Apparently we go straight to the bar. Okay, prostate cancer. The object of discussion today, to me, has a remarkably common pattern for a remarkably variable genetic disease. If you add up all of the genetic changes in prostate cancer and look at the various combinations in any individual patient, it is as complex as any epithelial cancer. Yet despite that, following local therapy, which can be curative, such as prostatectomy or local radiation therapy, a fraction of the patients will recur. And it's not really predictable based on that collection of genetic data. So something else is operating here. That subfraction that returns with disease after a variable time period, the double hash bars, they become subject to either watchful waiting or variety of relatively common therapies that you all use in your practices, which are basically suppressive but generally not curative. Okay? Now a man may die of something else, okay? Get the disease in old age and die of a second disease. Or have the disease suppressed sufficiently.
But a fraction of those patients will go through those anti-androgen blockades, and they will return with an aggressive disease we generally refer to as castration-resistant prostate cancer. This is really an operational definition. Okay? It's not linked to specific genetic changes. Some rare examples, counterexamples exist, but in general this is sort of a pattern of the disease writ large for almost all men who succumb to this. At the later, later stages of the disease, there are some subpopulation of patients that acquire a new histological variant, and that is as different from original prostate cancer as is any other aggressive end stage metastatic epithelial cancer. So what we're trying to apply is can we use T-cell-based immunotherapies to treat the early stage of recurrence when there's minimal residual disease in that first window shown on the slide. And of course, you could do that either with antigen receptor therapies targeting using antibodies to make chimeric antigen receptors, or with T-cell receptor based immunotherapy. And I'll talk about that today. What we're looking for, and I've used this word cautiously, but is curative intent to treat the disease when there's low tumor burden with a therapy which can provide longevity of treatment. And that's what the immune system can do that almost nothing else can. What's the target? Well, the target is prostatic acid phosphatase. Well-known target, been around for a long time. In fact, there was even a vaccine made, PROVENGE, which is in allowed clinical use, to try to excite an endogenous immune response against this disease.
Vaccine doesn't work all that well, everybody knows that. It is still used, but it's not a dramatic immune response which is mounted. So the antigen for this vaccine, we thought was still a great target for this adoptive immune therapy. The antigen is highly specific for prostate tissue. If you look real carefully somewhere along that baseline, you'll see a little tiny blip. That's the expression level measured by RNA-Seq type data in esophagus, which is barely perceptible, but is also seen very rare cells with immunohistochemistry. So we thought this was a great target. What modality do you use? We're thinking that we're going to go with T-cell mediated immune therapy. And the question is, do you try to use a CAR or do you try to use a TCR? Well, what we're most interested in here is not only the specificity of the attack, but the ability to get the immune system to persist in its therapeutic activity over long periods of time. So if there's regrowth of tumor, you have now a built-in immune system with immune memory that can come back and eradicate those small growing masses of tumor. So the other reason we like this idea of using TCRs is that they can have affinity constants and measurements of activation, which will allow the T-cell to not only differentiate and turn into killer T-cells, but also into memory T-cells.
If you go too high or too low in either of the ends of reactivity, you tend to get exhaustion or non-response. And so it's partly finding the sweet spot here that counts. So to go after that, former graduate student Mao, call him Dr. Mao now since he got his PhD, set out to determine the immunopeptidome of the prostatic acid phosphatase. That was actually never done in the creation of the vaccine PROVENGE. He mapped all of the epitopes for HLA-A*0201 that could be produced from the PAP gene. We have all that data and using it in various ways. And he set out to find TCRs. He was able to get those TCRs to react in a very specific manner to each of the processed epitopes of PAP, but we were very disappointed to find that they were specific, but incredibly weak. And if you look on the cytotoxicity data on the right-hand of this slide, the best of these TCRs at most could keep a population of tumor cells sort of in check. There was a sort of a balance between growth and death.
And this is nowhere near what you would need to take something into a clinical trial situation. And we were just about ready to give up on the project until I heard a talk by my colleague Chris Garcia, at Stanford, who had been working on something called catch bonds. Catch bonds, really if you get right down to it, is a specific type of mutation within the TCR alpha beta structure that does not change the forward rate constant of the TCR engaging and binding to peptide in the context of MHC, but dramatically changes the back reaction or disengagement of the TCR MHC peptide complex. It's like a fish hook going into the cheek of a fish. It's easier going in than coming back out because of the barb on the fish hook. The same thing is happening molecularly here, okay? And what that does is that it creates a very interesting mode of action. Now you get incredible efficacy for these TCRs. I won't go through all the data points, but you dramatically increase the efficiency of cytotoxicity, going now to ratios of one killer T-cell plated per 10 or 20 tumor cells, clearing the dish over some period of time. You can recharge, add more tumor cells, the same population will again be able to diminish and get rid of those tumor cells in these in dish assays. And we see dramatically enhanced efficacy in various in vivo models.
So the idea really works, okay? You can take a bad TCR and make it into an efficacious and useful TCR. To demonstrate the most impressive thing, we were able to establish a collaboration with Brian Evavold's lab at the University of Utah, and he can measure single molecule force relationships between any binding pair of proteins. And the key data is in sort of the middle of the slide there. This is the bond lifetime. How long does that interaction of the TCR with the MHC peptide target antigen last? And with these single point mutations in the TCR, these catch bonds, get a 40-fold enhancement or lengthening of the time of the back reaction. If you remember from your college K1 over K2 for... Okay, a big change in the denominator like that makes for a huge change in the overall effective affinity, and that's exactly what we find. So where are we going from here?
We've now credentialed these TCRs to the point where we think these are really worth putting into a clinical trial for minimal residual disease in patients of HLA-A*0201. And how to do it? Okay. Well, I've started lots of companies and that's an option, and we're considering that, but we really wanted to take this forward in an academic setting to prove the concept that this was a way to target early recurrence of prostate cancer. And we've been really fortunate to form a collaboration with the NCI Experimental Therapeutics Program, their NExT. We currently have all the materials, and they're processing them into GMP manufacturing, and will help us with early stage toxicity, and then I think even help us with portions of what will be the clinical trial. And so we're looking forward to taking this as a new way to treat prostate cancer instead of waiting for it to come back with radiographic findings by PET or any other measurement. We're going to want to propose to treat this disease at the first recurrence of PSA measurement, indicating there was residual tumor post-surgery and time to treat the patient now when you've got a much better chance at, let's say, curative possibility.
We'll have to do early testing in later stage patients for toxicity issues, but the goal is to get this into early stage treatment. Okay. Well, what we're trying to prevent, of course, is this progression after androgen deprivation therapy or other forms of androgen suppressive therapy to that late phase of the disease called castration-resistant adenocarcinoma, a fraction of which converts to this super aggressive small-cell carcinoma sometimes with neuroendocrine features. The best treatment for that disease is to prevent it from ever happening. Okay? That's a simple concept. I got interested in this as part of a large consortium funded by the Prostate Cancer Foundation, ACR, other agencies, with a multi-university group of people getting biopsies of late-stage prostate cancer patients. Many colleagues from UCSF were critically involved in this as well. Eric Small being the co-PI of the grant. The very first biopsies that were taken from these castration-resistant patients essentially showed all of the problem if you just looked at the slides. The prostate cancer went through a kind of a conversion where it changed its histological phenotype and began to acquire features that you would associate with what we call small-cell carcinoma, some with neuroendocrine features depending on the marker panel used.
This is very reminiscent of what happens in the therapy conversion of adenocarcinoma of the lung following treatment with certain tyrosine kinase receptor inhibitors to convert it into a phenotype we'd refer to as small-cell lung cancer. So this is a very dynamic response in different kinds of cancers, responsive to both therapy and treatment resistance. And what I'm going to show you is applicable across a broad range of cancers, not just prostate cancer. But to get into understanding what this was all about, something that I've done all through my career is say, "We have to build a model in which we can watch this biological process develop, rather than simply analyzing the end result by some correlative assays." And so this is the model we developed. It was published some years ago. 2018, time flies. And it's a very effective model for converting normal prostate epithelia into the most aggressive forms of prostate cancer, as shown here. It uses a combination of different genetic perturbance, it's not even important for today's story to tell you what they are, but they're all logically chosen. And you get these super aggressive cancers, as shown in the far right-hand panel by histology and some of the neuroendocrine markers. This enabled us to test not only prostate epithelia for this response, but also lung, bladder, and other tissues.
And we can basically take almost any epithelial tissue in the body and drive it into this super aggressive type of tumor tissue analog and use it to study the gene regulations and so on, what's going on. To give you an idea of how dramatic the changes are in how the chromatin has been rewritten and the gene expression change, look to the far right and you'll see a measurement of genomic openness or closedness measured by ATAC-Seq. And the simple answer is from the beginning of that tissue, that sample we took through the phases that represent more adenocarcinoma-like and then on to the small-cell carcinoma, the changes in open and closed chromatin are as large as going from a fertilized zygote to any somatic tissue in the body. Okay? So this cancer model is showing us how much of the genome changes during the evolution from an adenocarcinoma into the small-cell carcinoma phenotype. So what to do about it? Well, there's a whole bunch of studies where we've used this to look at kinetic issues, variation expression issues.
You name it, probably 15 papers have been published since then on that. But the one I want to talk about today is to look for genetic dependencies in this small-cell carcinoma state by using genomic analysis to look for dependencies. And there's a huge project called the DepMap, and the Broad Institute has been highly instrumental in this, to take thousands of cell lines and use the technique of CRISPR-mediated inactivation to look for genetic dependencies for growth of the cell lines in those tumors. These lines that we developed from the prostate cancer transformation model called PARCB, that I just showed you, can be used in this type of analysis as well. And when that was done, and the we of course is Liang Wang, who is a former postdoc in the lab, carried out the initial studies on this, now taken over by others in the lab. You get out a series of genes, some of which are already well known to be important dependencies in cancer. Those are the common essential genes. And then you get other gene sets, which we refer to as non-common or more special associated genes. And you go through the list and you categorize them and you look to see which one you want to spend the next few years studying. And the one we picked, because it was very high up in the analysis, is one member of the cell cycle control gene family of E2Fs. Okay? It's a specific submember, and as it turns out to be quite critical, called E2F3.
Okay? In this mathematical way of looking at the genes it was the third rank, use a different set of algorithms and it comes out in the top 20, but it's an important gene, caught our eye, and we set out to study it. Among the family of E2Fs, and these are important regulators of cell cycle control, both positively and negatively, the only one that scored in this assay above the threshold was E2F3. The little diagram is just to remind me to tell you that where do we think this works? Well, it works as a co-partner regulated by RB1, downstream from the cyclins and cyclin-dependent kinases, which are of course important targets for cancer therapy as well. Here's the type of data that we saw. If we went back and confirmed this activity, the E2F3 knockdown limits the growth of the PARCB cells. And I have to say, in the thousands of these kinds of experiments that have been done in my lab, this is probably singularly the best single gene inactivation cell growth response I think I've ever seen. Maybe BCR-ABL knocked down before we had drugs like imatinib would give you a similar phenotype for cessation of growth in these selected cell lines. But more impressively I think is figuring out that this particular gene target is, in a formal genetic sense, a synthetic lethal to the loss of RB function, which is a pretty common event in human cancer. We made isogenic cell lines out of PC3 which happens to have a wild type RB, made a knockout of RB, went through this sort of testing again. And the long and short is that if you're negative for RB function, you become susceptible to the loss of E2F3 as a therapeutic.
And I think this has broad applicability across human cancers of various types. And just one way to kind of say that is that in addition to looking genetically at loss of RB, you can have alternative events that regulate the efficacy of RB, and you can have a transcriptional signature for RB loss, which predicts E2F3 dependency across a huge number of cancer cell lines from publicly available DepMap data. So again, what I'm telling you is not just about prostate cancer. I think it's for lung cancer, ovary, bladder, et cetera. So what to do about this? Well, so while this work was being done, Evan Abt, he's an adjunct assistant professor in the group, was hard at work putting this paper together. What always happens in science? Somebody publishes something that looks disgustingly like what you're working on, and really annoying. There's a great paper, the one listed on the right-hand side, from Circle Pharma and associated academic investigators at Dana-Farber and elsewhere. And what they showed is that they had identified the E2F pathway in general as a therapeutic target site that's sort of been known for a long time, but they were able to make a small molecule inhibitor that actually works via regulation of negative regulators to introduce mitotic defects and the death of cancer cells.
It's a really clever chemistry in this, and those with more experience than me can go into it in detail, but it's a cyclic peptide decorated to take on more drug-like properties with very good bioavailability. But the mechanism that they talk about in the paper of how it works to induce cancer cell death is actually to increase E2F activity. Okay? And we just had a paper come out the other day that says if we reduce E2F3, a submember of the group activity, we also get cancer cell death. So I think this is one of those Goldilocks moments where you need the right amount of E2F activity, too much or too little can both be useful therapeutically, and I think there's going to be a lot more to come from this. So we were pretty happy with this and could have stopped there and it all would have been fine, but one of those nice things that happens in science is when the person who's doing the work remembers what he used to work on and connects it very nicely to the problem at hand.
So Evan Abt was a student and fellow in Caius Radu's lab where he was particularly interested in nucleotide metabolism, and he remembered data that he had done himself, his own hands, on the sensitivity of different tumor types to a particular inhibition of enzymes that regulate nucleotide pool, particularly pyrimidine pool size, the enzyme DHODH for which there's clinically available inhibitors that are used in autoimmune disease. And the lines that respond to the single point inhibition with these clinically available drugs that reduce pyrimidine pools were efficacious in only subsets of certain types of tumors, and that subset was overrepresented for things like small-cell lung cancer, which is the kind of cancer we're seeing grow out at the very end stage for prostate cancer patients. Turns out there are a couple of papers in the literature about this, including some very nice papers on zebrafish models, other nice papers using human tumor cell lines, but not much had been done since 1921, '22 in this field.
So we went back and Evan began to reinvestigate whether or not the cell line models and tumor PDF models that we were using might be affected by this already available drug that limits pyrimidine pools. And the answer is absolutely. Okay? If you add this drug to these cell lines, you can show that at different dosages you will inhibit cell growth, activate cytotoxic mechanisms, including getting cleaved caspase, et cetera, and really dramatically limit the outgrowth of these cells. So I don't think this is a cure for these cancers by any means, okay? But I do think it's a dramatic enough effect in a specific way of limiting the efficacy of E2F3 and its function that they probably will be worth trying, both in the clinic directly, single agents, or in combinations with current therapies or new therapies we can figure out based on how this works. So what we're doing now is looking at all of the things that might be affected by the addition of this DHODH inhibitor. And initial thought was pretty simple, "Well, you don't have pyrimidines, hard to make DNA. Well, that's the mechanism." Okay? It's like a cytotoxic agent that's limiting DNA processing in some way because you don't have the precursor. Well, like everything, way more complicated.
There's a whole bunch of things to think about. We're going through them and trying to figure out which of them will be most efficacious. And basically that's where we are in these two attempts to treat the earliest stage of prostate cancer post-surgery, recurrence, and thinking about things to help those patients who have reached these very difficult end stage tumors that have acquired the characteristic as dangerous as small-cell lung cancer. Lots of people to do this work, lots of collaborators.