2026-04-18-am-senescence-in-cancer

Date: 2026-04-18 Primary topic: single-cell-spatial-omics Cross-referenced from: single-cell-spatial-omics Word count: ~12,529

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 morning, everyone. I think we'll go ahead and get started. Thanks for grabbing your coffees and getting here early for this 8:00 AM session. We appreciate it. This is Educational Session Two, Senescence in Cancer. I'll be your chair. I'm Marcus Versetti, coming from UMass Chan Medical School. Just to go over some logistics, this is a 90-minute session broken into 30-minute increments. Each speaker will have 20 minutes presentation, 10 minutes for question. And we encourage a lot of discussion, so please either come up to the mics for the question, or you can scan this QR code or use the app and we can read your questions online as well. So we encourage you to have a great discussion with us. One little hiccup, so unfortunately, Jennifer, who was our first speaker, had an issue with her flight, so she can't make it in person, but luckily, she was able to record her talk. So you'll see her recorded talk. And Drew Pardal, one of her colleagues, will be answering questions for her. So we really appreciate Drew stepping in. And I'm also joined by my colleague, Clement Smith, as well. Just a quick intro to senescence. We know the field has really changed dramatically over the last 20 years. First described as a cell cycle arrest pathway, we know it to be a potent tissue remodeler, immune remodeler with both pro and anti-tumorigenic effects in cancer. The development of single-cell and spatial technologies now, and also senescence reporter mice, have really allowed us to dive in to this heterogeneity where we can find basically any cell in the body can undergo senescence from stresses we get every day or stresses that accumulate over a lifetime, and they can impact cancer therapies as well. So our hope is that dissecting the mechanisms by which senescent cells contribute to progression or clearance, we can design rational therapies you might have heard of called senolytics or senomorphics to target senescence to make it more tumor suppressive. As you'll see from the speakers today, I think it also takes a lot of multidisciplinary approaches from different cancer spaces to really tackle this complex question of senescence in cancer. So we will get started with our first talk today from Dr. Jennifer Elisseeff, who is Professor of Biomedical Engineering at John Hopkins University. Her talk will be a great intro to our session on defining senescence and implications for cancer. Good morning. I'm sorry I'm not there in person. I had some flight challenges on the last flights heading from East Coast. So, unfortunately, I'm recording, not there in person, but I did plant a collaborator in the audience to answer any questions that may come up. But I'm really happy to be able to share our work today in the field of senescence. Here are my current conflicts and previous conflicts, partially related to the presentation. So I'd like to present a little bit of background first, our approaches in tissue engineering, because they really influence how we approach senescence. And then talk about our work in senescence in articular trauma and arthritis and fibrosis. And from that experience, how it led us to look at ways we can better define what is a senescent cell and the concept of senotypes. And finally, I'll discuss senescence and what we've learned in those other models and the potential implications in cancer. So I come from the tissue engineering field, as I mentioned, and we spent many years looking at biomaterials that could serve as scaffolds for things like stem cells, and that we could integrate biological signals into these materials. And I mention particular stem cells because that really has guided a lot our concept of cell states and proliferation versus differentiation. A cell that's proliferating usually doesn't differentiate, and when we think of a senescent cell that's secreting actively lots of factors, it's really in a differentiated state. So we translated technologies in the area of cartilage and soft tissue fillers, and these experiences in the clinic really guided us to the immune system and the concept of regenerative immunotherapies. So today, we think of tissue engineering as manipulating immune stromal networks that are sensing and regulating tissue structure, and you'll see that senescence plays a role here, and we think it's an interesting node and a hub for these interactions. So we first started characterizing what is a natural immune response after tissue injury, and what does it look like in a healing wound versus a non-healing wound characterized by fibrosis and inflammation, and can we use this information to redirect that non-healing wound to a healing environment? And this has applications in cartilage and fibrosis, and we recognize that the tumor was described many years ago as a non-healing wound. In a healing wound, we can look at things like muscle and pro-regenerative therapies that are stimulating repair, but then work has shown that tumors that are responsive to therapies such as immune checkpoint blockade have a phenotype of wound healing and wound resolution. So if we look at wound healing and kinetics, the wound-healing process has been well-characterized in the context of the immune system, stem cell mobilization, wound closure. But senescence is also a part of the wound-healing process. So in a healing wound, you'll have senescence initially increase and then decrease. But in a non-healing wound or a wound characterized by chronic inflammation, senescence or markers of senescence remain high.So senescence in hard-to-repair tissues. As I mentioned, we started in cartilage repair for a tissue engineering application, but we recognized even if we had a perfect therapy to rebuild tissues, if there was inflammation, such as interleukin-1 beta, it would not repair very well. The tissue engineering wouldn't work. At the same time, collaborators Nat David had recognized that senescent cells could be an interesting therapeutic target for a number of applications, in particular osteoarthritis, where senescent cells had been seen in cartilage explants being taken from patients undergoing knee or hip replacement. So what is a senescent cell? Well, the history of the field has really looked at senescence in vitro. So first looking at the Hayflick limit and replicative senescence. So after cells have been cultured for a long time, they stop proliferating, as is connected with decrease in telomeres. It was also recognized there was stress-induced senescence that could be stimulated by oxidative stress, genotoxic, and oncogenic stress, and the proliferative stress being another mechanism. But much of this has been studied in vitro. So these cells were characterized with a number of markers. So beta-gal, for example, HMGB1, movement from the nucleus to the cytoplasm, and p16, cell cycle related gene, was seen to be upregulated in some cells, although well recognized to be an imperfect marker. But many of the tools are built around p16. So these cells are in growth arrest, but they are secreting a senescence-associated secretary phenotype. This SASP is what is implicated in the function of these cells to increase tumor production or related to the diseases associated with aging, such as heart disease, diabetes, and arthritis, as I mentioned. However, these cells are also recognized to be an important part of embryological development and tissue repair. So how are these cells both good and bad? Well, research has shown that senescent cells are a part of wound healing, and that is through secretion of PDGF, platelet-derived growth factor. And then our work on senescence as an example where the accumulation of senescence is associated with disease in the joint and arthritis. So from the arthritis perspective, we looked at an injury, anterior cruciate ligament transection, where the cartilage lining the surface of the joints degenerates after that injury. It's a chronic wound. Senescent cells, as I mentioned, were recognized to be a part of the arthritis process. And when we did senolysis or removal of senescent cells, either through molecules or through transgenic animals, we could restore the cartilage. And we observed reduction of SASP factors, reduction of pain, and we were very surprised, we also saw an increase in tissue repair when those senescent cells were removed. Now, luckily, I had an immunology collaborator who recognized some of these SASP markers, IL-6 and IL-1 beta, and told us, "Well, these molecules are connected to T cell differentiation." So we looked in that same model and saw increases in IL-17, both in the joint and in the draining lymph nodes, increasing with injury and disease. And when senolytics were added in both young and aged animals, a decrease in IL-17. So there seemed to be a connection between senescence and IL-17. So to further explore this, we induced senescence artificially in vitro and cultured these cells next to naive T cells. And in the presence of TGF-beta, we see a significant increase in IL-17 production by these T cells. We also took Th17 cells that we skewed in vitro and cultured them next to healthy fibroblasts and saw an increase in p16 and SASP factors, including IL-6. And we phenotyped these cells a little bit more and have a signature of an inflammatory-induced senescence from this in vitro model. So we can use inflammation as an additional mechanism for induction of senescence that may be more physiologically relevant than some of the others that I listed before. So again, when we think of tissue structure, this communication between the immune system and the stroma that's regulating the matrix and vascularity, senescent cells are playing a big role in that interaction. Now, continuing on the matrix discussion of tissue structure, senescence is associated with fibrosis in murine models and in human tissues. These are markers for p16 and p21, so associated with fibrotic diseases in these tissues. And we can see in other fibrosis models, this is a fibrosis around breast implant, and p16 is in pink here, and we can see p16 cells stretched out with a fibroblastic morphology. We also see p16 cells around the vasculature that is adjacent to this fibrosis. So we're interested in this network and understanding what a senescent cell is, and we're a part of the SenNet program at NIH that's looking to define what a senescent cell is in vivo in tissues, in both mouse and human tissues.So we first used the biomaterial model to study fibrosis, and this is a nice reproducible model where we can get consistent fibrosis. We have the same inflammation and abnormal vascularity. And you can see here that we have red p16 positive cells sort of adjacent to macrophages associated with the fibrosis stretched out again like a myofibroblast. But then we also see some associated with vascular structures, and you'll see that this is a common theme. So, yes, single cell, of course, is popular, and sometimes you can see p16 expression in single cell, particularly with the newer techniques. But in general, it can be hard to find p16, and again, as I mentioned, it's not a completely accurate model of marker for senescence. So in these fibrosis models, we can look at the cell types there. But in order to really capture p16 cells, we used a genetic model developed by Jan van Deursen and Darren Baker, p16 Cre. And with our biomaterial model, we could see that, number one, as it's recognized in the field, a small number of senescent cells are present. And when we look further at which cell types are senescent, we see a lot of stromal cells that are senescent, CD45 negative and CD29 positive and CD31 negative. There are also some I'll show you later in the tumor model that are CD31 positive. So when we pull out and sort just that subset of senescence, CD29 positive, we can perform bulk sequencing and get a signature which we call senescence signature. And when we look at what's there, we saw a lot of markers, enrichment in vascular-associated genes, but also chondrocyte differentiation genes. And while we worked on cartilage before, this is not a cartilage environment. When we look at how we can capture and identify senescence in single cell, we use transfer learning to then look at that signature, and in each cell, is that cell similar or different to that signature? So we can see senesc positive cells. This is the fibroblast population here, both activated and myofibroblasts. And then we see pericytes, a smooth muscle cell population that's also senescent. If we look at markers of these cells, in particular SAS factors in these subtypes, we can start getting phenotypic information of the activated fibroblasts versus the myofibroblasts. There's also an inflammatory fibroblast and a perivascular senescent cell or pericyte, which each senescent cell is expressing different markers. But single cell and the clustering is a way to bin cells into a certain phenotype, but it doesn't tell us much about patterns. So working with Ilana Furtak, we used her CO-GAPS program using non-negative matrix factorization to instead look for patterns of gene expression that can cross these cell identification clustering. And so we see patterns of gene expression within those senescent cell types, and we can connect that with spatial data to start seeing these niches where the senescent cells are and which cells are around there. So we have a fibrotic niche, a vascular niche, and an immune niche. And the cartilage niche is in the fibrosis, but it's expressing markers that are typically associated with cartilage and tendon tissue, which is avascular, similar to fibrosis. So just to give a summary, and this will be a preprint soon, we can look at transcription factors, cell surface markers that might be unique to these different subtypes of senescent cells, and SAS factors that can be associated with extracellular matrix, matrix degradation, immune factors, and growth factors of these different senescent cells or senotypes. Now we're very interested in how these cells are communicating. So senescent cells are a particular node in the network of cell communication. So we use a program that we developed called Domino Signal, which looks at transcription factor activation associated with receptors, and then connecting that with ligand expression. So a little bit moving backwards compared to some other standard receptor ligand programs. Now if we look at this in a muscle repair model, so good repair in a young animal versus poor repair in an old model. If we look in a young animal at communication, here's our senescent subtypes down here, we can see communication through a number of cell types, the immune cells, vascular cells. But in the old animal, it's really taken over by T cells, and this cartilage-like are really highly fibrotic cell types. So the senescent cell type and its communication is important. Now finally, interest here with tumors, the tumor microenvironment being characterized as a non-healing wound. Matt Wolf has done work looking at, for example, the lung tumor biopsy where we can see p16 positive cells in the stroma where there's fibrosis, so the pink cells here. And these are different than, of course, the tumor-associated senescence, which has been studied.And here's a close-up of these fibroblasts. But we still see these vascular-associated cells. And in the tumor model, we do see some senescent endothelial cells and pericytes. But I'm going to talk about something a little bit different. We see senescence also in fibroids, where we, again, have the fibroblast, vascular-associated in combination. So it's important when we talk about senescence in the tumor environment, that there are multiple senotypes. And the matrix these cells produce is important for enhancing tissue mechanics. For example, a fibroid is much stiffer with that increased senescence. And you're up-regulating mechanosensing related transcription factors. So we can build this model with our networks to look at communication. This is an example in the fibroid of senescence immune factors that are modulating the immune system and bringing in the immune cells, vascular-associated factors, again, contributing to both fibrosis and to the immune cell migration. But I mentioned early on, can senescence be beneficial, and what are the implications here? So, we looked at senescence in the context of different wounds and its association with immunotherapy response. So cancer immunotherapy has clearly been transformative, but many patients do not respond. So understanding why these patients don't respond and the mechanisms of resistance may help guide us in improving the impact of immunotherapies. So we were interested, coming from the wound healing tissue engineering perspective, how different injuries might impact the response of a distal tumor to immunotherapy. We looked, for example, at a muscle wound. That's a preprint that is available. So muscle wound will increase tumor growth and decrease responsiveness to immunotherapy. And the other wound model that we know how to do in the laboratory, as you know, is an arthritis model. So we tear the ACL at various time points before injecting a tumor. And what we found that opposite to the muscle wound, that the articular injury decreases tumor growth a little bit, but actually has a significantly great increase in response to immunotherapy. So a B16 tumor model, you see the tumor growth curve. When we have that articular injury and arthritis development, we see a significant reduction in tumor volume and increase in survival. How does this happen? Well, we first looked just at T cells. So this model is not very immunogenic. So in standard tumor model, we do not see many T cells. But when we have that articular injury, we see more T cells in that tumor environment. And these CD8 cells are tumor-specific using the B16-OVA model. When we look at vascular changes, we see some interesting differences here. So the tumor vasculature in the ACL wound, we looked at pericytes and endothelial cells, and you can see they're tightly associated with the arthritis. Without the arthritis, we have a gap and sort of a different stabilization or structure of that vessel. And we see more P-selectin in those vascular structures. So this obviously connects with that increased T cell infiltration. Now, we did look if the draining lymph node was required, and it wasn't. So then we further analyzed the serum with and without that arthritic injury and the tumor, and we saw significant increases in pathways associated with extracellular matrix organization, in particular vascular development and tissue development. When we look at single-cell analysis of these tumors, we were surprised that the P16 positive cells were not the ones that were positive for P-selectin. So this again shows the importance of communication here, where P16 positive cells were the senescent cells, but were activating a subpopulation. And this is true for both the endothelial cells and the mature pericytes. So that communication is important, and we have a number of candidates that might be responsible for this. If we look at the vascular structure more broadly, you can see the vessels are larger and more stable with that arthritic injury, and we have increased P16. So the red here, hopefully, you can see, with the pericytes and the endothelial cells, none really associated in the tumor itself, but really associated with the vasculature. And in the no surgery, just regular B16 tumor, you see very few vessels, less structure, and no senescence. And what's interesting, if we look at the perivascular cells and the endothelial cells with and without the arthritis, without arthritis, we have lots of genes associated with hypoxia and vascular development, VEGF production. So these cells are giving the signals, but the vessels just are not forming. And then in the perivascular and endothelial cells, we see the senescence upregulation.Now, looking up closely, you can see better here the pericytes and endothelial cells that are senescent. If we use the Incatak model to genetically delete senescent cells, you can see we lose the phenotype of the vessel structure, and we lose the phenotype of the T cell infiltration and P-selectin, and our tumors are no longer responding to immunotherapy. So I think I've reached my time limit, and I know virtual presentations are not quite as fun as in person. So I'd like to summarize that senescent cells can have many phenotypes, and the field and SenNet community is terming this senotypes. There'll be a review on this coming out in Nature Aging from the SenNet Consortium soon. These senotypes reside in different niches that help regulate tissue structure, and that tissue structure connects with the extracellular matrix production and things like fibrosis, but also the vascular structure and immune. And depending on the senotype, these cells can be beneficial or detrimental to many different things, wound healing and tumor biology and response. In addition to the senotype, it's also the kinetics. So, early on in a wound, for example, you do need some matrix production to rebuild the tissue, but chronic presence of those cells is associated with that fibrosis and disease. And I just wanted to mention one thing, bringing in sort of an interdisciplinary perspective. There's an interesting article that came out a number of years ago. You could see it took a little while to come out, 2021, it's a 2025 publication. And this article describes the pivot penalty in research when you change fields. And we often think about interdisciplinary work and innovation happening at the intersection of fields, and we need that to have discoveries, but there's a penalty with that, with publication citations, and grant funding, and things like that. So, let's consider as we bring in all these different perspectives. Many people to thank for the hard work on this, and collaborators, and in particular, the SenNet program, which has generated a lot of collaborations and interest in the field of senescence. So, thank you for the opportunity to present, and I hope you have a wonderful rest of the session, and feel free to reach out by email if you have any other specific questions. So, thank you very much for your attention. All right. So we have about five minutes for questions that I mentioned. Drew Pardal, Jennifer's collaborator, is willing to come up and answer some of those questions, but I think Clemens and I could also answer some senescence questions as well, if you have them. Instead of answering questions to Jennifer's talk, I would like to ask the first question- ... while you are still going to the microphones. Drew, the last point was truly interesting and a bit provocative. Can you give us more insights what it's needed on the osteoarthritis side in terms of inflammation versus senescent cells? And on the other side, seeing that there is an enrichment of neoantigen-specific T cells in the tumor now equipped with more angiogenesis, do the endothelial cells now serve as cross-presenters of that neoantigen in the tumor? So this is one of the elements that I can speak to. So this is still very early, but this is what we know and also what we don't know relative to these questions. So, what Jennifer didn't show in the interest of time was that if one uses a genetic senescence deletion model, so deleting P16-positive cells that can be induced by a drug, the anti-tumor effects of checkpoint blockade are mitigated. So, at least based on the definition, imperfect as it is, but not bad, of a senescent cell, that is certainly necessary. One also sees, of course, in this OA model and in human OA, senescent cells in the joint. And so one of the questions that's not resolved is if you eliminate senescent cells, how much of that is that you eliminate SASP components produced by the senescent cells in the joint versus components produced by the senescent pericytes in the tumor that are driving this normalization of vasculature P-select activation, P-selectin upregulation, and that translates into about a five to eightfold increase in traffic of lymphocytes out of the vasculature and into the tumor. Interestingly, the transcriptional programs of those lymphocytes in the tumor aren't that different between ACLT versus non-ACLT. So it really looks like a numbers game related to these alterations in the vasculature. But what the relative role is of SASP components versus other inflammatory componentsStill not worked out, and that's obviously a major area moving forward. Also, Jennifer didn't show an analysis of some of the Hopkins checkpoint blockade clinical databases. And certainly, if you have OA going into your anti-PD-1 therapy, actually your outcomes are dramatically better than if you don't. So this is good for somebody who has OA in their hips. This is the first positive thing associated with that. We'll take a question over here. Yeah. Thank you so much. This was a great talk. My question is regarding the integration of senotypes, as they're called, and temporality with SASP. So is it such that different senotypes have different temporal evolutions? I mean, canonically, the understanding is that in the early stages of SASP, you have a much more acute inflammatory sort of response versus this chronic inflammatory state that may be less beneficial in the context of tumor progression and response to therapy. So is there an integration between senotypes and that temporal evolution? Yeah. So this is a great question, and I think there's still relatively little that's known. So the variables are the cell that is producing the SASP, and different cell types, clearly different senotypes, call them. There are core components. So IL-6 and IL-8 are examples of-- They're fairly core to senescent cells of different types. And then there are also very distinct SASP elements that are cell type specific. So there's cell type. There is location. Some of these factors really can have systemic effects, but others of them, like IL-6 as an example, is a cytokine that acts relatively locally in that the concentration by the time you're eight to 10 cell diameters away from the cell producing it, that concentration is too low to be triggering the IL-6 receptor. And then there's time. And again, they're interrelated because one of the things that in some of these models that Jennifer sees is that the change in SASP with time is associated with ultimately loss of certain populations of senescent cells and other population, other senotypes coming up. So I think this is a hugely ripe area for the next 5 years of-- 10 years of investigation. And as Jennifer mentioned, there are actually some really impressive mouse tools, genetically modified mouse tools, that are available to be able to assess these questions. Excellent. Thank you. So we just have time for one quick question over here. I think mine is quick. I'm coming from outside this field, and what I wondered is whether the senescent signatures that are being generated give us improved markers to use in in vitro systems. So this is about can we develop better senescence markers? Yeah. And I wondered if that had emerged from the signatures. Yeah. So, absolutely. And certainly, in looking at senescent cells by single cell, really signatures composed of multiple components and this transfer learning approach of defining signatures in the mouse and looking at correlative cell types and settings in the human, I think that's going to be something that takes us a lot further than just p16, for sure. All right. We're going to move on to the next talk. Let's thank Drew for answering our questions and being a real trooper there. As you'll see, you'll have time for other questions as we go through some of these other talks. So Jennifer really gave a nice background on senotypes, introduced this concept of the senescence-associated secretory phenotype, or SASP. In my talk, I'm going to dig deeper into the SASP, particularly how it affects both cytotoxic lymphocytes as well as myeloid cells in both immune supporting and suppressive ways. And by understanding that regulation of the SASP, can we design therapies called senomorphics to remodel the SASP to get better immune responses and immunotherapy responses in cancer? These are my disclosures, though they're not relevant to this talk. So as Jennifer had alluded to, senescence goes beyond just aging, what Hayflick and Moorhead first defined it as. It's really a conserved physiological stress response to damage. There are a number of stimuli that can induce it, telomere shortening, DNA damage, reactive oxygen species, common oncogenes such as RAS, and common conventional cancer therapies like chemo and radiotherapy. The first really solid intrinsic hallmark that I think most of us are aware of is this stable cell cycle arrest that's mediated by these two tumor suppressor pathways, p53 driving p21 and p16 by inhibiting CDK4/6 kinase activity, activating Rb to both transcriptionally and epigenetically repress cell cycle genes.In addition to these cells being arrested, they're also persistent because they're resistant to cell death, such as apoptosis, by upregulating anti-apoptotic pathways such as BCL2 family members. Given these cells don't proliferate and they upregulate tumor suppressors, it was first surmised that this is a tumor-suppressive process. Indeed, there are a number of landmark studies in 2005, including from Dr. Smith's lab. They really were able to identify markers of oncogene-induced senescence in non-proliferative premalignant lesions. So what do we mean by that? So here's an example, KRAS-driven lung cancer from Manu Serrano's group. So on the right here, we have an adenoma. This is a premalignant lesion, hasn't become invasive yet. On the left here, we have an adenocarcinoma, which is considered malignant. Using this canonical marker of senescence, senescence-associated beta-gal, we can see the premalignant lesion stains beautifully for SA-beta-gal, but as the tumor progresses, it loses this senescence marker. And there's been really nice mouse models here from the Pandolfi lab in a different model here in prostate cancer losing PTEN to activate the PI3K pathway in these premalignant intraepithelial neoplasias, again, staining beautifully for beta-gal, and then having the need for additional tumor suppressor loss to bypass senescence, suggesting this might be a barrier to full malignancy. However, as we went into the 2010s and beyond, the development of a number of mouse models using two of our most canonical markers of senescence, p16 and p21, to track these to temporarily and spatially deplete them, and development of drugs called senolytic drugs to target these anti-apoptotic pathways or P53 pathways that allow senescent cells to persist, showed the exact opposite, that actually senolysis or removal of senescence can block tumor progression and relapse after therapy. So how can senescence have this paradoxical role? So there could be a number of reasons. As Jennifer alluded to, perhaps cells other than tumor cells are undergoing senescence and they're causing tumor promotion, and I'll get to that towards the end of my talk. You'll hear from Dr. Smith perhaps senescent cells aren't totally irreversibly arrested and could eventually proliferate again. But what my talk will focus on is the SAS, the ability of these damaged cells to convey their damage signals to the microenvironment. And the reason why the SAS might induce this paradoxical role of senescence, it has a lot of pleiotropy. There are a lot of different types of factors included in this secretome. These include proinflammatory cytokines, IL-1 alpha-beta, IL-6, chemokines like CCL2, but also growth factors and matrix metalloproteases that degrade the extracellular matrix, angiogenic factors such as VEGF, PDGFs, and also prostaglandins and other lipid species. And it's important to note that in addition to these secreted factors, senescent cells upregulate immunomodulatory receptors and ligands on their surface that can have immune consequences. For instance, activating NK cell ligands, activate NK cell surveillance, MHC class II expression of antigens that activate CD8 T cells, but also immunoevasive molecules such as PD-L1. And the SAS is regulated dynamically by the number of different transcription factors listed here. My talk, I'll focus on two of the more inflammatory ones, NF-kappa B and c-Gas, seeing the drive downstream IRFs and type I interferon signaling. We also know that there's also epigenetic regulation and enhancer remodeling to open up these loci as well as increase protein translation to drive the SAS. And so really, Judy Campisi, I believe, is the one who coined SAS as this double-edged sword in cancer. On the one hand, as Jennifer alluded to, the SAS can lead to fibrosis resolution, it can lead to wound closure, and it can also lead to attracting immune cells that come in and clear out that damage. But as the SAS lingers, for instance, during aging or other chronic diseases, that inflammation can be inherently immunosuppressive. It can also drive angiogenesis as well as invasion and proliferation of adjacent tumor cells. And so we really think this gets into this idea of acute versus chronic SAS. If you can acutely activate an immune response with a SAS that come in and clear out that damage, this can be tumor suppressive. But if this lingers, eventually the SAS can give rise to progression. So what I'll give you now is a few vignettes from my work and other works alluding to these tumor-suppressive and promoting aspects of the SAS, and then dig into how my lab has sort of looked into how the SAS is regulated and how we can push this response more to being tumor suppressive. So a number of works, particularly in hepatocellular carcinoma, have really shown us that senescent cells can be immunogenic, capable of both innate and adaptive immune surveillance. This is a work from Wen Xu, Lars Zeuner when they were in Scott Lowe's group. This is a model of HCC where P53 can be turned on or off with an shRNA. Reactivating P53 in the context of RAS mutations can induce senescence, as shown here by beta-gal. This can regress the tumors. And they showed that this was due to, at least here, innate immune activation, neutrophils, macrophages, and NK cells, induced in part by these SAS molecules shown here. So this SAS can induce innate immune surveillance to regress established tumors. More recently, Scott Lowe's group has looked actually not at the secretome, but at the surfaceome of senescent cells, what is on their surface that's upregulated. They found the interferon gamma receptor is highly upregulated on senescent cells. Why that's of consequence? Because interferon gamma in the microenvironment can bind these receptors and activate increased MHC-1, which can lead to better antigen presentation in the tumor, reprogram macrophages and also dendritic cells to be more antigen-presenting, and thereby also activating adaptive CD8 T cell surveillance. And a back-to-back paper that came out from Manu Serrano's group even showed that you could inject these senescent cells as a vaccine into mice as a prophylactic. So if you injected these cells, then transplanted a tumor, the tumor never formed. Again, showing that senescent cells can be inherently immunogenic.And there's also the context of acute therapy-induced senescence from many different types of therapies. So this is really nice work from Jim Jackson's group looking at chemotherapy in breast cancer models. They showed that downstream of NF-kappa B and IRFs, a number of immunomodulatory receptors upregulating, including a number of different immune checkpoints. And they found by combining chemotherapy with anti PD-1 plus anti-CTLA4, which are immune checkpoints, they can get better tumor responses in this breast cancer model. I also showed similar things when I was a postdoc in Scott Lowe's group in the setting of RAS-driven pancreas cancer. We found that a combination of palbociclib, a CDK4/6 inhibitor, and trametinib, a MEK inhibitor, could potently inhibit downstream RAS signaling to induce an Rb-mediated version of senescence. Interesting and similar to Jennifer's talk, we actually found the SAS in pancreas cancer was very angiogenic. There are lots of VEGFs, PDGFs secreted. This lead not only to increased vasculature, but better endothelial activation marked by P-selectin, VCAM, increased MHC II, that led to the increased infiltration of CD8 T cells. However, those T cells weren't really active. They upregulated PD-1 and the senescent cells themselves, though upregulating MHC I also simultaneously upregulated PD-L1 checkpoint. And we found that combining TP to induce senescence with anti PD-1 could at least lead to partial regressions in this model. So that's with acute senescence. But if we go back to this acute versus chronic scenario, I think it's a good example to look in the setting of oncogene-induced senescence. So this has really been work by Lars Zender's group here. They overexpress NRAS in hepatocytes to induce senescence. You can see these are senescent at day six. By day 16, they're gone. And they found that a combination of macrophages and CD4 T cells could eradicate these hepatocytes that are senescent before they can go on to form tumors. And from SAS profiling, they found that CCL2 seemed to be a main driver, particularly of the macrophage phenotype. But in later work from the Zender group, they also looked in more established HCC. And this is just a schematic describing that actually that chronic CCL2 that initially led to clearance early on, actually promoted the infiltration of immature myeloid cells, what some may call MDSCs, that actually inhibit macrophage differentiation and inhibited NK cells. So I think this is a good example of how chronic SAS can eventually be tumor promoting. And others have shown this. Really nice work from Sheila Stewart and others have shown particularly fibroblasts can secrete what I would call more of this pro-myeloid SAS, CCL2, IL-6, CXCL1 and 2, that promote these suppressive myeloid cells that can be tumor promoting. A nice example is here in the setting of prostate cancer from Andre Almonte's group. In this model, either of oncogene-induced senescence or chemo-induced senescence, the SAS heavily upregulated CXCL2, which drove in the infiltration of TAMS and also MDSCs expressing its receptor CXCR2. And Johan de Bono in the UK turned this into a clinical trial. So he looked at this high neutrophil to lymphocyte ratio, which is usually a marker of poor treatment outcomes and immune suppression. He found that high neutrophil to lymphocyte ratios correlated with different SAS markers, NF-kappa B, CXCL2, amongst others, and even enrolled patients with castration-resistant prostate cancer, which is the most aggressive form in hormone refractory portion of the disease with CXCR2 inhibitors and can get responses. So this is just SAS chronically can exist even beyond oncogene-induced senescence and could be targeted as a xenomorphic approach. And so when I started my lab in 2020, we were interested in really understanding this dichotomy of SAS, how it's regulated, how can we harness it for cancer immunotherapy. So some questions we were after that I'll get to in my talk. One, which SAS factors are actually important for antitumor immune responses? Here are just some of them. Some of them I mentioned implicated in activating NK cells and T cells, other myeloid cells. Ones in gray are very pleiotropic, they can do multiple things depending on the context. And can we harness just the SAS to activate, in particular, these cytotoxic lymphocytes, NK cells, and CD8 T cells that we know are important for antitumor immunity? Second, how is the SAS transcriptionally regulated? I mentioned NF-kappa B and STING are two of these more pro-inflammatory transcriptional regulators, but there have been many implicated. For instance, P38 MAPKs known to inhibit interferon signaling. Notches are known to drive a more TGF-beta-rich ECM remodeling SAS. In addition to the transcriptional regulation, we know that there's also dynamic epigenetic regulation as well that's worth exploring. And finally, how does the senescence inducer, meaning, for instance, the type of therapy or oncogene-induced senescence, the cell type, whether it's a tumor or non-tumor cell, and the resident microenvironment impact the SAS, and from that, its impact on immunity? So to get at that latter question, we did this interesting experiment where we treated mice with the same drug, but in different cancer contexts. So we went back to this trametinib, which is a MEK inhibitor palbociclib CDK4/6 inhibitor-induced senescence. We did this in KRAS-driven lung cancer models. Saw this nice NK cell response mediated by SAS factors like CCL2 and IL-15 that could attract NK cells, that could bind to the cell adhesion molecule ICAM-1 upregulated on senescent cells, leading to regressions that could be reversed by giving an NK1.1 blocking antibody. We also tested in a very similar genetic model, still KRAS, mutant and PD3 deficient, but now pancreas cancer. But we did not see the same inflammatory SAS induced. We did not see the same NK cell responses, and we did not see tumor regressionSo we're curious about this phenotype and digging deeper into that. First postdoc in my lab, Lo, really did this really cool experiment to try to address this question. She took mouse tumor cells that were from pancreas cancer, lung cancer, and she injected them into different organ sites. So she either injected them into the pancreas or tail vein injected these to form tumors in the lung, sorted out just the GFP tumor cells, looked at the SAS profile and found in this setting where pancreas tumors are grown in the pancreas microenvironment, they did not have the same SAS as seen in other organs. IL-15 and 18 that are important for NK and T cell activity, and CXCL10 and 11 that are important for their chemotaxis. Digging deeper in doing transcription factor motif analysis, she found of the differentially expressed genes in the pancreas, EZH2 was a top regulator. And EZH2 forms this PRC2 complex that leads to H3K27 trimethylation, which is a repressive chromatin mark. So could EZH2 upregulation of the pancreas lead to SAS suppression and hence NK cell inhibition? So she used shRNAs to knock down EZH2 in the tumor cells. We injected them, treated with these drugs, induced senescence. Now massive upregulation of the SAS, including many of these chemokines here that I mentioned are usually not induced in the pancreas microenvironment. With that, better NK cell infiltration, early activation markers. Here, six out of seven mice had complete responses now with knocking down EZH2 that could be completely reversed by depleting NK cells. So this we found was a way by taking off this epigenetic break to activate this inflammatory SAS, particularly chemokines, to activate NK cell immunity. But we're interested in the other side of the coin. What's so different about lung cancer compared to pancreas? Not surprisingly, some of the top transcription factor bindings were from RelA, which is the p65 subunit of NF-kappa B, and a number of IRFs that can drive downstream interferon signaling. So we thought, could we find a way to activate interferon signaling that's induced in the lung but not in the pancreas? To do that, we collaborated with Pravana Teegala Raleigh's group at UMass Amherst. They had made a lipid nanoparticle containing CDGMP, a STING agonist, and MPLA, a TLR4 agonist. She had previously shown the combination of these two pathways could synergistically activate IRF3 downstream to drive this type I interferon response. This is just showing that we can actually deliver these nanoparticles by tail vein injection and deliver them into the pancreas microenvironment. And with that, these orange bars indicate the combination of inducing senescence and activating IRFs could induce more interferon beta, IL-15, and other chemokines usually repressed. We also did long-term treatment responses in genetically engineered mouse models of pancreas cancer that form spontaneous tumors. While we did see some NK cell activity, we saw more of actually a CD8 T cell-driven response, as shown here. And while each modality alone could increase survival, we saw massive survival increases combining TP to induce senescence with these STING TLR4 agonist nanoparticles, including two complete responses that unfortunately relapsed off of drug, but still driving the proof of concept that activating the SAS, particularly the type I interferon arm, could be a way to lead to durable immune responses in pancreas cancer. And indeed, using an IFNAR antibody, which blocks a type I interferon receptor, we could reverse the T cell activity. So to summarize this portion, we know many different drugs can induce senescence, but they don't all induce the same SAS. We found that targeting EZH2 to take off the epigenetic brakes or targeting STING TLR4 agonists to activate the type I interferon arm of the SAS could activate both NK cells and CD8 T cells in pancreas cancer, leading to tumor regressions in this very immunologically excluded tumor type. But we're also interested in how different classes of therapies could modulate the SAS. So Lynn Joe, a former postdoc in my lab, did this in the setting of prostate cancer. This is just showing the different types of inhibitors she used, not only CDK and MEK inhibitors, but also a number of DNA-damaging agents, PARP inhibitor, aurora kinase, and conventional chemotherapies. What I hope you appreciate here is TP can induce some senescence, but these DNA-damaging agents seem to induce a much more robust senescence phenotype as shown by assay beta-gal, and also a much more robust SAS, both quantitatively and also qualitatively different, driving more of what I would call this pro-myeloid SAS of CSF, CXCL1, and CXCL2, particularly with this chemotherapy known as docetaxel. Doing RNA-seq, she dug deeper into this and actually found an addition to inflammatory signaling pathways being upregulated with DNA-damaging agents versus CDK inhibitors. Also ECM remodeling SAS, including MMPs, TGF-beta, that are known to remodel the ECM, but also drive what's known as EMT, which is a driver of invasion and metastasis. And indeed, in mouse prostate cancer models, docetaxel, which can induce senescence, can decrease primary tumor growth, but it has no effect on metastasis. And so she looked at what was upregulated with docetaxel. Not surprisingly, as I mentioned, BCL2 is this anti-apoptotic protein upregulated in senescence, but also IAP, inhibitor of apoptosis. And so she used these senolytic approaches to target these pro-survival pathways. This is showing we can induce cell death and decrease the p16 burden with these approaches. But interestingly, she found the mechanism of action was a little bit different between these senolytic approaches, with IAP antagonists not only inducing apoptosis, but an inflammatory version of cell death known as necroptosis. And with that in particular, she can not only reduce this myeloid chemokines, but actually also induce some of these pro-lymphocyte chemokines with that immunogenic cell death. And with that, with these senolytic approaches, activate more T cells, reduce immature myeloid cells, enhance antigen presentation by DCs, and now reduce metastasis. So this shows us that senolytic approaches could takeCare of this sort of ECM remodeling SASP by removing those senescent cells, and also suggests that the type of cell death induced by senolytic might induce different immune responses downstream of the SASP. And finally, we wanted to also get at how the cell type undergoing senescence also impacts the SASP. So Lynn, in a work published earlier this year, in her early model of prostate cancer, this oncogene-induced senescence, she found that p16 and p21 that we typically use interchangeably as senescence markers actually marked different senescence populations. p16 really marking epithelial tumor cells in the glandular structure, but p21 marking more stromal cells. Collaborating with my former PhD mentor, Hong Wu, we did single-cell analysis and found that CDKN2A, which encodes p16, was really in the epithelium, but p21 was expressed more broadly, including in myeloid cells and fibroblasts. And just to break down this busy figure, what we found by looking at the SASP was the epithelial compartment really wasn't expressing much. Fibroblasts, a little bit of TGF-beta and VEGF, but really it was the myeloid cells secreting predominantly this SASP. So senescing cells in the microenvironment might induce a different SASP, it might be tumor-promoting. To prove that functionally, we did this experiment where we transplanted a syngeneic prostate cancer cell line into this p21-RTTA model developed by Sundeep Khosla's lab here. p21 drives an inducible caspase VIII that allows you to give this pro-drug to remove any p21 cell. But because we're injecting a tumor cell line that doesn't have that allele, we'll only get rid of p21 in the microenvironment, not on the tumor per se. And with that, we can activate T cells, reduce immature myeloid cells, leads to tumor growth suppression, and importantly, block the SASP. We think the microenvironment SASP, perhaps because it's chronic, perhaps because it's more pro-myeloid, seems to be tumor promoting. And I'll just mention, we're not the only ones who have shown this. There are many in the field to look at this sort of non-tumor senescence. Darren Baker's group has shown this nicely in macrophages in lung cancer. Sheila Stewart's group has shown this in fibroblasts in breast cancer. And Andrea Alanti has shown this in neutrophils in prostate cancer. So there are many different cell types that can undergo senescence. This has really changed our thinking of senescence beyond the tumor and thinking about, could this stromal senescence be a link between stress, chronic inflammation, and cancer? Especially if you think of an aging individual accumulating senescent cell, could the senescent cell accumulation in benign tissues during aging create a soil for tumor propagation? So that's a nice thought, but we need a number of next steps in the field to really get to this point. We need better biomarkers to distinguish these senotypes, but also these different flavors of SASP. And from that, we'll need to design more specific senolytic and also senomorphic targets to target this pathological senescence. And we'll also acquire deeper characterization of how senescence and immune cells interact, particularly in niches, in particular at a spatial level. So to summarize the key points, what we found is that innate adaptive immune system can mediate potent clearance of senescent premalignant as well as established tumors through the SASP, through things like chemokine CXCL9 and 10, interleukins, interferons, but also have regulation of ligands that can activate NK cells and MHC class I molecules. We also know a chronic SASP can lead to myeloid-mediated immune suppression through factors such as IL-6 and CXCL1 and 2. And that SASP-mediated immune responses in cancer are dependent on a number of things. The resident makeup of the tumor microenvironment, the chromatin landscape of the tumor cells, particularly these marks that regulate enhancer remodeling, and the SASP transcriptional programs. I mentioned a few of these. We looked at pro-angiogenic that induce blood vessel formation, pro-lymphocyte that activate NK and T cells, pro-myeloid, and also ECM remodeling that could potentially drive metastasis. And it's also dependent on the therapy, whether we're talking about a cytostatic CDK inhibitor or cytotoxic chemotherapy can induce different SASP, and whether this is in a tumor cell or whether this is a stromal or immune cell undergoing senescence. And I'll leave you with this. I think senomorphic approaches to enhance type I interferon arm of SASP while suppressing the pro-myeloid SASP will maximize immunogenicity of senescence for cancer immunotherapy. So I'll just stop there and really acknowledge our lab, our great collaborators, mentors, and we're hiring at all levels if anyone's interested in joining our group, and I'm happy to take questions. Thanks. Thanks. Thanks, Marcus. That's not on? I don't know. Okay. Thanks, Marcus, for a great presentation. We have time for one or two quick questions. Hi, good morning, Marcus. My name is Jelani Zarif from Johns Hopkins. Great talk. I was wondering about your experiments using the MycAP models. Yeah. Were those animals-- And then you used the MP10. Were they castrated after tumor implantation? Those are castration sensitive. That's a great question. P10 we did, and that was a reviewer question. It might've been you. But we did it. No. It wasn't me. It's in the paper. So we actually had a better phenotype when we castrated the P10 model. But they're de novo resistant. They'll start growing immediately. So there's not a dip in a growth. They just keep growing. But they do even better. MycAP we didn't because we're a little bit afraid because if you remove androgen, you're going to also inhibit Myc, right? So I try to stay away from castrating the MycAP model. And then on top of that, you used the ABT inhibitor, I think in the MycAP model. Mm-hmm. Have you all combined that with the anti-PD-1 therapy, given that you saw a large influx of cytotoxic T cells based on your staining? Yeah, we haven't done that yet. That would be a really cool thing to do, but we haven't. Yeah. Great. Thank you. Yeah. Sheila Hey, Marcus. A great talk. So I understand this is a loaded question before I ask it, but when we think about P21 versus P16 senescence, admitting there's probably many more, do you think that it's more about cell specificity and/or temporally? Like one comes before the other- Ooh. ... like in a GEM model. My guess is temporal. My guess is that either tumor-secreted factors or SAS, I think P21 is a secondary senescent. I think it's secondary induced by acute stress, and they become hyper-inflammatory. It could also be induced by the microenvironment. They're getting into a hypoxic microenvironment. So my hypothesis would be that it's temporal, but we haven't proven that yet. Very last question, Barry. Yeah. Quick, please. Beautiful talk. I'm an immunologist, and this is new to me, so my brain is just- Mm-hmm ... going crazy now. And I was just wondering, beautiful mouse models, but have you kind of put the pieces together to look in the cancer atlas in patients to kind of understand where the jump is in terms of how much we could harness in patients here? I see. So translating what we're doing in the mouse model to the patients. Yeah. Yeah. We're wanting to do that. It's a little bit tricky. Some of the therapies like chemo, we can do. I think Clemens has really been the pioneer in the field of translating this into patients, so I think he can speak more to that. Maybe he will during his talk. But we haven't been able to yet. It's really hard to, I think-- Some of these markers are difficult to stain for. It's also hard to know fixed sections, where the signals are coming from, and how they change over time. So we would love to get there and love any suggestions from anyone in the audience, but I think Clemens is probably the best person to answer that. Maybe he will during his talk. Thank you. Yep. All right. So we'll move on to- Our third and final talk from Dr. Clemens Schmidt, who's professor at Charité University of Berlin, Max Delbrück Center of Molecular Medicine in Berlin, and Johann Kepler University in Linz. And we're looking forward to your talk. Thanks. All right. Thanks very much. Let me start off by thanking the ACR, the organizers, for selecting a presentation from my lab for this highly informative educational session on cellular senescence. So here are my disclosures. And I can only thank the previous speakers for so well introducing the topic that there's not much to add, but a little bit. In the clinic, we often don't know whether we look at a wound or cancer. So as obviously as we learned from the previous presentations, an intimate relationship between derailed wound healing and cancer. So senescent cells, as we all know, are an intrinsic component of the multicellularity of cancer as an organ. But I like to make the point, although it was pointed out by my previous colleagues, that in an acute fashion, senescent cells in wound healing are an essential component for proper wound healing. So the components listed here following the tissue injury, primary senescence, then SAS-mediated secondary senescence to take care of all damaged cells, not to propagate any more and to contribute to the parenchymal cellularity, is followed by SASP-called immune cell invasion to clear up the injured site. But there's also translational reprogramming into senescence-associated stemness, which sounds paradoxical, but in proper wound healing, it's the senescent cells which therefore may contribute to replenish cellularity once stresses have been resolved. And with this in mind, we may now look at senescence-promoted cancer features with some different perspective. In particular, looking at senescent cells not only as the source of SAS, but also being reprogrammed latent stem cells. And if this happens hijacked by a tumor, they may become de novo cancer stem cells. So with this, that's already obvious, I think, to this audience, there is, in an acute fashion, in response to activated oncogenes or in response to antineoplastic therapy, a beneficial tumor-suppressive form of senescence. But if those therapy-exposed senescent cells or oncogene-activated cells persist, then they may drive derailed wound healing inflammation. They are reprogrammed as stem cells. And if they get permission to occasionally re-enter the cell cycle, they may exert their stem-like function. There's immunogenic plasticity, immune evasion. We heard about this, and therefore, there are possible therapeutic interventions. We heard about senomorphics to limit, in particular, inflammation, and senolytics to actually catch the senescent cells at their root and eliminate them. I'd like to make a point which, in my view, is obvious but is not really deeply discussed. Senescence is not a binary state. It's not in or out of senescence. There is a pro-senescent trigger driving a cell into senescence. And actually, it may matter over time how long, how strong, and in what kind of trigger this trigger is applying to cause maybe initial senescence, light senescence, actual senescence, and perhaps something like super senescence. I'm not showing data on this, but I like to make this point that the depth of senescence, particularly at the level of epigenomic reprogramming, chromatin remodeling, in particular, the senescence-associated heterochromatin foci may decide about the quality of senescence and its depth, and in other words, what was already quoted before, reversibility of senescence. I would not claim senescence is reversible. I believe it's neither reversible nor irreversible, but there is a state beyond senescence. I like to share that with you. We have data why we believe this is not philosophical, so that if senescence is reversible, it was not senescence. If you have a cell deeply in senescence, and here's the machinery inaugurated by Monsieur Shin Narita in Scott Lowe, the SAFmachinery. E2F target genes are deeply silenced by an histone H3 lysine 9 trimethylated mechanism. We concluded if we would provide in senescence a demethylase activity, we may be able to drive cells out of senescence. That's what you see on the right side. If we induce Jumonji D2C, these firmly RAS-induced senescent cells can level off, become pS2 beta negative, and can reproliferate. And the mechanism behind is turnover of nucleosomes. This has to be permanently reestablished. So we allowed an H3 arginine 9 mutant to compete with H3 lysine 9, and you see it takes a while, but after a while there is an overtake of the R9 mutant, which can obviously not lysine 9 trimethylate, and that's sufficient to allow, again, senescent cells to get out of senescence. And you see these senescent cells have now a reduced load of H3 lysine 9 tri-methylation. We heard a lot about heterogeneity of senescence, and I think that's super important because the field is moving into single-cell omics technologies and tries to describe the various facets and types of senescence. I like to discuss with you whether there might be rather one full-featured uniform program that provides certain aspects in different contexts. So it's more or less the same senescence program, which is associated with key functionalities in given cellular context. Other features are present but less relevant. Or whether there are indeed multiple or many distinct types of senescence characterized by different lead functions where other features are less present. I'm not concluding either of the two is right. I guess it's a spectrum in between, but it's important for therapeutic interventions. So there is a core senescence machinery and a specialized machinery, which would mean if you look at a rather uniform senescence program, it's driven by the core machinery. If it's more a specialized program, it's driven by these additional lead functions. Core senescence markers we know, it's complicated enough to detect them or to determine whether a cell fulfills a sufficient amount of these markers or is positive for certain signatures, as we heard today. So this is complicated. But if we look at the more specialized senescence markers, we would have type-specific senescence vulnerabilities as compared to targeting the core machinery in terms of a shared senescence vulnerability. So the more uniform program is good and bad senescence throughout the body, while the more diverse or multiple types of senescence may give you specific senescence functions and you may argue, I understand that from a clinical perspective, you want to target the bad ones and want to leave the good ones intact. Whether that's achievable, that's a different question. I'm not making an argument for either point, but I like to go for the second half of the talk through an intense analytic approach where I share with you the addressing of a rather uniform aspect of cellular senescence and what the consequences of this might be. Before we get to this, looking at senolytics, what will happen? In acute senescence, or senescence in an acute fashion, I think we agree on this, is rather a beneficial program, and there is stemness associated with senescence in wound healing. There is immunosurveillance of senescent cells. But if senescence gets chronic, it's typically detrimental. I already referred to cancer cell-associated stemness, loss of functional parenchyma, and others before have talked about organ fibrosis. So if we eliminate the beneficial acute senescence, we may have problems. We may impair stem cell compartments. We compromise regeneration or may attenuate immunity. That needs to be seen. In the chronic detrimental senescence, that's obviously the space in which we like to eliminate senescent cells in order to see anti-aging effects, fibrolytic effects, and anti-cancer effects. So if I could choose, I would try to optimize dose scheduling in order to leave senescence beneficial in the first place and apply senolysis in a scenario where I can eliminate those which dictate unwanted disease pathologies. So that's what we aim for, to control cancer, actually even prevent cancer, and attenuate aging by pan-senolytic approach. And I'd like to share with you how we harness the redox imbalance in senescent cells for senolysis and what the clinical effects in preclinical models might be. For this, we teamed up with a medicinal chemist from the Free University in Berlin, Jörg Rademann, and colleagues, and set up a redox-based drug screen because we came across metabolic imbalances in senescent cells driven by NADH as a background, which could allow us to screen for an oxidative activation of candidate drugs in order to produce further oxidative stress to selectively eliminate senescent cells. So get a dual selectivity of an agent, which is now called Senlytic, for the elimination of senescent cells. So this agent we retrieved from the drug screen, Senlytic, is indeed activated by NADH. You see the higher the NADH we provide, the higher thePeroxide H2O2 output is, and this is countered by GSH. So in conclusion, high NADH, low GSH levels would activate this agent. And since it's producing peroxide, if we overexpress catalase, we reduce peroxide production, and we limit the cytotoxic efficacy of this agent. We made a derivative which is no longer activated by NADH. It's the de-Senlyt, the blue line, and you see this has no senolytic potential as compared to the original compound, the red line. Here you look at a number of senescent scenarios, therapy-induced, oncogene-induced, oncogene inhibition-induced senescence. And the common feature of all these senescent types is this altered NADH/GSH ratio across the cell types. Notice that senescent cells have an elevated pool of labile iron, and this LIP can be measured by Ferro-Orange. And you see that across different senescence inductions, we see this elevated pool. But senescent cells are protected from ferroptosis. We are coming to this in a second. If we provide a chelator here, Desferrioxamine, we can reduce these lipid iron pools. So if we use Senlyt, we can, in a dose-dependent manner, increase reactive oxygen species. You see in the upper left the lysosomal ROS or right, the mitochondrial ROS. It's the red lines to focus on for now. And lipid peroxidation increases or is increased, as I said, in senescent cells, and that can be utilized on the lower right by Senlyt to kill these cells out of this latent ferroptotic state. If we use another iron chelator here, B-pyridone, BIPI, the blue lines you see we reduce ROS production in response to Senlyt. We get less lipid peroxidation, and we don't see a senolytic effect. So this concludes that Senlyt helps to utilize iron in the so-called Fenton reaction to produce OH radicals and to kill. But I like to mention that another arm of killing by Senlyt is via classic apoptosis, as blockable here by cVAD-fmk. But in conclusion, we use two modes of cell death, ferroptosis and apoptosis, and even strongly apoptotically compromised tumor cells can still be killed via the ferroptotic pathway. Senlyt has high senescent selectivity. You look at four scenarios, non-malignant settings or tumor settings in the mouse or in the human species. And you can see from all these graphs that there is a huge senolytic index and great advantage to selectively kill senescent cells as compared to proliferating cells. That's reiterated here. It's difficult to read curve, sorry about this, but what it tells us is that Senlyt is at least as good or better, obviously, than other established senolytics in killing at low concentration senescent cells, but with limited toxicity to their proliferating counterparts. And there's a genetic senescence dependency if we test p53-proficient versus p53-deficient cells. It's the proficient and senescence-capable, but not the senescence-incapable cells we can kill by this agent. Just to go through a few more vignettes, this is replicative senescence in a setting where we have cells arresting after a certain number of divisions as compared to HTERT positive cells. It's only the RS cells which can be killed. We look at virus-induced senescence. Some cells can undergo virus-induced senescence. You see that on the left. Others don't, and only those which undergo this are susceptible to Senlyt. That might be the most important slide of the presentation for Senlyt, because you look at virtually all kinds of cellular senescence we tested, and whenever we achieve to drive cells into senescence, you see a blue square to depict senescence. And it's only in those settings where dose-dependent application of Senlyt will result in profound cell death. We can eliminate aged cells. Here, this is an example of lung fibrosis in old mice. Again, to remind you, there is equally an altered GSH/NADH ratio in these aged lungs. The aged lungs are characterized by high-level p16-positive cells and H3K9 trimethylation-positive cells, and these cells account for collagen cuffs around the bronchioli. And if we apply Senlyt to those aged animals, we not only remove the senescent cells, which is nice to see inside you, we also reduce the collagen cuffs, which is interesting because apparently you can eliminate senescent cells, and by the endogenous fibrolytic activity of the body, you even reduce collagen cuffs. So that's a demonstration of in vivo clearance of senescent cells. Senlyt is of very limited toxicity, to say the least. We don't see differences in body weight. We don't see differences across major cardinal lab or cell blood cell parameters. We can use Senlyt as an augmentative or consolidative agent upon chemotherapy. You look here at chemotherapy or TIS in a lymphoma model where only p53-proficient lymphoma cells with an apoptotic block underwent cellular senescence.And if we take those cells, apply senlytic, we kill the senescent, but not the equally treated p53-deficient cells. And we can test that in vivo. We administer lymphoma cells, lymphomas grow in recipient fully immune competent mice, and then we first apply chemotherapy and second senlytic. And it's senlytic which in senescent lymphomas, extends survival of the mice in virtually all settings. But if we use p53 null lymphomas, the addition of a second drug here, senlytic, is ineffective. We can look at oncogene-induced senescence, first in vitro. We take MEFs from LSL, MEFs and activate their BRAF. Then they turn senescent, and we can eliminate them by senlytic. We can do this in vivo in these very same mice, in which we activate BRAF by provision of hydroxy tamoxifen, CRE is activated, BRAF is on, we get nevus cell nevi pigmented skin lesions. You see that in the middle. And if we provide senlytic, we can reduce these p16-positive lesions. That is particularly nice to look at the ears of these mice, you see these dark spots in the ear, and if you provide senlytic, they disappear again. So another example for in vivo clearance of senescent cells. We can drive this a step further, and that's a pretty interesting model. Based on tamoxifen, we can turn on BRAF. But we can use tamoxifen in addition in this compound mouse containing two p53 ER alleles to restore functional wild-type similar p53 levels. In this setting, we get the same result as you just saw, but we have the opportunity to remove tamoxifen at some point in time, which means that there are BRAF-positive lesions now no longer with active p53, therefore on their way to form melanomas. And that's indeed what's happening in this model. You see the lesion at the back of the mouse. You see the skin tumor, and you can even see metastasis to inguinal lymph nodes in the lung. But if we now in this window of opportunity where we have only OIS nevus lesions provide senlytic, the result is the following. We see delayed and reduced melanoma genesis in this model due to senlytic. We can also target RAS senescent cells. This is simply retroviral provision of oncogenic RAS. And you see on the left, we can provide senlytic and eliminate those cells. If we do RAS activation in p53-deficient cells, there is no effect of senlytic. The interesting point is now using an old-fashioned chemical cancerogene, NMU, N-methyl-N-nitrosourea, because that causes RAS mutations from endogenous alleles. The result is same. We see senescence in response to NMU, and this senescence can be eliminated by senlytic only in p53 proficient cells. Now we took this to the next level and did this in vivo. So we treated mice, normal C57 black 6 mice with NMU. We get, among other things, RAS activating mutations. And all these mice will succumb to T-cell lymphomas in a limited period of time, as you can see. If you use a background here, Suva minus mice, which can no longer establish this H3K9 trimethylation mark. So they get instantly with no tumor blocking senescence, these T-cell lymphomas. If you do this in wild-type mice, it's much less impressive. About half of the mice will get over time T-cell lymphomas because there's a senescence barrier. But if you now provide senlytic after NMU provision to these mice, you virtually prevent tumor formation from incipient RAS lesions in this animal. So with this, let me conclude that we look at senlytic as a novel compound class that is activated by and kills via senescence-associated oxidative stress. It has senolytic activity across virtually all types of senescence with no obvious systemic toxicity. It extends lifespan, I didn't show that, and attenuates age-related pathologies. Prolonged survival after debulking chemotherapy, limits the progression of BRAF positive nevi to full-blown melanoma, eliminates OIS restored melanoma. I didn't show that, but you see the bullet point here. And it prevents RAS-driven tumor formation by eliminating incipient OIS lesions. So in summary, what does it mean for senolysis and pre-cancer? We can target an universal core feature of senescent cells here, and metabolic imbalance by an pan-senolytic that's feasible and effective. And using a short pulse dose schedule achieves clearance of senescent cells in vivo, whose proof is in clinical routine, certainly not trivial. Pan-senolysis covers multiple detrimental features of senescence, and I'm not seeing an advantage of a more selective senolytic approach, and that's of course open to a controversial discussion. Cancer preventive and aging attenuation treatments may require repetitive applications, and I can't tell you today, but I expect that in the range of every 6 to 12 months to be needed. No overt toxicity was observed, but long-term effects need to be monitored, especially upon repeated applications. With this, I think I have mentioned in particular the colleagues who contribute to this work, and like to finish my presentation to this double-edged session that it's maybe not so double-edged if we just eliminate in tight windows senescent cells once in a while. Thanks for your attention. Thanks, Clemens, for the very thought-provoking talk. The senlyt is super exciting. I guess I'll ask the first question. You've proven that it can eradicate senescent cells in so many scenarios. If you took it into the clinic, what indication would you first try, and why? So we are at the level where we try to push this agent through pharmacokinetics and different species, and are close to consider early testing in cancer settings. I'm grateful for the question since we are at the AACR, and heard about osteoarthritis. There are many diseases, age-related pathologies, where senolysis or SAS modulation may be very important. But we also talk about to bring those candidate agents actually at the bedside, bring it to clinical registration. And I believe as an oncologist hematologist, that probably the cancer space is the fastest way to demonstrate efficacy and bring these agents to the clinic, and then potentially into non-cancer settings. I will take a question here. Can I? Hi, wonderful talk. Thank you. Thank you. My question is maybe a naive question, but what about in cancer heterogeneity, the probability to select the most proliferative cells by using a senolytic drug? This possibility is given since we will be targeting obviously the senescent cells. But if I may say so from changing the perspective, one of the most critical problems in cancer is therapy surviving persister cells, and these persisters are typically low to non-proliferating cells, dormant cancer cells with overlapping features to cellular senescence. If we would get a handle on those cells by a senolytic approach, we would probably make a major achievement. And I'm certainly not arguing that this is all what it takes to eliminate potentially relapse-promoting cells. You might be very right, but I think the therapy surviving slow to no proliferating persisters are a central problem to target. Thank you.