Executive Summary of
the Cell Decisions in Response to DNA Damage: Survival vs. Programmed Cell
Death Think Tank
Current cancer therapy relies heavily on DNA damaging agents (radiation, DNA alkylating agents, etc.) to induce programmed cell death in cancer cells. The proven success of this therapy, albeit only in some patients, implies that cancer cells are more sensitive to killing by DNA damaging agents than normal cells. This increased sensitivity is believed to be due largely to defects in DNA damage response pathways within the cancer cell. Compelling experimental evidence suggests that it is possible to dramatically modulate the sensitivity of cells to DNA damage. Similarly it may be possible to modulate the cell-specific and stimulus-specific responses to DNA damage leading to cell death. These responses vary dramatically between cell and tissue type, metabolic state and genetic background. If the sensitivity of cancer cells to DNA damaging agents and the cell death response to them could be specifically increased (or the sensitivity of normal cells to these agents be specifically decreased) by just one order of magnitude this would lead to a significant increase in cancer cure rates.
The obvious benefits of such an outcome are:
A major goal of the workshop was to identify the knowledge and resources needed to optimize the DNA damage response leading to programmed cell death in human cancer. The Think Tank participants put forth a number of recommendations that can be summarized into 4 broad areas:
Introduction
The introduction of DNA damage is a major therapeutic strategy for killing cancer cells. However, cell death is not the sole option for cells in response to DNA damage. In brief, the damaged cancer cell has two options: To die by regulated programmed cell death (PCD), or to survive by preventing cell division until DNA repair can be completed. In their initial stages, both PCD and survival/DNA repair processes initiated by DNA damage share the same signaling cascades based on high-level protein kinases (e.g., ATM, ATR) and secondary kinases (e.g., Chk1 and 2) along with a number of other proteins involved in signal detection and Amediation@ of signaling and repair. What is less clear, however, is how, at the molecular/mechanistic level, human cancer (and also normal) cells that have sustained DNA damage assess its severity and make the ultimate cellular decision between death and survival. The aim of this Think Tank was to advance understanding of the mechanisms involved, with the ultimate goal of finding strategies that will select or enhance cell death.
Think
Tank Program
Session
I: DNA Damage and Repair
Discussion topics included:
¨ How do human cells (normal and cancer) detect
primary DNA damage?
¨ How are signals of primary DNA damage
amplified by the cell?
¨ What are common and distinct responses to
different types of DNA lesions?
The most lethal type of DNA damage is DNA double strand breaks (DSBs). DSBs can be caused directly by radiation or indirectly after DNA modified by chemotherapeutic drugs is processed by cellular enzymes. Both radiation and DNA-modifying chemotherapeutic drugs are currently used to treat human cancer.
The response to a DNA DSB includes DNA repair and DNA damage signaling components. DNA repair pathways are clearly important; however, DNA repair genes are not frequently mutated in human cancer and it is not clear that inhibiting these pathways will lead to cancer-specific killing. DNA damage signaling pathways (also known as DNA damage checkpoint pathways) include sensors, which detect the presence of DNA DSBs, transducers, which produce a DNA damage signal, and effectors, which induce cell death or cell cycle arrest (transient or permanent). Mutations that inactivate DNA damage checkpoint genes are extremely prevalent in human cancer, resulting in rewiring of the DNA damage signaling pathways and differences in the response of cells to DNA damage.
In the last ten years there has been tremendous progress in our understanding of DNA damage checkpoint pathways. Most of the known DNA damage checkpoint genes were cloned within the last ten years. Their function is just beginning to be understood and key research discoveries, especially with regard to the DNA DSB sensors, were published in the last year or are still in the process of being published. The emerging theme is that there is a small number of DNA DSB sensors and the response of the cell to DNA damage (cell cycle arrest or cell death) depends on which sensor is being used to recognize the damage. Crystallography analysis suggests that recruitment of at least one sensor (53BP1) to DNA DSBs is amenable to pharmacologic intervention. Inhibition of 53BP1 recruitment to sites of DNA DSBs would result in increased recruitment of other sensors (specifically, the NFBD1/MDC1-Mre11 complex), which is thought to be more potent in promoting cell death in response to DNA DSBs.
Cells also have at least two transducers of the DNA damage signal in the form of the protein kinases ATM and ATR. How these kinases are activated by the DNA DSB sensors is not yet clear. The pathway downstream of ATM and ATR is somewhat better understood, although there are still gaps in our knowledge there as well.
In conclusion, how cells detect the presence of DNA DSBs and transduce the DNA damage signal represents a major gap in our knowledge. The recent progress in this field suggests that we are at the verge of making considerable progress. Characterization of the DNA DSB sensors and transducers is expected to identify promising and pharmacologically amenable targets for development of cancer therapeutics.
Session
II: Checkpoints and Apoptosis
Discussion topics included:
¨ How do cells assess DNA damage information to
decide between survival and death?
¨ Is DNA repair linked to either checkpoint or
apoptosis induction?
¨ Are DNA lesions sufficient to trigger
apoptosis?
The p53-dependent pathway which induces apoptotic cell death in response to DNA damage is clearly relevant in terms of cancer therapy toxicity, since normal tissues exposed to DNA damage can undergo apoptosis. Stabilization of p53 following DNA damage results in its accumulation in both nuclear and cytosolic compartments. Cytosolic p53 can directly activate the proapoptotic activity of Bax to permeabilize the mitochondrial outer membrane. It remains to be determined to what extent this cytosolic effect, independent of transcription, contributes to p53-mediated death. If cytosolic p53 is apoptotic, the transcriptional upregulation of PUMA may provide another block to the anti-apoptotic functions of Bcl-2/Bcl-xl. However, in most solid tumors these pathways are unlikely to contribute to DNA damage-induced death, because in most tumors p53 is inactivated by mutations. Therefore it is anticipated that inhibiting p53-dependent apoptosis may enhance the therapeutic index of cancer therapy, although there are still reservations regarding whether such an approach would be beneficial in humans.
p53-independent cell death pathways are likely to be of greater relevance to killing of most solid tumors in response to therapy. Cell death in p53-mutant cancers may be due to apoptosis or perhaps more likely to progress through the cell cycle with unrepaired DNA DSBs, which occurs when DNA damage checkpoint genes are mutated. While in yeast and humans it is well established that defects in DNA damage checkpoint genes lead to extreme radiosensitivity (for example, in ataxia-telangiectasia patients in which the ATM transducer is mutant), a link between mutations in DNA damage checkpoint genes in human cancers and their radiosensitivity has not been systematically investigated (with the exception of the effect of p53 mutations). Recent studies have identified a p53-independent apoptotic pathway, involving p53-related p63 and p73, in mediating DNA damage induced apoptosis. Activation of p73 by genotoxic agents has been shown to involve the nuclear c-Abl tyrosine kinase and contributes to the apoptotic response to p53-negative cells. But in general, it is fair to say that we do not understand well how DNA DSBs kill cancer cells that bear mutant p53.
In addition, apoptosis is not the only way for damaged cells to die. Excessive or persistent damage can trigger other modes of cell death, including necrosis and mitotic catastrophe. It is also possible that these modes of cell death could offer opportunities for new therapeutic strategies.
Session
III: DNA Damage and Cancer Therapy
Discussion topics included:
¨ How can our knowledge about DNA damage
responses be best exploited to therapeutic benefit?
¨ What are the current roadblocks to
translating our knowledge about DNA-damage responses to therapeutic strategies?
Tumor cells start out as intrinsically more sensitive to apoptosis by virtue of their obligate oncogenic lesions, which push cells closer to their apoptotic activation threshold. This is thought to explain the innate sensitivity of most tumors (at least initially) to the crude and blunt types of classical therapies we currently use. Clearly, such innate apoptotic sensitivity is eroded during tumor evolution, and especially in response to the strong selective pressure of classical cancer therapies. However, such erosion is an evolutionary process and arises through a specific and restricted repertoire of anti-apoptotic mutations that the tumor acquires. Because of this, tumor cells remain chronically dependent on their restricted repertoire of anti-apoptotic mutations for their survival. In contrast, normal cells have no pro-apoptotic oncogenic lesions and are supported by a variety of trophic signals by virtue of their residing in their correct orthotopic somatic environments. By definition, tumor cells are outside their orthotopic environment and should therefore be acutely sensitive to therapies that negate their anti-apoptotic mutations.
Unfortunately, we know too little as yet of the range or repertoire of anti-apoptotic mutations that operate in different types of cancer cell. Since such mutations are acquired through natural selection, and since different selective pressures operate on different tissue types as they evolve into tumors, anti-apoptotic mechanism are likely to vary between different tumor types. This provides an unparalleled wealth of possible therapeutic targets, if we could but identify them. This must remain a priority.
Conclusions:
Pathways
A substantial body of data exists on the pathways of responses to DNA
damage in bacteria and yeast, and studies of pathways generally follow two main
strategies: mutations in pathways that trigger cell death, and responses of
cells (i.e., responses at the cellular level) to damage. Despite extensive information on some pathways
in certain organisms, many key gaps in knowledge exist in how cells
detect and repair DNA lesions and orchestrate these cellular responses. Further
investigation of DNA repair mechanisms in mitochondria and of the mechanisms of
target resistance and sensitivity is also warranted. We also need to increase our knowledge
of how DNA damage kills cells that lack a functional p53, and the role of
metabolism and microenvironment on cell death decisions in response to DNA
damage. We have accumulated clustered
knowledge in areas such as apoptosis and DNA repair, but we lack the knowledge needed
to link these clusters. The basic
mechanisms of DNA repair and molecular interactions in signal transduction
pathways are probably best studied in cellular models, including yeast and
mammalian cells, and advances in these
areas may assist in the identification of novel therapeutic interventions
(e.g., IR and targets of NHEJ repair).
It is also critical
in this area of investigation to understand the pathophysiology
of normal versus tumor tissue and the progression of normal tissue to tumor
tissue. It is assumed that mutations in
mechanisms that alter DNA repair or damage response pathways can potentiate tumor development. However, it is not clear whether this process
is (1) a stochastic phenomenon that “opens the door” to a cell progressing to
transformation or to cell death; or (2) an ordered, repeated process that
continues until reaching a specific threshold beyond which repair no longer
occurs, ultimately generating a cancer cell or causing a cell to die. Tracing the effect of a DNA repair defect
from its initial stage to the final outcome in a highly advanced tumor cell
would likely be highly informative in distinguishing between these two
possibilities and in identifying different death suppression mechanisms that
lead to variable sensitivity to DNA damage.
Comparisons of immediate responses to DNA damage, which appear to
involve posttranslational changes, and longer term adaptive responses, would
likely be informative.
Tumors may be considered as dysfunctional tissue. For example, tumor cells are attracted to a type of vasculature to which normal cells are not. In addition, specific lesions in tumors prevent terminal differentiation. A more sophisticated analysis of signaling pathways within currently curable human tumors would be enormously valuable. While these tumors (mostly pediatric leukemias, sarcoma, germ cell tumors, certain lymphomas) have been studied in the past, a more sophisticated approach, which compares and contrasts their signaling responses to chemotherapy (particularly DNA damaging agents) relative to incurable tumors, is likely to shed important light on key variables in treatment success.
Tools:
Considering that the NCI Developmental Therapeutics Program is being "reshaped," the NCI could now play a direct role in providing infrastructure and support for the development of new tools. The NCI should be looking hard at data already in hand with an eye to making it more accessible (e.g., cross-indices of chemical compounds vs. the ‘NCI60’ cell lines) or simply funding development to make needed reagents available (e.g., screening to identify panels of phospho-specific antibodies that work on tissue). A model is found with the development of the PubMed database, where the previous MedLine was rolled over into something that is web-accessible, readily searchable and easily elaborated to add additional features at the local level.
In addition,
improved, advanced computer and web-based systems modeling programs relevant to
DNA damage signaling also are needed.
Participants noted a two-dimensional computer model developed by Kurt
Kohn that incorporates the underlying biochemical mechanisms involved in the
cell cycle (a similar model for apoptosis also has been developed). Although these models can be helpful and
serve as a platform for further investigation, participants generally agreed
that a research model should be dynamic and should allow for contingency
pathways and multiple decisions by the cell.
It was noted that other more quantitative computer models are under
development. However, because of the
limited amount of data available, applications of these models currently are,
in turn, also limited.
Sophisticated mouse models that accurately recapitulate human neoplastic processes are also needed. Most existing transgenic and knockout models merely “mimic” cancer. Efforts, like that of the MMHCC, to more faithfully recapitulate relevant neoplastic processes should be strongly supported. Faithful mouse tumor models are essential since, as discussed countless times in the Think Tank, the response of somatic cells (normal and neoplastic) to DNA damage and/or other therapeutic insults is, in great part, dictated by interaction between the tumor cell and its dynamic somatic environment.
A particularly attractive suggestion regarding animal models was to engineer "reporter transgenics/knockins." Such mice would contain germline-encoded reporters (such as GFP, LacZ, Luciferase) that would respond to a variety of important signaling or transcriptional pathways (eg p53 responsive, NFkB responsive, HIF, hypoxia, and many more). Until molecular imaging becomes possible via PET (not likely for a while still), such animals could prove enormously valuable.
Human Tissues
Much of the emphasis in the past two decades had been on defining genetic differences between tumor and normal cells. This has come, quite naturally, from the emphasis on cancer as a somatic genetic disease process. The most likely way to identify truly useful differences between normal and tumor cells (i.e., those that when targeted would have the offer the largest possible therapeutic indices) is to determine which normal pathways in tumors are still functional and can be exploited in the background of the tumor to force specific endpoints, such as cell death. This analysis will be facilitated by the ability to assay the activity of damage response networks by protein expression levels.
Few tools are
available to study responses to DNA repair in tissues. However, such tools are necessary to map
processes in normal and tumor tissues. To
accomplish this goal ‘modern molecular pathology’ needs to be better defined in
this context. For example, what
functional imaging or histopathology markers are useful to examine endpoints
such as apoptosis, energy state and viability in vivo or in tissue/tumor
specimens from humans or mouse models?
Methods that cut across different types of specimens: e.g., functional
imaging, biopsy samples and archival tumor specimens (currently the world’s
largest untapped human tumor biology resource) would be particularly
valuable. The suggestion of phosphorylation site-specific antibodies that work on
tissue is a good example, as is simple staining for p53 abundance. This would be the best place to fund some
descriptive normal biology, which we desperately need to provide the
‘numerator’ for all of the tumor studies and begin to understand the tissue
correlates of therapeutic index. Use of
imaging as a noninvasive tool to examine molecular pathways in carcinogenesis
and in response to DNA damage is still in the early stages of development but
holds great promise. Expanded use of
skin (or biopsies) was suggested as a model for imaging studies and assessment
of the pathology of normal versus carcinogenic cellular processes.
At least two types
of effort are focused on mechanisms and markers of DNA repair in human tumor
tissue. They are academic-commercial
laboratory collaborations developing reagents (e.g., phosphoepitope-specific
antibodies) and researchers conducting tumor specimen analyses (e.g., chromatin
status or modification in relation to tumor type or stage). Forging better connections between the two
would facilitate movement of new analytical reagents and techniques into the clinical
setting and yield more information from human specimens.
Broader issues that
need further investigation include gaining a better understanding of the
persistence of DNA damage versus transient DNA damage systems; determining the
relative rates of proliferation and apoptosis in conjunction with DNA damage
signals in situ (e.g., in oncogenic mouse models); assessing processes in real time;
conducting longitudinal studies in humans; and understanding the different
mechanisms underlying chemotherapy and radiation. One notable gap that warrants further
investigation involves the distinct differences in the in vitro versus in vivo
responses to DNA damage and therapeutic interventions (e.g., radiation). An approach to gaining a better understanding
of these differences and bridging the gap between these systems may be through
the study of pathway reporters inside tumors (see Tools). Another general challenge to the field is the
standardization of technologies, assays, definitions, interventions, and outcomes
across research to improve the power of between-study comparisons.
Targets, Strategies
The emerging science indicates
that it will be possible to differentially modulate sensitivity to DNA damaging
agents in cancer and normal cells. Practically all cancer cells have mutations
targeting DNA damage response genes. Mutations in these genes are essential for
cancer development and result in rewiring of the DNA damage response pathways
in cancer cells. Because of this difference in rewiring, inhibiting the same
molecule is expected to have different effects in normal and cancer cells. We
need to support research to identify and characterize DNA double strand break
sensors and transducers. These proteins are promising targets for development
of drugs that could specifically enhance the efficacy of current cancer
therapeutics. Research to validate which of these sensors and transducers would
be suitable targets for development of cancer therapeutics should be supported,
including development of prototypical small chemical inhibitors to test proof
of principle. It is expected that the proposed research will utilize a
multitude of experimental systems, including human cells, human cancer tissues,
mouse models and lower eukaryotes, as each experimental system has its own
advantages and disadvantages and knowledge gained in one system can be easily
applied to the others. In addition, tumor cells have different sensitivities to
DNA damage and other apoptotic stimuli.
Sensitization models that focus on unique mutations and/or unique
clinical situations with high sensitivity may help determine the role of DNA
damage and repair in the carcinogenic process.
It is critically
important to explore ways in which sub-lethal pro-apoptotic signals can be
combined to overcome the apoptotic-buffering threshold in cells. Merely refining, exacerbating or increasing
the persistence of DNA damage alone may not accomplish selective tumor cell
death. No amount of damage signaling is
going to provide a therapeutically discriminate pro-apoptotic signal if, as in
many tumors, DNA damage-induced apoptosis has become compromised by the
mutation of p53. Exacerbating DNA damage also carries with it the future
problem of inducing further oncogenic mutations. While it is reasonable to argue that
therapy-induced mutagenic/carcinogenic risk is secondary to curing the
immediate neoplastic disease in patients, it would be
better to avoid the problem completely if a better way of triggering apoptosis
in tumor cells can be found. No tumor
cell has ever been identified that lacks the apoptotic machinery if hit hard
enough. This reflects our understanding
that the programmed cell death (PCD) machinery is highly redundant and
mechanistically defocused to employ several overlapping processes, including
mitochondrial dysfunction, pro-phagocytic signaling,
and caspase activation, each of which is alone
sufficient to ensure death of the affected cell. Therefore, it appears that PCD cannot be lost
by progressive mutation. Rather, tumors acquire refractoriness to apoptosis
through corruption of upstream signaling pathways and through lesions that
raise the threshold at which apoptosis is triggered. Thus, adroit stimulation of multiple
disparate apoptotic pathways, each at a sub-lethal level, might together trigger activation of the apoptotic program
Other research
questions of interest focus on the role of the cell cycle in DNA repair and
cell responses to DNA damage. We must
find out whether blocking the cell cycle is a good clinical strategy.
Specific Recommendations for the NCI:
¨ Analyses
of the human DNA-Damage Response networks (DDR):
o
Complete
the mapping of biochemical and regulatory circuitries that link the cell-cycle
checkpoints, apoptotic and DNA-repair pathways into the hDDR
with special emphasis on putative regulatory nodes.
o
Develop
a quantitative, computer-driven, systems-biology model for the hDDR network in human cells.
o
Complete
(or initiate) the integration of hDDR proteomics,
gene-regulatory and gene expression databases with more dynamic cellular
metabolite and energy profiles as functions of DNA damage.
¨ Identification of Molecular Targets &
Translational Strategies for Enhancing Programmed Cell Death over Cell Survival
in Response to DNA damage:
o
Support
more research on ways to modulate the activity/expression of early-time
(“upstream”) mediator and sensor proteins to activate the pathways of
programmed cell death.
o
Support
more research on the underlying mechanisms of p53-independent programmed cell death, because in most tumors p53 is
inactivated by mutations.
o
Validate,
as targets, mechanisms other than apoptosis that can induce cell death in human
tissue, such as those linked to metabolic pathways that are altered in specific
tumors.
o
Place
greater research emphasis on mechanisms that determine whether dual-function
(“downstream”) signaling proteins (e.g., p53, E2F1) signal survival or death
responses.
¨
o
Conduct
comparative proteomic studies of tumor cells/tissues, progenitor cells, and
normal cells/tissues to elucidate additional anti-apoptotic components altered
in cancer.
o
Characterize
and catalogue molecules from the signaling, checkpoint and apoptotic parts of
the hDDR network as possible therapeutic targets.
o
Develop
predictive profiles for radiation sensitive/resistant individuals; tumors with
or without functional p53, ATM or other key signaling proteins, in order to
tailor DNA-damage-based therapy to the patient.
¨ Reagents to Study and Technology to Study the
DDR in vivo.
o
Develop
a resource library of phospho-specific antibodies or
other “readout” reagents that can provide quantitative measures of DDR
signaling and output pathways (e.g., caspase
induction and apoptosis; focus formation and DNA repair). This should include phospho-specific antibodies that can examine the activities
of the DNA-damage signaling pathways in cultured cells and primary human
tumors, before and after treatment and at different stages in tumor
progression, etc.
o
Develop
small molecule inhibitors of specific targets implicated in the response of
cells to radiation, particularly where the target protein three-dimensional
structure is known.
o
Promote
dynamic imaging in the study of mechanisms of DNA repair in in
vivo and in vitro studies.
o
Develop
programs or incentives to allow for research access to early-phase clinical
drugs belonging to pharmaceutical companies and studies of pharmaceutical
reagents in mouse models of human cancer.
o
Support
application of molecular imaging capabilities to monitor the dynamics and cellular
compartmentalization of the hDDR.
o
Convene
a conference or think tank focused on new tissue markers and modern pathology
to improve connections between investigators in basic science and clinical
research.