Well, the murky beginnings were in about 1988, when Paul Brehm, who is now at Stony Brook, gave a seminar in our department. He was talking about bioluminescent organisms, in particular a coelenterate called Obelia geniculata. Coelenterates are the source of the protein aequorin, and the name comes from the jellyfish, Aequoria victoria. Aequorin is a bioluminescent protein that gives off blue light when calcium is added in vitro. But many coelenterates are not blue-light emitters. They emit a green light. Brehm said the reason for the color change was a protein called green fluorescent protein, GFP, and that shining blue light on the organism was sufficient to have the cells fluoresce green. At that point I thought to myself, "What an absolutely wonderful, wonderful compound this would be. I would love to put it into C. elegans and look at gene expression." In fact I imagined a really large number of things I would like to do if I had such a way of marking gene expression and protein localization in living cells. I got very excited and didn’t listen to another word of the seminar.

I spent the rest of the day and maybe the next day trying to find out who was working on the molecular biology or biochemistry of this protein and eventually found Douglas Prasher at the Woods Hole Oceanographic Institute. He was in the process of trying to clone the cDNA for GFP. I called Prasher and it turned out he was as excited about this as I was. "This is terrific," I told him. "We have the right organism to put it in. It’s a transparent organism. We have cell-specific promoters to drive it. This is going to be lot of fun." But he hadn’t finished the sequencing or isolation. I think maybe he only had a partial cDNA. He said he’d get back to me.

I didn’t hear from Doug Prasher for a while. And then I went on sabbatical and it turned out that Doug called while I was away, because he had the DNA ready, but nobody told him how to get in touch with me. Whoever answered the phone just said I wasn’t there anymore. Finally in about September 1992, a graduate student, Ghia Euschirken, one of the co-authors on the Science paper, rotated through the lab and I told her the GFP story. I said I really would have liked to have used GFP but that Doug never got in touch with me. So we went to Medline to see if there was anything on GFP, and Doug had just published a paper in the journal Gene saying he had the cDNA for GFP. l went and got the paper from the library and it had his phone number and I called him up and said, "What happened to our collaboration? Do you still want to do this?" Of course, he did. But for Doug, the protein wasn’t fluorescent. He said he could send the clone to me, and I talked to Ghia about how to make the construct. Maybe a week or two later, she came back and said she wanted to show me something. I went and looked in the microscope and saw green fluorescent E. coli. It had worked. That was pretty exciting. And I immediately got on the phone and told Doug and we started doing other experiments.

  Why did GFP fluoresce for you and not for Prasher or other researchers? What was the trick? Was it serendipity?

I have to set the stage for you; what the problems were at the time. GFP was first identified by Osamu Shimomura, who was at Princeton. He was the one who purified aequorin and had a paper in the early 1960s identifying another protein that fluoresced green. By the time I first spoke to Doug Prasher, it was understood that the fluorophore that made the protein fluorescent was actually formed from the primary amino acid sequence, but nobody knew how many enzymes, if any, were needed to make the final product. In Doug’s paper on the cloning of the cDNA, the protein product is referred to as apoGFP to signify this concern. I was curious to see if the protein could work on its own. Although others saw Doug’s paper and were trying to do the same experiments, we were successful because of the way we did the experiment. I was very concerned that we only use the coding sequence and no flanking DNA in our construct. We used PCR to obtain just the coding sequence and that worked terrifically. But the others who tried to use GFP simply cut the original lambda cDNA clone with restriction enzymes, and got additional bases associated with each end. These extra sequences turned out to be quite detrimental; expression wouldn’t occur. So we were fortunate in the way we designed the experiment.

  Why choose Science for the announcement?

Actually, since I am in the worm field, the first written report about GFP was in the C. elegans newsletter in October 1993, and, of course, we had also told several people. So even before the Science paper came out, we were sending out samples to people to let them try their hands with it. The reason we published in Science was to reach a broad range of scientists, because we were pretty sure that GFP was going to useful in many contexts.

  What is it that makes this protein and your Science paper so significant?

There are a couple of things. First of all, because GFP needs no substrate, it works just by shining a light on it; you don’t have to permeabilize the tissue to allow it to work. You can simply look at the cells and see whether they are expressing it with the virtually non-invasive technique of shining light on it. The second and maybe even more important factor is that this can be done in living cells, and so you can watch the processes occur in real time or even sped up using time-lapse photography. You can look at the localization of proteins in living cells as was first done in Tulle Hazelrigg’s lab, and whether proteins relocalize, e.g., from the nucleus to the cytoplasm, over time. That’s very useful. Then there are some other technical aspects that are quite nice. One is that GFP has a very high quantum efficiency. You get a lot of light out for what you put in. And there has been a lot of work since then to make the molecule fold more efficiently and so utilize light even more efficiently. Of all these, the real advantage is being able to look in living cells. Another advantage is that GFP is a relatively small molecule. GFP has 238 amino acids and exists as a monomer, except at very high concentrations. These properties mean, for example, that GFP can diffuse throughout cells. In contrast, beta-galactosidase forms a tetramer. It’s a very large molecule and it normally stays sequestered in the cell body. It won’t diffuse throughout nerve cell processes, for example, unless you do some tricks. GFP is able to diffuse throughout cells.

  How has GFP technology changed since your first paper?

We use it in quite a number of different ways. We identify cells by using promoter fusions to drive GFP expression to determine patterns of gene expression. We use it to look at where proteins localize within cells. We use it as the basis of mutant hunts. We have used it to characterize mutants. We use it to label nerve cells so we can then electrically record from those cells. That is one of the first things I said it would be used for. C. elegans is an animal in which every cell is identifiable, but once you cut open the animal everything gets a little messy. With GFP, you can label a particular cell or subset of cells and then go in and locate those cells and actually record from them.

  What’s the most imaginative use of GFP that you’ve heard of, in your opinion?

One of the most intriguing applications that is currently in progress is work by Bob Burlage, who is expressing GFP in bacteria off of a promoter that responds to TNT. The potential here is that these bacteria could be used to detect land mines because they often leak the explosive. Eliminating these remnants of war would be terrific. A lot of different things have been tried with GFP. It’s been into virtually every organism that you can think of. It’s been used in plants. It’s been used in prokaryotes and in eukaryotes, including mammals. It has been a lot of fun.

  Considering how significant GFP has been, why do you think no one else came up with it, while you were waiting for Doug Prasher to clone it?

That’s a very important point. In hindsight, you wonder why 50 billion people weren’t working on this. But I think the field of bioluminescence or, in general, the research done on organisms and biological problems that have no immediate medical implications, was not viewed as being important science. People were working on this, but it was slow and tedious work, and getting enough protein from jellyfish required rather long hours at the lab. They had to devise ways of isolating the cells that were bioluminescent and then grinding them up and doing the extraction on them. It’s not like ordering a bunch of mice and getting livers out and doing an experiment. It was all rather arduous. It’s quite remarkable that it was done at all. It was mostly biochemists doing it, and they were not getting a lot of support. In fact, as I remember it, Doug Prasher had some funding initially from the American Cancer Society, and when that dried up he could not get grants to pursue the work. I never applied for a grant to do the original GFP research. Granting agencies would have wanted to see preliminary data and the work was outside my main research program. GFP is really an example of something very useful coming from a far-outside-the-mainstream source. And because this was coming from a non-model-organism system, these jellyfish found off the west coast of the U.S., people were not jumping at the chance to go out and isolate RNAs and make cDNAs from them. So we’re not talking about a field that was highly populated. It was not something that was widely talked about. At the time, there was a lot of excitement about molecular biology, but this was biochemistry. The discovery really was somewhat orthogonal to the mainstream of biological research.

Yes, I thought it would be pretty important. I certainly hoped it would be important and that people would see it as an extremely useful tool. What’s really been quite wonderful is how many people have put their input into modifying, changing, and improving GFP so that it has become more and more useful every year.

  Would you consider GFP your greatest professional achievement or is there some aspect of your C. elegans work, your primary research, that you would rate above it?

I’m certainly very proud of this. There’s nothing quite as enjoyable as having an idea and finding out that it works. We have a lot of ideas that are best forgotten and usually are. But when you actually think of something and it works, that’s extremely thrilling. And then you always hope to have a breakthrough someday that will have implications or usefulness to a large audience, and this really did.