Scientist Stories: Ping Chih Ho, Immunometabolic Editing & Tumor Metabolic Evolution

Hi everyone, welcome. We're fortunate to have as our speaker today Dr. Ping-Chih Ho from the University of Lausanne. Ping-Chih got his start in science at the National Taiwan University. He moved to the University of Minnesota for his doctoral studies and then on to Yale University where he was a postdoc with Sue Ketchum. Since then he's been at Lausanne where he is currently in and he's associate professor at the university and an associate member of the Ludwig Institute for Cancer Research. He's won numerous prizes and awards along the way. I'll mention particularly that he's an EMBO Young Investigator, and he also recently received a CRI Lloyd J. Old Star Award. He's made a mark in tumor metabolism, not just the tumor cells themselves, but the crosstalk between tumor cells, CD8 T cells, Tregs, tumor-associated macrophages, and the metabolic signals. And he's been particularly innovative in finding ways to tweak cell metabolism that can then be applied in combination with anti-PD-L1 therapy, for example, to enhance control of tumors. So I think he'll tell you probably much more about this now. Why don't you start, PC?

So I just removed my mask. Okay, so first of all, I would like to thank Patrick's invitation, especially we tried to arrange this seminar like 2 years ago right before pandemics. Yeah, so it's a little bit interesting part, especially I still remember on the day I decided, okay, everything should be closed, and I received the email from Sue He said, see you next week in San Diego. I said, no. So that's a funny process. And actually, it's my pleasure to have the chance to come back again after almost— Is the microphone okay? What? You're a little quiet back here. Is the microphone okay? Yeah, microphone is okay. I don't know. For me, it's fine. Can you hear me in the back? Okay, so I will speak a little bit louder. Yeah, so I think this is a very special experience again to have the chance to come come back in after 2 years. And it's also a little bit difficult for me because I decided to change the topic. I should not present what I want to present 2 years ago. So that's why I decided not to present these exhaustion studies we already published almost 2 years ago. So I will switch to a new direction which kind of stick on the same thing what Patrick just mentioned, the crosstalk between tumor cells and immune cells. And I would like to convince you that there's a new type of immunoadapting which is very important are important to drive the cancer evolution and also why the cancer cell will be able to adapt particular metabolic features they display in vivo setting. Okay, so before I go into the detail, there are two things I want to tell you. First of all, you will only see one slide with facts data, so please don't be disappointed about that. And second thing, it will be more like a signaling cascade because I consider myself like biochemistry. Trained scientists rather than cellular immunologists. So let's see what I can do. Okay, so before I go into the detail, I want to introduce the concept of immunoadaptation. So this is basically the study did by Bob Streiber almost 20 years ago showing that actually cancer cells, if they have the chance to grow in the presence of immune systems, especially T cell response, they will have the ability to know how they can evade from immunosurveillance. So this was the experiment his group did. So they used the MCA to induce sarcoma formation in immunocompromised or immunodeficient Rag2 knockout mice and established sarcoma cells. As you can see in this slide here, this is the sarcoma cells growing in the presence of T cell and B cell. And the right-hand side are those sarcoma cells they established growing in the absence of T cells and B cells. And what they did was very simple. They engraft those cancer cell lines into the mice again. So you see that, you pay attention to the left-hand side. If they engraft those cancer cells growing in the presence of T cells and B cells back to Rag2 knockout or immunocompromised, you can see those tumors grow normally without any trouble. But the important part is on the right-hand side. If those tumor cells, they're growing in the absence of T cells and B cells, They can only form tumor 100% in the recipient which are lectin knockout. If those mice, they have immune system, you will see more than 50% of those tumors shrink after 1 week. So it suggests that the presence of the immune system during tumor progression tell the cancer cells how to control immune system. So when they encounter immune system again, they will have different ability to either shut down immune response or prevent T cell infiltration. So based upon this finding, people also have a following study and a study proposed that there are 3 E models during tumor progression. The first step, people believe that this is the elimination stage. In this stage, transformed cells will be able to tell our immune system they have certain trouble, they need to be recognized to be destroyed. In this case, I will pay attention to the T cell part. So in this case, T cell will be able to recognize those transformed cells based upon the neoantigen and then start to do the cytolytic activity or secrete interferon gamma. However, some of the cancer cells, they will be able to enter to the second stage called equilibrium stage. In this stage, those cancer cells, they have additional growth advantage. They can partially shut down immune response and drive PD-1 expression in those cells. But this is something people speculate with, we don't really know whether we will be able to catch this point in mouse model and human, but this is what people believe. In this case, then the immune system can get along with tumor cells. They can stay together. And in this case, immune system also partially restrict the growth, the outgrowth of cancer cells. And this can last probably decades, even a whole life in human. But most of patient we see in the clinics, they probably already enter to the last stage called escape stage. Those cancer cells acquire additional growth advantages such as antigen loss, MHC class I loss, or beta-2M loss. In this case, they become invisible to our immune system, and they also have the chance to coax the formation of regulatory T cells, MDSC, or M2-like macrophage to shut down the immune response in the tumor microenvironment, and the T cells become fully dysfunctional or exhausted. So basically, you can consider that the entire immune response, immune status in a tumor microenvironment is very high for the immunosupportive condition in the beginning, but this process declines. In contrast, the immunosuppressive features start to increase in the end of the tumor progression. So this switch indicates that the cancer cells will be able to learn how to deal with immune system, acquire multiple growth advantages over here. So this also lays out a foundation for the new new version of Hallmark of Cancer. In this new version of Hallmark of Cancer published in 2011, Doug and also Bob Weinberg, they proposed that there are 4 new features. And I will let you pay attention to the top 2. I want to make you believe they can be intercrossed. So one is that actually the cancer cells will have the deregulated cellular metabolic process. They will increase aerobic glycolysis activity, they will start to engage better metabolic flexibility to maintain their survival and type of apoptosis, et cetera. And on the right-hand side, those cancer cells also believed to acquire the ability to avoid immune destruction. But no one really know that's whether these two things can have the crosstalk, they can go hand in hand during tumor progression. So based on the study of Popchou with Su during my postdoc training, at the same time Eric Capillas also have a similar similar study revealed a similar concept. We revealed that actually the cancer cells will be able to use their metabolic activity to create an immunosuppressive tumor microenvironment. What they did was very simple. We know that cancer cells, they always upregulate certain metabolic features such as aerobic glycolysis. They can convert glucose into lactate. Unfortunately, T cells also use the same metabolic activity to to support their expansion and cytokine production. So based upon this concept, we demonstrate that there's a possibility the tumor microenvironment, based upon the high aerobic glycolysis activity of cancer cells, they will be able to deprive the nutrient in the surrounding environment. So when the T cells which can recognize those neoantigens come in, they will not be able to get proper nutrient they need for expansion and interferon gamma production. As an outcome of this entire crosstalk, tumor cells with higher aerobic glycolysis activity can shut down the interferon gamma-producing ability of T cells. So this raised a few questions based upon this after I published this work. I started to speculate. We know that cancer cells always upregulate particular metabolic activity in vivo. That's why we can use PET scan to see see those tumor cells. But we also know that cancer cells can display robust metabolic flexibility in vitro if we remove particular nutrients. They can, for example, if we remove glucose, they can survive with glutamine, with fatty acid. So this doesn't make any sense for the in vivo study why they tend to use a particular direction, metabolic direction to support expansion. So I start to speculate maybe something else happening in the the body would tell those cancer cells you have to engage a particular metabolic process. So I start to speculate if this is true, and based upon the concept of immunoadaptation, the presence of immune system, especially T cells, may provide a selection pressure to select those cancer cells with particular metabolic activity become the dominant clone. That's why they have the chance to become detected by PET scan, and that's why they have the chance to evade from immunosurveillance. Rather than changing their immunogenicity, they can change the metabolic process. So in order to understand whether this is the case, I initiated those experiments during my postdoc training, and I think they were very simple, we can really do it very quickly, so this is the only project I brought with me when I was a postdoc, and once I established my independent lab, but turned out it took us for 7 years. So it took many people to go through this process, but these 2 people in my lab, they are the main person to drive this entire project. One is my postdoc, and right now he is assistant professor in Taiwan, and the other one is my student who will graduate very soon. So in order to do that, we first cross the BRAF PTEN melanoma system we use into REC1 knockout background. So this is a genetically engineered melanoma system. We can use tamoxifen induction to introduce BRAF mutation and PKA deletion. So we have those two mouse strains, so we induce tumor and then we generate primary cancer cells from those tumor with intact immune system versus those mice, they don't have T cell and B cells. So we consider this is a non-edited cancer cells. So once we generate those cancer cell line, primary cancer cell line, we first check whether we can see similar things as Bob Schreiber reported 20 years ago. So those cancer cell lines generating in the immunocompromised, you can see here they grow normally in both wild-type or RAC1 knockout. However, those cancer cells growing in the absence of T cells and B cells defined as knockout cells, they only have the chance to grow in the RAC1 knockout recipient. So it suggests that actually we can use the same approach approach to generate edited versus non-edited cancer cells. And then we move to the next one. We try to understand whether those cancer cells display different metabolic activity. As you can see here, we first check the ATP level. Each dot here is one primary cell line we generate. They have very similar ATP level. However, when we use Seahorse analysis to see whether the cancer cells prefer to use oxygen consumption rate or lactate production, which indicates anaerobic glycolysis. We can see that those non-edited cancer cells, they have higher OCA to ECA ratio, which indicates they tend to use oxidative metabolism to support their ATP production. So it reflects that our hypothesis might be right. Cancer cells growing in the absence of adaptive immune response, they still can grow, but they will change their metabolic preference. They will use oxidative metabolism.

If you look at the growth in the Rag knockouts, you know, it looks like they grow more slowly generally.

Yes.

Is that true or does that statistically hold up?

I will show you the other data later because this is engraftment so still have some issues. Okay. Yeah. So this result suggests that non-ADK cancers will change their metabolic preference. They tend to use oxidative metabolism which is a slow process. Proliferating metabolic process. So back to the question, so this is the cell line we generate. Doesn't mean too much things because you have an issue that you culture them for a while so they probably change their property in vitro. So we decided to take a look whether we can see those metabolic profile change by collecting the primary tumor immediately. So we induce tumor formation and we wait for 5 weeks and then we get the tumor and perform metabolomic analysis and see what kind of metabolic profile we will see. So we performed a PCA analysis as you can see here. So this one is wild type and the red one is knockout. They have very different PCA analysis which indicates their metabolic profile for all the single metabolites can be different. But we have one concern, this is melanoma. So we might see something which is very specific for this genetic mouse model we use. So we get another two mouse model from our collaborator, Kellen DeVisser at NKI. So one is HER2 overexpressing breast cancer model, and the other one is P53 knockout and BRCA1 knockout breast cancer model. So we have both immune competent and immune deficient version. We can still see the same thing. Those tumor forming in the absence of adaptive immune response, they have different metabolic profile. So when we pay attention to those melanoma we'd like to use in the following study, we did the similar things like gene set enrichment analysis, but this time we did metabolite enrichment analysis. And you can see here— oh, sorry— those tumor growing in the absence of T cells and B cells, they enrich metabolite in TCA cycle, downregulate metabolite in glycolysis pathway, which indicates what we observe in the seahorse analysis is holding true. Those tumors growing in the absence of T cells and B cells, they tend to use oxidative metabolism and decrease their dependence to aerobic glycolysis. So back to the point, this is genetic knockout background, Rett1 knockout background. So we induce tumor formation. Those differences we observe can result from certain unknown or minor genetic difference between immune-competent Y-type mice versus REC1 knockout. So we decided to move the bar a little bit higher to use the same genetic background and then see whether we can still see the same difference. So in order to do that, we took advantage of antibody-based T-cell depletion because we think T-cells should be the major selection pressure. So we used the immunocompromised and we induced tumor formation by tamoxifen treatment. In this system we can see the tumor around, palpable tumor around 3 weeks post-tarmaciclin treatment. So 1 week before we can see the tumor, we start to inject anti-CD4 and anti-CD8 antibody to deplete T cells every 2 days for 3 weeks. As you can see here, those tumor without any T cells, they have very similar tumor burden compared with the control. So it suggests that without T cells, in the genetic knockdown background, in the genetic-induced melanoma background, they have a similar tumor growth rate compared with the immunocompromised. However, the lactate level drops significantly. We see more than 40% reduction of lactate in tumors. Okay? And we have one concern. As I told you, T cell can also use glucose and then generate lactate. So this difference actually can be a result of T cells disappear, T cells fail to produce lactate, and this difference is the lactate produced by T cells. So in order to examine this possibility, it's another case. We perform the same depletion, but this time we start the depletion later. So we let the tumor form for 5 weeks and then start the same duration of depletion for 3 weeks. In this case, we call it late depletion. If there's any editing process can be engaged, it can only happen in the early phase. The late phase, probably the editing is not that more important. In this case, we don't see any difference of tumor burden. We also didn't see any difference in terms of lactate production, which implies one thing: the presence of T cells in the early phase of tumorigenesis can determine whether those tumors have the chance to produce lactate. And we also performed the entire metabolomic analysis to see the global metabolic profile change rather than lactate only. And in this case, similar to the REC1 knockout background, we can see the distinct metabolic profile based upon the PCA analysis. And this difference cannot be seen if we use the LATE depletion. Most importantly, by using the same metabolite enrichment analysis, we see same pattern. Those tumor growing in the T cell, early T cell depleted mice, they upregulate TCA cycle metabolite, downregulate glycolysis metabolite. So which indicates that the presence of T cells in the early phase of tumorigenesis play a critical role to control what kind of metabolic features those tumor cells should engage, should display. And you may start to wonder, there are a few things in this entire case. First of all, it looks like the absence of T cells doesn't affect the tumor growth. Second thing is that those metabolic difference can result from vasculature change, hypoxia change, or HIF-1 alpha expression level change. And I will get back to the first point later, but I will try to tell you that HIF-1 alpha hypoxia doesn't matter in this case. So in order to check whether T cell depletion in the early phase of tumor progression can control metabolic activity of cancer cells due to hypoxia change or due to HIF-1α expression level change, we use pimonidazole staining to see whether the hypoxic severity can be altered when we wipe out T cells. As you can see here, no matter we wipe out T cells with antibody in the early or late depletion setting, there is no any difference of pimonidazole staining, which indicates hypoxia severity is identical. Moreover, in HIF-1 alpha expression pattern, we use Western blot, we also confirm that the T-cell depletion in the early phase or late phase doesn't change the entire HIF-1 alpha expression level in the tumor, which indicates those metabolic difference we observe may not result from those hypoxia severity change or HIF-1 alpha expression level change. It must be something else. But one thing we also need to remember, when we perform those metabolomic analysis, we use entire tumor. And there's a confounding factor is that the entire immune profile change when we deplete T cells. So you never know whether those difference is is due to the cancer cell only or due to, for example, macrophage increase when you deplete T cells or macrophage phenotype change when you deplete T cells. So to address this issue, we further cross our BRAF PTEN melanoma mouse model with the mTNG mouse model. In this case, the mTNG mouse model is a Cre reporter mouse. So we will be able to identify our cancer cells based upon the GFP expression. Since this entire mouse is digital metal positive, once Cre activity is engaged due to the reporter system, they will switch off digital metal expression and switch on GFP positive— GFP expression. So by using this approach, we perform early T-cell depletion. Similar to what I told you, from week 2 to week 5, every 2 days for 3 weeks. And then we isolate GFP-positive cells for ATAC-seq and RNA-seq. With this method, we'll be able to know whether they have any difference in terms of transcriptome, which reflects their metabolic difference. So this is ATAC-seq. These results suggest that those tumors growing in the absence of T cells display completely different chromatin architecture compared with those tumors growing in the absence of T cells. In the presence of T cells. And we see robust difference into the open chromatin region and closed chromatin region between these two groups. By using motif enrichment analysis, we further confirmed those tumor growing in the presence of T cells, the control group, they have a lot of C-MYC binding site become open, which indicates C-MYC should be actively participate in the transcription. To turn on those genes in those tumor growing in the presence of T cells. And we also see the AP1 upregulates in this case, which indicates they are more aggressive. So basically these results suggest that the presence of T cells during early tumorigenesis can change their malignancy, at the same time also change the C-MYC binding locus, which may reflect to their metabolic activity change. And then we move to the RNA sequencing analysis. We see the similar things as this ataxic indicates. The tumor growing in the presence versus in the absence of T cells, they have very different RNA expression pattern. And if you perform gene set enrichment, those tumor from the control group highly enriched miG target gene expression and down-regulate gene involved in oxidative metabolism. So which further indicates those metabolic activity we observe we observe is holding true. And this can change due to the entire transcriptional and also epigenetic landscape change when those tumor encounter T cells. So based upon this information, we propose that immunoadjuvants actually not only edit neoantigen expression, over MC class I expression. It can also edit the metabolic profile. So we named it immunometabolic editing process. We think this process, when we let the tumor, which define genetic mutation or driver mutation to form the tumor, to form a transformation event, if we wipe out T cells, those tumor cells will use whatever metabolic activity they can use. They don't care whether they need to engage aerobic glycolysis or oxidative metabolism, and they will be able to use whatever can support expansion and growth. However, if we see, if we let the tumor encounter T cells, the entire process will be altered. The presence of T cells will provide a selection pressure to force those cancer cells to start to use aerobic glycolysis and downregulate their dependence on oxidative metabolism. And by using this approach they will be able to evade from immunosurveillance. So this is our hypothesis. So we see a very beautiful change in the metabolic profile, but we don't really know whether this profile change can help the cancer cell evade from immunosurveillance or not. So in order to examine this possibility, we modify our design in the following way. Since we can use antibody to deplete T cells, we can also discontinue the antibody injection. In this case, T cells will start to come back. So we can see those tumors when they formed in the absence of T cells, how do they deal with T cells when the T cells re-emerge due to the discontinuation of antibody injection. So what we performed was very simple. We have the week 2 to week 5 early T cell depletion. So we divide them into 2 groups. The first, the first one here is that we further continue the entire depletion for another 2 weeks. And the last group is our experimental group. We stop, we let the T cell come back, and wait for another 2 weeks and see what happened. And this is the data we observed. In the control group without anything, just PBS injection, they have identical tumor growth kinetics as the continued T cell depletion. So it indicates as long as we deplete T cells, those tumors can still grow normally like the control group. However, those tumors forming in the absence of T cells during week 2 to week 5, once we discontinue the antibody injection, when the T cells come back, they fail to grow in the same kinetics and they have a smaller tumor burden, which indicates they fail to evade from T cell-mediated immune surveillance. And when we analyze what happened, we performed a series of immunohistochemistry staining to understand whether we can see any particular features reflecting the reduced tumor progression or apoptosis. As you can see here, by comparing this continued group versus a control, they have the same T cell abundance in circulation in spleen, in lymph node, and also in tumor. But we start to see very different features including proliferation features in the cancer. The discontinued group, the Ki-67 staining is much lower compared with the control. And this tumor also increased beta-gal staining which indicates they have senescence. At the same time, when we perform TUNEL staining, we can also see more TUNEL staining in this discontinued group which indicates those tumor, they're growing in the absence of T cells in the early phase. When they start to see T cell again. They fail to sustain their proliferation. They fail to avoid senescence and apoptosis. And this can be further confirmed by Western blots. You can see here P21, P16, and Clif-Cas3 expression level is much higher in the discontinued group. And this reflects one thing, whether this is due to the T-cell mediated immune response. Response. As I told you, the continued depletion, you can see those tumors grow normally, which indicates T cell is important. But how can T cell do that? How can T cell control those proliferation decline and apoptosis and senescence? And luckily, there's another paper published in Nature almost 10 years ago showing that T helper 1 cytokine, especially interferon gamma, can drive those features I just mentioned to you, decline proliferation, increase senescence and apoptosis in cancer cells. And this is the main reason interferon gamma is the most important cytokine to control tumor growth in both mouse and human. So we decided to take a look whether interferon gamma is the player to control those tumor growth in a discontinued group. In order to confirm we really see interferon gamma over there, we first check the targeting of interferon gamma expression in those tumors, which is PD-L1. So we performed immunohistochemistry staining of PD-L1. You can see here PD-L1 expression level can be seen by the green staining. By comparing the discontinued group versus the control, we did see higher level of the green signal, which indicates they have higher PD-L1. If we inject anti-interferon gamma antibody one day before we sacrifice the mice, we can decrease those staining, which indicates this indeed is interferon gamma-mediated PD-L1. Upper regulation in the discontinued group. So based upon this data, we performed cytokine arrays in those tumors by comparing control and discontinued group which have similar T cell abundance everywhere. But we still see the very robust difference in terms of cytokine profile. Those discontinued T cell depletion tumor, they have higher interferon gamma expression and also other cytokine. Which expression can be controlled by interferon gamma including ID10, as you can see here. So which indicates tumor growing in the absence of T cells during early stage of tumorigenesis, they will change their metabolic activity. And once T cells come back, they may fail to shut down T cell antitumor response due to their inability to reduce interferon gamma production in those T cells. So we next tried to come confirm whether this is the case. This is the only slide with FACS data. So we first confirmed that whether the T cell exhaustion pattern changed. Remember what I told you, they have the same T cell abundance. But in the discontinued group, you can see that those T cells, CD8 T cells in the tumor microenvironment, they expressed less PD-L1 and TIM-3. And you also see less percentage of double positive cells, which indicates they are less exhausted, terminally exhausted. And we also check whether they can produce interferon gamma in different kinetics and ability. As you can see here, those CD8 T cells from the tumor of this continued group, they actually have a higher ability to produce interferon gamma when we re-stimulate those cells ex vivo. So indicates that those T cells, when they re-enter the tumor cells growing in the absence of T cells, they would not be able to maintain the anti-tumor function and cytokine production. And the other very important feature we observed in this analysis is that actually the distribution of CD8 T cells are also altered. You can see here when we stain CD8 in these tumors, the control tumor, most of the CD8 T cells are in the margin of the tumor. In the tumor core, you don't see any CD8 T cells. However, the discontinued group, you start to see CD8 in the margin and you also see the CD8 T cell in the core, which indicates tumor growing in the absence of T cells in the early phase of tumorigenesis fail to create an inaccessible tumor microenvironment. They fail to prevent T cells infiltrate into the tumor microenvironment, which indicates those metabolic activity displayed by the edited cancer cells may allow them to create a harsher environment to restrict T cell infiltration. Okay, so with all the information I show you, it looks like interferon gamma should be the major players those cancer cells need to learn how to deal with. And metabolic process they display can be a very critical one to help them to deal with the interferon gamma producing ability by those CD8 T cells. So to confirm interferon gamma indeed is the one of those cancer cells when we discontinue antibody injection, we use the following step, we use the following design. We discontinue the antibody injection and then we start to introduce neutralizing interferon gamma antibody to block interferon gamma biological activity. In this case, we can restore the tumor growth again even they start to see CD8 T cell. In CD4 T cells. And we also make the tumor burden become very similar to the control group, which indicates interferon gamma is the main thing to shut down the entire growth of those discontinued tumors. And basically we also reverse all the phenotypes we observe. You can see here the beta-gal staining decreased once we neutralize interferon gamma in the discontinued group. We also reduce p21 expression, we also regain the KR67 staining. And this can be further confirmed by the Western blot. Interferon gamma blockade can easily reverse all the phenotypes. So which also indicates that what cancer cells, they need to learn during the early stage of tumor progression is to acquire the ability to reduce interferon gamma production in the CD8 T cells.. It also indicates interferon gamma should be the selection pressure provided by CD8 and CD4 T cells. So in order to confirm whether this is the case, we use the cell line we generate, the NACAR1 and NACAR7, they are non-edited cancer cells, and engraft them into wild-type or interferon gamma whole-body knockout. As I showed you before, in the wild-type they cannot grow. You will see they they start to grow a little bit and then regress. However, in interferon whole body knockout, even they can see CD4 and CD8 T cells. They have no problem to grow like the REC1 knockout background. So which indicates they only need to learn how to deal with interferon gamma, then they can start to grow again. So in order to know whether interferon gamma is the major player to provide this selection pressure to change the metabolic features of cancer cells. We modified the design again, and this time we didn't perform CD4 and CD8 T cell depletion. We just inject anti-interferon gamma antibody from week 2 to week 5 every 2 days for 3 weeks, and then check their metabolic profile. And we indeed see the same thing. Those tumor growing in the interferon gamma blocking antibody injected group, they upregulate metabolite in TCA cycle. Downregulate metabolite glycolysis pathway, which indicates the production of interferon gamma by our adaptive immune response play a critical role to control the metabolic features of cancer cells, which will allow the cancer cells start to upregulate glycolysis activity and downregulate their dependence on oxidative metabolism. So all the data looks pretty good and fits the hypothesis, but we still don't really know whether those cancer cells eventually have the chance to grow in the presence of T cells really need to change their metabolic process. So we took advantage of a few things. The first one is that if you remember what just Misha asked, if we engraft the cancer cells into the wild-type versus REC1 knockout background, Red1 knockout background, those non-edited cancer cells just grow normally. But the wild type, they will start to decline. But if we wait long enough, you will see some escaped cancer cells from these cell lines we generate. So by analyzing those escaped and non-escaped, we observe that by comparing with their parental cells, as long as those cells, they escape, from wild-type recipient, they start to downregulate their OCA to ECA ratio, which indicates they increase their dependence to aerobic glycolysis. And this can support their immune evasion. And we also use another model, which is genetic engineered mouse model. So if you remember what I show you, if we deplete T cell between week 2 to week 5 and we discontinue T cell depletion, tumor cannot grow normally. They have slower tumor growth kinetics. But if we wait for another 6 weeks, we can see 2 different tumors from this group. One is escaped, the other one is non-escaped. So this entire experiment is more like 3 months. And by analyzing this difference, we see the same thing. Once they escape, they change their metabolic profile compared with non-escaped. And by performing metabolite enrichment assay, Escaped tumor cells upregulate glycolysis metabolites and downregulate TCA cycle metabolites, which indicates these two things should go hand in hand. They have to change the metabolic process in order to escape. Okay, now let's go to the biochemistry part, which indicates interferon gamma— our data indicates interferon gamma can control the metabolic process of cancer cells, but how? How can interferon gamma do that? Do that. So we first check whether the cancer cells growing in the presence or absence of T cells can have different sensitivity to interferon gamma. As I told you before, interferon gamma mainly activates PD-L1 expression through the STAT1 pathway. So we first check whether the Y-type, the edited one versus non-edited one, they can respond to interferon gamma differently. By checking PD-L1 expression or MC class 1 pattern, we see the same thing. They respond normally, which indicates they still see interferon gamma. This is not due to the sensitivity issue. However, when we check the metabolic profile change, we start to see very different profile. This is the non-edited cancer cells. If we give them interferon gamma for 24 hours, you can start to see the OCA/ECA ratio change, which means— which indicates those cancer cells, if we give them 24 hours of interferon gamma exposure, they start to upregulate aerobic glycolysis and decrease their dependence in oxidative metabolism because their ATP level is very similar. However, those tumor cells already went through the editing process. They don't listen to interferon gamma in this case. They don't change the OCA/ECA ratio after 24 hours exposure even though they can still upregulate PD-L1. Which indicates that in the entire editing process, interferon gamma can trigger secondary signaling cascade to modify metabolic activity of cancer cells. So if you remember what I told you in the ATAC-seq, we observed that those cancer cells growing in the presence of T cells, they have higher C-MYC binding site become open. So we start to speculate actually C-MYC is the one interferon gamma regulate. So we give the non-edited cancer cells with interferon gamma for 24 hours, we can see the robust increase of c-Myc overexpression. And this is not the case in edited cancer cells. They already have very, very high c-Myc expression in the basal level, which indicates the editing process will let them sustain the c-Myc expression, and interferon gamma might be the trigger to trigger this entire sustained c-Myc expression pattern in the edited cancer cells. So in order to confirm whether the c-Myc is the player to regulate interferon gamma-induced metabolic reprogramming in cancer cells, we knocked down c-Myc by siRNA. As you can see here, if we give the cells interferon gamma, they will reduce OCA to ECA ratio as I showed you before.. But if they don't express C-Myc, you don't see this up— you don't see this downregulation, which indicates interferon gamma indeed controls the metabolic reprogramming in cancer cells by regulating C-Myc expression, which can— they can upregulate C-Myc. And more importantly, we also realize that the C-Myc upregulation in response to interferon gamma not only controls the metabolic preference of cancer cells, it actually can help the cancer cells to reduce senescence. This is the beta-gal standing. In the non-edited cancer cells, if we give them interferon gamma, they change their metabolic process. They increase c-Myc expression. And you will not see senescence. Okay, this is different from what people report. Interferon gamma exposure can drive very strong senescence. But indeed we can see what people report when we knock down c-Myc expression in those cancer cells, which indicates Interferon gamma can trigger c-Myc upregulation in cancer cells. This helps cancer cells change their metabolic activity. At the same time, it also prevents senescence induced by those cancer cells. So by using this approach, we further confirmed that actually this is the case with Western blots. The p21d1 can be upregulated when we get rid of c-Myc expression, and p16 also upregulates. So overall, our data suggests one very simple idea. When tumor cells encounter T cells during tumor progression, the presence of interferon gamma here will drive the signaling cascade, allow cancer cells upregulate c-Myc, and this will switch their metabolic activity, allow them to start to use glucose and then generate lactate production. And this also reflects what people already observed observed in patients treated with PD-1 blockade. Because in the PD-1 blockade, no matter it's Avi's paper or the other paper by Bob, by Dieter, they always realize that those patients in melanoma and breast cancer with PD-1 treatment, the first signature you will observe is that c-MYC target gene will be upregulated in an initial responder. They also observed in a further way, if those patients eventually become non-responders, they always have higher c-myc target gene expression pattern. But they don't really know why this is the case, which reflects that probably initially the first wave of immune response will use interferon gamma to trigger c-myc target genes and start to engage this new adaptation, allow the cancer cell evade eventually. So we need to understand how interferon gamma do this process. We may introduce a new treatment to prevent the acquired resistance of PD-1 treatment. So cut the story short, actually we realized that interferon gamma can control c-Myc expression by activating STAT3 rather than STAT1. Because STAT3 and STAT1, they have very similar homology. And we found that actually the non-edited cancer cells, if you give them interferon gamma, you will start to see stage 3 activation. But those addicted cancer cells, they don't listen to that. And if we give them stage 3 inhibitor, we can abolish c-Myc upregulation, which indicates interferon gamma in the first wave of immune response can turn on stage 1 and also turn on stage 3. For stage 1, which allow cancer cells to upregulate PD-L1, but for stage 3, which allow cancer cells to change the metabolic metabolic process by upregulating c-Myc expression. And you may wonder what's the reason interferon gamma can activate STAT3, but based upon the biochemical analysis did almost 20 years ago, people already know that if those cells they don't have STAT1 or their STAT1 level is very low, due to protein homology, JAK will activate STAT3 instead of STAT1. And actually that's the case between the edited and non-edited cancer cells.. The non-edited cancer cells, they have lower STAT1 expression, which indicates if we give them interferon gamma, they would tend to activate STAT3 greater than STAT1. And this is also holding true when we use the primary HCC cell line generating from immune-competent or immune-deficient mice. As you can see here, those knockout cells, they have a lower STAT1 expression. They also have a lower Seneca expression. So although our data indicate very simple idea, as I told you before, the first one for immune response, T cell will start to secrete interferon gamma. But interferon gamma will activate STAT1, then cancer cell will upregulate PD-L1 and try to shut down T cell immune response. However, this is not enough and also very dangerous for cancer cells because STAT1 engagement will trigger senescence., and PD-L1 upregulation also takes time and may not be sufficient to shut down the immune response in a timely manner. So the secondary pathway we identify here, which is the stage 3 c-Myc pathway, can coordinate with the PD-L1 pathway to fully shut down the T cell immune response by creating the metabolic stress happening in the tumor microenvironment. At the same time, the upregulation of c-Myc and prevents senescence and creates a perfect storm, allowing the cancer cells to evade from immunosurveillance more efficiently. So with this information, it looks like we have a new idea about how immunoadjuvant can change metabolic process in the cancer cells. But still, down to earth question is that if we will be able to identify particular the target, especially metabolic enzyme expressed by cancer cells, we should target because they use this enzyme to create this immune evasion process. So the way we try to understand whether we will be able to identify this entire target is to use the in vivo CRISPR screening. So we generate a guide RNA library targeting all the metabolic enzyme or metabolic related genes in mouse system and then transfuse the BRAF P10 melanoma cell line we generate. And then in this case, every single cell in this pool contains 1 to 2 GaIA only. And then we engraft them into B6 mice or REC1 knockout mice. If this metabolic enzyme is critical for immune evasion, you will see less biRNA-containing cancer cells show up in the B6 mouse because this enzyme is so critical allowing the cancer cells evade from immunosuppression. So we basically can analyze in this way to see which kind of RNA is abundant in the REC1 knockout but less abundant in the B6 Cas9 mice. And those genes, of course, those genes targeted by the guide RNA actually will be able to support the tumor immune evasion by suppressing anti-tumor immunity. And I will not go through every single enzyme because it's too detailed, but I will tell you that we identified more than 80 targets can be differentially expressed by this assay. And in this 80 enzymes here which can support tumor immune evasion, more than 50% they are C-mic directed. Target. Especially we see IDO1 is here and also other interesting candidates. And we already performed the analysis for two of the candidates. One is SLC23A2, which is a mitomycin transporter, and the other one is FASN. They will control the T cell differentiation pattern and also anti-tumor response. And those tumors without these two enzyme expression, they grow normally in the REC1 knockout background., but they will grow slowly in the immunocompromised. And we are in the process to understand why those two proteins and other proteins in this list can do this process and whether they can control the immune evasion or they can control the T cell differentiation pattern. So overall, I would like to give you the main idea of this talk. Immunoediting not only added the immunogenicity or neoantigen, it actually control— all the phenotypes which allow cancer cells to evade from immunosurveillance. And our finding provides the new information why cancer cells tend to stick on particular metabolic pathway in vivo. And this allows us to know which metabolic process everyone wants to target in the tumor microenvironment should be targeted if we can understand how this process is engaged. So overall, these results suggest that interferon gamma is the key players, and this entire process can be kicking all the time when you introduce new wave of immunotherapy. You just introduce a new type of selection, and they probably find out enzyme 1, enzyme 2, and enzyme 3 in different cycle of editing. So in the end, I'd like to thank people in my lab, especially as I told you, I initiated this entire project during my postdoc training in the last year, and I think it was very simple, but turned out it's It's not that simple and we need to generate so many different mouse models. And the experiment usually sometimes takes like 2 months to 4 months. And so without their support, their teamwork, it's not possible. This is the only teamwork paper in my lab. And also all the people support my, this entire project, especially Sue. We still collaborate in this particular project. And most importantly, the funding support for my lab in the past 6 years.

And I'll be happy to answer your question. Fantastic, thank you. I was wondering if you see this editing of the cancer cell metabolism by the T cells also on the level of the nutrients that they take up. So, do they reduce fatty acid consumption and things like that and switch to actually taking up more glucose?

And could that also impact then the T cells again? So, this is something we want to move into the next level. CRISPR screening data because it's difficult to analyze this, the usage of the nutrient in a single-cell base. But once we see the target, we can individually test how those targets affect this process. They may deprive the nutrient or they convert something and it generates an immunosuppressive secondary metabolite to shut down T cell response. So that's the reason we try to understand this entire process. But I think it's just The short answer is that we are not sure whether they only target T cells. There's one possibility is that those enzymes we identify here, they can change the fibroblast property as well, which will prevent the T cell infiltration by creating fibrosis features in the tumor microenvironment.

[Speaker:MALE_SPEAKER] Great talk.

It's really interesting.

Genetic models that you see when you deplete the CD4 and CD8 T cells, this sort of distinction where the metabolic editing is still occurring but the tumor burden was actually the same. Do you think that the sort of traditional immunoediting and this metabolic immunoediting are distinct phases where maybe the antigen-driven editing is already done and now it's doing this? It's just kind of interesting.

Normally you would see the tumor burden also go up. That's an excellent question. That's the question I was asked 7 years ago when I did my Chalk Talk, and I cannot answer this one, so I got killed by this interview. So initially when I see they have a similar tumor burden in between wild-type and BAK1 knockout or between the control and T-cell depleted mice, I also very— I'm also puzzled, you know, what's the reason? It suggests that the T-cell doesn't matter. And I dig into the more literature in the past few years, years, I realized that actually people identify the same thing no matter they use Sleeping Beauty system or like breast cancer model I show you or liver cancer. All of them can be, you can see all of them in those models wipe out T cells or using regular knockout background doesn't change the tumor burden and tumor growth rate compared with the immunocompromised. That's why if you go back to the literature, people argue that there's no any immunosuppression based on this data. But if you combine everything together, I think we have the answer. When you remove the immune selection pressure, the cancer cells use a slow proliferating metabolic activity, so they don't proliferate faster. When you remove immune surveillance, the tumor cell can grow faster. There's an assumption. Those tumors grow in the presence of T cells versus in the absence of T cells, they use the same metabolic process. They use the same they have the same growth rate. But turn out that's not the case in a genetic knock-out, a genetic tumor model. Remove the immune surveillance, they slow down their proliferation demand because they don't need to force themselves to use glycolysis activity. That's why they are very similar in terms of tumor growth rate. But we still speculate this is the reason. I think we need to have another way to temporarily control their metabolic process and know whether this is the case.

When you remove the T cells, so to speak, with CD4 and CD8, you of course do not remove all the T cells. And so have you looked at those, and might that be a reason why we don't see a difference in the tumor growth?

We check that, at least CD4, CD8 T cell, we see very, very low. In the RIK1 knockout, we know that we don't see it, but in the antibody depletion, we see very, very low level. So actually, before we perform those experiments, we kind of optimized the depletion approach we use to confirm we can deplete them, and also confirm that when we discontinue the T cell depletion, those T cells indeed can come back. So we actually optimized optimize that. I would say that the beauty for this is that in our mouse model, probably T cell is more important than other immune cells. If you use the mouse model, probably the NK cell is more important, or NKT, or other immune cells, they are more important. Probably will not be able to see this entire editing because the interferon gamma can be produced by NK cell or some other immune cells. So I would say that we just provide the possibility, but doesn't mean CDA is done. CD18, they are the only thing to provide a selection pressure.

And also, under immune pressure, the MHC expression of the tumor cells or their antigen presentation pathways might change, of course.

And so have you looked at any changes in those in association with metabolic changes? Cell line we generate, we know that after interferon gamma stimulation, they have the same level of MHC class I expression on the surface compared with the edited cancer cells. But we don't really know the antigen burden, whether they can really load antigen on MHC class I in a similar level. But it has been shown recently, not recently, 10 years ago, and also I know some researchers, they also try to understand this process. This is a process most people they forget. In the old biochemical study, people already noticed that when cells engage aerobic glycolysis activity, those cells will have difficulty to put antigen on MHC class I. This has been shown almost 20 years ago by a JVC paper. And in cancer cells, we see similar things but not super clear what happened, but it indicates that the metabolic enzyme indeed can control the antigen loading. I think this makes sense because it will control the ER and mitochondria activity. So I'm also aware in our screening some of the enzymes probably indicate this direction, so we also try to understand how they exactly perform this entire regulation to change the antigen-presenting load. So that's why I said that I put a very, I would say, vague way Huguet described the case night data under the CRISPR screening because we are not really sure whether they control T cell exhaustion, infiltration, or recognition. So we will tell you the answer probably in the next 5 to 10 years, individual one. [INAUDIBLE] Yes. [INAUDIBLE] Yes.

Why is it that the tumors—

sorry, it's sgRNA?

Yes, it's sgRNA. That give you greater tumor growth, is it, in RAG knockout mice?

Right. Why is it that they suppress antigen immunity, or why do you assume that they— So that means the cancer cells with this SGRNA expression, they have less chance to grow in the immunocompromised. Because if they have the similar— if this enzyme doesn't matter, you will see the guide RNA expression between RIG-I knockout and Cas9 is so abundant. So this is the— we check the guide RNA abundance in the tumor forming in the RIG-I knockout or in the tumor forming— Abundance. Abundance, yes.

I'm just surprised by how that you can inject antibodies so frequently and the T cells come back in a matter of, what was it, 4 weeks or 6 weeks or something?

So actually we need to titrate, we need to start to do, so the complete depletion is like you inject like 200 micrograms every 2 days, right? But if you want to let the T cell come back on the timing, You need to gradually reduce the dose because the half-life of antibody is like 21 days. Yes, so that's why I said— that's why I— right, so that's why I say we need to optimize the dose we inject. And when we stop the depletion, the T cell doesn't— those T cells do not come back in 2 days.

It still takes them like 1 week to catch up to the same level like the control. Not surprising, but so are those T cells that developed in that one week, do you think, or, and are they more tolerant because they grew up in a mouse with a tumor, or?

I don't think they are more tolerant, but that's one thing we expect that we probably need to perform TCR sequencing to understand their clonality difference between the control versus discontinuum. That's one particular caveat of this design, I have to admit.

Well, the rebound was just very surprising.

Yes. And by half-life. OK. Exactly.

OK. Do you know why the STAY-140 was reduced in the tumor? And then when you let T cell come back, does this STAY-140 go back?

Because STAGE1 is STAGE1's target gene. So there's a feedback. Once they see interferon gamma, they will upregulate, upregulate. So this has been shown by people, if you chronically expose cancer cells to interferon gamma, they will start to upregulate STAT1. And the data I didn't show you is that we have the non-edited cancer cells. If we give them interferon gamma 4 weeks in vitro, they will behave like edited cancer cells. And they change the STAT1/STAT3 ratio, they change the CMIC ratio, and they change the way they grow in vitro. From the immune-compromised when we re-engraft them. And we actually perform whole exome sequencing, RNA sequencing, or TEC sequencing to confirm the edited and non-edited cancer cells, they have the same genetic composition, same mutation, and same driver mutation as well, but their TEC-seq and RNA-seq pattern are totally different. So which indicates the entire process is probably epigenetic regulated by interferon gamma. [INAUDIBLE] In the screening or in RNA-seq? Either. In RNA-seq we see some mTOR pathway change, but not that strong. And in this one we don't really include mTOR because it will be super, super strong biases, so we target only the metabolic enzyme. [INAUDIBLE] We didn't do that for the acetyl-CoA reductase. Reason, because the late depletion, the tumor burden is already too big, we cannot really perform discontinuation because we probably only have like one week window can really do that. But I think based upon the one I show you, we do the early depletion and we discontinue, right, and then we start to see some escape and non-escape, so indicates that we just push the editing process later on. The editing always come back when the immune response coming back. So I think in principle this late depletion, if we introduce certain thing to boost the immune response, you will start to see the new wave of editing. And they probably acquire different enzyme should be expressed in this screening. Hi. Sorry, I must be late for the last question. What about the core associated environment microphage in the environment? I mean, they think of that first, at least. So when we perform the control and discontinuation, of course we know that the immune profile will change. So at least we did something to test. We confirmed that T cell abundance between control and discontinuation is very similar. And the DC, macrophage, B cell, and NK cell, they are very similar. But we didn't really perform single-cell RNA sequencing to know whether their functional alteration can be seen. But at least the abundance, they are quite similar. So with the interferon gamma neutralization, with the continuation of T cell depletion, I don't think those cells, they played a critical role there in our setting, I would say, in our setting.

Any other questions?

Thanks for this presentation. Thank you.