Skip to content

2026-04-19-neural-immune-crosstalk-in-cancer

Date: 2026-04-19 Primary topic: single-cell-spatial-omics Cross-referenced from: single-cell-spatial-omics Word count: ~11,558

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, and thank you for joining this particular session on neuroimmune crosstalk in cancer. These are my disclosures. So for many of us who are very interested in the tumor microenvironment, you can see all the usual suspects, the stromal cells, the cancers, the macrophage, the neutrophils, the T cells, and the B cells. However, something happened in Germany 175 years ago. Rudolf Wagner found that when he stimulated the nerve in a dog, the spleen moved. So he contacted his friend, Jacopo Andy, and he said, "Can you do an experiment for me with human data?" So his friend did the same thing with a decapitated human, something that I truly do not want any one of you to try. And then subsequently, a lot of people, including Paul, could start tracing these different nerves. So all of a sudden, for us who are cancer biologists, we start thinking about what if all that we've been doing, looking for the metabolic adaptation that would stimulate the immune system to improve immunotherapy or genomic instability, which would also be something that we really now know can stimulate the immune system against cancers. So this neuroimmune crosstalk in cancer becomes real because that stimulation of the nerves that the spleen moved, perhaps what we're doing is sending a message from nerves into the immune cells. And of course, there are the central nervous system and the peripheral nervous system, and mainly what we're going to be talking about is the peripheral immune system. We can have the autonomic nerves, which is splitting into the sympathetic, which makes norepinephrine, fight and flight, and then the parasympathetic make acetylcholine. But this is not really secretable, the acetylcholine from the T cells to other cells and in the neurons are mainly synapsis. So we will talk about today, the three of us, how these nerves actually play in, innovate, and crosstalk with the tumor microenvironment. Now, the field is slowly growing now. As of last year, 2,000 papers have been published on neuroimmune crosstalk. Out of that, about a quarter of them are now starting to ask questions with respect to what will be involved in the cancer setting. Today, we have three speakers who are all very seasoned in this field, and we have, as you can see here, written some reviews for those of you who want to get involved. I will give the first talk, Tumor Immune Microenvironment: Time to Chat. Professor Trumpp will talk about neuronal innervation of pancreatic cancer, and Professor Talbot will talk about nociceptor neurons impact on antitumor immunity. My journey came in 2008. I was at dinner at a metropolitan club in New York, and I met some neurosurgeons. They told me stories which made me so curious that we started working together as a group. The story was that although extrapolation from the German stimulating nerves, acetylcholine was indeed found about 100 years ago, and that if you actually stimulate the nerves, there would be more of this acetylcholine. But what was curious and mysterious was that up to todayIt has been very difficult to find cholinergic parasympathetic nerves within the spleen, the lymph node, thymus, and the liver. So the question then is, where is the acetylcholine coming from? As it turned out, that these autonomic nerves do usually hit a celiac ganglion. In the case of parasympathetic, they hit a celiac ganglion with acetylcholine, but then it comes out acetylcholine. But in some of these situations, like in the spleens and the liver, very likely, that the parasympathetic nerve comes in as the acetylcholine within the celiac ganglion made a switch. With time, you make norepinephrine. And these are the main organs where it had been very difficult to find fibers of acetylcholine. So what was easy to do- Several years ago, what was easy to do was to actually make a mouse with ChatGFP, not ChatGPT. And that, in a way, would be found that the CD4 cells and the CD8 cells were making a little bit of it. So to make a long story short, within the last 10, 15 years, we've been looking at the idea of how these T cells and these B cells make acetylcholine along the way. Norepinephrine can stimulate T and B cells to make acetylcholine, depending on the B cells and where they are. The T cells can be stimulated by norepinephrine, but T cell receptor, together with IL-21, is perhaps the main stimulus. So why? It's because when we have inflammation in certain part of the organ, the message has to go back to the brain, and the brain send back a message to exactly the same place where there is an inflammation or a cancer. And that way it gives the immune system an extra boost or an extra dampening signal so that we can control our physiology and pathophysiology. I just may give you an example. The two viruses many of you are familiar, they are called LCMV, and the Armstrong forms an acute infection where there are lots of short-lived CD8 effector cells. But the clone 13, which only differs by two amino acids and was originally found in the brain of a patient, has a persistent infection where there is exhaustion and CD4 cells get involved. So the number of CD4 cells that make acetylcholine in the normal mouse is about 2%, and CD8 cells is almost nothing. However, if you infect it with a lymphocytic choriomeningitis virus, about 50% of the T cells are pumping out acetylcholine. But what is interesting is the acute infections will jump from zero to very high and come straight back down to zero, whereas the protracted infection, the CD8 cells hang around to make sure it's clean, and the CD4 cells do the same thing, but they maintain a higher baseline. So we say, "What are these doing?" So Maureen Cox decided to delete acetylcholine transfer from the T cells only. So the whole mouse has every cell that makes acetylcholine the way it should be, but now the T cells cannot make acetylcholine. You can see definitely does this result reveal a genetic proof of function for Chat in T cells viral infection. What is the mechanism? It did look like that in absence of acetylcholine from the T cells, the blood vessels are smaller and fewer, and we suspect that the acetylcholine, in addition to chemokines and adhesion, will then interact with endothelium and then induce the muscles to dilate so that the T cells can go out. And it makes sense because only those T cells that had the T cell receptor amplified or stimulated should be going out. Now, just to kind of confirm this, Peder Olsson and us collaboratedTo show that in mice that have T cells making acetylcholine, the blood pressure is lower. And more recently, he was able to show that in patients with severe circulatory failure, those patients who do have acetylcholine do much better than those that do not. Perhaps have something to do with activation, dilation, hypoxia, whatever, we do not know. And in mice, when induced EAE, a form of MS, it is these T17 cells that parries the acetylcholine that causes the disease, and those that do not, actually do not. So it's become clear that T cells, and later I will show you the B cells, that this acetylcholine that is being conveyed to them allows them to basically function in a more proper way, amplifying or slowing down certain physiological functions. So I think this is really the beginning of an iceberg. So I am of the opinion that every single cell of your body needs these adrenergic and acetylcholine receptors at some point, at some time. You may not see them now, but something happens, they will come up or they will come down. The basic message is, somewhere something is happening pathophysiologically. The message goes to the brain, comes back, usually it's vagal nerve that hits a celiac ganglion, then express into a sympathetic nerve, and they will secrete norepinephrine to activate the T cells and the B cells together with the T cell receptors and the B cell toll receptors. They would then not secrete acetylcholine, but they will then basically make a synapse with almost every cell of your body at the right time, in the right place, for the right reason. Now, this is going to take a long time to solve. There are five muscarinic receptors, GPCRs. There are 16 nicotinic receptors, but some of them form heteropentamers, making the situation so complicated that I begin to believe there may be dozens, if not 100 of these receptors. And there are these nine adrenergic receptors, which are GPCRs. So Kai Qi in Beijing found that the alpha nine is actually required for the maturation of B cells into plasma cells. Shangfen Liu in our laboratory did a very interesting experiment. He found that the alpha nine is on double positive thymocytes. A message from a T cell within the thymus gives it an extra calcium signal, which would double the number of T cells in your body. So this is the way the brain is saying, as evolution progresses, we are exposed to more viruses, more bacterias, we need more T cells, and alpha nine provides that. Well, how do I think that? Because when he knocked it out, then there is less T cells repertoire, about half. And the T cell receptors composition and the repertoires are different. Such that, in fact, when given autoimmune diseases, it is less. Moving on to the germinal center, where T cells and B cells were given maybe three, four, five days to come up with not only the best cytotoxic T cells, but also the one with the highest affinity B cell antibodies to neutralize the pathogen. That race for time is completely controlled by acetylcholine from the T cell telling the B cells not to blow every single B cell, but only the ones with the highest affinity, so then they can become the even higher and higher and higher within the three, four, five days that you only have before the pathogen takes you away. At the same time, the B cell will tell the T cells what to do with another acetylcholine signal.The complications is so immense that multiple muscarinic receptors switch on to nicotinic receptors during that maturation in order that we actually do have exactly what we need. In my final seven minutes, I'm going to go into the liver, and this work done by two very talented post-doc, Nasta Ran and Chung Sheng. I had always thought that looking at sympathetic nerves in the liver, I never thought that there was anything interesting. By the way, nobody had found cholinergic nerves in the liver. But when I saw this paper by Adori et al., I was stunned. You could see that every part of that sympathetic nerves, which is red color, has only one friend. The friend is the portal system, which is the portal vein, the bile duct, and hepatic artery. What does this mean? It means everything that comes through the portal vein and the portal system will be soaked in norepinephrine, because norepinephrine has a half-life of two hours. You cannot escape it. To me, it must mean that the portal system and everything that comes through need norepinephrine. Otherwise, it's not going to be physiological. So what Nasta Ran did was to do the partial hepatectomy, which is very interesting. Your partial hepatectomy, your liver decides to divide or not divide in two to four hours, and then it will stop on day seven for a mouse. So there are many, many cells that are involved. And I'm taking the liberty of this really beautiful picture from Pardo in Madrid, where the hepatocytes is the princess. This princess needs many, many other cells that have to look after them. The smooth muscle cells, cholangioctyes, satellite cells, et cetera. But in there is also T cells and B cells. Oh, and the autonomic nerves. So what Nasta Ran did was to deprive the mouse T cells acetylcholine, and then cut out 70% of the liver. Guess what? They don't care. But when she deprived the mouse of B cells making acetylcholine, the mouse dies. That means B cells is needed for the liver regeneration. So what does this say? Without B cells making acetylcholine in the liver after partial hepatectomy, Cooper cells make less IL-6. That is the brake. That's the accelerator. And T cells make more gamma interferon. Without this, there's no liver hepatocyte regeneration. And with this high, it will stop. So norepinephrine, LPS, activate the B cells to make acetylcholine, to activate the Cooper cells to make IL-6, and now the hepatocytes can divide. That's the accelerator. In every case, when you drive a car, you need a brake. The brake is acetylcholine from the B cells, tells the CD8 T cells to make less gamma interferon, so the hepatocytes can divide. What about liver cancer? Well, with P53, p10 and MYC, Chung Sheng started the liver cancer model. If in this model, the T cells cannot make acetylcholine, you ended up with more liver mets. The mechanism is acetylcholine needs to keep the T-rex down and the PD1 down. In its absence, T-rex goes up, PD1 goes up, then immunosuppveillance is compromised. He actually was able to find some T cell receptors that he thinks are responsible. But more importantly, as again, only the T cells or mainly the T cells with the acetylcholine are actually functional. High fat diets, same story. In a Western diet, if the mouse cannot make acetylcholine in the T cells, you get more tumors.In human, the different receptors go up and go down, whether the tumor or the normal hepatocytes. And some are good prognosis and some are bad prognosis. So we're beginning to look at the complexity of this situation. And the complexity is perhaps beyond our immediate, clear understanding of a simple solution. My final slide. Different subtypes of breast cancer, basal, luminal A, luminal B, HER2, and normal, they all have different acetylcholine and adrenergic receptors. And they indeed make a difference with respect to even prognosis. So I think I have 20 seconds left. I would say, what are we going to do? Luckily, there are still lots of things to do. For example, in the gut, in the intestine, we think even the motility and tumor genesis also has an impact when some of these T cells or B cells not making acetylcholine. Acetylcholine, as I said, cannot be secreted. It's a synapse. The only surrogate you can use is choline, and choline acetyl CoA makes acetylcholine. Now, choline, of course, is not just for this, also with DNMT1, with methionine as epigenetic changes, but also for phosphatidylcholine, sphingomyelin, you also need choline. But nonetheless, these three doctors measured choline level of patients they treated with immunotherapy. And this is the result. Thank you very much. Now we have five minutes for questions. So if we are to target these cells, put it in the system, the T cell, would it affect these CMS for cancer? That's a very good question. Because a lot of the receptors are found in multiple organs. But their signaling is very different. Alpha 7, for example, is a positive signal in neurons, but it's a negative signals in myeloid cells and in T cells. We have to be able to target these in a very organ-specific manner, and we're working very hard on it. Thank you. Hi, Professor. Sorry, I think I was first. Okay, sure. Thank you very much for that talk. I was really struck by the phenotype where you showed that T cell-derived acetylcholine had no effect on the survival of mice after liver injury, but the B cell-derived acetylcholine did. From my very naive understanding, number one, it's the same acetylcholine regardless, and two, T cells are the predominant lymphocyte type in the liver regardless compared to B cells. So where does this difference arise? Is it because of proximity to the source of the acetylcholine, or is it because of maybe an autocrine loop? And I guess the follow-on question is, do you see other instances where there's a difference between B and T cell-derived acetylcholine? In every situation, the B cells and the T cell-derived acetylcholine are different. So we're beginning to think that there are very specific synapses with its target cells. That would be question number one, answer my last question. And in respond to your first question, after partial hepatectomy, there is a flood of B2 cells coming in through the portal vein system. So even though the B cells are outnumbered by T cells, the number of B cells making acetylcholine, both from the B1 within the liver and the B2 coming in through the portal system, apparently met the demand. So in that case, that's contingent on the acetylcholine only behaving in, only exerting its effects through a synapse, so it's proximal. Absolutely. Okay. Thank you. Thank you very much. That was a very interesting talk. I'm kind of new in this field, so I was wondering if, do you have any suggestions as to how would be best to identify this type of acetylcholine signaling transcriptomically, right? Because right now we have a lot of availability with spatial transcriptomics, so if we want to see if they're having this behavior. But I worry that trying to see this acetylcholine activity, you need to look at it at the mass spec level, but I don't know what your perspective is. Well, it's a very good question. Obviously, we cannot make a human with GFPSo the Per Olofsson's lab does qPCR on the T cells from these patients, and he thinks that the result is very reliable. And so for the time being, that's all we can do. Perhaps, better technology can be used to detect what I think is a quite crucial neurotransmitter that has expressed on a very transient basis. Mm-hmm. Okay. Thank you very much. Hi. Very nice talk. I was wondering if you wouldn't mind clarifying your very last slide, which you showed the Kaplan-Meier, which patients with high choline are doing much better. Is this TCGA data? No. No. It's our own patients' data. Okay. So this is actual measurement in serum? Yes. Okay. Thank you. Hi. Thank you for the talk. Regarding the acetylcholine interactions observed in the liver, especially since liver isn't really my area, I was just wondering if this was observed in the nerve system inherent to the liver, or if potentially you've looked at neural invasion, if that's something that happens in the liver as well as in other areas like the pancreas? It's a very good question. I think there is specificity within the liver content. For example, in our first paper by Colin Riordan, he found that B cells with Tolle and CCK in the spleen can make acetylcholine. But in his paper in 2013, he found that norepinephrine cannot activate B cells to make acetylcholine, but norepinephrine can activate T cells. Then Nastaran more recently looked at the B cells in the liver and showed that norepinephrine can do it. So I think it's quite context-dependent where these B cells are lying or the T cells are going. Okay. Got 24 seconds. And then the next speaker is Andreas Trump. Yeah. Thanks very much, Tak, for the introduction, and also appreciate that you pronounced my name is Andreas Trump with two Ps. It's quite important these days. And I'm really happy to be here, and thanks very much for the invitation for the 2026 AACR. So what I will do in the next 20 minutes, I will tell you about profiling and targeting innervating peripheral neurons in pancreatic cancer. And this is a story we have just finished, and I'll show you also some unpublished data. And the three heroes in my lab who actually have done this is Vera Thiel, then Simon Renders, a clinician scientist, who also actually now starts or runs the clinical trial, which I will summarize at the end, but we are planning here, and Jasper Panten, who has done all the bioinformatics. So this story started when Vera actually updated and optimized iDISCO protocol, which is used to actually make tissues or organs completely transparent. And you can read your newspaper through it, but the really beautiful part is that you can actually still label them with antibodies. And what she has done is used peripherin or other antibodies to label the peripheral nervous system. And then you can use light sheet microscopy to actually make 3D images of the complexity which happens on the peripheral nervous system in all the different organs of our body as well as in cancer, which is shown here. So as you all know, all healthy tissues, all healthy organs are innervated by the peripheral nervous system. The motor, sensory, autonomous components releasing neurotransmitters and trophic signals to main our tissue homeostasis and keep our organs functional. So it's not surprising at all that in arising cancer in all of those organs, actually you have a benign disease, it grows, it actually starts to contact blood vessels, but of course, it will also contact already preexisting neuronal cells actually in that tumor microenvironment. And therefore, it's clear that the PNS becomes automatically part of the tumor microenvironment. And that's been recognized by a number of researchers, and there has been as a large literature on potential interaction with metastasis, obviously cancer associated pain. There is a growing field of the interaction with the immune system, as we just heard from Tak, and inflammation, interactions with the tumor stroma, and of course, driving tumor growth. And the first data go back a little bit more than 10 years ago by Paul Frenette, who was the first in mouse models to actually cut the nerve to prostate cancer and actually has seen that actually by cutting the nerve, the tumor cells actually decrease growth, and there was this decreased metastasis. And by now, many others actually have followedAnd in principle, the sympathetic nervous system, adrenergic, motor, and sensory interactions are typically tumor-promoting, and parasympathetic ones are often inhibiting. But actually, there are exceptions, which we probably will hear later. So this generates now actually this new field of cancer neuroscience, which was initiated really by my old friend Paul Frenette and others, and is now driven from the central glioblastoma part from Frank Winkler and Michel Monje, Frank being actually also on my floor. And this is also in memory of my old friend Paul Frenette, who passed away much too early, and we were both working on leukemic stem cells, and he said, "Andreas, the real thing you have to do is actually work on cancer neuroscience." And this is now something which I, in fact, do. So if you look at the tumors, what you see is there is a lot of innervation, but what actually, in fact, you see in the tumor or in normal organs are the nerve endings. And these nerve endings are connected to an axon and connected at the end to the cell bodies. But where are the cell bodies? The cell bodies are actually not in the organs. They are not in the cancer, but they're in the peripheral ganglia. And so cell bodies of cancer or normal innervating neurons are located in peripheral ganglia, and for the PDAC, for pancreatic cancer, this is the celiac ganglion, CG, next to aorta, or the dorsal root ganglia, the sensory ones along the spinal cords. So knowing that, it is clear that the large data sets, TCGA, ICGC, 10X genomic studies, where you take tumor pieces and submit it to multi-omics, whatever, sequencing, actually, they all lack the information of the neurons because whatever you submit actually does not contain the peripheral ganglia where the cell bodies of the neurons are located. Now, pancreatic cancer, as you all know, is still a dismal disease. The five-year survival is in the range of about 10% to 12%. It's typically a RAS, mostly KRAS-driven tumor, and KRAS is during tumor progression. It's activated, it's duplicated, it's whole genome amplified, so RAS goes up. So the standard therapy is FOLFIRINOX, combination chemotherapy on paclitaxel and gemcitabine, in addition to surgical resection, which only 20% actually can profit from this. So all recent trials failed, but there is great hope with the new RAS inhibitors, and there are several talks also here at AACR from companies such as Revolution Medicines. They have really fantastic data, actually, that RAS might actually be the way to target pancreatic cancer in the future. So pancreatic cancer is characterized by a very high desmoplastic stroma, meaning that the epithelial cells can actually be only a minor component of the cancer, and stroma can make up to 90% of the tumor mass, which is quite amazing and actually generates obviously some issues. On top, by staining for neurofilament markers, in particular pathologists, but also surgeons actually realized that PDAC actually has hypertrophic nerves. But this is all in 2D, so how does that look like when we now look into an iDISCO and actually stain for the peripheral nerves actually in this human PDAC cancer, which has been grown endogenously by transplanting organoids onto endogenous mouse pancreas. And you see actually this very dense network. This is not blood vessels. This is really peripheral nerves which innervate the pancreatic cancer. But all those what you see are nerve endings, and actually none of the cell bodies of these are present within the tumor. And so what we then figured, well, we need to develop technology to identify and characterize the neurons at the transcriptional level. And so for that, Vera and Simon actually developed Trace and Seek, which identifies the transcriptomes or the proteomes of individual pancreas versus PDAC innervating neurons. And what they found is, or what they use is a tracer called Fast Blue, which turned out to be the best one. It's Fast Blue, which is injected into the pancreas or into the pancreatic cancer in mouse models. And actually, this Fast Blue is retrogradely transported back into the ganglia and actually stains exactly those neurons, which actually traject into the pancreas or into the pancreatic cancer. This can be isolated and looked at by fluorescence microscopy, but even more important, they can be isolated individual neurons. They can be sorted by Fast Blue and then sorted by a flow cytometer individually into each well. And from each well, you can then do downstream analysis, such as transcriptomic analysis, or we have also done now proteomic analysis. The technical details we have just now written together for nature protocols because this can be used for any neuron, for any situation, any mouse model, doesn't have to be cancer, can be heart attack or any other things. So it's actually widely applicable. So workflow consists of four key stages: the Fast Blue tracer injection, then the ganglion dissociation and enzymatic digestion, the neuronal enrichment, and single-cell FACS sorting into 384-well plates. And then you can do actually with these individual cells, any kind of multi-omics analysis you actually want to do. So these show us the Fast Blue injections. You get completely blue pancreas. There is actually no spilling over typically, and you can actually see that for the celiac ganglion, after 5 days, 7 days, 10 days, actually, it reaches actually the maximum at around 10 days, where the tracing actually reaches from the pancreas to the celiac ganglion. It takes a little bit longer for the dorsal root ganglion because they're also farther away.Now, this is how it looks like with this iDISCO. You have the pancreas, the spleen, and you have the celiac ganglion, the dorsal root ganglion over here, and this is how it looks through a microscope. And this is the celiac ganglion, which you have to dissect, and if you do trace and seq, you see that many of the cells have actually double labeling with a normal neuronal marker, but some actually don't. So you need specifically actually this trace and seq technology to separate neurons which go into your target organ and neurons which actually don't go into your organ. And then you do flow cytometry, you digest the ganglia, you use a live cell death, you use NeuO to separate or to specifically enrich for neurons, and then you use Fast Blue, and then these cells are then sorted into the 384 well plates and up for next analysis. And for the transcriptome, this is actually what you get. You get this complete heterogeneity of the neurons you find in the celiac ganglion, sympathetic ones as well as sensory ones. And you see actually the different subtypes you get within each of the ganglia. And on top of that, next to the identity of these neurons, you also get differentially expressed genes. And now we can actually fill in the missing piece by the genes which are expressed in the pancreas or in the PDAC, differentially in the celiac ganglion one or the pancreas or PDAC in the dorsal root ganglion. And we see a number of neuropeptides, neuroguidance molecules, DNA repair, signaling molecules, structural proteins, microtubule formation, and many others. So what is the real difference between normal pancreas and PDAC? And this is actually visible over here. So what you see is this massively increased sprouting in the pancreatic cancer. And this actually reminds us to these two trees. There's only one stem. Actually, the number of neurons is identical. There's no proliferation going on in the ganglia. So the number of neurons which traject into the normal or into the cancerous pancreas is actually the same. What you see is this massive sprouting, which actually increases the contacts between neuronal nerve endings and the pancreatic cancer cells. And this is actually shown over here again, in red, you see the image, and in yellow, you see actually machine learning enhanced image analysis. And you see actually this incredibly dense network of neurons within the pancreatic cancer. Now we, of course, validated this, and one of the genes we are very interested in is Lin28b. I'm a stem cell person. Lin28b is a classical stem cell gene, which is probably involved in reactivation of embryonic program in these terminally differentiated neurons, which probably increase neuronal plasticity, axonal regeneration, and synaptic reorganization. And we see also Robo slit signaling. It's classic for axon pathfinding, probably navigating of the neurons actually through the stroma, increased branching, and synaptic remodeling. Something we also saw that actually when we do this iDISCO and follow some of the nerves, that we see some going through the pancreatic cancer actually into the spleen. And so what Vera did is then actually do a double labeling, and indeed she finds that in the ganglia, then there are individual neurons which actually target both the pancreatic cancer or the normal pancreas as well as the spleen, which is quite amazing. And so we did then also double labeling of the liver and the pancreas, and actually you see that most of the neurons are individual to one of the organs, but you see also some which actually are targeting both, which means that PDAC reprogrammed neurons may also innervate the liver. They reprogrammed the liver, meaning that they might actually generate a pre-metastatic niche, and we know that pancreatic cancer actually very quickly and very effectively metastasizes to the liver. Maybe also the nervous system is involved in facilitating the growth of these cells in the liver. To summarize, we found that the neurons not only target and are in contact with the cancer cells, but they're also in direct contact with CAF cells, probably with vessels, and actually potentially also close to immune cells. Now, how important is this innervation? And can we actually block it? And do we have maybe already drugs which are already doing this to some degree? And so for that, we used 6-OHDA, which is a drug which is neurotoxic to adrenergic neurons, that's been used by many others. And if you do that and look at iDISCO, you actually only have the sensory ones left, but the sympathetic ones are gone. And if you do then growth assays in the PDX or in the KPC model, actually the loss of the neurons actually very much slow down the growth of these tumors. And then we wondered, well, how about now paclitaxel? Is that actually doing already something? Because we all know in the clinic that paclitaxel actually generates very severe neuropathy, so affecting the peripheral nervous system. And it's a microtubule stabilizer and thought to block mitosis, but maybe it actually also blocks these peripheral nerves. In support of that, there's a phase three clinical trial published where there is a linking neuropathy to PDAC outcome, and the patients who have very strong side effects actually had a much longer overall survival. Which doesn't prove anything, but actually suggests indeed that Abraxane may affect infiltrating neural function and thus may be inhibiting already some of the neurons in the cancer. And this we can actually also see in neoadjuvant setting whereWith a nab-paclitaxel treatment, these neurons actually are decreased. When we do trace and seq, what we see is that, indeed, the sensory neurons actually go down to zero in response to nab-paclitaxel and again, iDISCO, you see this untreated nab-paclitaxel. You have the sympathetic ones left, 6-HDA, you have the sensories left. It was not rocket science to actually combine the two. And if you do, then the growth assays in the mice, actually we see that there's a massive, really synergistic effect by actually blocking the entire nervous system, sensory and sympathetic. Actually, the tumors were barely growing at all. Now, also something fitting to the topic of the session here. We have denervation indeed turns immunosuppressive CAFs, cancer-associated fibroblasts, into pro-inflammatory CAFs. We see the proliferation is down, inflammation, T immune signatures are up. The lymphocytes start to actually migrate again into the pancreas. And by doing PD-1 inhibition in vivo, we see that this actually can be resensitized. So this may indicate that denervation may turn immunological cold pancreatic cancer into a checkpoint inhibitor responsive hot tumor. And there are also other data from Susan Kegg on beta blockers, which actually suggest that this indeed could be the case on top of the data we have generated. So then we decided, okay, so let's take all what we found and translate this into an early clinical trial, which we call synergy. It's a phase one, phase two multi-center trial, and we thought we have to do three things. First, to target the sympathetic neurons in the celiac plexus. This is actually shown here. This is a very busy area of the body, which also makes pancreatic cancer surgery so tricky. You can localize them by MRI or by actually PSMA PET to actually localize them, which allows maybe in the future for theranostic approaches. We then target the sensory neurons by nab-paclitaxel gemcitabine, which is anyway frontline or first-line treatment for pancreatic cancer. And we would add a checkpoint blockade, a PD-1 inhibitor after denervation. And so what we do now is actually we have to decide it, and maybe here, this is actually done as a proof of concept for the role of PNS. We thought actually to be able to block almost entirely the celiac plexus to see whether the microenvironment has actually a strong effect indeed, to pancreatic cancer. And what we use is actually a radiotherapy-mediated blockade of the celiac ganglion, using 25 gray. And this is a linear accelerator, which is coupled to a MRI, which is called MR-linac, which allow us to actually very specifically, only specifically, put the radiotherapy on the area of the celiac ganglion. And this approach has been used by others in the past. Not for therapy, but actually for reducing of pain, in patients who are terminally ill, and have a lot of pain, and this was very successful. And therefore, we thought that we can use this now and this approach, to combine with gemcitabine nab-paclitaxel and two arms of the PD-1 inhibitor. So the patients we will include are locally advanced pancreatic cancer patients. They are just not resectable anymore. They are hyper-innervated by neurons, and they have immune suppressive microenvironment. And the goals would be to test whether indeed the 25 gray actually indeed lead to a denervation together with the taxane. The primary endpoint is progression-free survival, and another endpoint we are quite excited about is secondary resectability. So the goal is to reduce the size of this tumor again to a moderate size, so that surgeons may be able to take out the tumor again surgically. And of course, response rate and immune response. So this is a phase one, phase two trial, which we run not only in Heidelberg, but actually it's a two-arm multicenter trial with six German comprehensive cancer centers. We're currently trying to obtain the finances for that and actually hope to start next year with this first cancer neuroscience trial on pancreatic cancer. So what comes next? Of course, the pancreatic cancer is not the only one which has an involvement of neurons. And we actually look at all of them. We analyze all of them and run trace and seq. We have done this already, so map the sensory innervation of healthy organs versus tumors. Here you see actually the data from the dorsal root ganglia. We have now identified 15,000 cells by flow cytometry and comparing healthy tissue, primary cancers as well as metastasis. And we are currently analyzing those data, in greater detail, which is actually quite difficult. And so we try to generate a neuro-pan-cancer atlas. We also will include now the nerves within the carcinoma ecosystem, which includes the CAFs, the vessels, immune cells, as well as the tumor cells to generate connectomes, to establish cancer nerve signals, signatures, which should establish common but also tumor entity-specific targets. We look at synapses. This has been identified now by our colleagues in Munich, Ekendemia, for synapses between sensory nerves and pancreatic cancer, but there might also be one with CAFs, is the electrochemical stimulation, and actually turn more of these data into early clinical trials. And with this, I thank you for your attention and actually take the credit. So credit is Vera Thiel, Simon Renders, who also will run, actually, this clinical trial synergy. Jasper Panten. We are in close collaboration with Frank Winkler, who is on our floor and our expert in neurology. This is done in my lab together with Martin Sprick. And this is all done at Heistem within the German Cancer Research Center. And this is Heistem. There are seven groups, two interested in stem cells, but also cancer neuroscience. And I'm just showing you while actually you prepare for questions, some more videos of organs and stained for the peripheral nervous system. Thank you very much and happy to take questions. Hi, Dani Engel from Salk Institute. Absolutely beautiful talk. So I was wondering, since the nerves can actually innervate both the pancreas and the liver, and there's such a high incidence of perineural invasion, have you actually mapped these nerves as paths of metastasis to that organ? Not yet. There's two things. What we have mostly looked at is actually the innervation within the pancreatic cancer. There's also perineural invasion, which is actually the opposite. So it's the nerve sheaths are innervated by disseminating tumor cells, which then integrate or basically migrate into the nerve sheaths, and that actually generates this massive pain in later stage cancers. And the cells actually also use the nerves as a freeway to actually move towards the liver. So this is a little bit of a different mechanism, but we are also looking at that. Hi, thanks for that. John White, UCSD. So the pancreas is also innervated by enteric nerves as well. When you did the tracing, of course you're looking at the- These we cannot look at. Those- You can't trace back- No ... using the same mechanism for- No, because they are basically inside the pancreas, right? So there are some cell bodies. That's what you're talking about, or? They go to the gut, yeah. Ah. Yeah. So not going to the ganglia, but actually the innervation coming from the enteric nerves from the gut that are reaching the pancreas. Yeah, so in principle they can also be traced, but obviously you need to know actually where the cell bodies are to actually exactly localize them and then isolate- For the install or take, yeah ... isolate the areas. So I think that's also relevant for translational science as well, because if you're looking at which nerves, of course, I'm sure you see that they are tracing, so a good portion I'm sure are going to these ganglia. But if you're looking to intervene and treat, then if there's a contribution from the enteric nervous system, maybe that's an access that you'd want to look at as well. Yeah, absolutely. Very good point. But it's amazing work. Thank you so much. Yes, thank you. Looking forward. Hi. We've been treating cancer for a long time. When you treat the cancer with the chemotherapy, what do you think is happening at the tumor neural interface, the tumor cell neural interface, if this input is coming and innervating them through the mediators, cholinergic or muscarinic mediators released into the TME? Is it getting more stronger, you think, like to make the tumor cells survive the chemotherapy? Yes, probably. There are many different factors which we are trying to figure out because there are clearly signaling pathways which link to the CAFs, which link also to the tumor cells. Definitely, there are more signals which actually drive tumor genesis as well as actually provide survival signals from the nerve to the tumor cells. Okay. And the second question is, these immune cells, they are moving, they are not fixed like CAFs. So, say it is like target on the go and the neural modulation of these T cells and the B cells the previous speaker spoke also, how is it able to regulate that input into those motile cells? And also one more thing is the tumor excluded immune cells, tumor included immune cells we talk about. Right. That's what I was showing in passing a little bit, sort of the nerve structures when the CAFs may actually lead to this immunosuppressive microenvironment, because when you block the nerves out again actually, then you change the CAFs in a way so that the immune cells actually can be able to migrate in again. That's something what we see, what we don't understand what the details are, and which we are currently looking at. Thank you. Thank you for the beautiful talk. You've dealt a lot in the immune cells, how the innervation affects the immune cells that are infiltrated in the tumor mass. Have you thought about how those innervation affect the tumor cells, how they interact with one another? And- How the tumor cells interact with other tumor cells? Or- With nerve cells With immune cells. With the? Nerve cells that are innervated. Oh, the nerve cells, yes. Well, there is now first evidence that synapse-like structures indeed actually are occurring next to trophic signals and specific cytokines which actually may be released. There might be also some atypical synapse formation, which is quite exciting. It has been shown in the central nervous system by Michel Monje and Frank Winkler. Right. But now there is more and more data that there are also synapses in the periphery, which actually add. So maybe the cells are actually fired, literally fired up by the nervous system. Yeah. My related question is, do you think that has to do with anything with neuroendocrine transformation that are often seen in pancreas? Cannot exclude it. The typical neuroendocrine is in lung cancer, in prostate cancer, and sometimes also in breast cancer. The neuroendocrine one is typically a different type of cancer. It's not a PDAC. I see. All right. Thank you. Real quick question. You're measuring the mRNA at the cell body level, but it's going to be different at the end of these neurons. Do you know anything about the different distribution of mRNA there than in the cell body? Very, very good question. In the neurons, there is also a transport of RNAs actually to the nerve endings, and there are ribosomes also at the nerve endings, which are producing. So we are currently actually collaborating with people at EMBL to actually look at that as well. So I think that's an important part which we are indeed partly missing by doing trace and seek. All right. Thank you very much. The last speaker is Sebastien Talbot. Okay. Well, first of all, thank you, Professor Mack, for the invitation. I'm super excited to be here today. So my lab study neuroimmunology, we do that in various contexts, including in allergy and asthma, as well as in cancer. And of course, today we're going to focus on cancer. So I'd like to start the talk by giving you a little bit of historical perspective that kind of underpin the field of cancer neuroscience. So here is data that actually led to Rita Levi-Montalcini 1986 Nobel Prize, where she cultured chick embryos, so sensory ganglia from chick embryo, in absence or in presence of sarcoma. And you can see that when it's in the presence of sarcoma, you see those neuron that actually start growing or projecting these axon. Eventually, what she found is that those cancerous sarcoma was secreting the nerve growth factor, and this was actually deriving this very strong production of nerve. So I think this is really underpinning the idea that cancer cell can really attract these nerve, to eventually infiltrate tumor. This is another picture from J.X. Mead in 1897 that actually showed, again in sarcoma, the presence of nerve. So we know for a very long time that there is nerve inside of cancer. We did not really know what were these nerve doing until the emergence of the field of cancer neuroscience as it was presented before by the two other speaker. But it has been postulated that the nerve may be able to directly interact with cancer cell and drive tumor growth. Perhaps they interact with blood vessel and control angiogenesis, or perhaps they actually interact with immune cells and control the function of the immune cells against the tumor. So my lab focus on sensory neuron, which we call also nociceptor neuron. Nociceptor are actually a subtype of neuron specialized in detecting danger. Classically, they will detect, let's say, noxious heat. They do that by expressing specific ion channel receptor. When these ion channel receptor are activated, you will have an influx of different ions, generation of action potential that travel from the periphery all the way to the brain. But something that's perhaps a little less appreciated is when these action potential travel actually from the periphery all the way to the cell body. When they reach axonal branch point, those axon potential can actually travel back to the periphery, where they will lead to the local release of neuropeptide. Neuropeptide is the thing that they use to communicate with other neuron or other cells in the tissue. Classic neuropeptide are substance P and CGRP. So what happened is locally, this substance P and CGRP will act on capillaries. They will increase blood flow, increase vasodilation. They can increase the chemotaxis and adhesion of immune cells, and eventually they will loosen the tight junction of these blood vessel, and this will allow the immune cells actually to get out of the capillaries and come in the microenvironment. And the consequences of this is it will put in very close vicinity immune cells and sensory neurons. So this allow for this local crosstalk between the nerve and immune cells. So this local crosstalk has been shown to be super important, control immune responses at various barrier site. Classic barrier site are, for example, the skin. A lot of work has been done by the lab of Isaac Chiu in the context of bacterial infection. There's also interaction in the lung. A lot of data has been done in the context of asthma, including from my lab and also at the gut, again during bacterial infection. So neuroimmune interaction are important to regulate immune function. For example, during, as I mentioned, bacterial infection or viral infection or wound healing. But what happen in the context of cancer? So we decided to study this because it was not really known. So we chose melanoma as our working model. Why we chose melanoma? Because it's a great model of immunosurveillance, and we know that the skin is densely innervated by sensory neurons. We are also not expert in cancer, so we thought this would be easy enough to actually measure tumor growth. You can just use a caliper and looked at the growth on the skin. So what we normally do is classic model, for example, B16-F10, that we inject orthotopically in nociceptor intact in ablated mice, meaning that we have mice in which we can specifically eliminate subtype of neuron. Then we typically look at the presence of itch or pain, which are marker of sensory neuron activation. Eventually, we looked at the presence of nerve that will infiltrate the tumor. At endpoint, we harvest the tumor, and we immunophenotype those tumor, and we looked at immunity. So here is one of the first experiment that we did, is we give melanoma to our nociceptor intact and ablated mice. To eliminate nociceptor, in this case, we use a TRPV1 cre diphtheria toxin. So this mice actually, as I mentioned, will actually kill specifically pain transmitting neuron and mainly the neuron that produce neuropeptides. So what we found is that in absence of sensory neuron, we have smaller tumor growth, and we have longer survival. And when we immunophenotype those tumor, we found that there were more CD8 T cell per milligram of tumor. But most interestingly, we found that there were higher interferon gamma positive CD8 T cell, and it seems that those T cell were less exhausted. So lower level of PD-1, LAG-3, and TIM-3 positive CD8 T cells. So this suggests that perhaps the neuron control immune function. But we ask ourselves, can the cancer cell also impact the function of neuron? So to study this, we decided to use a co-culture approach in which we co-culture DRG neuron in the presence or absence of melanoma.We did that for 48 hours, and then we FACS purified each cell type and RNA-Seq them. And we found that when the neuron were co-culture with the cancer cell, they drastically overexpressed many different neuropeptide, including Calcitonin which encode for CGRP, so this classic neuropeptide. So as I mentioned, neuropeptide are the mediator by which neuron impact immunity. So we ask ourself, can this CGRP would be able to impact T cell function? So we looked at this in many different ways. I'm going to show you one which is the simplest one. We harvest basically naive splenocyte CD8 T cell, we make them to be cytotoxic, and we expose them to CGRP. And what we found is that CGRP seems, again, to drive the exhaustion of those T cell, meaning they express higher level of PD-1, LAG-3, and TIM-3, and lower level of interferon gamma. This effect was actually absent when we harvested a RAMP1 knockout CD8 T cell, RAMP1 being the receptor for CGRP. Not showing you all the data, but we eventually shows that if you put CGRP on OT-1 CD8 T cell, they were less able to eliminate B16-F10 ova. So we really think that CGRP is able to drive exhaustion of the T cells. But to confirm this in vivo, we decided to use a RAMP1 knockout mice. RAMP1, as you may know, have no B and no T cells. So what we did is we transplanted either a RAMP1 wild type or RAMP1 knockout CD8 T cells, so T cells that do not respond to CGRP. So in these mice, in which the T cell cannot respond to CGRP, we found that there's smaller tumor growth. In some of our mice, we co-transplanted both RAMP1 wild type and RAMP1 knockout CD8 T cell. And using CD45.1 and CD45.2, we could track each subset of T cell within the same tumor. And what we found, I think, is pretty interesting. Again, lower exhaustion in our RAMP1 knockout CD8 T cells. So less PD-1, LAG-3, TIM-3 positive CD8 T cell. But to really confirm that, we actually FACS purify and then RNA-Seq those T cell. And again, in the knockout CD8 T cell, there's lower expression of Tox, Eomes, and other markers, suggesting again that there's perhaps lower exhaustion of CD8 T cell. So is this relevant for patients? So if you looked at TCGA data, you see that high RAMP1 expression melanoma seems to be bad for the patient. But most interestingly is when you looked at single-cell RNA-Seq data. So Aviv Regev had two paper published from melanoma patient. So because it's single cell, you can parse T cell in different populations. So we parse them either expressing high level of RAMP1 or low level of RAMP1. So within melanoma patient, if you express high level of RAMP1, the T cell, again, seems to be more exhausted. So higher expression of PD-1, TIM-3, LAG-3, CTLA4. Not showing you all the data. The paper has been published now for a few years. But what we found is that melanoma is densely innervated by sensory neuron, that melanoma activates sensory neuron by the secretion of SLEEPY. The sensory neuron will locally release CGRP. It will act on RAMP1 positive CD8 T cell and drive their exhaustion. And what we eventually found is that if you block the sensory neurons using different drug, for example, QX314, or if you block the RAMP1 receptor using the IBN4096, you prevent or you reduce T cell exhaustion, which are then able to fight better the tumor, so we have smaller tumor growth. Now I'm showing you some data that we looked at neuron signature. As we all know, when we looked at TCGA data, we normally looked at only one marker. So Ben HK, actually created at University of Toronto, created that tool which he called predictio.ca, which you can access. It's freely accessible. It's pretty cool, where you can look at different expression of different marker. So we wanted to look at a neuron signature, in this case, classic marker of neuron, including Trpv1, Nav1.8, RAMP1, and we found that our signature seems to be very bad for patient survival. It seems to be even worse than T cell exclusion. So having this neuron signature within cancer excuse me, seems to be very bad for patient. Then again, if we looked at the signature and whether a patient would respond better or not to IO, we found that only for melanoma, again, having this neuron signature seems to be a predictor for bad responsiveness to immunotherapy. So I'm going to show you data that's been generated or led by my colleague Alex Birbrair at Wisconsin Medicine. And he ask itself what happened with melanoma when they metastasize to the lung. So all our data so far were in the primary site. What happened in lung mets? So to do this, he used, again, B16-F10, in which he injected intravenously to a nociceptor neuron reporter mice. In this case, it's a Nav1.8 Cre TD tomato. And what he found in comparison to normal lung, the lung that were actually bearing those metastasis, they were more innervated. So there's perineural or perimetastatic innervation, but there's also intrametastatic innervation. So there's presence of neuron in and around those lung mets. So what would happen if you eliminate those neurons? So if you use RTX, which is a drug to eliminate sensory neuron, he said there were less metastasis. So then he wanted to confirm that using genetic model. In this case, he used chemogenetics. To do this, he used a Nav1.8 Cre-HM4QDI, in which you can specifically, in this case, block sensory neuron via the injection of CNO. And he found that if you inject CNO in those mice, you're actually decreasing, again, lung metastasis. So what is happening in those lung met? What is the crosstalk between the nerve and the lung mets? So first of all, if you looked at TCGA data in patient that have lung metastasis, you can compare a live patient and patient that eventually pass. In patient that pass, he saw that there were higher expression of RAMP1. And then again, if you looked at survival data, it seems that in patient that express high level of RAMP1, they seems to survive less.But what was super interesting is when you actually compare primary melanoma to a melanoma that actually metastasized to the lung. So if you harvest these guys and profile them using qPCR, we see that the melanoma that actually met to the lung drastically overexpressed RAMP1. This is about 10 or 20-fold higher expression of RAMP1. So then he ask itself, what happen if we remove RAMP1 expression of these cancer cells? So he used CRISPR-Cas9 to knock out RAMP1 out of YUM 1.7. And he sees that in absence of RAMP1 expression on those cancer cell, there's also less lung metastasis. Another thing that he discover is that RAMP1 is not specifically only expressed on the cancer cell. They seem to be also expressed by PDGFR beta positive cell. PDGFR beta is our marker of pericytes. And again, if you compare a normal lung to a lung that actually is bearing lung mets, the pericyte drastically overexpress RAMP1. Here it's 100-fold overexpression. Another thing that he discover is that CGRP positive neuron are in very close proximity to those pericytes, suggesting that perhaps the neuron is able to control pericyte function. So to look into this, he decided, well, he use the Cre-Lox toolbox to actually flux out RAMP1 out of the pericyte. And what he discover is that in absence of RAMP1 on those pericyte, there's smaller tumor growth. Then one of the question of one of our reviewer was, is this relevant for patient? And he asked us a very tough question where he wanted us to do a clinical trial in metastatic melanoma in the lung, which as you may think, this would be pretty complicated. So we're trying to go around this question where Alex was able to access about 100 million patient data set. And he looked at whether those patient were actually having melanoma or not. And he found about 40,000 patient that had melanoma, and some of those patient were actually treated with RAMP1 antagonists. So now there's RAMP1 antagonists on the market, and they are used to treat migraines, for example. So the acceptance of this drug is pretty recent, but he was still able to find about 120 patient. But the cool part was that those patient that were actually treated with those RAMP1 antagonists seems to actually be doing better than the control patient. So of course, these data have a bit of caveat. We don't know when the treatment was started. We don't know when the indication of the treatment. But nevertheless, it seems that using RAMP1 antagonists potentially is helping patient with lung metastasis, or actually metastasis in general. This is not only lung. So again, not showing you all the data, but something really cool that he found is that primary melanoma, when they met to the lung, they start overexpressing RAMP1. They can then respond to CGRP produced by these nociceptor neuron. He found that this will drive tumor growth. He also found that pericyte will also overexpress RAMP1. They will receive signal from nociceptor neuron, and that will drive proliferation and angiogenesis of the tumor. So now I'm going to show you, over the last five minute, another story that's now published. Wanted to see how nociceptor neuron control tumor growth beyond CGRP, and this is work that's been carried out in collaboration with Moran Amit at MD Anderson. So what we decided to do, and it's very similar to what Andreas actually showed you, is how melanoma or whether melanoma are reprogram, or whether how melanoma reprogram tumor innervating neurons. So to do this, again, very similar to what Andreas did, we injected melanoma in the INPA of the mice. Why the INPA? Is because we can specifically trace these neuron to L3 to L5 DRGs, harvest those L3 to L5 DRGs, purify nociceptor neuron using tdTomato, and then single-cell RNA seq them. And one of the thing that we found is that, well, there was a unique population of neuron in a melanoma-bearing animal, and there's a lot of differentially expressed genes across different subset of neuron. But really, this unique population of neuron were actually expressing ATF3. ATF3 is a key transcription factor that's expressed specifically in injured neurons. So this is very classic. If you have nerve injury, ATF3 is absolutely essential to allow neuron to actually regrow in the nerve stump. So finding this ATF3 in our tumor innervating neuron was very cool, so wanted to follow up on this. So first of all, we use a ATF3 reporter animal to confirm that, yes, tumor innervating neuron were actually expressing ATF3. But most interestingly, we wanted to see what happen if you flux out ATF3 from nociceptor neurons. If we make them resistant to injury, what happen? So we found that there were smaller tumor growth and smaller tumor weight. And then we ask ourself, how is this actually happening? So we decided to profile immune cells inside of those tumor, and we found that in our injury-resistant mice, we found there was higher expression of, or higher infiltration of interferon gamma positive CD8 T cell. And we also did single-cell RNA seq of this tumor. And what we found is that there's lower level of LAG3 expression of T cell, but also that tumor-associated macrophages, their potential to be immunosuppressive seems to be less, meaning they express lower level of IL1B, CD86, and CD80. Because of time, I'm going to just skip this one. But why is this actually interesting is because Moran had a population of patient, a cutaneous squamous cell carcinoma patient that were treated or not with anti-PD-1.And he had sample before, during, and after treatment. And of course, the patient might respond or not to the treatment. But the really cool part is that in non-responding patient, he found that there's a lot of perineural invasion. Perineural invasion, as it was discussed by Andreas, is basically, the tumor will kind of engulf the nerve, and that will lead to actually potentially injury to the nerve. And really this is what we actually found. We characterize those perineural invasion, and we're able to find that in the non-responder, they seem to overexpress ATF3. So both in our mice, but also in our patient that do not respond to treatment, ATF3 seems to overexpress in these neuron that have perineural invasion. So then we decided to create a signature for these injured neuron, and we wanted to see, well, is this nerve injury signature correlate with poor or bad prognosis for patient? And across different cancer type, including our cutaneous squamous cell carcinoma, PD-1 treated melanoma patient, gastric cancer treated with PD-1, as well as another cohort of melanoma patient treated with PD-1. In every cases, nerve injury seems to be correlated with poor prognosis for the patient. So based on this, we went back to our cutaneous squamous cell carcinoma patient, and we decided to do Visium transcriptomic on these slides. So we had three different gene profile. One that's nerve injury in green, one that's immunosuppression in red, and one that's good, so meaning good anti-tumor immunity in purple. And we find a very great overlap between nerve injury and immunosuppression. And when there's no nerve injury, we found really good presence of anti-tumor immunity. Meaning that, again, it seems that this nerve injury seems to be very bad for anti-tumor immunity. Again, not showing you all the data due to time, but what we found is that during tumor growth, including melanoma, the nerve become injured. They overexpress ATF3, this injury marker. Downstream of this, the nerve will overexpress galanin. We found that galanin actually drive T cell exhaustion, but we also found that the nerve overexpress interferon-1 and IL-6, which then will act on tumor-associated macrophages. So another cool part, the paper was published last year, but what was really cool to see that in the same issue, a different group completely found that also nerve injury was present and seems to be bad for glioblastoma progression. So it seemed that this nerve injury signature, it's pretty constant across tumor type and also perhaps in glioblastoma. So I think I'm out of time here. Just concluding, yeah, I'm just going to stop here, actually. All right. Thank you, guys. I'm going to thank the people that did the work in my lab. Mohamed, Mariam, and Tuani, our collaborator, Alex in Madison, Wisconsin, Moran at MD Anderson, Payver Mir in SUFA, Ronit Rapkin at UPenn, Nicole Schiff at Pitt, Vijay Kuchroo at HMS, Karen Dixon in Basel, and our student, Elisa Pretto. So on this, I'm going to welcome any questions. Sorry to went over a little bit longer. Really interesting talk. Is it true of the liver also, and is it the same mechanism, Ramp1 CGRP? It's a great question. Liver is a bit of a weird organ, actually. We don't see too much sensory neuron innervation in the liver. If you do, actually, I disco, at least I did some of that during my post-doc. I did not see too much sensory neuron innervation in the liver. That said, there's other cells that produce CGRP, so potentially that CGRP Ramp1 could also be present in liver cancer or liver in general. Second question was about Nav1.8 and whether it's another kind of neuron, and does it overlap with the Ramp1 axis? Yeah. So sorry. So Nav1.8 is a classic marker of nociceptor neuron. It labels about 80% of them. And within those Nav1.8 neuron, about 25% are CGRP positive, so the one that would act on Ramp1. So the one that would act on Ramp1 is a smaller subset of the Nav1.8. Thank you. Thank you. Yes. I wanted to write that down. Sorry about that. I hate to open the can of worms of T cell exhaustion- Yep ... but I was curious about your CGRP co-culture and how you defined CD8 T cell exhaustion, if you had a subset or a gene set. Yeah. In our case, we looked at PD-1, Tim-3, like 3 positive, interferon gamma negative, but, I guess, as you said it's a bit of a chemo war. One of the experiment that I liked, I think was the best one to show that it was a B16F10 ova OT1 CD8 T cell co-culture plus minus CGRP, and then that was actually decreasing by 50% the elimination of the cancer cell. So whether it's true exhaustion. So the way we rolled that is effect immuno surveillance, so it's kind of a little bit broader than, I guess, true or absolute exhaustion. Yeah. Quick mechanistic question. ATF3 can be activated through different pathways. Do you have an idea which pathway is responsible in this case? That's a great question. This is something we haven't really looked into. Yeah, I think that's really open to be tested. Yes. Thank you. Sorry. Yep, go for it. Talk about this tumor cells pumping various autoantibody exosomes. Yep. And in this case, the neural tumor cells can also exosomes. Yeah. And did anybody look at them? What are those, are those exact exosomes between the two-way transfer? Yeah, so Payver Mir actually had a Nature Comm paper where she showed that head and neck tumor were innervated because they were secreting exosome, and that was actually driving the innervation. And Moran had a 2021 Nature paper looking at exosome as well in head and neck, but he looked at a subset. It was a microRNA, which I don't remember right now which one it is, but if you look him up, he did look at a specific microRNA. That I'm sorry. Okay, thank you. So I just want to thank the speakers