Present Status Pathogenesis by Eric Fearon MD, PhD

SLIDES & TRANSCRIPTS
Tuesday, February 15, 2000

Present Status Pathogenesis
Eric Fearon, MD, PhD

Slide 1:

We are going to begin this morning by discussing the present status very briefly with some overviews on pathogenesis markers and therapy, and our first speaker is going to be Eric Fearon.

Eric is at the University of Michigan, associate professor in the Department of Medicine and Genetics, and he is going to talk about pathogenesis of colon cancer.

DR. FEARON: Good morning. I would like to thank Dr. O'Connell and Dr. Mayer very much for the kind invitation to visit. I guess I was a little less grateful last night around one-fifteen, I guess I should say this morning around one-fifteen as I was coming in from Dulles, but it is great to be here, and I hope to learn a lot today.

If I could have the first slide?

I thought I would start with this slide which is actually a rather old slide. I think it was from 1989, probably not visible to those at the back, so I will just very briefly point out what it is. It is actually an ad from the New England Journal of Medicine that ran and described at the top a paper that was published in 1988, by Burt Vogelstein and colleagues at Johns Hopkins describing genetic changes in colorectal tumors and it said that you needed to examine manuscripts like this published in the New England Journal before you examined, I guess, color enhanced versions of your patient's colon and, although I don't practice clinical medicine and didn't train very long myself in medical school, this seems a little bit anatomically incorrect. So again, whether you might want to take this seriously or not I hope that by the end of the talk and certainly by the end of this meeting that you will, in fact, if not already be convinced that understanding of the pathogenesis of cancer has major importance for thinking about clinical decision making.

Slide 2:

So what are some of the advances in cancer genetics? Certainly I won't be able to cover all of these. I want to spend the bulk of my time this morning talking about what we know about two different genetic mechanisms that give rise to colorectal tumors. Clearly there are significant implications when one understands the pathogenesis for understanding what individuals in the population are at the very highest risk of cancer development, what are some strategies that one might pursue to prevent cancer, what are some strategies that one might pursue to prevent cancers;, what strategies could one use, for instance, using somatic genetic alterations to look for early detection strategies, and certainly the hope is by understanding the pathogenesis well one might be able to identify novel targets or at least novel treatment strategies and combine these either with existing or combinations of novel strategies together.

Slide 3:

With regard to colorectal tumors, one of the things that really has been important in understanding how genetic changes contribute to development has been the appreciation that colorectal cancers arise from precursor lesions, adenomatous glands shown here which are the residual adenoma in this case that one can prove in molecular terms contain at least some of the cells that gave rise to this colorectal cancer.

So by tracking the genetic changes present in adenomatous lesions and comparing those to the changes present in carcinoma one can identify changes that arose early in the process. By defining changes that are present only in the carcinoma, not in the adenoma one might get a sense of the changes that contribute to progression, and this has been put forth in models that I will show you at the end in sort of a summary.

Again, one of the major questions out there still is is it only a small fraction or one of these cells very early in the development of the adenoma that is the precursor lesion for carcinoma or do in fact these lesions progress at a relatively late stage to carcinoma? And it is probably a relatively gray area still, with regard to exactly how early these populations of cells diverge, that is adenomas and carcinomas, during the development of the vast majority of colon cancers.

Slide 4:

So the models that have been put forth in simplest terms are consistent with models proposed even in the fifties by Leo Folds and Peter Knowell, and they are models of clonal evolution. I think the two important points to appreciate are the notions of somatic mutation and clonal selection, again, clonal selection being a very important one such that the genetic changes that arise at any one point in time will only be selected for and contribute in an important way, a causal fashion to carcinoma development if, in fact, they give rise to either more robust proliferative and survival properties in the cells.

So this genetic change arose, set the cell apart from its normal counterpart but certainly it wasn't until, for instance, in this schematic representation, a relatively last stage, when significant proliferative and survival properties were associated with outgrowth.

So why is this important? It is important because there are, in fact, preferred orders to the genetic changes that arise in colorectal and other cancers such that in colon, for instance, p53 mutation is a relatively late event that is only selected for, and undoubtedly mutations can arise early, but there is only a selection for p53 mutation probably at a relatively later stage. Whereas in other cancer types, for instance, esophageal cancers, the mutation seems to be associated with relatively earlier stages of the disease.

Slide 5:

So I want to focus my time on just two inherited syndromes, and I guess the understanding that they have offered us in sporadic forms of colorectal tumor development, familial adenomatous polyposis coli involving the APC gene on chromosome 5 and a collection of syndromes known as hereditary non-polyposis colorectal cancer which involve, as best we can tell in the vast majority of cases, germ line mutations in the DNA mismatch repair gene.

I will talk first about familial polyposis and the role of APC not only in inherited form of polyposis but the involvement of this gene in the vast majority of sporadic lesions, adenomas and carcinomas, then talk a little bit about polyposis and then try to wind it all up in some kind of summary touching on a number of other genetic changes.

Slide 6:

So why would I want to tell you about these syndromes? In aggregate they probably account for only about half a percent of all colorectal cancer cases, that is familial polyposis, and HNPCC probably only around 2 to 3 percent for the classic forms of this syndrome.

Various alleles in these genes may contribute to a fraction of these low penetrance genetic cases -- the exact proportion of all cases isn't exactly clear -B but again in the inherited setting they are relatively uncommon. Excuse me, in the aggregate of all colon cancer cases inherited mutations in these genes are relatively uncommon. However, they are evolved in the vast majority of all colon cancers as a result of somatic mutations of the gene.

Slide 7:

With regard to familial adenomatosis polyposis coli the gene was identified in 1991, molecularly cloned in 1991, by three groups, Ray White's group at the University of Utah, Uskey Nakamora at the University of Tokyo and Ken Kinzler and Burt Vogelstein leading a group at Johns Hopkins and the classic form of the disease is associated with the development of hundreds of thousands of adenomatous polyps arising by the late teenage years, early adult years and the fact that so many of these lesions arise and that medical management isn't sufficient to prevent their progression to carcinoma necessitates the removal of the patient's colon to prevent the development of colon and rectal cancer.

Slide 8:

The gene when it was cloned was a bit of a mystery in terms of what its function was. It encoded a large protein of around 300 kilodaltons, 2843 amino acids, and it didn't bear a strong similarity to any proteins in the database, and it is only through work in a variety of model organisms, as well as in mammalian cells, that some insights into its function have been obtained.

The important feature of the mutations is that they lead to premature truncation of the protein product. Germ line mutations, the vast majority, probably greater than 95 percent of the mutations associated with classic forms of polyposis lead to truncations of proteins prior to its halfway point in synthesis, and somatic mutations have essentially the same spectrum as the germ line mutations, again leading to premature truncation with intriguing cluster mutations in this region of the gene, again, truncating mutations.

There have been probably one-half dozen to maybe 10 different proteins that have been found to bind to the APC protein but the one that I am going to focus on initially to tell you about with regard to APC function is beta catenin of which APC seems to have multiple different binding regions and the ability to down regulate beta catenin levels in the cell.

Slide 9:

The models that have arisen are really not so complicated. So I will just take a moment to take you through the slide. In normal cells APC is broadly expressed in normal cells in a relatively low abundance protein. Beta catenin in contrast is a broadly expressed protein as well but is of considerably higher abundance, probably 1000 times the abundance in total level of APC.

The bulk of beta catenin is actually present at the cell membrane where it links the E-cad cell adhesion molecule to the actin corticocytoskeleton. A fraction of the beta-catenin protein is in equilibrium and not bound at the membrane but seems to be free in the cytosol, and this is the fraction that the APC protein seems to regulate.

APC doesn't regulate beta catenin on its own. It regulates beta catenin in concert with several other proteins including a kinase known as GSK3 and another protein that is an important cofactor on APC's regulation of beta catenin known as axin or conductin. Together these proteins are involved in phosphorylating the amino terminal region of beta catenin at multiple serine threonine sites, and these serine threonine phosphorylation events appear to target beta catenin for degradation as a result of ubiquitin ligases recognizing preferentially the phosphorylated form, transferring a ubiquitin polypeptide to beta catenin and that targets it for degradation via protein degradation machine known as the proteosome.

So again, the net consequence of APC's function seems to be to regulate the cytosolic as well as the nuclear abundance of beta catenin, and the consequence of APC inactivation is associated with increased levels of beta catenin in the cytosol and nucleus and an increased targeting of beta catenin to interact with a transcription factor family known as the TcF or leff factor family.

Why is this important? In colon cancers with APC defects there is evidence for constitutive activation of TcF transcription. In normal epithelial cells it is only activated in response to certain signals and in colon cancer cells with APC defect TcF transcription seems to be constitutively on.

This is the case in about 80 percent of all colorectal cancers and similar percentage of early adenomas. What makes this notion that APC has a critical role in regulating beta catenin more compelling is the fact that in a subset of the colon cancers that lack mutations in APC there are actually mutations in the amino terminus of beta catenin that render it resistant to phosphorylation of GSK3, because they are at serine threonine sites. They are at a point mutation changing the amino acids that are actual deletions of the amino terminal region. The net consequence again here to deregulate TcF transcription to constitutively activate it and to drive presumably important genes involved in proliferation or survival. I will tell you in just a moment about some other candidates, but C-myc has been suggested based on work from Burt Vogelstein and Ken Kinzler's lab.

So again, this gives you some sense of the notion. I want to take just a moment and tell you about how we pursue studies in the laboratory trying to identify additional genes and really trying to validate this model and whether it is perhaps even an important place for intervening in colorectal cancers.

Slide 10:

The strategy that one might use in the lab and that we have used in our lab is you take cancer-derived mutant alleles such as a mutation at a serine position at 33, changed to a tyrosine; it is again a potential phosphorylation site but not for this kinase GSK3, as well as amino terminal truncations and then to ask about which domain of beta catenin might be important if these alleles will behave as oncogenes, as might be predicted from that previous model, and ask about their ability to have certain features in immortal cell lines, much the same as the classic oncogene assays, in those case though using an epithelial line for transformation.

Slide 11:

The strategy here is to take mutant or variant cancer derived mutant alleles of beta catenin and ask if they, in fact, confer neoplastic growth properties upon cells when transferred into them.

You can see Lac Z, which is just a control gene, doesn't cause the cells to form piled up areas on the plate. Wild type beta catenin doesn't transform, but various cancer derived mutants transform to a greater or lesser extent. One can ask about the domains of beta catenin that are required, for instance, the region that binds to these TcF transcription factors or the C terminal regions involved in transcriptional activation.

Slide 12:

So one comes away with the notion that beta catenin actually meets the expectation. Again, these alleles were predicted to begin a function, and they behaved like oncogenes when tested in these kinds of in vitro assays, and you can ask about the requirement. Are TcF factors despite this correlative data, are they actually critical for transformation and what about, for instance, C-myc suggested as a target gene?

Slide 13:

So those kinds of strategies, again, probably not visible at the back but you just have to take my word or maybe that of colleagues in the front that there are significant differences on some of these plates. This is, again, a focus-forming assay using a strategy in which we use a so-called "dominant-negative form" of TcF, again, not a particularly attractive drug but a nice strategy in the laboratory that will bind the nucleic acid but won't bind beta catenin. So it blocks the sites that endogenous TcF might bind to and might be required for TcF to transform via beta catenin's effect.

S33Y transforms in the control line but not the dominant negative and similarly with amino terminal truncation, again, implicating TcF factors as critical partners in beta-catenin's ability to transform.

Slide 14:

What about C-myc? We have asked about C-myc expression levels in beta catenin transformed cells and found to our surprise that they are not uniformly elevated. In some cases they are considerably elevated compared to the control, but in other cases they are not activated. So this might suggest that myc is not a required factor based on this correlative data.

Slide 15:

We pursue these kinds of dominant-negative approaches and actually show that if you inhibit myc function with a dominant-negative mutant form, beta catenin will still transform.

In this case, in contrast to TcF, myc function doesn't seem to be required and may not even be a direct target of beta catenin at least in this system.

Slide 16:

This is the kind of strategy I have pursued, and our lab and others have pursued. As is always the case, models like these, although attractive, are clearly overly simplified and they don't take into account some of the other proteins in the cell that might, for instance, APC might interact with, one of which is a very close relative of beta catenin known as gamma catenin that has similar functions in cell-cell adhesion and was at least in invertebrate and simpler model organisms thought to have similar functions in this signaling process in the so-called "WNT" or wingless pathway. But because it wasn't mutated in cancers with APC mutations, it was considered a minor player and has been fairly much overlooked,

Slide 17:

but I will just show you a little bit of data to suggest that gamma catenin may be just as interesting and may offer some explanation for why there is this marked discord. Why are APC mutations present in 70 or 80 percent of colon cancers and beta catenin mutations present in somewhere around two to four.

As it turns out the gamma catenin protein when tested in these kinds of transformation strategies actually is an oncogene as well, and it does not need to be mutated at the amino terminus to be a transforming allele.

Wild type beta catenin doesn't transform. The mutant version does. This is wild type gamma catenin. It generates really quite large foci, perhaps a bit fewer in number but larger in size, and amino terminal mutations similar to those found in beta catenin at the similar position are also transforming, and if you abrogate certain of the domains in gamma catenin you inhibit its transforming function.

Slide 18:

The interesting thing about gamma catenin is although like beta catenin it requires TcF function, in contrast to beta catenin, which didn't require myc for transformation, gamma catenin does require myc.

Slide 19:

Consistent with that notion actually gamma catenin always leads to elevated myc expression in the transformed lines in probably 50-fold the level in the control lines in contrast to beta catenin.

Slide 20:

So why is this important? I think the observations at least provide us with some hypothesis that needs to be further explored, that APC has undoubtedly more than one function in the cell in colon cancer cells. In addition to the fairly well established function of regulating beta catenin's activity and inhibiting beta catenin's ability to activate transcription, I would like to suggest to you that it also probably regulates gamma catenin and that the consequence of deregulation gamma and beta are actually distinct, that some of the target genes activated as a result of gamma's deregulation, for instance, myc appear to be considerably less enhanced in their expression as a result of beta's deregulation. The other thing that I won't have time to tell you about is there are actually a number of genes that have been implicated as target genes and might represent intriguing or at least potentially promising therapeutic targets in large part because APC mutations are present in upwards of 80 percent of all colon cancers.

So again, if they are activated either as a result of beta deregulation, gamma deregulation or both, they might be potential targets for therapeutic intervention because of the sheer frequency of the activation in colon cancer. They include MMP7 or matrolysin PPAR delta, a recently suggested target that is inhibited by sulindac and may account for the ability of sulindac to act in Cox independent mechanism to inhibit colon cancer cell growth B- perhaps we could touch on that a little bit later B and a number of other candidate targets that might ultimately be pursued and validate and represent novel therapeutic target genes in this pathway.

Slide 21:

I would like to take just the last three or four minutes and tell you just a little bit about hereditary non-polyposis colorectal cancer because I think it represents a contrast to this APC pathway and some interesting observations there.

As I think most in the room are quite familiar, Henry Lynch and others made I think quite heroic efforts to identify families affected by hereditary non-polyposis colorectal cancer and really established the notion that this was a genetic syndrome likely to be due to a dominant gene with variable penetrance in these kindreds.

Slide 22:

The observation, I think perhaps the seminal observation, that led to the identification of the genetic basis for this syndrome was not very well shown here because perhaps of the light, but was the observation by a number of groups including Manuel Perucho, Burt Vogelstein and Albert de La Chapelle and Steve Thibideau and others that there were microsatellite sequence alterations associated with the cancers arising in individuals at HNPCC giving rise to new alleles in the cancer tissue as compared to the patient's normal alleles in the normal tissue, either expansions or contractions of these microsatellite short repeat tracks.

Slide 23:

The prediction from the yeast geneticist was that this would be accountable for the result of mutations in genes involved in recognizing and repairing DNA mismatches and a large number of genes and their protein products are involved in this process.

The most frequently mutated genes in the germ line of individuals with HNPCC

Slide 24:

actually summarized on this slide are the MSH2 gene on Chromosome 2P and the MLH1 gene on Chromosome 3P. Together these account for about two-thirds of the mutations involved in HNPCC and there are rare mutations in a variety of other genes which play a role in mismatch repair.

The notion is that the vast majority, perhaps 95 percent, of the families that meet the Amsterdam or ICG criteria for HNPCC will have mutations in mismatch repair genes or proteins involved in recognizing such defects.

Slide 25:

The thing I want to stress here in contrast to APC, where the germ line mutations appear to be relatively homogeneous there is a great deal of genetic heterogeneity in the HNPCC syndrome accountable for either as a result of MSH2, MLH1 or these rare genes, there appears to be no clear cut correlation between particular gene mutated in the germ line and the phenotype that one sees. Again, the predominant risk is of colorectal cancer in males and an elevated risk of endometrial and ovarian cancer and a few others in females and a lower risk of a smattering of other cancer types.

Slide 26:

The last point I think I would like to make here before just summarizing what I have tried to tell you and a few things I haven't had time to tell you is that this same class of genes is altered in about 15 percent or so of colorectal cancers. The mutations in the vast majority of sporadic tumors or apparently sporadic tumors that have apparent defects in mismatch repair are actually inactivation of the MLH1 gene, and it is not mutational inactivation. It appears to be inactivation by methylation of the promoter sequences. There is very nice work from a number of laboratories including Jean Pierre Issa and Steve Baylin, Richard Kolodner and others showing that methylation of the MLH1 promoter is essentially a way to functionally inactivate mismatch repair gene in most sporadic cancers.

Slide 27:

Some of the target genes that have been suggested include TGF beta type 2 receptor, BAX, a protein that is a pro-apoptotic molecule and presumably its inactivation will lead to perhaps a relative resistance to cell death, as well as some transcription factors, and the list goes on.

Slide 28:

So I will just wind up with this slide here and try to summarize a few things that I have tried to tell you, only a fraction of which may have actually been news to you today, but at least perhaps some of these points were revisited from things you might have heard earlier.

I tried to tell you a little bit at the outset about how well the natural history has shaped our notions of how these genetic changes may contribute, with some genetic changes contributing very early in the process such as APC mutations and other genetic changes appearing to contribute relatively later in the process and the notion of clonal selection for when these changes arise being associated with clonal outgrowth accounting at least in large part for this preferred order.

Now, again, it is not invariant. There are clearly exceptions where p53 mutations can be identified early, but again, these are the exceptions rather than the norm where the mutations really do show a preferred order.

The other thing I should point out, and I tried to a little bit with the APC story is that although we put specific genes up here they really are indicative of pathways as a whole, and it is the notion that one can mutate either the APC gene and deregulate presumably not only beta catenin but gamma catenin or one can mutate beta catenin which has at least some of the same effects as APC to account for these pathways being altered and that the notion might be that there are five or six, who knows exactly how many, critical pathways that need to be deregulated until one gets a fully invasive carcinoma capable of giving rise to distant metastasis and certainly other genetic changes contributing to metastasis.

I haven't had a chance to talk about alterations in DNA methylation which probably have a critical role as I tried to suggest in the case of MLH1 in altering patterns of gene expression. Although the initial suggested hypomethylation and perhaps activation, inappropriate activation of genes might be important, I think there is probably a larger body of data from work from Steve Baylin and others showing that increased methylation and inactivation of genes may be a perhaps more predominant mechanism of contributing to colon cancer development, and I have tried to stress at the end just a little bit of the notion that germ line mutations in mismatch repair genes again are relatively infrequent but upwards of about 15 percent of all colon cancer cases have an activation of one of the mismatch repair genes, most commonly MLH1 and that this inactivation presumably increases the tempo at which lesion-initiated clones progress on to carcinoma.

Again, I haven't had a chance to talk about ras or some of the genes on Chromosome 18. Stan may talk about these a bit as markers or at least Chromosome 18q but I hope I have given you some sense of how we are thinking about genetic changes and how some of these are being pursued to validate them as important targets not only in the biology but potentially as well important targets in preclinical and clinical settings.

So I will stop there, Bob, and I guess we will turn things over to Stan.