Scientist Stories: Susan Kaech, Memory T-cells

Hi everyone and welcome. It's my pleasure to get to— I'm Katie Wellen, and it's my pleasure to get to introduce the keynote session and chair the first session. So we're really thrilled today to have Sue Keck here as our keynote speaker. Sue is really a leader in the field of immunometabolism, and her work has led to a lot of key insights into how tumors suppress T-cell metabolism and how that contributes to an overall immunosuppressive environment in cancer. She is a professor at the Salk Institute as well as the—

sorry—

the director of the NOMIS Center for Immunobiology and Microbial Pathogenesis. It's a real pleasure to have you here, Sue, to kick off this meeting, and I'll turn it over to you.

Great. Thank you so much, Sue. Do I need to turn this on? Okay, great. Is it still on? There we go. Perfect. So thank you. I just want to thank Jerry, Katie, and Bart, who I haven't met either, for inviting me to talk to you guys about— and I love the cross-disciplinary nature of this talk. I am not a metabolic expert. I will admit that right from the beginning. I do know a lot about T cells, which I hope you guys will start to care about after we talk about them through my talk, Rusty's talk, and hopefully some others throughout the meeting. It's a fabulous system for which to study metabolism in cells. And so as Katie said, we've been very interested in understanding how T cells function, how they differentiate, and trying to integrate that with the metabolic regulation that the cells use as they undergo through immune responses. Today I'm going to talk to you about the work of a very talented postdoc, Shixin Ma. She gives a lot of humor to the lab. She has words of wisdom from the lab, and these are all of her quotes. We need to be positive in our lab for our lab. I like that one. We should all be replaced by robots. She has something about no vacation, keep going. Here's another one. My life is already short, so I must enjoy everything. And everything is epigenetics. And so that's what we're going to talk about today. Let's see. So one of the questions that we've been thinking about for my lab for many years is how cells undergo differentiation. And in particular, how T cells undergo differentiation. And I'll get to this in more detail in the next few slides. But what we started to think more about, in particular as we started to think about the nature of the microenvironment in which cells reside, you can think about our immune system in general and our T cells specifically as kind of like a liquid tissue. They permeate, they move, they go throughout all of our tissues. They infiltrate all these tissues. And as they infiltrate different tissues in different environments, they're going to be exposed to many different physiological environments and metabolic environments. And so several years ago, we started to think a lot more about how the nutrient availability, the nutrient composition, might influence the way in which T cells adapt to their environments and how that affects their differentiation and their function. And so we know that all of these three essential processes are intertwined. What nutrients are available will also influence the metabolic states of those cells, as well as just regulation of different metabolic activities in cells. And then that will feed in to change how these cells undergo changes in their gene expression, their epigenetic modifications, and how that leads to changes in differentiation. And so this is kind of the question we wanted to talk about and explore more in understanding T cell differentiation. So we've been studying and profiling single-cell aspects of T cells for years because by flow cytometry, we can really start to piece apart all the different types of T cells. But single-cell RNA-seq has also been just a massive advance in really actually helping us determine the finite numbers of types of T cells that exist in our bodies. So we know now because of single— biased approaches of single-cell RNA-seq, that there's not thousands of different types of T cells, there's not even hundreds of different types of T cells, there's probably on the order of a couple dozen or so when you look at all the different reproducible types of T cell responses and subsets that have come from single-cell RNA-seq. And so we know that the T cell populations can be very heterogeneous. This is looking at TILs from some of the first single-cell RNA-seq data that was produced by Zemin Zheng's lab where he looked at the TILs from lung, colorectal, and liver cancer. And what's been really fun for us is because we've been studying T cells for many years and really teasing apart what types of T cells exist in mice during infections and in cancer, what's been really pleasingly satisfying is that when we look at these types of correlations and the conservation of the different types of T cells in humans and in human tumors, they're actually quite conserved. And so I'm not going to go into all of these different types of T cells right now, but I just want you to realize that they're quite heterogeneous. And we can actually really, by these genetic signatures, identify kind of the different types of T cells that exist. And so this has really started to help us really elucidate tissue microenvironments and tumor microenvironments. And so one thing that we have been wanting to know then is given this integration of nutrient availability and epigenetic modifications, and knowing that you can lead to this diverse array of different types of T cells that have different epigenetic and transcriptional states, is how is this connected? And so we started to focus in on one of the key metabolites, acetyl-CoA, because that's going to be the metabolite that's used for histone acetylation that will lead to these epigenetic modifications that will then lead to the underlying epigenetic changes that occur as T cells differentiate into these different types of T cells. And specifically, we wanted to focus on acetyl-CoA because it offers a really interesting way to look at how T cells integrate the nutrients that are available, because acetyl-CoA can be metabolized and generated downstream of many different types of nutrient sources like glucose, lipids, and acetate. And so all these different nutrients can feed into the acetyl-CoA pool that's then used to lead to the epigenetic modifications that would then translate into the differentiation of T cells and their different functional states. And so we kind of wanted to ask then, is there an epigenetic metabolic code? Is it simply that all of the T cells and cells in general just listen to the total available pool of acetyl-CoA, or are they listening a little bit more carefully to where the source of this nutrient is coming from? Are they able to decipher if they're getting their acetyl-CoA from acetate, for instance, versus glucose. And so that's the question we started to try to look at within the T cells. And we've been guided by a lot of the beautiful work of Katie, Shelley Berger, and many others in the field, because we know that downstream of these different nutrients, downstream of glucose or downstream of acetate, we know that the synthesis of acetyl-CoA is generated by two different enzymes, ACSS2 which is in the cytoplasm, as well as ACLY, which is in the cytoplasm. And the fundamental research has shown from these pioneers that these enzymes both contribute to the acetyl-CoA pool that can influence the histone acetylation that occurs within cells. And it's also been shown that it's not just the cytoplasmic functions of these enzymes that can regulate the acetyl-CoA pool in these cells, but these cells can also— these proteins can also migrate into the nucleus. And have very local effects as well within the acetyl-CoA pool that's generated in the nucleus. So again, this kind of brings back a really interesting question. Are the cells able to sense and decipher where their acetyl-CoA is coming from, and then how does that then translate to potentially local activities within the nucleus? So that was really kind of the simple question that we set out to ask is, is there kind of an epigenetic metabolic code? Are these cells sensitive to the different activities and the ways in which acetyl-CoA is synthesized through ACSS2 or ACLY? And we focused specifically on these two enzymes, although it's also been suggested that pyruvate decarboxylase can also be within the nucleus as well to generate acetyl-CoA. And so we looked at this kind of question to see how this could influence T-cell differentiation, function, fitness and longevity. And so now I'm going to try to explain to you a little bit about T cell differentiation. This is why people hate, hate immunology, because we have all these different subsets and, and all these different types of T cells. But what's really kind of cool— I hope this is like the one thing you can really remember— is that our T cells are adapting to many different types of environments, and they adapt to the different types of immune responses that they're, that they're involved in. So if there is an immune response that's short-lived, like a vaccine an acute infection, something that you quickly resolve like the common cold. What typically happens in these types of immune responses is that you'll have— start off with a naive T cell and it will start to get activated and it will differentiate into different types of effector cells. These effector cells are differentiating into cells that produce cytokines, cytotoxic molecules. In particular, we're talking about CD8 T cells that differentiate into cytotoxic killer T cells.. And then this pool of effector, activated effector cells, actually has kind of two fates. One fate is that these cells can develop into these functional effector cells, but they die. So it's a short-lived fate. But there is another fate that a small proportion of these cells undergo, and that is to differentiate into a memory precursor cell that has the potential to then give rise to a pool of memory cells. These memory cells are long-lived. They're also heterogeneous. They go and infiltrate all different parts of our body. They circulate throughout our blood. Some of them go and reside long-term in our tissues like our skin, our guts, our brains. Every tissue that we've looked at, we can find these tissue-resident memory T cells. And so this is what generates then long-term immunity. This is what gives us our immunological memory after we've had a vaccine or, say, an acute infection.

Infection.

So this is kind of the classic pathway. But we know that there is also a change then if the cells, the T cells encounter antigens that they cannot remove themselves from. So if there's a chronic infection, a chronic virus that continues to produce antigens that the T cells cannot evade, they cannot control, they cannot get rid of and eradicate that virus. Or in the case of cancer where we know that the T cells are experiencing tumor antigens for many years decades, what we start to see then in this chronic stimulation of T cells is that they start to differentiate down a different path. And so while the overall architecture is kind of similar, initially these cells differentiate into what's called a progenitor cell, and these cells start to upregulate inhibitory receptors that are referred— like PD-1 or CTLA-4 or LAG-3. Many different inhibitory receptors get upregulated. But these progenitor cells then have the capacity, because they have some stem-like plasticity, these cells can then also give rise to a pool of effector cells, or they can further differentiate into a terminally exhausted cell. And this is what we think about a lot in cancer, because the abundance of these exhausted cells has been very well characterized to be found in tumors, and these cells upregulate even more inhibitory receptors in addition to PD-1. They upregulate TIM-3. CD101. But the most important thing is that as these cells are differentiating to this terminal exhausted state, they're losing their effector functions, they're losing their ability to secrete cytokines, and they're becoming what we refer to exhausted. And we use that to convey they're getting pooped out, you know, they're not able to function as well. But it's important to remember that these cells also do exert some immune pressure. They are actually still somewhat cytotoxic, so so they can kill cells. They've just lost their pro-inflammatory functions that could cause long-term collateral damage and pathology if these cells were to remain hyperfunctional for long, long periods of time. So this process is actually, I think, an evolutionary beneficial process in some ways to allow the immune response to adapt to the chronic stimulation, curtail the more pathogenic pro-inflammatory states, but fine-tune its activity towards the activities that would still allow some immunopressure. And also these cells then desensitize themselves from the antigen in the environment by upregulating these inhibitory receptors. So they're adapting and they're adjusting to this state. And so this understanding, this process though, of how these T cells undergo these differentiation states from this more progenitor state to this more terminally exhausted state state has been a very important part of understanding our, our T-cell response to cancer, for instance. It's also quite notable that these progenitor cells are the cells that respond to immunotherapies like anti-PD-1 and checkpoint blockade. So the presence of these progenitor cells is actually really critical to the responses to immunotherapy. So, so we were very interested then in understanding this question of how cells are integrating their metabolic states as they undergo these various differentiation states in the context of chronic infection and cancer, leading to this more terminally exhausted state. And it's been shown if you compare the effector cells that form in acute infection versus these cells that form during chronic infection or in cancer, there's been already many different notable metabolic changes and features associated with the T cells in these different states. States. So after acute infection, we know that the memory cells have substantially high levels of OXPHOS. They have high mitochondrial potential. They can induce aerobic glycolysis. They can proliferate. If they're restimulated, they remain quite functional. But these cells, as they undergo this chronic stimulation and develop into these more exhausted cells, some very notable features are that they downregulate their ability to activate mTOR. They have over time reduced glycolytic activity, and probably the most notable feature is they have greatly diminished mitochondrial mass and respiratory capacity. So these are some of the most notable features of T cells as they become this more terminally exhausted state. And so we've also been looking at various nutrients that we're going to focus on mainly as acetate and glucose within the environment.. And so if you measure during a viral infection, and here we're comparing— sorry, I shouldn't have used these labels. I should have said acute and chronic. But if you look at two different types of viral infections, one's LCMV, which is a virus that we work with a lot in the lab that's a natural mouse pathogen. This virus comes in two flavors, which makes it a really nice system to work with. Because one strain of the virus can establish an acute infection, but the other strain of the virus is clone 13, can establish a chronic infection. So this gives us a really nice way to study the same population of T cells to these two different types of infections. And if you look at an acute LCMV infection versus a chronic LCMV infection, you can see that we actually find slightly lower levels of glucose in the circulation and in the spleens of these animals. But if you look at acetate, you see the opposite. So there's nutrient changes occurring in these animals under these two different types of infections. And if we look inside tumors, we actually see a similar thing. It's been reported that glucose glucose levels can be lower in tumors, and that's what we see in the mirroring models that we work with, and we also then see an increase in the availability of acetate. So these T cells are exposed to different types of nutrients in these different types of immune responses. So we wanted to then ask some very simple questions. Do the different types of T cells have different modes of nutrient utilization? Do they have different preferences for different types of nutrients as they differentiate into these different states? How does this affect the epigenetic imprinting and the differentiation of the functional states? And for a therapeutic effect, can we start to reprogram these cells through rejuvenating the changes in which they utilize nutrients to generate and regulate their epigenetic status? Can we start to reprogram their functions through this type of knowledge? So we've been studying how these CD8 T cells differentiate, as I've been going into, probably too much detail, but we also have been learning a lot about their epigenetic changes as well. And this is just showing you there's been a lot of work on all the different types of modifications that are changing on the epigenomes of these cells. And this is just showing you a really nice example from Pam Schwarzenberg and Toichi Wu's lab where they looked at the histone acetylation of H3K27, the differential modified regions, found in some of these different types of T cells. So the important thing to look at on this graph here is that these first two are the ones that are found in acute infection. These latter two are the ones that are found in chronic infection. These are the terminally exhausted, and these are the progenitor cells I've been talking about, the ones that respond to PD-1 and the ones that become less functional and are probably the most— the least effective in fighting cancer. So if you look at then the changes in their epigenetic states, You can see that there are clearly some DMRs that are unique to the chronic setting, so they're becoming epigenetically altered. There are some that are in common between— in the T cells in both settings, but here are some of the ones that open up or become acetylated in the chronic setting. And then here are ones that are unique to the progenitor cells, so they do have unique acetylation of different loci in the progenitor cells., and these are ones that are unique for the terminally exhausted cells. So how are these patterns getting established? How are they being differentially acetylated at these different loci? And that's really what raised that simple question, which is, is it changes in the acetyl-CoA pool, or are there more specific activities locally within the nucleus by which ACSS2 and ACLY are generating acetyl-CoA leading to very specific locus-controlled gene expression and changes in histone acetylation that will regulate and modify the different types of genes that these T cells express that associate with their differential differentiation states. So again, how does ACSS2 and ACOY regulate the total pool of acetyl-CoA, or how do they operate in this locus-specific manner? So we started to get really excited about this because we came up with this idea to to probe this question: do T cells listen to the types of nutrients that they metabolize or are exposed to, and does this regulate the epigenetic code of the cells in a way that dictates their differentiation? But we actually came into something that was even unexpected when we looked at this, and that was by just looking at the expression of ACSS2 and ACLY as these cells undergo this differentiation state. Toward— between like in a functional effector cell or towards this more exhausted state. And so what we looked at by single-cell RNA-seq of either TILs or looking at the antiviral CD8 T cells is we saw a very similar pattern in that ACSS2 mRNA levels was greatly diminished in these cells as they became these more terminally exhausted dysfunctional cells. So the effector cells and the progenitor cells, these more functional cells have ACSS2, but this becomes diminished as they become exhausted. ACLY didn't seem to be changed too much. If anything, it actually went up a little bit in this more terminally exhausted cells. And so again, we saw this also when we looked at these antiviral cells as well. It's down in the antiviral cells, it's down in both of these cells that are found in the chronic infection, but it's even substantially lower in these more terminally exhausted cells. And then here, the protein, when we looked at this by Western blot, that was very clear. If you compare the TILs found in a tumor compared to the T cells found in the spleen, you saw that there's really diminished protein levels of ACSS2. Similarly, when we look at these two different infections of LCMV, looking at the acute infection or in the chronic infection at 8 and 21 days after infection, you can see that there also is a very substantial drop in ACSS2 levels.. And this is just looking at human TIL data as well. You can see that there is— if you look at ACSS2 on top, you can see that there are lower levels of ACSS2 in human TILs within these exhausted cells compared to the effector cells or compared to naive T cells. And for ACLY, we don't really see too much of a change. And again, if anything, there might be even just a little bit of a dip up in expression of ACLY in these more terminally exhausted cells. Infected cells. So very surprisingly, CD8 T cells are downregulating ACSS2 as they're becoming exhausted. So we wanted to know if this affected how they metabolize acetate. So we teamed up with Rusty, and he's been helping us do a lot of different types of analyses and metabolic flux analysis looking at these T cells. And so what we did is we labeled the T cells. We took them either out of infected animals from the acute or the chronic infection. Or we've done some experiments in vitro with cells that we can kind of push towards a more exhausted state, but I'm not going to show you that data today, but just show you some of our data isolating the cells directly ex vivo and looking at some of their abilities to metabolize acetate and glucose using flux analysis. And so one thing you can see is if you take these T cells from the acute infection or from the chronic infection, so these are more functional, these are less functional. Functional, and you feed them C-13-labeled acetate, you can see that the production of acetyl-CoA is reduced in the exhausted cells. So this decline in ACSS2— we know that they still have ACSS1. We haven't really seen too much of a change in ACSS1, but there is the reduction in their ability to generate acetyl-CoA from acetate in these more exhausted cells. And if anything, we're seeing that when we expose these cells to glucose, you can see that the exhausted cells are more dependent on glucose for generating acetyl-CoA. So there's kind of a switch then in the way that these cells can use acetate and glucose depending on if they're a more functional effector cell versus an exhausted effector cell. It became a little bit more noticeable when we looked at the synthesis of UDP-glcNAc, which is occurring exclusively in the cytoplasm. So this was a good marker for the cytoplasmic plasma pool of acetyl-CoA. And here you can see even a greater reduction in the ability of these exhausted cells to use acetate and make UDP-glcNAc from acetate. And then again, we see a little bit of a hint of an increase in labeling from glucose in the exhausted cells. So these effector and exhausted cells seem to have kind of switched the way in which they're preferentially utilizing acetate or glucose. So then we started to try to drill in a little bit closer in on the actual histones. So if they have different abilities to metabolize acetyl-CoA from acetate in the case of the exhausted cells because it's downregulated, how does this then influence their ability to acetylate actual histones? And so the way we did this is we took the T cells again from an acute or a chronic infection, so we're taking the more functional effector cells versus these more exhausted cells, and then we labeled them as as well with labeled glucose and acetate, and then we did mass— we isolated the histones and we did mass spec after about 4 hours of labeling. And what you can see here is, again, if you compare these effector cells in blue versus the exhausted cells in red, you can see that when the effector cells are— when they're exposed to acetate, you can see that the effector cells are able to incorporate acetyl-CoA and acetylate the histones with greater with greater ability than the exhausted cells. So they are able to funnel this acetyl-CoA downstream of acetate into the histones in these more functional effector cells compared to these exhausted cells. So again correlating with this decline in ACSS2 in these exhausted cells. And conversely we saw the opposite again with the glucose labeling where the exhausted cells are now funneling more of the acetyl-CoA downstream of glucose glucose into the histone acetyl pool. What was interesting, though, is that we didn't see a unique pattern of a specific histone acetylation site if it came from glucose or if it came from acetate. And that was a question we had. We wondered maybe there's going to be even regulation of the types of acetylated histones on the T cells by the different types of nutrients, but that's not necessarily what we saw. It seemed like for the most abundant histone acetylations that we could measure, they were being incorporated relatively equally between the two types of nutrients. It's just that we saw a difference in what the T cells preferred if they were more functional or if they were more exhausted. So then we started to ask, well, how does this affect it if you knock out ACLY and ACSS2 in these T cells? And so we were able to generate ACLY knockout T cells either using CRISPR-Cas9 activation, or in the case we also have genetic models, but what we have done here is we then again did the radiolabeling with acid— radiolabeled acetate, and we could see that if the T cells were deficient in ACSS2, both for histone H4 or for histone H3, we could see that the incorporation of acetyl-CoA from acetate was obviously greatly diminished when ACSS2 was lacking, so this showed us ACSS2 was a a necessary and important enzyme for allowing the histones to be acetylated in these T cells. But then if we knocked out ACLY, we saw the converse. We actually saw increased ability of these cells to incorporate acetyl-CoA from acetate onto their histones. And that's because a lot of the work that Katie and others have shown is that there actually tends to be compensation between these two enzymes, and so in the absence of ACLY, you can see that there's an augmentation of ACSS2 in the T cells, and if you look by a Western blot, that's exactly what we see as well. So getting rid of ACLY, these CD8 T cells are actually compensating and increasing ACSS2, and that's leading to an increase in their ability to phosphorylate the histones downstream of acetate. So what was interesting then is to just say, okay, well, what happens to the total pool of histones? We see that they have these changes— these preferences if they're exhausted or more functional effector cells, and which nutrients they can use to acetylate the histones. But let's just look at the total landscape of histone acetylation. What happens in these cells? And if you compare just the total T cells, but now you parse them apart into these two different subsets of T cells, these progenitor cells or these more terminally exhausted cells that become less functional, you can actually see that the total level of histone acetylation— we're using flow cytometry here to measure this— of at least H3K27 acetylation is reduced in these terminal exhausted cells. So the overall levels of histone acetylation is lower in these terminally exhausted cells. But now if you ask what happens if you get rid of ACSS2 or ACLY, you can see that they're getting rid of ACSS2 in the progenitor cells has a big effect on these ability of these cells to maintain their overall levels of histone acetylation. So you can see that the progenitor cells are quite dependent on ACSS2 for the histone acetylation that's found in these cells, whereas the terminal exhausted cells don't really care about ACSS2, and that's in part because ACSS2 is getting downregulated, if you remember, as these cells are becoming these more terminally exhausted cells. But then if you get rid of ACLY, you find that the progenitor cells don't really care about ACLY. ACLY too much to maintain their pools of acetylated histones, but if you get rid of it in the terminal exhausted cells, that's really all they have now, you can see that if you get rid of ACLY and you get this compensatory effect on ACSS2, we're actually getting an increase in histone acetylation in these terminal exhausted cells now that they lost ACLY. So there's different effects of these two different enzymes on the subsets of of cells and how they maintain their total pool of acetyl-CoA, where these progenitor cells depend more on ACSS2, and these terminally exhausted cells appear to not depend on ACLY, but if you get rid of it, you actually see this effect in which it's augmented acetylation by that compensatory effect of ACSS2. But now I just want to confuse you for a second. So you would think then, if the total acetylation levels are going down in these cells as they become more terminally exhausted and they're losing ACSS2 expression, you would think that these cells would have a lower acetyl-CoA pool. But actually our studies with Rusty showed that that's not the case. If you look at these exhausted cells, they actually have more total pools of acetyl-CoA. So they're not able to synthesize it as much from ACSS2, but they're actually probably not using as much acetyl-CoA. Acetyl-CoA as well. And looking at other types of processes that use acetyl-CoA like fatty acid synthesis, we find that that's the case. These exhausted cells have low levels of fatty acid synthesis too. So they're actually accumulating acetyl-CoA. They're not really utilizing it very effectively, and they certainly aren't able to utilize it very well through ACSS2. And— but yet they have this total reduction in histone acetylation, but actually have an increase in the pool of acetyl-CoA. So these are some of the major summaries I just went over. Relative to the functional effector cells, ACSS2 mRNA and protein are decreased in these exhausted cells. ACOY expression is maintained or possibly even elevated. These exhausted cells don't metabolize acetate as well as the effector cells. Seems like they prefer prefer and depend on a little bit more on glucose as a comparison between the two. Relatively, the exhausted cells prefer to utilize glucose for histone acetylation, whereas the effector cells prefer to utilize acetate. And these terminally exhausted cells have lower amounts of global histone acetylation relative to the effector cells and these progenitor cells, despite having a greater total abundance of acetyl-CoA pools. So we wanted to dive in a little bit into how these are affecting the differentiation of the T cells. And so we knocked out ACSS2 and ACLY using CRISPR-Cas9 in T cells that can recognize tumor antigens. And when we did this, we saw a very interesting phenotype. So I talked about the exhausted states that these T cells can go into, and this is I think the only fax plot I'm showing in the whole talk, but you can identify these terminally exhausted T cells up here in this upper upper right quadrant. These are the cells that express high levels of TIM-3, high levels of PD-1. They have more of these inhibitory receptors, are less functional. But when you get rid of ACLY, we actually see that there's fewer of these terminally exhausted cells forming. And if we get rid of ACSS2, we actually see that there's an increase in these terminally exhausted cells. Now, this latter result may not be too surprising because ACSS2 is actually declining in expression as these cells became more exhausted. Suggesting it's, if anything, the decline is driving a reduction— is driving T-cell exhaustion. But getting rid of ACSS2 genetically, we actually see that we can push these cells even farther into this more terminally exhausted dysfunctional state. But the knockout of ACLY was actually quite interesting because here we're seeing that these cells are no longer able to develop into these more exhausted cells. And in converse, When we look at markers for these progenitor cells, these are the cells that are more functionally responsive to anti-PD-1 checkpoint blockade, you can see that getting rid of ACLY actually boosts this progenitor population, and conversely, getting rid of ACSS2, we lose this population very profoundly. So we're gaining more exhausted cells when we get rid of ACSS2, in line with this expression pattern, but getting rid of ACLY was actually quite interesting. Because that actually prevented terminal exhaustion from happening and helped to maintain this more functional progenitor state in these T cells. And so if we then looked and put this into mice that had tumors and looked more carefully at how these T cells could control tumors, we found that getting rid of ACSS2, the tumors actually grew faster compared to the controls, and that's in great correlation with the phenotypes we see in that these cells are becoming less functional. They're becoming more exhausted, but if we get rid of ACOY where we're preventing this exhaustion from happening, we're actually seeing that these cells now have better anti-control of tumors. And so how does this then affect the histone acetylation in these CD8 T cells? We've been talking a lot about how they can regulate the different types of genes that lead to these different differentiation states. And so what we found, and then our prediction would be, based on the data I just showed you is that ACSS2 is actually important for sustaining the genes involved in the differentiation of these progenitor cells, so they'd be driving progenitor signature genes, and the ACLY would then be involved in driving the expression of these terminally exhausted genes. And so we started to look at histone acetylation patterns when we knocked out ACLY. Again, this is going to— the deficiency of ACLY ACLY is going to induce these cells to adopt more progenitor-like states. And what we found was that if you look at several loci that get upregulated or have increased acetylation in the absence of ACLY, you can see that there are several loci that gain acetylation in the absence of ACLY, and these had a lot of genes that were known to be genes that regulate the progenitor state of of these CD8 T cells. We also saw several AP1 transcription factors as well. Whereas you look at the ones that were actually, if anything, were downregulated, these were genes that were associated and signature genes of these terminally exhausted cells. So we're really seeing that getting rid of ACLY is changing which loci are getting acetylated, and across multiple different acetylation types, H3K27, H3K9, and H4, and we're seeing that the reduction of ACLY is leading to the maintenance or increase in abundance of acetylation on these genes that are associated with progenitor fates of T cells and a reduction in the genes associated with terminal exhausted. So the last thing I want to talk a little bit about is, well, we would like to know then more specifically about how this locus-specific control is happening between ACLY and ACSS2 in the T cells. So we'd like to try to find the binding sites for these in the genome. In the genome. We haven't really made that much headway on that, but we've also been interested then in how they might lead to the local deposition and production of acetyl-CoA, and which histone acetyltransferases would be then operating and sensitive to the local activities of ACSS2 or ACLY on these loci, leading to the differential gene expression of progenitor genes or of these terminally exhausted genes. So which HATs and transcription factors might these AclY and AcSS2 be operating with in T cells? So she had a really creative idea. She decided to just knock out a lot of different histone acetyltransferases and look to see which ones had a phenotype very similar to AclY or to AcSS2. And I'm only going to show you two of them. They're actually— most of these did not have much of a phenotype, but two of them did have striking phenotypes that correlated correlated with ACSS2 and ACLY expression, and that was P300 and Cat2a. And so we started to look at these more carefully in the nucleus using immunofluorescent microscopy. And if we stain for P300 and stain for ACSS2, or stain for Cat2a and stain for ACLY, what we're finding is that there is preferential colocalization of P300 and ACSS2. You can see these different puncta in the nuclei of the T cells. And if you look at the merge of the two, you can see there are some that are distinct, but there is an overlap of about 10 to 15% of the puncta in these cells are over— are colocalizing. And similarly, when we did Cat2a and AclY, we also found about 10 to 20% of these puncta are colocalizing. And by doing chromatin immunoprecipitation, we're also finding that P300 and at least ACSS2, we can see co-immunoprecipitation between these ACSS2, we haven't had a chance to get a good co-immunoprecipitation between ACLY yet. But moving forward, we then started to ask, well, what happens if we— now, knowing that ACSS2 is coming— is declining in these cells as they become exhausted, can we correct this? Can we reprogram these cells and rescue these cells from this terminal differentiation, this terminal exhaustion that's happening? By now forcing them to express ACSS2 and forcing them to express ACSS2 in the nucleus. And so Shuxian made a nuclear-localized ACSS2, and you can see here that it's preferentially in the nucleus, as we would hope when you have the nuclear-localized tag compared to the wild-type overexpression of ACSS2. And if we looked in the nucleus, we also saw very strong correlation— we saw much bigger puncta of ACSS2 in the nucleus. We're not quite sure what this represents, if these are condensates. If anyone works on condensates, I'm afraid to even use that word, but we do see this nice colocalization of P300 and ACSS2 in these nuclei now that we're overexpressing ACSS2 in the nucleus. And similarly, we can see a very strong effect on boosting the overall levels of histone acetylation in these cells that overexpress ACSS2. So what does it do to their function and to their epigenetic states? Well, if we overexpress ACSS2, we can see that there is a gain of histone acetylation on several signature genes that are associated with progenitor states, and we're seeing a loss of histone acetylation of several of the genes associated with terminally exhausted states. So you can see here when you do gene set enrichment, you can see that the genes that are expressed in these these T cells that now overexpress this nuclear form of ACSS2 more predominantly, that these cells are being induced to differentiate into this more progenitor-like state with this progenitor-like gene expression profile. And so is this functional to the T cells? Again, we put them into animals that had tumors and asked, what does this do to their anti-tumor activity? And you can see that the tumor volume, the tumor growth, was slower in the animals that received T cells that had these nuclear-localized ACSS2. So perhaps this could be a way to augment T cell for adoptive cell therapy, which is being done a lot right now in cancer. And if we looked at then their differentiation states, you could see that we were able to boost more of these progenitor-like states and reduce the terminally exhausted states that formed when we overexpressed ACSS2, and that was associated with greater effector function as well. And so with that, I'm just going to end with the final summary, and I'll show the final model. So ACSS2 nuclear localization was sufficient to induce specific histone acetylation at pro-signature genes, and it could boost the CD8 T cell antitumor functions and differentiations and suppress antitumor immunity, but surprisingly— and I didn't show you this— ACSS2 deficiency deficiency suppresses anti-tumor immunity. I showed you this at the very beginning, but surprisingly, ACLY deficiency boosts it. And ACLY suppresses the formation of these progenitor cells and promotes terminal exhaustion. So we've learned a lot then about trying to understand the dissection of these T cells and how they utilize and may prefer different nutrients as they differentiate into different types of T cells along these different states of T cells. And we're finding then that ACSST CD8+2 is playing an important role working in conjunction with P300 through the regulation of a lot of different loci in the CD8 T cells, and we presume this is also in cooperation with key transcription— AP1 transcription factors to regulate the expression of the effector genes that maintain an effector state and progenitor-like state that's supportive of an anti-tumor response where the maintenance of a T cell population in chronic infection and response to checkpoint blockade, whereas the cells as they differentiate into these more terminally exhausted cells, they're losing ACSS2 expression and they're maintained by this ACLY activity, but this is actually not just regulating gene expression in general, but is driving the expression selectively of these terminally exhausted genes. So this raises a lot of questions still. Why are these cells losing ACSS2 expression? We don't know why that's the case. Rusty had a really nice paper with Chris Hess where they actually could show that acetate itself was able to suppress the expression of ACSS2. As you might remember from the very beginning, we're seeing increased levels of acetate in tumors and in chronic infections, so perhaps this increased exposure to acetate is somehow leading to this loss of ACSS2. And it's also really interesting because if these cells become dependent as they're terminally exhausted on AclY activity, we know that the mitochondrial function is going to be declining in these cells as they become exhausted. So how is this then interplaying and affecting the activity of AclY is, I think, a really interesting question. And could it be that part of the reason why these exhausted cells eventually completely poop out and die is because once they start to now have impaired mitochondrial function, they're no longer even able to generate the citrate that's available— that's needed to sustain the overall genomic stability of these cells. And I think there's some other really interesting things to think about, you know, some of the work that's been done by Lydia Finley and the incomplete TCA cycle. Rusty and I have been talking about this, how this might also perhaps be related to her model of the incomplete TCA cycle supporting more of a differentiation state of a cell as opposed to maintaining stemness. And so maybe there's something like that going on here, but we've got a lot more to do. So I just want to thank everybody. Thank you for your time. Thanks for bearing a half hour of T cell biology. I really want to thank Shi Shen. She's been doing almost all of this work by herself. As I already mentioned, we've been doing a lot of this with Rusty and Michael Dahabieh in his lab, who's been just a really tremendous asset to work with. We've also been helped by a lot of people in our lab, as well as other people at Salk. And so thank you.

Hey, we'll take a couple of questions.

Thanks for a great talk. Yes, thank you so much. Very nice talk. I have a question that I don't know, probably you, you can— this, but I missed it. So the cytoplasmic nuclear shift of this enzyme seems to be key for sulfate decision. Is this actively exploited during differentiation? So is there some signal that tells these enzymes to go to the nucleus?

Katie might know more about that than I do, but the phosphorylation for ACSS2, I don't know if we know of a modification that really— but Katie has looked at phosphorylations of ACLY regulating its nuclear entry, so.

Yeah. So actually, the phosphorylation is happening in the nucleus in response to DNA damage. It doesn't actually regulate its localization, but yeah, so those are definitely active areas of research, sorry.

But since it was— Yeah, just a question.

Yeah, yeah, yeah, the experts here.

Yes, let's see, other questions? Yeah. Yeah, very, very nice study. So this most focused on the acetylation of the histones. The protein can be acetylated as well. So how does SCA1 and SCA2 affect protein acetylation? Yeah, so we haven't looked— broadly, that's actually one thing we're going to do next is look at the whole acetylene of the cells and try to look more broadly at this. We had a very focused study on the histones because of the roles that they play in the differentiation to connect with, but that would be an important next step.

Thank you. Beautiful work. Actually, based on this slide, I always think about how the— who make decision for the— to the specific locus? Have acetylase or maybe the ACS or the ACLY? They have like randomly have managed or just have specifically pick up the pattern?

[Speaker:JENNIFER_TCHAIKOVSKY] Yeah, yeah. I mean, we're not— we tried to do ChIP-seq for ACSS2 actually, and we failed to get really distinct peaks. So we can IP it from the chromatin with P300, but we— We can't really get good peaks from it, so that could be because, you know, we're just not very good at it. It could be because it's not really tightly associated with the chromatin. That's kind of my main thought. But what we suspect, and we can start to dissect more, is this going to be transcription factors are going to be guiding the complexes together at these sites to regulate the local histone deposition at certain regions of the genes. Okay, thank you. Yeah.

Okay, thank you. I think in the interest of time, we'll move along, but thank you so much for a fantastic start to the meeting.