2026-04-20-pm-late-microenvironmental-determinants-tumor-evolution-immune-escape

Date: 2026-04-20 Primary topic: single-cell-spatial-omics Cross-referenced from: single-cell-spatial-omics Word count: ~14,311

Auto-generated captions

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

Okay, I think it's time to get started. Thank you very much everyone for coming. And this is a session about microenvironmental determinants of tumor evolution and immune escape, another component of tumor microenvironment. And we will be your chairs for today, and we will have seven exciting talks in the program. And we will get started right now. Remember that please actively contribute to the discussions and questions, and our only requirements for speakers, try to stay on time. It's a 10-minute talk and five minutes questions. You will see the timer on the screen. When it's 10 minutes, it will be yellow color and your talk is supposed to end. When you run out of 15 minutes, it will be a red color and no more time for questions, unfortunately. And without further ado, I think we will have our first speaker, Yi Zheng from Columbia University, and he will talk about neuroinflammatory memory in gastric tumorgenesis. Thanks for the introduction. Good afternoon, everyone. I'm Yi from Columbia University. Today, I will talk about vagal sensory neuron inflammatory memory. We all know that H. pylori eradication reduces the risk of gastric cancer. However, a large RCT with up to 26 years of follow-up revealed important limitation. Even after a successful eradication, gastric cancer still occurred in about 2.6% of treated individual. A risk that remains potentially higher than that of general population. Interestingly, patients with dyspeptic symptoms derive little to no benefit from eradication therapy. Together, these findings suggest that eliminating the pathogen alone is not sufficient. Instead, they raise the possibility that a form of persistent inflammatory memory, potentially involving the gut-brain axis, remain after bacterial clearance. So before going further, let me briefly define what do we mean by inflammatory memory. Inflammatory memory can be defined as tissues or cells retaining a memory of inflammatory injury that acquisition to react more quickly and broadly to subsequent diverse secondary stimuli. In the pancreas and Crohn's, epithelial memory of inflammation limits tissue damage in subsequent episodes of acute inflammation while promoting tumorigenesis. So this brings us back to a central question. Can neuron themselves store inflammatory memory? For example, in a DSS colitis model, reactivation of insular cortex neural activity during primary inflammation is sufficient to induce secondary Crohn's inflammation. And these findings indicate that the nervous system can encode and reactivate information about prior inflammatory states. However, while a similar neural memory exists in the stomach and how it might influence long-term disease outcomes remains unknown. To study inflammatory memory in a physiologically relevant manner, we employed a set of complementary gastric regeneration and tumor model. For gastric regeneration, we primarily use a high-dose tamoxifen model, which induces parietal cell loss and chief cell loss, and triggers PEN, a kind of metaplastic response, and normally resolve back to histological baseline within three weeks. Using the high-dose tamoxifen model, we observed a striking memory phenotype in the stomach. We use immunofluorescence for parietal cell, chief cell, and spin cell, and proliferation. Mice that had experienced a single prior injury and fully returned to histological baseline displayed a markedly accelerated regenerative and metaplastic response upon a second injury. Importantly, the enhanced recall response was not transient. The memory effect persists for at least six months after the initial injury. A similar phenotype could be observed in a clinically relevant manner. In mice previously infected with H. pylori and subsequently eradicated, we induced the HDD injury. Although gastric histology appeared comparable to control before injury, mice with an inflammatory history showed significant accelerated regeneration and metaplastic response during recall with increased parietal cell and metaplastic cell. However, this accelerated recall response comes at a cost.While inflammatory memory enhances regeneration, it has deleterious consequences in the context of cancer. The memory effect accelerated orthotopic ACKP tumor progression, and a similar phenotype can be observed in multiple spontaneous cancer models, including the intestinal type and diffuse type. These findings indicate that the memory, although beneficial for tissue repair, creates a permissive microenvironment for tumor progression. So next, we sought to link the memory phenotype to the sensory nerve system. Of course, both regeneration and tumor model recall injury was consistently associated with an increased CGRP positive density fibers at the lesion site. In parallel, we observed a selective CGRP increase in the Trpv1 positive sensory neuron within the nodose ganglion, while neurons of dorsal root ganglion origin were not comparably engaged. To directly test whether vagal sensory neurons are required for the memory phenotype, we selectively transected the anterior vagal trunk, denervating the anterior gastric wall. Compared to the innervated posterior wall, the denervated region completely lost the accelerated regeneration and tumor-promoting memory effects. So whether this memory could be traced at the level of individual neurons using a force-dependent head-off laboring strategy, we permanently marked the neurons activated during the first injury and using tocilazole to stop laboring after the first injury. And then by using pERK as a marker of neural activation during recall, we found that reinjury activated a significantly larger number of nodose ganglion neurons than the initial injury, with a substantial overlapping with the neurons labeled during the first injury. To determine whether memory resides in a specific subset of neurons, we selectively ablated the neurons activated during the first injury using a force-dependent head-off strategy. In this setting, recall injury failed to induce regeneration and tumor progression, demonstrating that the memory-encoded nodose ganglion neurons are necessary for the recall. Finally, transportation of memory-encoded nodose ganglion into Trpv1 positive neuron deficient recipients was sufficient to restore the regenerative and tumor-promoting memory, indicating that this neuron persists over time and carry transferable inflammatory memory. So next key question comes whether this memory requires immune input, and if so, which immune compartment is involved. Using Rec1 deficient mice and Brg mice, which means Rec deficient but also deficient in IL-2 receptor gamma chain, we found that the memory phenotype is independent of adaptive immunity, but strictly depends on the innate immune system. Through systemic profiling, we identified gastric ILC2 as the only immune population capable of supporting the memory phenotype. During memory establishment, IL-25-positive ILC2 remained persistently enriched and was spatially positioned near the sensory fiber endings within the grand isthmus region and the tumor region. To directly test whether ILC2 are required for the memory formation, we selectively ablated ILC2 using the IL-5 Cre-IDT mice. Loss of ILC2 completely abolished the memory-associated phenotype. Conversely, transplantation of HDT-activated ILC2 into Brg mice was sufficient to restore inflammatory memory. Similarly, conditioned media from activated ILC2 was able to prime nodose ganglion in vitro and even with enhances the capacity to promote epithelial organoid growth. So this led us to ask a final mechanistic question: How do ILC2 write memory into sensory neurons? Using CGRP ELISA and calcium imaging of the nodose ganglion, we found that ILC2-derived IL-5 stimulates CGRP release from sensory neurons, and this effect is markedly amplified in memory-encoded neurons. This heightened responsiveness reflects the fact that the memory neurons express elevated levels of type II cytokine receptors. Interestingly, while a knockdown of IL-5 receptor only partially weakened the memory phenotype, only disruption of IL-13 receptor completely abolished the inflammatory memory. This result revealed IL-13 is the dominant signal that drives and stabilizes inflammatory memory with the sensory neuron.So, here are our memories for our research. First, we found that the sensory neuron encoded a long-term inflammatory memory in the stomach, and the neuronal inflammatory memory drives metaplasia and tumorigenesis upon a recurrent injury. A memory recall enhances CGRP release and gastric progenitor cell activation. ILC2 signaling exacerbates immuno-neural epithelial memory circuit. And finally, we found IL-13 epigenetically imprints and maintains neuronal memory programs. And thanks for listening. Please feel free to raise any question, and I'm glad to answer. Thanks. Questions, please. Maybe while we're waiting for more questions, I can't get over the novelty of an ILC2-dependent memory of nerve injury and injury to epithelium. What's going on with ILC2s? Do you think it's just the cytokine production that's somehow mediating this? Thanks. That's a great question. Because of the time limitation, we didn't show all the mechanism in our slide here, but I can provide the detail. After the ILC2 activation into the sensory neuron, because ILC2 release some kind of cytokine such as IL-5, IL-13, and IL-5A. And among them, IL-13 will cause some epigenetic change in the sensory neuron. We have confirmed that some obvious epigenetic changes such as H3K4 trimethylation and found some key regulators such as specific methyltransferases that will induce chromatin accessibility during the memory period. That's the mechanism basis of the memory-encoded neuron. Thanks. And another question I have with regard to sensory neurons, what are the ramifications of eating a lot of capsaicin in this model, and depleting sensory neurons in humans or in mice? There will be then less inflammation and no inflammatory memory? Thanks. That's a really interesting point. And we define the inflammation memory circuit as needed of the immune input. And so for the human, we didn't examine the sample about the inflammation. But we used a very similar clinical relevant manner, that is H. pylori infection and eradication. That's the same process as the human disease. And this model also confirmed that the memory from the sensory neuron are so previously related with the epithelial phenotype. Thanks. Thank you. Over there. Yeah. I think there's a small literature on TRPV1 positive vagal signaling in the inflammation that can contribute to pancreatic disease, in particular, diabetes and other conditions in the GI. Could this be a general inflammatory tract for GI tumorigenesis in the interface with ILC2, which is, as pointed out, very interesting? Thanks for your question. It's also an interesting point. We didn't check PD1 and PD-L1 that associated with the sensory neuron signaling. And for the immune input, we have some systemic profiling to identify the role of ILC2. Only after ablation of ILC2 that can induce the loss of the memory phenotype. So, for other kinds of immune cells, we can make sure that there are loss of the memory. Thanks. Okay. Thank you so much. Thank you. Our next presenter is Dr. Naoko Ohtani from Osaka Metropolitan University, and she's going to be talking about T cell exhaustion after steatosis in the tumor microenvironment. Thank you, Dr. Ohtani. Thank you very much for the kind introduction. My name is Naoko Ohtani. And first of all, I would like to thank organizers for giving me the opportunity to talk about the work here. Today, I'm going to talk about the mechanism of MASLD-associated hepatocellular carcinogenesis, particularly focusing on T cell exhaustion in steatotic tumor microenvironment. So I have no COI to disclose. Okay. Thank you very much. Over the past several decades, obesity has become more prevalent in most developed countries, including recognizing some major risk factor of several type of cancers, including liver cancer. However, the molecular mechanism of how obesity promote cancer have not been fully illustrated yet. So, we are particularly focusing on the MASLD-associated HCC, a typical obesity-associated HCC. MASLD is a metabolic dysfunction associated steatotic liver disease, so-called fatty liver.Around 25% to 30% of the global adult population, approximately 1.5 billion people, are estimated to have MASLD. And hepatocellular carcinoma arising from MASLD has been increasing worldwide, particularly in East Asia, Oceania, and Europe. However, unfortunately, ICF therapy, both monotherapy and combination therapy, are not effective in MASLD-associated HCC. And monotherapy even worsens the patients' survivals. So we questioned why MASLD-associated liver cancer is resistant to ICF therapy. So to this end, we first set up the system to investigate MASLD-associated liver cancer. Neonatal mice are treated with DMBA, a chemical carcinogen, which frequently initiates oncogenic mutation in the Ras gene, followed by normal diet or high-fat diet for 30 weeks. And very interestingly, almost all the high-fat diet mice developed hepatocellular carcinoma like this. And we are using this model as a MASLD-associated HCC model. Using this model, we previously found that the long-term high-fat diet increased gram-positive bacteria in the intestine, and deoxycholic acid produced by this gram-positive bacteria, deoxycholic acid, circulate through the intrahepatic circulation to provoke the serosinesis of hepatic stellate cells, fibroblast cells in the liver tissues. And moreover, this gram-positive bacteria also provided lipoteichoic acid, which is a cell wall component of the gram-positive bacteria, into the liver to stimulate the inflammatory signal through the toll-like receptor 2 signal to up-regulate the serum susp gene expression as well as Cox 2 expression to overproduce prostaglandin E2 to suppress antitumor immunity. In addition, we found that the caspase-1 and caspase-11 activity is high in the steatotic liver tissue, and this caspase activity created a gastamin D cleavage to create the small pore in the cell membrane to create the gastamin D internal mediated pore. Through this pore, several suspectors such as IL-beta and IL-33 exported to suppress antitumor immunity. And next we wanted to search for biomarkers for the MASLD-associated HCC. To this end, we have done the multiplex assay using the patient's plasma, high-fat diet fed mouse serum, and liver tissues. And we found that the CXCL9 chemokine was commonly high in these samples. And we found that CXCL9 is highly expressed in the LSEC, liver sinusoidal endothelial cells. And as you know, LSECs are located in the sinusoidal area, like this. And also we found that the microbial components such as LPS or LTa, as well as the bacteria membrane vesicles isolated from the MASLD patient's feces, were able to up-regulate the CXCL9 in the cultured LSECs. And so CXCL9 is a chemokine that can recruit the T cells from the lymph node into the liver tissue and through the CXCR3 receptor. So to block this axis, we treated mice with the neutralizing antibody for the CXCR3, and we found that this antibody treatment significantly reduced that liver cancer development, like this. Moreover, we found that the CXCL9 knockout mice as well as LSEC-specific conditional CXCL9 knockout mice, showed a significant reduced liver cancer development, like this. And when we investigate the immune status, we found that the ratio of exhausted T cells decreased and the activated CD8 T cells increased. However, the Treg status is unchanged in the condition of the CXCR3 antibody-treated status. So suggesting that the CXCL9 is important to develop the liver cancer. Next, we wanted to search for which cells are important for the liver cancer development, and we have done the single-cell T cell receptor analysis using the CD8 T cells isolated from the lymph node as well as the liver tissues. And we found that in the lymph nodes, TPEX, the progenitor-exhausted T cells existed in the lymph node, and showing some characteristic of TPEX, showing the exhausted markers as well as the memory markers. However, the same clone became completely exhausted when the cells entered into liver tissue, like this. And we have done the trajectory analysis, and this trajectory analysis showed a similar result, that the same clone from the lymph node, starting from the central memory T cells, became TPEX and losing the memory markers, and entered into the completely exhausted TECs, combining with increasing expression of exhausted markers.And in this analysis, we found that the TEpCs and as well as TEcs T cells highly express the granzyme K. So we are interested in the granzyme K, and this is one of the granzyme family. However, they are very different from the granzyme B, and this granzyme K has a caspase-1-like activity to mature the iron beta like this. And also we confirmed that the granzyme K is really correlated with the PD-1 expression. And as you know, caspase-1 is able to create the gasdermin D to create the immature domain to create the small pore in the cell membrane. And we previously found that through this gasdermin D cell membrane pore, that some suspect factors such as iron beta are exported from the hepatic stellate cells. So we treated the hepatic cells, isolated primary hepatic cells with the recombinant granzyme K, and we found that the iron beta expression is highly upregulated by the treatment of recombinant granzyme K. And also interestingly, we found that the iron beta is exported into the culture supernatant in this condition, suggesting that granzyme K facilitates iron beta upregulation as well as its export. And next we have created a granzyme K knock-out mice, and we found that the granzyme K knock-out mice shows the significantly reduced number of its development. However, the granzyme K overexpression increased the liver cancer development, suggesting that the granzyme K is involved in the mass 30 associated HCC progression. And next we used the human samples, and we isolated CD8 T cells using the human mass 30 associated HCC patient tissue. And we have done the single cell T cell sequencing analysis. And in these human cells, we found a similar trajectory pathway from the central memory T cells defects into the exhausted T cells. And similar to mouse model, TEpCs, TEcs exhibited the high expression granzyme K as well. And, finally, we have done the metabolome analysis, particularly focusing on the oxidized lipid oxylipin, and we found that a series of oxylipins are high in the high fat diet fed mouse liver. And a series of oxylipins are able to develop from naive T cells into the TEpCs like this. And also this oxylipin E, sorry, I'm not showing the real name, but oxylipin E was able to develop exhausted T cells from the TEpCs. So we have done this experiment because the steatotic liver is highly lipid rich, so we thought that the oxylipins may be involved in. So, this is a summary slide, and CXCL3 positive CD8 T cells are recruited from the lymph nodes by the CXCL9 chemokine produced by the LFX. And this TEpCs became exhausted in the steatotic liver condition and expressed the granzyme K. And the granzyme K facilitates the iron beta export from the hepatic stellate cells to facilitate the liver cancer development. And we first questioned why is the ICL therapy is, why mass 30 HCC is resistant to the ICL therapy. So maybe one of the answers is maybe granzyme K upregulation in TEpCs and TEcs in mass 30 HCC is involved in the ineffectiveness of ICL therapy. So this is the end of my talk, and thank you very much for your attention. Great talk. I have to ask whether you feel that this is a reversible T cell exhaustion. Can this be reversed in terms of dietary changes or any drugs that might be available yet? Yes. We haven't checked that point yet, but a TEpCs is thought to be reversible, maybe to some extent. So it is worth challenging, focusing the TEpCs to reverse into the active T cells. Yeah. Granzyme K is highly expressed in NK cells. Does any of your work address the potential role of NK here? We haven't checked the NK cells, but thank you very much for important information. We will check that. Thank you very much. And I had another question. So there were earlier publications that in NASH HCC, CD8 cells may be good or bad, and they may be bad by killing lipid-loaded hepatocytes and inducing compensatory proliferation. Is it possible that actually some of the function of CD8 is to kill hepatocytes and then to become exhausted- Yeah. -in the process? Okay, thank you very much. Important question, and I think that's possible because the chronic liver hepatitis background, so there's some CD8, active CD8 is involved in the hepatitis itself. So maybe it's possible, and we have to carefully investigate that. Thank you very much. I have one more question. Do you think that the factors driving T cell exhaustion in this model are systemic, or are they localized to the liver? Yeah. Actually, we are thinking that that should be local, and so we investigated a little bit metabolome analysis and found out some slight importance. But basically, there's a obesity problem, probably systemically. So maybe some other important factor circulating in the body should exist. And also in this model, can you estimate how much IL1 beta actually comes from CD8 T cells and the rest probably comes from myeloid cells, but maybe context is important that CD8s are interacting exactly with the cellular targets of IL1 beta, and that's why? Right. You are asking the IL1 beta origin itself. Yeah, that's important. And actually, at the same time, we were doing the spatial transcriptomic analysis. So we would like to check carefully the neighboring cells expressing IL1 beta. Thank you very much. Thank you. If there are no more questions, thank you so much. Thank you very much. Our next speaker will be Shuyang He from WashU, St. Louis, and she will talk about stress remodeling of enteric neural system in colitis to cancer transition. All right. Thank you for the introduction. Is the sound okay? All right. So cancer patients are undergoing all different kind of stresses from being diagnosed to all through all these treatments, and also worry about the collapse. So it is of note that one-third of the cancer patients will develop a mental health problem, and half of them will experience anxiety or depression. From large scale of epidemiological studies, we know that stress-related psychosocial factors associated with higher incidence and poorer survival in cancer patients. Social support, on the other hand, which can release the stress from the cancer patients, has been proved with longer survival in these patients. So as a systemic effect, it means that stress can affect our whole body. So my lab is of particularly interested in the gut system now. So it is not only because the gut is closely related to our daily life, but also because colorectal cancer is ranking number three in the United States, meaning a lot of people are stressed out about this. So although there are many evidence showing that genetic mutations like KRAS or APC are driving the colorectal cancer onset, there are more and more studies showing that these bad lifestyle factors are also associated with the early onset of colorectal cancer, including mental stress. So to further understand the relationship between stress and the colorectal cancer onset, we first looked into the published database from the UK Biobank, which has a huge amount of population data. For example, as shown here, when comparing these patients with psychiatric disorders, the risk of them getting colorectal cancer is much higher. Moreover, when grouping these patients into different disorders like depression or anxiety or psychological stress, the risk factors for them for the CRC incidence is also higher than the healthy controls. So to further understand how stress promotes the colorectal cancer onset, we went back to the mouse models in the lab. From the past several years, we and others have used different mouse models to study how stress promotes cancer progression. And we have identified that stress by activation of the HPA axis through the release of the stress hormone like glucocorticoids, where stress can activate the sympathetic nervous system in the mice. So to really mimic the stress response in the mouse, we're now using a physical restraint model to mimic stress. So as shown here, the physical restraint model can lead to much higher levels of corticosterone, which means that these mice are under stress. So with this model being established, we next combine it with the colorectal cancer model. So we first started with the Apc min mouse model, in which all the mice carry a loss-of-function mutation of the Apc gene, and this is also the first step for CRC onset in most of the cancer patients. As you can see here within this model, delayed physical restraint stress can significantly promote tumor genesis, and these mice have developed more numbers of the tumors. So this indicates that under the same genetic background, being stressed can promote the tumor initiation. So this phenomenon was not only limited in this Apc min mouse model, but can also be found in multiple mouse models, including the inducible CRC model and the spontaneous model. So in both cases, we have shown that daily stress can promote more numbers of the tumor happening throughout the gut, meaning that the stress can really promote the cancer onset. So before getting into any mechanistic studies, I would like to highlight here that gut is a highly innervated organ. So here I'm showing you a 3D whole mount imaging from this cleared tissue. You can see here the gut has extensive blood vessels, which are mediating the nutrient absorption. But around these blood vessels here is enteric nervous system. It includes the enteric glia cells labeled in red and the nerves stained in white.So these two types of the cells, the nerves plus the glia cells, has made this neuronal network within the gut, and they are playing critical roles in maintaining the gut homeostasis. So within this nervous system, we are focusing now on the role of glia cells, given its known role during the CRC development. So we first generated a gene signature expressed exclusively by the enteric glia cells, or the EGCs. When we applied this gene signature into the TCGA database for CRC patients, I hope you can appreciate that the patients who had the higher expression of this gene signature, meaning that their tumors probably are more infiltrated by glia cells, they actually had much shorter survival time. So to further understand or characterize the roles of EGCs during the CRC development, we next compared the EGC infiltration across different stages during the CRC development. However, to our surprise, we found that there are a lot of glia cells infiltrated into the polyps or the benign tumors at the initial stage. As you can see here, the glia cells are labeled in green with as 100 beta staining, and they're hanging around the cancer cells in red. However, when the tumors grow larger or become more advanced, we could barely find any glia cells inside. So this data has prompted us to think about the role of EGC in the initial stage. So we next decided to deplete them. We took advantage of a mouse strain called Plp1creERT RosatTTR mice, in which we could specifically deplete the glia cells in the gut following the injection of tamoxifen and diphtheria toxin. So, as confirmed here, we show that the glia cells can be largely depleted in the gut. So we then back cross this mouse strain within the ApcMin background and subjected all these mice into this stress-induced CRC model. So I think the data is pretty striking by just looking at this picture. So simply by depleting the glia cells at the very early stage could significantly reduce stress-induced tumorigenesis. So after identifying that the glia cells are in the tumor microenvironment to promote tumor initiation, we were then wondering, how do these glia cells promote tumorigenesis? So we then turn back to our in vitro culturing system. So we generated the organoid line from this ApcMin mice. We call it as a pre-tumor organoid lines. And at the same time, we also collected the conditioned medium from the stressed glia cells, meaning that we pre-treated the cells with the glucocorticoid or the GR receptor agonist. So what we found is that when treating this pre-cancer organoid with the conditioned medium from stressed glia cells, can significantly promote the outgrowth of these organoids. So to better understand at the molecular level, how do these glia cells promote tumorigenesis, we next perform the qPCR on this pre-cancer organoid. As you can appreciate here, all these tumorigenesis-related genes have been all upregulated when treating these organoids with the conditioned medium from stressed EGCs. So to further understand any clinical relevance of our study, we next compared the cross-species of our data. As you can see here, what we have found for these transcriptional changes in the mouse organoid can be very well recapitulated with a human database. So after knowing that the EGCs are in the microenvironment playing some role in promoting tumorigenesis, we next are wondering, is there any cell intrinsic response to stress? So to study this, we look at the early time point in our model. So getting inspired by previous studies showing that stress can induce reactive oxygen species, or ROS, in the gut, we next examined this oxidative damage within our system. As you can see here, with two different marker staining, we show that the gut tissue from the stressed mice do have elevated ROS, indicating that they are undergoing oxidative damage. So consequently of this oxidative damage, we also found that the gut cells from the stressed mice do show more DNA double-strand damage, as indicated by gamma-H2AX staining. But it's very interestingly that, although under all different kinds of damage, these gut cells from the stressed mice are also hyperproliferating. So here, as indicated by the Ki-67 staining, these abnormal damaged cells are hyperproliferating, and maybe this is the reason that they will undergo tumorigenesis. So in summary, our data has generated a stressed EGC-stem cell axis, and by activating this axis can trigger the tumor initiation in the mice. Magnetically, we found that the stressed glia cells, meaning that the activation of the GR pathways in these glia cells, are the driver in the gut microenvironment to trigger all these transcriptional changes in the stem cells, and finally becoming a tumor cell. So within this project, we also have some open questions for the future direction. First, the glia cells are really heterogeneous. So from recently published single-cell data, we know that there are at least four types of glia cells throughout the gut. So what we don't know is that if there is any location-specific or subtype-specific response to stress of these glia cells, and how this can be involved in cancer.So second, so for this project, we are working on understanding the glia's crosstalk with the stem cells, but they're also surrounded with other cell types. For example, whenever we did the staining for the glia cells, they always come along with the nerve cells. So what we don't know is that under stress condition, if there's any change of the nerves and whether and how this can be involved during the stress-induced tumor genesis. So with that, I would like to thank my lab members, especially Xiaoxiao, she's here, she's a mom now, and Chenchen. They are contributing significantly to this project, and happy to take any questions. Questions, please. Quick question about can you simplify your model, stress or no stress, if you just inject cortisol? Will it produce all the effects on glia and then on colitis or spontaneous tumor genesis? This is a very good question. I think my trainees will be very happy to know because they have to come to lab every day to stress the mice. And I think the simple answer is yes, because we tried in a different model with breast cancer, in which we injected dexamethasone to mimic stress response, and we saw the same results for breast cancer liver metastasis, but we haven't tested in the gut. Okay. Yeah. Tim Wong from Columbia. Really nice talk. Thank you. I like that you used the restraint stress model, which is very reproducible. So you focus a lot on the HPA axis, and you focus more on the glucocorticoid response, but you really didn't mention the adrenergic response and the effects of epinephrine and norepinephrine, which go up under stress and maybe under cold conditions. So did you look at whether or not adrenergic signaling played a role? Did you look at ADRP2 signaling, for example? Yes, very good question. So in another independent project with breast cancer, we actually look at the sympathetic nerves. We found that activation of the nerves specifically locally in the breast can also promote tumor growth in the primary site. But in the gut, it's a little bit different. So we try to identify the nerve types within the gut, and we found in the colon, these sympathetic nerves are really located in the smooth muscle layer. So when we dissected other neuron types, we found it's mostly like VIP or vasoactive intestinal peptide, VIP positive nerves, and endos positive nerves in the gut and also in the tumor. So we guess probably it's not the sympathetic nerves directly working on, but it is also possible because it's innervated into the smooth muscle layer, which has a lot of ganglia. So there's a possibility that it talks to the ganglia of the nerves in the gut and then transmit the signal into the tumor, but we haven't looked at it. Great question. Thank you. Please. Thanks for your fascinating presentation. I just have a question about the gut microbiome. A lot of the bacteria that are associated with psychiatric disorders are also associated with colonic tumorigenesis. And I was wondering if you can close the loop, whether the glial cells impact the bacterial populations that are in the gut, or how do you think about the microbiome in light of your work from a sort of systems point of view? Great. I think this is a very good question. So we actually have a project ongoing in my lab in which we are looking at how stress affects the colitis. And we have some preliminary data of this microbiome composition in the stressed mice, and we do see a lot of changes in terms of the species of microbiome. And we found some of them are closely related to ulcerative colitis in human. But what we don't know is that whether it's related to the glia cells or the stress itself, or it is because stress can induce a leaky gut, and then this leaky gut can promote local inflammation, which will change the microbiome. So we don't know yet, but this is very good. Okay, thanks. Okay, thank you so much. Next, we're going to move into the setting of head and neck cancer, squamous cell carcinoma progression, and hear about the epithelial myeloid circuit from Dichen Lin. Dr. Lin is from the University of Southern California. Thank you. Thanks so much for the introduction and invitation. I'm very excited to share some of the works from the lab that we are looking into the cancer development via the single-cell plus spatial characterization to understand, in this case, tumor monocytes crosstalk. Nothing to disclose. So switching gear a little bit, head and neck squamous cancer. So this is a pretty common cancer around the world. Not too much in United States, but every year there are 600,000 cases around the world. And then squamous cell is the most dominant cancer type, and then it's a very aggressive cancer, especially for the HPV negative subtype. The five-year survival, it's been stagnant for the far past, maybe two decades. It's been lower than 50%. So it's this pressing need for innovative research and drug development for this disease.So in terms of microenvironment, we wrote a little comment last year to highlight, just like many other cancer types, the diversity in terms of the microenvironment is pretty profound, and there are patient-specific tumor microenvironment, which explains the failure rate about 80% if you look at the ICB treatment in the clinic. And so in this case, we're thinking about the heterogeneous cell-cell crosstalk between different subtypes of cancer cells and the different subtypes of myeloid cells. And so specifically, we collected and we also sequenced our own patients. So you're looking at about 167 samples from 114 patients. This is led by Dr. Park, which sits there. So we have samples across the disease settings from normal premalignant tumor, metastatic tumors, and recurrent tumors. And so with that data set, we look into different cell clusters. And so paying more attention to myeloid cells, which you're looking at about 14 subsets here, standard UMAP annotation. So if you look at this, number 7, we call that tumor-enriched monocyte, because if you look at the x-axis from distal normal to precursor tumor, you see a jump in terms of the ratio, sort of normalized the ratio. And then we are very interested in this because across different cohorts, you see very consistent association between the high amount of this monocytic fraction and the poor survival in different patient cohorts, including ICB-treated cohorts. And then this subset also is very featuring the high expression of immunosuppressive genes, and then angiogenesis genes, so even higher than a commonly used positive control SPP1 positive macrophages. So this is even higher than that one. And then we looked through different cancer types, and you can see that across different cancer types that you see consistent enrichment of this cell type in the tumor specimens, especially also with the pre-inflamed or pre-cancer lesions. But you see that the increase is very tumor specific. And then if you look at the pseudo time and trajectory analysis, this is a public data set. So there are blood samples and the tissue samples, and this subset of monocyte, you can see that the slingshot shows that the trajectory is kind of deviated from the development from the CD14 classical monocyte to the macrophage to CD16 positive monocyte. So this is pretty distinct. So we believe that this is still a intratumor monocytic population rather than a fully differentiated macrophage. So, to understand the cell interaction, we took a published method to use the UMF to sort of distill the cell interaction. So basically, we nominate the nodes and the edges based on the weight of the UMF and then also pairwise correlation between the cell fraction. And then with that, we can basically identify modules. Here you can see these modules show very high correlation among themselves, and very interesting module six, it's very high in almost absent in any precursor lesions, but very high in the tumor. And why we're interested, because this is very high in terms of this monocytic population, and it's also enriched with this so-called hybrid EMT or partial EMT, which is recurrently now mentioned in the head and neck and other squamous tumors. And then the correlation is also very, very specific for tumor. You can see that the correlation is rather very random in the precursor or distal regions, but very enriched in the primary tumor and the metastatic tumor. And from here, the hybrid EMT or partial EMT tumors, also just like the monocytic subset, they're always associated with worse survival across different patient cohorts in head and neck cancer. And to further understand their interaction, we did VZMHD, the spatial sequencing to look at their locations. And here you can see the co-occurrence score based on anchoring the tumor at the center. And then you can see that the modeled score shows that the tumor-enriched monocyte is really the highest among all the module six subtypes. So it suggests the direct proximity between this monocytic cells and the hybrid EMT tumors, and you can visually see them across many regions in the tumor. They basically touch each other very often. And then this is the other way to look at it. So if you look at 50 micrometer, which is a reasonable distance, you see that across different-- So we also basically analyzed the subtypes of the tumor. So highly differentiated, cell cycling tumors and hypoxic tumors or oxidative tumors. So all different cell subtypes of tumors, the hybrid EMT or partial EMT always shows the highest correlation, the spatial closeness to this subtype of monocytic cells, and this is another way to look at the distance. If you rank them, the closest, the shortest distance... occurring the partial EMT cells. And we can also quantify the scores based on their distance. So we ask, for example, the tumor cells that are close to these monocytes versus tumor cells that are far away from the monocyte, which tumor cells have the higher, let's say, EMT score. And you can see that the more close they are, the higher score you can see from tumor cells. And then the same is true for these monocytic cells. I showed you the immunogenetic scores that is very high in these monocytic cells. And then when they're closer to the partial EMT tumors, they're also higher. So it shows that from this correlation analysis, you can see the spatial distance really dictates their functional status, at least from the gene expression level. And you can see that in the TCGA based on bulk data, the hybrid EMT score and the monocytic score is highly, highly correlated. So that suggests sort of a functional crosstalk between these two cell types. And then to understand that, we started the patient using organoid modeling. So basically, we took PBMC-derived monocytes, and we co-cultured them in patient-derived organoids. And then we created a matrix liquid interface so that these monocytic cells can really move around and interact with the tumors because the matrix is pretty dense, and monocytes, immune cells don't move very well, but in this case, they can actually move and infiltrate. So then after five days, you take these cells out, and you look at the molecular changes. So the first thing you see is that the signature scores of the monocyte compared to the culture alone, the monocytic cell alone, and the different patient organoids really enhance the expression of these monocytic genes up to 150-fold. So it's pretty huge. And then in the standing, you can also see that CD14+ monocyte, you co-culture with the... So we use ITGB4 as one of the marker genes for the HME, hybrid EMT cancer cells. So you see that life imaging shows that some of the... So this is a confocal. You see that some of the monocytes actually can infiltrate into the organoids, and you quantify that. And then we did one, two, three, four, five, six patients so far. So we stratified them by EMT high, EMT low patients. So you see that EMT low versus EMT high, you see that the hybrid, this is EMT high organoids, tumors infiltrate, recruit more monocytes into this tumor mass versus the low counterparts. And when you look at the organoid themselves, also you see that before and after the co-culture, you see the ITGB4, this one of the markers. So we see that increased also pretty significantly after they see the monocytes. So suggesting that they have functional crosstalk. So to further prove that whether hybrid EMT does impact the recruitment and the reprogramming of these monocytes, we use flow to sort cancer cells into EMT high, EMT low based on PDP and ITGB4. This is single-cell derived markers. And then after sorting them out, we incubate them with the monocyte, just like before I showed you. And then you see that compared to double, sorry about the resolution, it's kind of low. Double negative cells, which is hybrid EMT low, versus the double positive, which is hybrid EMT high. You see this very distinguishing difference. So the cancer cell, if it's double negative, it does go up, but very minimally, but with the double positive, it goes up to sixfold, suggesting that the cancer intrinsic hEMT status really can dictate their impact on the monocytic cell state. And then after the co-culture, we also took out these cancer cells, and we look at their hybrid EMT gene expression. And you can see the panel of genes here, some of the classical EMT genes, some of them are the squamous sort of identity genes here. And after co-culture with the monocyte, these cancer cells also express much, much higher hybrid EMT markers. And then you can also look at the math flow. So the hybrid EMT low organoids, they do go up, but very minimally, but after co-culture with the monocyte, the hybrid EMT high organoid, these patients, their EMT signature goes even higher from 10% to 20% to 30%. So that's basically showing that vice. So this is a bidirectional crosstalk between hybrid EMT tumor cells and this subset of monocytic cells. Basically, they impact each other to increase the, at least in the gene expression and protein standing level, the EMT markers and then the monocytic androgenetic genes. So to put that into more in the cancer context, we and others, we've shown extensively theThe function of hyper EMT in squamous biology, basically, it promotes not only the invasion metastasis but also drug resistance, radiation plus cisplat in this case, which is commonly used in treating head and neck cancer patients in the clinic. So you see that with cisplatting alone, the killing is about 50% ish, but in this case, we're using THP1 as a model system to show that. Okay, I'm done. I thought that's the time. That's it. Oh, I'm confused. Yeah. Yeah, I'm done. This is the last one. So with the monocyte being there, you see the cancers don't really die much. And then last one, this last piece. So basically in vivo you can see the tumor cell along, this is survival. And then if you co-implant them with the monocyte, they die much faster. But then the last piece is if you inject the monocyte into the circulation, after 20 days you actually see them being actively recruit to the tumor microenvironment through the bloodstream, so showing that there is active recruitment into the microenvironment. And that also reduced survival. So that's it. That's the little summary here. I'm going to skip that. And then lastly, I want to thank the lab members who are very dedicated and talented postdocs I'm blessed to work with. Thanks so much. I'll take any questions. Thank you. We have time for two quick questions. Go. One question I have, so EMT high cancer cells are more fibroblast-like, right? There is a concept that there is a tissue unit of fibroblast, macrophage or monocyte, and epithelial cell. Do you think if you incorporate into your cultures cancer-associated fibroblasts- Mm-hmm ... then monocytes will be uniquely skewed towards what you see regardless of the epithelial characteristics, and everything will be dictated by fibroblasts? That's a very nice question. I don't think so. I think there has to be cancer present, but in the computational analysis we do see there's myofibroblasts also in that community. So, hey, Dr. Wang. Yeah. Nice talk. So in terms of your TIMO monocytes, do you think they play any role in immune suppression? That's the biggest focus for the monocyte field these days. And for your TIMO monocytes, there've been a lot of failures in trying to deplete monocytes and treat cancer. Do you have any specific targets for TIMO? Yeah. Thanks so much. That's a very insightful question. Absolutely. So as I mentioned, we found the TIMO enrichment also is very strongly associated with ICB treated failure. And then we actually were starting to look at this, so when we do a trio culture, you put T cells in, put monocyte, we put the patient-derived tumor organoids, there is a dampening in terms of at least the activation markers of these T cells. We don't know the killing effect yet, but at least the flow level shows that there's decreased CD69, decreased granzyme B levels. So we believe there will be a somewhat impact on T cell activation. In terms of the dependency, yes, I think it's not been very fruitful in terms of targeting them, but we do think there's a metabolic vulnerability in terms of in this specific subset. I'm not ready to show any of the data here, but we do think that's a promising future lead to follow up. Thank you. Okay, I think we are out of time. Thanks so much. Okay. So our next speaker is Maria Hernandez from University of Florida, and chronic sleep deprivation promotes colon cancer, and she will tell us how microbiota factors in to this process. Good afternoon, everyone. Thank you for the kind introduction. I'm excited to share my graduate work at the University of Florida, and I have nothing to disclose. There's a common factor among many cancer patients, is that sleep deprivation. In fact, over 60% of patients have reported sleep disturbances prior to cancer diagnosis, and in fact, over 60% of them experience these disturbances after diagnosis and even into treatment. However, we don't know the biological mechanisms by which sleep deprivation affect cancer progression. However, we do know in the literature that sleep deprivation significantly increases the abundance of pro-inflammatory cytokines in the host that can act through the systemic circulation and nearby lymphoid tissues. These changes have been associated with immunosuppression within the immune landscape of the host. For example, through decreased cytotoxic T cell frequency, as well as decreased antigen presentation, which are both key for antitumor immunity. And these changes have overall been associated with increased risk in colorectal cancer development, as well as cancer risk. Similarly, the field has established that sleep deprivation can negatively modulate the microbiome by altering its structure, for example, through elevating the ratio of Firmicutes to Bacteroides. Now, the novelty of our work is aiming to understand how sleep deprivation induced dysbiosis drives colorectal cancer development and further can decrease treatment response. To study our central hypothesis, we employed a model of chronic sleep deprivation. Using these commercially available cages, we housed mice, where a bar sweeps across the floor of the bar every two minutes for 12 hours a day from 7:00 a.m. to 7:00 p.mThis works because mice are nocturnal, but it's meant to mimic a human being awake in the evening. The mouse models we used were Black 6 male mice, ages eight to 12 weeks old. And we employed a murine adenocarcinoma model of MC38. The majority of the samples in the readouts were stools for 16S rRNA gene sequencing, as well as tumor for qPCR analysis of mRNA gene expression, as well as using flow cytometry to analyze tumor-infiltrating lymphocytes. The first fundamental question we wanted to address was how sleep deprivation, now I will label it as SD, can promote colorectal cancer development. We pre-exposed the mice to seven days of SD, and the control mice were housed in the same room with the same light exposure and the same temperatures. After a seven-day exposure to the SD mice, we implanted both the SD as well as the control with half a million MC38 cancer cells, and we followed through with tumor growth until a significant endpoint. To assess how this regimen of SD was affecting circadian rhythm expression, we took endpoint tumor tissue and explored circadian rhythm gene expression. For example, as you can appreciate, one of the positive regulators, BMAL1, was significantly decreased in the SD mice compared to the non-SD counterparts. Similarly, the protein family Cry, both Cry 1 and Cry 2, were significantly elevated in tumor RNA compared to the non-SD counterparts. This was also observed in the other core protein, Per2. In the context of tumor growth, we see that SD mice were significantly more susceptible to tumor volume compared to the non-SD counterparts. This was also associated with a significant increase in tumor weight. And in the context of tumor immunity, we observed that the SD mice had a lower expression of IL17A, IL1B, and TNF alpha, likely alluding to an immunosuppressive tumor microenvironment. But more importantly than tumor progression, we wanted to focus on answering this fundamental question in another aspect of cancer care that's important, like chemotherapeutic response. 5-FU has long been standing as a key drug for human colorectal cancer. And so to assess this, we asked ourselves two fundamental questions. How is our chronic SD regimen affecting 5-FU efficacy? And how is a sleep-deprived microbiota affecting 5-FU efficacy? Similarly, as before, we employed the same model where SD mice were housed in their cages and pre-exposed to seven days of chronic SD. After the seven-day exposure, we implanted them with the same concentration of cancer cells. And when the tumors were palpable at day 10, we began treating them with 5-FU at a concentration of 25 milligrams per kg via IP route. The control mice were housed again in the same room and received the same tumor implantation and the same treatments starting at day 10 and until day 22. Endpoint was reached when the mice were at tumor endpoint. You can appreciate that the SD mice receiving 5-FU were significantly more susceptible to tumor growth compared to the non-SD controls also receiving 5-FU. This was also coupled with a significant increase in tumor weight. In the context of tumor immunity, we observed that the SD mice had a significant decrease in CD27 and interferon gamma in tumor mRNA levels, likely alluding to a decrease in T cell infiltration and cytotoxicity. You can also appreciate a significant increase in IL6, a pro-inflammatory cytokine associated with sleep deprivation. More exciting for us, we were able to observe using flow cytometry that the tumors of SD mice were significantly lower in frequency of CD4 positive T cells, which are effector cells key for anti-tumor immunity, as well as they had a lower frequency of CD8 T cells and interferon gamma CD8 T cells. Sleep deprivation also affected the infiltration efficacy of natural killer cells as well as granzyme A positive natural killer cells. And you can appreciate that the overarching theme is that this regimen significantly pushed the mice to a lower immune activation. It made it more susceptible to a decrease in 5-FU efficacy. Now, to understand how this regimen could be negatively affecting the microbiota, we employed a validated model of fecal microbiota transplantation that I will explain shortly. Using the endpoint stool of the SD versus the non-SD mice from the previous study I just showed you, we took those stools and we prepared a pooled slurry of them. Now, we took healthy Black 6 mice, all male, and they were housed in the same facility near each other, received seven days of antibiotic water cocktail, and after a 24-hour washout period, we orally gavaged them with either a pool of the SD stool slurry or the non-SD stool slurry. We allowed for the microbiota in the slurry to implant in the gut of the mice, and then we implanted them with the same concentration of half a million MC38 cancer cells on the right flank. Same as before, on day 10, we began treatment with 5-FU via IP and followed through for four more rounds and then measured the mice until endpoint. We collected stool at zero and at endpoint to assess changes in the microbiota diversity using 16S rRNA gene sequencing. You can appreciate that at time point zero, the microbiota of the non-SD mice compared to the SD mice clusters significantly different at the beta diversity level.These changes had also gone through until the endpoint. As you can appreciate, the non-SD cluster tightly, whereas the SD cluster separately at endpoint. So not only do we see significant changes in the microbiota sustained over a period of about 26 days, these changes were also accompanied by functional changes in the mice. You can appreciate that the recipient SD mice receiving 5-FU had significant tumor growth compared to the non-SD recipient mice treated with 5-FU. The SD recipient mice were also more susceptible to a decrease in TNF-alpha, a key pro-inflammatory cytokine important for anti-tumor immunity. In the context of tumor-infiltrating lymphocytes, we also saw a significant decrease in frequency of effector CD4 positive T cells, as well as cytotoxic CD8 positive T cells. So overall, our preliminary findings have demonstrated that our chronic SD model can significantly modulate the microbiota and create a dysbiosis or a lack of homeostasis that can potentiate the growth of cancer cells using an MC38 cancer model and a decrease in cornerstone chemotherapy 5-fluorouracil. I would like to thank my lab, the University of Florida, and our funding sources, as well as AACR for this wonderful opportunity. Thank you. That's very intriguing data. I have to ask, and I'm sure you've been asked this many times and have thought about it, but we all do our experiments during the day- Mm-hmm. When mice are supposed to be sleeping. Mm-hmm. So we've kind of reversed their night and day to begin with. So the animals you're using in this study presumably would've been conditioned under those circumstances before this experiment began. Yeah. How do you think that that may affect all of our data, let alone, then doing sleep deprivation studies? Well, it's incredible because before I started working on this, I didn't really understand how impactful circadian rhythm is. Every cell of our body has a circadian rhythm. Right. And some of the things like our gut microbiomes respond to that. So I think it's very important in all aspects of cancer research to incorporate these very large physiological systems that are taking place, like circadian rhythm, because prior to studies, in circadian biology, yeah, it's not really known, but- Yeah. Yeah. Thank you for bringing that out. Yeah. Very important work. Thank you. Yes. So, first of all, thanks, Betsy, you asked my question. Second question, but I have another suggestion. It's about the experiment that you show at the end where you do the FMT transfer. Mm-hmm. You took the mice at the end of the tumor growth. Mm-hmm. You took the FMT from mice with tumor. Yes. Have you done that with just simple stool of mice from sleep deprivation and then performed the experiment? We have not. But knowing that we have a preliminary phenotype, we're working into expanding and increasing the rigor of our experiments. So I appreciate you bringing that up. Thank you. Sure. Yes. Good work, Maria. Thank you. I was wondering if you have a confounding factor here that is stress. Mm-hmm. How can you differentiate between sleep deprivation or the machine is actually- Mm-hmm. Alarming the animals constantly? So I thank you for bringing that up. Preliminarily, that's why I checked for circadian rhythmicity to assess that it was affecting that axis of the host. But for our future studies, I will definitely incorporate checking plasma levels of certain stress hormones like glucocorticoid, or even other aspects of the parasympathetic or sympathetic nervous system. So I think that's very important. But something else about the mice and that I have read in the literature, they tend to acclimate to certain stressors. And because this is not an acute model and rather a stressful long-term chronic model, I anticipate that they're probably adapting to it. But I definitely agree that molecularly, I need to determine that whether this is a stress versus a sleep deprivation. So thank you for that. Or both together. Or both together. Absolutely. Thank you. It's a good talk, actually. I just want to comment on your experiment, where you used only the male mice. Yes. And circadian rhythm and bone colorectal cancer- Yes. They're affected by estrogen. Right. So, did you look into that also? Did you use female mice also? We have not used female mice simply because we know that females have the estrous cycle, and that does have fluctuation. So we wanted to minimize that confounding variable. As you can appreciate, the project is taking a very wide leap, so I think we're just trying to mostly concentrate on the microbiome as the highlight. But I definitely agree that there's to be consideration of the female sex and all of those factors, yeah. Thank you. Thank you. I have a question about the applicability of your model- Yeah. To the sleep problems we see in the clinic. Mm-hmm. Which are generally sleep apnea- Mm-hmm. Or mild to moderate, which is- Mm-hmm. Affects a surprising number of people. Yeah. Like 40%. Yeah, definitely. Like most- And the sleep apnea tends to track with visceral adiposity also- Mm-hmm. Which creates another thing to try and dissect out of what we're really measuring. Mm-hmm. How are you thinking about extrapolating from what you've done in the lab? So like most things, no model is perfect. Whenever I try to implement... There are lots of different sleep deprivation models in the literature. There's the jet lag model. There is a model that specifically affects REM sleep. And so we were trying to model something that a human or a cancer patient would be experiencing. So I think that's something that we should consider or maybedo our chronic SD and maybe incorporate other models that will likely... But yeah, it's not perfect. It was meant to be more of an umbrella term, but not a specific sleep condition because, as you know more, there are many, like insomnia, sleep apnea, et cetera. So yeah, not a perfect model, but- Thank you. Thank you. So sleep deprivation affects microbiota. When you do FMT, do clock genes change in immune cells or in the tumor? So I didn't show it here, but we have preliminary findings just from the tumor RNA that BMAL1 was significantly altered in the SD recipient mice. But we want to do further studies and evaluate more of those changes in specifically other tissues other than just the tumor. Okay. Thank you so much. Thank you very much. Next, we're going to hear about obesity-induced changes in adipose tissue in ovarian cancer by Dr. Sophia Hou from San Diego State University. You didn't have to come very far. Yeah. Yeah, thank you for the introduction and the invitation. I am a PhD student, to clarify for the audience. But yeah, I'm very excited to present my work to you, "The Impact of Obesity-Induced Changes in Adipose Tissue on Ovarian Cancer Progression." So ovarian cancer is the most lethal gynecologic malignancy and can be categorized into four different stages, where stages one and two is where the cancer is confined within the pelvic region. And then in stages three and four, the cancer has now spread to the omentum, and the omentum is an abdominal fatty tissue sheet that covers the peritoneal cavity. And unfortunately, at stages three and four, this is where ovarian cancer is most frequently caught due to its nonspecific symptoms and lack of accurate predictive screening tests. So, this makes understanding the biology of metastatic colonization to the omentum increasingly important. When ovarian cancer metastasizes to the omentum, the ovarian cancer cells establish a complex tumor microenvironment enriched of the cancer cells, pre-adipocytes, adipocytes, immune cells, fibroblasts that are secreting its own extracellular matrix composed of collagen or fibronectin. And there is cell-to-cell communication in the forms of IGF-1, IL-6, or TNF alpha as examples. And pre-adipocytes are precursors to fat cells, and they make up 15% to 50% of the stromal cell compartment and are an important component of visceral fat. And unlike lipid-rich adipocytes, their morphology is fibroblast-like, and their secretome is enriched in proinflammatory and tumor-supporting factors. So a previous graduate student wanted to understand the role of these pre-adipocytes in ovarian cancer tumor formation. And so she did a limiting dilution injection model where she injected 40 cancer cells to one pre-adipocyte, an equal ratio of one to one, and then a one to 40 ratio. And the injections of cancer cells alone were not able to form tumors, but with the presence of pre-adipocytes, this significantly enhanced tumor growth. And this one to 40 ratio is mimicking isolation stress, which is a condition characteristic of metastasizing cells that are colonizing the omentum where there is an abundance of pre-adipocytes. And we further validated with injections of mature adipocytes that both pre-adipocytes and mature adipocytes are generating tumors similarly, as well as tumor growth. And the cells were then subjected to co-cultures between pre-adipocytes and adipocytes, and there were significant hits for cellular response to stress and extracellular matrix organization, as well as regulation of IGF transport. And I want to emphasize that this RNA was taken from the cancer cells in co-culture. And specifically here, these pre-adipocytes enrich this phenotype implicating insulin-like growth factor signaling important for these cancer cells. And IGF has vital effects on proliferation and survival, and it was indeed increased in co-culture with pre-adipocytes. And IGF stimulation on cancer cells increased ECM remodeling genes like collagen and fibronectin. And in my preliminary experiments, I'm finding that IGF is working in both nutrient-full and nutrient-depleted conditions, activating phosphorylated AKT as well as phosphorylated p65, which is associated with the classical NF Kappa B pathway. And obesity increases the risk for several cancers, including ovarian, and its dysfunctional adipose tissue creates a special and permissive niche for ovarian cancer to spread and thrive. And in an obesity mouse model, these researchers found that there was increased metastasis to the omentum as well as to the mesentery, and these are adipose-rich sites. And in a different study, these omental adipose tissue were taken from obese and morbidly obese patients, and there was increased IGF expression as well as activation of the IGF receptor. So all of these potential links led us to want to understand this obesity-mediated IGF signaling in ovarian cancer metastasis. So I have hypothesized that obesity increases pre-adipocyte-mediated activation of IGF-1 signaling to enhance ovarian cancer growth and metastatic ability.So to test my hypothesis, I obtained normal BMI preadipocytes from female donors or a female donor, and high BMI preadipocytes from a different female donor. And I did either a co-culture or a co-injection model with GFP-labeled cancer cells. And for the tumor model, I processed the tumors with FAC sorting and also prepared tumor lysates. And for the tumor model, I collected at an early time point during the tumor's exponential growth phase, as well as a clinical endpoint. And normal BMI and high BMI preadipocyte injections generated tumors similarly. But at the early time point, the high BMI preadipocytes had a significant increase in growth relative to injections of cancer cells alone. And over the course of the whole study, the high BMI preadipocytes had a decrease in the probability of survival. When I collected these tumors for FAC sorting, I found that there was an equal abundance of GFP-positive cells within the tumors. So I collected these cells for RNA, and I did a qPCR. And I tested for SOX2 gene expression, and I found that there was an increase in the normal and high BMI setting. And this was interesting to us because SOX2 has been indicative of these tumor-initiating cell populations. And then at the later time points, there was a significant increase in the high BMI preadipocyte group. So this data is suggesting that the preadipocytes are enhancing tumor initiation through SOX2 potentially, but the high BMI preadipocytes are enhancing ECM remodeling through fibronectin. In co-culture, I found that there was increased IGF secretion between both normal and high BMI preadipocytes. And then in the tumor lysates, I found a trending increase of phosphorylated NF-kappa B activity. But as far as the non-canonical NF-kappa B, RelB processing, I found that it was high in the late tumors, but there were no clear differences among the groups. So this data, along with my preliminary serum deprivation experiments and IGF stimulation, indicate that IGF may provide a survival advantage through phosphorylated p65, and this may be modulated by the OB setting. I collected conditioned media from these preadipocytes, and I did a cytokine array, and I found that there was an increase in pro-inflammatory factors such as IL-6 or MCP-1, as well as increase in growth factors like HGF and FGF. But in the normal BMI setting, there was increased IGFBP binding proteins, and this is indicating that since these binding proteins are inhibiting free IGF-1, there may be more bioavailable IGF in the OB setting. So to summarize this talk, I found that IGF signaling is driving omental colonization by inducing NF-kappa B transcription factor activity, that these obesity-induced changes in preadipocytes are contributing to early tumor growth and a decreased survival probability in the mice, and that the high BMI environment may drive tumor initiation through SOX2 expression, as well as enhancing tumor survival and ECM remodeling through fibronectin. And my current goal right now will be to define the mechanistic role of IGF in an obesity-induced metastatic potential model in vivo. So with that, I would like to acknowledge my lab and my committee members, as well as our funding resources. And thank you to AACR for providing this opportunity to speak here. Thank you for your attention. Questions, please. Very nice talk. I have been thinking as you were speaking, and actually in some of the other talks, there seems to be a controversy emerging about the type of fat that matters, and I guess I'm thinking about brown or beige fat versus the white adipose tissue. And I wonder if that question is something that impacts the work you're doing. Does it matter- Mm-hmm ... what type of adipocyte we're talking about in terms of fat tissue? Yeah. Both brown and white adipose tissue have different thermogenic properties- Right ... for the body. But as far as the adiposity of the individual, I think, yeah, my preadipocytes are from higher normal BMI patients, but I think there can be a better clear distinction of this visceral adiposity versus a- Mm ... general increase in BMI. And I think that incorporating a better metric along with BMI, such as waist-to-hip ratio or waist circumference, for these ovarian cancer patients may be indicative of maybe their metastatic prognosis. So, yeah. And another question I had. So adipocytes are also known to interact with immune cells and sort of educate immune cells in environment. And IGF-1 data is clear that adipocytes can release it, affect cancer cells, and so on. Do you consider any role in immune education? Mm-hmm. Yeah, I think with-From my preliminary data, I'm also seeing that mature adipocytes also secrete IGF-1, yes. But with the immune crosstalk, I have not dove straight into that just yet. But with these preliminary cytokine arrays, it's indicating that, and also this classical NF-kappa B activation that I'm seeing with the high BMI, it could definitely lead me down towards that way too. So that's something I will be unraveling in my mouse model. Yes. Hi, that was a really interesting talk. I was just curious if right now you have it kind of bended low onto normal and high BMI. Mm-hmm. But have you gone through and maybe plot the BMI individually and see if there's a dose response for increasing BMI? No, I have not. Because the background data that I was showing, the morbidly obese setting significantly enriched for IGF, yeah. But I have not dove into a dose, like if you're underweight versus overweight versus obese and morbid. I have not gone into that just yet, yeah. Mm-hmm. Great question, though. Thank you. In addition to IGF-1, any role for metabolites released from adipocytes? Mm. What they can provide to cancer cells to make survival better. Mm-hmm. When we knock down RelB expression, we see changes in lipid metabolism in cancer cells, which is very interesting. But I have not gone into the BMI story of that yet, whether an enrichment of lipids and an abundance of adipocytes there, whether if we inject cells with RelA knockdown or RelB knockdown, whether that can change. But yeah, very interesting point. Yes. In the high BMI patients, frequently, patients have insulin resistance. Mm-hmm. And the insulin-resistant patients, fat is releasing fatty acids, and you get an inflammatory milieu. And so I'm just fascinated and interested that you were able to isolate IGF-1 in the sort of vast interconnectedness of metabolism and inflammation. Mm-hmm. Yeah, I think as far as fatty acids go, it's been published in ovarian cancer that FABP4 is an integral mediator of fatty acid intake into the cancer cells, and that this is mediated by the adipocytes. So what I'm really trying to unravel is the specific factors from pre-adipocytes that are less studied as mature adipocytes, and how they're able to regulate inflammation or potentially regulate metabolism in ovarian cancer metastasis. But yeah, very interesting point, and yeah. Just one other question. I noticed adiponectin was low- Mm-hmm. in these patients. Have you thought about using the drug that stimulates PPAR gamma to see if you can compensate that? We do have that in the lab. I've mainly been using it for adipogenesis studies for pre-adipocytes to adipocytes. But I think that would be interesting to further interrogate the role of PPAR gamma signaling and maybe with NF-kappa B too, if there's an interplay in metabolisms, yeah. Thanks. Thank you. Okay, thank you so much. So I believe our final speaker for today is Aram Liu from Fred Hutch Cancer Center, and he will talk about immunotherapy resistance to bone metastasis mediated by myeloid-driven mechanisms. I thank the organizers for this opportunity. My name is Aram Liu. I'm a post-doctoral fellow under the mentorship of Dr. Larry Fong at Fred Hutch Cancer Center. Today, I'll be presenting how myeloid cell-mediated immunotherapy resistance mechanisms evolve during prostate cancer progression, particularly in bone metastasis. I have nothing to disclose. Prostate cancer is among the most common male malignancies. Although immunotherapies, including immune checkpoint blockade, have revolutionized cancer treatment, patients with advanced prostate cancer remain largely unresponsive to these therapeutic options. For example, patients treated with atezolizumab, which is PD-L1 blockade, shown in blue, showed marginal efficacy compared to those treated with control arms. This therapeutic resistance is partly due to the presence of immunosuppressive tumor microenvironment. Particularly, tumor-associated myeloid cells are encouraging and promising immunotherapeutic targets. It's been well reported that immunosuppressive myeloid cells, including macrophages on the left and another suppressing myeloid population called myeloid-derived suppressor cells, become significantly enriched during prostate cancer progression. However, efforts to target these myeloid cells broadly, such as with CSFR1 receptor blockade, showed minimal efficacy in patients. So our lab hypothesized that if we identify distinct immunosuppressive myeloid subpopulations that signals a resolution, we may be able to enhance immunotherapy efficacy in patients.To address this question, we have previously investigated the primary tumor biopsies across multiple stages and identified a distinct macrophage population expressing elevated levels of SPP1 transcript, which encodes the protein osteopontin. These macrophages become significantly enriched during disease progression with enriched immunosuppressive molecular programs. Through preclinical studies, we demonstrated that these macrophages drive immunotherapy resistance, shown in blue versus red, via adenosine signaling pathways. So these findings inform the clinical trial showing that blockade of adenosine receptor synergize with atezolizumab, shown in red, compared to monotherapies. And these macrophage population may serve as predictive biomarkers. Although these longitudinal studies have provided significant clinical insights, these findings were primarily derived from primary tumors. And prostate cancer tend to spread to multiple organs, including bone, liver, and lymph nodes. And given majority of patients ultimately succumb to metastatic disease, we hypothesize that if we characterize myeloid departments across multiple metastatic organs, we may be able to distinct these immunotherapeutic targets. To address this question, we added more primary biopsies, particularly from metastatic organs, to existing single cell transcriptomes and conducted further transcriptional analysis. We found that a distinct organ has a unique immune and non-immune environment, notably bone metastasis, which is the most common organ of prostate cancer spread, showed a unique tumor microenvironment with neutrophils as a dominant component, shown in green. Importantly, we found that SPP1 high macrophages that we reported showed a marginal scarcity in bone metastasis compared to soft tissue sites, highlighting the need for organ-specific investigations. So we hypothesize that single cell characterization of myeloid cells, particularly neutrophils in bone metastasis, may reveal immunotherapeutic targets. To address this question, we combined prostate cancer bone biopsies with tumor-free bone samples and then performed further single-cell analysis. We identify multiple neutrophil subpopulations corresponding to their developmental trajectories. Notably, a mature neutrophil with elevated expression of IL1B became significantly enriched in tumor bone biopsies, and we orthogonally confirmed their presence at protein levels in bone biopsies. Using a published gene signature of immune suppression, we found that our target neutrophils show elevated immune-suppressive molecular programs compared to other neutrophil subpopulations. So we hypothesize that these IL1B mature neutrophil may play a critical role in driving immunotherapy resistance in prostate cancer bone metastasis. To investigate in more detail, we utilize preclinical models. We developed intraosseous prostate cancer model by engrafting syngenic mouse prostate cancer directly into the tibia. We collected the cells from tumor-bearing bones and then conducted single cell profiling along with characterization of cell surface antigen. Using systemic cross-transcriptional comparisons between the species, we found analogous mature neutrophils in our intraosseous models. And consistent with human data, our target neutrophils show elevated immune-suppressive molecular programs, suggesting that these neutrophils are well recapitulated in our model. Using cell surface markers revealed by our multi-omics sequencing, including CD62L and CXCR2, we successfully confirmed their presence at protein levels. To test whether they are immunosuppressive, we isolated these neutrophils and co-cultured them with activated CD8 T cells. The horizontal axis represents proliferation of T cells, and we found that these neutrophils potently suppress T cell proliferation and activation, at least in culture. To test the relevance in vivo, we generated two cohorts of mice, one with subcutaneous prostate cancer, another with intraosseous prostate cancer. It's been well reported that subcutaneously engrafted prostate cancer respond to immune checkpoint blockade, shown in gray versus red. So we reasoned that if our target neutrophils drive immunotherapy resistance, if we transfer these neutrophils from intraosseous prostate cancer into subcutaneous prostate cancer, then this introduction may confer therapeutic resistance. Indeed, when we introduced our target neutrophils, those subcutaneous tumors significantly became resistant to immune checkpoint blockade, suggesting that our target neutrophils mediate resistance to checkpoint blockade in vivo. To investigate the molecular mechanisms, we carried out pathway analysis using differentially expressed genes and then crossed the species to identify conserved activation of the pathways.And we found that inflammation-related pathways were highly upregulated in our target neutrophils. And further computation analysis implicated IL-1 receptor pathway as a key mechanism. To explore this possibility, we initiated co-culture of our target neutrophils with activated CD8 T cells in the presence or absence of CD1 receptor blockade. Indeed, immune suppression mediated by our target neutrophils were significantly reduced following IL-1 receptor inhibition. To test the relevance in vivo, we developed intraosseous prostate cancer and treated the cohort with immune checkpoint blockade in the presence or absence of IL-1 receptor antagonist. Consistent with human data, immune checkpoint blockade alone showed minimal efficacy, shown in blue. And IL-1 receptor inhibition also didn't show significant impact. However, when we combined these together, tumor growth significantly reduced, suggesting that IL-1 receptor blockade can enhance immune checkpoint blockade efficacy in intraosseous prostate cancer. Lastly, to investigate clinical perspective, we analyzed a published symposium consortium data, including bone biopsies. We stratified the patients using our mature neutrophil signature scores, and we found the elevated target neutrophil signature scores are highly associated with reduced survival, shown in gray versus red. Moreover, our target neutrophil signature scores also significantly correlate with increased immune suppression, as well as inflammation-mediated immune suppression, suggesting that our preclinical findings well align with those patient perspective. In summary, our lab has identified a distinct mature neutrophil expressing elevated IL-1 beta transcript in prostate cancer bone metastases. These neutrophils are functionally immunosuppressive, and the blockade of inflammatory signals, particularly IL-1 receptor, enhance the efficacy of these therapeutic options. An elevated IL-1 beta high mature neutrophil signature scores correlate with inferior survival. With that, I'd like to thank Dr. Larry Fong for his unwavering support and mentorship, and I thank our lab members and longstanding collaborators. And last but not least, I thank our funding sources, particularly Prostate Cancer Foundation, for supporting this presentation. Thank you for listening and happy to take questions. Yeah. I have a question. Is it black and white, SPP1 macrophages under the skin or orthotopically work, and then IL-1 beta neutrophils work in the bone and there is no role for SPP macrophages there? Yes. So the question, if I understand correctly, whether SPP1 high Macs are more restricted to primary tumors, whereas our target neutrophils are limited to the bone metastases. Am I understanding correctly? Yes. So other metastatic sites need to be further explored when I compare primary tumors versus bone metastases. Although I found multiple neutrophil subpopulations, I couldn't find these IL-1 beta high mature neutrophils in primary tumors, suggesting that those neutrophils, at least in the context of prostate cancer, may be more limited to bone metastases compared to other sites. Thank you. Please. Hi, I'm Amy Leslie. I'm from Stanford. And because we're talking prostate cancer, we got to talk about AR. And so thinking back to Amy Moran's and Felix Swings' paper on how enzalutamide and targeting AR changes the immune microenvironment and activates MHC-1, how do you think that recapitulates in terms of AR positive and AR negative and how that affects the immunosuppressive microenvironment, especially in terms of neutrophils? Are they affected by the AR signaling access or not? That's a great question, and it needs to be determined, at least for SPP1 macrophages. When I looked into AR expression in those macrophage populations and across the myeloid populations, their transcriptional expressions were quite marginal, suggesting that they may not be highly impacted by AR. But in terms of IL-1 beta mature neutrophils, I haven't done that analysis or experiment yet. So, thank you for the great question. Thanks. Mm-hmm. So conditioning with... Over to this mic. Tony Letai, NCI. Conditioning with cytotoxic chemotherapy is generally found to be necessary to have an effective CAR T cell treatment. It's often considered that this cytotoxic therapy is necessary for lymphocyte reduction. Do you think instead it could be an important effect is myeloid cell reduction? So if I received the question correctly, so whether these- I'm just saying we always give cytotoxic chemotherapy- Mm-hmm. ... to condition people before they get CAR T cells. Yes. And if we don't do that, they don't work well. Mm-hmm. Typically, it's thought that we need to give that chemo to do lymphoreduction, lymphodepletion. Mm-hmm. Is it possible, based on your findings- Mm-hmm ... that myeloid depletion is every bit as important as lymphodepletion, and that we're actually doing myeloid depletion with some of these chemotherapy regimens? Well, as a myeloid cell immunologist, not all myeloid populations are bad. During the disease progression, myeloid cells are reprogrammed into suppressive phenotypes. So I am relatively against the concept of broad ablation. I would rather more focus on how to reprogram bad myeloid populations into initially good, as part of innate immunity, myeloid populations. But I don't have experimental evidence, so I cannot say this confidently. But I think ablating myeloid cells could be playing a role there. Thank you for the question. Maybe I was just going to comment on this last question. I think one of the things we've seen, like even with checkpoint inhibition, actually, most of the efficacy that we see is in combination with chemotherapy. And I think for many folks studying immunotherapies, that's always been a bit of a conundrum. People think about potentially debulking the tumors, but there may be a myeloid reset that is associated with that. It may be a blunt tool to do that, but it's a provocative idea. Yes. Agreed. So prostate cancer isn't the only cancer that metastasizes to the bone. Breast cancer patients often have this happen. Do you think that what you're observing is restricted to prostate cancer, or would we expect some of these changes in, say, breast cancer? Yes. So, I would say the enrichment of our target neutrophils may not be restricted to prostate cancer. However, relative recent articles suggested that other subsets of neutrophils, inflammatory neutrophils expressing chitinase, particularly in the context of triple negative breast cancer. Phils may not be restricted, but it's not frequent or observed in at least bone trophic breast cancer. Okay. Thank you. So I'm charged with providing a summary, and I will be blessedly brief. I think we've heard some incredibly good talks, very exciting and future-directed, I think. And what impresses me the most is, even though this session was entitled "Microenvironmental Determinants," which is true, I'm struck by the systemic nature of the talks that we heard today in terms of how things like obesity, the nerve, brain stress, the microbiome, sleep deprivation, all of these clearly systemic events, inflammation, change from primary tumors to metastatic tumors, are affecting the tumor microenvironment, and I think that this has really opened up far more ideas than we originally thought of when we think about cancer hallmarks. There is going to be some great additions from the system, and I'm particularly intrigued myself with psychological stress and the role of nerves. But I'm impressed today with the incredible presentations that we saw on a variety of other factors as well. So I hope you all take home some new ideas and think about the systemic nature and control of the tumor microenvironment. And thanks to all the speakers. I'm really impressed with the quality of the presentations. We both are very happy with what we learned, and I hope you are too. So enjoy the rest of the evening, and thanks to all our speakers.