Good afternoon, everyone. Thanks for joining us. I'm Salveen Richter, a biotechnology analyst at Goldman Sachs. We're really pleased to have Keith Gottesdiener, Chief Executive Officer of Prime, with us. Keith, thank you so much for joining us. The company is a recent IPO, you know, for maybe for those less familiar with the story, could you just provide an overview of Prime Editing and your Prime Editing technology and your pipeline? Sure. Is this on now? Yes, I believe so. Prime Editing spun up in the summer of 2020. It came out of the laboratory of David Liu at Broad, who's a well-known entrepreneur and scientist. The technology really was a third-generation gene editing approach, there are many advantages of Prime Editing. We sort of lump them into four categories. The first is versatility. Not only can we inactivate genes, but we can edit genes and correct things. We can deal with any kind of base-pair mismatch that causes pathological disease. We can deal with small insertions and deletions. More recently, we've been able to actually put in very large multi-kilobase pieces of DNA, again, precisely targeted to a spot in the location, and also to loop out pieces of DNA as well. Really, when we think about what we can actually accomplish with a Prime Editor, frankly, there isn't much in terms of gene editing that we don't think we can actually do today. The second is, it's very precise. For a variety of reasons, the technique doesn't cause double-stranded breaks, is one important one, but also the fact that Prime Editing is a multi-stage process with checks along the way. It's very, very precise. It makes just the edit you ask it to do, and at the moment, it has minimal or no off-target activity. Obviously, that's preliminary, but we really hope that we can continue to develop that story. Third is, it's remarkably efficient overall. We can get very, very high levels of editing, really 60%, 70%, 80%, 90%. Many of our experiments we're doing to date, that's precise gene corrections, but it's also a once-and-done therapy. It makes the correction in situ at the actual location of the genome, so it stays under the endogenous control, which we think is very important for gene editing approaches. Last but not least, it's extremely broad. We've estimated that more than 90%, and that's an old figure, not even using some of the new improvements to the technology, we think more than 90% of genes or pathological mutations we can actually correct. Frankly, there are literally hundreds of places where this technology could be used, currently, we have a pipeline of 18 programs that we've been working with. 18 programs is probably quite a bit for a company our size and a company of our maturity, but the idea really was to figure out where Prime Editing could work well and to really start a little more broadly, and we expected that pipeline to narrow. One of the nice, very surprising, but things that gave us great joy is that so far, of those 18 programs that we started with, all 18 of them continue to move towards the clinic in a therapeutic sense. They fall into different categories, and the two big buckets we deal with them are things we call the immediate indications. These are indications where the genetics is well understood. There are thought leaders who want to work with us. There's great biomarkers, great unmet need, but probably most important is we can see the full path to the clinic, and that includes delivery approaches. We have eight indications that fall in that category, and we have eight that probably are a little farther behind that we call differentiated places where we think we could do something no one else can do. There's no question those approaches sometimes have a little bit more of a challenge towards moving them towards the clinic. Keith, there's multiple next-generation gene editors in development, including base editing and Prime Editing, on top of the ones that we've known for a while. How is Prime Editing differentiated from the other editing technologies, and can you just discuss some of the unique features and editing capabilities of Prime Editing that others may not be able to achieve? Correct. To some degree, I partially answered that question before. For example, CRISPR-Cas9 technology primarily is good at inactivating genes. We can do that as well, but we can go much farther beyond those. Base editing can correct four specific base pair mistakes or places where base pairs, in fact, are... cause pathological mutations. If you actually think and do the math, there's actually 12 different types of those mismatches that occur. We can do all of them as well, plus insertions, deletions, big and small. Those things aren't really available to most other techniques besides Prime Editing, it gives us a chance to work much more broadly. There are a number of companies out there who are doing things that look very similar to Prime Editing. Some of them have explained their technology. Some of them are very quiet about their technology overall. Some of it looks to us very suspiciously like Prime Editing overall, in practice at this point, it's probably too early to say. It is very important for us to emphasize that if you want to know how Prime Editing works and what it does, you could just go out to the scientific literature. There are literally a couple of hundred papers today that really outline Prime Editing by academics. There's a huge community where Prime Editing is done. You can understand the technique. You can look at its pluses and minuses. A lot of pluses so far, very few, if any, minuses, and you could really see what the power of it is. In some ways, that's been very useful to us because the technology has greatly advanced over the last couple of years. Delivery technology remains an area clearly where, there's ongoing optimization. Your pipeline includes both AAV and LNP delivery approaches, with some indications using ex vivo delivery. Can you just describe your strategy here and, you know, what else you might need to do when you're thinking about specific tissue targets that we don't really understand at this point? Yeah. Delivery is clearly something that's a challenge for the whole field, and I don't think it's any bigger challenge for Prime Editing than for any other approach. Our machinery is just a touch more complicated than base editing, which is just a touch more complicated than CRISPR-Cas9, but essentially all of these technologies share important commonalities. When we started with Prime Editing, once again, we went fairly broad. What we decided to do was to look and to try to understand and test potential therapeutic indications, excuse me, how Prime Editing interacted with different delivery modalities. Would it, in fact, work well with AAV? Would it, in fact, work well with lipid nanoparticles? Would it, in fact, be very useful to use an ex vivo approach? We didn't know the answer to that when we started, but I think the answer that's come through loud and clear over the last year or two is we can work with any one of those, and it works very successfully. Our ex vivo work, an area that can be challenging because currently that requires immunosuppression to put bone marrow cells back into the body, is really pretty extraordinary in terms of the levels of engraftment that we get and the ability for us to deal with and to edit stem cells without really impacting upon them in any way. Our LNP work so far is a little bit more nascent, but it's going together swimmingly, and we think it's a place where we've put some very big investments, plus we get some LNP technology and delivery from Beam, from our collaboration as well. AAV was a place where we weren't really sure whether it would work. We do require dual AAV to carry all the machinery, the same as base editing in many cases for CRISPR-Cas9. You know, when you're dealing with two AAVs that have to infect the same cell, pieces have to come together and editing has to occur, you're never 100% certain, is that going to work out particularly well? In our hands, actually, it's worked out much better than we could have expected. In our decks on our website, we show some data where if you actually use dual AAV, in our case, to infect the brain, the cortex, by injecting ICV into mice, that in fact, you can hit a good number of the cells. The important question we've answered is: If the cells take up one AAV, do they take up the second? Do the two pieces find each other, and do they create an edit? In a number of experiments that we included some of our corporate decks, we can show that of the transduced cells, those that are affected by at least one AAV, 90% or more go on and complete the whole edit. It just shows you that even though you're working with dual AAV, it's incredibly efficient, the process, and we can actually use dual AAV. We've done a lot of work with that, and we love our work. Not much of it is published or presented, but just two weeks ago, David R. Liu's laboratory put out a beautiful paper using AAV in animal models and others, and showed that it worked very well. We think our data internally is even a little bit better than theirs, but it's always nice to have a great laboratory do that. Maybe the reason our data may be a little bit better sometimes internally is a good number of the authors on that paper now work at Prime Medicine, so they brought over, you know, they brought over the technology, or maybe they learned it some bit at Prime Medicine, and they brought it back to the laboratory. In either case, it's important to realize we work with all. Now, this is a challenge. It's the problem we face with Prime Editing. If you have a great technology that seems to work everywhere, that doesn't cause off-target editing, you can use with any delivery technology, the biggest problem is: Where do you actually use it? We're just reaching the stage where we've set the universe in place about where Prime Editing can work. I'm sure we'll have more learnings in the future, but now we have to spend a little bit more time thinking about exactly where is the best place to use it. It's not a trivial question when you literally have hundreds of possibilities. Just to put that into context, I've told this story today to a couple of the folks who have had one-on-one meetings. At the end of last year, we had a company-building exercise where we gave a prize within Prime for whoever could come up with the best new indication to add to Prime Editing's pipeline. With 18 programs in a small company, we really didn't need another indication to add to what was there, to our intense surprise and great joy, we got 157 applications. At that point, the company was 200 people, and we actually had a science day where 60 of those were presented in posters. We talked it all through, and frankly, I could have started 30 more programs that day if I needed to. It just gives you some idea of what the promise of Prime Editing is and our ability to really move in many, many different ways. Maybe later with some of your questions, we could explore how we're going to deal with that particular problem as well. Can you help us understand in the context of that, when you're sitting here and when you decided on your near-term assets and your future assets, how you took into account your technology, where it may serve you best, and, you know, which areas, I guess, may be, already, you know, whether they're going to be treated by existing technologies and there's little ability to improve upon that, or there is an unmet need in the context of those. Like, how you put that all together with like, the ability to show proof of concept? Yeah. No, it's a great question, and it's a very complex answer, so rather than bore you for about 10 minutes with all the details, what I'd probably say is the following: Originally, it was really to explore the technology, 'cause from my experience, I've been in the drug development arena for at least 35, 40 years, I don't know exactly how long without counting, that people who get new technologies often hone in on a particular area that works well, and they leave all sorts of stuff by the wayside. I think our philosophy when we started was to go more broadly and to figure out where Prime Editing worked, and I think that strategy has been very successful. These days, other things come into consideration. Some of them have to deal with how we're going to deal with regulators. Another part of putting our pipeline together was to say, "Where should we start?" Our thoughts about that were, when you're going to go with a new technology to regulators, there are many things that make that process more direct and easier. One is you pick areas of very high unmet medical need. You pick places where your therapeutic intervention is likely to be more safe than what's being used today. Third is places where the approach really maximizes the chance for the regulators to get used to that particular technology. For example, we started with an ex vivo indication because you only edited cells outside of the body, and therefore, some of the off-target issues that one might hypothetically have had when we picked the pipeline would really be less important. Now, it's turned out our off-target activity has shown such so low amounts of off-target activity, in retrospect, that wasn't important, but we had to plan for that type of approach. Also places where the regulators could feel comfortable with the number of people that you were going to treat, because going out with a new technology that may carry risks and potentially aim to treat millions of people at, you know, with that indication, means that the bar is always higher in terms of the regulatory agencies. To some degree, we focused that particular part into what we were doing as well. We did all of that a little bit early. We're at the stage where a little bit of modulation of our pathway and our pipeline is almost certain to happen. We may drop one or two things off, not because they're not working, but because Prime Editing has improved each year and new possibilities opened up, and it may be time to put our focus in a few other places. A good example of that is at just last ASGCT, we did a beautiful job of showing that we can create CAR T cells in a way that no one else, we think, can actually do so, with a degree of engineering and design that just is impossible for other technologies. We're going to pursue that. It may mean that will take the place of one of the things in our pipeline. Last but not least, we're a little bit contrarian. Our general feeling was being the eighth sickle cell company probably wasn't the best place for us to be. There's some great companies that are in that space. Most of the areas where we've picked are places with only limited competition that could actually do what we think may be curative approaches. There are many people, for example, who are working in, you know, repeat expansion diseases or triplet diseases, but many of them were working indirectly. We can literally go in those diseases, we can literally snip out the pathologic repeats, we can put the DNA back together. Now, that's not mechanistically what happens, but when you're done, we've returned the DNA and the gene to wild type, to normal, and our working thought is that that's potentially a genetic cure, and hopefully, we'll hope to be able to show that it's actually a clinical cure as well. There are places like that where we know we can do things that absolutely no one else can do, and we hope to move to those kinds of programs very, very rapidly after we get the first couple programs into the clinic. You've also talked about partnerships and how that will be an important part of your approach going forward. How do you decide what to keep and work on on your own versus, you know, what you partner and when you partner? Well, to some degree, that's in our hands, and to some degree, that depends a little bit on, you know, potential partners. Our general feeling is that we probably need to partner at least a number of our programs in our current pipeline, because 18 programs in the clinic is just impractical for a company our size. Certainly, that's something that's on our mind, and we're looking down that list, in that regard. I think what we also want to do is to make sure that we form partnerships that potentially could expand. Our general approach to partnerships is, let's start with something small, but make sure the partnership agreements, excuse me, have the ability to expand if they turn out to be a great partner. I could talk about what a great partner is in just a minute. That when we go through this, they can begin to feel comfortable with Prime Editing. We could feel comfortable that they understand the Prime Editing. They're not going to do something that's going to, you know, what's the right word? Tear off the halo that Prime Medicine or Prime Editing currently has today, and really impact upon the technology. We need folks who are going to really approach things very, very responsibly and join with us in running these programs together. To some degree, we're a little bit more opportunistic on this, because while we love each of our 18 children, you know, I guess while we love even some of them a little more than the others, in the end, probably there's nothing on that pipeline we wouldn't at least consider as a single entity going out, as long as that's what it was and there was a chance to actually grow that relationship further onward. Actually, a lot of the interest comes from people who don't look at our pipeline. They come to us and they say, "We have a problem, or we have a disease, or we have something that we want to do. Prime Editing is the tool we want to work with that, and so what we'd like to do is to partner with you on X disease or Y disease. A number of companies have come and said, "We love Prime Editing. Let's figure out together where we should work, and let's see if we can put a small number of indications together in a particular area that we think is going to be important." Many of those things are ongoing. It's always hard in BD to know how those initial discussions are going to go, but the interest is very high, and our involvement with that is very large at the moment. Looking at your pipeline, you've guided to an IND filing as early as 2024 for your ex vivo program in chronic granulomatous disease or CGD. Can you just give us a brief overview of the preclinical data that you presented at ASGCT, and then how translatable you think that preclinical data would be to patients? Sure. We just gave a presentation about that. We've shown a lot of that data along the way, many investors have seen it from the early experiments all the way through. Just to remind people, chronic granulomatous disease is a disease mostly of childhood. The people who have this disease are unable to fight off bacterial and fungal infections because their white cells can't produce particular activity that actually is important in killing pathogens such as bacteria and fungi. What we do is we make a precise correction for people who have those mutations, and we actually return that particular activity back to normal again, so they can fight off infections. The process of doing any ex vivo approach is very, very similar. The idea really is that you take human cells in the preclinical studies, you actually edit them, okay, ex vivo, with your Prime Editor. Now we're doing that work with our actual development candidate. What you do is you put them into immunodeficient mice, and you let them grow in those mice for about 16 weeks, four months, and during that time, these mice can take human cells into their bone marrow and repopulate it. If you do that kind of experiment, what you do is you have an ability to test for engraftment, for editing efficiency, and eventually for actually the phenotypic activity that you'd like to do. Much of the data we showed was around that experiment, which we've now done literally a dozen times with many different donors. The idea really was to show a cross-section of that data, to show how robust our data is in that particular area. When you look at that, there are a couple of things that came out of the study that we focused on. The first was, is that we have high editing efficiencies. Over 92% editing occurs to fix the corrective mutation. The second thing is we seem to be incredibly mild on the actual stem cells that are part of the human CD34 population, and in fact, that's an important distinction that I think we can make with Prime Editing from some of the other gene technologies. When you finish the editing at the end and you take the CD34 population, only a key fraction of it is the part you care about, the long-term stem cells. More than 99% of it are partially differentiated cells, which generally are more robust and much easier to edit. What you really care about is not what the editing efficiency is right after you accomplish it, but what happens when you put them into the mice for 16 weeks and you actually watch and see what happens in the bone marrow when the long-term progenitors come in. Many other gene editing technologies have struggled to get very high levels of engraftment in those studies, and that's because the stem cells appear to be incredibly sensitive to insults, double-stranded breaks, or other kinds of approaches as well. What we found in the study that we pointed out in this work is that if you take mock-edited cells, the marrow was about 95% full, and if one takes, in fact, Prime-edited cells, you hit the same high level of engraftment. Then we showed that those engraftment cells hit that greater than 90% editing efficiencies, and that in fact, the activity returns and that these cells are pluripotent, so they make all the different lineages overall. The second thing that we showed in that is we reviewed a lot of our off-target activity from that program. Now we have IND-ready assays with the right degree of, you know, of specificity and sensitivity. With that, we were able to show that if we look with 10 or 12 different types of assays, we essentially detected absolutely no off-target activity. Probably that won't last as we go through many of our programs. We're reaching the sensitivity of the measuring methods. When you sequence, there's an inherent level of sequencing noise, and we're at that point because of the deep sequences we're doing. As of this moment, despite all of our efforts to identify even a single off-target example in our CGD program, we can't find any at all. That includes some very deep analysis of rearrangements and translocations and other things that have been seen with editing approaches. That was a major step forward for us from ASGCT. You asked the second part of it in my haste to explain very long-windedly, I can't remember what the second part of the question was. The translatability of the preclinical data. I do apologize. There is some short question that I don't love answering at length. I think there are parts of this that are very translatable and parts of it that will really be a little bit more difficult. The part that I think is translatable is this is just 1 of 18 programs where we get very high efficiency of editing. That part, I think that if we see it in this particular program, there's no reason to say that the Prime Editing approach shouldn't be able to hit that in other programs. I think the second part of it is that in all of our programs, we show literally from soup to nuts how our changes actually impact. It's not as if we make a genetic change, we say, you know, "Let's stop there," and the next thing you actually see is clinical data. We build that strong body of evidence, only part of which was in the presentation, that if you see this genetic change, this has a biochemical effect that results in a particular phenotypic effect. We do it in an animal model. It's all to build the case that this is all understandable, so you can look at things predictably. Since we're doing that in every program, if we're able to do that I think means that we're gonna be able to translate that in other places as well. Probably the most important thing is really about off-target activity. There's nothing about this particular edit or editing approach that we don't think is totally translatable to everywhere else. I think the only thing that may not be translatable is this is a very specific type of delivery mechanism. It may be that for others, LNP, AAV, it's very hard to predict if we're gonna get the same level of delivery efficiency. That one, I think, is more open. When you talked about some of the different diseases that you can focus on, I think one area is about introducing large-sized cargo into the genome through your PASSIGE technology. Can you just give us a little bit of an overview of this technology and how it could be advantageous when you think about CD19 CAR Ts in the context of the CAR-T genetic approaches? Sure. First of all, this is something that goes back to the original David Liu paper. In that original paper, there were examples of using this technology or something very similar to this technology, which means we think we have a very strong IP position from early on in this particular approach. The idea is simple but hard to put into practice, and we've been doing it within Prime Editing and Prime Medicine for a number of years, almost three years now, but really only spoke about it literally at the beginning of this year. The idea is what you do is you use Prime Editing to put a landing pad, a target sequence, or recombinase in the genome. One of the important things about Prime Editing is we can literally place that landing pad to the base pair exactly where we want it in the genome, and then we put in donor DNA with recombinase, and what that allows is for the integration of a large target right into that particular spot. We can do it with very, very high efficiency in the data we showed at ASGCT. We're able to put the landing pad with over 90% specificity, okay? It's important to realize with Prime Editing, we either put it in or we leave the DNA untouched, so we don't muck around with things in the process of getting that 90%. We're able to do 90%, and of those 90% of the cells, 60% of them took in the DNA exactly at the right spot as well. That was a completely unoptimized experiment, very early and preliminary. We think we can do better. Having said that, though, what we did is we began to create a new CAR T. Of course, to do a CAR T, what you need to do is put in the CAR protein, and you also need to inactivate the T-cell locus, and then you may want to do other editing as well. We're able to do is we're able to put the CAR T in exactly where we want, so we don't have to use a lentivirus and just put it anywhere in the genome randomly. We can put it in exactly in a spot to make sure it's really a safe locus. In this particular case, we actually destroyed the T cell location, the TRAC locus, and put our CAR T there. It now remains under the endogenous control of the T cell, you know, promoters and regulatory sequences, and at the same time, we took out the normal CAR T or the T cell receptor that would have been in there. It's kind of a very special twofer, okay? I think it's also the endogenous control that's a very important differentiator. What we also were able to show is multiplex editing, with other companies have done to be able to knock out other genes to do engineering. It's important to realize we don't only just knock out genes, we can correct, we can modulate them along the way. It's just an example of just how powerful the technique is to really edit these particular cells, to really make exactly the CAR T you want. Keep in mind that we do it with essentially no off-target activity. Regardless of whether you use it for an autologous approach or an allogeneic approach, you don't have to actually go around and try to select for places. We just don't find any off-target activity in most of our other cells. We haven't looked specifically in CAR Ts, to my knowledge, but it means that developing the tools and such are just much more powerful. CAR Ts play an important role in cancer. There's a lot of interest now in CAR Ts potentially playing a role also in autoimmune disease and other places as well. We probably believe at the moment, and this is not a final decision, that we don't currently have the expertise to really compete in the very crowded CAR T space, no matter how powerful our technology is in oncology. You know, you have to pick the right niche, you have to have the right relationship with investigators to get the patients, and you make one misstep in oncology these days with the competition, you know, you're done. Our general feeling, at least preliminarily today, is it would be better to partner with someone who has that expertise and let them carry it forward, and they may have very creative ideas, which we can translate with this powerful technology to potentially make new types of CAR Ts that matter. It's not just T cells. Potentially, one could do this with NK cells and other places that people are very interested. On the other hand, we do have a lot of expertise in autoimmune and immune, you know, other kinds of immunological diseases. We haven't made any decisions yet, so I don't want us to predict this is where we're going, but we're certainly giving a lot of thought about whether we can use that same kind of technology to explore places in that arena, which once again, may not be so easily reachable with the kinds of CAR Ts most others are making. Great. With that, any questions from the audience? We just have microphone to the front. Thank you. Elizabeth, it's not enough that Salvi asked you lots of questions, but you as well? Yeah. Yeah, go ahead. Your march up the chromosome programs, if you could touch on that and what kind of the future could hold for Prime Editing for personalized approaches. The March up the chromosome is our sort of name. We give a lot of names to many things we do at Prime Editing, so everybody knows what we're talking about, so apologize for the slogan, but the idea is very simple. Prime Editing is very modular. It's part of being programmable. You take the guide RNA, you switch it to another guide RNA. Much of the machinery stays the same, and so the ability to actually move between genes or within a gene is very important. There are a number of genes where there may be five or six very impactful mutations, but each ones are spread out along chromosome, and so you can't get at all of them with a single Prime editor. The idea was, let's build an approach as modular as we can possibly make it, so literally, the only thing that changes is the guide RNA, and within a gene, we switch from spot to spot within that genome. This requires a particular approach. It requires starting with one spot, a more common mutation, for example, and then working with regulators to introduce a second prime editor that's almost identical to the first, that can work on a second location without repeating everything one has to do in order to develop the second prime editor. Now, do we expect that the regulators are gonna, you know, fall at our feet and say, "What a wonderful idea?" We do not. There's a lot of precedent for ideas like that, so our general feeling has been, let's talk with the regulators with data. Some of our early programs, we actually are putting two Prime Editors into those programs to start. We're doing each of them in their entirety, doing every assay in duplicate, one for one and one for the others. With that, we're gonna have discussions with regulators as to, all right, there's an enormous amount of duplicative work here. What do we really have to do to do the third Prime Editor in this disease, or the fourth Prime Editor? We hope to make a data-driven decision with the regulators that the amount of data can become less and less and less. We certainly believe that we could put them all under one umbrella for clinical trials, and hopefully someday to actually include them in one NDA. That's a little bit in the future, but the idea really is that any edit that we could do, we'd like to do. Now, that also gives the opportunity for personalized approaches as well, because if you get to that point, that means any new edit that we want to make, even if it's a particularly rare edit, doesn't have this huge, you know, threshold that one has to go with. Frankly, in my old company, we were able to do that. At my company, Rhythm, we were able to convince the FDA that our drug worked in certain kinds of mutations, a little bit different, but then when new mutations came up, that we could extrapolate most of the data from the old indication to the new one, or the new mutation that we were dealing with, and they were willing to let us include them into the indication that we were doing. If we're, in fact, hardworking, and if we produce the data, frankly, if we're a little bit lucky, with the regulators, I think there's a possibility we'll be able to really move into places that are a little bit more personalized for people. Whether that's part of our business plan or not, I don't know. I can't imagine exactly how we're gonna make that work in terms of the cost of doing so. There may be many thoughts about that, foundations and other folks who may help us to get to that point. There's a lot of interest. Thanks, Elizabeth. Great. Well, with that, Keith, thank you so much. Thank you, everybody, for attending. Really appreciate your time. Really appreciate it. Bye-bye. Thank you.