Hi. I'm Lily Dara from the University of Southern California, and I'd like to welcome you to the 2021 State of the Art High Zimmerman Hepatotoxicity Lecture. It is my distinct pleasure to introduce this year's speaker, Dr. Laura Nagy. Dr. Nagy is Professor of Molecular Medicine and staff in Inflammation and Immunity and Gastroenterology and Hepatology at the Cleveland Clinic. Dr. Nagy also holds adjunct professor appointments at Case Western Reserve University and Cleveland State University. She is the Director of the Liver Disease Research Center in the Lerner Research Institute. Dr. Nagy's research program is in the area of signal transduction and innate immunity in non-alcoholic and alcoholic liver diseases. Her research has been continuously funded by the NIH and the DoD. Dr. Nagy is the recipient of an NIH Merit Award and is principal investigator for the Northern Ohio Alcohol Research Center. She is past chair of the Board of Scientific Counselors and currently serves as an advisory council member for the NIAAA. She is also the associate editor for both hepatology and alcoholism clinical and experimental research. Her talk is titled, Inflammation and Cell Death in Alcoholic Liver Disease. Take it away, Laura. Thank you, Lily, for the kind introduction and thank you to the ASLD for inviting me to present the Zimmerman Hepatotoxicity State-of-the-Art Lecture. My title today is Inflammation and Cell Death in Alcohol-Associated Liver Disease. I have no disclosures. For the outline of my presentation, first I'll provide a background on innate immunity and ALD and then tell you two short stories, basically. The first about inflammation with data that we've collected using single-cell RNA-Seq, and then another short story relating how inflammation impacts cell death with a particular focus on RIP3 and MLKL. I also want to emphasize that I have a few themes running through the presentation. The first is that by understanding mechanistic insights, we can also move to translational insights so that the basic science can actually move to translation. Also, I wanted to emphasize the tremendous value of collaborations that's been ongoing in the field of ALD in the recent years. First, to acknowledge the people in my lab, the work that I'll present today was primarily done by two postdocs in the lab, Adam Kim and Xiao-Huynh or Kieu Wu, as well as a previous fellow, Tatsunori Miyata. I'd also like to acknowledge my collaborator, Jiajia Li, here at the Cleveland Clinic, and then multiple groups for providing clinical samples from patients with alcoholic hepatitis, including the Johns Hopkins R24 Resource Center, Sally Sun, Idibops in Spain, with Pao Centro Brew, and Ramon Batele and Juan Caballé, the DASH Consortium, and also my local P50 funded Northern Ohio Alcohol Center. Of course, the work has been funded by NIAAA, and in particular, I'd like to thank the Alcoholic Hepatitis Network. The natural history of alcohol-induced liver disease progresses from steatosis to fibrosis, cirrhosis, with a high propensity for hepatocellular carcinoma. Not everyone progresses along this continuum, likely there are both environmental and genetic factors that contribute to whether a heavy drinker will progress along this chronic continuum. However, at any stage of the disease, an individual can move into the state of alcohol-associated hepatitis. This is characterized by neutrophil infiltration, inflammatory cytokines, and abundant hepatocellular death. Essentially, it becomes a situation of non-resolving inflammation. If we look schematically at the liver, we can see that multiple cell-cell interactions are involved in the development of an injured liver. In the normal liver, hepatocytes make up the bulk. While there are a number of cells within the hepatic sinusoid and the space of disc that contribute to homeostasis. We have hepatic stellate cells, which upon injury can become activated stellate cells, which generate extracellular matrix for fibrosis. We also have a number of cells of the innate immune system, including dendritic cells, NK, NKT, and the resident hepatic macrophages, the Kupfer cell. Upon injury, the hepatocytes undergo a situation of what you can call organelle stress, and are much more sensitive to death receptor signaling-induced cell death. We also see that the immune cells within the sinusoid are activated in terms of Kupfer cells or NK cells. We also see infiltration of neutrophils that become activated and contribute to inflammation as well as infiltrating monocytes. In today's talk, in the first section, I really want to focus on the changes that occur in the sensitivity of these innate immune cells, particularly the Kupfer cell and the infiltrating monocytes to activation by ligands. This is just a very schematic diagram of the interaction between inflammation and hepatocyte injury in ALD. We know that ethanol early on can activate the complement system, leading to the generation of the anaphylatoxin C3A and C5A. We also know that alcohol impairs the barrier function of the intestine, allowing for the influx of microbial metabolites such as LPS or beta-glucans into the portal circulation. The resident Kupfer cells or infiltrating monocytes are then exposed to these ligands and can be activated to generate inflammatory mediators. Here I'm just showing TNF-alpha as a prototypical cytokine, but I'll show you later the great variety of inflammatory mediators that are generated by these innate immune cells. These inflammatory mediators in turn can interact with receptors on the surface of hepatocytes contributing to steatosis as well as hepatocytic death via multiple mechanisms including apoptosis, necroptosis, and pyroptosis. We started on this adventure that I'll tell you about in terms of studying inflammation by investigating LPS-stimulated signal transduction in peripheral blood mononuclear cells from patients with alcohol-associated hepatitis or healthy controls. We had noticed earlier that hepatic macrophages isolated from animal models or mice or rats exposed to chronic ethanol exposure developed a sensitivity to LPS stimulation via TLR4. Importantly, we can model this or it's actually shown, it's not really modeling as the mice or the model. But anyway, in humans, we can see a similar response so that in response to stimulation with LPS, we can see that cells from healthy controls will of course make TNF-alpha in response. But cells isolated from patients with alcoholic hepatitis show an increased sensitivity to the LPS. I'd like to bring up an important aspect of our work is that we're working with very low concentrations of LPS. We've based this design on studies coming from patients where a single binge alcohol consumption causes short-term increase in peripheral blood LPS on the order of 100 picograms per mil. Similarly, patients with severe AH have a chronic ultra-low dose peripheral concentration of LPS, again, on the order of 100 picograms per mil. This is quite different than the typical studies using LPS that are modeled after situations of sepsis, where we can have concentrations of LPS in the milligram concentration. Adam came in my lab then proposed the question, why are monocytes from patients with AH hypersensitive to ultra-low concentrations of LPS? In order to answer this question, he made use of single-cell RNA-seq to reveal distinct immune cell functions in patients with AH. We basically asked two questions. The first is, are baseline or LPS-stimulated gene expression different in monocytes from healthy controls versus AH? Second, taking a relatively novel approach, we asked about immune surveillance by asking, is the expression of other PAMP or DAMP receptors increased in response to low-dose LPS? These studies were carried out in four patients with AH. This is severe AH with MELD scores greater than 21, as well as four age-matched healthy controls. They were stimulated with this low-dose LPS, 100 picograms per mil for 24 hours, and then we made use of the 10X platform to do single-cell RNA-seq. Adam mapped out the distinct immune cell populations in patients with AH and healthy controls. You can see we can account for nearly all the circulating cell types except neutrophils. We can't find neutrophils in the PBMC fractions. We have basically identified four clusters of monocytes that have distinct immune cell functions. We also have NK cells, cytotoxic T cells, as well as CD4 positive T cells, B cells, and a small population of dendritic cells. For today's talk, we're really only going to focus on the monocyte populations. I'm going to show you a series of violin plots, and I'll take you through this first one. That's a bit simple, but then later on, you're just going to have to look at the patterns of expression in terms of the black and the red. The way we have this set up is that the healthy controls are on the left, and then AH on the right. Basal, so non-stimulated is in black, and then LPS is on the right in red. For CD14 positive cluster 1, you can see that IL1 beta expression is induced a little bit in the healthy controls, but robustly increased in the PBMCs from AH. Similarly, the chemokine CCL2 is stimulated to a much greater extent in AH compared to healthy controls. The monocyte cluster 2 expresses both IL1 beta and CCL2 at baseline and in response to LPS, but again, it's more robust in AH. Monocyte cluster 3, again, this shows you the distinct functions of these clusters. So here we have IL1 beta expression, again, higher in AH, but the chemokine is not expressed at all within this cluster. The CD16 positive cluster, also we see quite distinct chemokine expression between healthy control and AH. So each cluster then in healthy controls is quite different in their response to stimulation with LPS, and then the AH patients exhibit a very different expression of IL1 beta and CCL2. Taking this a bit further, we can isolate looking specifically at chemokines in the CXCL pathway, and I think you can appreciate then in AH on the right-hand side in red, that LPS induces a much greater expression of these chemokines than in the healthy controls. The other aspect of the monocyte cluster 1 is that in healthy controls, in red here, you can see that they express antiviral genes, such as the IFIT family of genes and OAS1. In contrast, in patients with alcoholic hepatitis, they've completely lost this antiviral response to LPS. Similarly, the anti-inflammatory peptide that's expressed in healthy controls is not changed in AH. So we see a loss of antiviral as well as anti-inflammatory function in the patients with AH. So to summarize this part, we see that monocyte clusters behave quite differently in AH and healthy controls. Monocyte cluster 1 in healthy controls is antiviral and anti-inflammatory, but becomes essentially a pro-inflammatory monocyte cluster 2 in patients with AH. And then monocyte cluster 2, we see that the expression pattern is similar, but the level of expression of these chemokines is greatly increased in the AH. Moving to the second question about the innate immune surveillance, we wanted to ask what's going on, because if LPS is coming in from the circulation as been shown by a number of investigators, it's likely that other microbial metabolites or microbes themselves are entering into the portal circulation. And we wanted to understand what the impact of low-dose LPS was on the ability of monocytes and macrophages to sense these other microbial entities. So, Adam Kim then focused on a family of receptors called the C-type lectin receptors. It's another family of pattern recognition receptors similar to the family of TLR. And each of these C-type lectin receptors here, they're called dectins, or also go by nomenclature of CLEC, nomenclature CLEC4A, 6A, 4D, et cetera. And each of these C-type lectin receptors recognizes a particular type of microbe. So here, DCIR recognizes parasites, while dectin 3 recognizes bacteria. MDL recognizes virus. What's really interesting about these C-type lectin receptors is that they're clustered on chromosome 12 in a very small region of the chromosome that's called the NK gene complex or the NKC cassette. So, Adam postulated that, because of their close proximity on the chromosome, they might be coordinately expressed in response to particular stimuli. We started with an interest in minkle because of a collaboration with Jiajia Li at the Cleveland Clinic, where our groups had been studying the impact of an endogenous dam, this is called SAP-130, that's released from injured hepatocytes. It's recognized by the C-type lectin receptor on immune cells and can then stimulate inflammatory cytokine expression. So, Really, we started with an interest in minkle. What Adam did was to ask whether LPS, again, the low-dose LPS upregulates minkle in PBMCs from patients with AH. The idea being that if alcohol increases gut permeability, the LPS comes in, it could then increase minkle expression. Indeed, healthy controls are on the left. When we treat with low-dose LPS, we see more minkle expression. Minkle is higher even at baseline and is then increased even further in patients with AH. Because the minkle expression was increased, we predict that minkle ligands, such as the exogenous ligand called TDB, or the endogenous ligand called SAP-130, would increase the expression of inflammatory cytokines mediated by minkle receptor. Sure enough, looking at both IL-6 and IL-1 beta, we can see this is a log scale. The IL-6 is greatly increased when LPS-treated PBMCs are then subsequently challenged with the minkle ligand increasing expression. This is higher in AH patients compared to healthy controls and similar results are seen with IL-1 beta. mRNA expression. It's not just minkle that's upregulated. We went on to look at the expression of the different C-type lectin receptors in both PBMCs and liver samples. Many of the C-type lectin receptors are increased in AH patients in particular in response to low-dose LPS. Here the DCIR, dectin-2, dectin-3, and dectin-1 is higher only at baseline, not in response to LPS. For the liver, we made use of samples coming from the Johns Hopkins R24 Alcohol Resource Center where we can get liver samples from patients undergoing transplant, as well as portions of liver from healthy donors. You can see again that by mRNA that many of the C-type lectin receptors are increased in the livers of patients with AH compared to healthy controls. Because these were increased, we also from the literature, there's two different studies that have been published showing that minkle and both dectin-1 contribute to ethanol-induced liver injury. My lab in collaboration with Jiajia Li made use of minkle knockout mice to show that minkle was required for ethanol-induced liver injury in mouse models. Similarly, Bert Schnabel's lab interested in activation of C-type lectin receptor dectin-1 by intestinal fungi found that dectin-1 deficient mice were also protected from ethanol-induced liver injury. Together, these studies suggest that there's increased expression of the C-type lectin receptors and that they're likely contributing to the progression of liver injury. This really emphasizes the point that we might call this immune surveillance. Basically, low-dose LPS drives the coordinate expression of multiple C-type lectin receptors so that as that intestinal barrier function is disrupted, not only LPS, but other bacterial products and bacteria microbial products enter into the portal circulation, and then can interact with different C-type lectin receptors on the surface to enhance the uncontrolled inflammation that we see in age patients. Moving on then to the second part of my talk, so what's the impact of this inflammation on hepatocytes? Here what I want to do is talk a little bit about death receptor signaling and how death receptor signaling contributes to the organelle stress and death of hepatocytes. There's multiple pathways for hepatocyte death. Most well-studied, of course, is apoptosis, which is considered a relatively non-inflammatory mechanism, cell death. We also have necrosis, where we have cell swelling and plasma membrane disruption. This is an uncontrolled pathway, while necroptosis, which is morphologically similar to necrosis, is a highly regulated process. There's also emerging data that pyroptosis and ferroptosis might also be associated with ALD. I'm going to focus today though on TNF-mediated death receptor signaling via the necroptotic and apoptotic pathways. If we have increased amounts of TNF-alpha in the environment of the hepatocyte, it interacts with TNF receptor 1 to stimulate this proximal signaling via multiple interacting proteins. In a healthy liver, the TNF-alpha is actually hepatoprotective because of the expression of NF-kappa B. However, in situations of disease, we can shift the TNF signal to either apoptosis or necroptosis. Apoptosis and necroptosis are mediated by a family of serine-threonine kinases called RIP-1 and RIP-3. When activated for necroptosis, RIP-1 and RIP-3 will phosphorylate ML-KL, which moves to the membrane to form pores, allowing for a release of hepatocellular content and then basically mimicking necrosis. We know that with ALD, we have hepatocellular death. What is it that's shifting the signaling of TNF-alpha from this hepatoprotective function to apoptosis or perhaps necroptosis? We know that one of the primary mechanisms for maintaining the equilibrium between protection and death is the ubiquitination of RIP-1. In particular, this is a K63-linked ubiquitination. Tatsu Miyata, a visiting scientist from Japan in the lab, explored whether ethanol was impacting K63 ubiquitination of RIP-1. Sure enough, he saw that there was a difference. He made use of this assay or a system that's called total ubiquitin binding elements. It's a way to pull down specifically proteins that are ubiquitinated at K63 sites. He pulled down the K63 from livers of mice that were either controls or fed ethanol and then blotted for RIP-1. Essentially, in controls, we see a lot of ubiquitinated RIP-1. But after ethanol feeding, the ubiquitination of RIP-1 is substantially decreased. So these data suggest that the shift from hepatoprotection to death is mediated by a change in ubiquitination of the RIP-1, at least in part. So if we're shifting from hepatoprotection to death, is it apoptosis that's taking place? I'm not gonna show you the data here because it was published a number of years ago. But essentially, we made use of both genetic and pharmacologic interventions. So we had BID knockout mice. So BID knockout mice are, BID is common to both the intrinsic and extrinsic pathway of apoptosis. And then we also had hepatocyte-specific caspase-8 knockout mice, and also made use of pan-caspase inhibitors. While each of these interventions specifically blocked hepatocyte apoptosis in response to ethanol, there was really no effect at all on early injury in nearing models of ALD. Suggesting that even though we can see apoptosis occurring, it's not causal in terms of that early stage of injury. If that was the case, then we wanted to explore whether necroptosis might be involved in ethanol-induced liver injury. We first looked at the expression of RIT3 in mouse liver. And these are control mice, and you can see we and others have published that in healthy livers, there's very little RIT3 expression. But what happens is that after exposure of mice to high concentrations of alcohol, this is 32% of calories in the diet for four days or for 25 days, you can start to see an increase in RIT3 expression. Wenqing Ding's group has demonstrated that this increase is due to an inhibition of proteasomal degradation of the RIT3, allowing it to accumulate in the hepatocyte. We can also see this in livers of patients with different stages of ALD. Here you can see the expression of the RIT3 specifically in the hepatocytes, and both by histological scoring and morphometric analysis, we see that there's a significant increase in the expression of RIT3 protein in ALD. So our hypothesis then was that RIT3 and MLKL, it's downstream effector molecule, will contribute to both ethanol and high fat diet-induced liver injury. We included the high fat diet-induced liver injury here because the progression of both ALD and NASH in animal models is usually quite similar. And so we thought we could further explore what was going on in high fat diet related to NAFLD and NASH within these models. The results were quite surprising. We found that there were distinct functions for RIT3 and MLKL in mirroring models of ALD or NAFLD-NASH. So on the left, we have ethanol feeding, and this is this 25-day protocol with 32% of calories. And you can see that in wild type mice, I'm just showing a few phenotypic markers, ALT, hepatic triglycerides, and MCP1. So we basically see injury in terms of hepatocyte injury, steatosis, and a little bit of inflammation. And in the RIT3 knockout mice, completely lacking that necroptotic pathway, you see that they're protected at each of these markers of injury. In contrast, the high fat diet, here we use the high fat fructose cholesterol model developed by Greg Gores. You can see again, we have injury in wild type animals, ALT, triglycerides, and MCP1, but there's no protection at all. So this is illustrating that there's differential roles of RIT3 in ethanol feeding compared to high fat diets. Even more surprising is that when we take the MLKL knockout mice, and here we use, this is the orange bars, we use two different models. This is our 32% 25-day model, as well as the GALBINJ model of acute on chronic injury. You can see that the MLKL knockout mice are not protected in any situation from ethanol induced liver injury. In contrast, the high fat diet, the FFC diet, are completely protected in the MLKL knockout mice. So this means that we have distinct and non-canonical functions for RIT3 and MLKL in these different mirroring models of ALD and NAVL and NASH. In terms of chronic ethanol, we're exploring a mechanism related to how RIT3 independent of MLKL can activate the inflammasome to generate IL-1 beta and inflammation. I don't have time today to show you the data on that particular project. And then recently published by Jacqueline Wu in my group, in the high fat diets, Jacqueline found that MLKL was actually involved in the regulation of autophagic flux. And that was what was protecting mice from high fat diet induced injury in the absence of MLKL. So these are the mouse model data. But even though they're quite interesting mechanistically, we thought we could possibly use these differences to develop biomarkers to distinguish ALD from NAVL and NASH. So in the first case, it's interesting because there's several reports in the literature of RIT3, RIT1, and MLKL in the circulation being associated with chronic inflammatory diseases. And so what we did was we took samples from our P50 alcohol center, as well as samples from our UL1 collaboration with Pao Santo Gru and Juan Cabia at IDBAPS, and asked whether we can see differences in the circulating concentrations of RIT1, RIT3, and MLKL in Healthy Controls, AH, or NASH patients. And you can see that there's a quite robust reduction in RIT1 in AH compared to NASH, which was not different from Healthy Controls. In contrast, RIT3 was greatly increased in AH, but not in NASH, and modestly increased MLKL in AH compared to NASH and Healthy Controls. If we use ROC curves, we can readily use both RIT3 and RIT1 to distinguish AH from Healthy Controls, and also AH from NASH. MLKL is a little bit better. It's not as good, but it trends in the correct direction. We were concerned though, because many of the AH patients are higher severity disease. So what we did next was we took only moderate AH patients compared to NASH patients, and still we could see a significant difference between moderate AH and NASH, so essentially equivalent measures of injury. RIT3 was still reduced or higher in AH, and at this point, really a very modest difference in MLKL. And again, the ROC curves in particular, the RIT3 can distinguish moderate AH from NASH. So then to summarize the two stories I've told you about going from mechanisms to translational insights, in the first case I told you about our single cell RNA-seq data that revealed the impact of AH on the function of monocytes, leading to loss of antiviral and anti-inflammatory functions, and an enhancement of pro-inflammatory cytokine and chemokine expression. In the second part, we talked about how RIT1, RIT3, and MLKL have non-canonical functions in murine models of ALD and alpha-NASH, and also how they can be a potential biomarker to distinguish moderate AH from NASH. So thank you very much for your attention, and I look forward to questions in the Q&A later on today. Thank you.

Dr. Laura Nagy

Zimmerman Hepatotoxicity State-of-the-Art Lecture

Alcoholic Liver Disease

Inflammation

Cell Death

Monocyte Function Alterations

C-type Lectin Receptors