I would like to start by thanking the organizers for giving me the opportunity of giving this Hans Popper lecture. What I'm going to be talking to you today is about how we are modeling development and cancer using liver and biliary organoids. As I said, it's an honor for me to give that lecture in the name of Hans Popper, who, apart from being the father of hepatology, he contributed so significantly to the field with many accomplishments and one in particular that is really relevant to my talk is that he demonstrated that liver does not age. As I said, this is relevant to my talk because it's at the core of what we are doing in my lab. In my lab, we're trying to understand how tissues maintain and how they repair themselves and how this is deregulated in disease. We use the liver and other organs, but in particularly for today's talk, the liver as a model organ of high repair capacity. And we are aiming to study maintenance and repair of tissues in particularly in the context of the human. But as you can understand, that is for ethical reasons, many experiments cannot be performed. And we then turn back to the mouse models to ask some of these questions as well. We believe that if we better understand how tissues maintain and repair themselves, we're going to be able to build models that recapitulate organogenesis in a dish. And these models in return, in particularly in the case of the human, are going to be informing us how human tissues repair and how human tissues undergo disease. I don't need to explain and give any hints extra on how liver cellular composition is, how liver architecture is, just to remind you that in today's talk, I'm going to be talking about the parasites and ductile cells. I'm not going to be talking about the other very important supporting cells in the tissue, biliary ducts or endothelial or cover cells among many others. What I find most fascinating of the liver is that endometriosis, the liver, has very low cellular turnover, but yet it has a huge regeneration capacity upon damage. Which goes back to this concept that the liver has an age because it can regenerate itself, it always stays in the same original state. That has been exploited in my field, in the organoid field, to develop different types of liver organoids. I'm just giving you here a hint, a summary of the different organoid models that have been developed, starting with pluripotent stem cells. We have the pioneer work from Takanori Takebe, whereby he differentiated iPS cells into hepatocyte-like cells, put them in three dimensions, combined with other cell types and generated these beautiful liver bads that recapitulate liver bad development. This is not what my lab works on. My lab works on either taking cells directly from the tissue, either from the embryonic tissue or from the adult tissue, and generating either hepatoblast organoids or biliary duct organoids. And we've shown how to generate these ones in the past, and I'm going to be talking about that in coming slides. I also want to mention that the hepatocyte-derived organoids have been generated by the Clevers and Nurser labs, but I'm not going to focus on this today in my talk. As I said, I want to share with you three stories that come from my lab, how we developed and generated the hepatoblast-derived organoids that can model liver development, how we've exploited cholangiocyte organoids that we developed some years ago now to understand molecular and cellular principles that mediate drug-driven liver regeneration, and how we've used this technology to develop human primary liver cancer models for personalized medicine. I'm going to start with the first topic, which is how we've used hepatoblasts to generate hepatoblast organoids and model liver development in addition. As you know, liver development starts in the mouse around E8.5 with hepatic specification, liver diverticulum is formed around E9, and liver bat generation starts to be appearing around E10.5. If we, three days later at E13.5, this hepatoblasts that have been generating this liver, embryonic liver, are going to start to decide whether they are going to become ductile cells or whether they are going to become hepatocytes, and this being the work of Frederic Lemegre, Valerio Evans, and others. We noticed, though, when I started my lab six or seven years ago now, that the hepatoblast was considered a bipotent population capable of giving rise to both hepatocytes and ductile cells, but actually that we did not know whether that was at the single cell level or it was at the population level. In other words, we wanted to ask the question whether a single hepatoblast was indeed bipotent, or there were different committed hepatoblasts, hepatoblasts committed to the hepatic fate, hepatoblasts committed to the ductile fate, that meant that actually the population of a hepatoblast was bipotent as a population level, but not at the single cell level. To ask that question, we took advantage of performing lineage tracing experiments using LGR5 CRE as a driver. In that case, when we combine a CRE with a rosa TD tomato and the rosa reporter and we induce the maxifen, then we can label specifically the LGR5 population and also the descendants of this population, and particularly in that case, we would label them with TD tomato. We found that LGR5 was expressed at the embryonic liver E9.5, and then that would allow us to ask the question whether LGR5 was a bona fide hepatoblast pool. We need so that it is a bona fide hepatoblast pool because all the descendants of these LGR5 positive E9.5 cells are indeed marked by AFP, classical hepatoblast marker, and they also, 50 percent of them are proliferating. Then we ask the question, these are bona fide hepatoblasts, is actually this bona fide hepatoblast population truly bipotent? To do that, we took advantage of the confetti reporter, which is a rosa reporter, whereby when you induce the maxifen, you have the chance of getting one of the four colors, either green or yellow or red or blue. And all the descendants are going to have, the descendants of this labeled cell are going to have that particular color. So that allow us to perform the following experiment whereby we took LGR5-CRE combined with rosa reporter, mice, we took this compound mouse, injected the maxifen at E9.5 in development, and then we took the liver's postnatally and we asked the question, do we have cholangiocytes represented here with a triangle and hepatocytes represented here with a square? Do we have cholangiocytes and hepatocytes together marked on the same color, suggesting that they come from one single bipotent LGR5 progenitor, or instead we don't find clones marked by the same color, but we find clones marked always by different colors, suggesting that both ductal cells and hepatocytes are coming from two independent LGR5 progenitors. Nicole Prior in my lab, she took this project on and then in the end, she said, indeed, she performed that experiment and then she found clones whereby the ductal epithelium marked here by osteopontin in blue is also marked in red and nearby hepatocytes are also marked in red, indicating that this is a single clone, which we are going to see in this movie. This is a single clone that arises from a single bipotent LGR5 progenitor. We performed a very in-depth quantification and we just focused on those clones that are in the portal track and we looked at the cellular composition, we see that actually 50% of them, they are only made of hepatocytes, a single color, and 50% of them are made of hepatocytes and cholangiocytes that share the same color, suggesting that 50% of the clones arise from bipotent LGR5 progenitors. Then Nicole asked the question whether, OK, if these cells are bipotent in vivo, can we keep the self-renewing capacity and bipotency ex vivo in culture if we put them in an organoid medium? So she chose to isolate these LGR5 positive cells and put them in cholangiocyte media or in hepatocyte media and starting from one single hepatoblast, she could generate beautiful hepatocyte-like organoids when she put them in hepatocyte media and beautiful ductal organoids when she put them in cholangiocyte media. Here I'm showing you that cholangiocyte organoids are bona fide cholangiocyte-like cells expressing keratin-19, single-layer epithelium, while hepatocyte organoids grown in hepatocyte media, they have this stratified epithelium. We now also know the cells are quite polarized. They have bile canaliculi and they indeed secrete albumin, they express albumin, they secrete albumin, but they retain their embryonic nature because they are still expressing alpha-ketoprotein and secreting it into the medium. So as a summary of that part, what I can discuss with you is that indeed hepatoblast organoids, they can recapitulate hepatoblast self-renewing capacity in vitro and as well as their bipotency in culture. Now, I want to discuss with you how we've been exploiting cholangiocyte organoids to understand molecular and cellular principles of duct-mediated liver regeneration. As you all know, the liver has a huge capacity to regenerate if we perform hepatectomy, what we know that at the cellular level, hepatocytes are going to make hepatocytes, ductal cells are going to be making ductal cells. We also know that instead of hepatectomy, the liver is damaged by different damaging agents like hepatitis virus, alcohol, fat comes from the blood, are going to be the hepatocytes and the ductal cells, the cells are going to be damaged, mainly. And while many labs, Willem Brink and others, have also shown that there is self-replication if the damage is mild, my lab and others, in particular Sfort's lab, have also shown that in the presence of chronic liver injury or in the presence of a senescent liver, ductal cells go from being unipotent, only able to make themselves, to become bipotent and being able to make two cell types, hepatocytes and ductal cells. We don't know whether this is through a progenitor state or a transdifferentiation, but for us, what is most important is the ability of changing cell fate. They kind of reprogram from being capable of only making themselves to make two different, very different cell types. And that is true for both ductal cells and hepatocytes. This is very relevant for human liver disease because many human liver diseases occur with ductal reaction. This happens in chronic liver disease, hepatic fibrosis, hepatic cirrhosis, even in liver cancer, we find all this huge amount of ductal cells compared to a normal healthy human liver. Whether they come from a proliferation of ductal cells, whether there is transdifferentiation from hepatocytes, whether there is a stage two progenitor state, this is still under debate. Probably all options are possible and are happening at the same time. But for me, what is more interesting is that we can exploit adult liver organoids derived from the ductal compartment to ask the question, what is the mechanism by which these ductal cells arise and which is the mechanism of ductal reaction? We showed in the past already, more than nine years ago almost, that a given ductal cell, if we put it in culture in our organoid culture conditions that contains FGF, Wnt signaling, and extracellular matrix of madrigel, if we put these cells in culture, these cells are going to start proliferating, they're going to self-renew, they're going to make these beautiful three-dimensional structures that we can propagate in culture even for a year. And interestingly, if we start from one single cell, we can still generate both fates, hepatocytes and ductal cells, indicating that in culture, these cells behave like in vivo, these ductal cells that are capable of making and becoming bipotent and generating both hepatocytes and ductal cells. So exploiting these ductal organoids, we can now ask the question, how cells regulate the transition between different states? In other words, if we put it in the words of a ductal cell, how does a ductal cell regulate, how does a ductal cell mechanistically know how to transit to a progenitor state that is able to make different cell types, hepatocytes and ductal cells? Or what is the mechanism by which actually a given ductal cell is able to make different fates, in that case, hepatic and ductal fate? To address this question, Luigi Aloia in my lab, he took ductal cells, we all know that they are arrested in G0-G1, and he asked the question, how long does it take for any given ductal cell to start proliferating? So we could set up the stage for the following experiment. We observed that actually, if we look at this video, that is around 40 hours on average, in the video is going to be 30 hours, but that the cell goes from a G0-G1 arrested state to a S phase, and then it starts generating organoids and continues dividing. So to ask the question, what's happening in these 40 hours while the cell is sitting in the plate, Luigi took ductal cells at time zero, isolated right after isolation, or 12 hours, 24 hours and 48 hours, so before the cell divided, because the cell is dividing at 40 hours, and after the cell divided and at the organoid stage. He collected RNA and performed RNA-seq and asked the question, what is the transcriptional changes that occurred during this transition from a differentiated ductal cell to an organoid stage? And what he found, the line, the red line indicates before proliferation and after proliferation. And much to our surprise, what we found is that there is a huge dynamic changes that are occurring as the cell transits from a duct state to an organoid state. And we focused that actually, there were more than 40% known epigenetic modifiers that were changing dynamically on this transition. We wonder whether that was an artifact of the tissue culture. So we turned our eyes to the mouse, and we checked whether in vivo, there is also transcriptional changes. and to much of us surprise, indeed in the transition from a ductal cell to a progenitor state or proliferating ductal cell. He took, we saw before proliferation, there were already massive changes and there was a 70% overlap between the transcriptional changes occurring in a given ductal cell to a progenitor-like state or in a governed ductal cell to organoid state, 70% overlap. This was perfectly validating our ductal organoid culture system to understand this mechanistically what is the mechanism by which a ductal cell generates organoids or in other, in vivo, what is the mechanism by which a ductal cell starts proliferative progenitor state. Since there were more than 40% epigenetic regulators, known epigenetic regulators changing, we choose some of them, we perform an sRNA screen whereby we took ductal cells, we dumped more than 30 different sRNAs for 30 different genes. And we asked the question, which of them is relevant for organoid formation? We got different hits here, I'm just showing you the TAT family members in particularly Z1 inhibits fully organoid formation indicating that they are very important for this transition from a duct to organoid state. What is Z1? Z1 is a methylcytosine deoxygenase that catalyzes the conversion of methylcytosine, which is a repressive mark on the DNA for a hydroxymethylcytosine, which is an activating mark in the DNA. For any given gene to be expressed, methylcytosines have to be removed from the promoter region either entirely removed or changed for a hydroxymethylcytosine which enables transcription by favoring the arrival of all the transcriptional machinery to the promoter region. We found Z1 very important and interesting because Z1 is essential for induced pluripotent reprogramming, induced pluripotency. And at the end of the day, what we are experiencing is that these ductile cells are kind of reprogramming themselves from being unipotent to be potent. So we thought that Z1 could be an important player on this process. We looked what genes Z1 was binding and we found to our surprise that Z1 binds at the promoter region of important genes like Axin2, LGO5, and even very well-known genes of hepatic regeneration like some components of the hepa pathway like that. But it doesn't bind, of course, on those genes that are not expressed in our organoids like alpha-fetoprotein. So that gave us confidence that actually Z1 was modulating the expression of these genes and we confirmed actually all these genes are expressed in our culture system only after the cell has entered into this transition to a progenitor state, but not before. We asked the question, is that relevant during regeneration, right? Because of course this could still be an artifact of the in vitro culture. To ask that question, we took advantage again and we took a wild-type mouse and used the DTC diet model, very well-known in the field. We took the cells before the cells proliferate around day two, day three, and after the cells start to proliferate at day five, and we performed 5-hydroxymethylome profiling at the single base resolution. And what we observed to our surprise is that in vivo, during this transition between differentiated and progenitor, at day three, before the cells have started to divide, at day three, we have more than 3,000 genes are already acquiring the NOVO-hydroxymethylation sites in the promoter region, suggesting that these 3,000 genes are being targeted by TET1 to put this hydroxymethylation mark and are gonna have some importance in the regeneration. Interestingly, this is a transient mechanism because more than a third of them are already, they already removed this hydroxymethylation mark. And if we look at which are these genes, which means that there's gonna be induced in the transcription, we look which are these genes, we find very important pathways involved in liver regeneration like mTOR or RBB like UGF receptor, or as I mentioned before, the YAPIPO pathway, confirming what we had seen in the organoids that actually bona fide components of the pathway like TAD1 tasks are downstream of TET1. TET1 has to go to the promoter region, put the hydroxymethylation mark, and then they start activating transcription. Of course, all these genes are changing. There is a change on transcriptional profiling, but it's actually TET1 relevant for the function of for the regeneration itself, for the transition from this duct to a pathocyte. For that, we turned to the mouse model and that was developed or the damage model that was developed by the Forbes lab published in Raven et al in Nature in 2017, whereby they take a ductile CRE driver, in our cases, using the keratin 19 as they use in the paper we use from 1CRE. We combine it with a ROSA reporter, in that case, it was a ZS Green reporter, whereby all the ductile cells upon tamoxifen, all the ductile cells are gonna be green. And then we combine this reporter mouse with a TET1 flux, which allow us upon tamoxifen to delete TET1 as well. So in fact, upon tamoxifen induction, TET1 is gonna be depleted and ROSA Green is gonna be expressed. Then we induce senescent to these mice and we also induce liver damage. And we ask the question, in the wild type scenario, we know from the Forbes results that ductile cells are gonna proliferate and are gonna give us hepatocyte clones because they acquire this potency and they go from being only able to make ductile cells to make ductile cells and hepatocytes. And we ask the question, what happens when we deplete TET1? Are we gonna be able to generate hepatocyte clones? Are we gonna be able to generate ductile cells? What we observed is that actually, as we expected from Stuart Forbes data, when we have TET1 proficient, we could perfectly reproduce Stuart's data. Ductile cells generate beautiful hepatocyte clones from up to even 120 cells in very readily and present in all over the parenchyma. However, when we have a TET1 mutant, we hardly find hepatocyte clones. And when we find one label hepatocyte coming from a ductile cell is always clones made from either one or maximum of two cells. So that suggested that actually TET1 is involved in liver regeneration. In the last minutes of my talk, I want to show you how we've exploited liver cancer organoids to model liver cancer and use them for personalized medicine. That makes me bring to remind you that already in 2015, we showed that we can take a biopsy from a human healthy donor, someone that donates the liver for transplantation. We could take a biopsy of these donors, put them in culture, in our improved human liver culture organoid media, which is very similar to the mouse, but contains two additional factors, especially TGF-B, the inhibition and cross-calling. And now we could also establish human liver organoids that would self-propagate and we could keep them in months and months in culture for more than six months in culture. We've done that for now, I would say more than 40 donors. And we also observed that these organoids started from ductile cells as well, have come positive population and they still retain this biopotency. They are capable of generating both hepatocytes and ductile cells. And these hepatocytes are hepatocyte-like cells that retain some of the function of the human liver. They, for instance, have cytochrome activity, they produce bile acid, but they are not bona fide hepatocytes. I would say that now hepatocyte organoids, in that sense, have much better functions of hepatocytes. Yet this system, again, in the human setting, allows us to say that we are, we are capable of activating this biopotent state in vitro, also for human cells. We took advantage of that system to try to see whether we could generate tumor organoids. In other words, we could generate organoids derived from a patient that had liver cancer. We were lucky that our collaborators at the Erasmus Rotterdam Center and in Addenbrookes Hospital in Cambridge, they gave us samples from three different subtypes of liver cancer, hepatocellular carcinoma, mixed subtype and cholangiocarcinoma. And we put them in culture, that was the work of Laura Brottier in my lab. And what we found is that indeed we could expand the three subtypes. I want to point here that very well differentiated HCCs, we could not expand if they have less than 5% proliferation and proliferating cells in the tissue, we didn't manage. But when this amount of proliferation is higher than 5%, we could readily establish organoids from the three different subtypes. Interestingly, we could expand them long-term in culture. We have had them now, I would say, for more than two years in culture. And when we look at the histology, while a healthy donor is a single-layered epithelium of ductile cells, type of bipotent progenitor that can generate both the pathocytes and ductile cells. The tumors, they grow very differently. They grow like a solid structure. They have these pseudo rosettes. When they are bona fide HCCs, they are completely compacted in the case of some of the mixed subtypes. In the cholangiocarcinomas, we always have carcinoma in situ, cells invading into the lumen. We asked the question, how do our organoids resemble? How well do they resemble the transcriptional changes that occur in the tissue? First thing, we wanted to know how well they resemble to the original patient. And actually, we were very surprised to see that actually each single, by performing RNA sequencing at the genome-wide level, each organoid line has maximum correlation with its own tissue it was derived from. So actually, they recapitulate in vitro the patient-specific tumor profile, marking it with the biggest square, which is the maximum correlation. In other words, cholangiocarcinoma organoid line from patient one correlates the best with the tissue of the patient one, and it correlates less with other cholangiocarcinoma patients' tissues. And even, of course, it doesn't correlate at all with hepatocellular carcinoma tissues. And the same is true for other of the HCC lines and all the other lines. Interestingly enough, all these cells, all these organoids, all these organoid lines are grown under same media conditions, and yet they retain the characteristics of the subtype of the tissue, of the tumor. For instance, if we look at AFP, AFP is expressed by the hepatocellular carcinoma, it's not a cholangiocarcinoma marker, and we only find it in hepatocellular carcinoma cultures and in some mixed subtypes, and always, if it is present in the tissue, it's present in the organoids, and if it is not present in the tissue, it's not present in the organoids, like in the cholangios. The same is true for cholangiocarcinoma markers, like keratin-19, it's expressed in the tissues, expressed in the organoids, but it's not expressed in the HCC type of tumors, suggesting that they match very well the patient's profile in vitro. And we asked the question how well they match the genetic mutations or genetic alterations that are present in the original tumor, and we found that more than 80% of the cancer-related somatic variants that were already present in the patient tissue were retained in culture. If we look even at particular cancer mutations, we look, for instance, KRAS, was only positive in the tissue of one of the mixed subtypes and in the tissue in one of the cholangios, and it was, similarly, this mutation we found it in the organoids of the corresponding patient. Similarly would be, for instance, for beta-catenin. We had beta-catenin in a specific of HCCs. One of our HCCs had this mutation in the tissue, we also found it in the organoids. But overall, 80%, more than 80% of the somatic variants were present, were retained in culture. Then we asked the question whether they retain also the histology, the histological characteristics of the tumor. In order to do that, we decided to take this patient organoids, these cancer organoids, and transplant them into the kidney capsule and also subcutaneously to immunodeficient mice and see how they grow as a tumor in vivo. This is example of a cholangiocarcinoma organoid line. We also did it for hepatocellular carcinoma organoid. And as you can see, the pathologist, when we showed that to the pathologist, she could not distinguish whether that was the tumor that had grown in the mouse or the tumor that had grown in the patient. As you can see, they even mount this huge stromal reaction. I show you now the patient that was, from which these organoids that were transplanted in the mouse derived from, and you can see there is almost indistinguishable one from the other. Then once we had this beautiful model that now mimics in vitro many aspects, not all, but many aspects of the tumor of the patient in vivo, we asked whether we can use them for personalized medicine. We collaborated with Matthew Garnett and Haley Francis, and we ran a panel of 30 known anticancer drugs. And we found that some lines are sensitive and resistant to some of the drugs. We took one drug, an ERK inhibitor, that had not been used in liver cancer treatment. And we asked the question whether actually it would have an effect in vivo. So we generated tumors in the mouse again, and while the tumor injected with the vehicle grew, as we expected, the tumor injected with the compound actually undergo full apoptosis, and the cells were all turned positive. So with that, I get to my summary. I want to recapitulate what we've been talking today. We said that actually a single hepatoblast can be potent in vivo, and also in vitro, we can retain this property in vitro in hepatoblast organoids. We've also talked that cholangiocyte organoids have allowed us to identify a mechanism by which, in that particular case, that one regulates the transition from being a differentiated unipotent cell to a vipotent cell. This happens in vitro, but I also show you it also happens in vivo. And I've also shown you that we can use human liver cancer organoids to recapitulate many, not all, but many aspects of human liver cancer. And with that, I would like to thank all the people that has worked in these projects, starting with the embryonic hepatoblast organoid project. We have Nicole Pryor, which she's here, Nicole Pryor and Chris Hindley in the lab. They were instrumental to make this project further. Germain, Mikel and Luigi, they were instrumental in the TET1 project, identifying this mechanistic and molecular mechanism for the transition to a vipotent state. And Laura Brottier, Gianmarco and Robert contributed strongly, all of them, into this liver cancer project. With that, I would like to thank you for your attention. Of course, thank you, my funding. Thank you for your attention. I'll be happy to answer your questions. Thank you.

Hans Popper lecture

development modeling

cancer organoids

liver repair

biopotency

liver regeneration

personalized medicine