Dr. Peter Fecci from Duke University explores T-cell dysfunction in glioblastoma.
Uh, thank you very much, Wiley. It's been a pleasure, uh, working with you over the last bunch of years. Uh, Wy and I go back to about 2018 or so now, I think at this point. Um, I recognize that, uh, typically on a Monday morning, maybe you guys don't start your talks or your days with T cell dysfunction talks, uh, so I will try to make this, uh, as engaging and as exciting as basic science can be on a Monday morning, um. Uh, I, and I do think it's more part of the reason I don't have any relevant disclosures is because this really is a research talk, uh, and just to make sure that for your sake I obviously do have clinical, but none of those are relevant to this conversation. Uh, so, uh, and there it is actually, so again, not relevant to this, OK. A lot of you folks will know that there are a variety of immunotherapies out there that you see advertised on television all the time, things like, uh, nivolumab, uh pembrolizumab, you know, you'll see the name brand versions of those which we won't mention for the purposes of CME, uh, and those, those drugs are very, very common. They're checkpoint inhibitors. They're used in a variety of cancers, melanoma, lung cancer. But today there are still no immunotherapies that are FDA approved for GBM, and there's a variety of reasons why immunotherapies have been less successful. A lot of it has to do with the fact that the immune system in these folks is truly shot, and that's a little bit interesting when you think about the fact that these tumors are confined to the brain and really don't metastasize out of the brain. And yet despite that, what I'll try to convince you today is that glioblastoma in particular exerts a tremendous influence over the immune system, not just locally in the intracranial compartment, but systemically, and some of the things that it does are actually not specific to GBM but actually a feature of any tumor that pops up in the brain. Even if it is a metastasis from another cancer, so we've known in GBM for many years, actually dating back to the 1970s, there's two guys here, Brooks and Rosman, who are actually out of Kentucky. One of them recently passed away unfortunately, but their work really dates all the way back to the 1970s when people actually recognized that the immune system in these patients didn't work very well. They lost their DTH responses, T cell number was reduced, uh, and then T cell function sorry, was also reduced. And they kind of just characterize these as quantitative and qualitative deficits, but a lot of people presumed that the reason the T cells weren't there or weren't working was because essentially you had all these therapies like chemotherapy and radiation and steroids, uh, but it turns out actually that even if you start with patients that are brand new diagnosis with GBM and they show up in your emergency department long before they've had steroids, chemotherapy, radiation, uh, their T cells really don't work and so. Back in the 1970s we just said well they were reduced in number and function, but now we can start to think about things in more modern modes where we can look at how dysfunction is described by basic immunologists and so this was a paper that our group put out some years ago that kind of takes all the things that immunologists know about T cell dysfunction and looks at brain cancers and other cancers and divides T cell dysfunction into a little bit more of a scientific grouping. And so this, this is the grouping here, uh, kind of going into each of those and what they look like is beyond the scope of this. Uh what I will say is that we spend a lot of time now on the bottom two categories, ignorance and exhaustion, and so that's what I'm gonna go through with you today. So what is T cell ignorance? It's basically exactly like it sounds. The T cells are kind of stupid because they don't really know what to do. So why would a T cell not be able to do its job, uh, secondary to kind of ignorance? Well, it might not have ever seen its antigen, or the T cells might be stuck somewhere or they might be unable to get to their targets because those targets live in a privileged compartment. We'll go over that more in a second. This was the key paper we put out about 6 years ago in Nature Medicine that talks about this idea that T cells in patients with GBM and in other intracranial tumors actually will become trapped in the bone marrow, unable to get out, and therefore unable to travel to the brain in order to do their job and fight the tumors. So talked about why might T cells be ignorant of their target. Well, they might be inexperienced, they may never have seen that target. The privileged site like the brain harboring a tumor where the T cells really can't get in, that may be a factor. We certainly know that the brain is a little bit privileged. It's really more immune distinct than it is privileged, or those T cells may be trapped somewhere where they can't get to their target, and that's of course what I'm going to talk to you about here. So again we kind of noted, I guess our little animations don't work, but that's OK. Hopefully that won't be a problem for the rest of the slides. Well, imagine that when I push a button, all these little green dots disappear, meant to reflect the idea that essentially T cells are reduced in number in patients with GBM. And again, we've known that from the 1970s, but people had again had always said that that was likely due to all the therapy that those folks have been exposed to. What we found actually is that the average patient that comes into the into the emergency department with the GBM has about a 30 to 50% reduction, sometimes more in their T cell counts, and actually 20% of patients that show up in our emergency department with a newly diagnosed GBM will have CD4 counts that are less than 200, which is essentially the demarcation for progression to AIDS in. Infected individuals, so that's pretty severe lack of T cells and as a result, we did some studies because people will say, well, does that mean that they can suffer opportunistic infections and it turns out that they can. In fact, 25% of PCP pneumonia cases at Duke Hospital in the last 30 years have been in patients with GBM. So they do in fact suffer things like optimistic infections, and we said, well, where are all these T cells going? Maybe they're being trapped in the spleen, other lymphhoid organs. So we started to look at abdominal scans of patients with GBM and we looked in mice with GBM and actually it's exactly the opposite. The average person who comes into the emergency department with the GBM also has a 30 to 50% reduction in the size of their spleen. Uh, for which there is no medical term because that just never really happens, so we have coined it hyposplenia. The only people really prior to that would be people with sickle cell anemia who have splenic infarcts. Uh, and when we go to mice with GBM we see exactly the same thing lympopenia, lack of T cells in the blood, and we see that their lymphoid organs, including things like their, uh, lymph nodes and spleens and thymus actually all dramatically shrink in size. So where are all the T cells? Well. We decided then that if they're not in all these other lymphoid organs, maybe there's a problem with production, so we turned to the bone marrow and we're very surprised to find that instead there were 56, sometimes 10-fold expansions of T cell numbers in the bone marrow and that's not just true in mice. So when we went to patients, this was a study that I was working on when I was in residency at Mass General. so we uh we started a clinical trial. Up in the Harvard system where we started taking bone marrow aspirates from patients while they were with consent. It was a study. We didn't just do it under general anesthesia for their resection, and we started to compare that to spinal control patients who were there for fusions and were getting marrow taken out and lo and behold, it was the same thing was true in people. It's not 100% penetrant, but sometimes you can have 20-fold expansions of T cells in the bone marrow of patients with GBM. What's interesting then is that we said well maybe this is a glioma specific phenomenon, you know, uh, and so we decided to put a bunch of different tumors into mice because we do see the same exact phenomenon in mice uh as you saw. So we started putting, see if the little arrow comes up here, yeah, OK, so CT2A is a glioma. EO771 is a breast cancer. uh B16 is a melanoma, and LLC is a Lewis lung carcinoma. So these are all the different histologies that you might find, uh, as, as responsible for brain metastasis, let's say. And if we put those gliomas either sub subcutaneously in the flank, or intracrannially, all these tumors in those various compartments, what you see is that these phenomenon bone marrow T cell sequestration, the same was true of lymphoid organ regression, lympopenia, they happen with all of these tumors, but only when these tumors are placed within the brain. And even if you place gliomass in the flank, uh, it doesn't happen. So that really begs the question, is this really actually a problem with the tumors or is it more of a function of the brain, i.e., when you put something in the brain, you elicit the capacity for the brain to start a mechanism that essentially keeps T cells out. Is this a novel form of immune privilege? Well, if that were the case, we would expect other inflammatory insults in the brain to do the same thing. That turns out also to be true. So when we. Look at models of stroke. We can include the MCA in mice and give them an MCA hemisphere stroke. What we actually find is exactly the same thing that within about a week or so we see shrinkage of their spleen. We see lympopenia, we see sequestration of T cells in the bone marrow. We now know that this also happens with things like traumatic brain traumatic brain injury, viral infection in the brain. And this is a paper that we are continuing to work on to try to understand the upstream and downstream mechanisms, and we do think we have it worked out and things with stroke and traumatic brain injury tends to be transient because there's an insult and then the animal recovers. With tumors, it tends to be marked and progressive because it's a chronic process that doesn't go away. What we ultimately found is that the proximal cause of T cells becoming trapped in the bone marrow was that globally they lose a receptor called S1P1 and that receptor, uh, beyond being a pain in the butt to work with, uh, also happens to be a reasonably popular receptor. During the last 15 or so years, because people have recognized that it is actually an exit visa for T cells that allows them to leave the lymphoid organs and get back into the circulation. And so if you lose that receptor, the T cells in fact in turn become trapped in the lymphoid organs. So that begs the question, well, if you can stabilize that receptor, prevent its loss, will that allow T cells to enter the circulation more where they can perhaps be better served by immunotherapy? Uh, and so we have a mouse that we were able to, uh, essentially, um, build slash obtain. That has a knocked in S1P1 receptor that cannot be internalized. It stays on the surface and serves its exit visa uh uh role. And what we found is that just doing that alone does get T cells out into the circulation, but that's not sufficient, of course, to cure brain cancers by themselves. But if you now give those mice out of a noctin receptor and immunotherapy that doesn't work very well, you know, only kind of 10%. Long term survival. Now suddenly about half of those mice will survive long term from these glioblastomas, uh, and then if we now add things like checkpoint blockade, antiPD1, which is that novolumab drug that you see on the uh television all the time, now you see really dramatic extensions of survival and these therapies that have not worked previously in brain tumors now start to actually work. I'm very fortunate for two reasons. Number 1, S1P1 is a G protein coupled receptor and as a result, has a fairly kind of systematic and studied way of being internalized. And then Bob Lefkowitz, who is a Nobel laureate at Duke, and his Nobel Prize was for discovering G protein coupled receptors back in the 1960s. His office is about 50 yards from mine. And so our labs are now very much in a strong collaboration to work on developing a drug that can prevent this receptor from being internalized, and you can't touch the receptor on the outside, which is normally what you would want to do with the drug. You actually have to touch the internal machinery here to prevent the receptors internalization. And so that's a little bit trickier, but we now have some drugs that are actually doing what we want them to do, and we're beginning to build chemical analogs to make them more stable in vivo. We work with a company in the Triangle down by us called RTI and we've started farming out some of the medicinal chemistry and we'll hopefully have something that comes into the clinic within the next couple of years. Luke Waksmith is the MD PhD student who's been working on this of late. So that is the drug side of it, but then the real question is how do you get from a brain tumor to global changes in your T cells so that they can lose a receptor and become trapped in the bone marrow? Well, We figured whatever's happening in the brain to elicit this mechanism, it has to be some type of cell type that's specific to the brain right there because it doesn't happen when these tumors are in the flank and then it's also eliciting some type of communication between the brain and the rest of the body. So what are the forms of communication that the brain can have with the rest of the body? Well, a lot of you are familiar familiar with the synthetic nervous system, so fight or flight, right, uh, kinds of things that make your eyes get big when you're when you're stressed, etc. make you sweat, all that. Uh, and then the HPA axis, uh, is the other route. We had a lot of reasons again beyond the scope of this talk to suspect that maybe the sympathetic nervous system was involved in some of this, so we decided to study that a little bit more and try to figure out is there perhaps hyperactivation of the sympathetic nervous system in mice with brain tumors in people with brain tumors of any histology and all these other processes like traumatic brain injury, etc. and could that be causing some immunosuppression. I will skip all the details of this hypothesis, but we did have this kind of, you know, somewhat long winded hypothesis for how this might be working, but what it really boiled down to was that at the end of the day we needed to show if this hypothesis were true that there's increased levels of catecholamines, which are the chemicals that are kind of representative of the sympathetic nervous system. That bind the beta and alpha adrenergic receptors that are present on the surface of cells like T cells and then we had all this long winded hypothesis for how doing that would lead to essentially reduction in things like S1P1 levels and immunosuppression and all this other stuff, uh, but we had to really start with the question of do we find systemic catecholamines elevated in the setting of intracranial tumors specifically? The answer turns out to be yes. So in those same models glioblastoma, melanoma, lung cancer. If we put tumors in the flank or in the brain, we actually see dramatic rises in epinephrine and norepinephrine in the blood of these mice, but only when those tumors are placed within the brain and not when those tumors are placed within the flank. Uh, Selina Laurie, Luke Waksmith, uh, are the two grad students who were working on this, uh, and so then we said, well, what about people? So we went to patients with GBM, whether newly diagnosed or recurrent, looked in their blood, and sure enough we see, uh, increased levels of epinephrine, norepinephrine. Uh, and then we wanted to kind of have a good sense of, well, you know, there's a variety of different receptors that can respond to increased catecholamines there's alpha 1, alpha 2, alpha-3, beta 1, beta 2, beta 3, so forth and so on, uh, and the receptor that seems to be present in the tumor and on the lymphocytes in these folks is mostly the beta 2 adrenergic receptor. So we kind of were interested in, well, what role is the beta 2 adrenergic receptor playing in all this. And so what we found is that uh intracranial tumors can actually elicit a very profound systemic T cell dysfunction, and we knew that, but what we also found is that we could recapitulate that exact mode of dysfunction simply by placing pumps into the flanks of mice and pumping them filled with adrenergic uh compounds, and that could either be um uh things like catecholamines which recapitulated this, or we could actually use beta 2 and beta 1 selective. Adrenergic agonists and what we found is that beta 2 agonists in particular seemed to recapitulate a lot of the dysfunction we see in T cells, which are the kind of things that you're seeing pictured here. I won't go into too much detail here, but needless to say, those T cells don't function well and beta 2 adrenergic stimulators are capable of recapitulating that same mode of dysfunction. So that begs an important question. Does that mean you can reverse a lot of this immune function simply by giving a beta blocker, i.e. the drugs that you've heard of all the time, propranolol, atenolol, metoprolol, all of those, right? Well, propranolol is the one that probably has the most beta 2 activity. It's non-selective for beta 2 and beta 1, but we started giving propranolol to mice with brain tumors, and we showed that actually we can reverse a lot of the immune dysfunction that they have simply by putting mice on propranolol. So that also reverses not just the systemic immune function but it kind of reshapes the tumor milieu, the tumor micro environment and reverses some of the dysfunction that we see there, um, uh, this is just more of the same. But more importantly, if we give mice with beta blockers that have brain tumors and then now start giving these mice immunotherapies that again didn't really work before, it's not the most impressive thing you've ever seen, but you wouldn't expect it to be this potent, and it is. We can get you know 20% long term survival. This is a CT2A glioma. It's pretty tough to cure. You can see that, um, you know, the immunotherapy, this green line is an immunotherapy that we typically give, doesn't do much, uh, but now by combining that with beta blocker we actually get some long term survival, uh, and same thing over here with antiPD1. So that's true in mice and that's, you know, not the most impressive thing we've ever seen, but what's, what's going on in people. Well, uh, Quinn Ostrom, who runs CB Trust, which is kind of like the national database for patients with glioblastomas that do, and then she has access to a lot of the kind of Medicare data through like the SEAR databases, and so we decided to look at patients with glioblastoma through the Medicare databases and figure out whether those that had been on beta blockers, now there's limitations of looking retrospectively like this, and so I'm not going to convince you that this is the cause, but it's certainly interesting that it will be associated this way. And we looked at nearly 9000 patients with GBM. This is Medicare data, so these are all folks over the age of 65. So you're already kind of at the lower end of the survival curve here. uh, but what you do see is that those patients who are on beta blockers actually did have dramatically longer survival. Again, I'm not saying that's the cause, it's retrospective data, but it's the association that you would expect to see. Now there are no FDA approved immunotherapies for GBM. We talked about that, so we can't really look to see, you know, how did patients with GBM on immunotherapy do if they're on beta blockers, but patients with brain metastases do, and particularly lung cancer and melanoma are folks that have had immunotherapies delivered to them and so we can go back and look through the same SER Medicare databases and look at these folks and say, you know, OK, if they were on um beta blockers, did their immunotherapies work better? And lo and behold they did. So if we take, uh, this was actually um. About 25,000 patients with lung cancer brain metastases, uh, and we looked at the therapies and you can see that patients who are on immune checkpoint blockade alone and then but also on a beta blocker had uh almost 13 longer median survival, uh, and then for melanoma it's even more dramatic, 63% longer median survival if they were on a combination of beta blocker and checkpoint blockade compared to checkpoint blockade alone this is. A big deal. And what's interesting is we're not going to try to say that beta blockers don't also help with the immune response against cancers outside the brain. They do for a variety of reasons also beyond the scope of this, but the survival benefits are not as pronounced for those extracranial metastases as they are for the intracranial metastasis. We expect that because of these intracranial metastases that are probably eliciting elevated catecholamines and things that are restricting the immune response, and you're essentially fighting that with the beta blockers. So we're really trying to understand now the upstream parts of that. So how do you get from a tumor to hyperactivation of the sympathetic nervous system, because we haven't explained that yet. We have some collaborations with some groups out at Stanford working to understand that and ultimately I think we really suspect that we may have the answer to that, but it's probably, I imagine that we won't have that paper done for another year or so, but it really could unlock some really interesting things about intracranial pathologies in general. So, uh, this is a multi-part talk, uh, so if you were hoping you're done, I'm sorry, I still have a little bit more to go, but, uh, exhaustion, I said we're interested in the kind of bottom two categories here that was ignorance. Now I'm gonna work into exhaustion and then the last part of this, after I tell you that everything is very sad, uh, we do have some kind of hopeful parts at the end for some things we found that may help immunotherapy really work, uh, significantly better, so. Exhaustion What is T cell exhaustion? Well, this is also kind of what it sounds like where the T cells become tired after they've been activated. Uh, this was a paper that we put out a few years ago, uh, kind of uncovering that exhaustion, uh, like everything is kind of worse in GBM. That's not a surprise. But also providing some mechanistic insights into how exhaustion may develop. So we have been really focused in the brain tumor immunotherapy world for a long time of, well, the whole problem here is we just can't get the immune system into the brain because the brain is immune privileged. If we could simply get T cells through the front door, everything would be fine. But it turns out that simply getting T cells to the front door of the tumor is not sufficient. Our T cells are not that gifted, and part of the problem with that is that they don't function well. And one of the big problems with their function is exhaustion. And interestingly, checkpoint blockade therapies are really aimed at reversing T cell exhaustion, or at least staving it off, but the exhaustion is particularly severe in GBM, which is potentially why those drugs haven't worked in glioblastoma. So exhaustion was initially characterized in chronic viral infections in mice, uh, and again it, it's a tiring out of the T cells, so they have to initially be activated and then they they become exhausted in a very stereotype way, uh, that is a description scripttional program within T cells. These markers here, uh, CHA 4 PD1, uh, where's that little area you are, uh, it's going across screens anyway. Uh, CLA4 PD 1, and then all the others here, these are kind of typical markers of T cell exhaustion. CA 4 and PD1 are the canonical ones. Uh, these, uh, oh good, it, it skipped right over the next part of that animation which showed all the secondary messenger cascades which are not very exciting anyway. Um, but really this has arrived as essentially an evolutionary advantage for individuals because it's a way of kind of modulating your T cell response if it's, if it's a little bit too active and you haven't really gotten rid of the pathogens, so. It really is a a a stalemate between your immune system and the pathogen when there's something chronic around that you're not clearing because if you keep activating your T cells at a time when you haven't really been able to get rid of the pathogen, then you're gonna end up with a lot of kind of uh collateral autoimmune damage. Uh, so, uh, we did our homework and made sure that we proved that everything that was going on in GBM was in fact exhaustion. You have to show all these different things. Uh, we showed that this was indeed bona fide exhaustion, the same type that you would see in viral infections, uh, but, uh, we also showed a couple of other interesting things. So first off, so CT2A and SMA 560 are glioma models. This is again breast, melanoma, lung, and you see that this red heat map shows how bad the exhaustion. is on T cells in glioma. Uh, what's also interesting though is that no matter what tumor we place and no matter whether we put it in the brain or the flank, each tumor kind of elicits its own very specific exhaustion signature as far as what the T cells infiltrating the tumor express. And so you can, we could almost take a tumor out, not knowing what we'd put in, match it up to one of these signatures, these fingerprints, and we could tell what tumor we'd put in simply by the type of exhaustion profile uh it elicited. We now know that there are two subsets of exhaustion, so there's kind of a progenitor exhaustion and a internal exhaustion, and uh uh T cells progress through those en route to that terminal exhaustion that's not very helpful. Uh, and the reason that progenitor exhaustion is kind of important is because progenit exhaustion is still this kind of renewable state where checkpoint blockade is able to work on the T cells and make things OK. Thermal exhaustion really not so much, although those T cells can kill tumors, but they're kind of like a dead end, if you will, uh, and so the idea is that you really kind of want more progenit exhaustion. You really don't want a lot of terminal exhaustion at the end of the day. And probably the greatest discovery uh that I think I've ever had in science is that we have found that the progenitor exhaustion to terminal exhaustion ratio or the PEter turns out to be like really important for what's going on with the tumors and so we actually have this paper out my grad students hate it because they hate having to put my name in it but in any event. Uh, this paper is about to come out next month in immunity, and we will show that actually the PE ratio turns out to be very important and uh and if you increase that ratio, i.e., increase the amount of progenerative exhaustion relative to terminal exhaustion, you can actually improve immune responses, uh, and so these are just kind of, uh, uh, sort of involved figures that really show how things progress over time using things like single cell RNA sequencing, uh, and showing like what types of functions progenerative versus terminal exhaustion maintain. But the real thing is that we show that actually it is tumor specific T cells, T cells that are specific for tumor antigens that do tend to become more exhausted, which is not surprising because they have to be activated first, and so we can actually track T cells that are specific for a tumor that has a model antigen like TRIP 2 in it. We can show that it's all the TRIP 2 specific T cells that become exhausted. Most people think that the reason T cells become exhausted is because they're seeing their antigen chronically on the surface of tumor cells, right? And so. That antigen presentation by a tumor cell on its surface in the context of like MHC molecules is kind of suboptimal and then the T cells see it in a suboptimal way and become exhausted and don't function. We wanted to kind of understand what really is the role of tumor man and presentation in eliciting T cell exhaustion, so we knocked out a component of MHC molecules in those tumor cells by eliminating something called beta microglobulin. We did that with CRISPR, which is just a way of eliminating genes. Some of you have probably heard that because that's also a very big discovery in science. And what we've now found is that. In fact, lo and behold, if you take away the capacity of tumor cells to present antigen to T cells, it actually has no impact on exhaustion whatsoever, and this had been the model for years is that people just presumed it was T cells seeing their antigen over and over on tumor cells that caused them to be exhausted, but it turns out actually that's not the case. What it actually is is all of the myeloid cells in the tumor micro environment like macrophages, etc. uh, which are not the best antigen presenting cells, uh, that actually elicit that and so we did these very complicated experiments with bone marrow transplants creating bone marrow chimes between beta2 beta2M knockout. And wild type mice so we could remove the capacity of all the myeloid cells in an animal to essentially present antigen but leave antigen presentation by everything else including the tumor intact. And when we eliminated the capacity of myeloid cells to present antigen, lo and behold, exhaustion kind of went away. Uh, and so now we've shown that when we do this in other cancers as well, whether that's going to be, uh, melanoma, um, uh, actually this is glioma, this is melanoma down here, uh, if we remove macrophages actually and remove myeloid cells or remove the capacity of myeloid cells to present antigen, we see an increase in the PE ratio within those tumors and we now see immunotherapies actually start to work better, uh, and we see less exhaustion, uh, and so that's that's the paper that will be coming out next month. And for us that's a big deal because it really undoes the model for how people have supposed exhaustion arises in the tumor microenviron. It also provides you the capacity to target the interactions between myeloid cells and T cells with a variety of drugs or blocking agents, etc. in a way that now also should make immunotherapies work better in these models. Uh, so Jessica Weibo Plania was the grad student that just, uh, uh, finished her PhD with us that, uh, she and Zana Migelbrink, uh, were the two, first authors on this paper. Uh, Zana is still around, and, and she actually, oh, OK, I don't have this in this one. Great. Uh, good one. Spare you 15 minutes of pain, but in any event, Zana is working on a novel marker for exhaustion that we've uncovered that's not listed in those ones that we showed you that actually does appear to play a very strong role in the transition from progenitor to terminal exhaustion. Uh, and we've just discovered and that if we knock that out, we actually can prevent that transition and we can actually make, uh, drugs that actually target that, uh, make, uh, some of the immunotherapies work better, and that papers in revision, uh, right now at a journal called Clinical Cancer Search. So the last part of this is you know, some hope for the future, if you will, and this is a paper that we put out in Nature Cancer last year, and it, it really kind of undoes the textbook, if you will, for the last 50 years on how T cells actually kill cancer and I think provides us a novel route for targeting certain capacities that T cells have that we didn't know that they had. Uh, so this was the paper. Emily Lerner is the MD PhD student, uh, who just finished her PhD in the group as well. Uh, she's the first author on this paper. Uh, so. The historical model for and if you kind of remember your high school or college or uh subsequent biology that you've been taught is that T cells in order to work have to be activated. Only in the context of seeing their antigen presented by MHC molecules. So in other words, a T cell isn't going to recognize like a receptor on the surface of a tumor cell, a whole protein like that. The reason that T cells work is because every cell in the body is chewing up in all the proteins that those cells are making. Into little peptides and then they present those peptides on their surface in the context of an MHC molecule. It's like a little flagpole that they put out and say, hey, look, this is what's going on inside of me, and the T cells are what go around and say, yeah, that looks OK, or no, and then kill the cell. And so for the most part what T cells are really designed to do is to get rid of viral infection because any kind of viral production that's going on inside a cell, foreign proteins. The cell puts that out on a flag and says, Hey, come and check out what's going on here. T cells come and say nope and kill it and that also, however, is the mode of immunity for how T cells recognize cancer cells because when you think about cancer cells in order to outstrip normal growth patterns, they have to either misexpress proteins or express mutated proteins. Those mutated proteins will be recognized as foreign. So T cells, we've been taught for 75 years as far as the model of anti-tumor immunity, must see their antigens in the context of MHC molecules. They must be shown that antigen first by an antigen presenting cell, and then once they see that, they go out and find that target on the surface of a tumor cell again and if it's a CD8 T cell in the context of an MHC1 molecule specifically. And as a result, one of the key things that we've been taught over the years is that in order to avoid the immune system, tumors will take down all the flagpoles, and they do. So a lot of tumors will actually down regulate their expression of MHC molecules, and we have presumed that that is a way in which tumors can avoid the immune response. That turns out not quite to be true. So this was our WTF moment a couple of years ago in the lab. Uh, so this is what a typical survival curve looks like in an immunotherapy experiment in the lab where we kind of game the system in our favor. So we take a glioma like a CT2A and we chock it full of uh an antigen that can be kind of viewed by the immune system as foreign, in this case TRIP 2, and those tumors we make it so that every single tumor cell in the tumor expresses the TTIP 2 antigen. And so it expresses it. It also processes it and presents it and puts little peptides on its surface, uh, and so that's how T cells recognize it. And you know, unlike normal tumors in the human body where like maybe 10% of the cells express a rejection antigen, in this case 100% of those cells express the rejection antigen, and that's why we can combine immunotherapies and get a pretty good survival rate here uh in this model. So I think if you remember back to the exhaustion part of the talk, I said well we created these tumors that didn't have class 1 on their surface because we wanted to study what impact that had on exhaustion. So we did this beta 2M knockout model where now the tumor cells don't have the capacity to present antigen. They still might express antigens like TRIP 2, so like if you went inside the cell, you could find TRIP 2, but they're not presenting it, so it doesn't really matter in theory to T cells. So one of the grad students just had this idea about what the hell, let's just see what happens when we put these class one knock out tumors into mice and give them immunotherapy with the suspicion being that that should essentially eliminate all immunotherapy responses, but instead the opposite occurred, and these mice actually did really, really well with immunotherapy, better than this kind of standard models and that didn't make any sense so we said, well. Maybe it's a problem with gliomas or maybe it's, you know, it's the brain or who knows, so we created beta 2M knockouts for uh CT2A again and put them in the flank, uh and uh you know in case that was a brain specific thing, but sure enough, you know, we still were able to eliminate those tumors uh perfectly well. The immunotherapy and then we created a B16 melanoma model where we did the same thing loaded it with an ova antigen uh but then knocked out beta2M so that ova could not be presented to T cells and still sure enough we eliminated those melanoma tumors with immunotherapy fine. So the whole thing is that you would suspect, and one would suspect that these tumors, even if they're getting killed by immunotherapy, cannot be getting killed by T cells because there's no class 1 on the surface of those. Uh, so we said, well, the cells in the body that normally would get rid of tumors that lack class one would be NK cells. So we'll eliminate NK cells that will eliminate this whole thing. So we knocked out NK cells from mice and, uh, uh, and actually that had no impact. And eventually when we got rid of we eventually got around to getting rid of CD8 T cells and that took away the capacity of the immune system to kill these tumor cells and that didn't make any sense whatsoever because in theory the only way CDA T cells could kill these tumors was via C class one expression of 1 presentation of tumors. Uh, I'll skip some of these very very complicated experiments that we did, uh, but the real point is just kind of what we learned from them and not necessarily how we set it up. So the trick is that we found that these immunotherapies that worked against tumors that had the TIPI antigen but couldn't present them, uh, if we gave a trip to specific immunotherapy, we still needed that trip to antigen around, even though it didn't, it shouldn't have seemed to matter because that antigen wasn't being presented to T cells. So the antigen specificity component was still important, which was a little bit surprising. Uh, and so we did a variety of experiments to kind of show, you know, how is this happening and what, you know, what's necessary, uh, and so we found that actually in order for these for T cells to kill tumor cells that lacked uh class one on their servers. Those T cells still had to be activated via their T cell receptor, which is the business end of that T cell that recognizes antigen in the context of MHC, and that you needed cells in vitro with those T cells to activate them, so like macrophages, for instance, or dendritic cells, presenting antigen to those T cells to activate them via their T cell receptor. What was really interesting though is that once those T cells were activated via their T cell receptor, they didn't care what they saw, they could kill anything. And so for instance, and this was kind of a money experiment for us, if we took T cells that had a T cell receptor specificity for ova. And we activated them with overpulsed macrophages, so now you've elicited T cell activation by over presentation. Once activated, those T cells could kill CT12 gliomas that didn't expressova but expressed TTERP 2, but it didn't matter because they didn't present it. So the really surprise factor here is that once a T cell receptor is activated, it was an indiscriminate killing machine and could kill anything whether it had the antigen of interest or had MHC on its surface. This is truly something that completely undoes, you know, essentially nearly 100 years of understanding of T cell killing. So we showed this in vivo as well. We did some other kind of mechanistic experiments, but the key point of what we wanted to figure out is how is a T cell killing a tumor cell that doesn't have its target and doesn't have MHC on its surface? Uh, what it turns out is that T cells also express a variety of things that NK cells express, and NK cells would normally be the cells that could do that. Uh, and so we did a variety of experiments to show that what's up regulated on these T cells in tumors that lack MHC is actually an NK cytotoxicity molecule called NKG2D. Uh, and what's interesting about NKG 2D is that its ligands are essentially stress inducible ligands that pop up on cells that are stressed either due to heat, cytotoxic stress, etc. Tumors also tend to have a lot of these ligands on their surface. And so what we found is that if we gave in either in vitro or in vivo in either mouse in vivo and in vitro or human cells in vitro, if we gave a blocking antibody to NKG2D now we could completely abrogate this killing mechanism and we now prevented T cells from being able to kill tumors that lacked MHC or the antigen of interest. And so we also wanted to say, well, OK, if NKG2D recognizes NKG2D ligands and we're able to kill these tumors, these tumors must have lots of ligands. That turns out to be true. And in fact, as tumors lose MHD, they tend to actually have interestingly more of these ligands. And so this is the mechanism by how T cells are able to do it. And what's particularly interesting is that if we take mouse NKG2D ligands and plug them into human tumor cells that don't have MHC on their surface, we can now make mouse T cells kill those human tumor cells just for the presence of the mouse NKG2D ligands.