Scientist Stories: Jim Allison, Pioneering Immunotherapies

Hello everybody. Bonjour. Konnichiwa. Guten Tag. Ciao. Hola. Ni hao. Aloha. Salaam. It's my pleasure to welcome you all to another Global ImmunoTalk where we have an extraordinary special guest today, Dr. Jim Allison, who needs no introduction to the field. But I'll give you some background on Jim and his life because he's got a very interesting life and career, one of the most illustrious and distinguished scientists in the field of immunology. So, Jim was born in, you know, in Alice, Texas. So, he's a Texan, has been born and raised and loves Texas. He started his career in— at UT Austin. Then he moved to Berkeley where he became a full professor and HHMI investigator. Went on to move on to Sloan Cancer Center where he was a chair of immunology and then wound his way back to Texas again MD Anderson, and he's now the Regental Professor and Chair of Department of Immunology, Vice President of Immunology, Executive Director of Immunotherapy Platform, Director of the Parker Institute for Cancer Research, Deputy Director of David Koch Center for Applied Research in Genitourinary Cancers, and the Ogle Keith Weiss Distinguished University Chair of Cancer Research. I don't think, Jim, I've ever heard anybody that has more titles than that. He's a very busy person. He's also now the founding director of the James P. Allison Institute, which was established a couple of years ago and is going to be a leading institute in cancer immunology. Dr. Jim has spent a distinguished career studying the regulation of T-cell responses and developing the new strategies for cancer immunotherapy, as we all are aware of. He earned the 2018 Nobel Prize, for which he shared that with Dr. Tasuko Honjo for their discovery of cancer therapy by inhibition of negative immune regulation. And as we all know, his most notable discoveries in the pioneering the understanding of how the T cell receptor signals, its structure, how co-stimulation, the discovery of CD28, and how CTLA-4 operates to inhibit CD28 signaling and how that regulates T cell activation, autoimmunity, and anti-tumor immune responses. And obviously, that were some of the most seminal findings that paved the way to his work for developing the first immunochip blockade molecule and antibody against CTLA-4 that was approved by the FDA and has gone on to now save the lives of many cancer patients. He's also received other notable awards, of course, National Academy of Science, the Lasker Award, Tang Prize, and many others. But Jim, we all know that you are a musician. You love to play the harmonica. Many of us have seen you play the harmonica at meetings, and you are an extraordinary musician. And so I wanted to share this slide for those of you who may not have seen Jim himself playing the harmonica. This is a slide that I've been able to see some of your own personal videos of playing the harmonica on stage with Willie Nelson. Here he's also in this other picture on the right getting the Nobel Prize award. So Jim, I have one personal question before I ask our globally— I mean, I'll talk question. Which are you more proud of?

Uh, no, no, it's pretty much the thrill of playing with Rilly. Um, so I don't know, I'd rather not take a stand.

I'm just teasing you. Of course, we're so proud of both of you, uh, for both, uh, both your musical, um, talent and, uh, of course, your scientific talents. Um, so the, uh, talk we would like our, our listeners to hear from you— I mean, the question we'd like to have our listeners hear from you today is if you were to think back on your career and you maybe made a decision that was the most pivotal in influencing your scientific career, what would that be?

I think it was the most pivotal and the most dangerous at the same time, I think, was when I, due to hard work of a number of people in the lab, you know, when we came up with a notion for our CTLA-4 blockade, you know, before there were checkpoints. CTLA-4 was the first. And, you know, having dabbled in the field for about a decade before that and just seeing sort of marginal responses and stuff, and when we showed that almost any tumor that we threw this at, it would just wipe out, they'd just melt away, and the mice were immune. I'd never seen anything like that before and was convinced that, you know, this really had to go to the clinic. And met with a lot of opposition there. And, uh, um, and so I devoted close to 4 years, really, I would say. Um, my friends all talked about how, I don't know, fanatical I was about it. But, but the thing that happened that made it a pivotal decision, I think, was that I got so focused on that, you know, some of the other work slid, and, and I kept doing more stuff here rather than, you know, my usual thing, which is to dive deeper and move on. But it was one of those things where if I won, I won. If I didn't, I was done for, you know, that would have been the end of my career. But, you know, luckily, you know, it worked out. So found somebody that would help me get into the clinic. That was the whole deal. There was so much contempt for the notion of the immune system and immunology, the immune system working in cancer therapy at the time. People just laughed at it, even a lot of immunologists would score, you're saying, well, Jim's a tumor immunologist, you know. But I think that, that was probably the most pivotal. Yeah.

Well, thank you for sharing that. Well, we're excited to hear your talk, so please go ahead and share your screen and we'll go forward.

Okay. I think my screen— wherever I lost it here. There we are. So, what I want to do today is really concentrate on just one thing really, and that's just talking about the mechanisms of CTLA-4 and PD-1. There are these two major checkpoints now, some minor ones I'll get to briefly at the end, but in the clinic, because PD-1 had fewer and less severe overall immune-related adverse events in patients that were treated. It became the dominant one. But we pursued this from more of a fundamental immunology view, looking at really how it works, the kind of cells that are involved, some of the signaling that's involved, its effect on indirect effects on the myeloid cells, and finally, just the inherent properties of memory that are associated with initiating these two kinds of therapy on T cells, taking it completely out of the context of tumor immunology, just asking, okay, what do these do to memory and memory development through the mechanisms I'm sure everybody on this call is aware of? So, that's what I'm going to do. There are my— mine and my spouse Pam Sharma's disclosures. The work I'm going to show today, I always have to thank Max Krummel and Alan Korman. Max, who as a graduate student was the one that showed CTLA-4 was a negative regulator, then Alan, who helped us get over the top in making— helped us make the ipilimumab antibody. It was the first clinical reagent. The work I'm going to show you today is largely due to 2, 3 postdocs— actually largely due to 2 Spencer Wei and Stephen Mock. And I'll acknowledge others as I go through with the constant work of help of Pam Sharma and also with the immunotherapy platform at MD Anderson. So, as Sue said, you know, I began in the '80s actually by trying to look at how T cells were activated. After showing the structure of the T cell receptor, the field quickly became aware that that's not sufficient to activate a T cell. Additional signals that they called co-stimulatory signals required. Without going into all the details, we, after a few years, showed that the major, if not the only one that fits the criteria of a second signal required to turn on a naive cell, prevent a clone from becoming anergic, is CD28. And others show that the ligands were B7 molecules, and only subsets of cells, notably Of course, dendritic cells can do that effectively. And when a T cell gets those signals, it turns on all this stuff that allows it to enter cell cycle progression, turns on survival factors, turns off inhibitors of proliferation. So, the cells begin to proliferate. You have to do that, of course, because, you know, the 10 to the 10th or 10 to the 11th, whatever number of different clones you have, only a few get activated by given tumor or virus, if you will, you've got to generate an army, so they've got to divide. And so that's what that's all about. But this other molecule that we started working on because it's been cloned by others, but its function was really not known, CTLA-4, quickly reared its head because it's structurally homologous to CD28. And Peter Lindsley showed that it reacted with the same ligands as CD28. So, that led some interest into it. First it was thought it was another gas pedal, you know, like CD28, but it quickly became clear through Max's work that it actually inhibited T-cell proliferation. And so we began studying its anti-tumor properties. But later on we found that it's even— it's not just something that you exploit there, it really is fundamental in terminating the proliferative phase. You know, at the time we were doing this work, people thought it was just activation-induced cell death. That of course does something— FAS-mediated cell death. But when you get rid of those, there's not a hell of a lot of impact on immune response. But when you get rid of CTLA-4, it's lethal. You know, mice die when they're about 3 weeks old. So tecmactidine, and my lab showed that all showed that at about the same time. And the way it works is basically by reversing a lot of the things that CD28 does. But it comes on, the level of protein accumulates. There's a lot of controversy in how it works. We think it just outcompetes because it's got about a 3-log or so higher affinity because of some structural peculiarities of CD4 compared to CD28 and just outcompetes CD28 for the B7 molecules and just shuts it off. Any event, with that tumor context, we've been doing a lot of work showed that tumors failed to elicit immune responses, not because they lacked antigens, but because they couldn't co-stimulate. We put B7.1 or B7.2 into a number of tumor models and tumor in mice and showed that that made them completely immunogenic, induced immunity. And so that wasn't the problem. And so this led— gave rise to this model that we began working on, and that is that, you know, CTLA-4 is going to come up in every immune response. It's hardwired to do that. Its job is to do that, to stop proliferation. But that once it starts coming on, it's a race. And if you generate enough T cells to take care of the tumor, which has had a head start, then you know, the immune system wins. But if you don't, then the tumor wins. And so we had the idea just taking the brakes off with an antibody might allow the immune system just keep going as long as you needed it to. So as I said, we carried out some early experiments that worked in almost every tumor model that we did. I'll show one in a moment. But there were two things that were, you know, just scientifically compelling too. One of them is that we're not treating the tumor. The tumor's over here. We're actually treating the immune system, immune cells, you know, inhibitory molecule on T cells. So this could work against any kind of tumor effectively. But the other thing is, since all this priming starts with necrotic cell death and activation of the innate immune system, when you kill tumor cells or when you provide a vaccine or you kill tumor cells with chemotherapy, radiation, whatever, you're going to start that process, but CTLA-4 is going to shut it off. So you could also very readily use this as combination therapies, and this is happening now. So, which I'll come to later, in combination. So, you know, it could give memory and polyvalency and all these things that immune therapies have that conventional therapies don't. So, this just shows one of our early tumor experiments. This is an experimental tumor growing on the backs of mice. If we injected blocking antibodies to CD28, you can see the tumor grows faster, showing that there is a nascent immune response, just can't get it done. When you inject anti-CTLA-4 monoclonal antibodies, the tumor grows for a while, but then I think as the T cells accumulate, it begins to, you know, take the tumor out and vanishes, and those mice are immune for the rest of their lives. So very effective. We did this in a lot of different tumor models. Anyway, this led to a Phase 3 trial. It was of survival. Overall survival is the endpoint. It took almost 5 years. It was finished in 2010, and it was in stage 4 melanoma. And just to put this in context, this is untreated. These are placebo controls, which was not unethical in those days. There was no standard of care for melanoma. No drug had ever prolonged survival of melanoma. The median survival was about 7 months. In 2011 after diagnosis. That's just what it was. Fewer than 3% of patients were alive at 5 years. Anyway, you can see here this top line is the IPI control, and you can see that it moved that median survival over a few months. But the cool thing is it also flattened out here at about 20% after about 2 years. Anyway, this alone would have been sufficient for FDA approval because no drug had ever done that. But this tale was really intriguing. An interim study a few years after that, a total of almost 4,000 patients for whom there was 10-year survival, showed that this survival— again, this monotherapy lasted for about at least 10 years in patients that had received ipilimumab antibodies. But why is it 100%? I could give you some mechanistic reasons for that, but one of the obvious ones, certainly more interesting ones scientifically, was there another checkpoint. And Kataska, as Sue mentioned, who had been working on a molecule called PD-1 that he had cloned and thought played a role in negative regulation in thymic selection, hence the term program death 1. Arlene Sharpe and Gordon Freeman actually showed that it had two ligands, PD-L1 and L2, and was actually another checkpoint, different in that, as Li-Ping Ching and Sun showed earlier, that's ligands induced on tumor cells by gamma interferon. And then that means that T cells, which come, have specificity for a tumor and come near them, make gamma interferon, and they upregulate— the tumor upregulates PD-L1. It binds to PD-1 and through the T cell receptor, we think, turns them off. And so it also— both these drugs worked in a wide variety of patients, not exactly By the time I talk about some interesting differences. But anyway, both of them worked in, you know, not just melanoma, but lung cancer, bladder cancer, head and neck cancer, throat cancer, cancers, a lot of cancers that was caused by mutations. I'll come back to that later. But anyway, this shows some of the data for phase 1 trial that Suzanne Topalian carried out. So here we have The response rates with PD-1 are a lot— the responses are faster because, again, it works on terminally exhausted cells, as shown in the cartoon, but also not complete. So, what do you do? Well, the obvious thing was to put them together. So, Mike Curran, when he was a postdoc in my lab at Sloan Kettering, combined them in mice and he showed they're at least additive. And this actually is not the most recent follow-up. I thought I'd updated this, but anyway, this is a combination trial done by Jed Wolchok that was started 8 years ago now, and the survival there is still over 50% with no additional treatments after this early period. And so that's quite an improvement from the 3% that you got with the monotherapy with IPI. Or the survival that was obtained with PD-1. But I think that we could start to think about curing cancer in melanoma. That again, we went from a disease in which only 3% of patients would be alive here up to more than half being alive and essentially free of worry, many of them that I've met, which is— well, given that for a while, but an obvious benefit for quality of life in those patients. But it's not just melanoma. This is a partial list too. There are too many approvals now and different indications to put them all up here. But ipi was approved in 2011, and then Merck's PD-1 in 2014, Bristol-Myers Squibb's PD-1 in 2014 later, and then the combination of ipi and PD-1 in melanoma based on the data I showed you and others in 2015. 2015. More recently, and I'll talk about this a little bit later, PD-1 has been approved with another molecule called LAG-3. And this is— other than that, there really haven't been any more different molecules. These, all of the things you all hear, are other PD-1s, or maybe one or two of them are other CTLA-4 antibodies. But any event, there's survival advantage benefit. And not just melanoma, as I said, but lung, renal, colorectal, head and neck, lymphoma, and pancreatic. It goes on and on. But these range from 15 to 25, or in the case of bladder, maybe 35, 40%. But lower that, and then some cancers like pancreatic and glioblastoma really haven't responded very well. There's some hints lately that give us some hope but haven't responded very well. So we've still got a lot of work to do. And so I think one of the things that we've decided to focus on is really understanding how these drugs work and others so that they can be combined in a way that makes sense instead of just randomly, as seems to be happening in a lot of cases now. I'll spare you my soapbox for a little while. In any event, so it's useful to compare the properties of these. I mentioned CTLA-4 is hardwired. It's involved every immune response, every T-cell response that you get. PD-1 is an induced resistance mechanism that's thought plays an important role in protecting the developing embryo from destruction by the maternal immune system, and that tumors have co-opted it some way. As I said, CTLA-4 works during priming— sorry, targets CD28 co-stimulation, therefore works during priming, whereas PD-1 seems to work on the T-cell receptor pathway. There's A little bit of controversy there, but I think the overwhelming data suggested that that's the way it works. Works on fully differentiated effector cells that can make gamma interferon. That's how it's induced. Anyway, because of these differences, PD-L1, CTLA-4 blockade can expand clonal diversity. We showed a long time ago, Cynthia Chambers in my laboratory at Berkeley in the '90s showed, and others in the lab using variety of peptides of different strengths that CTLA-4 can lower the threshold of avidity that a peptide MHC needs to fully activate a T cell. And that has the effect of expanding clonal diversity where PD-1 just seems to reawaken those differentiated cells that you already have and allows it to work. Again, because those differences CTLA-4 is often slow, PD-1 rapid. So, I'll show you CTLA-4 in monotherapy primarily affects CD8s— excuse me, CD4s, although in combination CD8s. And PD-1 as monotherapy absolutely only affects CD8 T cells, although I'll show you it could also affect CD4s in combination. I'm not going to have time to show you the data today, but Pam Sharma in the immunotherapy platform here at Anderson has shown that CTLA-4 could move T cells into cold tumors, immunologically cold tumors such as pancreatic, and then in some bladder patients that are cold, where PD-1 cannot do that. Again, adverse events with CTLA-4 are quite frequent, much less so with PD-1. And also, this begins to give us a hint that there's something more fundamental going on because disease recurrence after CTLA-4 is very rare. Generally, you know, if you make it that 3 to 5 years or so, that's it. But disease recurrence after responses to PD-1 are significant and can be, you know, 25 to 35% after 2 years or so. And another difference is the way they're given, and this will become important later. And again, it's because of how they work. But CTLA-4 blockers are given usually just a few times. Partially that's because of toxicity, but mechanistically that's maybe all that it needs, I'll show you later. Whereas PD-1 is given depending on the supplier and everything, the company, repeatedly as often as 2 weeks, sometimes for as long as 2 years, or until patients progress. So again, a number of differences between, between these 2 ways of treating cancer. So, Spencer Wang, when he joined the lab a few years ago, decided to— this is when CyTOF was relatively new, so I'm not going to explain to you these days how that works, but basically we have a CyTOF panel using all 43, actually 44 possible flags, which comprise of those to eliminate, and the data I'm going to show you, just to exclude cells other than T cells. And then have the major differentiation markers and antibodies to transcription factors that delineate different types, functional types of T cells, and also markers for proliferation, for cell death, things like that, so that we can really get an idea of what's going on. And we also kept track of all the data in individual mice as we were gathering it so that we could you could correlate the changes that we saw with the phenotypes with what's going on in the tumor. So, I'm going to show you data that we got using the MC38 tumor. This is a methylcholoanthrine-induced colorectal carcinoma, very high mutational burden, very much like human melanoma, actually. And it responds very well to CTLA-4 and PD-1 monotherapy— blockade monotherapies. We also did the experiments with the B16BL6 tumor, which responds very poorly, and we had to use a vaccine along with that tumor, which has very low mutational burden. And anyway, but the data that we got, the cell types involved in the mechanism was essentially the same in both. And we also did some human patients, not as extensively as the mouse, and it replicated best as we could tell, in humans as well. So, just to show you the annotated different clusters of cells that we saw. Now, we've got to spend a few minutes on the Treg cells because this needs some explanation. But anyway, there's two kinds of FOXP3-positive CD4s in this analysis, KLRG1-positive and KLRG1-negative, these being the more mature, we think. These are definitely depleted by anti-CTLA-4, at least by 9H10, which is the antibody that we made hamsters that we're using in these studies. Whether or not it does anything to these, probably naive cells, we're not sure. But anyway, it definitely depletes these functional ones. There's also some decrease in the numbers by PD-1, but I think that's by depletion. Anyway, the absolute numbers of T-REX, of course, is associated with bigger tumors because they're inhibitory. But so far, the work that we've done, in humans, at least in melanoma and bladder where we've looked pretty extensively. As has Toti Ribas in melanoma. We've not seen any evidence for this going on with ipilimumab. And it does seem— and Alan Korman is really— and Mr. Bashkwip saw the same thing. We could deplete Tregs in mice models, but it doesn't seem to happen in humans. But I'll just— I won't show the cartoon again, but there is this cell-intrinsic pathway that Arlene Sharpe and my lab both studied extensively in the '70s. So, before the Treg thing was discovered. So, there is a cell-intrinsic as well as this extrinsic way. So, in humans, only that extrinsic one is functional up to date, but there are efforts being done now by a number of groups trying to make CTA-4 antibodies in humans deplete Tregs, which is not that hard because they have about 2 logs higher CTA-4 than effector T cells, and therefore can become targets of antibody-dependent cell-mediated cytotoxicity by macrophages in the case of mice and humans in the case of patients. Anyhow, so Tregs, in mice but not really in humans. If you look at CD4s here, there's basically two different types. These seem to be, again, naive, unactivated cells. But a major population of what we call Th1-like is expanded, as I'll show you here, by ipilimumab, but not by PD-1, as shown here in the blue and green dots here. And we call that Th1-like because we know from other experiments they make gamma interferon and TNF-alpha, and they have T-bet, but they also have ICOS at pretty high levels. And it makes them stand apart from what you normally see as effector CD4s that you're used to seeing in virus studies and normally. So, I'll have to come back with that, but these cells are linearly associated with smaller tumors. If you look at the CD8s, what you see is that there are really two populations where the numbers are associated with smaller tumors. These cells, which are effector cells, they have some PD-1 because they've been activated. They have CD69, T-bet, other activation markers. But these are conventional effector CD8 killer cells. These are expanded by NACLA-4 and also by PD-1. PD-1 doesn't do anything to these CD4s but does expand these effector cells. And then there's this other population, which I'm sure everybody knows about, these cells that express high PD-1, TIM-3, LAG-3, so 3 major co-stimulatory molecules. And these cells are considered exhausted by John Wherry and others. And have epigenetically been changed to be more or less permanently, although there's some debate about that, locked into this phenotype. But CTLA-4 doesn't really touch those, but PD-1 does expand them a bit. And so, I just point out that these cells still have that phenotype when you give the antibody. So, these results agree with with the work of John and many others. And so this is, I think, why you have to keep giving the antibody because they can— and we know from these studies that while the antibody's there, they'll proliferate. But when you take— when the antibody goes away, they'll stop proliferating. That's why you have to keep giving it maybe until that last tumor cell's dead. So let's just summarize it with CTLA-4 again, these ICOS Th1-like cells., and these effector CHs expand. And then in PD-1, these differentiated effectors also, and these exhausted cells. Of course, since the combination is now becoming the predominant treatment over the monotherapies in many cases, the question is, what happens when you put them together? Is it the same, or is it a modification or something completely different? So, We'll go through this same thing. We add another group which has combination therapy. So let's just look first at these exhausted cells, PD-1 high, LAG-3, TIM-3, that are not expanded by CTLA-4 but are expanded by PD-1, and again are associated with smaller tumors. If you add CTLA-4, you actually see that the number of those goes down significantly, way down below control, untreated mice even. Below that of CTLA-4 alone. So, something's happening to those guys, and I'll come back to that in a minute. But if you look at the terminally differentiated cells that are expanded by CTLA-4 and PD-1, a little bit more by CTLA-4 than PD-1, when you do have a combination, you expand them a lot more, as you might expect, because you're having the effect on them during priming, but also Taking, taking, uh, you know, taking it off after activation with PD-1. So just to go back to the exhausted cells for a minute, this is what the Spearman coefficient is. If you look at association with tumor volume, negative association either in untreated or PD-1 monotherapy. But when you add CTLA-4, they not only, as I showed you earlier, do the numbers decrease, but they become irrelevant. To the clinical activity. So, a bit of a surprise. And but if you look at the ICOS-positive CD4s, again, those go up in combination therapy even more. The biggest expand from CTLA-4 working during priming, but also again, because they activated, they have PD-1, it expands their ability to proliferate after activation through inhibiting the PD-1 that's on them at that point. So anyway, you collapse this, this, you know, 2-cell type with one different with the monotherapies down to this ICOS-positive CD4s and just the effector cells, not the exhausted cells when you do the combination therapy, which has some implications, I think, to the way this is done because now patients get an initiation dose of both antibodies, and then they have maintenance doses of PD-1, which may not be necessary depending on what's going on with these cells. And so the question is, you know, what really happens to those exhausted cells? Are they converted back into CD8 effector cells? Well, I think since they're epigenetically fixed, that may be difficult. John Wherry, I think, thinks that that's not very likely, but others think it— and just, I think, fix it. It might be possible if they're not completely fixed. Anyway, it's an open question right now. But I think another possibility that I find very interesting is somebody who's interested in just the fundamental importance of co-stimulation is that maybe you can prevent exhaustion by— if CD28 co-stimulation is readily available because you continue to block CTLA-4 and have to remove its inhibitory effects, we have even more robust CD28 co-stimulation. So, we don't know the answer. We're looking for mechanistic signals. But I want to talk a little bit about these inducible co-stimulator-positive cells. ICOS is a member of the CD28 CTLA-4 superfamily, as is PD-1, as you can see here, the phylogenetic tree. It has its own ligand, ICOS ligand. But of course, immunologists, when you see a CD4 cell with ICOS, will say, well, it's either a follicular T helper cell with Bcl-6 and involved in Th2 cytokines, or, or, um, got a 3 with TH2 cytokines of class switch recombination with B cells, etc. Or it's one of the regulatory FoxP3-positive CD4s that, that make IL-10 because it regulates IL-10. But their role in cancer was first shown by Pam Sharma in 2006 in a neoadjuvant trial that she was doing of ipilimumab in bladder cancer, way before it was even approved. So what she found there in her clinical samples from the tissues— this, this was a neo-adjuvant sample because it was a workaround at the time. You really weren't able very easily to get biopsies from patients on trials for research studies. You could only get them for, you know, prognostic or diagnostic value. So Pam pioneered the use of, not only neoadjuvant studies where you give them before and after surgery, but also in these studies doing, doing the studies before surgery in patients with localized disease that were going to be treated by their surgery, namely patients with pancreatic— sorry, excuse me— with, with prostate, bladder, and kidney cancer that were going to removal. So in bladder cancer, she got the whole bladder after surgery by giving them a couple of doses before they went in for the standard of care. And was able to look at what went on. One of the things she found was that 3 of the 12 patients in this small trial were cytogenetic and pathogenetic complete responders. There was no sign of tumor when they finally went to surgery. In any event, that finally led later to a full-on clinical trial, which led to the approval of these for treating bladder cancer. In any event, what she noticed was there's a 2-fold increase of these ICOS-positive CD4s in tumor, but also in blood. And that she showed by using NYESO1 as a model target that the patients whose tumors expressed NYESO1 and had CD4-positive T-cell-specific red wine, so on, those that produce gamma interferon TNF-alpha were all in the ICOS-positive fraction of CD4 cells by, by flow cytometry. And then while— right after she made this observation, this is while their Phase 1 was still going on— she shared that data with Jed Walchok when I was still at Slow Caring. And what Jed and Pam found was that the sustained increase of that phenotype— they had to be up or for the entire period of treatment with surgery, or there was no clinical benefit, there was no survival benefit whatsoever, suggesting that it's at least a pharmacodynamic marker of VIP's biological activity. But subsequent mouse models, well, I'll just show you the mouse models here. So, the question is, is this really playing, you know, a role? Having made that observation in the patients, we can take this system back to the clinic and test the hypothesis that it plays a role. In the clinical activity of NSC2A4. This shows with this B6 model I told you about where we use a GM-CSF-expressing vaccine with it, cure 85 to 90% of the mice. But if you do that same therapy at mice that lack either ICOS or its ligand, you lose half, at least half of the therapeutic benefit. So, That shows that ICOS does contribute to the clinical activity of CTLA-4. And so, we wanted to go on further and ask, could you use ICOS as a target to increase efficacy? And so, we collaborated and made this vaccine where we altered the cells, these B16 vaccine cells, to express ICOS ligand.. And so the experiment is pretty obvious. We just gave the mice tumors treated with anti-CTLA-4 with irradiated, just vanilla tumor cell vaccines, or with the tumor cells which have been transduced to express ICOS ligand, and then looked at what happened. And what you can see here, it's very interesting, is these are all the monotherapies, but this is the, the ICOS, the vaccine that does not have access ICOS ligand. And this is where you have ICOS ligand. You go from a little over about 25% efficacy back up to 90%. So, you get a tremendous boost of the therapeutic effect by engaging ICOS. And this is through ICOS signaling because if you do exactly that same experiment in mice whose T cells do not have ICOS, having that receptor around doesn't make any difference. It does require signaling. Through the ICOS molecule to realize that clinical benefit. And so, this is a potential way to benefit. So, oh, we do have some work from Naveen now. We recently submitted this. So, we wanted to look and see what happens here. So, we looked at single-cell RNA sequencing to look at what happened in CD4s and CD8s. So, just to orient you a little bit here, Tregs up here. It does look like there's some depletion when you treat with anaciklafor. These are the effector CD4 cells here. These Th1-like effectors that you see when you get CTLA-4, these red populations, you can see those are greatly increased when you have IVACs around. And the complexity of those goes up. There's just an even more functional type of those. If you look down at CD8s, what you see, again, it's almost a disappearance of the exhausted T cells and an increase in the effector-like CD8 cells. Big bang for the buck at both CD4 and CD8 responses when you add that IVAX to the C4 backbone. If you look at myeloid cells, it's also very interesting because here is all the— the cost of these dendritic cells here, monocytes and macrophages here. These seem to be the suppressive kinds with arginase and other inhibitory genes. And these down here seem to be the more inflammatory type.. But if you look what happens with the combination, the inflammatory— sorry, the suppressive myeloid cells both almost go away, and you greatly increase these inflammatory myeloid cells. So this combination therapy really shows your improvement both in the T-cell compartment but also in the myeloid compartment, I think, which again provides support for doing this and also looking for ways of additional ways of influencing the myeloid cell. So one of the things that was curious about this was, you know, you see this with NACLA-4. We did not observe these cells in mice treated with PD-1. And so we went back, and that's if you just looked at the simple standard, you know, multicolored flow without all the markers that we had for CyTOF. But if you go back and use those markers. And look at what happens when you knock cells out. You get this pattern. So these are wild-type CD4, wild-type, and HETS in green and blue. But if you knock it out, you've got these patches of cells that don't exist that we can't detect at levels, you know, at about a half percent or a percent, maybe, degenerous, including particularly this continent down here, which is not connected with the mainstream at all, mainland continents at all. You know, very, very different, almost Australia-like in its isolation from the rest of T cells. So, just to show you what those are, the cells which are not detectable are these clusters here where the blue and the green are on this axis here, undetectable. And so, there are 4 types of those, as you can see here, this one, this one, this one, this one. And if you look at what they are, These are precisely those that you see in the mice treated with anti-CTLA-4 or the patients treated with anti-CTLA-4. You can also not find these cells if you do the same analysis on PBMC from humans. So these cells don't appear to exist normally, at least unless you either, unless you get rid of iCaS, sorry, of CTLA-4 either with an antibody or genetically here as I've shown you. When you do that, you generate this new phenotype of T cell, which is again has T-bet. So, it's a Th1, has ICOS, it makes a lot of TNF-alpha and gamma interferon. You also have these other three, and I'm not going to go into it, but it's just some time. But this cell has BCL-6, GATA3, and ROR gamma T, three different major transcription factors, which according to the textbooks shouldn't be in the same cell. But those cells— Spencer came up with a way of sorting them, and they probably could recover the cells for functional studies, and they're making all of those cytokines actually. So, you know, some new cell types in the CD4 population to think about that are present when you don't— we've not confirmed these other three with antibody treatment. This is the only one that increases enough by antibody treatment versus to see. But again, back to— for those non-immunologists, this is an immunologist's view of T-cell development. They come out of the thymus as naive cells with CD4, with a class II restricted MHC receptor. And depending on the signaling pathway that they get, the kind of cytokines that are around, the strength of signal, etc., it's pretty complex. But you end up with cells at the end of the day that, you know, this is a partial listing of them, but that have a particular transcription factor that regulates cytokines, and different cytokines adopted, you know, adapted to different functions. And again, T-bet associated with gamma interferon and in these cells that we got. And something I'll point out is that Kim Bottomley in the '90s, as soon as the Th1, Th2 cytokine paradigm had been established. She showed that stronger TCR signal strength, everything else being the same, tended to drive cells towards a Th1 rather than a Th2. Anyway, so I'll come back to that in a moment. But what our data suggests is that it's really not quite this binary or simple, and they're probably intermediates that might be what we're capturing. But we think what's going on You know, what we've ruled out is alterations in the thymic development. The thymus from the CCLIF4 knockout mouse is absolutely indistinguishable in any way from wild-type thymus until the cells have exited and undergone priming. We also don't see any enhanced proliferation or activation leading to these cells. And so, it seems to be the deregulation of phenotypic constraints may remove barriers which constrain the types of the fates that are available normally to developing CD4s. And we think that may be due to removing CTLA-4, which allows effective— more effective CD28 co-stimulation. If you think about it, the TCR signal is a combination of the initial TCR signal, which the— along with the amplifying signals from CD28, which has been shown to do essentially the same genes, just in more quantities. We think that this may— by removing that constraint, you get cells that have this supernatural, if you will. I don't mean Halloween, I mean levels above what would ever happen in nature, because you never have an absence of CTLA-4 genetically. You may have various levels of it, but it's never absent because that's lethal. With the antibody, you can inhibit that, and we think you're removing those constraints and revealing these new T cells. So, anyway, it suggests also that those other cells that we showed, that there may be, you know, some back and forth as suggested a long time ago by John O'Shea and the late Bill Paul. But anyway, it tells you that you need to really think about what's going on when you treat with a drug that especially, you know, that can affect the differentiation of T cells. Having said that, Having said that, I will point out that if you look at the CD8 compartment of CTLA-4 knockouts, there's difference in ratios of different CD8 types, but there's none that aren't found in germline. And if you knock out PD-1, again, you see difference in ratios, but you don't see anything appearing that wasn't in wild type or in HEDS there. So, this, what I've just shown you, is unique to CTLA-4. Knockouts, which again is, I think, because of where it works in T-cell development. It works the earliest steps of priming rather than on functional aspects of the cell after it's been terminally differentiated. Okay. Anyway, so this last thing that I talked to you about is that people— this memory. And so one of the things that people have noticed is that CTLA-4 does seem to be the durability of more durable responses. So Stephen Mock in the lab has been addressing this for a while now, and this just sort of shows what you can get if you give this irradiated tumor cell vaccine, no live tumor cells, but NIPD-1 or CTLA-4, and wait 70 days or so and then give them tumor cells. What you can see is that the naive cells, the T cells, the tumor cells grow pretty fast, With PD-1, they slowed somewhat, but with CTLA-4, most of those mice are totally immune and reject the tumors. But there's probably— there's perhaps some problems with residual cells. So we did it also with irradiated cells, a different kind of vaccine that didn't have GMC in itself, and actually basically observed the same thing. But a purist would say you still have to worry about tumor cell bits, especially if you do this with live tumors, little micro metastasis. So you get a continual stimulation. So we did the best we could and irradiated— sorry, this is just to go to the end here— immunized with peptides, ovalbumin peptides, and then test the responses with using tetramers. And what you can see here is the initial phase, and then after memory, when they're re-challenged just with peptide, what you can see is the recall response with CTLA-4 is much higher than it is with PD-1, which is barely better than wild type. And if you take those cells from there and transfer them into mice and look at— actually, excuse me, if you take those cells at the end of priming, which is about right here, and ask what's going on in those cells, what you see is a pretty startling difference that we think may account for this That's by measuring TCF1 and TOX levels. And you can see here with CTLA-4, the predominant factor is TCF1. PD-1, very few of the cells have TCF1. And the reciprocal is true when that initial bump was with PD-1. Very few TCF1-positive cells, mostly TOX-positive cells. We think that may be the fact that makes them more powerful. So if you take those cells, the ones that had more of the TCF1 than the toxin, you transfer them on equal cell number basis, then re-challenge the mice with the tumor expressing OBA. This is the first time tumor's ever been in that mouse. You can see that those that, that have the higher TOX levels, much more capable of really getting rid of the tumor than those that— sorry, TCF much higher than those that had predominantly TOX. So I think it's this ratio, the fact that CCLA-4, even in the early response, induces more cells with TCF than with TOX, explains this memory thing. And so anyway, I'll just close there and just point out that this figure, much to dismay of a lot of us in the field, is not very different than the one that we would have shown you in 2011 or 2014. The only molecule besides CTLA-4 and PD-1 that's moved over into the FDA-approved category is LAG-3 that I told you about. That's part of this I mentioned earlier. It's part of this triplet, LAG-3, TIM-3, and PD-1 that are really highly expressed on exhausted T cells. The others really haven't moved, even though these molecules in in vitro studies and often in preclinical studies look like they're pretty powerful, including some new ones. One of the ones that we've been looking at now are predominantly myeloid cells express members of the LILR family, LILRb2 and b4. And so, we've been— we've shown that LILRb4, for example, is not only on inhibitory myeloid cells, as are these other two, VISTA and LILRb2, but it's also a subpopulation of FOXP3-positive regulatory cells, which lose their regulatory ability in LILRb4 knockout cells. And so do myeloid cells. Anyway, so we've got all these possibilities, and I think that what we need to be doing is more combinations but analyzing them in a different way. So there are trials going on now, about 3,000 at one check, one count a few years ago of PD-1 plus other things. And the, the standard way was to take two things, you know, and do a clinical trial, 50 patients or so. And see if you get a clinical signal. And I think that's a lot to ask when we know some of these molecules, like VISTA for example, doesn't even appear to be induced until after you've given CTLA-4 or PD-1. So a lot of these molecules haven't had single agent activity, so they're not tested that way. You wouldn't expect them to because they're often not induced until you give another checkpoint. But everything changes after you give a checkpoint. So giving two things together and asking for a clinical response is asking a lot. But again, we now have examples I've shown in the mouse of positive things like ICOS. Anyway, we're beginning to see more conventional therapies. More recently, there were combinations with surgery in these neoadjuvant trials, both in melanoma and in lung cancer, have had response rate as high as 70%. Beginning. And it's very important— so it's been shown to be very important to start with the checkpoint blocker prior to the surgery and continue it afterwards for maximum effect. But that's a really promising area. And I think that as we come up with ways of changing the way that chemotherapy is given, that radiation is given, so you don't try to kill that last tumor cell and in the process kill all the immune cells, we could make them more friendly and try to use those conventional therapies as vaccines, I think we're going to see a lot of power in those kind of conditions and perhaps targeted therapies as well. But there's still a lot of work to do. We still need to base all this on basic fundamental understanding of basic fundamental biology, not just of T-cell biology but myeloid cells and other, other members of the adaptive and innate immune system. But I think where— what we've learned is that, you know, for the— since the dawn of the age of chemotherapy, oncology has been taking large amounts, large numbers of patients, treating them, and seeing if a statistically significant number of them show higher median survival. IPI was the first immune-based drug to do that, again, 4 months in melanoma. But it also had a 10-year tail on it. Plus, actually, there's a couple of patients now that are 21 years out and no retreatment. This one is an aspirational line. Instead of concentrating on this, we need to be working on combinations that are done along the way that, again, that I'm showing on the immunotherapy platform and pioneer that is treat a small number of patients with standard care, take— get biopsies, look at what happened. There's a new checkpoint, take that back to mice, or anything significant you observe, a new transcription factor and a new genetic regulator, epigenetic regulator, and then just take them back to mouse models, see if there's any effect, and then bring them back to patients again. I think we're going to I have to use that kind of intensive scientific study to do this properly and reach that goal. But the good news is I think we know it can be done if we just, you know, work hard, pay attention to basic biology and learn from it. And we can reach a point where we not only raise the response rates to the tumors that I've shown you and talked about, but also done something in glioblastoma and pancreatic cancer. So with that, I'll stop and thank you all for attending. As I understand, there's no questions. So that's it.

There'll be some questions on Twitter, I suppose. Anyway, now you. Yeah. Thanks, Jim. That was wonderful. The differences in memory are really exciting. Makes me think about a couple of things like, you know, potentially different dendritic cell interactions or the differences in maybe PI3 kinase signaling that might be emanated by blocking CTLA-4 versus PD-1, because that would be a fun, you know, a major regulator of TCF. So that's really, really cool. Thank you so much for sharing this. Yes, so please ask your questions, uh, to, to Dr. Allison, uh, by Twitter. You can go to his own Twitter site at Jamalison or go to the Global Immuno Talks, and we will try to get those answered for you over the coming days. And again, Jim, thank you so much for your time and for talking to the world. And we're just very, very, very inspirational. Thank you. Thank you, Sue. Bye-bye. Bye.