Scientist Stories: Uri Alon, Simplifying Inflammation and Fibrosis

Record. Okay, good. So recording just started, so we can start. This is a wonderful honor and pleasure to welcome Uri Allen for this Philosophy in Biology and Medicine seminar. As you know, one of the main goals of the Philosophy in Biology and Medicine network is to promote conceptual and theoretical approaches to biological phenomena and medical phenomena as well. And I think Ori Allen is a fantastic illustration of just that, as he's one of the world's leading scientists having developed and promoted systems biology approaches, which, as you know, have led to fascinating results both at the theoretical level and the experimental level. Ori is the author of a remarkable textbook, the subtitles of which is at least as important as the title itself, I would say. It's called An Introduction to Systems Biology, and the subtitle is Design Principles of Biological Circuits. And the second edition of that textbook was published in 2020, and I highly recommend that textbook. And the idea of identifying underlying principles that are common to various biological phenomena is one of the most central ideas of Ori's work. As also illustrated by many of his papers, including the paper Network Motifs: Simple Building Blocks of Complex Networks that was published in Science in 2002 and I think has been cited more than 8,000 times. Uri has a PhD in physics. Then he did a postdoc with Stanislas Leibler at the Department of Molecular Biology at Princeton University in the US. Uri is now a full professor at the Department of Molecular Cell Biology at the Weizmann Institute of Science in Israel. And the last thing I wanted to say is that we also like in this seminar series to have sort of regular motifs, so to speak. And you certainly remember that last month, our speaker was Stephen Frank of UC Irvine, who spoke about robustness. And there might be some echoes of robustness and complexity question in the talk today. You may remember that in September 2020, we had as a speaker in this very seminar series, Ruslan Medzhitov from Yale, who spoke about homeostasis, inflammation, and disease. The video is available if you're interested. I think that very interestingly today, Uri will tell us more about a topic on which he has collaborated with Ruslan Medzhitov, namely inflammation and fibrosis. The very exciting title of Uri's talk today is Simplifying Inflammation and Fibrosis. Uri, thank you again so much for accepting this invitation. I'm very excited about it, and the floor is yours.

Thank you. Thanks for working with me and to make this finally happen and really taking care of me with a lot of emails and the explanations, and I'm grateful., and I hope to one day visit you in person also. I apologize that I can't do that today. So, um, I want to talk about an issue that cuts across medicine, uh, and that is, uh, the fibrosis, which means excess scarring where you have an organ turn into scar, and this, um, this causes the organ not to function well and In fact, some people connect about 50% of deaths to fibrosis, which is, uh, can be, you know, heart attacks, liver fibrosis, lung turns to scar, lung fibrosis, and kidney fibrosis. And all these together are, uh, are major causes of death. But usually studied the different fields, heart fibrosis, liver fibrosis, and now we want to unify to see, to look what's unified. So Across, across medicine, you can see a pattern, which is injury, injury, and that leads to inflammation, swelling, redness, heat, and pain, and the immune system comes in, et cetera. And here we can go into one of two outcomes. So one possibility is in organs that can regenerate is healing. So you know that, like, you remember when we were kids, we got a, we got an injury that it was red, inflamed, and there was a scar. And after 2 weeks, scar is gone, everything's healed, perfect. But if we got a little bit deeper scar or a repetitive or persistent injury, and we have a scar that doesn't go away, and that's an example of fibrosis. We have some of these scars on us until today. I'm sure we all know those kind of scars. So One question we can ask right off the bat is, how can one system in healing have two outcomes, right? That's one major question we can ask. Second thing I want to say is that, um, fibrosis, it's not, it's not bad. It's an evolved physiological response. Is It's to prevent a hole in the tissue. You have a heart attack, you don't want a hole in your heart, so fibrosis fills it in. Or if you have a virus you can't get rid of, like in the liver, hepatitis, you have fibrosis around it. If you have a, a splinter or a bullet inside you, you coat it with fibrosis. So it's an evolved physiological response of the, of the body. But in fibrotic diseases, it goes out of what it's meant to do and replaces an organ with scar. And then you have, especially in old age, these diseases often happen at old age. Something goes out of whack and you have these fibrotic diseases. And we can understand how age, we'll talk about how age figures into this process. And across medicine, there's another interesting fact that there's a limited time window or golden period of let's say 2 days where if you remove inflammation, there's no fibrosis. But if you, if you don't remove inflammation, there will be fibrosis, even if you remove the inflammation later. So where did this golden period come from, or this time scale, and where you, after surgery, you want to reduce inflammation so there won't be fibrosis, etc.? So we want to understand that. And in this talk, I'm going to go from a mathematical theory to experiments that actually provide specific molecules that can reverse fibrosis in mouse models in the heart and in the liver. So we'll go all the way to understanding molecules a little bit towards the end, but we'll start with a conceptual picture. So that's— we will talk about the biology of fibrosis, a conceptual mathematical picture, and then some experiments and strategies to reverse fibrosis And I promise you all will understand everything, everything I say. And if you don't, we're going to have chats for you to send. And there's already 3 chats, 3 chats here. We do not see the screen slides.

People were expecting, I think, classic slides and now they got it.

Then there's chalk talk and then with no chalk. How cool. Great. Okay. So, I also wanna say this, fibrotic diseases, it's an unmet need. There's no drug nowadays that can reverse fibrosis in any case. So it's a huge unmet need to prevent and reverse fibrosis in humans. I mean, you can do things in mice, but to go to humans, I think you need new concepts. So I'm gonna start with a little drawing out a little bit about the biology of fibrosis. I'm gonna draw for you a little, tissue here. Just let's say the epithelial tissue. Some tissue. These are the cells. These are the nuclei maybe here of the cells. And here below, I'm going to draw, let's draw, yeah, I want to draw a little blood vessel here. And I'll make a little window just for you. I'll make a window inside it. Here. And we can see maybe a red blood cell here. And in the tissue, like all tissues. So if you look at tissues in general, by the way, biology now has— there used to be two fields, cytology and histology. Cytology, how cells are different, and histology, how tissues are different. There's no more cytology. There's cell biology. We look at what's in common to all cells. All cells have nuclei. All cells have Golgi. But there's still histology because we study tissues one by one. That's why we study fibrosis in each tissue separately. But if you look, there's also— there's a need, and Russell Medzhitov says this very nicely, there's a need for tissue biology where you look what's common to the tissues. And what's common to the tissues are, for example, that they're made of the same 4 building blocks. There's red— there's the blood vessels, there's fibroblasts that make the fibers, There's macrophages, which are like immune regulators, and there's the endothelial cells that do what— that are— that do what the tissue is supposed to do. They're like the violinist, and if the sink is broken, the violinist doesn't repair the sink, it calls the plumber, right? So that's the macrophage calling the fibroblast to repair the tissue. So these are the way tissues are built. So we have fibroblasts and macrophages. Now what happens when there's an injury, right? So now we have an injury, cells are dying, they secrete signals. The signals do two major things. First of all, they cause the fibroblasts to become big myofibroblasts, which are— myo means muscle. These are fibroblasts with fibers like muscle. This, the, the pull on the tissue to contract the wound, myofibroblasts, and they secrete a lot of fibers which are the scar. The second thing that happens are signals that tell macrophages or monocytes from the blood to go out of the blood and go into the tissue. And tissue certainly has 10 times more or more macrophages than it than the usual tissue-resident macrophages. And they come in to eat up the bacteria and they come in to do many tasks like that. There's also neutrophils. I'm going to ignore neutrophils that come into eat up the bacteria. They last about a day. And these fibroblasts and macrophages also interact with each other. And they're going to be the stars that we're going to talk about. So I'm going to talk about fibroblasts, actually myofibroblasts. And macrophages. These will be the variables in my theory. Okay, when there's healing, those extra fibroblasts and macrophages, after 2 weeks, vanish. They're gone. But if there's fibrosis, they stay, right? So that's— here I'm stopping the biology of And now I'm going to go to the concepts. But before I do that, I'm going to stop until somebody asks me a question by chat. So this is your opportunity to look into the chat and ask me a question. Otherwise, I will not continue. And of course, I do this to get a communication going and to see that I haven't lost you. So who's going to be brave and chat a question for me? I am.

Oh, ever—

you do it, you enable me to continue for everybody's benefit. Okay. I see some, I think. All right. So, however, the endothelium is active, not included in the concept. So the comment is the endothelium, which is the blood vessel, is not one of the variables I put in here. Did I understand the question? Yes, of course. So it's correct. And maybe I should say that Definitely the blood vessel gets changed by the inflammation and opens up and there's a lot of swelling and stuff. And the blood vessel, the cells themselves are very smart and they also secrete things and sense things. Also the endothelial cells are smart and they become senescent and they change and they regrow, et cetera. And now we come to a philosophical point, how you make a model and how you choose which variables to include and which variables to ignore. And of course, the model is a fiction, a lie that hopefully helps us find the truth. And the reason you might say that I want to start here and simplify and use only those two cells are they're the most dynamic. Their numbers change by, as I said, a factor of 10 or more, and they vanish. And that for me is an indication that if we make a population dynamics model like I'll do right now, they're going to be the important. I hope I understood your question. Angioendocrine role, or angio— can you precise the—

angiocrine roles.

Angiocrine roles. Angiocrine roles.

So angiocrine, it's a new concept that's out that in fact the endothelium in the vessel is an active organ that sends a lot of cytokines out and reprograms the tissue and also inflammation.

Sure. And that's seen now in single-cell gene expression studies, how much the blood vessels change your gene expression in inflammation. And they also, the adipocytes do that. And also almost every cell type now is becoming almost like an endocrine cell in itself, right? And then of course there's VEGF and all these calling in the immune system and all these. So maybe we should say that fibrosis is so complex and it involves thousands of molecules and dozens of cell types that, for me, it's an invitation to find a direction where there is simplicity. It doesn't mean to say that everything in fibrosis is fibroblasts and macrophages, but maybe from that point of view, we can understand answers to these questions of two outcomes and the time scales, et cetera. Maybe there there we can find some answers without pretending that they're the only thing there. All right. So the variables are indeed the number of cells. That's another question here. And I want to go on now to talk about what experiments we did. These experiments with Ruslan Medzhitov and my graduate student then, Miri Adler, Now she's a faculty member in Jerusalem. And Xu Zhu and other co-authors. This is a paper in Cell 2018. And then they also, iScience 2020. That's the theory behind it. So look, so what, what we did was to take an in vitro approach, which means out of the body. And grow in small plates fibroblasts and macrophages at different initial ratios, sometimes more macrophages, sometimes more fibroblasts. And then under the microscope, observe what happens over 2 days of growth. So this allowed us to make, to use a concept from the Enable Systems phase portrait that, and we constructed it as follows. This is the amount, the number of fibroblast cells in the plate, and this is the number of macrophages in the plate. Suppose I started with this initial condition, a lot of fibroblasts and a little bit of macrophages, and over 2 days, this is the change. I draw a little arrow. And if I started here, this is the change observed after 2 days. If I start here, it's like this. If I start here, it's like this. I can start on the x-axis with only macrophages. I can start on the x-axis with only fibroblasts. And I can do this. And I get, in one experiment, this entire phase portrait. And in fact, in a certain sense, it gives us a sense of the dynamics in one picture. And I'd like to walk you through it. So if I start with only macrophages, they crash to zero. And if I start with fibroblasts below, only fibroblasts below this critical concentration here, they also crash to zero. So a little bit of cells, they crash to zero. But if I have only fibroblasts above this circle here, concentration, they can support themselves in a steady state where fibroblasts divide and die. But they persist. And there's another state where fibroblasts and macrophages together persist. So if I move away from there, they go back. And by the way, when we did the experiment of doing from day 0 to day 2, and then from day 2 to day 4, and we can see that these arrows are consistent, meaning the system is time invariant. And in fact, there's a line in this phase portrait that separates all the conditions that crash to zero and the conditions that don't. And this line is called, in the language of dynamical systems, the separatrix. So what we have here is a situation with several fixed points. This fixed point here, we'll call it the healing fixed point. Why? Because there's no fibroblasts and no macrophages, like at the end of the healing process. And then these two points with high fibroblasts, we're going to call— give them names. These are new names for the field. This one we're going to call it hot fibrosis because there's fibroblasts and macrophages. Hot, and you can think like hot tumors that have immune systems. This means we call it hot fibrosis. We invented this term. And this fixed point we'll call cold fibrosis. It's only fibroblasts. All right. So we, from this picture, we can imagine there's two kinds of fibrosis. By the way, for those of you who like math, I want to say that in this particular experiment, the cold fibrosis fixed point is actually semi-stable. So it's, if you have only fibroblasts, it's a fixed point. But if you add macrophages, it goes up. So there was only, so this one is a stable fixed point. This one is semi-stable. And now this, by the way, feel free to chat. And there's a question about state of activation. So I want to say another complication is the macrophages can be in different states, M1, M2, pro-fibrotic, anti-fibrotic. Fibroblasts have a whole continuum of states, et cetera. So again, another simplification is I'm only considering their number here and not their molecular states, even though we have studied their molecular states. And I can tell you a lot about that. But for this purpose, we're only studying their number. So if I look at this face portrait, I can immediately draw the circuit. Fibroblasts can support themselves. That means they must be able to secrete growth factors from themselves. This is called an autocrine loop. And it's going to be very important for us when we try to reverse fibrosis soon. And fibroblasts also send factors to macrophages, and macrophages send factors to fibroblasts. And in the plate, we could identify what these are. This is CSF-1 and PDGF, et cetera. In the body, we also identified them, and I'll tell you in a second. But this is the circuit underlying this phase portrait. And the circuit is bistable. It has an off state, healing. And it has an on state. And there's a set of conditions that crash to zero and a set of conditions that go high. So to address this question, how can one system have two outcomes? Let's do a little simulation here on the screen of a wound, of an inflammation. All right, let's do that together. So let's imagine that I need to choose a color. So let's choose, let's say, orange. So I started with a little bit of fibroblasts and macrophages that are the tissue resident ones. And now I have an injury. Macrophages flow into the tissue from the blood vessel. And now my initial condition is here, but it's inside the separatrix. So what happens is they're going to incite the fibroblasts, but it's not enough. And they're going to crash, by the way, by this unstable fixed point, they slow down, slow down, slow down, slow down. And then they, and bam. So this is what healing looks like. Let's imagine now that my injury was a little bit longer. I have 3 days of inflammation. Now it's here. Now, no matter what, if I even stop the inflammation, I'm inside the basin of attraction of downstate. So I'm going to go like this near this unstable fixed point, slow down, slow down, slow down, and go here. So that's how one system can have two outcomes. If I have a repeated injury, I start healing, and now I have a repeated injury from here. Let's say I have another injury. Boom. Also, that's how repeated injuries can lead to fibrosis, whereas transient or weak injuries lead to healing. And this property, of course, is known as bistability. And it gives us several phenomena at the same time. The existence of two outcomes, the fact that mild injuries heal and repeated or prolonged injuries go to fibrosis. It also even gives us an interesting thing. This scar healing takes a couple of weeks and fibrosis maturation takes months, even though the cell turnover time is a day. And the reason is this unstable fixed point here. Near this unstable fixed point, like climbing on a hill, things slow down. And that helps us even explain the timescales of this problem. So we're very satisfied not saying this is the whole truth, but at least that this contains mathematical properties that are robust enough to be satisfying for us to explain some of these big questions. And I'm going to stop now until you ask me some things in the chat. So I'm not going to continue until we go forward. Oh, there's a new message here. It looks like the FNM condition going to hot fibrosis are reaching an equilibrium. Does it mean this in vitro model there's an optimal ratio? So I suppose what you mean, you mean that this on state is an equilibrium. So I would say more precisely, it's a steady state where cells are continuously dividing and dying. We know that from the in vitro that they're dividing and dying. And the ratios indeed from different initial conditions converge to the same ratios. That's correct. We started with 30 different folds. Yes. A vast— I don't remember now— 2 orders of magnitude difference in initial condition ratios, and they converged to within a factor of 2. So absolutely. Did I answer your question? How do we get to cold fibrosis? Good. So in this picture, we never get to cold fibrosis unless we start with only fibroblasts. But I want to tell you that in this circuit, there are parameters. Let's call this one parameter A, the autocrine loop. This, let's call this parameter B, and this parameter C. And if parameter B becomes weak enough, then the fixed point, stable fixed point becomes cold fibrosis and not hot fibrosis. And we believe that the different tissues are programmed to have either hot fibrosis or cold fibrosis. We know, for instance, that heart attacks, we know from experience we did together with collaborators that heart attacks, the fibrosis there is cold fibrosis. And we also know that in the kidney, glomerular fibrosis is hot and tubular fibrosis is cold. So the same organ can have two different outcomes. And I think it's by tuning this parameter, which is basically strength of secretion of factors and turnover times, et cetera. Did I answer your question? How do you quantify this increase of cell number, cell matrix deposition? I'm not sure I understand the question, but I wanna say that cell numbers you can definitely measure. And I also wanna say that in this on state with the hot fibrosis and the cold fibrosis, there's also much, much more deposition of fibers in the plate. We measured that. And in fact, cold fibrosis is even more than hot fibrosis. And macrophages degrade the fibers. So hot fibrosis is different kind of fibers stiffness than cold fibrosis. Cold fibrosis is tougher, hot fibrosis is softer. And in the field of dermatology, keloid scars are hot fibrosis and hypertrophic scars, which the flat ones, are cold fibrosis. So we think there's a whole world for pathologists to think about. Okay, thanks. We should consider this point would be modified in particular fibrotic diseases. Yeah, okay, let's continue. So I'm really grateful for the questions and I feel like I'm having a conversation with you. I hope I'm answering the questions. And if you guys say yes, you answered my question, or no, that will also help me. So please do that so I can know, because I can't see your face, so, you know, I'm trying to communicate into the void.

Yeah, I think Nick said you answered his question about the cold fibrosis, and I think Jonathan wanted to add something. So I don't know if Jonathan, did you mean that the equilibrium you talking about would be different in different fibrotic diseases. Maybe you can explain that in the chat, Jonathan, if you're still here.

Yes, sorry, I'm here. Can you hear me?

Yes.

Yeah, yeah. So the answer, yeah, was good, but I was just adding that maybe my question was whether this point, this equilibrium, this steady state would be modified in certain diseases, like, you know, having fibrotic— But how often happens the speaker can't fully understand the question, so it's very good to clarify. So we don't have a lot of examples so far because it's so new, but we did do experiments on this. We repeated this experiment with Ruth Chaval on fibroblasts from different organs, and we observed that the positions of the cold and hot fibrosis fixed points were very similar. I wouldn't say very similar, it was within a factor of 2. For fibroblasts, for different tissue rather than fibroblasts in this experiment. So it's definitely the case that different tissues should have different parameters that will determine different positions for the, for these. And I would say that's a wonderful question for research because there's not enough data about that. But definitely they can move, they can move around depending on those parameters. Did I kind of get your question? Yeah, and there's a question here in cancer. So one of the things we're doing right now is using this concept to understand the tumor microenvironment, which looks quite a lot like fibrosis but is also different. And I can talk about that maybe towards the end of the talk too. But definitely the same picture we think applies to cancer microenvironment, except you need to take into account the T cells and the B cells, the parasites and the mast cells, and all the, all several other endothelial cells and all, and the cancer cells. So it's more complicated, I think, but maybe these concepts can help there. I mean, I know that they'll help there because we're in the middle of it, but okay. Now I want to talk about how we can reverse fibrosis, how we can use this concept for healing fibrosis on the computer. So to speak. All right, so the, the way to do that is to— maybe before that I'll say something about aging. Aging changes these parameters. For example, aging reduces turnover rate of cells, and that means it moves the separatrix. And it moves the separatrix— we analyze this in such a way to reduce the basin of attraction of healing so that injuries that used to be healing are now going to be fibrotic. So general things that aging does, like reduce replication rate of cells and increase inflammatory factors in the, in the body because of cellular senescence, reduces the space of attraction, and you have more chance that a given event will reach fibrosis. And that can help explain the exponential rise in incidence of fibrotic diseases with age. And I could talk about the exponential too, but if you want, but basically exponential comes from stochastic processes crossing, like tunneling, tunneling through a separatrix, and that gets exponentially larger with the barrier, which reduces with age, linearly with age, you can say. So you have exponential rise in these diseases. If you're interested in aging, I can, we have a whole theoretical and experimental project on aging, why we age, and the incidence of age-related diseases, and mathematical essence of aging that I can do a different time with you. But fibrosis is one aspect of aging. Let's now talk about a medical problem of how can we prevent fibrosis and even abolish existing fibrosis. All right. And to do that, what we did is on the computer, we varied the different parameters, A, B, and C, and see what they did to the face portrait. And of course, you can even analytically do it. And we find that there's one parameter more important than all others, which is the autocrine loop strength. If you reduce the autocrine loop strength, what happens is Let's look on the x-axis. Fibroblasts now need more fibroblasts to support themselves because on the x-axis, only fibroblasts is only this autocrine loop, what they secrete to themselves. So if you reduce it, you move this point here. So I need a new color. Let's do, we're going to heal. So let's do like some kind of blue here, a nice soothing color. And that moves the entire separatrix with it. So if I reduce the autocrine loop, fibroblasts can no longer support themselves. They need higher numbers to support themselves. And this increases the situations that go to healing. And if I increase it a little more, this point collides with this point. And now the separatrix looks like this. And now we have a huge region that goes to healing. So we like that. I really apologize because this could be one of my daughters and we have a medical situation. So please, hello. It was the phone company. So sometimes my daughter comes from calls from an unidentified number. That's why I have to answer. All right. Sorry about that. So what we're going to do is we're going to inhibit the autocrine loop. And another interesting thing that this theory says is you don't need to inhibit it to zero. You need to inhibit it just beyond a certain point. And that point is, it's just like in COVID, you know, you want the replication number to go below 1. We want each infected person to infect less than one other, right? Same thing here. You want each fibroblast to allow the production of less than one other fibroblast. And that's very hopeful. And in the history of science, those messages are very hopeful. Like in the beginning of the 20th century, when there was eradication of malaria, a similar model like this said that you don't need to kill the mosquitoes or prevent the mosquitoes to zero, only below a threshold number. And that, you know, with all those nets and stuff, that was a key result in epidemiology. So similar to here. So when we gave this talk, there were a couple of experimental groups that could understand. I hope you're understanding me, for example, right now. I'm making a big effort to be understandable. And they said, this is important enough for us to do experiments using our mouse models. And what they did is identify the autocrine in vivo, the autocrine factors, and found a way to inhibit them and then tested this results on established models of heart fibrosis and liver fibrosis. There's a question. Which autocrine loop? There are many. PFGFs, PDGFs, TGF-beta. Yes, yes, exactly, exactly. So which of these should we intervene, right? And maybe there's— so we all know, people who look at fibrosis know that if you knock out TGF-beta or intervene, it doesn't work, right? So a lot of things have been tested before, it doesn't work. But maybe we can find something that does work here. So this is exactly the question. Thank you. I wonder who, who said that. It's me. Yeah, thank you. And are you an expert on fibrosis?

Yes, I'm expert in also on cancer and blood vessels also.

Super. Okay, so honor for me to have your inputs. Okay, is what I'm saying so far, um, make sense to you, Andreas?

Yeah, yeah, it's fine, it's, it's great. All right, so, uh, just another question. So which system you are using, human or mouse?

Yeah, let me— I want to get to it right now. Right, so far this was all in mouse. Okay. So then, so I'm gonna talk about two, two experiments, one on the, on heart attack, which is called, of course, myocardial infarction. And the other one is gonna be on liver. And the, this is what's called cirrhosis. It's— that's the word in liver for fibrosis. So the work on mouse was done with and his student. And what is going on here? So what they did was they take the mouse and the mouse has this artery, LAD. Which feeds, gives oxygen to the heart, part of the heart, and they tied it up with a piece of string. That's called ligation. And therefore these cells here don't have oxygen and they die, and that's a heart attack. And then they're replaced with scar tissue. So Shobhat first asked, is this hot or cold fibrosis? And he built a mouse where the cardiac fibroblasts and myocardial fibroblasts are in green and red, GFP and RFP, and he could find that you have an increase in both and then a crash to cold fibrosis. So this is a situation of cold fibrosis in the heart. And then he wanted to ask, like, address what is the autocrine loop, which molecule. And for that purpose, he did single-cell RNA sequencing in both mouse and human and looked for, what he looked for was factors that fibroblasts, RNAs, their molecules that fibroblasts both secrete, and also that they have receptors for. And he found, and he looked at that list and he crossed it with human. He also wanted to find things that are autocrine loops of myofibroblasts in the heart for humans. And he looked for molecules that aren't secreted by any other cell, not the normal fibroblasts, not the normal blood vessels, only the myofibroblasts in the scar. And he found, and Drast's list here, he found FGF, PDGF, TGF, and then he also found a molecule that was less studied as a growth factor, tissue inhibitor of metalloproteinase 1, which usual function is to inhibit the molecular scissors that macrophages secrete that chop up the fibers. But he showed that this molecule is actually a growth factor for cardiac myofibroblasts. And then he used an antibody that inhibits it and he injected, he did this heart attack to mice. So he took the mice, he did the heart attack, and then he injected this antibody immediately after the heart attack and also after 3 days. So this antibody inhibits one, one of many autocrine factors of the fibroblasts. So we think it's a dominant one in this case. And then he cut up the hearts. So this is a cross-section of a mouse heart, and here is the Sirius red staining for fibrosis. This is a normal heart attack. And with the antibodies, you had fibrosis that was restricted, didn't go through the heart, and a 50% reduction in fibrosis. So inhibiting the autocrine loop here made a very large reduction in fibrosis in this experimental animal. And okay. All right, and I want, I want to talk about the liver cirrhosis experiment. This was done with Scott Friedman at Mount Sinai. By the way, that's Dwight's man. And this paper was published, Science Translational Medicine 2023. This one, you can find this one on bioRxiv. It has not published yet. You want to take a look? Here, it's also an interesting story. Here I'm drawing a liver. And you know that liver cirrhosis used to be mainly in people who had alcohol use and damaged your liver or had hepatitis that made the virus. And nowadays liver cirrhosis is rising very much because of a fatty liver. If you have fatty liver, and 30% of the world has fatty liver, you can go to inflamed liver, NASH. It's called NASH. I'm sure maybe some of you have heard of this. And this NASH in a fraction of people goes to fibrosis. And that means sometimes the only cure is to replace the liver. And in about 7% of those people with fibrosis, it goes to liver cancer, which is a terrible deadly cancer. It's rare, but one of the most deadly cancers. Oh, here's a question. Does TIP1 inhibition, besides blocking fibrosis in the heart, results in additional non-fibrotic injuries in other, is there less fibrosis but the epoxic tissue is still functional? So the question here, if I understood correctly, is, We have less fibrosis here, but what happened to that heart tissue that died? I mean, heart can't regenerate. So what happens? Is the heart still functional? Did I understand the question? Yes. So what happened actually is quite surprising. It looked like the heart muscle, instead of, it didn't divide, but it stretched itself and reintercalated and made a perfectly functional heart. And we cannot detect any reduction in, ejection fraction or any heart function. Very interesting. I think that because it's one of the first times you can see what happens if you prevent fibrosis in a non-regenerating organ. It's not regenerating, but it's coping. I think that's interesting too. Now, the liver is regenerating. Okay. So we have a different situation here. So what Scott Friedman is expert in is these NASH mouse models. That you take the mouse and for 28 weeks you give them a toxin, a liver toxin, CCL4, together with high-fat diet. And after 28 weeks, they have cirrhosis and their liver looks like a piece of rubber. It's full of fibrosis. So they did the same experiment using single-cell RNA-seq in mouse and humans, they found that the, my fibroblasts in the liver, they're called hepatic stellate cells. And there's normal ones, and there's a new cluster of hepatic stellate cells that's found only in the fibrosis. And they have an autocrine loop. In fact, both in the heart and in the liver, the fibroblasts, if you look at all the interactions, the fibroblast autocrine loop is the most massive interaction in the tissue. It has the most amount of RNA for the secreted molecule and receptor. It's huge. The autocrine loop of myofibroblasts is the most dominant interaction in both tissues. And you look what factors are there. Again, you find addresses FGF, PDGF, TGF-beta, the usual suspects. And a new one that was intriguing is called neurotrophic factor 3 and its receptor, which is called NTRK3, which is a cancer receptor, actually a tyrosine cancer receptor, because it turns out that these stellate cells look like neurons, have neuronal projections, and they touch each other like this, the fibroblasts, and one projection secretes and the other one receives the signal. And in this case, to inhibit the autocrine loop, they used a drug that is a cancer drug that inhibits the kinase here. It's called LOXO. It's a drug. And then after 28 weeks, they gave this drug 4 weeks, and then they looked at the liver. So normal or control mice, full of fibrosis. In the treated mice, you have 50% reduction in fibrosis again, and the liver has full of regenerating pockets of healthy liver. We're now studying it, but it looks like this treatment also reduces the amount of liver cancer because these mice, the male mice always get liver cancer. So it's a very nice model also for liver cancer prevention of fibrosis. We're beginning to see this is preliminary, also prevents the liver cancer in this model. So what have I told you in two different situations, an acute injury in a non-regenerating tissue and a chronic injury of regenerating tissue inhibiting the autocrine loop, seems to, not seems to, inhibiting the autocrine loop reduces fibrosis significantly. And, and we're very excited about that. We want to expand this to, as I said, the cancer microenvironment and the many other kinds of fibrotic diseases that are important in medicine. And, okay, so there's more questions. When TMP1 inhibition outcome, does it mean the different autocrine loops are not fully redundant? Yeah, so the question is, the autocrine loop is composed of many different factors. I think I wouldn't call them redundant because it's likely that they have many functions, not only to make the fibroblasts replicate, but also change their gene expression. And TMP1 has an entirely other function, which is to stop enzymes that cut up the ECM. So I don't think they're redundant. I think they have specific functions. But like I said, with the malaria or the COVID you need to cut down that arrow below threshold. So it could be that we just managed to inhibit enough of the total proliferative effect below threshold, and then we could see this effect. And maybe we could have done it by a combination of other ones too. I hope I understood your question. And you can say yes or no. And Fridolin Gross says, is a fibrotic state an unavoidable byproduct of building a circuit for generation, or can you think of a smarter way of building the system? So is fibrotic inevitable? And we know it's not inevitable. For example, in embryos, if you damage them, they will not get fibrosis. Well, after their— and also mice, until day 7, their heart doesn't get fibrosis. After day 7, they do it. I think fibrosis is an evolved response to the fact that there are pathogens and dangers not in the utero but outside. And I think it's definitely designed and evolved, and that's, I think, also why we can intervene and abrogate or remove it, I think. And I think the different tissues will have different set points for their circuits, and I don't think it's inevitable, but I think it is designed. I think it is a physiological design that just with age gets abused. The problem is causality, how to construct a causal network to simplify the interactions. This is a comment by Andreas. So maybe you can say more, Andreas. I'm not sure I understood your comment, but if you can write some more, I maybe can address. Okay. Yeah, I think it's a good time to stop and maybe go to I hope I didn't forget anything. You see, it's all the lectures in a single whiteboard here so that we can see everything I said. And yeah, this is a great time to learn from you. So please tell me more. So what do you think and what questions do you have? We can maybe go to voice now.

Sounds great. So the first thing I suggest is that we clap to thank Ori, and then indeed we can move to the general discussion where maybe it's easier if people just talk and speak instead of write. But both options, I'm sure, are possible. So if you prefer to write in the chat, it's also perfectly possible. One thing just to start is that, I mean, Ori, you mentioned several times what you were doing right now about cancer, and many of us are interested in cancer repair and fibrosis. Can you say a little bit more, or it's just too secret to share that with us right now?

And I don't believe in keeping secrets. I mean, we came to science to share. And, uh, and I mean, of course there is theft and cheating in science, but I think the more we share, the more that becomes, uh, uh, as a community we can, um, be more effective at reducing that phenomenon. So the work on cancer microenvironment is together with Ruth Scherz, Chauval, and Weitzman. And we started with the same experiment as with Ruslan, growing cancer-associated fibroblasts and tumor-associated macrophages in a plate. And we found basically the same phase portrait. And then going into the single-cell RNA sequencing, we could map out the circuit between multiple cell types. And in fact, to our wonderful surprise, it looks like it just turned our idea of the cancer microenvironment on its head because what we used to think is that cancer is the main player rewiring the environment. But when we look at the network that we find, we find that it looks like regular fibrosis, fibroblasts at the top with an autocrine loop instructing macrophages and all the other cell types and cancer kind of in the middle and it's mainly talking to the immune cells, T cells and B cells. So it looks like to us now that cancer is riding on an evolved machinery that is wound healing, and it's manipulating it to say, I'm a particular kind of wound. I'm a wound that's more like some kind of tissue repair that shouldn't be attacked. I'm not a pathogen. And that's very interesting to us. And then we're trying to characterize these interactions in order to see how we can modulate the cancer microenvironment into desirable states. That's very— so that's where we are, and, and our model there is breast cancer. But really excited to see where this concept can take us in, in, in cancer.

So, so maybe I can ask, uh, So question after this. So, so the talk was great, thank you very much. Thank you. Was really nice and, and I like it a lot. So the, um, so the thing, there are in fact two is—

there in fact three or four or five or six or 10 issues, maybe 100.

Yeah, there is, there's so many things to, to talk, but the thing is that the, uh, one, one thing is that a tumor is not the same in one organ or the other. So you have an organ-specificity. For instance, when you use— when you look on pancreatic carcinoma, ductal carcinoma, so they have huge fibrosis around. And this is why, for instance, the T cells, they cannot enter. It blocks around. So they are— and inflammation is blocked. So in fact, in the PDAC, it's a cold It's a cold fibrosis.

Are you sure it's cold? It's cold in terms of T cells, but is it cold in terms of macrophages?

It's also down. I mean, it's not, you know, the big problem is that the inflammatory cells, they cannot go in. Yeah. And then you have the other type. Let's see, this is the opposite. You have hot tumors. So you have, for instance, the glioblastoma. Yeah, it's a hot tumor. So in glioblastoma, you have a lot of influx of all kinds of cells, except that you have, of course, an, um, let's say, immunity, an adaptive immunity that is not as strong as in colon and so forth. But at least it's a— so you have these extremes. And I think the things that you have and the models that you get are not universal. Okay, so they should be in fact the specificity in each of the models. Yeah, some of them, if you look on brain, if you look on pancreas, if you look on liver, if you look on lung and so forth, it will be very different. And I think also the factors are very different.

So this is an invitation to go beyond a simple and universal picture to an organ-specific and more—

Exactly, because there's the organ-specific tissue, and now, for instance, in the vessels you have organ-specific endothelium.

Sure.

Okay, so it's not at all the same endothelium that you have in the lung and so forth, in that tumor and so on. Yeah, people have done it with, with single-cell sequencing So the thing is really present in the lung.

There's in the healthy lung, there's two kinds of capillaries, one for gas exchange and the other one for— I forget what. But this single-cell sequencing is opening our eyes to see the tissue specificity of all these factors. And of course, in this model, you can have a situation where it's depending on the parameters, one tissue will have cold fibrosis, another tissue will have hot fibrosis. So this model can incorporate different situations for different tissues within a single picture. It's also important to incorporate more cell types like blood vessels, et cetera, and maybe use this as a core to see how dream towards a model of the full organ that's organ-specific and has all the complexity. And I think that's definitely the direction for going for accuracy. I completely agree. Great. Thank you.

Other questions? Maybe from the philosophers in the room. Many people are interested in aging. I was thinking that maybe there would be some questions about aging, maybe not. But Uri was sort of launching a sort of invitation to ask more questions about that. He has a full theory about this. I don't know, Fridolin, if this is what you're going to talk about.

But anyway, Fridolin.

Yeah, in a way, I mean, I wanted to ask about this, like, to what extent do you think this is kind of an universal principle or theory underlying aging.

This idea of, if I understood it well, it's kind of noise-induced transitions toward a kind of diseased steady state.

Yeah, I'll be happy if you invite me to do a full talk on aging and the— in a similar style, right? Yeah, that, that's definitely something I enjoy doing and would be open to. But, um, the way we think about aging is, um, is, is, um, first of all, amassing as many quantitative patterns as we can find, like The fact that the risk of dying rises exponentially with age, or the incidence of disease rises exponentially with age and then drops at old age for, like, we have databases of hundreds of diseases like that. Or that if you give an intervention in fruit flies that increases lifespan, you go from the bad to the better survival curve within a day. So there's a slow timescale of the lifespan and the fast timescale. And all these together, just like here, I use the fact that there's two outcomes to think of a bistable model that's supported by mechanism. There we're trying to think of all these constraints. What kind of mathematical model can— it's one mathematical model can explain all those different quantitative patterns. And then we also do experiments to, to test that. So the, the model that we converge on is that, um, there is a damage causal for aging that is produced by something that accumulates linearly with age. We have ideas what that is, and it's removed by a process that gets saturated by the damage itself. And mathematically, this creates a situation where production that inevitably exceeds the maximal removal rate and you have a catastrophe. And when you add noise to this, any kind of noise, you get a stochastic or statistical process that has all the features that I mentioned and more, many quantitative features. We're actually amazed at how many quantitative features this explains and we keep testing it. And it also suggests very specific experiments that some of them we've done that give kind of surprising outcomes. So I'm right now very excited about this particular mathematical model of damage production removal that also has features that can be universal. For example, bacteria and yeast and worms and stuff like that, each one of them has a different chemical species that is damaged. But the property of production rising with age and removal that saturates, I think, is universal. So you have similar statistical properties in different organisms. And to someone like me who comes from physics and has this problem that I believe that there has to be simplicity, and then it's very satisfying. But, or maybe it has, maybe it's more like a disability because I can't do science without this belief. I hope I, I hope I answered the question. And, and of course, maybe I make a little advertisement. I'm, I've written a book called Systems Medicine, a new one, and, and it's going to be published hopefully in November. But if you want to look at it, you can take a look at my website, uh, on the course called Systemic Medicine, and you can see the lecture notes there are basically the same as the book. And there's 3 chapters that develop this aging theory. There, there's one chapter on fibrosis, which is basically the same talk I gave it to you today. So this is the style of the book. Chapter 5 is what I gave— I talked with you today. And chapter 6, 7, 8 are about aging. And, and, and then there's something about hormone circuits and diabetes and the stress axis and autoimmune diseases. There's an explanation of autoimmune diseases and everything is derived analytically. Yeah. So this model is simple enough that you can solves this phase portrait analytically. You can do the nullclines and solve everything analytically. And you can find that the autocurrent loop is important analytically. And for some reason, this model has the property that you can solve it analytically. I have a suspicion that biology evolution converges in circuits that you can solve analytically. Because otherwise they're too unreliable. And if you can solve it analytically, they have a very, very stable kind of deep kind of structural stability that is useful for systems to work in the real world. I think engineers have a similar belief sometimes. If you need simulations, you never know if you simulated every possible situation, but analytically you can prove it. But that's maybe kind of a belief system that comes maybe also from physics. But for people who like to solve things on backs of envelope like me, this is a great field to be in.

Rory, I have a very practical question. You mentioned at the beginning of your talk neutrophils and the fact that neutrophils would not be at the center of this talk. Can you say a little bit more about that? I'm asking this question because some of us worked on the role of the immune system in tissue repair, and neutrophils were really the key cells. I mean, macrophages were very important and fibroblasts were very important, but neutrophils were absolutely crucial in the process of healing. And I'm just wondering if you can say a little bit more about the role of neutrophils in the therapeutic process.

In what stage and how are they essential?

So I don't know if people more competent than me who are right now in this virtual room want to say more about that. But the— as so anyone Okay, so, oh, Marie-Lise, you want to say more? Would you agree to say more?

Yes, of course. I don't know if we said that mutual feelings are really the most important cells in the process, but it was very important very early in the process. So that was the point of our study and our publication because Actually, we showed a very dynamic process and that it was— but finally you show also quite similar things with your very elegant demonstration. But time is very, very important. And at the beginning, neutrophils are very important and they can also change their face as macrophages do. I don't know if I'm clear. Absolutely. And I agree. As far as I know, I'm not an expert, in this initial rise here of macrophages, I don't know, can you see my thing? So in this, in this initial rise here of macrophages, which takes a day or two, the first responders are neutrophils and they go in. And as far as I know, they also go down faster. So maybe there are some situations where that's not the case, but that's what I understand. And therefore you can say that this is philosophically, how do I choose fibroblasts and macrophages as my variable? One of the reasons is the time scale that I'm worried about, or that I'd like to understand, which is the time scale of weeks. If I'm interested in the time scale of weeks, maybe I can ignore things that are important in the time scale of days and then not so important And I agree. Initial conditions. So that's one method to focus on fibroblasts and macrophages and ignore neutrophils to a first approximation. Thank you.

Okay. I don't see any other questions.

Are there other questions that people—

I have maybe one. Go ahead, Andreas.

Okay, so when you look on the macrophages, you have in fact a number of subtypes, as you already mentioned, M1, M2, and even M3 and M4 now. But the thing is also you have the resident macrophages, and often the resident that are not coming from the blood are more abundant. So have you done anything So on that, to look which type is it more, the resident macrophages or the influx of the—

So here, these residents are here, the green ones here, right?

Yes.

And the red ones are the monocyte-derived macrophages coming?

Yes.

Well, in the— where we looked most carefully, which is the heart MI, it's clear to us that there's a huge influx coming in from the blood, and after 2, 3, 4 days, it crashes, and you get to a situation where there's myofibroblasts that are persistent, but there is a small population that looks to us like tissue-resident macrophages in the end. So the cold fibrosis has no— has zero— let's say this axis now, let's make it more precise. This axis is now monocyte-derived macrophages. Yes, they go away. So, um, yeah, it's very interesting, uh, distinction. By the way, another thing I can say is that another line of work— there's, in biology, there is a lot of tendency to think categorical, but when you actually look at gene expression —of single cells, you find that often it's a continuum. And so the way we mathematically describe continuum actually is by looking at the shape that encloses this continuum and thinking of these cells as archetypal cells. And there you find M1, let's say M2, M3. If it's a tetrahedron, you find M4. And they have very, very special programs, but there's also some cells that are generalists in the middle. So this entire topic, we call it Pareto task inference because it talks to us about trade-offs in gene expression between cells. And we find that both macrophages and fibroblasts have this kind of continual enclosed inside polyhedra and that That helps us to understand things in a more, let's say, I think it's more biological way. Of course, sometimes you have two clusters of macrophages and then it's clearly two different kind of problems. A lot of times it's a continuum.

And maybe just the final, so final question for me because I speak too much often. So just the final is what is in fact triggering the autocrine loop? Because of course you have the transformation from the fibroblast to the myofibroblast.

We know that.

But what are the cues doing that?

So if you ask what transcription factors, for example, activate—

yes, or either transcription factors or factors from the outside. Yeah. Because you have to shift from the fibroblast, from the normal ones. Yeah. The—

and the—

and the myofibroblast or the cancer-associated fibroblast. Correct.

So I don't know the answer, but it's clear that the autocrine loop for fibroblasts— and so it's clear that the autocrine loop of fibroblasts is different from the autocrine loop of myofibroblasts. And we assume that in this differentiation process, there's special myofibroblast transcription factors that are triggered and epigenetic changes that allow the expression of these myofibroblast-specific growth factors, like TMP1, for example. Yeah, yeah. And we don't— we haven't studied, but it's possible to do, of course, to see what's the trigger. I understood your question. I don't know the answer. Okay, thanks. All right, I think it's good if we could stop here, Thomas.

Yeah, that's, that's, that's very good. Um, and really, uh, we love your talk, Uri. I had several comments in the chat just to me. People really love the talk and also the methodology of the talk. So really, thank you, thank you so much for that. And, and we hope to invite you in Bordeaux and that it's going to be possible for you to visit us in person in the, in the near future.

I'd love to and to meet you in person and learn from you. And I already learned things here, so, um, seems like we have a good set of common interests. And I appreciate it. So thanks again, Thomas, for keeping in touch and making it happen. Thank you.

Thank you very, very much. Have a great day and see you soon. Bye-bye. And thanks again.