See related article by Kastan et al., Cancer Res 1991;51:6304–11.
Visit the Cancer Research 75th Anniversary timeline.
The focus of the DNA damage and repair field prior to 1991 was largely on mechanisms by which environmental carcinogens (e.g., aromatic hydrocarbons, ultraviolet light), ionizing irradiation (IR), oxidative stress, or chemotherapeutic agents cause DNA breakage or damage to DNA bases and how cells could repair such damage. There was little focus on or mechanistic understanding of how DNA damage and repair processes interface with other cellular processes, including cell-cycle progression or signal transduction pathways. The insights described in this article built upon a nascent field developing in studies of yeast where cell-cycle changes or transcriptional changes induced by DNA-damaging events were just beginning to be described (1). Similarly, elucidation of human tumor suppressor genes was just in its infancy, with discoveries about the Rb gene relatively new (2) and the switch of perspectives of the p53 gene from that of an oncogenic driver to that of a tumor suppressor gene (3, 4). This article on p53 and DNA damage responses was published at a time when these fields were in an early stage of development and initiated the new and important field of study that we now call the “DNA damage response” (DDR; ref. 5). This article, and the two immediate follow-up studies (6, 7), were the first critical links to tie together a number of disparate fields, including DNA damage/repair, carcinogenesis, cell-cycle control, tumor suppressor genes, and radiobiology.
Having done my graduate work studying mechanisms of DNA repair and noting that there must be important requisite relationships between DNA repair, DNA methylation, and DNA replication (8), I became intrigued by the necessity of cells to cope with DNA damage in settings where it was also trying to replicate or transcribe its DNA. The landmark observations by Weinert and Hartwell describing a G2 arrest in yeast (Saccharomyces cerevisiae) caused by ionizing irradiation and its dependence on the Rad9 gene led them to coin the term “cell-cycle checkpoint” (1). It was also clear that DNA damage of mammalian cells also caused cell-cycle perturbations (9). I thus became interested in elucidating the mechanisms by which DNA-damaging events could impact on cell-cycle controls in mammalian cells. Until that time, the primary focus of DNA damage–related cell-cycle studies in higher eukaryotes had been on how DNA damage would impact on replication fork progression. However, the Rad9 insights in yeast suggested to me that DNA damage events could also engage signal transduction pathways to alter cell-cycle controls beyond the direct impact on replication forks.
Thus, I set out to identify DNA damage–induced changes in expression of proteins that could alter cell-cycle control in higher organisms. As an initial step, I thought it important to first fully characterize DNA damage–induced changes in cell cycle in human cells. I adapted a flow cytometric technique that had just been described in the flow cytometry literature simultaneously assessing DNA content, utilizing propidium iodide, and DNA replication, utilizing measurements of new bromodeoxyuridine incorporation. Using this assay in human cells at various times after DNA damage allowed me to get a very dynamic picture of perturbations of all stages of the cell cycle (10). This approach to assess cell-cycle checkpoints became quite common in the scientific literature after this article. I initially used UV irradiation as the DNA-damaging agent in these studies, but it quickly became clear that the bulky thymidine dimers could physically obstruct replication forks and major cell-cycle arrests would occur because of this perturbation rather than signaling-induced alterations in cell-cycle control. I then switched to IR, which causes DNA breakage rather than bulky DNA adducts, as the prime DNA-damaging agent to investigate DNA damage–induced changes in cell cycle and in signaling proteins. Using IR as the DNA-damaging agent and the BrdUrd/PI assay for cell-cycle stage and progression permitted detailed characterization of the perturbations and kinetics of cell cycle with and without DNA breakage. Specifically, IR caused discrete arrests of cell cycle in human cells in G1, S, and G2 phases of the cell cycle. Importantly, this was different than cell-cycle changes characterized in yeast, where G1 arrests do not occur and G2 arrests dominate the cell-cycle perturbations after damage. As we later demonstrated that p53 was required for G1 arrest, but not for G2 arrest, this created some initial confusion as yeast did not exhibit this checkpoint. However, full characterization of damage-induced cell-cycle changes in the human cells meant that any future discoveries of candidate gene products could easily be evaluated for their impact on DNA damage–induced cell-cycle control.
The p53 tumor suppressor protein was not an obvious candidate for a focal control point in DNA damage–induced cell-cycle arrest. In 1990, as I initiated these studies, p53 had only recently been reclassified as a tumor suppressor gene (3, 4). However, a report came out in 1990 showing that overexpression of a p53 transgene could cause growth arrest of human cells (11). There were also technical challenges in studying p53 protein, because it was expressed at such low levels in nontumor cells. Immunoblots were relatively insensitive at that time (prior to the development of ECL technology), so p53 protein could not be detected by immunoblots. It was serendipitous that I had developed flow cytometric techniques to detect intracellular antigens (12) and p53 was one of the proteins that I had characterized with this assay (13). I examined changes in numerous different proteins after IR using this flow cytometric assay, but the only protein that detectably changed after irradiation turned out to be p53. This increase in p53 protein levels after DNA damage was validated with 35S-methionine labeling and immunoprecipitation.
Once the observation was made that p53 protein levels increased after irradiation, we were quickly able to demonstrate that p53 function appeared to be required for G1 arrest of cells after irradiation by comparing cells with and without p53 mutations. Interestingly, p53 was clearly not required for either the S-phase or G2 arrests. The kinetics of p53 induction was rapid and consistent with the kinetics of the G1 arrest. In addition, caffeine treatment of cells blocked both p53 induction and G1 arrest after IR, thus further linking these two events and providing the first clue that IR was inducing a signal transduction pathway that was increasing p53 protein levels, which was in turn required for G1 cell-cycle arrest. We also demonstrated that the IR-induced increase in p53 protein levels was a posttranscriptional event as levels of p53 mRNA did not change during this time frame after IR. The identification of p53 as a mediator of these responses gained wide interest in part because it was becoming clear that p53 was one of the most commonly mutated genes in human cancer (14). Thus, defective responses to DNA damage were linked to a commonly mutated gene, lending further evidence of the importance of DNA damage in the genesis of human cancers.
This article was rapidly followed by two other publications from my laboratory that definitively established p53 as a mammalian cell-cycle checkpoint protein (6, 7) and established this DNA damage response as a signal transduction pathway, initiated by DNA breakage (15), involved signaling through the gene defective in the cancer-prone disease, ataxia-telangiectasia (A-T), induced p53 transcriptional targets (GADD45 was the first one we identified), and was required for G1 cell-cycle arrest (7). The next year, another p53 transcriptional target was identified, p21CIP1, an inhibitor of cyclin-dependent kinases, which was clearly implicated in damage-induced G1 arrest (16, 17). It was also soon demonstrated that p53 induction after DNA damage could not only lead to cell-cycle arrest, it was also required for rapid radiation-induced programmed cell death (18), thus suggesting that this pathway could also be an important determinant of tumor responses to radiation or chemotherapy.
The stage was then set for further characterization of this damage-induced signal transduction pathway, which included activation of the gene defective in A-T, the ataxia telangiectasia mutated (ATM) protein kinase, in the phosphorylation and induction of p53 protein (19). This signaling pathway became much more robust as numerous targets of the ATM and related kinases were identified and characterized as critical mediators of a complex DNA damage response pathway (20). As caffeine can inhibit ATM and related kinases, our initial observation of the effects of caffeine on IR-induced p53 induction and cell-cycle changes in this first article now made biochemical sense.
In summary, this first report implicating p53 in cellular responses to DNA damage had a significant and long-lasting impact on the field: it was the first established step in the now accepted DNA damage response pathway; it provided the first functional role for p53, still the most commonly mutated gene in human cancer; it demonstrated the presence of cell-cycle checkpoints in higher eukaryotes; and it provided fodder for decades of subsequent studies elaborating pathways that contribute both to the development of human cancers and to the responses of tumors to chemotherapy and radiotherapy. The experiments were initiated because of insights gained from studies of yeast and demonstrate the importance of applying knowledge from model systems to studies of human disease. Insights gained from years of these and subsequent DDR studies have now led to identification of cancer susceptibility syndromes, improved prognostication of tumor responses to therapies, and are now the basis for the development of novel cancer therapies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.