n this interview, Dr. Michael Kastan talks with ESI correspondent Gary Taubes about his highly cited work with p53 and cancer. Seven papers authored by Dr. Kastan have been cited a total of 7,832 times, placing him among the top 20 most-cited scientists of the 1990s. Dr. Kastan is the chairman of the department of Hematology/Oncology at the St. Jude Children’s Research Hospital in Memphis, TN.

  You seem to have been working on p53 for most of your career. What got you started on it?

I am a graduate of a combined M.D.-Ph.D. program and did my graduate work in the areas of DNA damage and repair and DNA methylation. After I went through my clinical training to become a Pediatric Oncologist, I went back into the lab and worked in the field of hematopoiesis. When I joined the faculty of Johns Hopkins, I started my own lab and initially studied how oncogenes control hematopoietic cell proliferation. Because I had worked on DNA damage responses in the past, I became interested in studying the links between DNA damage and DNA replication. It seemed to me that it would be very important for cells to repair damage prior to replicating DNA.

This was 1989-90, and the major mechanistic work in the literature investigating the relationship of repair and replication had been done in yeast. The rad9 gene had been identified as a "cell-cycle checkpoint" gene. It prevented cell cycle progression from G2 to M if DNA was damaged. But there was really nothing known about control of the cell cycle in mammalian cells after DNA damage. That was what I began to study and I first characterized the cell cycle changes that occurred in human cells after irradiation. With irradiation I was able to distinguish arrests in G1, in S and in G2 phases of the cell cycle; the question that came up is, "what are the molecular changes induced by radiation that cause these cell cycle changes?" I screened a number of proteins that were thought to either positively or negatively regulate the cell cycle, and one of them turned out to be rapidly induced by irradiation, and that was p53.

  Was this the work that became your most highly cited paper: "Participation of P53 protein in the cellular response to DNA damage?" (Cancer Research, 51[23]: 6304-11, 1 December 1991).

That was it. The first observation in the paper was that p53 levels went up after irradiation, and p53 had only recently been characterized as a tumor suppressor gene. For a decade, it had been thought to be an oncogene, but just as I was doing these studies, it had been reclassified as a tumor suppressor gene. And it was known that when you over-expressed p53, it would cause cell-cycle arrest. No one had any idea when the cell would use p53 to cause the cell-cycle arrest. My initial observation was that irradiation increased levels of p53, then I asked whether that contributed to any cell-cycle arrests we saw. I compared cells that didn’t have p53 to cells that did and was able to show that p53 played a role in G1 arrest and did not play a role in initiating the S phase and G2 arrest.

  Why do you think the paper had such a major impact?

These findings immediately tied together a lot of observations. First of all, p53 was known to be commonly mutated in human cancer. Linking p53 to DNA responses in this paper explained why loss of p53 might be important. In effect, if the cell didn’t have p53, it could continue to go through the cell cycle and replicate damaged DNA. So p53 was a potential source of genetic instability and the paper provided the first molecular clue of how the cell cycle was controlled after DNA damage; it also provided the first insight about what signals p53 responded to. One of the reasons the paper made such an impact is that it influenced a lot of different fields. It made a mark in the field of tumor suppressor genes, in radiation biology, in cell-cycle control, in carcinogenesis, and in the general field of tumor biology. In addition, it had clinical implications, both for tumor development and for therapy.

  What was the greatest challenge in doing the research?

One difficult part initially was that the available assays in 1990 weren’t very good for looking at p53. I actually used a flow cytometric assay for looking at the p53 protein, because you couldn’t detect normal p53 by Western blot in those days. The assays weren’t sensitive enough. It was only because I had developed this flow cytometric assay that I was able to do this. I had developed that assay for different reasons. So there was really a lot of serendipity: asking the right question at the right time with the right system, with the right assay. The other major difficulty was gaining acceptance for the ideas—this data led me to push the idea that DNA damage was initiating signal transduction pathways, which were typically thought of as responses to cytokines and growth factors, but not responses to things like DNA damage. In addition, at the time of our initial publications, most of the focus in the p53 field was on its role as a transcription factor and tumor suppressor gene, and it took a little time for the notion that it was a DNA damage response gene product to become generally accepted.

  How did your research progress from that initial observation?

That Cancer Research paper effectively showed that p53 played a role in DNA damage response and in controlling the cell cycle. In our next paper, which we published in PNAS ("Wild-type p53 is a cell-cycle checkpoint determinant following irradiation," PNAS, 89[16]: 7491-5, 15 August 1992), we took it to the next level and actually made isogenic cell lines. We manipulated p53 in the cell and showed it controlled the cell cycle after DNA damage. That’s when we called it a cell-cycle checkpoint protein. A few months later we published a paper in Cell ("A mammalian-cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia," Cell, 71[4]: 587-97, 13 November 1992) that took it to the next stage. In that paper, we did the final confirmation of p53 as a checkpoint protein by using cells from p53 knockout mice; we showed that induction of p53 was abnormal in cells from patients with the disease, ataxia-telangiectasia (A-T). We took cells from these patients, whom we knew were abnormal in how they responded to radiation, and we showed they did not induce p53 abnormally. That said that the A-T gene, whatever it was going to be, somehow signaled to p53. The third part of that paper identified the first gene downstream of p53, called GADD45. We showed that there is a binding site for p53 in the GADD45 gene and showed that its induction was dependent on p53. This result established this p53-GADD45 connection as a checkpoint pathway. You irradiate a cell and somehow a signal goes through ATM (which is now known as the product of the A-T gene) to p53 and that p53 signals to GADD45 and somehow you then get a G1 arrest. We published those three papers—in Cancer Research, in PNAS, and in Cell—all within an 11-month period and we basically established p53 as a checkpoint gene and AT-p53-GADD45 as part of a checkpoint pathway.

  How did you decide where to send the papers for publication?

I had initially sent the first paper to Cell, but it was not even sent out for review. I believe that it was such a new concept that the editor was not sure why it was important. Being only a second-year faculty member and not really savvy at publication "games," I did not argue nor try for one of the other top two or three journals. I was very familiar with the journal, Cancer Research, and felt comfortable sending it there because I thought it would reach the "carcinogenesis" audience, who should be interested in it. The fact is that most people did not understand the significance of the result at that time. I had actually presented the data initially several months earlier at a Keystone Meeting in February 1991. Only a handful of people came by the poster who understood what it meant, and one of them was Ted Weinert, who had discovered the yeast rad9 checkpoint gene. It was his papers that I had read that led me to ask these questions in mammalian cells. He saw the poster and we’ve been friends and talked about checkpoint pathways ever since. Another interesting thing is that same poster session was the first time that telomere length was linked to aging. It was a good poster section. The subsequent two papers on p53 and DNA damage responses were published in PNAS and Cell, respectively, and represented increasing sophistication and clarification of this signaling pathway.

  Your recent papers all concentrate on the ATM gene. When did you switch and why?

The 1992 Cell paper established that whatever the gene was that was defective in A-T, it signaled to p53. The gene wasn’t cloned until Yossi Shiloh at Tel Aviv University accomplished this difficult task by positional cloning in 1995. Then it took another two years to have a full-length cDNA that could be manipulated. During that time we continued to work on p53 and we showed that p53 gets phosphorylated in response to DNA damage and we mapped one of the sites that gets phosphorylated. We also showed that that phosphorylation does not occur normally in A-T cells, so it implicated the A-T gene in this phosphorylation event somehow. Then we developed assays that were able to demonstrate that ATM directly phosphorylates p53 on this site in cells. So now it’s a direct signaling pathway that radiation activates the ATM kinase: ATM phosphorylates p53 at this site and also phosphorylates other proteins. P53 gets induced and then causes either cell cycle arrest or apoptosis. So that basically explains the switch, although skipping a few things along the way.

  If you were to look 10 years in the future, what would you like to achieve?

As a scientist, I’d like to understand the molecular steps in these stress-signaling pathways and determine what accounts for cell-type specific differences in responses. As a clinician, I want to be able to use these insights to enhance therapy. And since these signaling pathways are those induced by chemotherapy and radiation therapy, the better we understand the pathways, the more effectively we can use our therapeutic agents. In an ideal world, I’d like to be able to take these basic science discoveries and develop new approaches to therapy and optimize current approaches to therapy. In a sense, we’re already doing that to some extent because we already know the importance of p53 in some tumor types. If tumors mutate p53, there are some therapies the tumors don’t respond well to, and we have a better understanding of why that is. And we’re identifying other molecules to try to target. For example, the ATM gene becomes a wonderful potential target for radiosensitization. If we inhibit its function, cells become radiosensitive.

  What other work in your field do you consider promising and noteworthy?

I think that understanding the tumor microenvironment is going to be a big area where we need to make breakthroughs. The microenvironment includes the vasculature: blood vessels and ongoing angiogenesis; it also includes the effects of hypoxia and oxidative stress as well as nutrient and glucose deprivation – all these things that put the tumor cells in a unique environment. Understanding the unique microenvironment of the tumor could provide us with a handle of a common selective difference between normal cells and tumor cells, which is what we need to exploit to get an optimal therapeutic treatment. We have to selectively kill tumor cells, and here the selective difference would be in the distinctive microenvironment of the tumor cells. The microenvironment is something I would like to develop new ways to approach, but it’s a very tough area because there are a limited number of good models for what actually happens in vivo. Even animal models are not perfect for this. Xenografts are not totally representative of the situation and the differences between humans and rodents in tumorigenesis are exemplified in many different ways. Nevertheless, elucidation of the microenvironment is a "biggy."

  Are you optimistic that cancer therapy will improve dramatically?

Frequently when I give talks, especially if there are lay people in the audience, I show a slide of a Gary Larson cartoon: cavemen looking through their hands at a rock. The caption says "early microbiologists." And I say that this is the stage we were at in 1971 when Richard Nixon declared the war on cancer, because we had neither the technology nor the knowledge base to address the problem. It was naive to think we could make an impact because we didn’t even understand the disease we were dealing with. The majority of clinical advances made in the last 30 years have come about because of well-designed clinical trials, not because we’ve designed therapies specific for tumor cells. However over the last 30 years, and especially over the last 10 years, what we’ve learned about tumors has increased dramatically. Now we’re in an exponential phase of understanding the molecular biology of tumors, and that will definitely lead to new ways to approach therapy. Then we have to go back and do the same thing we did with the original therapeutic agents, which is to do innovative clinical trials again with the new therapies. Hopefully we can speed things along at that point. So yes, I am optimistic that this incredible amount of understanding that has occurred in the lab will pay off. We are just at the cusp of being able to apply this knowledge in the clinic. That’s why we see all the excitement. We can now understand what questions to ask and we can begin to answer them.

Michael Kastan, M. D., Ph.D.
St. Jude Children’s Research Hospital
Department of Hematology/Oncology
Memphis, TN, USA