Skip to content

2026-04-21-pm-immune-stromal-reprogramming-micro-macroenvironment

Date: 2026-04-21 Primary topic: single-cell-spatial-omics Cross-referenced from: single-cell-spatial-omics Word count: ~12,884

Auto-generated captions

This transcript is Vimeo auto-generated English. Expect misheard names and technical terms. Use the ripgrep/search approach for exact quotes.

Good afternoon, everyone. We're going to start our session, Immune and Stromal Reprogramming Across the Tumor Micro and Macro Environment. So the field of the microenvironment, and I see many of the people involved in this research, which is the closest to my heart and my main topic, has been around for several decades, and I think for now, we don't need to convince anyone that the microenvironment is important after decades of research. But it also has become very clear that there's more to it. Tumors are ecosystems, and they're part of an entire organism, and there's also a lot of what we call now macro environment factors that affect tumor growth, metastasis, and very important, response to therapy. And this is what we're going to try to integrate. It's a huge amount of topics, but we're going to try and taste and integrate this in this session. And we have two speakers. Yuval Shaked will tell us about the host response to immunotherapy and how it changes the microenvironment, specifically ECM. Mitchell Fain from Fox Chase will tell us about a very important systemic factor at the level of the organism, which is aging, that affects tumor growth, and mostly it affects metastasis, as Mitchell will tell us. And then we'll finish with my presentation talking about stromal and immune interaction in organ-specific metastasis, and again, I hope you all enjoy it. What I didn't say is that at the end of each talk, you can ask questions, but you can also... There it is, the barcode. If you're too shy or if you're online, scan the barcode, and then text the questions. Okay. Let's start our session with our first speaker, Yuval Shaked. Thank you very much, Netta. I'd like to thank the organizers and the chair of this session for inviting me to this meeting. Today, I'm going to talk about immune checkpoint inhibitors and their association with the extracellular matrix, and all with respect to host response to treatment. So my lab in the last, I would say, two decades, is interested in resistant to treatment. And while the majority of studies focusing on how tumor cells are being affected by the treatment, we were more interested to see how the therapy affects many different host cells. They can be at the bone marrow or other organs, and they can be even at the tumor microenvironment. These are not side effects that we are talking about. I call it silent side effects. They are changing the tumor microenvironment, sometimes, as I will show you, to the tumor advantage. I call it the host response to anticancer treatment. By way of example, if you take non-tumor-bearing mice and you treat them with chemotherapy, and then you extract the plasma from these mice, and after that, you look into a different variety of assays that looking at the characteristics of cancer cells, what you can find is sometimes that the tumor cells become more aggressive when they are cultured with plasma from treated mice compared to plasma from control. So something in the plasma, in response to the treatment, promoted the activation of cancer cells. They are becoming excited. They are becoming proliferative and invasive. Another way to show it is that if you prime non-tumor-bearing mice with chemotherapy and then inject tumor cells to the mice, the mice that will show higher mortality rate are those that were first treated with chemotherapy. Again, there is something in the host in response to the treatment that promotes the aggressiveness of tumor cells. And these effects is not only in preclinical models. We can actually find them in the clinic, in samples from patients. Indeed, what we've done is we took plasma samples from colorectal cancer patients before and after treatment with FOLFOX chemotherapy. Before baseline and after treatment, it's 24 hours after the first treatment, and we did the same assay that I just showed in mice. 70% of the patient will show increase in metastatic properties of the tumor cells by means of invasion and filtration throughout the Boyden chamber, as you can see here. So this means that the host, in response to the treatment, promotes a lot of regenerative factors and proliferative factors that support the growth of the tumor. It can explain tumor relapse and metastasis. And throughout the years, we have done work not only with chemotherapy but also with radiation and targeted drugs such as bortezomib, anti-angiogenic drugs, surgery, and more recently, with immune checkpoint blockade. In each and one of these treatment modalities, we do see these effects. There is infiltration of a lot of immune cells into the tumor, and not only immune cells, but here we show only the immune cells. They secrete a lot of different factors that makes the tumors more immunosuppressive, and that could actually help tumors proliferate and metastasize. But one component that we neglected so far was the extracellular matrix. And we ask, what happened to the extracellular matrix? And the reason why it's important, because there is a huge interplay between different cell types and the extracellular matrix.Cancer-associated fibroblasts that support ECM remodeling. The cancer cells themselves. Immune cells, mostly myeloid cells that also support. And the interconnection between all these cells can promote ECM remodeling. And why ECM remodeling is important, because we know that it's pro-tumorigenic. It promotes resistant cell death. It escape the immune system. It is being used as a barrier, as a physical barrier, not allowing T-cells getting into the tumor, promote invasion and metastasis, and induce angiogenesis. So really, we're interested in the ECM. And when we say ECM remodeling, is not only the expression, the novosynthesis of different micromolecules that support the ECM, that make the ECM, but also post-translational modification of enzymes. And many of these enzymes are known to degrade the matrix. These are the MMPs, cathepsin, ADAMs, heparans, and others. They will degrade the ECM, the macromolecules. But in addition to them, so far in the world literature, we know about two enzymes that assemble the ECM. This is the lysyl oxidase, which is five members of one family, and the BMP1, which is upstream to LOX. They are known to assemble the ECM to cross-link collagen and elastin. And indeed, we found that there is increase in LOX expression in the lungs of mice that underwent chemotherapy. And this could explain why after chemotherapy, we do see sometimes more metastasis in mice that have been first treated. Long story short, what we showed in a previous study is that chemotherapy can induce ECM remodeling, and this support the seeding of cancer cells in the lungs due to the stiff tissue in the lungs. What we found is that CD8 T-cells expressing lysyl oxidase are the one that support chemotherapy-induced ECM remodeling. And when we talk about CD8 T-cells, what questions that we really were interested then was what happened when you give immunotherapy. Would you get a similar effect or changes in the ECM? And the story that I'm going to tell you is about the immunotherapy-induced ECM remodeling. And why it's important? It's important because we know that only 30 or 40% of patients will benefit from immunotherapy, and the other patients will not. And there are many different reasons for resistant mechanisms for immunotherapy. Among them are those that are associated with immunosuppression activity. The ECM is immunosuppressive. It stores a lot of factors that are immunosuppressive. And also, as I mentioned, it serves as a physical barrier at not letting T-cells infiltrating the tumor. So we really were interested to study whether immunotherapy induced ECM remodeling. So in order to do it in an unbiased manner, we were looking for a homogeneous population of patients, cancer patients, that we have paired samples before treatment, usually at biopsy, and after immunotherapy, usually when you have tumor debulking. And we found in triple-negative breast cancer, a study that was published in "Nature Medicine" a few years ago. And we found 12 paired samples that we have the same patients. So we looked at the immune cells within these tumors and the expression, the differentially expressed genes between before and after treatment. And what we found that there are about 3,000 genes that are differentially expressed when you look at the whole immune system. And among these biological pathways that are enriched in these cells are usually cell matrix adhesion, of course, immune modulation, ECM remodeling, and epithelium development. And what really struck us was the fact that we found that the majority of the ECM remodeling and cell matrix adhesion were found in the T-cells. So we really were interested to see whether ECM remodeling can be recapitulated in animal models. So we used the 4T1, the triple-negative breast cancer syngeneic tumors, and we asked what happened when you give anti-PD-1. What happened to the ECM in the tumor? As you can see here with Sirius Red staining, there is dense ECM collagen I and III staining after anti-PD-1 compared to the IgG control. Here is another tumor model, such as Lewis lung carcinoma. You can see again these fibers, massive dense fibers following anti-PD-1 compared to control. Since our suspect are the T-cells based on the single-cell RNA sequencing, as well as based on the previous studies we have done, and the fact that it's anti-PD-1 treatment, we ask what happens when you don't have T-cells. We use SCID mice that are lacking B and T-cells, and we did the same experiment. As you can see here, after anti-PD-1, there is no changes in ECM in the tumors. And in fact, when we now adaptively transfer T-cells from normal mice into these SCID mice, you can restore the ECM remodeling after anti-PD-1 therapy, suggesting that T-cells indeed responsible, either directly or indirectly, to the ECM remodeling. So we wanted to study what could be the reason, and we looked at the different enzymes, post-translational modification of the ECM, and we found that BMP1 and LOX are upregulated in the T-cells.We validated it by looking into conditioned medium of CD8 T cells, but we also find it systemically. We found that also in the plasma of anti-PD-1-treated mice, we have higher levels of BMP1. What is BMP1? It's a bone morphogenic protein 1. It plays a role in the formation and development of the extracellular matrix. How it does that? It does it in a variety of ways. It cleaves procollagen, usually I, II, and III, into active collagen, fibrillar collagen. It also cleaves pro-LOX, make it active LOX, then it can cross-link collagen and elastin, and it also release TGF-beta from the ECM. BMP1 in tumors, overexpression is known to be pro-tumorigenic and support metastasis. And there is an inhibitor for BMP1, a small molecule that developed by Pfizer, for fibrosis, not for cancer, and it's called UK-383367. I will call it from now on UK. So the obvious experience that one would like to try is to see what happens when you combine anti-PD-1 with the BMP1 inhibitor. So the BMP1 inhibitor as a small molecule and short half-life, you need to give on a daily basis, whereas the PD-1, you give twice a week. And again, you use the 4T1 breast cancer model that is resistant to anti-PD-1. But once you use this combination, now you can see the tumors become sensitive to anti-PD-1 therapy. And why they are sensitive? They are sensitive because if you look at the ECM, there is reduced ECM in the combination treatment compared to the anti-PD-1. And what else you can see, that there is more infiltration of activated T cells into this tumor, suggesting that blocking BMP1 in combination with anti-PD-1 can actually make resistant tumors sensitive tumors to therapy. But we wanted to really, because we saw that in the plasma there is increase in BMP1, so there is systemic increase in BMP1, but not only in T cell. We really wanted to figure out whether the BMP1 in T cell is as responsible. So what we did next was we did an experiment where we basically took the SCID mice, and we adaptively transferred T cells either expressing or not expressing BMP1 using CRISPR-Cas9, and so you can see that there is knockdown of BMP1, and then we gave anti-PD-1 treatment. And you can see that when you don't have BMP1, there is no ECM remodeling in these tumors. Suggesting that it's sufficient to block BMP1 just in T cells in order to get the effect on the ECM. So it doesn't necessarily have to be a systemic effect. It could be really local effect with the BMP1 in the T cells. And we figured out the mechanism, how actually it works. So if you have anti-PD-1, it actually activate the T cells. In the granules, you have stored BMP1 that is secreted after activation. The BMP1, as we said, act on the ECM and makes collagen, activate LOX, and therefore, make more ECM or ECM remodeling. But it also, as I mentioned, release the TGF-beta from the ECM. And once it release it from the ECM, it can bind back to the receptor, the TGF-beta receptor, that now will exhaust the T cells. So when we have kind of a positive loop that will inhibit the activity of T cells once you give anti-PD-1. And to prove that, here I'll show you that when you take T cells, dump them cells in vitro, and you add anti-PD-1, you get activation of the T cells. But if you do it on the experiment when you have ECM in the culture, then you see that after anti-PD-1 treatment, there is no activation of CD8 T cells. But you can restore the activation if you block either the TGF-beta receptor or you block BMP1 activity by the enzyme, as I mentioned. So what you have here is not only inhibition of the physical barrier, but you also have ignition of the T cell, making sure that they not undergo exhaustion and actually work better in this case. So due to these studies, we decided to develop an antibody-drug conjugate. I call it the first-in-class antibody drug conjugate because usually the antibody drug conjugate, what do they do? We have three components that are very important. The first one is the specificity, which is based on the antibody that you use. You have a linker that is very important to know how many payloads you put on each antibody and if there is a steric interference with the antibody. And you have the payload that, in our case, is a BMP1 inhibitor rather than a cytotoxic drug. Currently, most of the ADCs that are in the clinic are based on payloads of tubulin inhibitors and DNA-modifying agents, which are cytotoxic agent. So what we decided to do is to use an enzyme inhibitor as a payload, and that we will know that we'll harness the immune system and will reduce the frequency of drug administration. We don't need to give the UK on a daily basis. It's on the antibody, so it may stay for a longer time. So then we tried the same treatment, but this time using the ADC. And you can see that just by adding the ADC, you get, again, tumor growth retardation in this 4T1 tumor model. We've tried additional models such as melanoma and Lewis lung carcinoma. And why it happens? It happens because it reuses the ECM content or the remodeling in the tumors with the ADC, and you get ignition of-Or a full activation of the T cells that infiltrate into these tumors. So you don't have the physical barrier, and you have more activation of the T cells due to this ADC. Last but not least, what we were also interested in is to see, like this is all preclinical models, what happened in clinical situation. So we were collaborating with Roche on a clinical study that was done in Essen, by Sven Brandau, and we actually looked at small number of patients where we, again, have a biopsy and after treatment with immunotherapy of head and neck squamous cell carcinoma. And you can see that almost all patients show higher levels of ECM remodeling after treatment with immunotherapy. And if you look on the immune cells, you can see that CD8 T cells express higher levels of BMP1 after treatment, suggesting that what we have seen in preclinical model might be relevant to the clinic. And this is, I would like to conclude that ECM remodeling after immunotherapy is associated with ECM remodeling. This is done due to de novo synthesis. I didn't show you, but we have done also work on showing that there is more fibrillar collagens that is highly expressed and activation of post-translational modification by enzymes. In addition to cancer-associated fibroblasts, myeloid cells are also known to support ECM remodeling. But here we show that also CD8 T cells can generate factors that can remodel the ECM. Blocking BMP1 in combination with immunotherapy, it can sensitize tumors that are otherwise resistant to immunotherapy. And tumor ECM remodeling was found also in clinical samples using head neck squamous cell carcinoma. And we propose here first in class, first time antibody drug conjugate that encompass a PD-1 with an inhibitor of ECM remodeling enzymes, the BMP1, in order to make resistant tumors to immunotherapy become sensitive. And I would like to thank the people that were involved in this work. First and foremost, Bar and Josephine, that equally contributed this work. Josephine already a post-doc in the UK, but she continued working on this work. And many collaborators from the Technion, that have done the work on the single-cell RNA sequencing, Sven Brandau, Keren Yitzhak, and also Karbian Odivalla-Dilke, that has done some of the staining that I showed with the tumor samples from the UK. With this, I would like to thank you for your attention. Thank you, Yuval. We have time for questions. Please go to the microphone if you have questions, and introduce yourself before asking. Maybe by the time people are walking in, I'll start with a question. Yuval, did you check whether the thing that you're seeing with the changes is specific to any organ? Did you look at different metastasis, for example, because 41 will metastasize to the lungs, but it can also metastasize to the brain. So did you see whether this mechanism is specific to a certain organ, or is it general response? So in this study, we looked at the tumor, at the primary tumor. But I can tell you that indeed, we looked also on other organs. We look at the lung, and in the lung, we also see changes in the ECM. And in fact, if you look at studies, these are reviews that were published, clinical reviews. They started to say that after immunotherapy, there is a pulmonary fibrosis in some patients. So this could explain what we see here, could explain this. So we looked at the lungs, and we found it. And in the lungs, the reason we think that in the lungs you will see it mostly is because the lysyl oxidases need oxygen to work. So that's why this perfused tissue, oxygenated tissue, where you will see mostly the effect. So in the lungs, we do see this effect. Okay. And we have a question there. Hi. Very interesting data. I am Devashish Sain from Washington University, St. Louis. I was just wondering that, have you checked whether this PD-1 treatment leads to also changes, like through T cell, leads to changes in fibroblast phenotype in the tumor microenvironment? No. This, we haven't done that, and I'm sure that Netta will say that I need to do it as well. Yeah. So we haven't done that. So we haven't looked into what happened to the fibroblast, and I'm sure that also will go some kind of changes, but we haven't looked into that. Thank you. Hi. Xavier Tecpley from Oslo University Hospital. Very interesting talk, looking at different interaction between different cell type and cell states. I had a question about BMP1 expression across different T cell state indeed, and if you had the chance to look if in different single-cell RNA sequencing or different datasets, which T cell may express BMP1. Yes. And as a continuation of that, within the dataset that you basically mined in the initial single-cell RNAseq experiments, did you look if the responders versus non-responder had a different expression of ECM-related genes? So yeah. Thank you for these questions. Yes, we have done that. So first of all, most of the T cells that express are in the states of memory T cells that express that. That's where you see BMP1, but I have to say BMP1 is not highly expressed on these cells, so it's really only after the activation that you can start to see the BMP1 expression. And regarding the other question, yes, we have looked also into BMP1 levels and maybe by BMP1 levels, predict response or non-response to immunotherapy. I don't have a clear conclusion. That's why I didn't put it on the slides, because I think that we don't have enough samples to test it, because we had to look to samples that when you have the same patients before and after. You cannot just look at the levels of BMP1 because it could vary between a patient and between samples. So it was very important to look before and after the same sampling, and we didn't have many of those, so we didn't get into conclusive results about that. That's why we don't want to say that. Thank you. Thank you. Charles Earnshaw from Cancer Research UK, Manchester. I really liked the story and the BMP1 idea from CD8 T cells to TGF-beta. Obviously, TGF-beta activation from the latent band form is known to involve things like GARP and integrins as well, possibly on other cell types. Do you think there is a link between the CD8, a physical link, maybe tumor cells, fibroblasts, or monocytes that might be also involved in the activation step that you're seeing? This is an interesting view. We haven't looked into that, but I will go back to the board and look into that. Thank you. Thanks. Yeah. Last question. Matt Maxwell. I'm a postdoc at Tempus AI. I was curious if you've looked at clinical data to see if, not just non-responder versus responder, but is BMP1 predictive of hyperprogression following ICI treatment? So not just not responding, but rapidly progressing. Yeah. We have a lot of studies we have done on hyperprogression as well, not with respect to this study. And, so again, I don't have answer for that, for these questions. But I do want to mention, remember that even in the patient data that we showed with the head and neck squamous cell carcinoma, we showed that the majority of the patient show higher levels of BMP1 in the T cells. Okay? Majority of the patients, even though they responded to treatment. So I'm not sure that we can count on this as a biomarker. However, remember that also the treatment was a very short time of treatment, because it's before tumor debulking, so the immunotherapy is given, I think once, and after six weeks you get the surgery. So, maybe if you give consistent treatment of immunotherapy, maybe then it start to kick in and then you will see differences, but we don't have an answer for that. Great. Wonderful talk. Thanks. Thank you. Thank you, Yuval. And our next speaker is Mitchell Fain from Fox Chase, and he's going to tell us about the role of aging in immune-mediated reactivation from metastatic dormancy. Go, Mitch. Yes. Thank you so much for that introduction. It's a real pleasure to be here. We opened our lab about three years ago, so yeah, I'm very excited to talk about the role of aging in metastatic dormancy progression, particularly in melanoma. So I have nothing to disclose, and if there's one thing that you can take from the talk that I'm giving today, I hope it's the fact that cancer is indeed a disease of aging. We see that aging is one of the highest prognostic risk factors for the majority of cancer, and if we simply just look at cancer incidence over time, for a large number of cancers on the left here, and then we have melanoma specifically on the right, which is what I'm going to be focusing on today. You can see past the age of 30, we start to get this really rapid increase in cancer incidence. But what's also really interesting and what I got interested in when I first started the lab is that once you get to around the age of 85, you actually have this inflection point where the majority of cancers actually start to decrease with age. So we got really interested in this difference between these middle-aged patients versus these geriatric patients. And, unfortunately, I won't have time to talk about this decreasing trend, but my postdoc, Dr. Kelly Coton, is speaking right after this in the aging micro and macro environments and tumor progression and therapy session in ballroom 6A. She's a great speaker and has a really, really interesting story. So go check that out if you're interested. But today I'm going to be focused on more on the role that middle age plays in driving this increase in incidence, progression, mortality, and all of these different factors, particularly within melanoma. So who cares about aging? It's interesting statement. A reviewer has actually said this to me before, and if you look at healthy aged tissues, so no cancer or anything like that, what you tend to see over time in the lung, in the liver, these really common metastatic sites in healthy aging is you get dysfunction of a huge amount of different stromal cells. So fibroblasts change the soluble factors that they secrete with age. They change the way in which they remodel the extracellular matrix. We get really dramatic differences in angiogenesis. And what I'm going to mainly focus on today is this dysfunction in immune cells with age. And when we look at, and as I said, in a normal healthy human, these dysfunctional phenotypes that we see in these stromal cells almost mimic exactly what we see in a very aggressive tumor microenvironment, be that a primary tumor or even a metastatic tumor as well. And one of the reasons why this is a particular issue in the field is that I would say about 90% to 95% of studies are done in preclinical animal models in these eight to 12-week-old mice. This is the clinical equivalent of a 20-year-old human for a disease where the median age of diagnosis is 65. So there's a famous saying, we could treat and cure cancer in mice, but we can't do it in humans. And I think one of the reasons why translation is so abysmal is because we're putting these therapies into these young mice that have a fully intact immune system, don't really have any of these age-related deficits. Whereas when we see in the clinic, a lot of these patients are in their 60s to 70s, and they have that age-related dysfunction. So that's something that I wanted to really start looking at when we set up the lab. The other issue is that there's often not very good models to look at colonization or metastasis across a wide variety of different organs.Particularly in melanoma, the majority of studies are done in the lung because there are some syngeneic melanoma cell lines that will disseminate into the lung, and you can study for metastasis, or you could do a very simple tail vein injection and the cells will go to the lung as well. A very understudied site in melanoma is the liver because there aren't great models of colonization. And one of the reasons why this is important is there are studies now in the clinic that show if patients have localized liver metastasis only versus patients that have localized lung metastasis only, the patients with liver metastasis will actually respond much worse to immune checkpoint blockade compared to those with lung metastases. So it's important to look at multiple different organs to really increase the ability for us to give personalized patient care, particularly with regards to age. So I was very fortunate to join Dr. Oshani Wiraratne's lab in 2017, for a six-year postdoc. It was the best six years of my life. And what we actually did is we would get these syngeneic melanoma cell lines, these YUM 1.7 cells, so the exact same genetic model. We would put it into the exact same genetic mouse background, a C57 Black 6, fully immunocompetent, and the only difference would be the age of the mouse. So we looked at these eight to 12-week-old mice, young mice, versus these middle-aged 12 to 16-month-old mice, because this is the time in which the incidence really starts to dramatically increase. And what we saw very interestingly is that if you form primary tumors in these mice, they grow much faster in the young skin relative to the aged lung. However, if you look at dissemination and outgrowth via the primary tumor or you do these colonization assays, you see the exact opposite effect. Melanoma cells are able to grow much more efficiently in the aged mouse lung compared to the young mouse lung. So indeed, organs seem to show this age specificity in how they regulate melanoma growth. And this is a lot of the studies that we were doing during my postdoc. So I'll just quickly show what we found during my postdoc, so I can show you some of the more unpublished data we've been able to get at Fox Chase in the liver. But basically what we found is that melanoma cells were able to disseminate into the lung in young and aged mice. In the young lung, they would undergo this prolonged period of dormancy driven by factors inside melanoma cells like WNT5A expression and the AXL tyrosine kinase receptor. However, in the aged lung, what we found is that aged lung fibroblasts actually secrete very different soluble factors compared to young lung fibroblasts. We were able to show this in human and mice. And one factor that was upregulated in aged lung fibroblasts was SFRP1. So SFRP1 was able to basically inhibit WNT5A and AXL signaling. It would drive expression of this MerTK receptor and really aggressive outgrowth and proliferation. And not only was that simply just the stromal microenvironment inducing really aggressive outgrowth in the aged lung, but then these reactivated melanoma cells would also secrete other factors that would drive really aggressive outgrowth. So one factor we found was protein S1. It was able to act on other dormant melanoma cells in the lung and induce further reactivation and drive really aggressive outgrowth. So what we didn't address in this study is any immune changes with age. Unfortunately, in that particular study, we had to use a fluorophore to track full-on metastatic dissemination and outgrowth. So we didn't look at the immune system, and what we've done since we've come to Fox Chase is we now have these 20 color flow cytometry panels where we can look at differences in lymphocytes, myeloid cells, and some downstream kind of signaling as well within these cells. And then we only looked at the lung. So we wanted to ask the question when we started at Fox Chase, what about the liver? So we're very fortunate to recruit Dr. Kelly Couton, a really amazing postdoc who came from a uveal melanoma lab. So the majority of uveal melanoma cases do metastasize to the liver, so she had great experience in the liver microenvironment, but we're doing cutaneous melanoma. So the way in which we wanted to look at this, unfortunately, there's just no syngeneic melanoma cell lines that disseminate from the primary tumor into the liver to study this effect. We're trying to create some right now. But in the meantime, what we would do is these invasive surgeries where we open these mice up, we inject directly into the hepatic portal vein, and the cells go directly into the liver. So we were able to do this successfully. You can see we only had to inject 50,000 cells, and after 14 days, we got really aggressive outgrowth in the liver. The problem is right before we started doing these studies, we started to look at the different ages of the mice we were using, and we wanted to avoid doing these invasive surgeries because, I mean, 12 to 16 months versus these geriatric 22 to 26-month mice, we thought that they would have very different variable immune responses, even regardless before the cancer is in there, and we wanted to avoid doing that. So we were very fortunate to stumble across this hydrodynamic tail vein injection method. So a normal tail vein, and you can see we use luciferase vector for this analysis right here, but all the rest of our experiments are done without any fluorophore or luciferase agent. But in a normal tail vein, you're putting a very small amount of liquid containing melanoma cells into the vein, and at least with melanoma cells, almost 100% of them go into the lung and start to grow out. Hydrodynamic tail vein, on the other hand, however, you're getting a very large volume of liquid with the same number of melanoma cells, and you inject it rapidly, and what it does is it increases intravascular pressure in the superior vena cava. It induces cardiac congestion, and then rather going to the lung, it actually drives blood flow in a retrograde direction in these mice for about two minutes, and it drives all of these cells into the hepatic vein. So you can see if we use the exact same number of melanoma cells and the exact same amount of time as a tail vein, all of these cells are now going to the liver. So now we had this nice colonization model where we didn't have to perform these surgeries, so we asked the question: Is there any difference in outgrowth between young and aged mice? And lo and behold, what we're able to show is that these young mice weren't able to grow out very efficiently in terms of their ability to colonize the liver, but we got significantly increased outgrowth in the aged liver. We repeated this across three cell lines, and they had different metastatic variability, but indeed, we saw across all three cell lines that they were able to grow much more efficiently in the aged mouse liver.So hopefully what I've just been able to show so far is that the lung and the liver in young mice basically suppresses the ability to grow out, whereas in the aged mice, they have much higher metastatic propensity, and the ability to colonize as well. So the next question we had is, what are the actual immune changes that are occurring with age? So we had two specific aims we wanted to accomplish. The first was to look at the pre-metastatic healthy lung and liver to see if with age, they're actually creating this niche that once the melanoma cells can get in there, they're able to grow out without being affected by effector T cells, NK cells, and these other lymphocytes. And then the second was to look at whether these changes actually also occur in the post-metastatic niche. So we were specifically looking for consistent changes in the lung, liver, and also systemically, so either the blood or the spleen. And we simply started by looking at CD45. And what was interesting is that in these healthy mice, in the actual peripheral tissues, we get a significant decrease in CD45 infiltration, but if we look at the spleen or blood, we actually get an increase in the amount of CD45 cells. So this is kind of not really surprising to us. I think we see in humans over time, you get this low-grade increase in persistent chronic inflammation. But it was quite interesting that in the tissues you actually have less resident CD45 cells. So the first thing we wanted to find is a more of a prognostic biomarker, so some of these cytokines, chemokines, and other factors that may be increased or decreased with age, which might be effectors of the particular phenotype we were seeing. So we did a screen, and these were some of the factors that we potentially thought could have a role in age-induced metastasis. And the one that stood out to us when we did flow cytometry was arginase 1. Arginase 1 was significantly increased in CD45 cells in the lung. It was significantly increased in CD45 cells in the liver. And interestingly, arginine levels in humans really dramatically decrease with age as well. So what is arginase 1? It's an enzyme that breaks down L-arginine to L-ornithine and urea. And basically, arginine is very critical for T cell and other innate lymphocyte function. So when arginine gets broken down, these T cells are no longer able to proliferate very well. You get a decrease in their cytotoxicity as well. And a lot of these studies have been done in CD8 T cells in cancers like PDAC. And interestingly, there isn't much studies done on NK cells, which I'll go into a bit later for our talk. So in the literature, they've largely been produced by myeloid cells as well. So we were interested in finding the specific myeloid cell that might be increasing arginase 1 with age. But before we did that, we went to the human literature first to see whether or not ARG1 was actually an age-induced factor in melanoma. So we looked in melanoma patients in the TCGA data set, and we found that patients with higher levels of ARG1 had decreased survival. And then we also stratified patients to being under the age of 50 versus over the age of 55, and indeed, ARG1 is increased in these older patients. So we then wanted to ask the question, what immune cells is it actually increasing in? So we looked at these kind of myeloid cells, and we wanted to see whether they were increased in neutrophils, MDSCs, macrophages. We looked at every single immune cell, and what was interesting is we couldn't pinpoint one particular immune cell. It just seemed to be dramatically increased across all myeloid cells in the healthy lung and in the healthy liver as well. So we went ahead with these kind of CD11b positive myeloid cells. When we look at these metastases, we can see that they're also significantly increased via flow cytometry. And when we do these IHC stainings as well on the metastases, we can see that there's much more ARG1 in the aged lung and liver compared to the young lung and liver as well. What was also really interesting is we went back to our previous study, which looked at young versus aged lung fibroblasts. So these are from healthy human patients, and these are secreted soluble factors from these healthy human patients. And previously, we were interested in SFRP1 and factors that drove melanoma proliferation, but now we went back and looked at some of these cytokines and chemokines which might change, and lo and behold, ARG1 was actually being secreted very heavily by aged lung fibroblasts. So it doesn't just seem to be an immune specific effect with age, but it also seems to be this fibroblast specific effect as well. So we went and we got whole lung lysate from young and aged mice, and when we did Western blots on them, ARG1 was really dramatically increased in the aged lung of these healthy mice. So just to recap so far, hopefully what I've been able to show is that the young lung kind of maintains dormant melanoma or doesn't allow very efficient colonization, and we have these myeloid cells with age secreting large amounts of ARG1 along with these aged lung fibroblasts. So the next question we wanted to ask is, if we flood these young mice that have low levels of ARG1 with ARG1 using recombinant ARG1 protein, can we actually reactivate melanoma cells from dormancy in young mice? And then can we actually inhibit ARG1 in aged mice to stop aged-induced colonization? So we'll start with young mice first. We basically formed primary tumors with these mice to see whether or not recombinant ARG1 actually affected them, and interestingly, there was no change in primary tumor growth. So we then used mCherry to basically look at whether or not we saw an increase in metastasis. So this is our PBS treated control, and what we would always see in young mice after about five weeks is that you could only see these single cell colonies. So these red cells are the melanoma cells. But if we treated with recombinant ARG1 throughout that entire primary tumor process, you can see we're now getting reactivation from dormancy and some metastatic colonies forming. So that was really interesting to us. We next moved on to the liver. As I said before, these cells don't disseminate to the liver, so we had to do hydrodynamic tail vein. And when we would treat with recombinant ARG1, we saw that there was a significant increase in colonization in the liver. So we then wanted to look at what effector immune cells were potentially being kind of regulated by ARG1. We thought it would be CD8 T cells, but the main immune cell that we saw getting down-regulated in melanoma was actually NK cells. So you could see this is our PBS control. These are a particularly rare subtype, these NK cells, but they do infiltrate quite dramatically in young mice into the metastases. But when we treat with recombinant ARG1, we get a really dramatic decrease in NK cells infiltrating into these tumors now.So we wanted to ask, does ARG1 directly regulate NK cells? So to do this, first we did a simple experiment. We got healthy young mice, we treated them with PBS or with recombinant ARG1, and just looked at NK cells. And overall, we saw a significant decrease in NK cells in the liver and in the spleen. So this seems to be a particular effect that's occurring even in the absence of melanoma as well when we increase ARG1. So we then went back to our aged mice, and we looked at young versus aged healthy mice. And lo and behold, what we can see is that NK cells significantly decrease in the lung, in the liver, and in the spleen. So it seems that even in aged healthy mice, we're getting these decreases. And when we look at the metastases, we also see that NK cells decrease via flow cytometry and also via IHC in the metastatic lung and liver as well. So we then started to look at some of these cell surface markers that may indicate exhaustion, activation, inhibition on these NK cells during metastasis. And what we found is that PD1 and TIGIT expression was significantly increased. We saw an increase in these inhibitory markers like KLRG1 and LY49, and we saw a decrease in these activation receptors as well, like DNAM1. So with this, we then wanted to test whether or not ex vivo, if we got human NK cells and we treat it with ARG1, does it actually regulate their function? So we did this NK92, which is a human NK cell killing assay with human melanoma cells. So A2058 are our human melanoma cells. This is an Acceligen assay, so no NK cells, they grow just fine. If we add NK in a one-to-one ratio, it limits their growth. NK cells with a one-to-five ratio significantly decreases it anymore. And what we were going to do first is treat these NK cells with recombinant ARG1 prior to killing. So we did three days of NK cell treatment, but what was really interesting to us is before we even did these cell killing assays, we actually found that treatment with it significantly stopped their ability to grow. They basically had the same number that they had when we first seeded them before the ARG1 treatment. So we have done sequencing on this, and I'm happy to discuss this afterwards. I won't have time to show it today. But basically what we did is we got these cells, we seeded them with the same density as the PBS control. And what we found is that no NK cells versus our PBS control, you can see that we get efficient killing. But if we pre-treated with ARG1, we actually significantly decrease their ability to kill these melanoma cells. So the summary with that basically shows that we have these NK cells that are very high in the lung of young mice and they're maintaining dormancy. Whereas we have ARG1 being secreted by things like aged lung fibroblasts and these CD11b cells, which is inhibiting NK cell number and function and driving really aggressive outgrowth. So the last question we had is, can we actually inhibit ARG1 to stop aged-induced metastasis? So thankfully for us, there is this inhibitor, a CB-1158. It's a small molecule inhibitor of ARG1 and ARG2. And it's been shown to be very active in PDAC cancers where ARG1 expression is quite large. And what we found is that when we did this in aged mice, we would get these aggressive metastases in our negative control. But when we treat with the ARG1 inhibitor, we're able to significantly decrease liver colonization. So this was very exciting to us. We then wanted to see whether or not it was affecting NK cells. And when we treat with this inhibitor, we not only significantly increase NK cells, but we also decrease PD1 expression on them. We increase some of these activation markers as well. So the final experiment we did is in young mice. Given the fact that they have low ARG1 levels and high NK cells, we wanted to see if it could still be effective. So we did this experiment, and when we treated these young mice this time, we didn't see any significant difference. So this is an age-associated effect, and when we looked at NK cells, NK cells didn't change, and we didn't see a change in some of these exhaustion markers. There was a change in DNAM1, which is an activation marker, so that was quite interesting to us. So in conclusion, hopefully what I've been able to show is that there is these very age-related effects that drive different tumor dynamics across multiple tissues. And that looking at this ARG1 therapy, for example, we're basically able to treat these aged lungs and livers, and we're able to see a significant decrease in their ability to grow out. But this isn't working in the young lung and liver. So we really need to look at different aged environments and different aged metastatic tissues to get proper personalized therapy to different patients. So with that, I just want to thank all my collaborators. I was supposed to take my dog out of that, so sorry, but he is very helpful in the lab. But I want to thank everyone. Dr. Kelly Couton did a lot of this work, along with John Passamonte as well, all of our collaborators at Fox Chase. And then thanks to Dr. George Zhu, who is our kind of collaborator on the human side of NK cells. So yeah, happy to take any questions. Wonderful. Yeah. Yes, I see we have questions already. Very good study. There is a very big redundancy in the system. And how do you feel confident that arginase is really what drive these changes? Been tried many times, different intermediates, and it always end up with the same story, that other factors took over. And the second question is a little bit connected with that. The model which you have, is it model of dormancy or just metastasis? Because I didn't quite get it. How do you establish that you have a dormant tumor cells? Yeah. Great question, Dimitri. So the first one, I do agree there are some functional redundancies with just ARG1. That is definitely one factor. What I didn't have time to show today is that the melanoma cells are also secreting factors like PROS1, and this is really dramatically increasing myeloid cell infiltration as well. So it's not just the ARG1 and the aged lung. I think that's one of the factors that maybe allows these things to grow out efficiently initially. But then I think the melanoma cells take over and start to secrete, not only some of these factors. ARG1 seems to be a factor that they can secrete, but there's a huge amount of different cytokines and chemokines that these particular cells are secreting to inhibit things like CD8 T cells as well. So I didn't have time to show that today. And then, yes, unfortunately, this is a 100% a model of colonization, not dormancy. We have some of that kind of dormancy effect going in the lung. The only way we can really look at it, though, is putting mCherry in, and this does affect the immune system. So we are trying to figure out how we can potentially do this without using fluorescent proteins. But right now what we're trying to create in the liver is, basically, we have a GFP luciferase vector in some of these cells. And when we do these liver colonization assays, we basically take the cells from that liver, and we are reculturing them, and we're hoping to be able to actually get them to disseminate to the liver and look more at that dormancy side. Again, it's probably going to affect the immune studies a bit, but that's probably the best way we're going to do it. And we also have created a vector, which is this dual blue fluorescent protein on top of H2B, and we could basically track whether or not these cells are growing out over time or if they remain dormant, and we have FUCCI with that as well. So I'm happy to discuss that afterwards. But yeah, in terms of the dormancy side, we're trying to get to that stage, but right now we're more on the colonization side of things. Okay. Adriana? Thank you. Very great talk. Very nice studies. I was wondering what happens if you induce even in younger mice some sort of stress-induced aging like smoke for lung or alcohol or liver cirrhosis for the liver. And the other question is, do you know if the natural killer cells, besides changing those markers, IPD-1, make more or less granzyme and perforin, so if they're still lytic? Thank you. Yeah, that's a great question. So the NK side of it, we have done the sequencing, and yeah, granzyme, perforin, all of that is really dramatically decreased with treatment with ARG1. We're currently in the process of trying to get something approved where we can actually get NK cells from healthy and melanoma patients from young and aged, just to see whether that kind of tracks as well. In terms of the stress, we do have some collaborators like Lucia Borrello, who's looking to do smoking in these young mice to see whether or not it induces an age-like phenotype. She's doing some UV studies as well, where she treats some of these skin fibroblasts as well. I definitely think things like that high-fat diet inducing liver cirrhosis would potentially induce an aged phenotype as well. So we're not necessarily doing them right now, but I definitely think that trying to change the microenvironment to have these effects would mimic what we see in the aged and probably drive much more aggressive metastasis in young mice and indeed in young patients as well. So don't smoke and drink. Okay. I have a quick question about... You found the arginine one. What about mitochondrial arginine two? Is there any contribution for the phenotype you see? The problem with that, we were trying to incorporate it. There's just not very good flow antibodies for arginase two, and so we're trying to do that now on more of an IHC level. I think from what I've seen, there's redundancies in ARG1 and two, where if we decrease ARG1, ARG2 might come up. So I think that the actual functional inhibitor inhibiting both does a good job of that. But yeah, we don't have the answer for ARG2 yet, unfortunately, but I'm hoping eventually we can see whether or not it tracks the same or if there's actual differences between that. Yeah. So ARG2 basically is a M2 macrophage phenotype, a marco, right? Mm-hmm. And how are you going to differentiate it? Well, so we were hoping that these ARG1 levels that were changing, we would actually be able to define into one population. We thought polymorphonuclear MDSCs or all these macrophi-- But it just increases across everything. Mm-hmm. So yeah. Again, we haven't been able to look at ARG2 yet, but I kind of agree with age. I think it is pushing macrophages towards M2. I think it's pushing neutrophils towards MDSCs. So I think we are seeing that effect. I think maybe we're going to have to go down the single-cell route or something and really look at some of these populations to differentiate them. Yeah. Thank you. We have time for one last quick question because we're out of time, so be short. Okay. Sorry. I wanted to basically try to put these studies in perspective of genetics somehow. So if you had the chance to look at accumulation of potential mutation from the dormant cell from one point of view, and also if you, I don't know if it exists in mice as well, but if you had chance to think about clonal hematopoiesis in the aged mice and Tet2, DNMT3, which have been associated with ARG1 expression in MDS. So we haven't been able to look at it yet. The problem is we already have the variable of changing young and kind of age. So then adding variability on top with different mouse strains or something like that kind of makes it tough to point out what you want to look at. I'm hoping to potentially discuss with you afterwards how you might be able to do that a bit better. But yeah, we're kind of lacking on that front right now. It's a really great question, and I think genetics obviously does play a huge role. And even in the story that we have, where in very old humans, we see this decrease in incidence, and we're always asking this question, is it because these organs are starting to break down so much that the cancer cells can't grow? Or are these very physically fit genetic humans getting past the age of 80 to 90, and they're just not predisposed to cancer? SoLots of really good questions on that front. We need to figure out how to kind of look at them a bit better. So yeah, great question. Thank you. Maybe we discuss it. Yes. Thanks. Thank you, Mitch. If you have more questions, you can discuss later. Thank you. Okay. The next speaker is me. That's crazy. And I'm going to talk about stromal and immune plasticity in the metastatic microenvironment. I have no disclosures. And the reason that my lab is focused on studying metastasis is because cancer mortality is almost exclusively a result of tumor metastasis. This is data from breast cancer, but this is true for most solid tumors, where mortality rates go up and survival goes dramatically down when tumors are diagnosed after they had already disseminated. And that's why we believe that understanding the biology of metastasis is the key to improving survival. And we're doing this by focusing on the metastatic microenvironment. So I guess if you're here in this session, I don't have to convince you that the microenvironment is important. But a lot of our knowledge on the microenvironment actually comes from studies of primary tumors, and we know a lot less about the metastatic microenvironment, which really affects the growth of disseminated cells. We know that the changes can precede metastasis, that some of them occur early, but to the most cases, this is kind of like a black box, the early events of metastasis. And we're trying to kind of open up this box and understand what is happening in the metastatic microenvironment. And we believe that metastasis and the metastatic microenvironment should be studied in an organ-specific manner. So the cells that are disseminated, they're already cancerous and tumor cells. But to thrive at the metastatic sites, they must adapt to organ-specific microenvironments. In a sense, they kind of need to learn a new language because growing in brain or in bone is not the same as growing in breast or in skin. So they kind of need to learn to now speak lung or speak bone, and this may take time, which may explain also some of the delays that we see in metastatic relapse. And to adapt to organ-specific microenvironments, they need to start developing crosstalk and interactions with different cells than they did in their original organ. And what we also see is that organ-specific microenvironments can instruct common adaptation mechanisms in very different cancer types, which I'll show you today, and my talk today will focus on brain metastasis. And the main questions that we're asking is how is a hospitable microenvironment instigated at organ-specific metastatic niches, and what is the functional role of stromal and immune cells in facilitating metastasis? And our previous studies showed us, if you kind of integrate what we have learned, is that the resident stromal cells of the tissues, such as fibroblasts or osteoblasts in the bone or astrocytes, which are the stars of my talk today in the brain, are actually many times the first responders to changes when cells disseminate, and that they are capable of mediating recruitment of immune cells. They mediate immune suppression response to therapy, and many other things that I'm not going to go into. But understanding these interactions between stromal and immune cells is a key factor in understanding organ-specific metastasis and the reshaping of metastatic microenvironments. And specifically today, I'm going to talk to you about brain metastasis. They're actually a lot more common than primary brain tumors, and they are very deadly with a median survival of less than one year. These are the main cancer types that metastasize to brain, and I'm going to focus on melanoma and breast cancer metastasis to brain. So in the lab many years ago, Hila Doron, who was then a PhD student, established a model of spontaneous melanoma brain metastasis in immune competent mice, where we inject the cancer cells intradermally, and then resect the primary tumor and wait for spontaneous brain metastasis, which takes up to six months. And not all the mice get macro metastasis, which is why in parallel, we're also working with experimental brain metastasis where either melanoma or breast cancer cells are injected intracardially, and then they form brain metastasis. And we have created by in vivo selection, brain tropic variants of melanoma and breast cancer cells, which is what I'm going to tell you about today. And what we found is that the brain tropism of melanoma cells is mediated partially by hijacking astrocyte-derived inflammatory signaling. We saw that astrocytes become activated even before melanoma cells get to the brain using the spontaneous brain metastasis model, and that they start secreting CXCL10 that then recruits melanoma cells that express their receptor, the CXCL3 receptor of melanoma cells. So we wanted to know, since this seemed to be systemic signaling, what are these melanoma-derived soluble factors that activate astrocytes even before melanoma cells reach the brain? And what we found, I'm cutting a long story short, but we focused on a secreted protein called lipocalin 2, or in short, LCN2. It was-Highly upregulated in the blood of mice with brain metastasis. And the reason we decided to focus on it, even though, as you can see here, it was not the only secreted factors that was upregulated, not even the top one, is because it was shown to be an activator of astrocytes, and part of the innate immune response to bacteria, actually, by sequestering iron. And it was shown to be important in brain pathologies and in cancer, but there was not much known about it in brain metastasis. And so using, again, these two models of brain metastasis, melanoma and breast, what we found is that LCN2 is systemically upregulated in the blood and also in the CSF of mice, and the levels of LCN2 correlated with brain metastasis, suggesting that it is functionally important. And we always try to validate our findings in human patients just to see that we don't get lost in our mouse models. And what we found is that the blood levels of LCN2 were higher in melanoma patients with metastatic disease. It was correlated with worse survival, and even more interestingly, it was correlated with those patients that eventually developed brain metastasis, but not metastasis in other organs, again, showing how there is organ-specific effects. And when we looked at the levels of LCN2 in the blood of patients with brain metastasis, again, it was higher in their blood. And when we did this longitudinal analysis of patients with brain metastasis, we saw that whenever the levels of LCN2 in the blood rose, this was the last blood test because shortly after the patients passed away. And you can see here the correlation with survival. And we also had a cohort of lung cancer patients. And again, levels of LCN2 were associated with poor survival in these patients. So we went to do a functional experiment in mice, and what we saw was that when we injected melanoma cells to LCN2 knockout mice, they had significantly less metastasis. You can see here the quantification of the metastasis with the MRI. So imaging, and also because the cells are labeled with mCherry, we can also quantify their levels using FACS analysis of mCherry. And this shows us that LCN2 is really functionally important for the formation of brain metastasis. And looking at the mechanism, what we saw is that LCN2 was important for pro-inflammatory activation of astrocytes. Because in LCN2 knockout mice, not only did they have less metastasis, but also their astrocytes, you can see here on the right-hand side, expressed a lot less pro-inflammatory factors. And these factors were important for recruitment of bone marrow-derived granulocytes and monocytes to the brain because in LCN2 knockout mice, this recruitment was significantly inhibited. And we looked at what was the main source of LCN2 by sorting different cells, and we found, both in melanoma and in breast, that granulocytes were actually a main source of LCN2 secretion. And looking at a human data set here too, neutrophils, so this is a data set from Joanna Joyce's lab. Neutrophils were a major source of LCN2. So to see whether these cells were functionally important for brain metastasis, we did bone marrow transplantation from LCN2 knockout mice to wild-type mice, so now these mice do not have LCN2 in their bone marrow-derived cells and in their granulocytes, compared, of course, with wild-type bone marrow. And what we found is that the astrocytes in the brains of these mice looked like wild-type astrocytes. So LCN2 from bone marrow-derived granulocytes was essential for pro-inflammatory activation of astrocytes for the recruitment of monocytes and granulocytes, and also for brain metastasis. Because what we found is that mice that were transplanted with LCN2 knockout mice bone marrow had significantly less brain metastasis. So basically, just to summarize this part of my talk, what we found is that crosstalk between bone marrow-derived neutrophils and brain-resident astrocytes was important in brain metastasis, in facilitating brain metastasis, and this is mediated by granulocyte-derived LCN2. And in the last minutes I have in the second part of my talk, I want to quickly take you through this very exciting, new, unpublished project in the lab that was done by Yael Zeit, a PhD student in the lab. So because of our interest in astrocytes, we did single-cell analysis of astrocytes and also T cells, which I'll get to later, that we isolated from mice with brain metastasis. And as expected, we found that astrocytes are heterogeneous, that we found many different subpopulations. And one subpopulation that really attracted our attention were these neuroprotective astrocytes.And the reason that we found this so exciting is because these cells expressed genes that have to do with MHC class II complex. So MHC class II is usually expressed on professional antigen-presenting cells, which astrocytes are not known to be. So we thought this is very exciting and new, and we wanted to see whether this is indeed the case. Can astrocytes function like professional antigen-presenting cells? And so it's not enough to kind of express MHC class II, you also have to express costimulatory molecules. And we checked and validated at the protein level, and what we found is, again, in different models of brain metastasis, there was actually a subpopulation of metastasis-associated astrocytes that upregulated MHC class II expression as well as the costimulatory molecules. We also found that interferon gamma can upregulate the expression of MHC class II on normal astrocytes when we do this in vitro. And interferon gamma is highly upregulated in brains with metastasis because of the neuroinflammation that is associated with brain metastasis. And so, we kind of went to a lot of immunologist friends. We consulted with immunologists. If we had to convince you that these cells are professional antigen-presenting cells, what would we have to do? And we did everything they advised, and I'm really not showing you everything, just kind of bottom lines. We showed using DQ ovalbumin, which is a fluorescently silent probe that becomes fluorescent upon proteolytic degradation. We showed that astrocytes that are induced by interferon gamma, so the AP astrocytes, can actually uptake and process antigens. But they also need to, in order to be professional antigen-presenting cells, they need to be able to uptake, process, and present to T cells. So we also did a series of experiments, again, showing you just a little bit of it, so I can show you the highlights. We showed that AP astrocytes induced with interferon gamma can activate CD4+ T cells that they are co-cultured with, and these CD4+ T cells were isolated from vaccinated mice that were vaccinated with tumor cell lysate. So they can present antigen to CD4 T cells in a tumor antigen context. And we saw this by proliferation of T cells, expression of CD44, and also secretion of interferon gamma by these activated T cells. But can they actually interact? So we used image stream analysis to see whether astrocytes that are isolated from brain metastasis can be physically associated with T cells. And what we found is that, yes, AP astrocytes can be physically associated with T cells. And then we looked in patient samples from melanoma, breast cancer, and lung cancer, and to see whether we can kind of catch them in action. And we did. So this is the quantification. What we found is that AP astrocytes, but not astrocytes that are not expressing MHC class II, like we said, this is just a subpopulation, are more likely to be found in contact with T cells in brain metastasis. And so because we, like I told you, also did single-cell analysis of T cells, again, finding, I'm not going into this, but finding many subpopulations of T cells, we looked at the predicted interactions of T cells with AP astrocytes, and we found that these neuroprotective astrocytes, which are the population that contains the AP astrocytes, are most likely to form many interactions with T cells. And so we wanted to see whether this would have any functional consequences for tumor cell killing. So we did this ex vivo assay, again, with T cells isolated from vaccinated mice with breast cancer cells, and we co-cultured them with AP astrocytes or with normal astrocytes, CD4 T cells that I showed you are activated by astrocytes, and CD8 cells, and we looked at tumor cell killing. And what we found is that CD8 T cells that were cultured with AP astrocytes enhanced their secretion of granzyme B, interferon gamma, and TNF alpha, which shows that they are activated. And when we culture them then with cancer cells, this really significantly enhanced the killing of cancer cells by these AP astrocyte-activated T cells. But until now, everything I showed you, all these ex vivo assays, were with astrocytes that we induced by interferon gamma. What happens if we isolate astrocytes from end-stage metastasis? And there the plot thickened. What we found is that when we isolate astrocytes from late-stage metastasis, they were actually immunosuppressive. They suppressed the proliferation of T cells, and they started expressing high levels of checkpoint ligands such as PD-L1 and CD155, suggesting that something is suppressingTheir function as AP astrocytes. And indeed, when we looked at the correlation between MHC expression and brain metastatic burden, we found that the expression of MHC was correlated with low brain metastatic burden. So something is happening at later stages. I'm not going to go into what is happening. I'll just tell you that it looks like the secreted factors from tumor cells are suppressing the antigen presentation by astrocytes. So we wanted to see the bottom line, and these would be my final pieces of data. What happens if we manipulate AP astrocytes? Are they good? Are they bad? Do they enhance or inhibit brain metastasis? So for that, we created mice that had MHC II knockout specifically in astrocytes by crossing GFAP-Cre mice with MHC Lox mice, and then we injected these mice with cancer cells and looked at their metastasis. And what we found is that AP astrocytes actually inhibit brain metastasis, because in these mice where we knocked out MHC class II specifically in astrocytes, so we inhibited the antigen presentation by astrocytes, they now had a lot more brain metastasis. Suggesting, again, that AP astrocytes are important. So just to summarize what I've shown you, that antigen-presenting astrocytes can orchestrate antitumor immunity in brain metastasis. Their expression of MHC class II is induced by interferon gamma, which is a hallmark of neuroinflammation in the brain, and they activate T cells, enhance tumor cell killing, but later, as metastasis prevails, tumor secreted factor, I didn't show you that part for lack of time, inhibit the expression of MHC class II, causing immune suppression and progression of metastasis. So I want to thank all the members of my lab and our collaborators and, of course, funding and Yael is really-- So the crown moved, but Yael is really the queen of this project. And one very last thing that I want to keep on while we have a discussion is I want to encourage you to attend the MRS meeting, the Metastasis Research Society that will be on June 28th in Utah. It's going to be an amazing meeting. There's still early bird registration, and I encourage you to attend, and I'm happy to take any questions. I have a couple of quick question. First one, I may be missing. When you knock down lipocalin 2 and show very nice and strong effect on metastasis and survival, what happens with primary tumor? So we knocked out lipocalin 2 using the experimental metastasis model. So the data I showed was when cells were injected intracardially, but we also did this with the primary tumor, and there was no effect on primary tumor growth. It's actually published, so it's in the paper. No effect on primary tumor growth. Okay. Very interesting. It was a specific mechanism for brain metastasis. And just related to that, so effect on astrocytes, it's clear. What happens with their counterpart in other tissues like fibroblast on effect on macrophages? Do you see something different? It's only this astrocyte-specific effect of lipocalin, or it's little bit more than that? So it is a little bit more than that. Lipocalin is actually secreted also by fibroblasts in the primary tumor. It has different effects on organ-specific metastasis. We focused on brain metastasis, but there are other papers showing that lipocalin 2 has effects also in other organs, just different effects. Yes, and that effect is anti-tumor? The other papers. No. No? It's always pro-tumorogenic. It's pro-inflammatory, yeah. Okay, thanks. Okay, Sergey. Yeah, very nice talk. About astrocytes early and astrocytes late, did you look at their interactions with Tregs among other CD4 T cells? Yes, we have, and they do interact with Tregs, but I don't think we compared early and late. So we did this only in late metastasis. Yes. Thanks for the presentation. My question is, precedes the metastatic organ or tissue, the niche is not solved yet. But for the research purpose and the clinical study, any good idea or suggestion in terms of how to choose the organ? What timeline is the best to figure out those changes of the niche? Mm. A lot of resources, energies were put on it. So any idea about it? Yes. So when working with mouse models, it's an important question. Thank you for that. It needs to be carefully calibrated to find what are the important time points and how to define early and late.And this changes with the cancer type, the organ, it really changes in each system, and it's important to calibrate it in each study before you call this early and this late. But it's very important to do that. It's just very specific to tumor type and metastasis site. Thanks. Yes. So it was just about the LMC2 buildup. Is there an actual buildup in the primary tumor before you reach a threshold that's able to activate these astrocytes in the brain? Or does the tumor have to grow to a certain size before you're actually getting the astrocyte activation, or is it something else? Yeah. So it is secreted from the primary tumor, and it's gradually increasing in the blood. And at the time when we resect the primary tumor, we see a decrease in the blood in LCN2, but the astrocytes are already activated, and then the main source is actually granulocytes that are recruited to the brain, and the blood levels go up again. So I guess the ask is, is there anything changing in the tumor to increase this, or as soon as the tumor starts, the primary tumor, you're already getting that LCN2 into the blood? Yes. Okay. And it gradually increases as the tumor grows. Thanks. Yes. Thank you for the talk. Have you looked at the blood-brain barrier? How does it react? How T cells and tumor cell, how do they get in there? Yeah. So this is not our studies, but many people have looked, and one of the hallmarks of brain metastasis and of neuroinflammation is breakdown of the blood-brain barrier that allows recruitment of granulocytes and T cells from the periphery to brain metastasis. So this is something that is known, and we also saw it in our system. Thank you. Thank you for the very interesting talk. I was just wondering, have you compared this antigen-presenting ability of AP astrocyte with the professional antigen-presenting cells? I was just wondering why they are coming and- Are you reviewer number two? I'm still a PhD student, so no. Yeah. So it's a cool question, and we're doing it now because we were asked by one of the reviewers to do that. Okay. Thank you. Welcome. I want to ask a question. I know the time is up. Okay. But you are the chair, so you can- Yeah ... choose to continue. One last, last question. So new role for astrocytes, APC astrocytes. So do we have dendritic cells in the brain? And if yes, what do they do? There are dendritic cells in the brain when there is brain metastasis. I'm guessing that they do the same, and this comparison is exactly what we're trying to do now. Mm-hmm. It's a really good question. Thank you. Okay. So thank you all for listening and for a great discussion, and thank you for our speakers, and enjoy the rest of your day.