Resistance to Therapy: p53 and Chemosensitivity in Gastric Cancer by James M Ford, MD

SLIDES & TRANSCRIPTS
Tuesday, March 6

Resistance to Therapy: p53 and Chemosensitivity in Gastric Cancer
James M. Ford, MD

Slide 1: p53 and Chemosensitivity

I am going to be talking about P53 and chemosensitivity. I think I have two challenges. One is being the last speaker in a long session in which all the speakers actually were here. The second is to try to briefly try to summarize the role of P53 in chemosensitivity, which is a very complicated subject. For those of you anxious to get to the airport, I can summarize the talk by saying that, despite truly an enormous literature, the specific role of P53 in clinical chemosensitivity really remains not well understood, it there is any all, regardless of its fairly clear role in carcinogenesis and prognostics.

Nevertheless, the biology and model systems for studying P53 in sensitivity and drug resistance are very provocative. What I am going to try to do this morning is to try to talk about some of our newer understandings of the molecular mechanisms of regulation of P53 and its targets, and a couple models that are provocative in ways to try to exploit P53 in this process.

Slide 2: p53 and Cellular Response

Let me start by summarizing some of the things you have already heard. P53 in normal cells, normal mammalian systems, is a central player in response to DNA damage of a variety of sorts. And traditionally this has been studied with types of radiation, such as X-irradiation, UV-irradiation, but this is equally applicable to most of our commonly used chemotherapy drugs, which are DNA-damaging agents.

So in a normal cell, P53 at very low levels, usually barely detectable, but following DNA damage is stabilized and activated and the protein level increases, which is the basis for the immuno-peroxidase studies of tumor tissues. And after activation it serves as a transcription factor to regulate a variety of downstream genes which are involved in important biological effects of cell cycle check points, apoptosis and DNA repair. Therefore, in tumors mutant or null for P53, these downstream effects are defective and you can certainly see why this is central to the process of tumor formation and also how this certainly may regulate the response to chemotherapy in cell death.

Slide 3: Frequency of p53 Gene Alterations

So why is P53 important? As is often cited, it is mutated in nearly 50 percent of many common tumors, and you can see many of the GI tumors fall into that range, including stomach cancer, as in this graph.

Slide 4: p53 Mutations

More close studies have confirmed this. Indeed, in most esophageal, GE-junction and gastric cancers, more than 50 percent of them over-express protein.

Slide 5: p53 Gene Map

The region where the majority of known mutations occur is, in fact, the region critical for transcription factor activity and DNA binding activity, suggesting that the effect of these mutations, therefore, is to disrupt that activity. But it is important to note that P53 has other activities in terms of binding other important proteins, though very few mutations occur in those regions.

Slide 6: p53 Signalling Pathways

This is a larger schema of the regulation of P53 both upstream and downstream, and I think it is one of the major points I want to make today, which is that simply looking at P53 status in isolation is not surprisingly confusing in terms of trying to make predictions, given this very complicated regulation. So in response to a variety of stresses important for today, types of DNA damage, but also many other stresses such as oncogene activation, hypoxia, P53 is activated through a huge number of upstream factors that I am not even showing on this slide, and don't want to get into today, but serve to phosphorylate, acetylate, regulate P53. As we well know, P21 and other genes are involved in cell cycle. There is quite a list of genes involved in P53-dependent apoptosis. We've heard of newer genes, such as PERP and pig genes, P53-inducible genes. And what I will tell you about briefly, since this is not as known information, are several genes directly involved in DNA repair and some of the processes that you have just heard about.

Slide 7: DNA Damages

So I want to spend just a few minutes talking about DNA repair since we have heard this brought up with respect to chemotherapy agents, particularly cisplatin, and how P53 can regulate this. The first point is that there are as many DNA repair processes as there are types of DNA damage. This slide just simply demonstrates some of the many types of DNA damage and strand breaks from x-ray, various alkylation and adduct damage, oxidative-induced DNA damage.
The mechanism of DNA repair that I am going to talk about today is nucleotide excision repair, which has been historically studied in response to UV irradiation, which produces dimers, but is also relevant to the drug cisplatin and various alkylator agents.

Slide 8: Substrates

So the commonly known substrates for the repair process, and I will show you mostly data involved with UV, but importantly also inter-strand cross-links, alkylation and oxidative damage. The regulation of nucleotide excision repair therefore may very well affect the cellular response to these types of chemotherapeutic drugs.

Slide 9: Nucleotide Excision Repair

The process for repair is simply shown here, which involves several steps. One is recognition of a DNA-damaged adduct, which can be caused by a drug such as cisplatin, excision of this whole piece of DNA that contains that damage, and then re-synthesis of the DNA to replace it based on the template strand and restore the normal DNA sequence.

Slide 10: Important Enzymes

In humans this enzymatic process is very complicated and I just want to point out a few of the enzymes important in this. Several of them have been discussed already and I will discuss several more today. The initial recognition step is activated by two complexes, one called XPE and one called XPC, and this contains a gene called P48. Following this recognition, this complex recruits a number of other genes and complexes, which you can see here, including this whole TF2H complex and I just want to point out ERCC-1, which you were just hearing about as a predictive factor for platinum sensitivity. As you can see, ERCC-1 is just but one of a large and highly regulated complex, and so how this one gene plays a role in this larger one still remains to be sorted out. What my lab and others now have identified is that both of these very upstream recognition complexes are transcriptionally regulated by P53. So XPE and XPC are DNA-damage inducible, and that is through a P53-dependent process, and in P53 mutant cells, the expression of these is deficient and those cells are actually deficient in nucleotide excision repair.

Slide 11: Two Pathways for DNA Excision Repair

So this global nucleotide excision repair pathway I am talking about is consistently deficient in most of the P53 mutant mammalian cells, though those cells retain another repair pathway. I don't have time to get into that uses many of these enzymes, but specifically targets lesions within the transcribed strand of expressed genes, and this is called transcription-coupled repair. I bring this up only because I think this is a potentially important target in P53 mutant cells, in that they have lost one repair pathway and are relying on this, making this a potentially susceptible target for interventions.

Slide 12: Effect of HPV E6 on Strand-Specific Repair

This is some data showing an example of this in primary fibroblasts. This is just a fibroblast cell line from humans which have been transformed with the human papilloma virus E6 gene, which serves to target the P53 protein for degradation and knockout. SO this is functionally a P53 knockout cell. This is just a graph showing the repair of UV damage in global genomic DNA in the normal cells, which is quite good over 18 to 24 hours, and is completely deficient in the P53 knockout cells. Whereas this repair of a transcribe gene remains the same.

Slide 13: Effect of HPV 16 E6 on Clonogenic Survival

This results, in these types of models, in increased sensitivity to DNA damage. So these are the same cells showing a significant increase in the sensitivity by clonogenic assay to UV radiation. So in this model, P53 mutant cells are now hyper sensitive to DAN damage. I want to focus on this because we find different results, depending on different cell lines, with regard to P53 status.

Slide 14: p53 and Cellular Response to DNA Damage

Let me just summarize this by saying that the model is now being fleshed out in that many of these downstream P53 target genes, P53 transcriptionally regulated genes are now being identified, that confer these downstream biological effects of the cell cycle check points, apoptosis and excision repair. But the regulation of these and where these play an important art in which tumor types, in which cell types, is very complicated and I think explains, in part, the diversion effects we see of P53.

Slide 15: p53 and Cellular Response to DNA Damage

Let me then turn to talking about some models for P53 dysfunction and how we can try to predict the effect of this on drug sensitivity. If we go back to this general model and think about what happens when P53 is mutant or lost in a tumor cell line, on all these downstream effects, we can start to see how it is very complicated to predict chemosensitivity. One obvious thought is that loss of P53 interrupts DNA damage-induced apoptosis so that these cells will undergo less cell death after a chemotherapeutic drug. But at the same time, loss of P53-dependent cell cycle check points or P53-dependent DNA repair may have just the opposite effect. So it is the regulation of all of these downstream effects and which ones have a predominant role in any tissue that we need to understand better in terms of trying to predict these effects.
Just to make an example of that, let me show you several models so far.

Slide 16: Cellular Models

Some of the seminal work in this area, and what has resulted in one of the dogmas of the fact that we think P53 is a drug mutant that results in drug resistance comes from the early work of Tyler Jackson and Scott Lowe published in Cell and Nature in the early nineties, in which murine embryonic fibroblasts from P53 knockout cells were transformed with oncogenes, and those cells exhibited a decreased sensitivity and apoptosis following a variety of DNA damaging agents -- X-rays, various chemotherapeutic drugs. Similarly, thymocytes from these P53 knockout mice were quite resistant to X-irradiation compared to wild type thymocytes. So this was a very nice model and provided the basis for all of the following studies, but it is important to point out that these are not necessarily perfect models for cancer cells.

Slide 17: Cellular Models

More recently and potentially more relevant to human cancer, were important studies done by Fred Bunns in the Vogelstein lab, creating homologous somatic knockout of both P53 and P21 in a colon cancer cell line. This was published a couple of years ago in JCI. In this cell model, those P53 null colon cancer cells showed resistance to 5FU, though they actually were more sensitive to X-rays and adriamycin. Again, though, it is important to look at the cells that this was done in. These were HCT-116 cells, which are actually mutant for the mismatch repair gene, MLH-1. So these exhibit micro satellite instability. They are a mismatch repair deficient cell and the loss of P53 in a mismatch repair defective background, as you know from colon cancer, is very rare. It is very odd to have both of those things in a single cell line. So again this model lacks certain aspects to know how to predict what this means in the clinical situation.

Slide 18: Fibroblast Cellular Models

Now my lab and Al Fornice's lab and others have done a lot of work looking at a lot of work looking at cell fibroblast cells lines, as I showed you, and the effects of P53 loss on DNA repair, and furthermore how that affects cell death. An as I showed you in the E6 transformed cell lines, those DNA repair deficient P53 null cells were hypersensitive to UV radiation and that is also true for cisplatin. With Al Fornice we showed that P53 knockout murine embryonic fibroblasts are also deficient in DNA repair, are more sensitive to UV radiation and several chemotherapy drugs shown here.

Slide 19: Sensitivity of Murine MEFs

So this is IC50s in a variety of mouse MEF knockouts, but you can see the P53 homozygous knockouts are quite sensitive, to cisplatin, to melphalan, compared to the wild types. Now this data from our groups has been used quite extensively by others to suggest that, therefore, P53 mutant cells are actually more sensitive to DNA damaging chemotherapy drugs. I just want to make the point that I think that is too big of a jump. You cannot make that. These are not good models for tumors., These are models for DNA repair in fibroblast cell lines. So the real question is what happens in tumor cells and in clinical patients.

Slide 20: Human Tumor Cellular Models

In terms of human tumor cell models there is a huge amount of data and the bottom line is that you can find any result that supports any hypothesis you want in one cell like or another, and it is very confusing given the genomic instability in those cell lines and the many other mutations they all have to sort out what the direct effect of P53 is.

Slide 21: Clinical Tumors and Correlations

In terms of clinical data it is confusing as well. Again, there is a huge literature out there on measuring P53 status in various clinical samples and trying to correlate that with outcomes and drug sensitivity, but there is a whole series of caveats in interpreting those. One I brought up already is stating for P53 is potentially misleading because not seeing expression of the protein in your sample can either be due to the fact that the protein is wild type or that it has been deleted altogether. So opposite effects. So sequencing is really critical in this, but of course this is difficult and rarely is it done in the whole gene and so there are very few studies in which there is a large amount of data on this.

And then there are many other problems that I don't want to get into too much today, but different mutations within the gene can have different effects. There are genotype-phenotype relationships. There are dominant negative mutations in P53 and then all the other upstream and downstream factors that I have been talking about. So I think it is very difficult to interpret from these straightforward P53 measurements in a tumor cell line what the result is. Nevertheless, the general trend in the literature is that mutations or loss of P53 tend to result in resistance to most of the commonly used DNA damaging drugs, though this has hardly been proven.

Slide 22: p53 and Cellular Response to Damage

So I want to go on for the rest of the talk and finish by just telling you about some data from our lab, trying to look more specifically at some of these downstream biologic effects and predict how they may affect drug sensitivity, particularly with regard to DNA repair and cell cycle check points.

Slide 23: p53 and p21 Protein Levels

To do this we developed one model in which we could regulate the expression of wild type P53 in a cell line very carefully and thus ask more careful genetic question. This just shows that, in a P53 null human fibroblast we stably expressed a wild type P53 regulated by a tetracycline responsive promoter. So simply by manipulating tetracycline in media you can induce P53 and ask what the effect is in various experiments,

Slide 24: DNA Damaging Drugs

and this just shows the sensitivity to several DNA damaging drugs in these cell lines. The blue lines are cells which are mutant for P53, whereas in the same cell line, when one induces wild type P53, we see an increased sensitivity to doxorubicin, to cisplatin, to nitrogen mustard.

Slide 25: Anti-Microtubule Drugs

However, in the same cell model we see the opposite effects on sensitivity to taxol. SO here we see a profound increased sensitivity in P53 mutant cells to the anti-proliferative effects of taxol compared to the P53 wild type cells. We did not see this from another microtubular inhibitor, vinblastine.

Slide 26: UV-Sensitivity

So that was a fibroblast model. We have now done this in gastric cells lines we actually got from Gary Schwartz, and you have heard about these already. These are some of the same cell lines. So again we have found in a panel of gastric cancer cell lines, the GT5 cell line you heard about, which is mutant for P53, and another gastric cell line, MK74, which is a wild type for P53. Again these mutant P53 cells lines are DNA repair deficient. They are deficient in the expression of those P53 regulated repair genes I told you about. And as you can see here, they are quite sensitive to UV compared to the P53 wild type cells. However, this does not translate into something simple, like cisplatin sensitivity, as you might expect, making the situation more complicated in cancer, as we would expect.

Slide 27: Chemosensitivity in Gastric Cancer Cell Lines

So here you can see antiproliferative curves to cisplatin and doxorubicin in these two cell lines, and we see very little difference in drug sensitivity. If anything, the P53 mutant cells ar slightly resistance to platinum.

Slide 28: Chemosensitivity in Gastric Cancer Cell Lines

But again, this taxol effect holds up. So the P53 mutant gastric cancer cell liens are several fold more sensitive to taxol in experimental microtubular inhibitor, epithiolone, than the P53 wild type cell.

Slide 29: Cell Cycle-p53 Induction

What is the mechanism of this? This has not been fully figured out yet. Our lab has been interested in a P53 regulated mitotic check point important for taxol responses, and this just shows some flow cytometry data showing that in P53 wild type cells, in response to taxol, these cells tend to undergo an arrest in G2M and post-mitotic state.

Slide 30: Cell Cycle-p53 Suppressed

Whereas when wild type P53 is suppressed, when these are mutant P53 cell lines, following treatment with taxol, these cells proceed through G2M and cycle back into G1 and start replicating and this seems to trigger a massive apoptosis in those cells, explaining that increased sensitivity. Now why this occurs molecularly remains to be sorted out,

Slide 31: p53 Mitotic Checkpoint

but as I said there appears to be therefore a role for a P53 mitotic check point, in addition to the other known P53 check points, in response, in regulating microtubular structure and chromosome alignment and in the absence of functional P53, these cells exit this mitotic check point and result in aberrant DNA replication, polyploidy and this endo re-duplication seems to result in apoptosis.

Slide 32: p53 Checkpoint Genes

So I think this points out another set of downstream P53 targets for further exploration, in addition to the ones we have talked about so far. And these are genes that appear to be involved in this mitotic spindle check point regulated by P53 and these are starting to be identified, and they include the MAD genes and the BUB genes, which are homologous of yeast genes involved in that process.

Slide 33: Slide not available

Let me just finish by saying we are trying to explore this on other levels as well and this is just an example of using micro-array analysis to try to look in a more global way at the effect on gene expression of things like taxol in cells line in which you can manipulate P53. This just shows an example of a cluster of genes induced by taxol specifically in P53 mutant cells, including some of these G2M and mitotic check point genes, cyclin A and CDC2, which you have heard about. I don=t want to go into the data of this, but I think that exploring the overall effect on expression and activity of these genes in going to be key in figuring out the effect of P53 in this process.

Slide 34: Translational Research Program

So let me finish by just saying that, like many of you, we are using these leads to try to help identify gastric cancer patients who may or may not respond to certain treatments. In several pilot studies at my institution we are looking at trials of adjuvant or neoadjuvant chemotherapy in gastric and esophageal cancer in which we have samples of these tumors and in which we are planning on characterizing some of these molecular markers, as well as looking at overall gene expression profiles in an attempt to try to better identify which tumors may respond better to, for instance, platinum-based regimens versus taxane-based regimens, this based on some of the pre-clinical data I just showed you.

Slide 35: Targets for Modifying Drug Sensitivity

So let me just finish by pointing out that I think there are still a number of novel targets related to P53 that have yet to be fully exploited for therapeutic gain, but we need to think of them somewhat in terms of whether they are in tumors that are wild type for P53 or mutant. So for instance, in a wild type P53, with in tact DNA repair, these may be important targets to knock out since theoretically these tumors may retain some of their P53 dependent cell death mechanisms. As I pointed out, I think some of these G2M and mitotic check point genes will be important as well.

On the other hand, in the very common situation of P53 mutant cells, the obvious goal is to try to restore apoptosis through those genes, but very difficult to do. So some other approaches, such as exploiting the check point dysfunction, which I just showed you with the taxol data, or potentially inhibiting this remaining repair pathway, transcription coupled repair, would take advantage of the Achilles= heel of these P53 mutant cells.

Slide 36: DNA Damage Response Signal Transduction Pathway

So I want to finish with that and just show you the complexity of the situation in terms of the regulation upstream, the transduction of the signal, and ten the transcriptional effect, all regulated through this central P53 pathway and having profound downstream effects on DNA damaging chemotherapeutic drugs