, and find out whether 7 that DNA which is -- I mean, you need a positive 8 control for these experiments. 9 We could all create a lot of experiments 10 that would result in no tumors and say that the 11 residual DNA in this vaccine prep didn't do anything 12 bad. What about some DNA that we know can cause 13 neoplasia, run it through a vaccine preparation, put 14 it into a animal that's known to be sensitive? 15 If those animals don't get tumors when 16 inoculated with spiked preparations that have been 17 processed, then I think the idea of working backwards 18 from a positive control to what's likely to be a 19 negative experiment might be more logical. 20 AUDIENCE PARTICIPANT: Well, I think that 21 the question -- One of the questions that you're 22 asking is, in fact, something that we thought about a 23 little bit. That is the possibility of using -- to 24 get around this whole issue of the problem of DNA is 25 to use the same concept that was applied to the 227 1 business of viral clearance. 2 That is the concept of DNA clearance. In 3 other words, you would spike DNA with DNA containing 4 defined or known viral oncogenes, if you will, and 5 then show that the manufacturing process actually 6 clears that DNA to a certain -- with a certain 7 efficiency, and then using those numbers, you can then 8 make a prediction as to what else would be removed 9 during the same procedure. 10 So that's one way that you can think about 11 possibly getting around this business of testing. The 12 other thing that comes to mind when you're thinking 13 about risks associated with elements that are 14 components of vaccines is the fact, if you -- and 15 we're talking about tumorigenesis -- but the fact, if 16 you worry about what is in the tumor cell and the 17 possibility that the tumor cell might contain an 18 adventitious agent or, in face, is induced by an 19 adventitious agent -- and I think one of your 20 examples, that's a ras sarcoma cell -- you infect ras 21 sarcoma cells in chickens, and the chickens get ras 22 sarcomas, but they also get ras sarcoma virus. 23 So the fact that a cell line makes a tumor 24 could be obscuring -- The tumor itself could be caused 25 by some sort of oncogenic agents, especially if you're 228 1 dealing with an unknown cell line. 2 The tumor could be obscuring the presence 3 of an oncogenic agent. The oncogenic agent could be 4 there whether you get a tumor or not, simply because 5 the animal model you're using is not sensitive to 6 that. 7 So if you wanted to go out and begin to 8 poke at the issue of how to assess the transfer of a 9 neoplastic event or some type of infectious event, 10 then you need to compare the tumorigenicity of the 11 cell line with the possible tumorigenicity of the X 12 rac and things like that. 13 So there are models that -- things that 14 you need to do, but I think the question is how can 15 you do those quantitatively and at what point in time 16 do you need to begin to define where you go in that 17 direction; because this is not a small amount of work. 18 DR. FRIED: Okay, thank you very much. I 19 want to thank everybody here. I think we've overrun 20 our break. I feel I should summarize, but maybe we'll 21 leave it to the last session. 22 I think we're much more amenable that DNA 23 is not as big a hazard as we might have thought in the 24 beginning. 25 (Whereupon, the foregoing matter went off 229 1 the record at 3:00 p.m. and went back on the record at 2 3:19 p.m.) 3 DR. ROSENBERG: If everyone could take 4 their seats, we would like to get started. 5 We're going to be shifting gears only 6 slightly, I think, in this afternoon's session, and 7 then continuing in the same vein tomorrow morning to 8 a discussion that will highlight for you issues that 9 relate to retroviruses and ways in which retroviruses 10 are able to alter particularly the growth of cells and 11 ways in which retroviruses can change during their 12 replication in cells. 13 I want to very briefly introduce this by 14 showing a couple of slides, and these will just 15 highlight for you -- I think if we could have the 16 lights down, please -- some of the aspects that will 17 be coming up in the talks this afternoon and this 18 morning. 19 Certainly, you're all aware that 20 retroviruses, unlike some of the agents that have been 21 discussed already, have as a natural part of their 22 life cycle the ability or an obligate need to 23 integrate into the DNA of the host. As a consequence 24 of this, we appreciate that there are a number of 25 things that can occur. 230 1 These are illustrated here. Here we see 2 an insertion and an activation of an oncogene, leading 3 eventually to the development of a mammary tumor in 4 this mouse. 5 Other things can happen as well. While we 6 may be focusing on these events, I think it's 7 important to remember, these are not the only 8 consequences. Two other inactivations, one an 9 activation rather seen here, results in the hairless 10 phenotype in this mouse, and an insertion here results 11 in up regulation of amylase, which has been suggested 12 at least by some, including the people who made this 13 figure, perhaps to explain our great love for 14 starches, including some that were just consumed in 15 the hall. 16 So there's a diverse set of things that 17 can happen when retroviruses integrate. To illustrate 18 this in perhaps a more mechanistic fashion, because of 19 the cleverness, or at least to me the cleverness of 20 these agents, a number of ways that the integrated 21 virus, shown here, can impact on genes nearby, it's 22 illustrated. 23 Here we have a circumstance where we have 24 expression of the neighboring gene under the influence 25 of promoters here. 231 1 Here we have a circumstance where strong 2 enhancers contained within viral control elements are 3 again affecting the expression of a neighboring gene, 4 and we certainly appreciate, although the slide 5 illustrates that this gene looks very close, that this 6 can occur over reasonably long distance. 7 Here we have another circumstance where 8 viral control elements identical to these but located 9 at a different position in the viral genome again are 10 affecting expression of the gene. 11 So these insertions don't have to be 12 completely specific with respect to cellular genes in 13 order to impact upon them. In addition, in 14 circumstances where replication occurs, not only can 15 there be effects on gene expression. Here we have the 16 integrated provirus, just as in the earlier slide, and 17 a cellular gene. 18 In this case we have transcription, as I 19 showed before, mediated from the virus into the 20 cellular gene, creating a hybrid viral cellular 21 transcript, and subsequent, although all the steps are 22 not diagramed here, packaging of this transcript into 23 a virion, subsequent replication, and in this case 24 incorporation of part of the cellular sequence into 25 the virus, then having the capacity to be transmitted 232 1 to cells subsequently if this structure is contained 2 within a cell with a replicating virus. 3 Clearly, this is a very low frequency 4 event, but not one that occurs at such a low frequency 5 that we haven't been presented with enough examples of 6 these types of viruses to employ many retrovirologists 7 over very long periods of time. 8 Finally, another event that we will hear 9 about today involves the ability of retroviruses to 10 recombine, and this slide illustrates recombinations 11 that go on in the generation of a particular set of 12 pathogenic murine leukemia viruses. However, this 13 slide could apply to other situations with other types 14 of retroviruses as well. 15 Here we have acquisition of sequences in 16 this region from endogenous viruses, and -- hopefully 17 the colors show up, although I can't see them from 18 here -- changes also affecting the LTRs of this virus. 19 As a consequence of these, the virus 20 tropism changes, and confers to this virus a highly 21 tumorigenic phenotype. Other kinds of recombinations 22 can occur as well and, again, alter both host range 23 and tissue tropism. 24 This is an extremely important way in 25 which viruses can infect new hosts or new cells, again 233 1 leading to diverse consequences. 2 So I think, as we discuss all of these and 3 present model systems in which these are well 4 characterize, we'll hopefully set the stage for the 5 discussion that will follow these retroviral sessions. 6 With nothing further, I'd like to turn 7 this over to Sandy Ruscetti, who will present the 8 first talk. 9 DR. RUSCETTI: I'm going to move down 10 here, because I understand you can't see your slides 11 otherwise. 12 Thank you, Naomi. As Naomi mentioned, 13 when one uses neoplastic cells as substrates for 14 vaccine development, one can inadvertently get viral- 15 viral or viral-cellular interactions that could have 16 unknown biological consequences. 17 So the organizers here asked me today to 18 talk about a model system in the mouse, and that's the 19 generation of MCF retroviruses as a model for 20 understanding how viral-viral and viral-cellular 21 interactions can result in the generation of new 22 viruses that can have pathological consequences. 23 Now mouse cells are just loaded with 24 endogenous retroviruses of various subtypes. Some 25 strains have ecotropic MuLVs in small copy numbers 234 1 and, when these are expressed, they are involved in 2 the development of leukemia in these mice. 3 The AKR mouse is an example of such a 4 mouse, but all strains of mice contain many copies of 5 xenotropic, polytropic and modified polytropic MuLVs. 6 Now these viruses really don't cause any problems to 7 the mouse or the mouse cells, but when they are 8 expressed with ecotropic virus or if ecotropic virus 9 is injected into the mice, recombination can occur 10 with these endogenous sequences, particularly the 11 polytropic sequences, and this can result in the 12 development of a pathogenic virus. 13 These viruses are called MCF viruses, and 14 what I'm going to do today is to give you a background 15 about MCF viruses in the mice, how they're generated 16 and what their characteristics are, and give you a 17 little food for thought about whether we should worry 18 about these in using mouse cells as substrates for 19 vaccine development. 20 Now these are just some properties about 21 MCF viruses. MCF viruses are recombinant viruses 22 between an ecotropic virus and endogenous sequences 23 present in mouse DNA. 24 These viruses contain ons genes. So they 25 are derived from polytropic or modified polytropic 235 1 MuLV sequences, and they contain oftentimes LTR 2 sequences that are also derived from the endogenous 3 sequences in the mouse or certainly modified so that 4 they can be expressed at high levels in certain 5 tissues. 6 Now one of the properties of MCF viruses 7 that gave it its name is that they can induce unique 8 changes, and they were called foci, but not 9 transformed foci, in mink lung fibroblasts. But other 10 studies have shown that not all MCF viruses have this 11 property. 12 One of the interesting things about MCF 13 viruses, and another reason why we're interested in 14 them today, is that compared with ecotropic MuLVs, 15 these viruses have an extended host range which allows 16 them to replicate both in mouse cells and non-mouse 17 cells. This is because MCF viruses use a different 18 receptor than ecotropic viruses to enter cells. 19 Now when one looks at the host range of 20 MCF viruses compared to other classes of murine 21 leukemia viruses, one can see that it has a host range 22 that's a combination of them. Ecotropic MuLVs are 23 mouse-tropic. They can only infect mouse cells and 24 not non-mouse cells. 25 Xenotropic MuLVs cannot infect mouse 236 1 cells, but they are very efficient in infecting non- 2 mouse cells, including human cells. Polytropic or MCF 3 viruses can infect both mouse and non-mouse cells. 4 Now it's still unclear how infectious MCF 5 viruses are for human cells, and I thought it was 6 worth bringing that up at this conference. Working 7 with the prototype MCF virus, AKR MCF 247, some 8 studies have shown that certain human cell lines can 9 be efficiently infected with this virus. 10 Other studies show that they can't, and I 11 think, importantly, four different primary human cell 12 cultures could not be infected with AKR MCF virus. So 13 it's unclear just how infectious MCF viruses are for 14 human cells, and they are certainly less infectious 15 than the xenotropic or even amphatropic MuLVs. 16 The prototype model for the generation of 17 MCF viruses is the AKR mouse. Now these mice develop 18 thymic lymphomas at a high frequency when they are 19 about six to 12 months of age. 20 It was shown early on that these mice 21 contain in their genome an endogenous ecotropic MuLV. 22 It's called AKV or several copies of AKV, from which 23 infectious ecotropic replicating MuLV is produced at 24 birth. 25 Now it's always been a mystery why these 237 1 animals who produce lots of this virus do not get 2 leukemia for six to 12 months, especially since early 3 studies indicated that AKV was certainly important for 4 the generation of this leukemia; because you could 5 take AKV and express it in a low leukemic strain, and 6 they would become a high leukemic strain. 7 Well, in 1977 Jan Hartley discovered a new 8 virus from these tissues of these diseased mice that 9 provided an answer. What she and her colleagues were 10 able to do were isolate this new class of MCF viruses 11 from the pre-leukemic and the leukemic thymuses. 12 Further studies showed that the MCF 13 viruses were recombinants between the AKV and 14 endogenous MuLV sequences, and replication of both the 15 AKV ecotropic and the MCF viruses in these thymuses 16 were what were responsible for the development of the 17 leukemia that occurred when the animals were six to 12 18 months of age. 19 Now how are these MCF viruses generated? 20 Studies have shown that there are three endogenous 21 MuLVs that are involved in the generation of AKR MCF 22 viruses: AKV, which is the ecotropic virus I just 23 mentioned; VXV1, which is an endogenous xenotropic 24 MuLV; and various copies of polytropic and modified 25 polytropic MuLV sequences. 238 1 The generation of MCF virus in the AKR 2 mouse is thought to involve several different steps. 3 First of all, the AKV is thought to recombine with the 4 om genes of one of these endogenous polytropic MuLVs, 5 generating a new virus which looks basically like AKV, 6 but it has new om gene sequences that are derived from 7 the polytropic virus. 8 Then the second even occurs. This 9 recombinant recombines with the endogenous xenotropic 10 virus to generate a new LTR sequence. 11 Now it's not clear which of these events 12 occur first, but both of them apparently need to 13 occur. Finally, alterations appear to occur in the 14 LTR sequences that the viruses acquired from the 15 xenotropic MuLV, which often involve the duplication 16 of the enhancer sequence, obviously making this now a 17 pathogenic virus in the AKR mouse. 18 Now the AKR mouse is not the only mouse in 19 which MCF viruses are generated. You can generate MCF 20 viruses in low leukemic strains of mice if you inject 21 them with ecotropic MuLV. Two good examples of that 22 are Frend and Moloney MuLV induced leukemia. 23 If you take newborn mice and inject them 24 with Frend MuLV -- this is an ecotropic MuLV -- you 25 get the generation of MCF viruses, and these animals 239 1 develop erythroleukemia several weeks later. 2 If you take Moloney MuLV and inject those 3 into newborn mice, you also get the generation of MCF 4 viruses, but these animals now develop thymic 5 lymphoma. A lot of studies have been done with these 6 two systems. 7 It's a little bit easier to do studies 8 with the input virus, because one can inject that into 9 the mice when it's not just relying on AKV sequences 10 in the AKR mouse, and it was shown -- several things 11 that I think are important. 12 The type of disease induced is determined 13 by the input LTR. So if you have a Frend MuLV LTR, 14 you'll get erythroleukemia. If you have a Moloney, 15 you'll get thymic lymphoma. This was determined by 16 making chimeras between the two viruses. 17 It's thought that perhaps the MuLV -- 18 Frend MuLV LTR may replicate better in erythroid 19 target cells, and the Moloney in T-cells, and that may 20 be why you get that particular type of tissue tropism. 21 But also, too -- Actually, Leonard Evans who is here 22 in the audience -- he showed that Frend and Moloney 23 ecotropic viruses actually recombine with different 24 endogenous polytropic om gene sequences to generate 25 MCF viruses. 240 1 So the particular sequences that they 2 recombine with may determine the type of disease 3 that's induced. 4 Another thing: Adult mice are resistant 5 to leukemia development by injecting these viruses. 6 You need to inject newborn mice, and it's thought that 7 the reason you need to do this is because adult mice 8 may either have fewer target cells for the virus or 9 may have an efficient immune response that allows you 10 to clear either the input virus or the generated MCF 11 virus. 12 We showed a number of years ago with Frend 13 MCF that you could treat adult mice to increase the 14 target cells or decrease the immune response, and they 15 would be susceptible as adults. 16 Finally, there are strains of mice that 17 are resistant to the development of leukemia induced 18 by these MuLVs, even if you inject them into newborn 19 mice, and these strains of mice do not replicate MCF 20 viruses; because they contain genes that suggest they 21 are MCF genes which prevent the replication of the MCF 22 viruses. 23 This makes the case even stronger that 24 efficient generation and replication of MCF viruses in 25 these mice are responsible and are necessary for the 241 1 development of the leukemias that follow. 2 Now the generation of MCF viruses after 3 injection of Frend or Moloney is a little less 4 complicated than in AKR mice. Just showing this same 5 slide with AKR, with this being the ecotropic MuLVs. 6 The ecotropic MuLV does recombine with 7 endogenous polytropic MuLV sequences to generate a new 8 om gene. Again, particular polytropic sequences they 9 recombine with are different, but usually the viruses 10 do not need to acquire a new LTR. 11 The ecotropic Frend and Moloney LTR seems 12 to work fairly well to allow this virus to replicate 13 in the target cell. Now sometimes there have been 14 observed some changes in the enhancer region of the 15 LTR, and these may be to allow the particular MCF to 16 replicate to high levels in the target tissue. 17 One point I wanted to make was that both 18 in AKR mice and in MuLV infected mice, one may 19 generate lots of different types of MCF viruses, and 20 maybe only a certain subset of those MCF viruses are 21 actually pathogenic, ones that have acquired all the 22 right changes to cause disease. 23 Now these studies have all been done where 24 MCF viruses are generated in situ in the mouth. Now 25 what happens if you take the MCF virus from the 242 1 diseased tissue, purify it away from the ecotropic 2 virus, and inject it back into mice? Nothing really 3 happens. 4 That's because MCF viruses need to be 5 pseudotyped with ecotropic MuLV in order to cause 6 disease when you take them out and try to put them 7 back in. For example, AKR MCF virus will accelerate 8 leukemia in AKR mice, but it will not cause leukemia 9 in the low leukemic NIH mouse unless these mice 10 express an AKV ecotropic virus as a transgene. 11 Okay. Frend MCF virus will not cause 12 leukemia in mice unless it's pseudotyped with 13 ecotropic MuLV. Also I put this in here, because we 14 had showed a number of years ago that, if we infected 15 a packaging cell line inside two lines with Frend MCF 16 virus, we could get a virus preparation that would 17 cause leukemia, erythroleukemia in NIH Swiss mice. 18 So this suggested that the ecotropic MuLV 19 may not have to replicate in the mouse once it gets 20 the MCF there, and we actually showed that we did not 21 regenerate an infectious ecotropic MuLV from these 22 studies. 23 Now it's thought that pseudotyping may 24 prevent the inactivation of MCF viruses by maybe serum 25 factors complement or it may allow the virus to bypass 243 1 an effective immune response that might be against the 2 MCF virus. 3 So how to MCF viruses cause leukemia? 4 It's not completely clear, but there have been several 5 hypotheses, and it may be actually a combination of 6 several of these. 7 It's possible that the MCF viral envelope 8 proteins may interact with cytokine receptors on 9 hematopoietic cells and alter their growth and 10 differentiation. Also integration of MCF viral genome 11 into the host DNA may lead to activation of certain 12 proto oncogenes whose inappropriate expression in 13 hematopoietic cells may also lead to the alterations 14 in their growth and differentiation. 15 Also, since MCF viruses use a different 16 receptor than ecotropic viruses, and since both 17 viruses are present in the diseased animal, that would 18 allow two MuLVs to get into the same cell and increase 19 the changes of an activation event due to viral 20 integration. 21 So the data with studies of MCF viruses in 22 mice indicates that you can generate in mice 23 recombinant MCF viruses, and that these are really 24 crucial for the development of leukemia in these mice. 25 But the question that's more relevant for 244 1 this conference is can these viruses be generated in 2 vitro, and if they can, is that a problem when using 3 mouse cells as substrates for the development of 4 vaccines? 5 Also, even if these could be generated, 6 could they cause a biological effect if they could 7 actually replicate in humans? I think this will be 8 perhaps addressed by some of the other talks in this 9 session, and I think certainly will be a topic of 10 discussion when we have our panel tomorrow, late 11 tomorrow morning. 12 So if you have any questions in general 13 about MCF viruses and how they are generated and how 14 they are used as a model for viral-viral interaction, 15 I'll be happy to answer those for you. 16 (APPLAUSE.) 17 DR. EVANS: I'm Leonard Evans from Rocky 18 Mountain Labs. 19 I wanted to clarify about the 20 recombination with distinct sequences. Those studies 21 were done where the viruses were initially selected on 22 mink cells. 23 We found subsequently that the viruses -- 24 many more viruses that could be found on mouse cells 25 than on mink cells, largely because they were 245 1 serotyped. When we analyzed the larger population of 2 viruses, we found there was a significant overlap. 3 Frend does not specifically recombine with something 4 different than Moloney. 5 There are differences in the preponderance 6 of the viruses, but not in the identify of those 7 viruses. 8 The second point I wanted to make is that 9 we can inject Frend MCF into NFS mice, and they do 10 develop erythroleukemia. 11 DR. RUSCETTI: In the absence of eco? 12 DR. EVANS: Yes. Markedly cloned and 13 transfected. 14 DR. RUSCETTI: Well, I know one of the 15 questions that always arises: When you purify an MCF 16 virus out of a diseased tissue, are you purifying the 17 pathogenic one or not? 18 Possibly, some of the reasons why you 19 might not be able to induce disease with a given MCF 20 virus is because you haven't actually cloned the 21 pathogenic MCF virus. But I know our studies with our 22 MCF virus, we were never able by itself to, but I 23 agree, that doesn't mean that you always have to have 24 that. That's generally a problem, though, with MCF 25 viruses replicating once you take them out of the 246 1 animal. 2 DR. EVANS: Yes. It's also been shown 3 that there's a tremendous range of infectivities for 4 different heterologous cell lines like mink rhesus 5 mouse. It can be vary over several orders of 6 magnitude in terms of the ratio, the relative 7 infectivity, and there's such a heterogeneity of 8 polytropic viruses that to generalize that they do not 9 infect human cells is probably a leap, you know. 10 They don't readily affect them. Maybe 11 someone -- 12 DR. RUSCETTI: All right. Now I wasn't 13 trying to say that. 14 DR. EVANS: No, no, I know. 15 DR. RUSCETTI: I just think that people 16 might have an impression that they are very infectious 17 for human cells, and that may not be the case. 18 DR. EVANS: Some of them might be. 19 DR. LINIAL: Maxine Linial, Hutchinson 20 Cancer Center. 21 So there are a whole bunch of other 22 endogenous retro elements in mouse cells like DL-30s 23 which, at least in one defective virus, have been 24 shown to be recombined, and there are mouse endogenous 25 viruses.d 247 1 So has anyone looked in replication 2 competent MCFs, maybe not the ones that grow out but 3 in the pool, to see if they have acquired other 4 endogenous retroviral sequences? 5 DR. RUSCETTI: Not that I'm aware of. I 6 don't know. Somebody else might be able to address 7 that, because I'm really not working in that field. 8 DR. ONIONS: David Onions. Hi, Sandra. 9 I didn't quite understand your last point 10 when you were suggesting that the -- the possibility 11 that the pseudotyping with ecotropic envelope was 12 necessary to evade the immune response or perhaps to 13 avoid a complement factor. 14 I would assume, as in other pseudotyping 15 situations, the envelopes are chimeric with both 16 ecotropic and MCF type sequences. So wouldn't that 17 make them susceptible to complement, and wouldn't it 18 also make them susceptible to the immune response? 19 DR. RUSCETTI: Oh, no. It would just be 20 the MCF genome that's packaged in an ecotropic coat 21 would be able to get past those inactivating factors 22 in order to infect the cells that it needed to infect 23 to cause disease. 24 DR. ONIONS: So you think that 25 pseudotyping is absolute, as in there's a complete 248 1 ecotropic coat there? 2 DR. RUSCETTI: Right. Well, as Pug 3 mentioned, that may not be absolute, but certainly it 4 appears to be the common event that occurs, that you 5 do get pseudotyping, but it's actually -- It's a 6 complete ecotropic coat. It's not a chimera between 7 the eco and the MCF. 8 DR. ROSENBERG: Thank you. The next talk 9 will be by Clive Patience on MLV packaging systems. 10 DR. PATIENCE: Okay. Firstly, I'd like 11 to thank the organizers for inviting me to be here to 12 present some data. If I could have the first slide 13 and the lights down a bit, it would be great. 14 The work I'll be presenting was primarily 15 performed in Robin Weiss's laboratory in cancer 16 research in London. However, recently -- it does not 17 account for my jetlag, which I don't have -- I've 18 moved to BioTransplant, a xenotransplantation company 19 in Boston. 20 Now when I was invited I made the 21 assumption that I'd be talking to a diverse audience, 22 ranging from those which know very little about 23 retroviruses up to those which are astral heights of 24 editors of retroviral textbooks that should remain 25 nameless. 249 1 So what I'll try and do is give you a 2 story which I hope everyone will be able to follow, 3 and it's going to deal with the murine leukemia based 4 virus packaging systems, gene therapy models, and 5 basically estimation of potential recombination. 6 Okay. For the uninitiated, this is the 7 basic system, which I shall try and get in focus for 8 you somewhat. I think that's as good as we're going 9 to do. Basically, there's three components. 10 You have the packaging cell line at the 11 top here. The function of the packaging cell line is 12 to create the virus. There's generally three 13 components, the gag and pol expression plasmid which 14 make the body of the virus, and the envelope which 15 makes the outer covering of the virus which determines 16 what sort of cells you will ultimately infect. 17 Into that virus, rather than packaging the 18 normal virus genome, you put in your therapeutic 19 vector. You make your virus, and that virus then is 20 used to infect your target recipient cells. Where the 21 RNAs converts to DNA, integrate it and your gene is 22 expressed. 23 A little bit more detail on the actual 24 structure of a packaging cell line: Basically, we use 25 the immortalized cell line HT1080 for the particular 250 1 cell line I'm going to describe. 2 We compared this human cell line in 3 comparison to a 3T3 based murine based system. It's 4 called FLY. It's got nothing to do with little 5 insects. It's actually the initials of the 6 discoverers, or I should say the inventors. 7 Okay. Firstly, you put in your gag pol 8 expression plasmid, which basically produced the core 9 of the virus. You then cotransfect in your envelope 10 plasmid. So you can then make a complete virus 11 particle with all the machinery required for 12 retroviral integration. 13 You then insert your vector of choice, and 14 basically that can be a therapeutic vector or in the 15 laboratory a reporter vector. The particular plasmids 16 which I'm going to discuss here, you can see, is 17 diverged somewhat away from its wild type parent. 18 It has a Moloney murine leukemic virus, 5 19 prime LTR, as well as Moloney gag and pol. It then 20 has basically a gap through here after the stop code 21 on drug selection. So basically you get a read- 22 through here. 23 About five percent of ribosomes will read 24 through this stop code and go on to transcribe your 25 drug resistance gene. So if you select the drug 251 1 resistance, you should, and indeed you do, select for 2 very high expressors of the gag/pol proteins and, 3 therefore, high levels of virus. 4 Top down then we have a polydentylation 5 signal which I believe was SB40, if I remember right. 6 The envelope comes out very similar, 5 prime 7 retroviral LTR, no packaging signal. This region here 8 determines whether the RNA construct will actually be 9 packaged into virus particles. The envelope is 10 expressed up here, and again SV40, and your reporter 11 vector -- the important part here is the presence of 12 a packaging signal. So it should be efficiently 13 packaged into virus particles. 14 Okay. We were trying to get some sort of 15 feeling for the degree of recombination that could 16 occur, and our logic was that it would be greatly 17 increased if there was any generation of replication 18 competent retroviruses in the packaging system. 19 Now the recombination could either come 20 from the -- The viruses -- sorry -- could either come 21 from a combination of DNA present in the packaging 22 cell lines, and you may have noticed that all of the 23 plasmid constructs have quite a number of regions 24 deleted or mutated. Very elaborate cell lines have 25 been produced now over many years. 252 1 The alternative is that the recombination 2 events could actually occur during reverse 3 transcription of the RNA in the virus particle. For 4 that to occur you need expression of the unwanted RNAs 5 in the packaging cell line, and that in turn will be 6 enhanced. This is significant similarity between the 7 vector and the endogenous sequences. 8 Some bad news to you people. About one 9 percent of your genome is retroviral. Yes, we're 10 derived from apes, but we're also derived from 11 retroviruses. 12 Basically, so about one percent of your 13 genome is retroviral. It probably represents the 14 ancient remains of past retrovirus infections where 15 germ line cells have become infected. Over the 16 millions of years, most of these copies of the various 17 viruses, which range number from one up to thousand, 18 have become deleted and mutated. So they can no 19 longer encode for infectious virus or, in most cases, 20 even proteins. 21 Of particular interest or those which have 22 received particular attention are the HERV-H or RTVL-H 23 family of viruses. There's approximately a thousand 24 copies of these in our genome, about 200 full length, 25 about 800 deleted versions, very highly transcribed in 253 1 general terms. 2 HERB-K, another family, has been the focus 3 of a lot of attention, primarily because it has pretty 4 well conserved open reading frames. It still has the 5 ability to encode envelope proteins, gag proteins, pol 6 proteins, and there's been a lot of effort, people 7 trying to grow this guy and correlate it with disease, 8 etcetera, with not too much success compared to the 9 amount of effort which has been put in. 10 So this is the basic scenario we were 11 looking at. We were looking at the packaging cell 12 line, which was human. Normal situation is that you 13 would like for your therapeutic transcript to be 14 packaged into the virus. We were looking to try and 15 determine whether the human endogenous retroviral 16 transcripts could also be co-packaged and get into 17 your virus particle. 18 So the first scenario or the first 19 requirement is that you have HERV expression in your 20 cell lines. Now we compared two cell lines, the old 21 and inverted comers M-12 and murine 3T3 based system 22 and the human based system which we've developed in 23 the laboratory. 24 This is basically an RT PCR. For those of 25 you who are not familiar with PCR, basically a bright 254 1 band means a positive. A lack of band is a negative. 2 Therapeutic -- so that what we were looking at was 3 actually reported back to LacZ, and you can see there 4 it's expressed in both the human and the murine cell 5 line. 6 We performed a degenerate PCR approach, 7 basically using PCR primers designed to conserve 8 regions in the virus and try and pull out as much as 9 we possibly could and identify as many different 10 sequences as we could. 11 In the human cell line we managed to 12 identify sequences which showed similarity to the B 13 and B-type retroviruses as well as the RTDL-H family, 14 which I have mentioned earlier. 15 In the murine cell line we again managed 16 to pull out the D type sequences and also another 17 retroviral sequences called BL-30 which is an 18 endogenous retroelement of mice which has been shown 19 to be packaged in murine cell lines, not by myself 20 but by Damian Purcell and others maybe -- and others. 21 We basic scenario we then went to do was 22 we had confirmed RNA expression in the cells. We then 23 wanted to try and identify whether this got as far as 24 the virus particles. And if it gets into particles, 25 it may be available for recombination events. 255 1 So we took these from the various 2 packaging cell lines and spun it through a sucrose 3 density gradient to purify and concentrate the 4 particles with respect to their density. What we 5 could then do was analyze the various fractions where 6 I've put up the enhanced RT assays or PERT assays to 7 identify the fractions with the highest virus content, 8 and then also examine exactly those fractions for the 9 presence of unwanted retroviral transcripts. 10 This is a typical section across a 11 gradient where you see banding of virus particles at 12 a density appropriate for mature retroviruses. 13 Okay. If we look at the old system, to 14 start off with, the RT assay here identifies a 15 concentration of particles, at one point 174 16 approximately grams per mil, which is appropriate for 17 virus particles. We see, as expected, the reporter 18 transcript as it possesses a packaging signal co- 19 purifying or co-concentrating in the same fraction. 20 What we also see is significant levels of 21 the endogenous retroviral element packaging in the 22 particles. 23 When we looked at the human system, again 24 we get a RT fraction, positive fraction, here with 25 high levels of beta galactoside as reporter vector. 256 1 However, in comparison to the murine system we don't 2 detect any packaging of endogenous sequences, RTVL-H 3 family, BLD type sequences or use in pan-retroviral 4 sequences. 5 We do see a faint band or found a faint 6 band when we were looking with C-type primers, and 7 I'll come back to the identity of what this product 8 was a little bit later. 9 So this phase basically identified RNA 10 expression in the cells, and we had in some cases 11 shown that some of the endogenous retroviral sequences 12 could be packaged in particles. 13 The next thing we really had to do was try 14 and put the numbers on this. Really, a plus and a 15 minus wasn't very informative. So the methodology 16 which we undertook was some in vitro transcription 17 whereby we cloned some of the PCR products' retroviral 18 sequences, cleaved before your 5 prime end, 19 effectively your ATG, if it were there, and 20 synthesized RNA from the transcripts. 21 We could then use this RNA to actually 22 determine accurately the sensitivity of our RT PCRs, 23 and that would allow us to eventually estimate the 24 concentration of RNA in the virus particles. 25 Shown in this slide here is the 257 1 sensitivities that we found for our different PCRs, 2 and it was really -- We were glad we did this, because 3 we got a huge range of sensitivities down from 4 approximately 1 femptogram reaction way up to in the 5 500+ femptograms a quad in the RT PCR to get a 6 positive signal. 7 For the reporter vector, which was the 8 least sensitive, it didn't appear to be due to 9 primers. We varied primers quite considerably and 10 didn't really see any improvement in the 11 sensitivities. It was probably due to the actual 12 sequence that you're trying to amplify any reactions. 13 So we then basically determined the 14 sensitivities of the various PCRs, whether it be 15 through reporter or for the different families of 16 endogenous sequences. 17 What we then went on to do was to perform 18 dilutions of the peak fractions. Basically, once we 19 had the sensitivities, we could combine these with the 20 dilution series of the fractions to determine the RNA 21 concentration. 22 If we look at the murine system at the top 23 here, the beta galactosidase, twofold dilutions, 24 looked to be in very similar concentration to the VL- 25 30 endogenous retroviral elements. 258 1 In the human based system we saw 2 significantly more beta galactosidase present in the 3 fraction, which is reflected in a higher titer of 4 virus, and relatively little C-type product. 5 So basically, that allowed us to estimate 6 the ratio of the desired therapeutic or recorder 7 transcript in comparison to the unwanted endogenous 8 retroviral transcripts in the cells. 9 The peak fraction of the murine based 10 system contained 6.3 times 103 femptograms and nearly 11 900 femptograms of the endogenous retroviral sequence. 12 If you do the mathematics, convert these to 13 molarities, that meant, basically speaking, for every 14 seven virus particles which contain the LacZ genome, 15 which you want, you have one virus particle which 16 contains the endogenous retroviral sequence. 17 Obviously, that means for every 18 integration, every cell that you treat, one in seven 19 is going to be the wrong molecule being integrated 20 into the genome. In fact, it's slightly more 21 complicated as the retro-particle actually carries two 22 copies of viral RNA, but the mathematics is still 23 basically the same. 24 When we looked at the human FLY packaging 25 system, you see we saw much, as I had mentioned 259 1 earlier, higher level of reporter vector and 2 basically, looking at the endogenous retroviral 3 sequences, these were much cleaner than the seven to 4 one which we saw with the murine system. 5 For the C type product which we saw, we 6 had calculated that about one in 19,000 virus 7 particles would contain this retroviral sequence. We 8 went on to identify what this was. 9 We cloned and sequenced it and, actually, 10 it turned out to be on sequencing identical to the 11 gag/pol expression plasmid which we were using to 12 actually create the particles. So we inserted that as 13 representing basically the packaging of random 14 cellular transcript. There's no real way of getting 15 around this. You will get random RNAs incorporated 16 into viral particles at a certain, albeit low, degree. 17 What I'd like to -- Oh, sorry, I should 18 state that, since performing this data, we've gone on 19 to try and address the potential problem with this 20 approach, whereby we're revising our sequences to 21 known retroviral sequence. 22 What we've done is to actually perform a 23 random PCR on CDNAs made from the virus particles, to 24 try and identify unknown viral families which may be 25 being packaged and other viral sequences. 260 1 I don't have any slides, but a summary of 2 the results would be that in the murine based system 3 the predominant sequence which we identified was VL- 4 30, as you might expect, because we could even detect 5 it with this system. 6 In the human based system, again we 7 identified the gag/pol expression plasmid and a whole 8 range of various, apparently random, in many cases 9 unidentified RNA transcripts, very few of which had 10 open reading frames, and most of which were quite 11 small in size. So the significance of that data I'd 12 have to come back to in perhaps a year's time or 13 whenever. 14 Regards another potential risk of 15 retroviral gene therapy, we tried to address or what 16 we wanted to address was whether, once you had treated 17 a cell, if the endogenous retroviruses in that cell 18 were expressing gag proteins, whether they could pick 19 up the RNA transcripts from your integrated 20 therapeutic vector, package them, and potentially 21 immobilize them and bounce them around the genome. 22 So this would be a disadvantage due to 23 insertional mutagenesis, and there are some mechanisms 24 which we had talked about earlier. 25 In our bodies there are very few cells 261 1 which express significant levels of retroviral 2 particles. One major exception is the placenta which 3 is absolutely heaving with BD type virus particles 4 butting from the few cyncitial trophoblast layer into 5 the cytotrophoblast which forms basically the barrier 6 between the mother and the fetus. 7 There are all of these particles as yet to 8 be determined. It's a lot of research going into 9 that. So the cell line which we analyzed to try and 10 assess this case was one of the few cell lines where 11 we do get lots of virus production, HERV-K based BD 12 type virus. 13 What we did was to look for the -- There's 14 a Gh cell line for those that would like the 15 information -- again, performed sucrose gradients on 16 the virus particles. We could see indeed endogenous 17 retrovirus particles being produced of the B and D 18 type sequence, which were banding at densities 19 appropriate for BD type viruses. 20 Interestingly, in those cells we could 21 also detect the packaging of RTVL-H sequences, another 22 endogenous family of sequences. The significance of 23 that we're really not sure, and haven't taken this 24 work any further at the moment. But the reassuring 25 point was the murine leukemia based virus vectors 262 1 which we had introduced into these cells did not 2 appear to be packaged into these particles. 3 Absolutely blank. So that was somewhat reassuring. 4 So in conclusion, I think we have to, and 5 will always have to, put up with the problem that gene 6 therapy packaging cell lines, whether they be derived 7 from human or murine systems, dog, you name it, will 8 express endogenous retroviral sequences which can 9 potentially interact with your gene therapy vectors. 10 The FLY line, human baseline, and probably 11 other packaging cell lines seem to produce virus 12 particles which are significantly less contaminated 13 with endogenous retroviral sequences and do murine 14 cells. 15 As a result, these are probably less 16 likely to produce replication competent recombinant 17 retroviruses which could be a problem in such 18 therapies and, encouragingly, again human endogenous 19 retroviral gag proteins, core proteins, only interact 20 weakly, if at all, with the NOV based viral RNA 21 transcripts. 22 Once again, thank you for the invitation 23 to speak, and I'll welcome any questions. 24 (APPLAUSE.) 25 DR. COFFIN: It's an easy one. I'm John 263 1 Coffin from Tufts. It's also been reported that MLV 2 based vectors produced on certain packaging cell lines 3 ex situ contain very high levels of an MLV related 4 RNA, from Jim Cunningham's lab some years ago, for 5 example. Did you see any of those -- any of that? 6 Did you look for that? 7 DR. PATIENCE: We didn't look, actually. 8 It should be done, actually. 9 DR. KAPPES: John Kappes from UAB. You 10 said you identified that the gag/pol transcript was 11 derived from the plasmid you transfected. Through 12 your sequencing, were you able to gain any information 13 that helped you understand how it recombined? 14 DR. PATIENCE: I don't think it was a 15 recombination at all. I think it's basically a 16 nonspecific packaging of RNA transcripts, which you do 17 see in retroviruses. You can pick up viral sequences. 18 I don't think it was recombination there. 19 DR. KAPPES: If you looked in infected 20 cells for evidence of recombination -- Have you looked 21 in infected cells for evidence of recombination? 22 DR. PATIENCE: Oh, yes, sure. We've 23 looked for both the transfer of the gag/pol construct 24 and rescue of -- Basically, we targeted cells with 25 envelope only, introduced our high titer virus, and 264 1 looked for rescue of a reporter vector by the 2 envelope, which we had already put in, and the 3 gag/pol. 4 We could detect very, very low levels of 5 recombinant viruses -- very, very low levels. 6 DR. PURCELL: Damian Purcell. With the 7 endogenous viruses that you've detected packaged, the 8 human elements, have you been able to establish 9 transmission and expression for those endogenous 10 elements? 11 DR. PATIENCE: We haven't seen any co- 12 packaging and, therefore, we've been unable to detect 13 any transmission. We did, unfortunately, do the 14 experiment back to front where we infected cells and 15 then looked for human transcripts, and couldn't see 16 any, which I guess again is even more reassuring, but 17 nothing is getting across. 18 DR. ROSENBERG: Thank you. The next talk 19 is by Leonard Evans on, again, mixed retrovirus 20 infection. 21 DR. EVANS: Okay. What I want to talk 22 about today are mixed retrovirus infections between 23 two different types of viruses that have already been 24 discussed. That's the polytropic viruses and 25 ecotropic viruses, and these are experiments that 265 1 we've done in mice. 2 I guess the take-home message from this is 3 that it's really going to be difficult to try to 4 anticipate what the effects of a mixed retrovirus 5 infection might bring. 6 The initial experiments I'm going to -- 7 How this actually started out was not look at mixed 8 retrovirus infections. What I wanted to look at was 9 recombination, and I wanted to look at the 10 recombination of a polytropic virus when it was 11 inoculated into a mouse and see whether or not it 12 underwent further recombination. 13 The reason we wanted to do this is because 14 we saw that the population in the individual mouse of 15 recombinant viruses that came up was plural. There 16 were a lot of different types of viruses, and we 17 wanted to know whether this was a sequential situation 18 or whether or not they were independent events. 19 Now the virus that we chose is a virus 20 called Fr98, and Fr98 was constructed in John Portiss' 21 lab or Rocky Mountain Labs, and it's constructed by 22 taking the envelope of the polytropic isolate and 23 sticking it into a backbone of Frend murine leukemia 24 virus, and in that way it was advantageous for this, 25 because we knew that there were no other alterations 266 1 in the genome other than the envelope and part of the 2 integrase gene. 3 The other reason we chose Fr98 is that it 4 replicates very well in mouse genes. So we thought 5 that -- I mean, if we're looking for recombination, we 6 definitely have to have something that's replicating. 7 The other had to do with the antigenic 8 properties of this virus. That's shown -- I'll 9 explain that on the basis of this slide. 10 Okay. This slide describes the monoclonal 11 antibodies that we used in these studies. There's 12 four monoclonal antibodies here. The first three of 13 these react specifically with polytropic viruses, and 14 the last one reacts with the Frend ecotropic virus, 15 which is the ecotropic virus we used in these studies. 16 The first two antibodies here, hybridoma 17 7 and 516, define two antigenic subclasses in these 18 viruses. Almost every virus that we ever looked at, 19 including our isolates and other people's isolates -- 20 virtually all of them react with one of the two of 21 these antibodies. 22 They're mutually exclusive, and they map 23 to the same position on the Su proteins and alternate 24 amino acids. We've also looked at endogenous 25 sequences, many of them, not all of them, and all of 267 1 the endogenous sequences, including the modified and 2 the regular polytropic viruses, correspond to either 3 this one or this one in terms of the amino acid. 4 Now what was unique about Fr98 is that it 5 was a virus that didn't react with either one of these 6 antibodies, although it did react with 514 which 7 reacts with all polytropic MuLVs. 8 S for our purposes, it was really a nice 9 virus, because any emerging viruses we thought that 10 would arise by recombination would either react with 11 hybridoma 7 or 516. So we would be able to watch 12 this. 13 Now another property of Fr98 is that it's 14 a neuropathogenic virus. It causes a neurological 15 disease in IRW mice, but at the doses we were going to 16 use we didn't think we would probably see a disease 17 for about 30 days. We weren't using IRW mice. We 18 were using NFS mice. 19 When we inoculated Fr98 into NFS mice, we 20 were able to actually follow the disease or follow the 21 mice for the generation of new polytropic viruses that 22 react with hybridoma 7 or 516 for about three or four 23 months, because it turned out that these mice didn't 24 get neurological disease with Fr98. They developed 25 erythroleukemia with a latency somewhat longer than 268 1 you see with the ecotropic virus. 2 During the course of that time, we didn't 3 see anything that suggested any emerging viruses with 4 hybridoma 7 or 516. Everything was reactive with the 5 514, which is what the Fr98 is reactive with. 6 It occurred to us that maybe what was 7 happening is, because we're putting in a polytropic 8 virus, even if we generated new polytropic viruses, 9 because of the vicinity that we're apt to be 10 generating this in and it would be a high level of 11 polytropic infection, that there some interference so 12 that we would never see the virus, because it was 13 never, ever able to spread. 14 Then it further occurred to us that, if 15 this actually was happening, then maybe if we 16 inoculated an ecotropic virus with the polytropic 17 virus, that we could suppress the development -- the 18 de novo development of new polytropic viruses. 19 So we co-inoculated Fr98 with Frend MuLV, 20 and that's the next slide. Okay. Now I expected 21 there might be -- You know, I hoped there might be a 22 little suppression. What I didn't expect was the 23 complete turnoff of any de novo polytropic viruses. 24 Normally, when you inoculate Frend MuLV, 25 you get -- It's predominantly hybridoma 7, but there's 269 1 some -- two or three percent that are -- of the 2 polytropic viruses that are reactive with 516. 3 We got about 60,000 here on the average of 4 12 days, and we got absolutely nothing, no hybridoma 5 7 and no 516. We have -- In fact, in any co- 6 inoculation experiments, we have never seen any of the 7 normal -- any hybridoma 7 or 516 polytropic viruses 8 coming out. 9 We did, however, see a huge amount of -- 10 This is the sera. We saw a huge amount of virus that 11 corresponded to Fr98 that's reactive to 514 only. 12 This is nearly 107, but there was a second observation 13 that was really striking, and that's showing on the 14 next slide. 15 We found that the mice developed -- The 16 reason those experiments only went for 12 days is 17 because that 12 days after, the mice are gone. We 18 found out that, when you inoculated Fr98 with fMuLV, 19 we got a very rapid development of neurological 20 disease, very similar to what's seen in IRW mouse 21 except for the onset. At this dose in IRW mouse, they 22 would normally come up about here. 23 The mice appear really normal for about 24 ten days, and when they're suckling, there didn't seem 25 to be any problem. Then from 11 to 15 days these mice 270 1 get sick. They become ataxic. They fall over. They 2 have tremors, and then they die. 3 It's so rapid that, I mean, if -- The 4 range here appears to be about a week, but in fact, 5 when the first mouse in a litter develops symptoms, 6 usually within a few hours all of the mice will 7 develop those symptoms, and within 12 hours they will 8 all be dead. 9 Okay. John Portiss looked at the brains 10 of these mice. He could find no spongiform 11 encephalopathy at all. There's no hemorrhagic 12 differences. John is a pathologist at our lab. So -- 13 which is very similar to what you see in IRW mice. In 14 fact, a higher levels of virus inoculum -- and this is 15 with Fr98 alone in IRW mice -- you can actually 16 shorten the time of incubation here to come very close 17 to this curve. 18 Well, we wanted to know what was going on 19 in these mice. So we had a hint from the fact that 20 the viremia was 107 at 12 days. So we decided to 21 follow the virus load in these mice, in mice that were 22 inoculated with Fr98 alone, with MuLV or with the 23 combination. 24 So we looked at both ecotropic, compared 25 the ecotropic with the co-inoculation, and the 271 1 polytropic virus load and the co-inoculation. Next 2 slide shows those experiments. 3 This is the ecotropic virus. This is off 4 course here a little bit, but basically with ecotropic 5 virus we could see no differences in the -- This is a 6 serum. This is the spleen, and this is the brain, 7 which are the tissues we looked at in these mice. 8 We could see no difference in the 9 ecotropic virus replication in these mice. It was the 10 same, whether or not it was inoculated by itself or 11 whether or not it was inoculated in combination with 12 the polytropic virus. 13 Completely different situation, though, 14 with the polytropic virus. If I can get the next 15 slide. Okay. Here in the serum there is about a two- 16 to-four hundredfold difference in the level of viremia 17 between the co-inoculated mouse in the red and the 18 mouse inoculated with simply Fr98. It's even more in 19 the spleen in terms of infectious centers. 20 These are cell free viruses. These are 21 infectious centers. A little less pronounced in the 22 brain, but still a ten-to-twenty-fold difference in 23 the level of infection in the brain. 24 So it appeared that the viruses were, 25 obviously, interacting in these mice, and one of the 272 1 things a person might suspect would be pseudotyping 2 between the viruses, even though they were inoculated 3 as separate -- well, as a mixture of stock, but they 4 were inoculated as separate viruses. 5 So could I have the next slide? We looked 6 at pseudotyping in these mice. There's a couple of 7 things I've got to point out about this, a couple of 8 striking things. 9 One is -- that I'll come back to in a 10 second -- the fact that the polytropic virus appeared 11 to be pseudotyped completely, almost through the whole 12 course of infection, and pseudotyped by the ecotropic 13 virus. These assays are done by comparing the titers 14 on 3T3 cells versus the titers of the virus on 3T3 15 cells that are infected with fMuLV. 16 So that if it's pseudotyped because of 17 viral interference, it won't score on the fMuLV 18 infected 3T3s, but it will score on the 3T3s that are 19 not infected, because both eco and poly will go into 20 those cells. 21 The first striking thing here is the 22 pattern of pseudotyping in the brain. The pattern of 23 pseudotyping changed abruptly right before the onset 24 of disease, and this is one of the few instances where 25 we see polytropic viruses that are not pseudotyped by 273 1 ectotropic virus, and the simplest explanation of 2 this, which we're currently trying to resolve, is that 3 the polytropic virus has spread to a new population of 4 cells, and either that population is not coinfected 5 with ecotropic virus or that population of cells has 6 a different property in that it does not promote 7 pseudotyping of the polytropic virus by the ecotropic 8 virus. 9 The other puzzling thing here is that -- 10 I can now extend these curves clear back to here for 11 all this -- is the level of pseudotyping of the 12 polytropic virus by the ecotropic virus -- the level 13 of pseudotyping is nearly 100 percent here, and at 14 this point in tim the level of infection is very, very 15 low. We're getting around 1000 titers. 16 So the very first viruses that we're 17 detecting in these mice are, in fact, pseudotyped. 18 Well, the only way you can get pseudotyping that I 19 know of is to have co-infected cells. So this 20 suggested then that we actually have a very small 21 population of cells that are infected by both viruses, 22 which is kind of remarkable, considering they both use 23 different receptors. 24 It occurred to me that maybe this was 25 because of the route of inoculation, because we were 274 1 mixing the viruses and sticking it in a compartment, 2 actually in the peritoneum of the mouse. So we 3 thought that possibly it might be that. 4 So what we did was to inoculate the 5 viruses subcu. as different stalks of viruses. So 6 Fr98 is inoculated in one spot subcutaneously, and 7 then like in the right haunch and the left shoulder, 8 the ecotropic virus was inoculated. 9 When we did that, we found the same thing, 10 complete -- almost complete pseudotyping by the time 11 we could get a good measure of the virus. 12 Well, that suggested to me that it might 13 be a mobile population of cells that was actually 14 being infected by these viruses. So -- and that at 15 some point in time a person should be able to find 16 cells that were not pseudotyped, so that you could 17 find cells that were not co-infected. 18 So we did a -- God, more experiments than 19 I want to know to look at this at very early times. 20 That's shown on the last slide. 21 We could, in fact, find a time during the 22 infection where the pseudotyping was minimal. This 23 computer moved this curve. So it's supposed to be 24 like this. But basically, at about a tenth of a 25 percent of there cells infected we could find free 275 1 polytropic viruses. 2 After that there was a significant level 3 of pseudotyping. So it looked to us then that there 4 must be a very small percentage of circulating cells 5 that are the initially infected cells by both of these 6 viruses. I guess one of the messages here that this 7 implies is that just because the virus's dose and 8 infection is a quite low dose, it does not preclude 9 fairly immediate interaction of the viruses. 10 I suppose I should say something about the 11 possibility of this interactions in humans, not these 12 interactions but the possibility of interactions of 13 different types of viruses. 14 There have been some examples of co- 15 infection in different types of viruses in humans. I 16 mean, it's well documented, HtLV and HIV, but there is 17 no good documentation, even though there was a lot of 18 suggestion earlier, that HtLV may accelerate HIV. 19 To my knowledge, looking at larger cohorts 20 there, there was no evidence of that, and also SiV and 21 StLV infections in an experimental situation didn't 22 show any synergism. But definitely, the tools for a 23 mixed retrovirus infection exists. 24 I mean, we talked a little bit yesterday 25 about cells -- or viruses that could infect humans. 276 1 There's a lot of viruses from a lot of different 2 sources that can infect heterologous cells, and in 3 fact, that may be the rule instead of the exception. 4 With the murine system, the only ones that 5 have never been, to my knowledge, demonstrated to 6 infect human cells are the ecotropic viruses. So we 7 know there's -- from the bovine, from the feline 8 system, obviously from the higher primates, there's 9 all kinds of viruses that can affect human cells. So 10 that's the one prerequisite for setting up such a 11 mixed retrovirus infection. 12 The other prerequisite is for these 13 viruses to interact. Now they're somewhat restricted 14 in the interaction in terms of packaging, but even 15 that is not a foolproof situation. C-type viruses 16 appear to package C-type genomes more easily, but 17 there is still -- Moloney has been demonstrated to 18 package the genome of HIV and also the unspliced -- or 19 the unspliced and the spliced messengers. 20 So it may not be a specific packaging, but 21 it definitely was found within the virion, although it 22 wasn't shown to be passed, and it wasn't tried in 23 those particular experiments. 24 In terms of utilizing glycoproteins and 25 incorporating surface proteins in different 277 1 retroviruses, they're very promiscuous. Again, you 2 have to search for places where it doesn't happen. It 3 doesn't happen very easily with HIV and MuLV, because 4 -- presumably because of the long cytoplasmid material 5 of HIV, although SIV will incorporate into -- which 6 also has a long cytoplasmid tail, will incorporate 7 into MuLV particles. 8 You also have to be concerned about 9 heterologous pseudotyping between different kinds of 10 viruses. It's been shown that fowl plague virus, 11 which is an influenza virus, can pseudotype MuLV. 12 Sendai virus, both proteins in Sendai virus can 13 pseudotype MuLVs. E. bola virus can, but that's 14 probably a moot point. 15 So I think that the potential for a mixed 16 retrovirus infection is definitely there. Whether or 17 not it can be easily avoided is a question. Thank 18 you. 19 (APPLAUSE.) 20 DR. ROSENBERG: Are there questions? 21 Thank you very much. 22 The next talk is by Damian Purcell. 23 DR. PURCELL: Can I have my first slide? 24 So we've been introduced to gene therapy and the 25 complications of co-packaging between multiple RNAs, 278 1 two RNAs into the retroviral particle. 2 What I want to talk to you about today is 3 some studies analyzing some of the early gene therapy 4 vectors and packaging cell lines that would -- using 5 what are now fairly unsophisticated reagents and the 6 vectors based on Moloney murine leukemia virus and the 7 N2 vector containing the packaging signal of Moloney 8 and the helper constructs based on the amphotropic 9 MuLV 4070A; and they have deletions in the packaging 10 signal, but these constructs have regions of hemology 11 where two recombination events can regenerate 12 replication competent retrovirus. 13 So the emergence of these replication 14 competent retroviruses was observed, and the safety of 15 -- potential risks from those replication competent 16 retroviruses in primates has been studied. I'll just 17 review quickly the studies that have been published. 18 There are essentially three of them. The 19 first one by Coronetta, et al. in 1990 examined a 20 replication competent retrovirus coming from sax 21 retroviral vector producer clone, which is essentially 22 the same as the one on the previous slide. 23 So in this study there was an intravenous 24 infusion 7 by 107 focus forming units of this 25 replication competent retrovirus or RCR, and though 279 1 that entity wasn't really well characterized 2 molecularly, presumably it had an ecotropic Moloney 3 MuLV gag/pol with the ampitrophic envelope. 4 Four rhesus macaques were infused 5 intravenously, three normals and one immunosuppressed 6 with prednisolone and cyclosporin, and in these 7 studies the virus was cleared in 15 minutes with, 8 obviously, no detectable pathology after 15 year 9 follow-up. 10 So even within this study it was 11 recognized that in these four animals infused with 12 replication competent retrovirus produced in a murine 13 cell line that probably a xeno antigen effect led to 14 virolysis by complement where the enzyme alpha 13 15 galactosol transferase adds the alpha gel residue onto 16 the glycoproteins of the virus particles, and 17 antibodies that are raised in humans, petrophile 18 antibodies, recognize this activate complement and 19 lead to a lysis of these murine cell derived viruses 20 very quickly. 21 So two ways to avoid this is, one, to 22 reduce the virus in human or primate cells where you 23 don't have the enzyme, and you also acquire passively 24 into the envelope of the viral particles molecules 25 that protect against the activation of complement such 280 1 as CD46, CD55 and CD59. 2 The other way is to introduce the 3 replication competent retrovirus in vitro into 4 autologous cells and then implant the infected cells. 5 This is what was done in this study where 6 by 107 6 fibroblasts from a rhesus macaque were infected with 7 the recombinant retrovirus, both subcu. and IP, and 8 also for good measure there was 3 by 108 focus forming 9 units of the prototypic amphotropic virus MuLV 4070A 10 injected IP into a single macaque. 11 This animal was immunosuppressed. The 12 animal was viremic for two days, and virus was 13 recovered from PBMC for several weeks. PCR could 14 detect the presence of the amphotropic MuLV envelope 15 up to 200 days, but there was no detectable pathology 16 after five years of follow-up. 17 In the second study, also published by 18 Coronetta et al. in '91, the following year, they took 19 the examination of the safety of these replication 20 competent retroviruses derived from the mouse and 21 murine system in primates one step further where they 22 examined the transplantation of replication competent 23 virus infected autologous bone marrow cells, in this 24 case four cynamologous macaques. 25 So in this protocol 1 by 106 focus forming 281 1 units of the replication competent retrovirus, again 2 from a sax derived vector but carrying the ADA gene 3 this time, were used to infect bone marrow, and they 4 used supernate directly from these producer cell 5 clones. 6 The animals for the purpose of bone marrow 7 transplantation received very strong immunosuppression 8 with 1000 rads of total body irradiation, gamma rays, 9 immediately prior to transplantation. On follow-up 10 there was no detectable viremia at early or late time 11 points, no detection of the amphotropic envelope by 12 PCR for up to three years, after long term follow-up 13 no detectable pathology out to seven years. 14 So on the basis of those studies, the 15 replication competent retroviruses are probably 16 thought of as being fairly innocuous in primates. So 17 a large study was done in the lab of Art Niehise at 18 the NIH examining the transduction of autologous bone 19 marrow in a group of ten rhesus macaques. 20 In this protocol, using the vector 21 packaging system on the first slide, they had a new 22 twist to increase the titer of virus introduced into 23 the bone marrow cells, and that was an amplification 24 technique that they termed the ping-pong amplification 25 where they basically bounce the vector between 282 1 ecotropic and amphotropic packaging cell lines in a 2 co-culture system. 3 This increased the titer of the N2 neo- 4 vector carrying neomycin marker gene in the study, and 5 also produced low levels of replication competent 6 retrovirus, which they knew existed at around -- up to 7 103 focus forming units, although the number does 8 bounce a bit between the papers. 9 So if we describe the largest number, 106 10 focus forming units of the replication competent 11 retrovirus, this was infected into 108 CD34 selected 12 bone marrow cells in vitro in a system that cultured 13 the bone marrow cells with the vector for 80 hours, 14 and then these were infused into ten rhesus macaques. 15 There was immunosuppression supplied prior 16 to harvesting or just after harvesting the bone 17 marrow, 5 FU for nine days, and then one and two days 18 prior to transplantation there were two doses of 500 19 rad of total body gamma irradiation, so total again of 20 1000 rads of irradiation. 21 So the surprise was that chronic 22 retroviremia appeared in three animals, and these 23 animals were noted not to have antibody against the 24 GP70 or P30 capsid proteins of replicating competent 25 retrovirus. 283 1 These three animals went on to develop T- 2 cell lymphomas after 200 days, and these lymphomas 3 were, obviously, rapidly fatal. 4 So just to summarize the animals there and 5 the data that I've just described to you, we had an 6 opportunity to follow up in particularly two of these 7 animals the types of viruses that were administered 8 through this protocol. 9 One animal in particular had very well 10 preserved samples, this 15445, the other 88049, had 11 suffered a degradation post mortem, but we were able 12 to use these samples from animals to see what had 13 happened -- what had generated the problem here. 14 So the first thing we looked at was the 15 replication competence of these viruses in diploid 16 rhesus cells. We wanted to avoid selecting particular 17 murine viruses by culture on rodent cells, and we 18 could find that readily infect this rhesus lung 19 fibroblast cell line with -- directly with plasma or 20 with virus pelleted out of plasma. 21 What was the surprise when we took that 22 virus harvested out of the rhesus cells, along with 23 virus similarly cultured in rhesus cells, and that 24 virus being the 4070A amphotropic retrovirus -- so the 25 prototypic amphotropic retrovirus -- normalizing 284 1 inoculant, reinfected that back into rhesus lung 2 fibroblasts, and the virus that was present in monkeys 3 had a much greater capacity for replication in the 4 rhesus cells compared to the prototypic amphotropic 5 retrovirus. 6 So when these viruses with the same 7 inoculant were applied to Morsoduni fibroblasts which 8 are very permissive for the growth of murine leukemia 9 viruses, this growth differential was not evident. 10 So summarizing the properties in terms of 11 infectivity of these viruses in various cell types, 12 reverse transcriptase activity was directly detectable 13 in the plasma of these animals. As I've said, the 14 rhesus lung fibroblasts were regularly infected, 15 bronchus fibroblasts. Human helo cells were also 16 infectable, and rodent fibroblasts and Morsoduni 17 cells. 18 What were not infectable after many 19 attempts were rhesus PDMC, which we had activated with 20 THA, and this is a bit of a pity, because the PDMC are 21 obviously the easiest cells to examine from animals or 22 patients for replication competent retroviruses, but 23 may probably provide a poor source of deriving virus 24 back. 25 So we next undertook a molecular 285 1 characterization of the types of virus present and 2 transmitted and expressed in the host animal. We 3 chose to us an RT PCR protocol and focusing on two 4 areas of the virus, the envelope region which, we have 5 already heard, determines tropism of the virus, and 6 also the LTR, particularly the U-3 region containing 7 the promoter and the enhancer sequences, because these 8 also determine the tropism of the murine viruses, as 9 we've heard. 10 So we extracted the RNA directly from the 11 tumor tissue, made poly-A RNA from it, synthesized 12 CDNA. Then we chose, after a lot of sequence 13 alignments, sequence primers that would amplify as 14 many of the murine leukemia viruses that we could pull 15 out of the database and stuff into an alignment. 16 So we're essentially going for full length 17 envelope sequences here, and we were successful using 18 different samples of RNA from the rhesus tumor and 19 amplifying full length envelope CDNAs, and we 20 amplified no bands from the rhesus lung cell line. 21 I guess what is important to point out in 22 this slide is that what amplifies is mostly the full 23 length envelope, not other shorter products that might 24 indicate the transduction with cellular components, as 25 we might find after passage of the virus in the rhesus 286 1 lung fibroblast cell line. 2 So we sequenced -- or we cloned those PCR 3 products and sequenced the clones that we derived. 4 Essentially, three types of envelope structures were 5 observed, the amphotropic envelope which was directly 6 the same as that derived from the packaging cell 7 construction 4070A with clear recombination back into 8 Moloney right here around the R protein region of 9 Moloney. 10 There was a variance within this type of 11 envelope that had deleted six base pairs in a key 12 region determining the tropism that's in the VRB or 13 variable receptor binding domain of the envelope 14 protein, and this led to an amino acid sequence change 15 within that crucial amino acid loop structure. 16 What was a surprise in this animal was to 17 find a new player, an MCFY envelope where the bulk of 18 the envelope sequence had come from the Frend MCF 19 strain with sequence towards the junction with the 20 4070A amphotropic sequence, resembling a cluster 21 alignment with xenotropic envelopes. 22 So this was in one animal. When we looked 23 in the other two where we had less good starting 24 material to do an RNA analysis, looking instead at 25 DNA, it was clear that all of the animals, all three, 287 1 had the amphotropic envelope sequence present, but 2 only the one animal had the either MCF sequence with 3 primers within the MCF region only or the 4 recombination between the amphotropic and MCF 5 sequence. None of these animals amplified with 6 xenotropic primers. 7 The recombinant amphotropic MCF virus 8 could be traced back to the producer cell clone and 9 its emergence appearing between the 22nd and 25th 10 week. So it seems that its derivation was from the 11 packaging cell line. 12 So next we examined the LTR structures 13 within the various players here. So firstly, I draw 14 your attention to this line here, which is 15 amplification across the U-3 region of the LTR. So 16 we're looking at the structure of the core enhancer 17 and the promoter region. 18 The amphotropic virus has a single core 19 enhancer, giving rise to this size fragment here. The 20 A2 producer cell line has predominantly the Moloney 21 LTR which drives the expression of vector and also the 22 recombinant retroviruses from the previous slide, and 23 Moloney has a duplication of the core enhancer, and we 24 see a larger size PCR product. 25 Also evident was another faint band which 288 1 was further duplication of the enhancer element being 2 expressed within RNA in the producer clone. 3 In the two rhesus samples that we 4 examined, the predominant forms of LTR expressing RNA 5 were those that had the three copies of core enhancer 6 and four copies of core enhancer was also evident in 7 these cells. 8 We infect the same virus into the Mosduny 9 cell type, the indicator, it more closely resembles 10 the expression seen in the A2 producer. So this 11 greater addition of enhancer sequences seems to lead 12 to preferential expression in rhesus. 13 So just to give a cartoon version, this is 14 the amphotropic type U3 region. This is what Moloney 15 has. These forms here seem to be expressed in the 16 tumors. 17 So knowing the structures of the LTRs, we 18 did PCR analyses that would match the envelope types 19 to the LTRs and, as we expected, the three envelope 20 types characterized earlier, each had the LTRs that I 21 described, each having duplications of the core 22 enhancer. But when we used a specific primer for the 23 MCF type LTR, the U-3 region, we asked whether this 24 was also being expressed at very low levels in this 25 specific PCR for the MCF endogenous virus itself, we 289 1 indeed found expression of this in the tumors. 2 So I just wanted to point out that these 3 viruses here, the MCF viruses, we couldn't detect 4 infection in either the Mosduny indicator cell type or 5 the diploid rhesus lung, but these do grow in mixed 6 cell lines. 7 So if we look at the genetic structures 8 that we pulled out, so that the main player is the 9 recombination between the vector and helper virus. So 10 it's integrated into the DNA and does express RNA in 11 the lymphoma. 12 Also, we find recombinations between 13 endogenous retroviruses in two types. We see the MCF 14 endogenous envelope represented in this main 15 replication competent entity, but also we see just the 16 expression of the endogenous retroviral elements 17 themselves as RNA. 18 As Clive Patience mentioned earlier, we 19 also looked for VL30, and we found lots of VL30 DNA 20 present in the rhesus tumors, but no RNA expression 21 from these endogenous elements. We also looked for 22 the co-packaging of cellular RNA species or DNAs, 23 which is not that easy to do, and we didn't find 24 anything. 25 So if we want to summarize the factors 290 1 that might account for the pathogenesis of the 2 replication competent retroviruses in three of the ten 3 monkeys in this study, first we would have to question 4 the use of mixed cell populations or the co-culture 5 technique that might have fostered additional rounds 6 of replication to generate novel replication competent 7 entities. 8 The high level of post-immunosuppression 9 is something that's also being pointed to, although 10 this has been debated quite a bit, whether that was 11 particularly high or more significant than previous 12 studies where this didn't seem to contribute to the 13 emergence of pathogenesis. 14 Certainly, the lack of response or lack of 15 antibody response to the presence of the replication 16 competent retroviruses in those three animals is 17 another thing that seems to indicate a problem for 18 that particular animal, and probably also associated 19 with the emergence of the chronic retroviremia. But 20 the adaption of the replication competent viruses is 21 also, we think, a very important feature. 22 So the genetic factors, particularly, the 23 recombination events that arise and are associated 24 with elements from within the packaging cell lines, 25 particularly the expansion of the core enhancer 291 1 elements in the U3 region of the LTR, the modification 2 of the BRB domain is a consistent feature between the 3 animals, and we're unsure of the relevance of this, 4 but probably enhances the infectivity to rhesus cells, 5 and also the recombination with endogenous MCF is also 6 an interesting thing, but doesn't correlate with each 7 of the tumors in all of the animals. 8 Epigenetic factors, the alpha gel, so the 9 lack of addition of the 1-3 galactose on the 10 carbohydrates when virus is derived out of primate 11 cells in this final study compared to the first one, 12 clearly makes the difference here. So the viruses 13 that emerge out of the autologous cells are resistant 14 to complement lysis. 15 There was a very multiplicity of 16 integrations after the prolonged in vitro infection, 17 the 80 hours of exposure of the bone marrow cells to 18 the vector and RCR mix. 19 So the implications are to select 20 packaging cell types with the fewest endogenous viral 21 RNAs, that the properties of the replication competent 22 virus derived that emerge that combine the endogenous 23 elements may not be totally predictable, are certainly 24 not the same as prototypic viruses that go up to make 25 the elements placed into the packaging cell lines, and 292 1 this has implications when selecting the indicator 2 cell lines for detecting the presence of RCRs in males 3 that have implications of prime transfer across 4 species. 5 Certainly, the immune compromising or 6 immunosuppression of recipients for gene therapy 7 protocols would place them at greater susceptibility 8 to spreading infection with RCRs. 9 Now because the organizers have put 10 implications of this work for viral vaccines, I've 11 included this slide here to try raising some of these 12 points. 13 Endogenous retroviral elements that are 14 present in many neoplastic cell lines, and 15 particularly the murine cell lines, as we've heard 16 earlier in this session, might receive help function 17 from the expression of functional viral proteins which 18 may be a component of a vaccine producing a functional 19 viral protein that might interact with elements, be 20 they RNA or protein, in the cell line. 21 The expression from viral genes may lead 22 to repair of replication competent retrovirus, and 23 that really is the really nasty element that is the 24 most obvious thing to guard against, the greatest 25 danger, if you like. 293 1 RCR might have novel cell tropism, an 2 enhanced infectivity in vitro compared to prototypic 3 retroviruses due to modifications in the envelope and 4 LTR. Live attenuated retroviral vaccines may 5 recombine with endogenous elements during production 6 in cells expressing these endogenous retroviral 7 elements, and mixed cell populations may facilitate 8 rounds of replication that foster the emergence of a 9 replication competent retrovirus. 10 So whether this argues maybe that some 11 cell lines may be better because there's definitely a 12 clonal population, that's another way of interpreting 13 this. 14 I just wanted to thank my co-workers who 15 helped with this work while at the lab of molecular 16 microbiology at NIH, Norton Martin's lab, Elio Vannen 17 from Gene Therapy, Inc., who did another similar study 18 also published, and Art Niehise who is now at St. 19 Jude's. Thank you. 20 (APPLAUSE.) 21 AUDIENCE PARTICIPANT: Have you looked at 22 the integration sites of those lymphomas? 23 DR. PURCELL: That was -- Some studies 24 were done with that, but it was appreciated that in 25 the lymphoma there were up to 35 integrants from the 294 1 replicating competent retroviruses, and it was 2 considered to be at that point too difficult, too many 3 integrations. So that's clearly the undercurrent of 4 the pathogenesis, the insertional mutagenesis, from 5 having so many integrations into the genome. 6 DR. COFFIN: John Coffin from Tufts. Your 7 talk and several before emphasized a very important 8 but not often completely appreciated point about 9 retroviruses, and that is that their really very large 10 genetic flexibility allows them to respond to all 11 kinds of different selection pressures in all kinds of 12 remarkable and surprising ways. 13 So when one is thinking about handling 14 systems that have retroviruses in them, one should 15 think not only about the genetic variation, the 16 mechanism of genetic variation that you undergo, but 17 also and probably much more importantly the kinds of 18 selective pressures you're putting on these viruses to 19 enhance their growth. 20 For example, in the AKR mouse it's really 21 remarkable that every single animal in lockstep 22 virtually undergoes the sorts of changes that we saw 23 in slides earlier, and the same kind of remarkable 24 changes that you see here, and there are many, many 25 other examples of these kind of things in 295 1 retrovirology. 2 So I think we always have to be very 3 careful in our thinking about how this goes, not only 4 for understanding molecular mechanisms but also for 5 understanding selective pressures that we're putting 6 on the viruses and the cells to produce them. 7 DR. PURCELL: I couldn't agree more. I 8 mean, also an interesting component of this things is 9 when you put these viruses together in vitro where you 10 don't have immune selections and you're greatly 11 enhancing the start of the virus to recombine and make 12 a whole host of new entities, do we immune suppress 13 the animal? That gives it a new environment in which 14 to make changes and get going. 15 DR. ROSENBERG: The last question, because 16 we are running and want people to stay and hear our 17 final speaker. 18 AUDIENCE PARTICIPANT: I'll speak quickly. 19 Two very short questions. 20 Damian, do you know if anyone actually 21 measured the levels of natural antibody in the primate 22 recipients of the various viruses? 23 Secondly, have the complement inhibitors 24 ever been directly shown to protect viruses produced 25 from potentially sensitive cell lines? 296 1 DR. PURCELL: There are studies published, 2 I believe, that look at the lysis of murine derived 3 retroviruses by human complement, and ascribe that 4 activity. But I'm not sure that it's been conducted 5 in molecular rigor. 6 AUDIENCE PARTICIPANT: hose studies 7 certainly do implicate alpha 1-3 gal as the culprit 8 molecule, but I don't think anyone has ever shown that 9 the complement inhibitors work directly on the virus 10 particles by accidental packaging of the inhibitors. 11 DR. PURCELL: Yes, I have to agree with 12 that. I mean, my slide there was more deduction of 13 the data rather than a good study to explain those 14 results. So, yes, more work needed. 15 DR. ROSENBERG: Our final talk this 16 afternoon is by Paul Jolicoeur on defective 17 retroviruses. 18 DR. JOLICOEUR: Yes. I should thank the 19 organizer for giving me the opportunity to talk. I'm 20 going to talk mainly about pathogenesis of defective 21 viruses, as I was asked to talk about. The first 22 slide, please. 23 What are defective retroviruses? This is 24 illustrated here. They are a shorter genome and, in 25 general, they have deleted or mutated genes. 297 1 Therefore, they cannot replicate by themselves, and 2 they do need a nondefective, which we call helper 3 retrovirus for full replication. 4 Some of them, however, are not a gene in 5 general, which I think capture many of the oncogenes, 6 and we'll see a little bit later what they are. These 7 defective viruses can be expressed normally, once 8 integrated, and we count helper proteins. This is 9 important for pathogenesis of some of them; and if 10 expressed, they can also be rescued by other 11 nondefective viral proteins, even those encoded in 12 other defective viruses. 13 In other words, in helper free stock you 14 can generate one-cycle replication, and you heard 15 about that earlier, in the packaging system where two 16 defective viruses complement, in a way, each other. 17 Why are there concern about defective 18 retroviruses in vaccine preparation? Again, you heard 19 about it from the previous speaker, that we do have 20 several copies of endogenous defective retroviruses 21 which are either nondefective or nondefective 22 retroviral sequence, including in humans where most of 23 them, if not all of them, are defective. 24 These endogenous sequences participate in 25 recombination events, give rise to novel viral 298 1 entities, novel viruses, again as you heard from 2 previous speakers. And several strains of defective 3 retroviruses have been shown to be pathogenic. 4 There are broadly two classes of defective 5 viruses, the one which essentially encode only viral 6 genes such as this one, and the best known examples 7 are the Frend virus and the duplan strain virus or the 8 MAIDS virus on which I will talk just a little bit 9 more; and obviously, the one which encodes V onc, 10 which have captured a nonviral cellular gene, and from 11 which we have learned most about oncogene and from 12 which the first oncogenes were derived from and labels 13 on the FBJ, the RAS Harvey and Kersten and so on, and 14 where you do have a V onc. which has been captured in 15 the middle of the defective viruses. 16 The SFFV containing only viral sequences 17 is defective in the sense that it has gag and pol -- 18 several gag and pol mutations. It has a complete and 19 open reading frame, but its end sequences are a form 20 of two types of sequences, nonecotropic and ecotropic 21 recombinant, plus an insertion, a one base pairs 22 insertion here, giving rise to this new protein which 23 is able now to recognize a structure which normally 24 none of the parents recognize, that is, the CD EPO 25 receptor. 299 1 The GP55 which is encoded by this Frend 2 virus would stimulate the EPO receptor and give rise 3 to this massive proliferation and erythroleukemia. 4 This is an example of a virus which is defective and 5 is able to participate in tumor formation very easily. 6 I have listed here some of the 7 characteristics of defective retroviruses. We are 8 going to go through some of them, and I will 9 illustrate many of them in a model which we call the 10 MAIDS model. 11 First, defective viruses can be generated 12 in a lifetime of an animal, and this has been the 13 hallmark of the V onc. defective viruses which all 14 occur in one animal and were isolated in this animal, 15 although very rare genetic events are required to 16 generate in general these viruses, sometimes two or 17 three recombinations of them. 18 Defective and recombination retroviruses 19 appear to be generated more easily in vivo than in 20 vitro. Defective viruses or pathogenic viral genome 21 can emerge from nonpathogenic viral sequences. 22 As I said previously, two defective 23 retroviruses can complement each other. Individual 24 retroviral sequence expressed in absence of 25 replication in defective systems, such as the 300 1 packaging system, can be pathogenic or, in the context 2 of transgenic where you express only one gene. 3 In general, those viral genes interfere 4 with -- or participate in cell signaling. We also 5 know that defective viral sequence such as VL30 can 6 enhance V onc. activity or such as gag in gag oncogene 7 fusion protein. It has been shown, for example, with 8 a gag Abelson fusion protein where gag is important 9 for myristylation and location in the membrane. 10 Finally, defective viral genomes are as 11 effective as nondefective viral genomes as insertional 12 mutant, and we'll see some example of that. 13 Just to show a cartoon here, how can rare 14 events -- very rare events, in fact, get -- occur in 15 a live form of an animal as a capture of an oncogene 16 where it has been alluded before by Naomi. First, it 17 has to -- This is a model which has been postulated by 18 many labs, including Mike Bishop, and has been 19 reviewed here. 20 You have to have -- It is postulated that 21 there is an insertion in front of a C onc., that you 22 do have deletion which by itself is rare. First event 23 is rare. Second event, this deletion is relatively 24 rare, and you do have a transcript. 25 You make a packaging sequence which has 301 1 been conserved here so that you can package this with 2 an infection of a new cell where you do have 3 transcription and a jump, which is also a relatively 4 rare event, where you do have a jump to capture this 5 sequence and put it between two LTR in chicken and 6 mice, mainly in these two species. It has been well 7 documented that you do see that in the lifetime of a 8 single animal. 9 I will not go through that, because many 10 people have talked about the packaging system where 11 two defective genomes finally can be produced in a 12 one-cycle event. 13 Let me focus now on an animal model which 14 is called MAIDS, which has been induced by defective 15 viruses and where we can see the plasticity of what 16 these viruses are. 17 MAIDS stands for murine AIDS, essentially 18 a lymphoproliferative disorders with splenomegaly and 19 lymphadenopathy where you do have severe immune 20 dysfunction of TNB cell lymphocyte. 21 There is massive lymphoproliferation in 22 these mice. You can have as much as two or three 23 grams of lymph node tissue, due to this massive 24 proliferation. It's basically caused by defective 25 viruses which has a single open reading frame in the 302 1 gag region with severe deletion in pol MRV regions. 2 So essentially it's only a viral gene, and 3 one gene -- one viral gene has been kept open. The 4 origin of this virus is interesting. It arose after 5 successive passage of cell-free extract for mix-free 6 induced thymomas in C67 blacks. 7 The crude stock themselves contain several 8 substrains of MuLV initially, and it is likely to 9 originate from endogenous proviruses. In fact, some 10 of the parents have been partially identified. 11 One is a family of endogenous MAIDS 12 related virus, MRV, identified in the mouse genome -- 13 In fact, there are five members of that -- by Neil 14 Kaplan, where the gag is intact and where a deletion 15 of pol MRV is present, very much like the MAIDS virus. 16 The gag region is homologous to the MAIDS 17 virus and harbors some additional mutation, however, 18 and the LTR is this thing. The MRV is a xenotropic 19 LTR, while the maze virus is an ecotropic LTR. 20 There is another family of expressed 21 genome, EDV, where the gag is homologous to that of 22 the MAIDS virus if you allow a frame shifting. 23 Therefore, it appears MAIDS is a recombinant virus, 24 probably of at least three distinct endogenous 25 retroviral sequences, which again occur during passage 303 1 of irradiated tissues. 2 It encodes a single protein, which is a 3 gag protein, which is not cleaved in matrix or capsid 4 or nucleocapsid proteins. It is not cleaved. This is 5 a peculiarity of this gag protein. 6 It is myristylated and well produced in 7 cells, and our hints that this system could be 8 interesting to study the pathogenesis came when we, in 9 fact, used a packaging system to produce helper free 10 stocks of these defective viruses and to show that, in 11 fact, these helper free stocks, which are replicating 12 incompetent except for one cycle, would cause disease 13 to a high frequency -- this is a Type 2 virus here -- 14 in fact, at 100 percent as compared to a replication 15 competent helper virus. 16 This is remarkable, because when you think 17 about it, when you showed this prep in IP, it gets 18 into the peritoneum, and the very first cell it's 19 going to see and infect will be the last one, and 20 these cells will cause the pathology you just saw. 21 We are talking here of very low titer. In 22 the range of 103, 104 will give rise to this disease. 23 So 1000 cells to 10,000 cells in the body at maximum, 24 if we don't account for the loss of infectivity going 25 through the peritoneum, are responsible for the 304 1 disease. 2 What's happened when you do that and look 3 in the spleen, for example, or lymph node, you do very 4 early see little clones of proliferation which are 5 independent. So a little polyclonal proliferation in 6 distinct nodes and lymph nodes which eventually, as 7 the time goes on, will fuse and get massive polyclonal 8 proliferations. 9 We have been able, through different 10 techniques -- I'm showing only one here; this is 11 mainly a typing by immunocytochemistry to show that 12 these cells are B cells. This is a B220 staining as 13 opposed to the control, different controls in lymph 14 node or a spread of cells here. This is a positive 15 cell, because you see the blastic-like inside two 16 positive cells, and the noninfected, more mature B 17 cells which are smaller. So that you have the 18 features of blastic, and they are B220 low. 19 Another interesting feature of this growth 20 is that it's thought in many mice as polyclonal. This 21 is a blot where we look at integration site, on the 22 junction of integration site, and you can see that 23 different nodes in the same animals -- in this animal, 24 for example, or this animal -- you do have a 25 polyclonal or different -- not polyclonal but distinct 305 1 integration. But in some animals the same clone has 2 taken over the whole animal here. 3 That is to say, a single clone here, a 4 single cell really, took over and appears in these 5 lymph nodes. 6 Looking at that, we hypothesize that in 7 fact not only the gag was providing a growth advantage 8 to these cells -- that is quite obvious, because you 9 can see it polyclonally and very quickly -- but also 10 the genome itself was an insertional mutagent. 11 We test this hypothesis by cloning some of 12 these integration events and asking whether these loci 13 were occupying a certain percentage of the tumor, and 14 we did find at least two loci like that which we call 15 Dis-2 where we did find common integration. 16 We have not yet identified the genes 17 involved here, but the Dis-1 -- and you can see here 18 the rearrangement and different tumor or lymph node 19 enlargements, if you wish, in the band of the Dis-2 or 20 Dis-1 locus here. 21 The other locus that we have identified is 22 the PU.1 locus which is a transcription factor which 23 is specific for B cells in macrophage. You can see in 24 a certain percentage of tumors that the integration 25 occurs at the same place where there were initially 306 1 identified in erythroleukemia and the Frend's systems. 2 Therefore, this defective virus 3 participates in growth of these cells, not only by 4 providing the gag protein but also by serving as an 5 insertional mutagent, and this is an example of a 6 defective virus genome which can serve as an 7 insertional mutagent. 8 Now we did the structure function of that, 9 and again this illustrates a point. When we do point 10 mutation in this protein very early at the subcode 11 down here within a few amino acid of the beginning of 12 gag, disease capacity of that virus is lost. 13 If you do other point mutation along the 14 protein or the sequence, you do see that this viral 15 genome is pathogenic, but at the same time we found 16 recombination in these viruses. 17 So, in fact, these are nonmutant, because 18 they are revertant, and you will see what kind of 19 version we found. So the only two viruses we could 20 use where we found loss of capacity to induce disease 21 was this early mutant stop coder and a myristylation 22 negative mutant, as you can see here. 23 Surprisingly, despite the fact that we had 24 a single point mutation in the middle of this 25 sequence, the type of reversion we got was not a 307 1 reversion of that mutation, but in fact was a 2 recombinant. 3 As you can see here, this is the virus we 4 put in with the capacity to encode now a very short 5 protein as opposed to P63 the encoder protein of P21. 6 This is in fibroblasts where we produce a stock of 7 virus. 8 In every mouse inoculated with this virus 9 where we looked in every lymph node, what we found is 10 a recombinant having the molecular where it closed to 11 PR60, and you can see they act differently in each 12 other. 13 So again this illustrates the point with 14 virology that John was coming at a few minutes ago, 15 that those are very plastic and even, surprisingly, 16 with a point mutation like that, what you achieve is 17 you achieve a highest recombination event presumably 18 with other endogenous sequence, and we know this was 19 not -- we know it was not generated in vitro. 20 The myristylation negative mutants tell us 21 something about that. First, it tells us that 22 myristylation itself is required. So, therefore, 23 attachment to the membrane is important for the 24 pathogenicity of that virus, and this suggests that 25 PR60 molecule is the only protein required for disease 308 1 induction. 2 Two, it suggests that the processing of 3 that protein or that antigen, if you wish, by B cell, 4 which are APC, is not required for this effect, 5 although some people could argue about it; and PR60 is 6 used as a docking side. 7 The way we see it now is that you do have 8 this protein here, and presumably, by interacting with 9 some protein, you do send two messages. One is 10 proliferation, as illustrated by the very quick, 11 within a day or two -- very quick proliferation of B 12 cell polyclonal and, two, by reprogramming of these B 13 cells the sense that B cells which proliferate will 14 not induce total anergy of T-cells, as we do find in 15 these mice. 16 So, therefore, there is something special 17 about these cells which make them energize T-cells 18 which are not themselves infected, and there are some 19 motif, especially in the P12 domain of this protein, 20 which is a very proline rich protein as opposed to 21 either P12 protein of other MuLV, and where you do 22 create pyridivous H-3 finding site, and as well some 23 H-2 phosphotyrosine binding sites, as you will find in 24 signaling molecule. 25 So to search for a putative factor binding 309 1 to PR60 gag, we use the yeast-two hybrid. As a bait 2 we use the PR60 in the LXA system, and the very first 3 clone we got was an interesting clone. It was the 4 only small clone containing the SH-2 - SH-3 domain of 5 C-ABL. 6 We did some coimmunoprecipitation and 7 other studies that I don't have time to go through 8 here, showing that this interaction was -- could be 9 seen in vitro and in cells infected with the virus 10 itself, as you can see here with an anti-IPA both, for 11 example, and coming back with anti-CA. 12 Therefore, we do see this molecule, PR60 13 gag, as a docking molecule where protein or a 14 signaling molecule such as C-ABL will dock to, or 15 other putative molecule to send a single proliferation 16 and reprogramming of these B cells. 17 So I will conclude with that, to state 18 again that this is probably a docking site for C-ABL 19 through the SH-3 domain, and the translocation of C- 20 ABL to the plasma membrane in this system may release 21 its block on the cell cycle progression and may give 22 it access to novel target substrates for tyrosine 23 phosphorylation. 24 The way we see the pathogenesis of this 25 virus is summarized here in a cartoon where the 310 1 defective viruses, whether with a helper or without a 2 helper, as we have shown, can infect a B cell which 3 will start proliferating locally in the germinal 4 center -- I forgot to mention that these cells are 5 located -- these B cells are located in the germinal 6 centers -- which eventually will progress, and at the 7 same time influence other cells, either to progress or 8 attract them. 9 I should mention that the infected cells 10 in the huge lymph node sometimes represent only five 11 percent of the cells of the node, a little bit like 12 you will find with the Reed Stanford cells where the 13 majority of the node is composed of normal cells. 14 This is also the case here in most of the time, which 15 is by itself an interesting biological feature of the 16 system. 17 Consecutive to the growth of these cells, 18 you do find in the body a defect in non-infected 19 cells, which are themselves T-cells where you do find 20 total anergy of all the B bearer T-cells, as well as 21 defect of D cells. Obviously, the infected B cells 22 are themselves un-normal as well. 23 So again, I'm just putting the same slide 24 you saw before, which illustrates some of the 25 characteristics and, in fact, you find many of them in 311 1 this system of the MAIDS virus. 2 I'm not going to go through all of them 3 here, but again to point toward one important aspect 4 is that defective pathogenic viral genomes can emerge 5 from nonpathogenic viral sequences, and that this 6 viral gene products are involved in cell signaling and 7 are able to interfere with the cells. 8 I will stop there, because time is running 9 short, and I will thank my collaborator who has worked 10 on this system. Thank you. 11 (APPLAUSE.) 12 DR. HENEINE: Walid Heneine, CDC. 13 My question is about set of data you 14 mentioned on the insertional mutagenesis with this 15 virus, and has relevance to what we've been discussing 16 earlier about transfer of DNA. 17 If I get it right, you said that one 18 mechanism also involved -- of the disease involves 19 insertional mutagenesis, despite the use of very small 20 virus inoculum. Right? 21 DR. JOLICOEUR: Yes. 22 DR. HENEINE: And you mentioned -- What is 23 the number you mentioned about those viruses that you 24 inoculate with -- the number? 25 DR. JOLICOEUR: Number of virus particle? 312 1 Infectious virus particle are in general 103, 104. 2 I should stress here that the insertional 3 mutagenesis is not an essential component of the 4 disease. In other words, you do have disease when you 5 do have a polyclonal growth of many clones. 6 The fact that some of these clones emerge 7 and take over is not a feature -- is not a necessarily 8 feature of the disease. In other words, to anergize 9 T-cell, you don't need this novel event, if you wish, 10 but it does occur presumably because there are so many 11 cells being in proliferation that some of them are at 12 an advantage over others. 13 DR. HENEINE: So it has no causal 14 relationship, you're saying? 15 DR. JOLICOEUR: Not for the disease 16 itself, no. 17 DR. HENEINE: Not for the disease. 18 DR. JOLICOEUR: No, but it occurs. 19 DR. HENEINE: Thank you. 20 AUDIENCE PARTICIPANT: I was very 21 interested in your apparently pathology of B cells 22 from the peritoneum into the germinal centers of the 23 lymph nodes and ultimately the systemic tolerance of 24 anergise in the T-cells. 25 Did you look at all in the bone marrow to 313 1 see if any of the B cells eventually resided, as I 2 say, into the marrow? 3 DR. JOLICOEUR: If the B cell, the 4 infected B cell? 5 AUDIENCE PARTICIPANT: Yes. 6 DR. JOLICOEUR: We take over the whole 7 body. We know that initially it's not a B cell from 8 the bone marrow which is infected. It's a B cell 9 which is localized at the site of inoculation. We 10 have done time course. 11 If you inoculate IP the very first cells 12 are within the mediastinal lymph node. If you 13 inoculate in the foot pad, it's in the right or left, 14 whatever you inoculate, lymph node and so on. 15 So the target cell is localized at the 16 draining lymph node. 17 AUDIENCE PARTICIPANT: Right. But do you 18 know if any of the B cells from the germinal center in 19 the lymph nodes eventually get back to the marrow 20 where they could have -- 21 DR. JOLICOEUR: They do, yes. They take 22 over the whole body. Oh, yes, all over, including -- 23 and quite infiltrating. Some of them infiltrate the 24 liver, infiltrate the brain and so on. Some get quite 25 aggressive. 314 1 DR. ROSENBERG: Thank you. 2 I'd like to thank all of our speakers -- 3 Oh, one more? Okay. 4 AUDIENCE PARTICIPANT: What is there -- 5 I'm Carlo Russo from Merck. 6 What is the receptor on the B cells? 7 DR. JOLICOEUR: Oh, this is the eco- 8 receptor. We use -- In each case the helper has been 9 the Moloney helper or the ecotropic helper. So, 10 therefore, the cell that would get infected is 11 infected through the eco-receptor. 12 Other group have used the amphotropic 13 receptor for entrance, and they get the same disease. 14 AUDIENCE PARTICIPANT: And the mechanism 15 of T-cell anergy, you think, is mediated by what? 16 DR. JOLICOEUR: This is a big mystery. 17 One theory has been that it is a super antigen effect, 18 but I think this has been quite ruled out by our group 19 and other groups now. 20 It's totally mysterious. It's mysterious 21 whether you need the contact with B cells to get this 22 or whether it's a cytokine type of release from the B 23 cells. 24 DR. ROSENBERG: If there are no more 25 questions, we'll adjourn, I think, until eight o'clock 315 1 tomorrow morning when there will be more presenters. 2 (Whereupon, the foregoing matter went off 3 the record at 5 :30 p.m.) 4 - - - 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25