I'd like to welcome you to our virtual SIG symposium at the ASLD entitled The Basic Mechanisms of Hepatotoxicity and Their Clinical Implications. We're fortunate to have this morning some leading investigators in the field who will be focusing on more the basic molecular mechanisms involved with hepatocellular injury and mechanisms of activation of the immune system. Our initial talk will be entitled The Hepatocyte Stress Response, Mitochondrial Dysfunction and Oxidative Stress, will be presented by Dr. Neal Kaplowitz of the Keck School of Medicine. Mitochondrial injury is recognized as a key component in drug-induced hepatotoxicity. The next talk, entitled Liver Injury Alteration and Bile Acid Homeostasis, will be presented by Dr. Bruno Stigler of the University of Zurich, and his talk will review the important role of bile acid homeostasis and its potential contribution to drug-induced liver injury. The next talk will be entitled Adaptive Immune Injury and Neoantigen Formation by Dr. Jack Utrecht at the University of Toronto, and this will overview the important contributions and recognition that activation of the immunologic systems is involved with liver injury. And the final talk, entitled Epigenetics and Genetic Pathways to Hepatotoxicity, will be my opportunity to review for you the different pathways that have been implicated in hepatotoxicity. I'd like to remind the audience that questions can be asked through the chat box, and hope that you enjoy this exciting symposium. Thank you. Good morning, everyone. The title of my talk is Hepatocyte Stress Response, Mitochondrial Dysfunction, and Oxidative Stress. And I have nothing to disclose. I'll start with an overview of the stress responses in liver injury, particularly hepatocellular stress. So intracellular stress is mediated mostly by three major factors or stressors. Reactive oxygen species, which is oxidative stress, ER stress, which is usually due to ER overload or malfolding of ER proteins, and mitochondrial impairment and the mitochondrial stress response. Each of these induces a variety of physiological adaptive responses. And when these adaptive responses are insufficient to dampen the intracellular stress, pathological responses occur. Reactive oxygen species are detoxified by the antioxidant response, which is driven by transcription factor. ER overload and malfolding induces the unfolded protein response in the ER, which, through a variety of transcription factors that derive from the ER, induce chaperone production, decreased client protein overload by reducing protein synthesis, translation arrest, and degrading the messenger RNAs. In addition, inducing ER biogenesis, making more ER, and autophagy, removing damaged ER. The mitochondrial impairment induces its own unique unfolded protein response mediated by transcription factors, which promote chaperone production that's mitochondrial specific, as well as the induction of import machinery that permits the uptake of newly synthesized proteins into the mitochondria, as well as transcriptional regulation of mitochondrial biogenesis. In addition to that, mitochondrial removal by mitophagy, removal of damaged mitochondria, and modulation of fission fusion. So these are dampening responses, but when overwhelmed or insufficient lead to a variety of pathological consequences, particularly cell death. But in general, stress kinases are activated by all of these intracellular pathways. And J and K is the particularly important stress kinase that mediates many of the pathological responses. So this slide sort of illustrates the complex interplay that exists. So that when, for example, in DILI, when reactive metabolites are generated in the parasite, all of these stress responses can be activated. So either mitochondrial stress or ER stress can promote reactive oxygen species and J and K activation. Conversely, reactive oxygen species and J and K can induce mitochondrial stress and ER stress. So there's a complex interplay, and all of these are generally involved to some degree. And aside from reactive drug metabolites, it's other things that can induce this type of interplay include saturated fatty acids and bile acids. The consequences of this are that this process may kill hepatocytes or induce nonlethal consequences. So intercellular stress can promote release of factors which promote inflammation, fibrosis, and metabolic dysfunction. So this is illustrated then by the J and K MAP kinase pathway and J and K activation through MAP3 and MAP2 kinases activated upstream by reactive oxygen species or ER stress. J and K activation or phospho-J and K can elicit physiological responses or pathological responses. This largely depends on the duration and intensity of phospho-J and K activation. J and K has many targets that are involved through transcription factors that are part of the AP1 family, including proliferation, inflammation, altered metabolism. Non-transcriptional factors are very important in the acute responses and include activation of pro-hepatotic BCL2 family members and inactivation of anti-hepatotic BCL2 families, as well as phosphorylation and degradation of the catalytic subunit of GCLC, which is rate-limiting for glutathione synthesis, impairing glutathione metabolism, and targeting an outer mitochondrial membrane protein that I'll talk about more, which promotes reactive oxygen species production. A number of years ago, we noticed that J and K activation was a common feature of toxic insults in the liver. And in particular, in the acetaminophen model, we identified a early activation of J and K, shown here by the immunohistochemistry, mainly in the central lobular area, which is the area where subsequent necrosis occurs. But before necrosis, J and K was activated, knockdown of J and K-102 or J and K in small molecule inhibitor prevented the liver injury without affecting acetaminophen metabolism or initial glutathione depletion. So let me go to the next slide here. So as we were conducting experiments on the role of J and K in liver injury using a variety of different toxic insults, we identified that sustained J and K activation was generally associated with translocation of J and K to the surface of the mitochondria. Activated J and K bound to the surface, unactivated J and K didn't, and activated J and K did not enter the mitochondria. So we were quite interested in what the significance of this phenomenon could be in relationship to the mechanisms of toxicity. So to make a long story short, a number of years ago, a group in the UK headed by Will Shire identified an outer membrane mitochondrial protein, which bound activated J and K. And they did this in fibroblast cell lines, but never really explored any further as to the significance of that finding. Subsequently, we identified their paper and identified SAB as being a key outer mitochondrial membrane protein in the liver and largely found in mitochondria rather than other cellular sites. SAB and activated J and K co-immunoprecipitated following different types of toxic stress, including acetaminophen. And SAB was phosphorylated by J and K. And this is a cartoon just to show that SAB has one transmembrane spanning domain, a short C-terminus, and a long N-terminus, and it's phosphorylated on the cytoplasmic side where it binds activated J and K. We then went on more recently to knock out in an inducible fashion, hepatocyte-specific knockout of SAB and J and K1 and 2. And in both cases, using the acetaminophen model as an example, the marked necrosis seen with acetaminophen was abrogated in the knockout of either SAB or J and K, indicating that both are required to mediate the interplay that is necessary for necrosis. We then, along the way, wanted to understand what is happening to the mitochondria when J and K, activated J and K, binds to it. So in this case, I'm showing you some seahorse data, which is a way of measuring oxygen consumption in the mitochondria to show you that unactivated J and K doesn't do anything to mitochondria. This is identical to what one sees with control normal mitochondria. But activated J and K impairs state 3 respiration, oxidative phosphorylation, and maximum respiratory capacity. But as you can see, using mitochondria from SAB knockout mice, there is no effect of J and K on the mitochondria. And a peptide which blocks the interaction of J and K with mitochondria was able to restore normal mitochondrial function in the presence of phospho J and K. And that peptide blocked reactive oxygen species production in isolated mitochondria. So directly exposing mitochondria to activated J and K led to a burst of reactive oxygen species production, which was quite sustained. And also, the blocking peptide inhibited that. So this cartoon just is to exemplify that either knockout of SAB or this peptide blocker prevented decreased oxygen consumption rate, effect of phospho J and K, and also prevented the increase in reactive oxygen species. So the big question, then, is what's going on inside the mitochondria that's leading to this mitochondrial impairment? So to make another long story short, this summarizes work that we published in 2016 in hepatology describing the intramitochondrial signal transduction pathway that's activated when J and K, activated J and K binds to SAB and phosphorylates SAB. It leads to the release from the intermembrane portion of SAB, the release of a protein tyrosine phosphatase, PTPN6, which otherwise is known as SHP1. This phosphatase dephosphorylates intermembrane phosphosarc, which is the active form, to an inactive form. And it requires the presence of DOC4, a docking protein, which brings the sarc and the phosphatase together on the surface of the mitochondria. Activated sarc is known to be required to maintain electron transport by phosphorylating various components of the electron transport chain, whereas dephosphorylation and inactivation of sarc impairs electron transport, promotes reactive oxygen species, which then feedback on the MAP kinase cascade, continuing to activate the cascade to activate J and K. This sustained J and K activation in the acetaminophen case promotes excessive reactive oxygen species production in the mitochondria, and ultimately, the mitochondria already being damaged to some extent by the acetaminophen, then undergo mitochondrial permeability, transition pore opening, and the steps that ultimately lead to necrosis. On the other hand, sustained J and K activation also promotes phosphorylation of other targets, for example, the BCL2 family to promote apoptosis. So although I've shown you data with acetaminophen, also a knockout or knockdown in this case of SAB also markedly protects against TNF-galactosamine-induced massive apoptosis, agonistic FAS monoclonal antibody-induced massive apoptosis, protects against ConA, concanavalin A-induced necrosis, which is T cell-mediated toxicity to the liver, markedly prevented. In all these models, it's known that J and K plays a pivotal role, and SAB, therefore, prevents sustained J and K activation. The furosemide model is another model where high-dose furosemide induces central ovular necrosis, and this is prevented by a knockout of acetaminophen of SAB, sorry. So we then wanted to increase the expression of SAB to see if that would play a role in whether the level of expression of SAB is an important factor in toxicity. So we have wild type SAB knockout and delivery of increasing doses of adenoviral SAB to the knockout to restore and increase SAB levels. These mice were then exposed to acetaminophen, and two hours later, there was a dose dependent related to the SAB level, dose dependent increase in J and K, sustained J and K activation, and just a progressively increasing severity of liver injury so that the normal mice had about 50% to 45% necrosis. They were protected by knockout, but the reintroduction of SAB to different levels increased the area of necrosis and the levels of ALT. And finally, I just want to point out that the previous slide dealt with male mice. And it's been, it's known that female mice are resistant to the toxicity of acetaminophen. And so we hypothesized that this might be due to a difference in SAB expression. And indeed, female mice expressed very low levels of SAB compared to male. And we elucidated a pathway from estrogen receptor alpha to P53, which is regulated by ER alpha, to microRNA 34A, which is regulated by P53. MicroRNA 34A represses SAB expression. And one can manipulate any one of these steps to reverse the phenomenon, to make females susceptible or make males resistant, depending on the expression of these. So in conclusion, sustained J and K activation and reactive oxygen species both cause and result from mitochondrial and ER stress. SAB activates an intramitochondrial signaling pathway, which impairs mitochondrial electron transport, increases reactive oxygen, and sustains J and K. The level of sustained J and K activation and severity of liver injury are directly related to the level of SAB expression. So the phospho-J and K mitochondrial SAB reactive oxygen species activation loop is a central node in acute hepatotoxicity. And this is to acknowledge all the contributors, and particularly to single out Sanda Nguyen and Tin-Ong Than, who really are responsible for most of this work. Thank you. Welcome to the presentation, liver injury alterations in bile acid homeostasis. My name is Bruno Stieger. I'm training a PhD in biochemistry and cell biology. I work at the Department of Clinical Pharmacology and Toxicology at University Hospital in Zurich. And my research interests are physiology and pathophysiology of bile formation. I would like to thank the organizers of this symposium for giving me a chance to present here. I will divide my presentation into four parts and start with enteropathic circulation of bile acids. You could also call it bile acid homeostasis. Bile acids are secreted by hepatocytes into bile ducts. The bile duct drains into the small intestine and bile acids are reabsorbed along the small intestine into the portal blood. They travel with the portal blood back to hepatocytes in the liver, where they are taken up again from the portal blood plasma and their journey starts again. Hence the name enteropathic circulation. The human bile salt pool is about three grams and it cycles four to 12 times per day between the liver and the intestine. 95% of the secreted bile acids are reabsorbed into the portal circulation and back to the liver. About half a gram of bile acids are lost by fecal excretion and this proportion is replenished by the novosynthesis in the liver. Urinary excretion can be neglected. In the liver, in hepatocytes, bile acids are taken up by NTCP in a sodium-dependent manner and by OATPs in a sodium-independent manner. They are secreted in an ATP-dependent manner by the bile salt export pump PSEP into canaliculi. In the small intestine, bile acids are taken up in a sodium-dependent manner by ASBT into enterocytes and released from enterocytes in a sodium-independent manner by Ost-alpha-beta. Some situations may lead to elevated intercellular bile acids, for example, in hepatocytes. For this, hepatocytes have salvage systems to export bile acids across the basal lateral membrane back into the portal plasma. MOP3E4 is an ATP-dependent transporter, respectively 2 and Ost-alpha-beta releases bile acids along its concentration gradient. The key regulator of bile acid homeostasis in enteropathic circulation is FXR or farasol-X receptor. This is a nuclear transcription factor. FXR in hepatocytes activates or upregulates the expression of PSEP and indirectly via SHIP downregulates NTCP. Upon binding of bile acid in the small intestine, FXR downregulates expression of ASBT and upregulates expression of Ost-alpha-beta. Hence, these transporters are regulated in such a way that intercellular bile acid concentrations stay normal or low as they may be toxic. In addition, in the intestine, FXR upregulates the expression of FGF19, which is secreted into the portal blood, transported to the liver, binding to hepatocytes and thereby represses bile acid biosynthesis. Now, what are important physicochemical properties of bile acids in bile acid homeostasis? Bile acids or bile salts can be seen as detergent. You see here torocholate in an upper view and in a lower view, and you can clearly see it's an amphipathic molecule. Now, amphipathic molecules are detergents, and this is illustrated in this experiment where we incubated in vitro isolated directly with canalical membrane vesicles. And once torocholate reached a concentration of about 8 millimolar, you can see phospholipids, predominantly phosphatidylcholine is released into the supernatant. Hence, torocholate acts here as a detergent. Intercellularly, bile acids can also impair due to their detergent action the function of mitochondria. This experiment was done with isolated red liver mitochondria, which were exposed to increasing deoxycholate, chino-deoxycholate, or lithocholate, and the activity of the respiratory chain was measured. I give you here the example of complex 1-2, where you can see that the activity lowers in a dose-dependent manner when bile acids are present. Hence, bile acids can be toxic to mitochondria. What are now negative impacts of bile acids in cells? At the cellular level, they can cause ER stress, including unfolded protein response. They can lead to mitochondrial impairment, which in turn leads to increased reactive oxygen species or lowered ATP levels in cells. They can start inflammatory responses in cells. If they persist at higher concentrations, they may drive a cell into apoptosis or even into cell death or necrosis. And of course, they lead to dysregulation of many proteins, for example, proteins involved in bile acid biosynthesis or bile acid transport. At the organ level, meaning in the liver, increased intercellular bile acids lead to cholestasis. Of course, there is an adaptive response, and that is activation of transcription factors, which I illustrated before by FXR, but there are additional transcription factors which are responsive to bile acids. Cholestasis can be discriminated between inherited forms of cholestasis, acquired cholestatic liver disease, and cholangiopathies. I'm not going to talk about cholangiopathies. It's well known currently that we have five different genes which can lead to progressive familial intrahepatic cholestasis, namely severe liver disease. And important genes involve, for example, BCEP and MDR3, but also tight junction protein to mention a few. There are also milder forms like benign recurrent intrahepatic cholestasis. These are patients which have recurrent episodes of cholestasis, and in between, they feel normal. Acquired cholestatic liver disease is, for example, intrahepatic cholestasis of pregnancy or drug and or herbal-induced cholestasis. Drug-induced liver injury is a diagnosis of exclusion, and it's typically diagnosed first by measuring serum parameters. And there is a classification into cholestatic mixed or hepatocellular based on the earliest identified pattern in a patient. And this R value or ratio between alanine aminotransferase and alkaline phosphatase, if it's low, it's classified as cholestatic liver disease. If it's high, it's classified as hepatocellular drug-induced liver injury, which can also or is hepatitis. And then there are intermediate forms. In my view, alkaline phosphatase can be seen as an indirect readout for elevated intercellular bile acids. What are potential mechanisms of drug-induced liver injury? Drugs or their metabolites can impair cellular functions. That means they can, for example, inhibit enzymes or transport proteins. Drugs can also be metabolized into reactive drug metabolites, which in turn bind to proteins and impair protein function. Such drug protein adducts may activate an immune response and thereby causing cellular damage. These are key or general mechanisms of drug-induced liver injury. Up to 50% of drug-induced liver injury patients may have a cholestatic component. This is a reason for publication in gastroenterology. A study in our department a while ago identified, not too surprisingly, antibacterials and anti-cancer drugs as major causes for drug-induced liver injury. And taken together, about 30% had a cholestatic component, meaning the type classified was either cholestatic or mixed. Hence, drug-induced cholestasis or acquired cholestasis, which may be caused by the detergent action of bile acids, is a relevant clinical entity. Now, if we look at canalical bile formation, it requires the concerted action of three transporters, MDR3 for phosphatidylcholine translocation, BSAP, and FIG1, which is an aminophospholipid translocator. These are three types in case of mutations leading to progressive liver disease. Now, it's evident that not all mutations completely block the action of the protein. We had one case in our clinic, which was a composed heterozygous of an E297G mutation and an ARF432 mutation. We expressed wild type and these two mutated BSAP in insect cells to the same level and characterized its kinetics. And you can see, even if you add up the Vmax value here, it does not reach the Vmax value of the wild type. On the other hand, KM values didn't differ significantly. As this was a BRIC2 patient, we conclude there may be a minimum level of BSAP, which does not lead to overt liver disease. And of course, instead of mutations, you can also inhibit the activity of BSAP and thereby get low levels of function. We then took several drugs known to induce drug-induced cholestasis and tested its inhibition on rat BSAP and human BSAP. And some of the studies were also done by other labs. And in all cases, we found a competitive inhibition by BSAP. So cyclosporine is a competitive inhibitor of BSAP and leads to drug-induced cholestasis in susceptible patients. Later studies expanded the number of compounds tested, as I see here, Morgan. He found about 10% of the drugs tested to be potent BSAP inhibitors and 70% to be moderate inhibitors. Needless to say, drugs can also inhibit MDR3. We developed for this a cell line expressing MDR3, BSAP, and NTCP. And we measured phospholipid secretion into the apical medium. And you can see that itoconazole, ketoconazole, and cyclosporine inhibit the activity of MDR3. Please keep in mind, mutations in MDR3 also lead to cholestatic liver disease. Hence, some drugs like cyclosporine may not only inhibit BSAP, as I showed before, but also in parallel MDR3 and hence aggravate the toxic action of bile acids in hepatocytes. Now, I mentioned before the key regulator of BSAP expression. This is FXR. And it has been demonstrated that some drugs can work as antagonists of FXR. And this study investigated several non-steroidal and inflammatory drugs. And this is a list of the drugs which actually were antagonists of FXR. And you can see ibuprofen, for example, has a quite low IC50 with one micromolar. The two drugs highlighted in red were actually functionally confirmed to act as FXR antagonists in, sorry, antagonists in hep G2 cells. Another study investigated more drugs. And you can see they found tele-positive and tele-negative drugs. Drugs I highlighted in red are also known to be BSAP inhibitor. So, Bosentan, for example, not only inhibits competitively BSAP, but it also antagonizes FXR, which in turn will downregulate expression of BSAP. Hence, drugs may interfere indirectly with bile acid homeostasis. Drugs tend to be metabolized predominantly by the gut microbiome. We have primary bile acids. They are synthesized in hepatocytes, kinodeoxycholate and cholate, which then are conjugated to glycine or taurine. Secondary bile acids are produced from hydrolyzed bile acid conjugates by the gut flora. And they include deoxycholate, lithocholate, urtodeoxycholate. Hydrolysis of conjugated bile acids is predominantly achieved by lactobacilli, enterococci, bifidobacteria, hydroxylation, leading to the secondary bile acids actually by Clostridium. It's important to keep in mind that different bile acids have a different affinity or potency of activation FXR. Kinodeoxycholic acid is by far the best one, followed by deoxycholate and lithocholate, which are much better than cholate. Hence, a disbalance of the ratio of these two primary and secondary bile acids may lead to alterations in activation of FXR and hence again affect bile acid homeostasis. Needless to say, the gut microbiome is affected by drugs like antibiotics, beta blockers, metformin, proton pump inhibitors, tricyclic antidepressants. Bacterial strains highlighted in red tend to be involved in hydrolysis of conjugated bile acids in green of hydroxylation. Arrows meaning up, they increase, the relative proportion down, they decrease. Hence, this leads to a complex response in the composition of secondary bile acids. I would like to come to the last point. That's the barrier function between sinusoidal blood and bile. And these are tight junctions. If you look at concentration gradients from blood to bile, you see that bile acids are concentrated up to 500 fold. And it's the tight junction here, highlighted by arrow, which maintains this concentration gradient between canalicular bile and the portal blood plasma. In black, you have lanthanum chloride, which was infused via the portal vein. Now we studied a while ago the effect of ethanol estradiol treatment on type junction function in rats, and we found that horseradish peroxidase can, after five days of treatment, cross the paracellularly into bile. We also counted the number of type junction strands, and you can see they were significantly reduced from chrontrol to ethanol estradiol. This effect was dependent on the molecular weight, so the type junctions are not completely open. We did not test for the barrier function with respect to bile acids, but it is conceivable that under such conditions, bile acid may flow back into the space of this and hence elevate bile acids in the systemic circulation. There's very few literature testing drugs on the integrity of type junctions. I would say this is an area open to investigation. So to come to a conclusion, bile acid homeostasis includes the function of transporters and transcription factors. Bile acids act intercellularly as detergents and as such may interfere with many different organelles. Bile acid homeostasis is an interplay between the liver and the gut, and in the gut, of course, it's the microbiome which is very important. Drugs may interfere both with homeostasis in the liver or homeostasis in the gut. An impairment of type junction barrier function by exogenous factors at the moment is poorly understood. With this, I would like to thank you for your attention. Welcome to this virtual presentation. The title of my presentation is Adaptive Immune Injury and Neoantigen Formation. It's not exactly the title that I would have picked, but it's close enough. My name is Jack Utrecht. I have a PhD in organic chemistry from Cornell University and an MD from Ohio State University. I did my internal medicine residency at KU Med Center followed by a clinical pharmacology fellowship at Vanderbilt University. I'm currently a professor and Canada research chair in adverse drug reactions at the University of Toronto, and my research focus for a long period of time has been on the mechanisms of idiosyncratic drug reactions, including idiosyncratic drug-induced liver injury. I've done consulting for many different companies. The most recent are listed here, but none are related to the content of my presentation. As you all know, idiosyncratic drug-induced liver injury is a significant cause of liver failure. There are many mechanistic hypotheses, such as inhibition of BSEP and inhibition of mitochondrial electron transport chain, but the idiosyncratic nature of idiosyncratic drug-induced liver injury makes mechanistic studies difficult, so it's hard to test hypotheses and the mechanistic details are unclear. Even though the details are unclear, there's very strong evidence for an adaptive immune response, and here I will concentrate on hepatocellular idiosyncratic TILI. Most important are the HLA associations and other genes involved in immune regulation, such as PTPN22 and interleukin-10, that are associated with an increased risk of idiosyncratic TILI. The histology is characterized by an inflammatory infiltrate consisting mostly of lymphocytes and sometimes eosinophils. An inflammatory infiltrate could be a result rather than a cause of the injury, but since CD8 T-cells are prominent, CD8 T-cells are involved in killing cells that are infected by viruses or are cancerous, they are not involved in the repair of injury. So I think that's very good evidence that an adaptive immune response is responsible for the injury. Sometimes there's a positive lymphocyte transformation test or the presence of anti-drug antibodies, but there are many reasons why these could be falsely negative. The general characteristics are easiest to explain on the basis of an immune response, and these include a delay in onset with rapid onset on rechallenge. However, I want to emphasize that an immune mechanism does not exclude other mechanisms, such as BCEP inhibition, that may contribute to the induction of an immune response. There's also evidence for the involvement of reactive metabolites, but this is mostly circumstantial evidence. The fact that most drug metabolism occurs in the liver is likely one reason that the liver is a frequent site of idiosyncratic drug reactions. Reactive metabolites can produce drug-modified proteins that the immune system recognizes as foreign, and reactive metabolites can also cause cell damage or cell stress, leading to the release of danger-associated molecular pattern molecules, otherwise known as DAMPs, which can activate antigen-presenting cells. However, it's difficult to prove that reactive metabolites are responsible for idiosyncratic dilly, and many drugs form more than one reactive metabolite, so there's often an argument about which reactive metabolite might be responsible. A few drugs associated with idiosyncratic dilly, such as Zymalogatrin, don't appear to form a reactive metabolite. And although the formation of a reactive metabolite is considered a risk in drug development, the relationship between the amount of reactive metabolite formed and the risk that a drug will cause an idiosyncratic drug reaction is far from perfect. It's likely that what the reactive metabolite binds to and whether it causes the release of DAMPs are important factors. Another important concept is immune tolerance. It appears that the dominant immune response to drugs, especially in the liver, is immune tolerance. Drugs that cause idiosyncratic liver failure always cause a higher incidence of mild liver injury that often resolves despite continued treatment with the drug. This is referred to as adaptation. But if the injury is immune-mediated, this resolution must involve immune tolerance. Checkpoint inhibitors that impair immune tolerance increase the risk of idiosyncratic dilly caused by co-administered drugs. And we've developed an animal model in which blocking immune checkpoints unmasks the ability of drugs to cause liver injury. So in this figure, I show in the gray bars animals treated with amediacone. And you see in these gray bars an increase in ALT, but it resolves despite continued treatment. If we add anti-CTLA-4, anti-CTLA-4 is another immune checkpoint. Then we see greater liver injury that does not resolve despite continued treatment. The ALT isn't all that high, but it's sustained, and there is decreased liver function, sort of like a Hyde's rule, with an increase in bilirubin. And the histology looks just like idiosyncratic dilly in humans and is characterized by piecemeal necrosis. We know that this injury is mediated by CD8 T cells because if we deplete CD8 T cells, it's protective. So what makes idiosyncratic dilly idiosyncratic? Although the risk of idiosyncratic dilly for several drugs is associated with a specific HLA haplotype, even if a patient with that haplotype is treated with the associated drug, they're unlikely to develop idiosyncratic dilly. There must be other risk factors. And although there are other known risk factors, their contribution appears to be small. A drug-modified peptide must not only bind to HLA on the energy-presenting cell, it must also bind to a T cell receptor. The T cell receptor repertoire is generated by random gene recombination, so it's different even in identical twins. The number of potential T cell receptors is almost limitless because of this random gene recombination, but the number of T cells is not unlimited, and therefore each patient has a limited repertoire of T cell receptors. Although it's more difficult to study the T cell receptor dependence of idiosyncratic drug reactions such as idiosyncratic dilly, it's been demonstrated that the serious skin rash caused by caromazepine is dependent on a specific T cell receptor. So this leads to a postulated sequence of events leading to idiosyncratic dilly and other idiosyncratic drug reactions. You have a drug, it forms a reactometabolite, and that reactometabolite produces drug-modified proteins that can be seen as foreign, and it can also cause the release of DAMPs, which activate energy-presenting cells. So this is an innate immune response. In most cases, there won't be an HLA or T cell receptor that recognizes the drug-modified peptides, and so you get no adaptive immune response and no idiosyncratic drug reaction. Sometimes there's moderate binding to HLA and to a T cell receptor, and so you get mild idiosyncratic drug reactions, which often resolve with immune tolerance. And it's only if the patient has HLA and T cell receptors with high affinity that you get a strong adaptive immune response and a serious idiosyncratic drug reaction. So now I want to talk about involvement of the innate immune system. The adaptive immune response requires an innate immune response to activate energy-presenting cells. Most patients, and even most animals, form the reactometabolites involved in the induction of an innate immune response. Therefore, most humans, and even animals, are likely to have an innate immune response to drugs that cause idiosyncratic drug-induced liver injury. That means that the early events in the induction of idiosyncratic DILI can be studied in humans who do not develop liver injury as well as in normal animals. So do we have evidence that this is true? Well, much of our work has been done with clozapine, which the major idiosyncratic reaction is agranulocytosis, but it can also cause idiosyncratic DILI. Most patients treated with clozapine have a transient innate immune response with an increase in serum IL-6, IL-1 beta, and a paradoxical neutrophilia. The neutrophil count actually goes up rather than down. And clozapine causes an innate immune response in rats, very similar, with a paradoxical neutrophilia and an increase in the chemokine CXCL1. Another drug that we've studied is nevirapine. Nevirapine is associated with a relatively high incidence of idiosyncratic DILI and very serious skin rashes, including toxic epidermal microlysis. We know that the skin rash is caused by a reactive benzylic sulfate formed in the skin because unlike most cases in female brown norovirates and only in female brown norovirates, nevirapine causes a skin rash. And if we apply a sulfatransferase inhibitor topically, it prevents the covalent binding in the skin and it prevents the skin rash where it's formed. In contrast, the covalent binding in the liver is due to a quinone methide, and that's presumably the reactive metabolite responsible for idiosyncratic DILI. Nevirapine also causes mild liver injury in our impaired immune tolerance model, but not in normal mice. When we treat normal mice with nevirapine, after one day we see a marked increase in inflammatory lysic C monocytes in the spleen. So they're stained brown and you see there's much more brown staining in the treated spleen than in the spleen from a normal untreated animal. We also see in three days a marked decrease in the size of the inguinal lymph nodes, a rather dramatic finding. We also see in lymphocytes a decrease in the expression of the immune checkpoint, PD-1, and another checkpoint, TIM-3. Now later on in the impaired immune tolerance model, the immune checkpoint molecules actually increase. So this is a transient decrease associated with activation markers. Going back to clozapine, we know that clozapine is oxidized by myeloperoxidase in neutrophils. It causes agranulocytosis, and also in THP1 cells, which is a human monocyte cell line. And this leads to activation of inflammasomes with the production of IL-1 beta. So here you see release of IL-1 beta from the THP1 cells, which increases with increased concentration, and these concentrations are within the therapeutic range. In contrast, olanzapine—oh, and I should mention that a caspase inhibitor inhibits this, so we know that the IL-1 beta is coming from inflammasomes—olanzapine, which has a very similar structure and also forms a similar reactive metabolite, doesn't activate inflammasomes, and doesn't lead to release of IL-1 beta. But now nevirapine can't be oxidized by myeloperoxidase. It requires cytochrome P450. And we see that there's no production of IL-1 beta by the THP1 cells. This is just the positive control. However, if we take nevirapine and incubate it with hepatocyte spheroids and take the supernatant from that, the supernatant does activate inflammasomes with the production of IL-1 beta, again, prevented by the caspase inhibitor, not prevented by the sulfotransferase inhibitor that prevents the skin rash, but it is prevented by a P450 inhibitor, aminobenzotriazole. Again, this is just the positive control. So, again, this suggests that one of the early steps in the innate immune response is activation of inflammasomes. But, again, these animals do not develop liver injury. This is a clinically silent innate immune response. So, again, back to the sequence of events, we think that, in general, again, the drug forms a reactant metabolite, which can produce drug-modified proteins and also danger-associated molecular pattern molecules. This leads to innate immune response. And this we can study even in patients that don't develop liver injury and in animals. And it's only if the innate immune response leads to an adaptive immune response because the patient happens to have both an HLA and T cell receptor that recognizes the drug-modified peptides that you get a serious idiosyncratic drug reaction such as liver injury. Having said that, the immune system is really complicated, people are complicated, and they're probably exceptions to this general principle. For example, some idiosyncratic GILI does not appear to involve a reactant metabolite. And the details, this is just an overall sketch, the details of the innate immune response are unknown at this time. But since we can study these details in animals and humans, hopefully we can work out the details. Excuse me. So, the takeaways are most idiosyncratic GILI is mediated by the adaptive immune response. Again, I think there's really conclusive evidence for this, maybe not in all cases, but certainly in some. It's idiosyncratic because it requires a specific combination of an HLA molecule on the antigen-presenting cell and a specific T cell receptor on T cells. But the adaptive immune response requires an innate immune response to activate antigen-presenting cells, and this innate immune response is not idiosyncratic. Therefore, we can readily study it in humans, in which a drug doesn't cause idiosyncratic GILI, and also in animals. This provides a method to study the initial events that in rare patients result in liver failure. And so, the ability of a drug candidate to produce an innate immune response may predict GILI risk in drug candidates, and if this is true, that would be a major advance in drug development. So, I want to thank the people who actually do the work. Amir Matushi and Alistair Mack worked on the impaired immune tolerance model. Very good scientists. Samantha did the recent work on Clozapine. Teshin and Allison did the recent work on Nivirapine. Professor Kato, who spent a year in my lab, he's from Japan, did the in vitro and flammasome work. And of course, I have to thank the source of my grants, the Canadian Institutes of Health Research. And I thank you for your attention and I hope I've given you some new things to think about. So I'd like to thank the ASLD for giving me this opportunity to present the title of my talk called Epigenetics and Genetic Pathways to Hepatotoxicity. I have no disclosures related to the presentation or with any commercial interest. So in the next 15 minutes or so, I'd like to do is to review for you some of the basic mechanisms of hepatotoxicity as an opportunity to introduce to you potential gene groups that may contribute to drug-induced liver injury, referred to as DILI from now on. Within this, identify specific genes associated with drugs and herbal and dietary supplement-induced injury. Briefly review then the role of epigenetics in both drug and toxin-induced liver injury. And finally, summarize some of the future role of genetics and epigenetics and hepatotoxicity to put this in some sort of context. Now, when we consider then the nature of hepatotoxicity, there's a class of agents that we refer to as direct toxins. And these are compounds that predictably cause liver injury and typically are not species-dependent. And during drug development, these drugs are clearly excluded from further development for clinical use. The classic hepatotoxin that's been extensively studied is acetaminophen. And just to review for you then, the process for acetaminophen causing liver injury, it's an orally ingested agent. There are negotiopathic metabolism in the process. Some of the acetaminophen is conjugated with glucuronides and sulfates as shown here that lead to its excretion. Yet that portion which is not metabolized will then undergo further metabolism by the cytochrome P450-2E1, which then leads to development of reactive metabolites leading to glutathione use and subsequent liver injury. In the process, there's activation of cell death pathways and activation of the immune system as well. The agents of drug that we're more concerned about clinically are those that are associated with referred to as idiosyncratic drug-induced liver injury or IDILI. And these are liver injuries that are rare and unpredictable and they may occur either due to pharmaceutical agents or to herbal supplements. And these typically cannot be reproduced in animal models. Now there is, in looking at the classes of agents that have been associated with IDILI, there's data suggests that both hepatic metabolism and drug dose can actually predict the likelihood. And this would be mediated by the same ADME genes as referred to earlier on. And the current feeling is that these drugs may elicit a mild liver injury that may resolve with time, generating intracellular stress, which then may activate the DAMPs or damage-associated molecular pattern proteins, which then triggers an activation of the immune system. And then there's a process by which the immune system, once activated, there is an unknown process by which either injury can continue, which then can lead to toxicity that we recognize clinically and even can occur in the absence of the agent, or that the body adapts to this, leads to down-regulation of liver injury, and therefore there is then adaptation. And specific HLA genes have been associated with this. Now in reviewing the ADME gene, we're talking about a family then of different genes that perform different functions. These include the absorption, and it's important to note that the intestinal uptake and metabolism is another important factor in considering exposure to a specific agent. There's distribution that also occurs depending on transporters, allowing compounds to get into the liver or other sites. And then once the drug or herbal supplement enters into the liver, there's also metabolism. And there's been a major focus on the initial metabolism in terms of liver injury, as that this is when there are potential generation of toxic metabolites in select individuals and generating intracellular stress. And initially it was thought that the idyllia, those rare variants, was likely due to variations in the ADME gene that placed people at risk if they had rare genetic variants in their metabolic pathway. These process of initial metabolism is referred to as phase one, and typically we consider this phase of metabolism as when reactive metabolites are generated. This is then necessary to allow for then a conjugation with either water soluble molecules, which then constitutes phase two. These conjugated compounds, which are now more hydrophilic, then can be excreted out through transporters located either the sinusoidal or the canalicular domain of the hepatocyte, and this is referred to as phase three. It's important to note that in order for these various genes to orchestrate the uptake metabolism, conjugation, and then excretion, that there are also master transcription factors listed here that may also be important and might then be potentially other sites that may alter metabolic response to different agents. Now, if we look specifically at the phase one enzymes, those that lead to modification of functional groups, the predominant family that's involved with this are the cytochrome P450s, and it's thought that they represent anywhere from 75% to 80% of drug metabolism as shown in this slide. There are 57 genes, families noted in this group. However, only families, members of one through three, are typically responsible for the exogenous compounds, that is the drugs or herbal and dietary supplement. And of those, there's only a select number as listed in this slide that are actually thought to contribute, and as you can see in the figure to the right. It's important to note that one of these cytochrome P450-2E1 is also alcohol inducible, and therefore, those people that drink may place patients at risk for those drugs which may be metabolized by that. And it's well recognized that functional variants in these CYP gene families differ significantly between race and ethnic groups, and therefore, this may explain some of the variation that could be seen in different populations. The phase two group of reactions mediate those involved with conjugation and excretion. They typically will take endogenous water-soluble molecules such as the glucuronides and the sulfatransferases, sulfates, and mediate by the UGTs and the sulfatransferases to promote these reactions. Another important family of the glutathionase transferases involved in detoxification, and it's well recognized that variants in these phase two can lead to severe toxicity, and an example of that is the TPMT deficiency leading to bone marrow failure in those that are exposed to those agents. One in particular I'd like to discuss is the arylamine N-acetyltransferase, the NAT2, as it has been implicated in INH and obesity. So the NAT2 gene is known to be highly polymorphic. There are significant variants in different populations. It has been roughly determined that in terms of their enzymatic activities, that one can group these in three groups of either rapid, normal, and slow acetylator phenotype, but it's important to note that the substrates may also determine which of these functional activities are present. Evolution has suggested that these slow acetylators actually are associated with agricultural society as opposed to the nomadic lifestyle, so depending on which populations, there may be differences in function. And although the role of NAT2 polymorphisms in INH hepatotoxicity is controversial, there are some studies in specific populations which have identified that the NAT2 variants can actually predict the incidence of anti-TB hepatotoxicity, and as shown on the right in a study from Japan, when patients were dosed for the INH dependent on the phenotypes of their acetylator status, it was noted that the incidence of pharmacogenomic-based therapy was lower than those who just received routine sequencing. The other major advance in terms of the eye deal has been the discovery of the close HLA association, and this is one of the earliest studies, a GWAS study looking at the drug glucloxacillin, and a very strong, as evidenced here in this Manhattan plot, a very strong association located on chromosome 6 at the site where the HLA family resides. This is typically used in Europe with the development of cholestatic hepatitis, typically in females and often more elderly. And in this nice study done, the features of the glucloxacillin was it would occur within 1 to 45 days of onset. This study was strictly limited to patients of European ancestry, and important to note that this HLA B5701 had extremely high odds ratio of 80, and a replication set was also confirmed to have similar events. But it's important to note that despite this strong association of the HLA B5701, if one looks at it at the flip side and estimates how many patients that have this HLA will, in fact, develop glucloxacillin dilly, it's been estimated that only 1 out of 500 to 1 out of the thousands are likely to develop it, suggesting that the HLA is necessary but not sufficient by itself to cause this. Now, another agent that has received further GWAS studies is that associated with amoxicillin clavulanic acid, as this is a common agent that has associated with drug-induced liver injury in multiple places. And as you can see here in this study, limited to, again, Caucasians, that there was, again, an HLA association, as shown in the panel A. But when looking more closely in panel B and more with detailed analysis, there appeared to be actually multiple HLA regions that were associated with dilly, and these were independent of each other, an important point because the HLA region is highly linked, and therefore, one cannot assume that there isn't linkage equilibrium, that is, a carrier one HLA may actually carry the other HLA. And in the study, in looking at these different HLAs, it was determined that there were associations in both class I and class II, that these were independent of each other, and that they afforded about a two- to three-fold increased risk in those that had it. Now, one interesting question, given the fact that AC dilly can present with both an hepatocellular and a more cholestatic, was to determine whether or not HLAs might be associated with specific path of injury. And a nice corollary study included in this initial, in this manuscript was that, when looking at Spanish patients, a more homogeneous population, that the HLA B1801 was more frequently associated with a patocellular injury and less likely cholestatic, suggesting that the HLA may, in fact, detect patterns of injury. This has been an active area investigation, and as seen here, there are multiple HLAs that have been associated with specific drugs, and importantly, there are some HLAs that are associated with different agents, so that the same HLA may be causing, associated with injury due to different agents. Now, with this identification of HLA, we then want to look and see if there are other genes that may be potentially playing a role. And in this recent study, including approximately 2,000 patients, there was an all DILI examination looking for genetic variants. And as you can see in panel A, there's a strong HLA association, as predicted, given the role of AC in contributing to DILI, but there is also another peak that loaded on chromosome number one. And as we see then on this slide, it then asks, particularly in those patients with AC DILI risk, what the role was of either the HLA and the HLA with the PTPN22, which I'll describe a little bit later, as to what risk that may be associated. So, on this slide, as you can see, in those that carry neither the HLA, which is the first and second in the carriage group, or the PTPN22 mutation, there was no increased risk for mutation. And yet, when we then looked for the patients who carried both HLAs that are independent of each other, we noticed there's approximately a sevenfold risk. If we look at the PTPN22 variant, that, by itself, there seems to be no increased risk. And the PTPN22 is added in addition to those with the two HLA risks. We see that the risk of the odds ratio increases to 13-fold. So, in essence, this PTPN22 is functioning similar to if you would think of it as a rheostat to modify the activity of those with delay risk. And PTPN22 is a gene. It's a protein phosphatase involved in cell signaling. It's important to note that this SNP actually is associated with an amino acid substitution, and, therefore, it suggests a functional consequence, as shown on the slide on the right, that PTPN22 has been associated with both T and TOL-like receptor signaling, and, therefore, it makes sense for this to be a candidate. And it's previously been implicated as a risk factor for both autoimmune diseases, including RA or disease. Now, finally, in the last couple of minutes, I just want to review briefly epigenetics and pathotoxicity. And, as shown on the right there, epigenetics refers to changes in the structure, methylation allowing for gene transcription to occur, and also microRNA and regulation of expression. Little is known in DILI as the liver samples are needed to really analyze and cite two genomic changes, but it has been noted in mouse studies that environmental hepatotoxins are capable of introducing epigenetic changes that are present in the offspring. And, finally, drugs that are associated with DRESS are associated with activation of latent viral infections, which suggests that there may, in fact, be epigenetic changes occurring to allow expression. An example of one such agent is aflatoxin as well, and here we see that the aflatoxin, a mycotoxin, is known when infected with those with chronic hepatitis B and increased incidence of HCC. And we now know that the aflatoxin undergoes metabolism to generate a free radical, which then leads to specific CT inversions leading to genomic... So, in summary, if you look at the impact of genetics and epigenetics on DILI in the future, we can consider that ADME screening is currently being used for drug design, both in looking at inhibition of B-cep activity as well as to make sure that those variant cytochrome P450s are not metabolized in drug of agents. GWAS studies have identified molecular mechanisms of drug-induced injury, therefore allowing greater insight into mechanism and potential for future biomarkers to be developed. It's known that some of the HLA testing is required through drug administration, although not for liver injury, and it may have a role if a patient has DILI on multiple medication and one can find a specific HLA. And we've seen now with combining variants of both HLA and, for example, PTPN22, there may be future benefits. And individual epigenetic changes due to environment, diet, toxin exposure are likely to influence the risk of developing. So the key takeaway points are that variants in the ADME genes rarely appear in human GWAS hepatotoxicity studies, multiple HLAs associated with drug DILI, and some shared with different agents. The risk is not sufficient to explain DILI development than any individual. Future combination of genetic variants may inform those individuals at risk for DILI, and epigenetics are likely to play an important role, but there's little current evidence for that. Thank you. Hello, I'm Adrian Rubin, and I'm the current chair of the Hepatotoxicity Special Interest Group. I'd like to say a few words in closing after our wonderful symposium today. First of all, I'd like to give warm thanks to Vic Navarro and Andy Stultz for organizing the symposium. As you know, Vic was formerly the chair of this SIG and is currently the newly appointed chair of medicine at Einstein Medical Center in Philadelphia, and Andy Stultz is on the faculty of the University of Southern California. I'd like to thank all of you for attending, whether you had to get up in the early hours of the morning or stay up into the early hours of the morning to be with us. I hope that you find it both enjoyable and informative. I want to give thanks to our wonderful faculty for presenting. Neil Kaplowitz gave his usual excellent discussion of mitochondrial dysfunction and oxidative stress in hepatotoxicity. Bruno Steiger stepped in at the last moment from the University of Zurich to give our talk on liver injury alterations in bile acid homeostasis. Jack Utrecht of the University of Toronto gave a wonderful discussion of adaptive immunity and neoantigen formation. And Andy Stultz ended up by discussing some epigenetic and genetic pathways as they relate to hepatotoxicity. I'd like to remind you too that we hold a monthly hepatotoxicity seminar on Zoom that Wen-Zhen Ding from Kansas Medical Center and I have organized since August. These occur on the second Thursday of each month and will now continue from December through June of next year. If you don't already have access, please contact our liaison at ASLD, John Lingefeld, and he will arrange to give you access. And finally, I hope that you will all stay safe in this current pandemic crisis. As they say in the UK these days, hands, face and space. Wash your hands, cover your face and keep your social distancing. I look forward to seeing you all perhaps in person next year at the liver meeting in 2021. Thank you all again for attending. And if you have any other questions, please voice them at the symposium. Thank you very much.

Hepatotoxicity

ASLD

SIG symposium

Molecular mechanisms

Hepatocellular injury

Immune system activation

Mitochondrial dysfunction

Oxidative stress

Bile acid homeostasis

Adaptive immune injury

Epigenetics