Scientist Stories: Sonja Schrepfer, The Future of Stem Cells and Hypoimmune Cell Therapies

I remember back in 2001, a bit of history, President George Bush, George W. Bush, banned federal funding of research on newly created embryonic stem cell lines. This was the first point at which I really became aware of stem cell therapy. And in 2004, that sort of probably followed on from that, the California Stem Cell Agency was created after California voters approved Proposition 71, which was to fund in-state stem cell research directed toward cures. Just a couple years after that, Shinya Yamanaka, who by the way, was, had a lab at Gladstone Institutes affiliated with UCSF, described how to reprogram somatic cells to become induced pluripotent stem cells, for which he later won the Nobel Prize along with John Gurdon. And the deal with pluripotent stem cells is that they can propagate indefinitely and generate cell types for any tissue in the body, which makes them incredibly promising as as cell replacements for defective cells that, that cause devastating diseases. But we have immune systems for a reason, which is to reject foreign biological material. So how do you get these stem cell therapies into the body to cure these, these terrible diseases without inducing an immune reaction, which is why we've invited Dr. Sonia Shrepfer, who's head of Hypoimmune Platform at Sano Biotech, also an adjunct professor— correct me if I'm wrong, these are just the current titles I have for you— adjunct professor at UCSF in the Department of Surgery, in which she also directs the Transplant and Stem Cell Immunobiology Lab, and founder of Transplant and Stem Cell Immunobiology Lab at Stanford. Sonia received her MD and PhD degrees from the Universities of Würzburg and Hamburg, where she was awarded the Heisenberg Professorship for her work to understand the immune barrier in stem cell biology for regenerative medicine. Sonia and her team subsequently decreased the immunogenic potential of stem cells to generate hypoimmunogenic tissues and organs, work that was funded by CIRM and NIH, among other sources. We're really happy to have you, Dr. Shrepfer. Please tell us all about hypoimmune stem cell therapies.

Thank you so much, Kasper. I'm really excited to be here today. So I start with my disclosure slide, but it was already correctly said everything. So I am at Sana Biotechnology at the moment and adjunct professor at UCSF. I wanted to start really in the early days to show you the whole journey in that field because I'm now at Sana, but it's not that we just started the last 2 years to work in that field. We started much earlier in the early 2000s to do that work. And my background is in solid organ transplantation. I was trained in Germany, in Munich, by a cardiac surgeon to do cardiac transplantations. And I got frustrated because the surgery itself seems to be really easy, but the hard part starts afterwards when the organ gets rejected and how to control the immune system of the patient. So I decided back then to leave the clinical field and to really focus on the immune part and understanding the mechanisms of rejection. And I moved from Germany to Stanford in 2008. 2005, and that was an exciting time because, as Kasper said, the human embryonic stem cells were available back then. And I remember that we had stickers on the centrifuges which were not supported by federal funding, and that allowed us to work with the human embryonic stem cells. The work I show you today in the slide deck was all done with iPSCs, so induced pluripotent stem cells, that we wouldn't have to deal with those ethical issues with the embryonic stem cells. That's to the background. This slide here is quite old. That was really the early work around 2006, 2007, when we used back then pluripotent stem cells. We started with the human embryonic and then later we repeated it with iPSCs. We wanted to understand how those cells get rejected. In these days, the field was discussing that pluripotent stem cells, because they are so early in development that they wouldn't get recognized by the immune system after transplantation, and therefore, rejection wouldn't be a problem. So in the early days, we looked into that, and we tried to understand if they do get rejected and which immune cells are responsible for that. And this slide shows you an experiment where we just transplanted pluripotent stem cells into different mouse strains with different immune cell populations to understand that immune hurdle and to identify it better. And what And what we learned back then was that if you transplant into a mouse with a complete immune system, so that mouse has functional T cells and B cells, that's what we call adaptive immune cells, so they recognize the foreign antigen, and if it's foreign, they would start rejecting immediately. And this mouse also has innate cells, the NK cell population, and NK cells are usually looking for virus loaded cells, for example, to clear them. So it's a different defense mechanism of the human body. And what we learned, when the mouse has all immune cells intact, those mice are capable of rejecting pluripotent stem cells. So it, it showed us clearly that pluripotent stem cells, although they are early in the development, they do get recognized by immune cells and they do get rejected. And you see here within 5 days that the transplanted cells were rejected here in an assay where we basically monitored survival of the transplanted cells. When we then transplanted them into a different mouse strain with no T cells but functional B and NK cells, we saw that only 2 out of 9 mice rejected, and that showed us for the first time that the T cell population seems to be the major contributor in this allograft rejection of pluripotent stem cells. Then we use different models. I just show one more here, the PAF1 knockout mice, where you see the mice who have T cells and B cells. They don't have functional NK cells. And you see those mice are capable of rejecting the graft. And the reason why I put this slide in is to show you that the first step for us was just the understanding what is happening. And we clearly identified the T cell population, so the adaptive immune system, as responsible for the major rejection of the graft. So in our next step, what we did is we started thinking about what antigens are responsible to trigger a T cell response, an adaptive immune response, and we tried with genetic engineering to knock these molecules out. Or back then, you see that it was published 2011, but we started working on it around 2008. Back then, we didn't have CRISPR-Cas9 technology available, so what we did is In order to control the T cells, we said we want to wipe off the fingerprint of our pluripotent stem cells, that we can then see if the T cells, if they don't see the fingerprint, they shouldn't react and we would overcome the T cell rejection barrier. So what we did is we knocked the fingerprint, the HLA class I molecule down, and we used a silence RNA, Andrew Fire at Stanford back then had the lab, and he had developed the silence RNA. So we tried to learn from the technology, and we combined it with an intrabody that we put into the endoplasmic reticulum of the cells. And whenever the MHC molecule would get assembled, the intrabody should pull it down, and it shouldn't go to the cell surface and show something foreign to the immune system of the recipient. And this technology kind of worked to knock down the fingerprint, but it never worked to completely have a knockout. So what you see in that graph is in red, this is the fingerprint, the HLA-1 molecule, and you see it goes down and then it stays down quite for a while. And that showed us for the very first time that gene editing is possible in the cells, but we would need better technology or modify the molecules to really achieve knockouts of the fingerprint and then overcome the T cell barrier. So the idea that then came out of those experiments was we want to have one healthy iPSC donor. So these are pluripotent stem cells we generate from one person out of, for example, skin cells or blood cells. And this one iPSC line we want to make hypoimmune, that we evade the recognition of the immune system when we transplant this line into any patient. So the goal is really to have a very well-characterized master bank that we can transplant, that it's easy to manufacture. We have good quality controls because it's only one cell line you have to, to the QC. And in the patient, the goal is immunosuppression-free, so no immune rejection and to provide the technology really to anyone, anytime, and anywhere. So with this goal in mind and the first experiments we did, understanding that that fingerprint and the T cells are the main barrier in that transplantation, we looked into nature and we wanted to understand where in nature do we have a situation that there is an allograft. That means basically engraft with surface proteins from another person, but the graft is surviving. And that is in fetomaternally tolerance during pregnancy. Because during pregnancy, half of the proteins of the fetus are from the father. So the mother's immune system doesn't know those proteins and sees them as foreign and would, in theory, start to reject the fetus. But pregnancy works. We know that because of the evolution, and we try to understand those molecules. We know that the mother's immune system is seeing the proteins or the antigens, but it's not rejecting the fetus. And we spent many, many years in understanding those single molecules, and the goal was to identify the fewest number of those molecules we would need to create a hypoimmune cell product that we then could transplant into any patient without immunorejection. And what you see here on the right side is really the basic principle we learned, and that is that on an allogeneic cell, so our iPSC donor product, and that cell does have a fingerprint. So all of our cells have it. That's how our immune system distinguishes between foreign and own, and it would immediately reject and form fingerprints. So that fingerprint, those are the HLA molecules. And we do know that those HLA molecules, they trigger T cell responses, and that leads to killing of those cells. But what we also know is at the same time, this fingerprint is inhibiting the other immune cell population, the innate cells, the NK cells and the macrophages. And by connecting with the NK cells, inhibiting the NK cells and there is no killing. So it's a killing molecule for the adaptive immunity but protective for the innate. So by now wiping off that fingerprint, that's what you see here on the right side, you can easily overcome that hurdle of T cell recognition and B cell recognition. But now we have the problem that the NK cells are not inhibited anymore, and then NK cells would kill those allergenic cells with the missing self immediately. The reason for that is in nature when we have, for example, for example, a virus infection, the virus is knocking down the HLA molecules. And innate cells, like NK cells, they're permanently looking for cells with a missing self, where the fingerprint is not there, and they're triggered to kill them. That's how, for example, we can clear virus-loaded cells. So the goal was we needed to understand which molecules we need to now control the NK cell hurdle. And if we would overcome this, we most likely would have a product that could be transplanted without immunosuppression. And over the years, we had studied many different molecules to inhibit these NK cells. What I show you here are videos. And in these videos, the green cells you see here, the dots, those are cells where we had removed the HLA, so they're HLA negative. And then we overexpressed molecules that were discussed or thought of inhibiting innate cell killing. And that is HLA-E on the left or HLA-G. We also overexpressed PD-L1 or CD47 here on the right. And then we put NK cells, these are the little gray round structures here, into that cell culture. And we wanted to see which of those of these molecules is able to inhibit the NK cell killing that is triggered by the missing self. And I start the video. The overall video is like 6 hours. And what you see during that timeframe is that when we overexpress the molecule called CD47, that the stimulated NK cells that are sensing the missing self, and they see there is no fingerprint, they are still not attacking. So CD47 seems to block that killing of the NK cells. So therefore, the data I show you today are cells, iPSC cells, where we wiped off the fingerprint, the HLA class I and the class II, with a genetic knockout. And here we use the CRISPR/Cas9 technology for the data I show you. And then we overexpressed CD47 to inhibit the NK cell killing at the macrophages. And the idea is that these iPSCs, you could now differentiate into any cell type you are interested in. I have, for examples here, the T cells, beta islet cells, cardiomyocytes, or clear progenitors, and to provide off-the-shelf therapies and transplant those cells now without immunosuppression into any patient who needs them at any time and anywhere.. And what I thought today is that I put together the slides where we already then use differentiated cells. We have lots of data characterizing the iPSCs in mice and humanized mice, but I think the differentiated cell data are very interesting, and I thought I would like to share them with you today. So the first data set you see here is we differentiated those cells into cardiomyocytes. And we are here on this slide in a mouse model. So these are mouse iPSCs, and we differentiated them into mouse cardiomyocytes. And when we do that, usually the mouse reacts with an immune rejection because the cardiomyocyte is from a different mouse. And you see that here in red. This is a T-cell assay, and the red bar here is quite high. The blue is the background of that assay. That is basically the transplantation of the cell in the way into the VSM mouse. So that's like an identical twin. There is no immune rejection. And you see cardiomyocytes get recognized by the immune system because of the fingerprint, high T-cell activation. The next assay here is an antibody binding assay. So that shows you the B cells or then plasma cells produce a lot of antibodies. So the red is much higher than the blue. On the bottom, these are the so-called hypoimmune cardiac cells. So here we had used mouse iPSCs with an HLA or MHC class I, class II knockout and CD47 overexpression. And what you see now is that the red graph is in, in the background of the blue. So it is basically, it gives you the similar response like when you would get your own cells into your body.. So it shows you that those cardiac cells are able to evade the immune recognition. We then wanted, of course, to show that in a disease model. So we used the mice, we created a myocardial infarction, and then we injected either the cardiomyocytes, which we didn't do any gene edits, or the hypoimmune cardiomyocytes or cardiac cells. That's what you see here on the right. And what you see in bioluminescence imaging is we can visualize those cells. And on day 0, you see what we transplanted. And over time, the signal when the cells get rejected because they're unmodified disappears. And the blue graphs here, they show you each single mouse. You basically, you see that each mouse rejects cardiomyocytes if they're not hypoimmune. But on the right side, you see when we transplant hypoimmune cardiomyocytes, the immune system doesn't see the cells. It's immune evasion. And the cardiomyocytes, they are surviving. The follow-up I have here in the picture is 28 days. We then looked if we improved the cardiac function in those mice. And we performed pressure-volume— pressure-volume loops. What you see on the x-axis is the volume the heart is pumping.. And on the y-axis, you see the pressure. The black loops show you a healthy heart. So you see the heart is pumping out volume. And we are measuring the volume that is basically in the heart after the blood got ejected. And you see on the y-axis, the pressure the heart can, can build up. When we are now performing myocardial infarction, you see that the volume increases in the heart because the the pump function is decreased and you see the pressure goes down. But when we transplanted hypoimmune cardiac cells, that's what you see here in red, you see that the loop goes back to the left. More volume gets pumped out of the heart and you see the pressure goes up. The heart is able to have an increased pressure. It's really hard to study engraftment or regeneration in mice. But what we wanted to show here with the data is really a proof of concept that if the hypoimmune cardiac cells survive in the heart, that they can contribute to regeneration. Then another model we did, and all those data were done at UCSF, is that we put endothelial cells into mice. And what happened, what was quite intriguing when we saw that, is we transplanted those endothelial cells endothelial cells into our mice. And first, we were interested, do they survive? And they did survive. But what we then saw is that they're not only surviving, they are functional as endothelial cells. And that's what you see here in the histopathology after 2 weeks. You see in blue the nuclei of the cells. In red, you see VE-cadherin, that is a marker for endothelial cells. And in green, luciferase, that's the enzyme we overexpressed in our transplant cells. So whatever is green, you know, those are the cells we put into the mice. And what they started around 2 weeks, but much more obvious around 3 and 4 weeks, is that they started to form those rings, those premature vessel structures. And then later we even found erythrocytes in those vessel structures, what you see as pink rings.. So what we then thought is if they're not only surviving, but they seem to function as endothelial cells, can we use that in a disease model to cure mice with a certain disease? And we picked the hindlimb ischemia model. So what you do in that model is that you remove the blood flow, and here in this case, on the left leg of the mouse to the foot. So these are Doppler perfusion images, and you see on the left leg here, there's no blood flow. Blood flow would look like here on the right, that you have this red tube or line to the foot. And what we did is we removed the whole artery, so we knew there is no blood flow to the foot. Then we transplanted wild-type, so unmodified endothelial cells. And as expected, they did get rejected because they were allogeneic. And eventually, we couldn't rescue the limb of the mouse. And this is a model to study, for example, in patients, peripheral artery disease. So those patients usually need amputation, they lose their leg. But when we transplanted now hypoimmune iPSC-derived endothelial cells, we saw on day 1 the same picture. We had removed the artery, there is no blood flow. After day 7, you see here, it starts to look very chaotic, a lot of color, but not very organized or structured at all. Around day 14, the same. And then around day 21, day 28, if you look, there's blood flow to the limb. So the blood flow is restored, and we basically preserved the limb so that the mice didn't lose the limb. And that showed us in another disease model that hypoimmune cells regenerated are not only surviving and evading the immune response, they're capable of treating in a mouse disease model the disease. So all the data I showed you so far, they were generated in UCSF when I was at UCSF full-time. We then started— I then started to do the work at Sana Biotechnology. And going forward now, all the other slides will be the work I did at Sana Biotechnology since I joined in 2019. So this, these data here show you now one field we are very interested in, and that is using the hypoimmune idea for CAR-T cell therapy. And the idea is that you would create an allogeneic CAR-T cell, basically off the shelf, that you could use in any patient in any hospital to cure cancer. And The model I show you here is a CD19 model. The tumor expresses CD19 and the CAR-T cells that were generated are against CD19. They see the tumor and the idea is that those CAR-T cells are killing the tumor. The question we had here is, do those three hypoimmune edits, the HLA class I knockout, class II knockout, and the CD47 knock-in, do they alter the activity or the function of the CAR-T cells. So these mice here are NSG mice that received a tumor, the NAM6 tumor, and you see on day 0, the tumor is everywhere, is mainly in the lungs of those mice, but then by day 27, you see the tumors everywhere, the whole mice, they are red. When we transplant unmodified CAR-T cells, the CAR-T cells are against the CD19, so they they do see the tumor, and you see they're killing, they're clearing the tumor in those mice. And when we are using hypoimmune CAR T cells, what you see is that the tumor clearance looks comparable to the unmodified CAR T cells. And that is an indication, was a first indication for us, that those three hypoimmune edits are not impacting the CAR T cell function. When we then studied the immune response, we saw in those mice that humanized mice, when they received unmodified CAR T cells, which are allogeneic, they're from a different donor, the mice react with T cell activation and very high blue graph. But the hypoimmune CAR T cells here, they stay in the background. So also for the CAR T cells, the hypoimmune CAR T cells, they evade the immune recognition by the recipient. And we also have here some antibody binding assays. So again, the unmodified CAR T cells, CD47 CAR T cells, you see antibody binding, but they're not binding to the hypoimmune CAR T cells. We paid a lot of attention to the innate immune cells. And what you see here are NK cell killing assay where we had a CAR T cell where we just knocked out the HLA or MHC class I, class II molecule. And you see there is immediately killing by NK cells because the fingerprint is missing. When we overexpress on the CAR T cells CD47, you see this green line stays up.. And this has not killed by NK cells. When we are now blocking the anti-CD47 with a blocking antibody, you see the signaling to the NK cells doesn't work anymore. And now we see killing again. So that shows you the relevance in that model for CAR-T cells as well of the CD47 molecule to inhibit the NK cell killing. So what we then did at Sana was The goal, of course, is to translate the technology into the patient. So we said, we see now it works in certain mouse disease models, the ones I shared today, but the goal is really the patient. So we were aiming for a higher bar translational model, and we choose the non-human primate model. The non-human primate model from the immune perspective is a very high bar because the immune system of those non-human primates is very reactive. So what we did is we had two different groups. The green cohort here had four NHPs, and we had used here unmodified allogeneic iPSCs. That means from another monkey. And we transplanted those into the muscle of the monkeys. And after 6 weeks, we then transplanted hypoimmune IPSCs into the very same monkeys. And then the vice versa experiment is cohort 2, first receives hypoimmune iPSCs, and then after 6 weeks, we transplanted unmodified iPSCs. And we really wanted to understand the immunobiology around it. And what those data show you, these are the immune data first, is that in the top row, when we transplanted unmodified iPSCs, allogeneic, from one monkey into the other, what we see is pre-Tx, that means before we transplanted the cells, there is no T cell activation. The reason is because the monkey T cells hasn't seen the antigen yet, so they aren't activated. But one week, what we label with 1W, one week after the transplantation, you see a huge increase in T cell activation.. And then the T-cell activation over time declines. It never goes below the background. And that, of course, for the follow-up we have here, was kind of expected because the T cells, they build memory. That if you would now inject the very same cells the second time, the T cells could even react more potent and faster. So the blue graphs stay up. However, on day— on 6 weeks, what we here labeled Day 0 for the red group, for the high immune group, but we have the red arrow there. What you see is the hypoimmune cells. Also, the T cells know the antigen, they are sensitized. They don't recognize that we transplant hypoimmune. There is no T cell activation against the hypoimmune cells, which is easy to achieve because we wiped off the fingerprint. There is nothing the T cell can see. For the antibody production, we see similar data. When we transplant unmodified allogeneic iPSCs, You see after 1 week, high IgM antibody response. It declines. And what happens is after 2 weeks, the IgM antibodies, they switch to the IgG. And that's what you see in those monkeys. So that shows you the monkey immune system is reacting like we expected in a transplant patient. But what you also see is that when we transplant the hypoimmune cells, when we have the red arrow, there's no de novo antibody production. You see the red. Grafts, they stay down. And I think the most exciting data are here under the IgG antibody production, because what you see is after 6 weeks when we transplant hypoimmune cells, these monkeys have really high, still very high levels of IgG antibodies. So that means those monkeys are sensitized, they know the antigen, and still there is no de novo IgG antibody production. And that shows shows us that we can think about redosing hypoimmune cell transplants, and we can think about going into sensitized patients. That could be, for example, a patient with autoimmune disease like type 1 diabetes. What we learned from those experiments on the bottom is these are the 4 NHPs that received the hypoimmune iPSCs, and what you see is that there is no movement when we transplanted hypoimmune and T-cell activation, no antibody production for IgM or IgG. But 6 weeks later, when we now transplant the unmodified iPSCs, the immune system recognizes them as foreign and starts to react with T cell activation, IgM antibody production with the blue graphs up, and IgG antibody production. And that shows you that the hypoimmune transplantations we did before does not immunosuppress those recipients. The immune system of those NHPs is capable of mounting a normal immune response to the unmodified allogeneic iPSCs we transplanted. And the innate immune response is really something we are studying in detail. And what you see here are the killing data out of those NHPs for the macrophages on the left and the NK cells on the right. The top row is just the control showing you that when we wipe off the fingerprints, so no HLA-I or II, that the cells get killed by macrophages and NK cells. But when we overexpress the CD47, you see here the green line stays up, there is no killing by macrophages, no by NK cells. We can reverse that by blocking the CD47, then you see macrophages and NK cells are killing. What is exciting for us, we did for the first time is that we had labeled the stem cells we put into those monkeys and we could follow them in vivo, in the living monkey over time. What you see here are the images from the first cohort, the 4NHPs that had received first the unmodified iPSCs and rejected them. On day 0, you see in this colorful areas, the transplanted iPSCs, they're all there. But within 3 weeks, you see that the signal disappeared, those cells got rejected. If you look here on the right side, what I labeled with summary, The blue line that shows you those monkeys we imaged every week, and we could show that the grafts get rejected within 2 to 3 weeks. And then we did a retransplant of the hypoimmune iPSCs into the very same sensitized recipient. And you see on day 0, all the stem cells are there, and those stem cells are surviving, although the very same monkey rejected the unmodified ones before. And you see here pictures 4 weeks after or 8 weeks after the transplant. Transplantation, and in red here, when you look at the graphs, these are the individual follow-ups when we measured every week. So this is quite exciting because the recipients are not immunosuppressed, and we are transplanting those hypoimmune cells into a very sensitized immune system. The other group of monkeys received first the hypoimmune iPSCs, and you see on day 0, the signal is nicely there. Here we show the follow-up for 4 months in 2 monkeys, where you see pictures of 5 weeks into other monkeys. And on the right side, what is in red, you see those are the hypoimmune iPSCs. So they do survive and even here proliferate in vivo without immunosuppression. And when we now transplant after 6 weeks unmodified iPSCs in blue, you see they do get rejected very fast. So while that monkey's immune system is rejecting the unmodified ones, the blue lines go down fast, the hypoimmunes still don't get recognized by the immune immune system, and they evade the immune recognition, and they persist and survive during that time. So the data what we have generated over those many years is really that hypoimmune cells with the 3 edits I showed you can evade allogeneic immune rejection. I showed you mouse data. We had humanized data and the NHP data. And they do not activate the missing self-response by macrophages or NK cells.. They can be transplanted into sensitized recipients. That opens now the opportunity for redosing or diseases where patients are sensitized. And they're not altering the recipient's immune system. So this was for us very exciting. These are brand new data because it hasn't been shown in an NHP model before without immunosuppression that an allogeneic graft survives. And thus, we think the cellular transplantation without immunosuppression appears to be an achievable goal with the hypoimmune cells. And what we are currently doing, Sanaa is moving directly into the hypoimmune engineered cell therapy programs for the T cells, for pancreatic islets, and for cardiac cells. So this is my last slide showing you that those 3 hypoimmune edits are not impairing the cell differentiation. So we show you here an example for cardiac cell differentiation or islet cell differentiation. The upper row are unmodified human iPSCs, the bottom row are hypoimmune, and we could differentiate them as well as the unmodified iPSCs. So meaning that those three edits are not impacting the cell differentiation.

And with that, I would like to end my talk and answer any questions you have. Thank you. Great. Well, it's top of the hour Thank you very much, Sonja. I think, well, I certainly appreciate it and know everybody else did, and best of luck at Sana, what you're doing in the future.

Thank you very much for inviting me, Kaspar.

Thank you. You're welcome.

Bye.