Hi, everyone, and welcome to Kronos Bio's Virtual R&D Day. Thanks for joining our webinar. I'm Stephanie Yao, Executive Director of Investor Relations and Corporate Communications. Please note that the press release we issued earlier today is available on our website, kronosbio.com. During today's webinar, all participants will be in listen-only mode. The slides and a replay of our event will be available on our website shortly following the live broadcast. I'd like to remind you that this presentation includes certain projections and forward-looking statements as of the date of this presentation. These projections and forward-looking statements are based on the beliefs of our management, as well as assumptions made and information currently available to us, and reflect our current views and are subject to business, regulatory, economic, and competitive risks, uncertainties, contingencies, and assumptions. In light of these risks, uncertainties, contingencies, and assumptions, the events or circumstances referred to in the forward-looking statements may not occur. None of the future projections, expectations, estimates, or prospects in this presentation should be taken as forecasts or promises. The actual results may vary from the anticipated results, and the variations may be material. We have planned a great event for you today. Our President and CEO, Norbert Bischofberger, is going to kick us off with a welcome and introduction to Kronos Bio. He will then hand it over to our guest speaker, Dr. Eytan Stein. Eytan is the Director of the Program for Drug Development in Leukemia at Memorial Sloan Kettering Cancer Center. We're honored to have Dr. Stein join us, given his role and experience conducting clinical trials for novel therapies in AML. After Dr. Stein, our CMO, Jorge DiMartino, will unveil our development strategy for our portfolio of SYK inhibitors, and our CFO, Yasir Al-Wakeel, will outline the potential SYK opportunity. We'll move into a 20-minute Q&A session with our speakers. If you would like to submit a question, you may do so at any time using the Q&A text box. After a short break, Jorge will discuss the opportunity we have to target MYC using our CDK9 inhibitor, KB-0742. He will be followed by Charles Lin, our SVP of Biology, and Chris Dinsmore, our Chief Scientific Officer, who will highlight the capabilities of our platform and the progress we're making with our discovery pipeline. We'll hold another Q&A session for about 20 minutes to answer your questions about these topics. Finally, Yasir will say some brief remarks to close out our event. With that, I would like to now turn it over to Norbert. Well, thank you, Stephanie, and welcome everyone to our virtual R&D Day. At Kronos Bio, we are committed to discovering and developing therapies that have the potential to change the lives of those afflicted by cancer. We do this by targeting dysregulated transcription factors and their associated transcriptional regulatory networks or TRNs. These have long been known to play a causal role in cancer but have been difficult to drug. With a robust pipeline, we have a SYK inhibitor portfolio that has the potential to address two transcription factors, HOX and MEIS, which are dysregulated through mutations in about 2/3 of patients with acute myeloid leukemia or AML. Then we have a CDK9 inhibitor, and that has the potential to address another transcription factor, which is amplified in about 30% of solid tumors, and that's MYC. We have a proprietary product engine, which consists of a high-throughput screening methodology called SMM or Small Molecule Microarray, that is ideally suited to target these transcription factors and associated with that, we have deep computational biology expertise. Finally, we've been able to recruit an executive leadership team with a lot of expertise and experience, with collectively more than 25 therapeutic product approvals under their belt. Kronos Bio was founded only three years ago, we have made enormous progress, particularly over the last year. Just a year ago, we were only a research-stage company, we acquired a portfolio of SYK inhibitors, which immediately propelled us into a clinical-stage company that allowed us to do successful financing last fall, both crossover and IPO in October. Meanwhile, our internal program, KB-0742, has made progress. We published preclinical data in chemical and cell chemical biology. We filed an IND last December. The first patient with this drug was dosed in February. Finally, we announced a positive end of phase II meeting with FDA, with ENTO or entospletinib. This will allow us to go into a phase III registrational study in the coming months. This shows our pipeline. As I said, ENTO is our lead program. We're going to a phase III registrational study in frontline FLT NPM1 mutated AML. We have another compound behind it, lanraplenib or LANRA, we're going to call it. We intend to initiate two studies, one later this year, the other one early in 2023. The first one will be in relapsed/refractory FLT3 mutated AML. The other one will be in the frontline setting, unfit patients, NPM1 mutated with or without the FLT3 mutation. Finally, as I said, KB-0742 is making great progress. It's currently in a dose escalation and schedule-finding study, and we hope we will be able to share from the first couple of cohorts, both PK and PD data towards the end of this year. We're making impressive progress in our research discovery. You will hear a lot more about this with the subsequent speakers. With that, it's a great pleasure for me to introduce the next speaker. It's Eytan Stein. He is Director, Program for Drug Development in Leukemia at Memorial Sloan Kettering Cancer Center. Welcome, Eytan, and he will talk to you about both the AML disease and also the treatment landscape. With that, Eytan, over to you. Thanks, Norbert, so much for that kind introduction. I'm very excited to speak to you today about the management of acute myeloid leukemia in 2021. As was said, my name is Eytan Stein. I'm an attending physician on the leukemia service at Memorial Sloan Kettering Cancer Center. I also direct what's called the Program for Drug Development in Leukemia, which is our dedicated phase I program for acute myeloid leukemia and related diseases. In my role as the Director of that program, I have the opportunity to participate in clinical trials, which what I would consider are the most novel therapies that we have for AML. As many of you already know, but some of you might not know, acute myeloid leukemia is a hematological malignancy that affects approximately 21,000 patients per year in the U.S. When a patient is diagnosed with acute myeloid leukemia, one of the ways we know they have it is they have a bone marrow biopsy, which you can see on the left-hand side, where we extract bone marrow from the posterior iliac crest. On the right-hand side, you can see those ugly-looking, very large cells that are bigger than the red blood cells that are sort of the pale red thing in the photomicrograph, and those are myeloblasts. The reason this is abnormal is because you're not supposed to have myeloblasts in your peripheral blood, and you shouldn't have more than 1%-5% myeloblasts in your bone marrow in a normal, healthy individual. If you have more than that, then that is indicative that something is abnormal. If you have more than 20% blasts, that's consistent with having the diagnosis of acute myeloid leukemia. When you look at this photomicrograph, you can see that all of these cells look approximately the same. Many years ago, in the 1800s and the early 1900s, when a doctor suspected a patient might have acute myeloid leukemia, they would see this under the microscope, and they would say, "Yeah, those are too many blasts, and therefore, we know this patient has AML." In 2021, it's become clear that AML is much, much more complicated than just saying a patient has too many blasts. How you get to the phenotype of acute myeloid leukemia, that is how you get to having over 20% blasts, can come in many different forms. This is a pie chart from the European LeukemiaNet guidelines on the management of acute myeloid leukemia, looking at all of the different cyto and molecular genetic abnormalities that can lead to a block in myeloid differentiation, which is what acute myeloid leukemia is. I want to draw your attention specifically to the bottom blue wedge of the pie. That's the group of patients who have mutations in NPM1, otherwise known as nucleophosmin. This molecular alteration is the most common molecular alteration in patients with acute myeloid leukemia. It's not simple enough to say, "Hey, 30% of patients with NPM1 mutant AML." Because if you look below this blue part of the pie, what you can see is that those NPM1 mutations can coexist with other mutations. About 40% of patients with NPM1 mutations will also have mutations in FLT3. About 50% of those patients will have mutations in DNMT3A, about 15% of patients will have mutations in IDH1. Not only can you have double mutations, like an NPM1 mutation and a FLT3 mutation, or an NPM1 and DNMT3A, you can have triplets. You can have a patient with NPM1 mutant disease, with FLT3 mutant disease, with IDH mutant disease. Based on what the co-occurring mutations are, that is what dictates the prognosis and the outcome and the expected response to therapy in any particular patient. How are we doing when it comes to the overall survival of patients with acute myeloid leukemia? These are survival curves from about 10 years ago, and what I want to show you here is how the survival has improved and not improved in patients with acute myeloid leukemia. What I'm showing you here is very similar to what I could show you in 2021. I just like this survival curve because I think it's laid out quite nicely. You can see on the top part in panel A, that for patients younger than age 60, the 10-year overall survival in 1970- 1979 was about 5%, and that's improved to about 53% between 2005 and 2009. For patients older than age 60, the outcome is really dramatically worse. Almost no patients survived back between 1970 and 1979, and about 20% of our patients survive in the current era. Why is panel B important? Panel B is important because the vast majority of our patients are older than 60 years old. The median age of patients diagnosed with acute myeloid leukemia is about 68. Almost all of our patients are going to fall into these survival curves with an overall survival of 5-10 years of about, or five years, of about 20%. Going back up to panel A, I want to draw your attention to the following important point. Certainly, the overall survival has improved between 1970 and 2009. That's very, very clear. If I told you, or if I told myself that I had a 50% chance of not being alive in five years, I wouldn't be happy about that. Right? Survival has improved, but it's improved compared to just how dismal the survival was back 30 years ago. You can also look at survival and stratify that survival based on molecular genetics. The survival of any particular patient, the prior slide is sort of overall survival. This is survival based on the molecular genetic alterations that we see in groups of patients. Basically takes that pie chart that I showed you a couple of slides ago, and it looks at the survival of the patients based on the genetic abnormalities that they had at the time of diagnosis. Again, I want to draw your attention here to this green line, which is sort of in the middle. That's the group of patients with NPM1 mutant AML. You can see the overall survival of those patients is just about 40%, maybe a little bit higher than 40% at five years. Although we do consider nucleophosmin mutated acute myeloid leukemia without any other mutations to be favorable risk, that favorable risk leaves a dramatic opportunity for additional improvement because we want to get that number from 42% up to 100% survival. In addition, whether a patient survives with NPM1 mutant acute myeloid leukemia, as I told you previously, is heavily dependent on the co-occurring cyto and molecular genetic alterations. In these survival curves, what I'm showing you is the overall survival of patients with NPM1 mutant disease, either with or without other cytogenetic abnormalities. AK here means an abnormal karyotype, which means additional cytogenetic abnormalities. NK means a normal karyotype. You can see for the patients with an abnormal karyotype, the overall survival of those patients, even if they have NPM1 mutant acute myeloid leukemia, is just terrible, with an overall survival of just about 20% at four years or even three years. Again, just to highlight the points I was making before, the survival is heavily dependent on the mutations that we see at the time of diagnosis. Even patients with so-called favorable risk acute myeloid leukemia really still have a pretty bad outcome, with only 50% of those patients surviving. For NPM1 mutant acute myeloid leukemia, the overall survival can be all over the map, depending on the co-occurring cytomolecular genetic abnormalities that that patient has. How do we treat newly diagnosed acute myeloid leukemia? I'm going to take you through 50 years of clinical research in one minute. Up until about five years ago, we looked at our patients, and we decided, are they fit to get what's called intensive induction chemotherapy with 7+3, or are they candidates just for low-dose therapy with a hypomethylating agent or low-dose cytarabine? Maybe they're not candidates for anything at all, and all we can offer them is supportive care with supportive transfusion, prophylactic antibiotics. Over the past five years, what we've done is we've added drugs onto the backbone of 7+3. For newly diagnosed acute myeloid leukemia, for those patients fit to receive intensive induction chemotherapy, we've added other drugs onto that backbone of 7+3. In this case, midostaurin or gemtuzumab, or we've altered 7+3 a little bit, and we now give it for certain kinds of acute myeloid leukemia in a liposomal formulation. For the older patients with acute myeloid leukemia, we've added on to the backbone of hypomethylating agents and low-dose cytarabine, and we've added on venetoclax in some cases, and in other cases, we've added on the Hedgehog inhibitor glasdegib. For patients with relapsed and refractory acute myeloid leukemia, we now have targeted therapies against three specific molecular genetic alterations, ivosidenib for IDH1 mutant AML, enasidenib for IDH2 mutant AML, and gilteritinib for FLT3 mutated AML. Let's look at some of the survival curves for these patients who get targeted therapy for these specific genetic mutations in the context of relapsed and refractory AML. Gilteritinib, the FLT3 inhibitor, was approved based on the results of the ADMIRAL trial, published in the New England Journal of Medicine in 2019. This was a trial that randomized patients with relapsed and refractory AML with a FLT3 mutation between receiving gilteritinib monotherapy or receiving salvage chemotherapy. You can see the median overall survival favors gilteritinib, 9.3 months median overall survival compared to 5.6 months. The important point here is that gilteritinib itself is not curing anyone in the long term. These curves come together around two years. While you can get improved overall survival compared to chemotherapy, you're not, at the end of the day, making more patients live past two to two and a half years. What about IDH1 and IDH2 mutant disease? These are the overall survival curves from the trial enasidenib, the IDH2 inhibitor in relapsed and refractory IDH2 mutant AML. Again, median overall survival here of 8.8 months, very similar to the 9.6 months we saw with the FLT3 mutant disease. If you look at patients with IDH1 mutant acute myeloid leukemia, again, relapsed and refractory, the median overall survival in this group of patients is also just about nine months. Okay, we know in the setting of relapsed and refractory acute myeloid leukemia, that with targeted therapy that you give, you get a median overall survival with whatever targeted therapy we have approved to date between nine and 10 months. Whenever we have a drug that is active in the setting of relapsed and refractory disease, you want to give that drug up front. You want to move your best therapies and the therapies that are most active into earlier lines of treatment where you're going to cure more patients. Let me give you some examples of that when this was done with both FLT3 and with IDH. With FLT3 mutant disease, there was a clinical trial called the RATIFY study that combined a different FLT3 inhibitor called midostaurin in combination with chemotherapy for patients with newly diagnosed FLT3 mutant acute myeloid leukemia. You can see here that the five-year overall survival favors midostaurin. The difference between these two lines is about 7%-8% overall survival benefit. What happens if you combine an IDH inhibitor, either an IDH1 or IDH2 inhibitor, with intensive chemotherapy for newly diagnosed IDH1 or IDH2 mutant acute myeloid leukemia? This is what we did in a phase I clinical trial that we just published in Blood, where we combined either the IDH1 inhibitor ivosidenib on the top or the IDH2 inhibitor enasidenib on the bottom with intensive induction chemotherapy with 7+3 and with consolidation chemotherapy. Again, this is a non-randomized phase I study. If you look at the overall survival curves that were just published, you can see that the overall survival moving to IDH1 and IDH2 inhibitors in combination with intensive chemotherapy for newly diagnosed disease are really, I would say, quite impressive. You can see the overall survival out in about two years here is about 70%. If you compare it to historical controls, you wouldn't expect to see outcomes like this with just chemotherapy alone. It gets to the question of if this is a good way to do things, to take targeted therapies for patients with relapsed and refractory acute myeloid leukemia, picking off specific mutant targets, are there other targets that we can go after? One of the targets that I've been very excited about, as I sort of alluded to in the earlier slides, is nucleophosmin, is NPM1 mutations. Why is this? One of the reasons is that I read this paper published in Science in 2020 showing that NPM1 could be targeted, at least in this paper, with menin inhibitors. This is a preclinical paper looking at a mouse model of acute myeloid leukemia that also, at least in panel D, harbors mutations in FLT3. Actually both a FLT3 tyrosine kinase domain mutation and a FLT3 internal tandem duplication. What you can see here is that the survival curve in these mice at least, is really dramatically better than the survival curve in the control mice who did not get a menin inhibitor. We're now starting to see clinical data that menin inhibition may be effective in actual human beings with NPM1 mutant acute myeloid leukemia. This is data that I presented with the Syndax menin inhibitor about two or three weeks ago, showing that when you give this menin inhibitor to patients both with MLL rearranged and NPM1 mutant acute myeloid leukemia, you can get responses. Just focusing on the NPM1 mutant patients at the very bottom here, there aren't that many patients on the study. There are only seven patients on the study with NPM1 mutant disease to date. Of those seven patients, two of those patients had a response. My general rule of thumb to think about whether a drug might be active in acute myeloid leukemia is that if you get a response with a single agent in relapsed and refractory AML against a specific genetic mutation that is predicted pre-clinically where it would be effective, that is very, very encouraging, and that's something that is really screaming for further development, further study, and to be moved up into earlier lines of therapy. My final point is really about modern-day challenges with clinical trial design in AML. As I mentioned now at least three times, AML is a heterogeneous disease with many different molecular alterations with different predicted outcomes. The smaller you slice the pie, the harder it is to get enough patients to meet a traditional endpoint like overall or event-free survival. Trials in molecularly defined subgroups can take many years to accrue. The RATIFY trial took over 10 years before we had results of that study. It is not fair to the patients, honestly, to think that when we're trying to do studies in an NPM1 mutant population or other molecularly defined subgroups, that we're going to be able to wait 10 years for an overall survival endpoint. We need novel endpoints such as MRD negativity. That means the patients have the absence of measurable residual disease. They can act as surrogate endpoints so that we can get answers to these important clinical questions and maybe even get drug approvals more quickly. The reason I think MRD is a good surrogate endpoint for survival is based on a whole bunch of different data. The data I like the most was published by the Dutch HOVON Group in the New England Journal of Medicine in 2018. What this study did is it basically looked at a large number of patients who had been treated for newly diagnosed acute myeloid leukemia on clinical trials that were done by the Dutch HOVON Group. When a patient achieved complete remission morphologically, the investigators asked the question, what was the relapse rate of these patients if they had or did not have measurable residual disease? The second question was, what if we looked at the different types of measurable residual disease assessment? One being next-generation sequencing, NGS. One being MFC, which stands for multi-parameter flow cytometry. You can see here that the highest rate of relapse in the purple line is in those patients who are MRD positive. Again, morphologic, they're in a morphologic complete remission, but they're MRD positive by both next-generation sequencing and by multi-parameter flow cytometry. The lowest rate of relapse is for those patients who are negative for MRD, both by NGS and by flow cytometry. Patients who are somewhere in the middle, meaning they have either NGS or MFC, or MFC and not NGS, they fall into an intermediate relapse category. When you relapse with acute myeloid leukemia, that directly impacts your survival because patients with relapsed acute myeloid leukemia are predicted not to survive very long. It's why I believe that using MRD as a surrogate endpoint to read out your clinical trials is going to be an important thing that we're going to be doing for many clinical trials over the next 5-10 years. In summary, I want to just summarize all of my points here. Molecular studies are part of the routine assessment of patients with newly diagnosed and relapsed acute myeloid leukemia. Mutations rarely occur in isolation. They commonly co-occur in predictable patterns, such as NPM1 mutations with FLT3 mutations, and how they co-occur can predict the overall survival. NPM1 mutant AML specifically has a variable prognosis, but the overall outcomes in these patients, even in the best-case scenario, is woefully insufficient, with only about 50% of these patients surviving. We're slowly chipping away at molecular subgroups that can be targeted, such as FLT3 and IDH1 and IDH2. I showed you how we're using these drugs as part of standard clinical practice for relapsed and refractory AML. We're now starting to use these drugs for patients with newly diagnosed acute myeloid leukemia with the mutations of interest as well. I believe nucleophosmin is really ripe to be the next target, not only because there's a proof of concept that there are drugs that are going to work against this target, but also because it encompasses 30% of patients with acute myeloid leukemia. Finally, as I just said, modern-day clinical trials for AML really need to use surrogate endpoints such as MRD negativity, which can predict relapse and survival earlier than time to event endpoints like overall and event-free survival. With that, I want to thank you very much for your attention. I'll be happy to take questions at the end of the presentation, and I want to pass it on now to Jorge DiMartino. Thank you very much. Great. Thanks, Eytan. That was a really great overview of the current AML landscape. I'll be touching on some of the elements from Eytan's talk as I go through the next 20 minutes or so our SYK development strategy in AML. I'll start by giving a brief overview on the biological rationale for SYK as a therapeutic target in genetically defined subsets of AML, including NPM1 and FLT3-mutated AML. Now, some of you may already be familiar with ENTO, our lead SYK inhibitor that's going into a registrational trial this summer. Today, I'll share with you our view on LANRA. How we're planning to deploy both SYK inhibitors to really maximize the impact of SYK inhibition across AML. As Norbert said, Kronos Bio is a company that's really focused on drugging dysregulated Transcriptional Regulatory Networks in cancer. This can take a couple of forms. As you'll hear a little later from Charles and Chris, we do have a discovery platform that really lends itself to being able to drug transcription factors and their associated proteins directly. Moreover, we have a great deal of expertise in mapping these networks, and what that leads us to from time to time is identification of other vulnerabilities that we can target therapeutically. That's really where SYK fits into the Kronos strategy. Just to briefly walk you through this, what this cartoon depicts is a lot of data from a number of publications, a couple of which are listed here. In a nutshell, what it suggests is that this transcriptional regulatory network around the transcription factors HOXA9 and MEIS becomes dysregulated as a consequence of NPM1 mutations, which occur in 30% of AML patients, or rearrangements of the MLL gene, which occur in 5%-10% of AML patients. As a result of that, SYK becomes massively overexpressed and constitutively activated. This drives signaling down its own cascade, which drives cytokine-independent growth genes. There's also a positive feedback loop to MEIS1. SYK really sustains its own overexpression by continuing to drive MEIS1. All of this suggests that SYK is functioning as what we refer to as a central node within this TRN that lends itself to being a therapeutic target in subsets of AML with high HOXA9/MEIS. This is the biology that we're planning to exploit, and I'll go more into that in the subsequent slides for our planned phase III trial. Beyond this, there's very good data that SYK and FLT3 collaborate to drive leukemogenesis. It's this cooperativity that really suggests that the combination of a SYK inhibitor with a FLT3 inhibitor will be very active in AML as driven by FLT3. I will touch more on that in subsequent slides. We were very pleased just last summer to be able to announce that we had acquired a couple of clinical-stage SYK inhibitors from Gilead. These are both what I consider to be sort of next-generation SYK inhibitors with oral bioavailability, very good potency against SYK itself, but very selective against other kinases. There are some differences between them, mostly relating to the BID versus QD dosing that I'll touch on later. Importantly, they both come with a solid package of clinical data. Particularly, ENTO comes with a package of clinical data, phase II data in AML that I'll share with you in a few slides. It's that data package that is really driving us to launch a registrational trial around ENTO this summer. The strategy that we're going to be laying out for you today starts out with that. It starts out with how we're going to capitalize on the data that we have already in hand with ENTO to launch a registrational trial this summer that will help us get ENTO to patients as soon as possible. However, we think really this is just the beginning. There is a much broader opportunity for SYK inhibition to address genetically defined AML patient populations across the full spectrum of age and fitness levels by combining not only with the 7+3 backbone, but with the VenAZA backbone that's used in older and less fit patients. Perhaps even more enticingly than that, we think there is the opportunity to create rational combinations of genetically targeted agents for AML, starting with the combination in FLT3 which I'll talk about more later. Again, because these are novel combinations, we'll be starting out in the relapsed and refractory space. Needless to say, if the activity looks spectacular there, we would seek to move those forward towards a label. In parallel, moving those combinations into the frontline space where I think we could really start to have a significant impact on patient survival. A good part of the reason that we're confident enough in ENTO to move it into phase III is a combination of the clinical data that I'll show, but also the strength of the biological rationale. A couple of slides ago, I showed you a summary of a lot of data really implicating SYK as a downstream target in HOXA9/MEIS-driven AML. Much of that work was done with mouse models. Completely independent of that work, it turns out that Brian Druker, as part of the Beat AML Consortium, was looking at how patient AML samples ex vivo respond to various agents. This was an unbiased drug sensitivity screen with 122 drugs. One of them happened to be entospletinib. Samples from 572 AML patients were sequenced, so he knew what the genetics of each sample was. Really, he was trying to correlate sensitivity to specific agents with particular mutation profiles. What Brian noted and went out of his way to point out to us was that the samples that were responding to entospletinib ex vivo all had NPM1 mutations, either as the sole mutation or co-mutated with FLT3-ITD or with the DNMT3A mutations. Really, it was in the context of this preclinical understanding that we were able to interpret the clinical data emerging from the Gilead phase I-B/II study of entospletinib in AML. This was a fairly large study. It enrolled about 148 patients on three different study arms. One arm looked at entospletinib monotherapy in relapsed and refractory AML. A second arm enrolled patients who were unfit for intensive induction and treated them with entospletinib with a hypomethylating agent. The data shown on this slide is from an arm where 53 newly diagnosed AML patients who were eligible for intensive induction received a 14-day lead-in window of ENTO monotherapy followed by cytarabine daunorubicin induction, the so-called 7+3 regimen, together with entospletinib. There were a number of important learnings from this trial. First of all, one of 10 MLL-rearranged AML patients in this arm of the study achieved a complete response with ENTO monotherapy. There were an additional two patients on the relapsed and refractory arm who achieved complete responses with ENTO monotherapy. A total of three MLL-rearranged patients out of 23, combining the frontline and relapsed patients who achieved single-agent CRs with ENTO monotherapy. As Eytan said earlier, when you see responses in the genetically defined subset where you expect to see it based on your preclinical understanding, that really is significant. It tells you that you're on mechanism and that the hypothesis is likely to hold true. This is certainly supported by the combination data. Now, when these patients received entospletinib together with 7+3 induction, about 70% of them achieved either complete response or a complete response with incomplete hematologic recovery, so-called CRi. This was a relatively high-risk group of patients. Many of them were older than 60. A number had secondary AML and/or FLT3 co-mutations. Again, a relatively high-risk group of frontline patients. 70% is roughly what you might expect with 7+3 alone. However, when you look at the patients with the mutations that drive high HOXA9/MEIS, NPM1, MLL-r, or even NPM1 with FLT3 co-mutation, the CR rate is substantially higher than that 70%. It's close to 90%. Even more importantly, it's much higher than the CR rate in the patients lacking those mutations. This also is seen in the survival data shown in the Kaplan-Meier curves at the right. In this retrospective analysis, patients who had higher than the median HOXA9/MEIS expression shown in the blue line had significantly superior overall survival to patients with low to normal levels of HOXA9/MEIS expression shown in the pink line. Again, really suggesting within the context of the single-arm study that the benefit of adding ENTO to 7+3 was occurring preferentially to patients with either high HOXA9/MEIS-associated mutations or high HOXA9/MEIS itself. Very importantly, the safety profile of ENTO when combined with 7+3 induction was considered to be quite tolerable. About what you would expect with 7+3 alone. Count recovery occurred on time. There was no evidence of enhanced myelosuppression or other significant toxicities. When you see data like this, the obvious next step to take is a randomized trial, and I will share that with you. Before, I think it's important to touch on some of the insights that we had that informed how we designed this phase II trial or, sorry, this randomized trial that made it very attractive to us. As I said, one of the findings from the Gilead study was that high HOXA9/MEIS really drives the benefit of adding ENTO to 7+3. If you were to design or develop a patient selection platform based on high HOXA9/MEIS expression, it's doable, but it's a heavy lift. We recognize that NPM1 actually is a much better way to select patients for this study. This is because it accounts for the lion's share of high HOXA9/MEIS expression. It's 30% of AML. We know that it's a gatekeeper mutation. In other words, it marks the leukemic clone, and mechanistically, we know that it drives high HOXA9/MEIS expression. You can very reliably enrich for high HOXA9/MEIS patients by selecting with NPM1. Moreover, it's a fairly simple or non-heterogeneous set of mutations, small insertions within exon 12 which are readily detectable by next-gen sequencing or PCR-based methods. There are already plenty of existing platforms in clinical use today to identify patients who are NPM1 mutated. The other advantage of the sort of genomic simplicity of these mutations is that they lend themselves very nicely to detection of MRD. As Eytan mentioned, MRD has become an increasingly useful part of modern AML therapy. What this chart shows is that you can be in CR, in complete response, with less than 5% blasts and still have a very substantial burden of leukemic cells left in the body. Those are patients that we would expect to relapse fairly shortly after achieving remission. You could have much deeper responses. In other words, you've knocked those leukemic blasts down by several logs, in which case you'll have a longer relapse-free survival and potentially a longer overall survival. Over the last several years, methods have emerged that allow us to reliably detect the presence of these leukemic blasts in patients who are in morphologic CR quite reliably. The most reliable of these methods are molecular methods, where you're using polymerase chain reaction or next-gen sequencing to detect the leukemic cells. In order to do that, you need a change in the DNA that is found only in the leukemic cells, not in the normal cells. NPM1 mutations essentially provide that. Because they're so readily measured, many studies over the last 10 years have looked at MRD negative CR in patients with NPM1 mutations after intensive induction and have shown consistently that if you have undetectable NPM1 after induction, you are likely to survive much longer than patients who still have detectable measurable residual disease based on NPM1. This table here just shows a few of those studies. I can tell you, in fact, that there are many, many more in the literature, and it's largely on the basis of these studies that the European LeukemiaNet now recommends routine monitoring of MRD for patients with NPM1 mutations as a way to assess the depth of response after induction therapy. The FDA has also been paying attention to this space. Over the last several years, they've really engaged with industry partners to lay out what the requirements for MRD as a regulatory endpoint should be. Most recently in 2018, blinatumomab was approved in relapsed and refractory acute lymphoblastic leukemia based on an MRD endpoint. In January of last year, the FDA issued definitive guidance for industry partners to lay out what they're looking for in MRD as a registrational endpoint. We actually followed that guidance very closely as we approached FDA back in February with the design that's shown on this slide. This really sums up how these insights impacted the study design. First of all, focusing on the NPM1 mutated subset, we're able to use existing assays to select patients so we can start the trial right away. We will perform the necessary validation so that one of these assays becomes the companion diagnostic that's launched with ENTO so the trial will be successful. Finally, we proposed MRD negative CR as a primary endpoint for accelerated approval, with a secondary endpoint of event-free survival, since we'll continue to follow these patients up for relapse and survival. Suffice it to say, we had a very productive discussion with FDA. Based on that discussion, we have a path forward, and we'll be initiating this trial as designed starting this summer. As I mentioned, this is really just the beginning. In the next few slides, I'll talk to you about the next couple of studies we have planned and how LANRA fits in as part of this overall plan in AML. Both ENTO and LANRA are highly potent selective oral SYK inhibitors, but there are important differences between them that lend themselves to specific treatment settings in AML. Specifically, LANRA is once daily dosing, and it doesn't have the contraindication for proton pump inhibitors, so patients can continue to take those if they need them. There's no food effect, so it can be taken fast and fed. We think all of these features really lend themselves to treatment paradigms that are more chronic in nature. As opposed to where ENTO will be used, which is during induction and up to three cycles of consolidation, really a defined period of treatment. LANRA can be dosed, for example, in relapsed refractory disease or frontline elderly unfit combinations where the standard practice is to dose to progression. This can be a period of a year or more, depending on how effective the treatment is. However, LANRA doesn't have any data, clinical data or even preclinical data at the time that we acquired it in oncology indications or AML specifically. We really wanted to be sure that LANRA was going to have anti-leukemic activity that was comparable to what was seen with ENTO. In order to achieve that, we relied on treatment of primary patient AML samples, either from peripheral blood or bone marrow. You'll recall this is the system that Brian Druker and his team used initially to discover the activity of ENTO in NPM1 mutated AML samples. We were very pleased to see that when we looked at LANRA and compared it side by side to ENTO in this model of cancer, we saw essentially equivalent activity. This is shown by the curves on the left, which are representative curves from the many samples that we looked at. These are shown in the panels on the right, where on the x-axis, we look at ENTO activity and on the y-axis, LANRA activity, really looking for the samples to sort of fall along the diagonal, showing that they're relatively equivalent. This is, in fact, what we've seen. We're very confident that this is going to translate also into clinical activity of LANRA in NPM1 and FLT3-mutated patients. We use the same system to study the interaction of LANRA with other agents. In this case, we're looking at the combination of LANRA with gilteritinib in primary patient sample across a range of concentrations that includes the clinically relevant concentrations. What we're seeing here predominantly across most of the concentrations is additivity. However, in the range of the clinically relevant combinations, we're seeing some mild synergy. On the basis of these data, we're going to be starting up towards the end of this year, the trial that's shown on this slide. This is going to be a phase I/II study of LANRA in combination with gilteritinib in patients with relapsed and refractory FLT3-mutated AML. It's going to be in two stages. In the first stage, we're going to be trying a couple of different doses of LANRA in combination with the standard approved dose of gilteritinib really predominantly to establish the safety of the combination and to select the dose of LANRA in combination with gilteritinib that we want to move forward into the expansion cohort. Here we'll continue to enroll patients at the recommended phase II dose of LANRA in combination with gilteritinib, to assess more of the safety, but really to get a handle on what the antileukemic activity looks like in terms of the composite CR rate and the duration of response. In terms of benchmarks for what we would consider to be a successful outcome of this expansion, we're relying on the data from the ADMIRAL study. This is the study where gilteritinib was approved as a single agent in relapsed and refractory AML, showing a complete response rate of about 21%, with nine months median overall survival and just under three months event-free survival. These are the benchmarks that within the context of a single-arm study, we will be evaluating ourselves against to determine whether to move forward into a potential randomized registrational trial. Similarly with LANRA, we've looked at the combination of LANRA with venetoclax. This is using the MV-4-11 AML cell line, which is an MLL rearranged cell line. Again, here we see across a range of concentrations, including clinically relevant, clinically achievable concentrations that we're seeing anywhere from additivity to mild synergy with concurrent treatment with both agents. The trial that we have planned for LANRA with the VenAZA backbone we anticipate will start in 2022. This will enroll patients who are either over the age of 75 or who have a comorbid condition that precludes them from receiving intensive induction with 7+3. These will be patients who have either NPM1 mutation, a co-mutation of NPM1 with FLT3 or FLT3 mutation. We expect to see activity across all of those subsets, the first stage of the trial, much as the gilter trial, will really be about safety and dose selection, looking at a couple of different LANRA doses in combination with the standard VenAZA dose, followed by then an expansion, a single-arm expansion at the recommended phase II dose, to assess antileukemic activity. Our benchmarks here are based on the VIALE-A trial, the randomized trial of AZA with or without venetoclax, where the complete response rate was about 36.7%, with 14.7 months overall survival. Interestingly, NPM1-mutated patients in this trial had pretty much the same outcome as the overall population. In other words, NPM1 did not seem to be a favorable prognostic factor in this trial. We think that these benchmarks from the overall population are still really valid for interpreting this trial. In both this and the gilter trial, we'll also be incorporating MRD negativity as an exploratory endpoint to help us get a handle on the level of activity. This really covers the plans for ENTO and LANRA over the next couple of years. I'm going to hand it over now to Yasir Al-Wakeel, our Chief Financial Officer, and he's going to walk you through how we're viewing the market opportunity for SYK inhibition in AML. Thank you for your attention. Thanks, Jorge. I'm Yasir Al-Wakeel, CFO and head of corporate development here at Kronos Bio. It's my pleasure to walk you through how we're viewing the SYK opportunity. What's particularly exciting about this is that SYK is a key player in a diverse set of biological functions. As such, inhibition of SYK has therapeutic potential in a number of different diseases. In particular, given its role in cell signaling in both the myeloid and lymphoid lineages, it is a relevant target in both. While the focus of today's SYK conversation, as you've heard so far, is on AML, I did want to make the point that the opportunity is potentially broader, and we look forward to discussing additional indications, such as ITP, further in the future. As we think about the AML landscape from a mutational perspective, SYK has the potential to address more than 2/3 of patients. NPM1 represents approximately 30% of mutations in AML. There are currently no approved treatments specifically targeting this mutation. This, as you have heard, is the primary focus for our ENTO phase III trial. In addition, mechanistically, SYK has the potential to add benefit to patients with FLT3 mutation as well as MLL rearrangements. These offer an incremental 30% and 9%, respectively. As you heard today, we have our first trial with LANRA in patients with the FLT3 mutation. Another way of looking at the AML landscape is in terms of lines of therapy. Each year in the U.S., around 20,000 patients are diagnosed with AML. Within the front-line setting, standard of care differs depending on whether or not patients can tolerate intensive chemotherapy, with the current split being around 50/50. With the addition of the LANRA trial in combination with VenAZA, we now have the ability to offer a targeted therapy to the entire newly diagnosed NPM1 patient population. ENTO with 7+3 in the fit population, and LANRA with VenAZA in the unfit population. Furthermore, in this front-line unfit setting, where no FLT3 inhibitor is approved, we will also be recruiting patients who have the FLT3 mutation. This opens us up to potentially benefiting 60% of the unfit front-line opportunity. In the relapsed or refractory AML setting, which comprises approximately 15,000 patients per year in the United States, we will also be targeting FLT3 patients through a combination trial with gilteritinib. As you heard from Jorge, this is an area where there is substantial opportunity to improve patient outcomes. Hopefully, what has become clear today is that our development strategy allows us to differentiate the positioning of both therapies very clearly in a way to best serve patient needs. ENTO is being developed for shorter-term use in the only population where cure is still possible, with the intent to increase cure rates. It benefits from already having a body of data in the AML setting and as such, could be quick to market. LANRA, on the other hand, is being developed for more chronic treatment to progression-type settings in patient populations with the poorest outcomes and highest unmet need, with the intent to increase responses and offer continuous disease suppression. Our strategy is very much aligned with the characteristics of each therapy, with LANRA having a more patient-friendly profile for chronic use. I'll close by summarizing some key points. As you heard from Dr. Stein, outcomes for patients with genetically defined subsets of AML, including NPM1, are woefully insufficient. With our SYK inhibitors, ENTO and LANRA, there is strong biological rationale to address mutations present in more than 2/3 of AML. As Jorge walked you through, ENTO and LANRA are differentiated clinical-stage SYK inhibitors. Our strategy in AML, ENTO in combination with induction regimens with curative potential, and LANRA in combination with regimens for continuous disease suppression, allows us to maximize the impact of both investigational therapies. More importantly, the availability of ENTO and LANRA allows us to tailor these therapies to address patient needs. Now, I'll turn it over to Stephanie to moderate our first Q&A session. All right. Thanks, everyone. If you do have a question, I just want to invite you to type it into the chat. To start off, we have a few questions from Mike Yee at Jefferies. His first one is, Dr. Stein really emphasized heterogeneity even within large groups such as NPM1. Maybe this is for Jorge. What can we say about the heterogeneity here and how we're looking at enrolling our ENTO study? Yeah. Certainly, there is some heterogeneity, as Eytan pointed out, with regards to age. The study is designed with the stratification factors of age, so we'll be stratifying for age 60 and above or below 60. We'll also be stratifying sorry, for FLT3 status in the sense that we are allowing patients with FLT3 mutations to enroll in areas where midostaurin is either not approved or not available. To the extent that we do enroll those patients, we'll be stratifying them on the basis of FLT3 mutations. Those are the main drivers of differential outcomes we're looking at. Great. A question for Dr. Stein. Various drugs have had a hard time showing OS benefits despite various CR rates. How do you think about CR rates in MRD negative CR for 7+3 chemo, which are already pretty high, and the ability to improve upon that and drive longer-term OS benefit? Well, if I'm understanding the question correctly, I do think that benchmarking things to MRD negative CR has a better chance of being a legitimate surrogate for overall survival because of the data that I showed you in the second to last slide, where those patients are much, much less likely to relapse. I do think that using MRD negative CRs is really probably the only way forward as time goes on and as we break things down into smaller and smaller molecular subsets, because just the trials will never get done if we don't do it that way. As I said a second ago, I think they're a very good surrogate for overall survival. Stephanie, maybe I can quickly add to that as well. Mike's right. Certainly, the majority of patients do achieve morphologic CR. What we've seen in the literature is even if 70% of patients are in CR, anywhere from 30%-40% of those patients will still have detectable MRD at the time that they're in CR. We think that that provides ample room for improvement. Great. We'll move on to our next question from Geoff Porges. Maybe this is for Norbert, and then Jorge can pick up after Norbert comments. With the development history of our SYK inhibitor portfolio, what are some of the information or the new information that we know now that perhaps Gilead did not know then when it chose to prioritize the SYK inhibitors, and then how we're thinking about the development based on the new information, and then also maybe thoughts on combination with CD47 and what that might look like. Geoff, good to hear from you again. It's a pleasure. We made certain assumptions at Gilead that were completely correct at that time, and those assumptions made this a not very attractive opportunity. The two assumptions were, number ine, we needed a companion diagnostic to identify high HOX/ MEIS patients. That, of course, did not exist and had to be prospectively validated in order to identify the patients that could be enrolled in the study. Secondly, the endpoint was going to be event-free survival, which is large sample size, long timelines. As Jorge nicely presented here, we now know that NPM1 mutation is a perfect marker of high HOX/ MEIS, and that then allows us also to use MRD as the endpoint, which is a much smaller sample size and shorter timelines than we assumed. The other thing you asked, Geoff, was what are the conditions? This was an asset acquisition. We do owe royalties and milestones to Gilead, but they don't have any further rights to these compounds. Jorge, do you want to make a comment on CD47? Yeah, CD47 certainly seems to be an interesting agent. It's really not targeting a genetically defined population as best I know. Again, we would think of it in terms of being part of the backbone. I think the advantage to genetically defined targeted therapies such as SYK is that they can be added onto various backbones. Presumably, as CD47 finds its place in the treatment paradigm, we would probably be looking into that. Great. I'll move on to our next question from Andrea Tan at Goldman Sachs regarding our LANRA trials. The first question is, what type of delta would we want to see above gilteritinib and monotherapy in order to move forward with that trial? If we are going to assess MRD in the trial and is FLT3 a marker that can be used in the same way as NPM1? Yeah. Essentially, we're taking a Simon two-stage approach in that trial, same thing with the VenAZA trial. The benchmark CR rate is a little over 20%. That Simon two-stage approach is basically designed to. That sort of sets the null hypothesis. It's designed to exclude a CR rate of 20% with about 80% power. Essentially targeting an alternative hypothesis with about a 40% CR rate. Obviously, we're going to be looking very closely at safety and durability and other aspects, including MRD negativity. FLT3, I think there are now platforms of it. One of the challenges with using FLT3 mutation as an MRD marker historically has been that it's a relatively late mutation and there's a lot of the normal FLT3 allele around. However, I think those challenges have been resolved technically, at least. There's at least a couple of different trials that I'm aware of using FLT3 MRD as a marker. We would be very keen to do that as well. Great. Maybe a question that we get a lot, maybe this is something that both Dr. Stein and Jorge can comment on. How we're viewing the AML landscape and all those various therapies that are in development and what makes maybe our inhibitor portfolio unique in a sense, also, thinking back to our development strategy potential for rational combinations. Maybe Jorge you can start and then Eytan can also chime in. Yeah. It's an exciting time, right? I think one of the differentiating aspects right now is I'm not aware of other frontline trials specifically targeting NPM1. I think the developments around the menin inhibitor are very exciting. For now, those are in the relapsed and refractory space. At this moment we're really focusing on that frontline population. Ultimately, sort of in my dreams, I would see rational combinations of genetically targeted agents being able to perhaps even place some of the more toxic aspects of the standard chemotherapy backbones. Where there are co-mutations such as NPM1, FLT3 or NPM1 and IDH, to put together these rational, genetically targeted therapies and really take advantage of what we've learned over the last few decades. Yeah. I would just add to that, or agree that certainly the space has become a little bit more crowded. That's obviously good for patients because we have more options. Like Jorge was saying, what we're really going to have to do to make significant progress with this disease, given its molecular heterogeneity, is really understand how to take agents that are active against in genetically defined populations and combine those agents both together and then combine those agents onto various backbones that are less toxic. I hear a lot, or I get a lot of questions about, well, you've got so many drugs for AML now, what are the opportunities? The answer is there is still a ton of opportunity. This is a disease that is by no way taken care of, which is what I tried to show in the earlier slides. We have a lot of work to do to get the outcomes of patients to be better. We're going to need new therapies and new combinations to get that done. Great. We have a question here from Bill at Cowen about long-term plan for ENTO and LANRA and how we position them with our development strategy. Maybe, Jorge, if you want to talk a little bit more here about maybe if there's a potential for LANRA to move into the first line fit population and how we're viewing both of the investigational therapies. Yeah. I think we sort of laid out the broad brush strokes here of the strategy, which is to start with the beachhead. It's basically taking advantage of the available clinical data with ENTO to launch that in the frontline fit setting as quickly as possible. Also a kind of the next step opportunities combining with VenAZA in the less fit population and then starting to explore combinations of targeted agents in the relapsed refractory setting with the goal of hopefully eventually being able to move those into the frontline setting. Now, if the question is about LANRA sort of displacing ENTO in the frontline fit setting, right now, we don't have any plans to pursue that. However, we'll continue to look at what the opportunities are. Great. Another question from the audience, maybe this is for Eytan and Jorge. If we know the HOXA9/MEIS1 expression profile in FLT3 and MLL-r AML. Certainly in MLL-r. That was always part of the defining part of the gene expression profile for MLL rearranged AML is high HOXA9/MEIS. In FLT3 mutated AML, it's co-mutated with either MLL or NPM1. I would expect to see high HOXA9/MEIS end up with FLT3 alone. FLT3 with MLL-r, NPM1 or DNMT3A, which I think covers a lot of FLT3, we would expect to see high HOXA9/MEIS. FLT3 without any of those co-mutations, I don't know that I've seen data specifically. Yeah. If you just have a FLT3, I've not seen data about HOXA9/MEIS1 in that setting. Great. A question here probably for Jorge. Can you talk about, again, just to summarize the different characteristics of ENTO and LANRA and some of the maybe safety profiles of each product? Sure. We touched a little bit on that in the presentation. Again, both single digit nanomolar against SYK and-- One key difference, ENTO also seems to hit FLT3 within a little over twofold of the IC50, biochemical IC50 for SYK. LANRA does not. LANRA's next closer sort of off-target would be ZAP70, and then both hit JAK2 and Src 10 to 15-fold above their SYK IC50. Other than that, not a ton of off targets. The other key difference is half-life. The half-life of ENTO really requires twice-daily dosing to achieve adequate target coverage. LANRA has a longer half-life and is suitable for once-daily dosing. ENTO has pH-dependent absorption, and so it can't be dosed concurrently with proton pump inhibitors. They will reduce exposure. Patients can continue to take H2 blockers and then Carafate to manage gastritis, but PPIs would be prohibited. There's a food effect where in a fed state reduces exposure to ENTO. LANRA doesn't have any of those features, so it's perfectly fine to take proton pump inhibitors, fed or fasted. I think those are the main differences in terms of the chemical or the pharmacologic properties of the two drugs. Great. Maybe one last question before we head into a short break. Are any NPM1 or associated mutations also present in high-risk MDS? Maybe Dr. Stein and Jorge can take this. Sure, I can start. We do see NPM1 mutations in myelodysplastic syndromes. They're not as common as in patients with acute myeloid leukemia. There are certain other mutations that can occur in high-risk MDS. If the question is, I think that certainly if there's activity of ENTO in an NPM mutant AML population, I would expect that in a high-risk MDS population with an NPM1 mutation, you would see something similar, although it's not super common. Yeah, I agree. I think the number I've seen is somewhere around 3% of MDS will have NPM1 mutation. Great. All right. With that, we will go ahead into a short 10-minute break, then come back, and Jorge will talk about our work with our CDK9 inhibitor, KB-0742. I'd also like to thank Dr. Stein for being with us this morning or this afternoon and presenting on AML. It was really great to have you. Thank you, Dr. Stein. Thank you. Thank you Eytan. [Break] Welcome back from the break, everybody. I'm Jorge DiMartino, Chief Medical Officer at Kronos Bio, and over the next 30 minutes, I'm going to walk you through our CDK9 inhibitor program. This is the overview about the presentation. In the first part, I will share with you just briefly some data showing that tumors that are highly dependent on the oncogene MYC have a vulnerability to CDK9 inhibition. I'll talk a little bit about our clinical candidate, KB-0742, and its properties that we feel should be advantageous for its clinical development. Finally, I'll touch a bit on the ongoing phase I trial with KB-0742 that is looking at the safety pharmacokinetics and pharmacodynamics of this molecule in cancer patients. MYC is a bit of an unusual oncogene. It was one of the first oncogenes to be discovered, and it is seldom actually mutated in the way that other oncogenes are mutated. MYC becomes deregulated through overexpression, and that's illustrated on this slide. Again, like most oncogenes, MYC has a normal function in cell biology, and at steady state, it drives expression of critical housekeeping genes. It can be transiently upregulated during responses such as wound healing when cells need to divide more rapidly. However, in these physiological settings, it's normally rapidly downregulated, so you just get a burst of MYC activity. The way MYC is deregulated in tumor cells basically is the cell finds ways of producing very, very high levels of MYC persistently. It goes up to super physiologic levels, and it just stays there. The effect that this has is that you see distribution of MYC to super enhancers that represent typically lower affinity binding sites that are not typically bound during normal cell function or only transiently during wound healing. In this case, these are locked in the on state, they drive cell growth and survival. This ends up creating a vulnerability to CDK9, that's illustrated on this slide. The vulnerability comes from sort of a double whammy, if you will. MYC is a very short half-life mRNA. It needs to be actively transcribed all the time just to stay at the same steady-state level. For this, it requires CDK9. CDK9 binds to the MYC super enhancers along with a number of other factors. We found that even intermittent partial inhibition of CDK9 can cause MYC mRNA levels to drop. Beyond this, MYC is a transcription factor itself, and it also requires CDK9 to carry out its function of driving expression of genes that support cell growth and survival. It's this double whammy that we speak of as a vulnerability to CDK9 for tumors that are addicted to super physiologic levels of MYC. The consequences of this are that it should be possible to inhibit CDK9 partially and intermittently. This is different from the way that other kinases are targeted, where you're trying to inhibit it almost completely every day. With CDK9, if you were to do that type of dosing paradigm, you would end up with a lot of toxicity. We feel that this is not necessary for the anti-tumor effects of CDK9 and, in fact, that it's not desirable. This has significant implications for the way that we're planning to dose CDK9 and for the selection of patients that we're planning to pursue with KB-0742. KB-0742 came off of our internal discovery platform, the small molecule microarray, that you will hear more about in the next section from Charles and Chris. Suffice it to say that what this platform allows us to do is to detect binders to transcription factors of interest or their associated complex proteins. In this case, this screen for interactors with the androgen receptor complex identified a hit that bound to CDK9. This is a cofactor of the androgen receptor. What made this hit very interesting to us was its degree of selectivity for CDK9 over other CDK isoforms. We think this will be important, as I'll show you in the next slide, in terms of helping to achieve a therapeutic index. The molecule, the original hit, also had very nice drug-like properties. It was small, it looked to be orally bioavailable, our medicinal chemistry team was able to evolve it into a clinical candidate in a fairly efficient 18 months with fewer than 300 analogs synthesized to arrive at KB-0742, a molecule that remains highly selective for CDK9 over other CDK isoforms and other targets that has oral bioavailability and some properties that we think will be favorable for its clinical development. I'll go into that in the next few slides. This slide shows the selectivity profile of KB-0742. It is single-digit nanomolar IC50 for CDK9, at least 100-fold selective for most of the other CDK isoforms. Very importantly, selective against CDK4 and CDK6, which are cell cycle CDKs. CDK9 inhibitors that end up engaging these other off-targets, I think, have to contend with the toxicity that comes along with inhibiting CDK4 and CDK6. The selectivity and the avoidance of the cell cycle CDKs, I think will help us to avoid these off-target toxicities. As I mentioned, the molecule has PK properties that we think will be favorable for its clinical development. What we know about the molecule in terms of its PK properties now is strictly from preclinical species and from in vitro studies such as microsome and hepatocyte PREPs. To summarize, what these studies tell us is that we think as much as 75% of an oral dose will enter into the bloodstream, that the molecule is very stable, it's low turnover in these microsome and hepatocyte PREPs that typically look at how stable the molecule is in plasma. It seems to have a relatively low projected clearance and a moderate to high volume of distribution. What all of this means taken together is that the molecule is likely to have a fairly long projected human plasma half-life. This is important because it really fits into the dosing scheme that we think we need to achieve to really maximize the therapeutic index of the drug. As I'll go into later, we think that the key to hitting CDK9 is intermittent dosing, but with intermittent dosing, having a long plasma half-life allows you to achieve accumulation over a dosing period of three days, in this case. What that lets you do is it gives you this nice smooth entry into this key therapeutic window where you're above the threshold that's needed for efficacy based on preclinical models, but where you're avoiding going above this red line, where you may start to run into off-target toxicity. Again, we think that this is really going to be helpful as we dose escalate in order to define therapeutic exposure while avoiding on-target and off-target toxicity. As I've mentioned, intermittent dosing we think is critical. We don't think continuous dosing is necessary, and that's supported by some models that we've worked with, in particular the MV-4-11, which has been our workhorse model for the anti-tumor effects of CDK9 inhibition. This is a MYC-driven AML xenograft model. In a nutshell, what we've been able to find here is that when we dose the drug for three days out of the week, three days on, four days off, as long as we're delivering a similar total dose that would otherwise be delivered over seven days on daily dosing period, we're getting essentially equivalent tumor growth control. As I'll show you later, avoiding many of the toxicities that come along with daily dosing. What these workhorse models also allow us to do is to measure the degree of target inhibition that is associated with anti-tumor effects. In this case, we can do it by looking at phosphoserine II on RNA polymerase II. This is the mark that CDK9 catalyzes on RNA polymerase II. We can also look at things like MYC downregulation and MCL1 downregulation. Suffice it to say that at the efficacious dose of 60 mg/ kg, we see target engagement of over 50% that lasts for at least 12 hours and starts to come back up sometime between 12 and 24 hours. Again, we do think that having a sustained period of MYC attenuation is what's critical for efficacy here, as opposed to, say, shorter duration, deeper MYC modulation. It's the sustained inhibition that's important. We have developed assays that we can use in the clinic in the peripheral blood mononuclear cells of patients that are being dosed with CDK9 inhibitor KB-0742 as we dose escalate. This slide shows the results of some ex vivo studies that were done using peripheral blood mononuclear cells from healthy donors that were exposed to KB-0742 in vitro, and we are able to measure very robust downregulation of the phosphoserine II mark on RNA polymerase II. In addition, we can look at gene expression signatures dependent on CDK9, and this is a custom gene expression signature that is measured on a NanoString platform. When we treat peripheral blood mononuclear cells from healthy donors ex vivo, we can see reproducibly reduction in this gene expression signature. These are the assays that will be used as we dose escalate in the clinic to measure target engagement. In terms of selection of tumor types to look for efficacy, this is where our preclinical work using primary patient-derived cell lines and patient-derived organoids becomes very relevant. We recently presented these data in a poster form at AACR, but just to recapitulate here, we see significant growth inhibitory effects across a broad swath of patient-derived tumor cell lines. These are primary cell lines, not immortalized. They haven't been in culture for a long time. These are from patients who have received multiple lines of therapy. In particular, lines from patients with breast and lung cancer show activity in the cytotoxic range. This also applies to patient-derived organoids. These are cells from patient tumors that grow as three-dimensional organoids in culture, in this case, derived from patients with small cell lung cancer on the top and triple-negative breast, the last two rows along the bottom. Some of these patients are treatment-naive, but a number of them have gone through treatment and relapsed after standard of care regimens with etoposide and cisplatin. As you can see in the green lines, many of these organoids no longer respond or respond very poorly to these standard of care agents. However, they do respond quite nicely to KB-0742 when treated ex vivo. This is also seen for the triple-negative breast lines. We also see these types of responses with in vivo models. These models are patient-derived xenograft models of small cell lung cancer and triple-negative breast cancer, where dosing KB-0742 at the 60 mg/kg dose on a three day on, four day off schedule achieves tumor regressions in the case of the small cell lung cancer model, comparable to what is seen with the standard of care cisplatin and etoposide. Again, very good tumor growth inhibition in this triple-negative breast cancer model, again, comparable to what is seen with standard of care agents. In terms of trying to narrow down the focus and really selecting tumors that are dependent on high levels of MYC, we've found that copy number amplification is something that tracks very strongly with sensitivity to KB-0742. These data are from the Broad PRISM cell line panel. This is a panel that contains close to 800 cell lines from various solid tumor types and looks at growth inhibition in a high-throughput format. What you can see here is that when you look at all solid tumor types that are within this array, you see a greater sensitivity or lower IC50 with KB-0742 in the models that have extra copies of MYC in the genome. This is seen more strongly as you narrow down the tumor types to all lung cancer, and in particular, non-small cell lung cancer, where we see the biggest difference in sensitivity between cell lines that have extra copies of MYC genome and cell lines that have the standard two copies of MYC. There is a trend with the small cell lung cancer lines. There are fewer of these, and this doesn't achieve statistical significance. MYC copy number gain or amplification, as it's known, does lend itself as a potential platform for patient selection. This is because this is something that's detected by next-gen sequencing, and it's relatively common. A publication from a few years ago looking at all the tumors in the TCGA dataset found that on the order of 30% of them will have extra copies of MYC genome, anywhere from a handful of extra copies to dozens or more, and these include many of the common tumor types today that require new treatment options. This is a result that is reported out by commercially available tumor sequencing platforms, for example, the FoundationOne Platform and the Tempus platform that are in clinical use today, so we do have access to these data as we enroll patients. It does correlate well with the overexpression of MYC RNA and protein. As I mentioned earlier, the intermittent dosing schedule that we've landed on, the three days on, four days off, which delivers really good tumor control in preclinical models comparable to what we see with daily dosing, is much more well-tolerated. Our non-GLP tox that we did prior to the IND-enabling tox, the dose range find was on the seven day continuous dosing or 14-day continuous dosing schedule. At that schedule, we found fairly significant effects on neutropenia and GI tox. For our IND-enabling tox, we went to the three day on, four day off schedule that was shown to be active in preclinical models. With this schedule, we were able to dose up to 2 mg/ kg in dogs, 10 mg/kg in rats, to achieve fairly good exposures. These exposures did result in some GI tox and some bone marrow tox with red blood cells and neutrophils. However, the toxicology reviewers viewed this as non-severe toxicity. This allowed us to establish a no adverse effect level and to arrive at a human starting dose of 10 mg, flat dosing, three days on, four days off, which provides a fairly good margin over the no adverse event level. This is the phase I/II trial that started enrolling in February of this year. We're currently dosing with the three day on, four day off schedule weekly. We plan to conduct a PK and PD analysis after enrollment of the second cohort, and on the basis of this, we'll continue to escalate either on the three day on, four day off schedule, or potentially consider alternative intermittent schedules, all with the goal of getting to a selected phase II dose and schedule at the end of this stage. This is enrolling relapse and refractory solid tumor patients and non-Hodgkin's lymphoma patients without specifically selecting for MYC overexpression or other biomarkers. The goals of this part of the study are really about understanding the safety, the PK and PD, and I'll elaborate on that in the next slide. Our plan, once we've defined that recommended phase II dose and schedule, is to go into these expansion cohorts, which I'll also go into more detail on subsequently. This is where we're really expecting to start to see anti-tumor effects in selected patient populations if our hypothesis holds. This illustrates the PK and PD analysis that we hope to be reporting on by Q4 of this year. We'll conduct it after the first two dose level cohorts have completed evaluation. This should be approximately six patients. We're going to be looking at the plasma PK. I've mentioned that we think that having this type of a PK profile with a long plasma half-life and accumulation is really going to be critical for being able to achieve a therapeutic index. Up until now, we have really the preclinical data to base that on. It's going to be really key for us to look at that clinical PK in patients and see what that looks like and see if we're in the ballpark of where we hope to be. We're going to be looking at target engagement markers in peripheral blood mononuclear cells. Again, the RNA polymerase II phosphoserine II, which we have assays for, and this CDK9-dependent gene expression signature in peripheral blood mononuclear cells. Very importantly, we're going to be looking at the safety. Our hypothesis states that if we're able to achieve this sort of exposure profile and start to see target engagement activity with a manageable safety profile, then we're well on our way to defining this recommended phase II dose and schedule. At that point in Q4, these data should help us to address some of the questions shown here. Again, does the clinical PK look like what we expected to see based on our preclinical data? Are we observing evidence of CDK9 inhibition, so target engagement in the peripheral blood mononuclear cells? What does the incidence and severity of adverse events, specifically neutropenia, GI tox, fatigue, at these dose levels, does it seem to be manageable? Is the four-day off period enough to allow recovery from any neutropenia that we see? Finally, based on these data, do we continue escalating with three days on, four days off? Based on the PK that we're seeing in the clinic, do we explore different intermittent dosing schedules? If we were to do this, we would do it in parallel rather than serially. Once that phase II dose and schedule is defined, we would initiate enrollment into the expansion cohorts, where we would anticipate seeing responses in some of these sensitive patient populations. Cohort A is really going to focus on high MYC as an indicator of transcriptional addiction or CDK9 dependency. I mentioned earlier, high MYC can be defined based on MYC copy number, so extra copies of MYC in the genome, which typically are associated with high levels of MYC RNA and high levels of MYC protein. We will have a patient selection assay in place for initiation of this expansion cohort that will select for MYC-high patients. We'll be looking at multiple ways of defining high MYC. We've been collaborating with a company called Tempus that has access to genomic and transcriptomic data from real-world evidence to really understand the epidemiology of these three different measures of being high MYC. We hope to understand which of these best associates with response to KB-0742. In the second cohort, we're going to be looking at a basket of less common tumor types that have other biological hallmarks of being transcriptionally addicted and therefore dependent on CDK9. This includes tumors such as soft tissue sarcomas, of which there are several different types. Many of these have as their founder mutation, a chimeric transcription factor such as EWSR1 or PAX-FOXO1. There's actually quite an extensive list of these, even though they're fairly rare collectively, there are a number of them. Again, the fact that these tumors have a chimeric transcription factor, so a deregulated transcription factor as their founder mutation, sometimes few or no other mutations suggest that transcription was really key for the growth and survival of these tumors, and therefore inhibiting CDK9 may have an anti-tumor effect. Another example of tumors in this basket includes chordoma, this is again, a relatively rare tumor that can occur in the midline. It has no really curative therapies available at present. Our Senior VP of Biology, Charles Lin, whom you'll hear from in a few minutes, identified in chordoma that there's a dependency on a transcription factor called Brachyury. This is a transcription factor that's normally expressed during embryonic development, but not postnatally, except in these tumors where it persistently stays high. What Charles showed is that this dependency on Brachyury translates into a dependency on CDK9, as shown with several CDK9 inhibiting tool compounds. With these rare tumor types, if we do see objective responses that are durable, the high degree of unmet medical need here could potentially position KB-0742 for an accelerated approval pathway. In conclusion, I think I've shown that we're interested in targeting dysregulated MYC. This is implicated in approximately 30% of various common epithelial solid tumors with MYC copy number gain, and it creates a dependence on CDK9, which we think we can exploit therapeutically. We're planning to do this with KB-0742, a CDK9 inhibitor that was discovered internally using our small molecule microarray screening platform. The unique combination of selectivity for CDK9 and favorable pharmacologic properties, I think, positions this molecule well for the challenge of demonstrating a therapeutic index and benefit from inhibiting CDK9. Our strategy is in two parts. The first part is to show that we can, in fact, achieve a therapeutic index with an intermittent dosing schedule. We hope to be reporting out on that at the end of this year in Q4. The second part following that will be then to take that intermittent dosing schedule into tumors that are selected for sensitivity to CDK9 and determine whether we see, in fact, objective responses. We'll report initially on this in Q4 of this year and subsequently, if we're able to achieve that, to develop the drug in common tumor types as well as rare cancers that have transcriptional addiction. That's the end of my piece. I will hand you over now to our CSO, Chris Dinsmore, and our Senior VP of Biology, Charles Lin, who will talk to you about our discovery platforms. Thank you, Jorge, for the introduction. It's my pleasure to talk to you today about Kronos' discovery product engine. My name is Charles Lin. I'm the SVP of Biology, and I'm a computational biologist by training. I've spent most of my career in academia really studying and thinking about how to target transcription factors in cancer. I think the why of transcription factors is pretty clear. They've advertised themselves as some of the most important genes in human disease and especially cancer. We've known this for decades. Look no further than MYC, the most commonly amplified oncogene, or P53, the most commonly lost tumor suppressor. These are both transcription factors. Transcription factors in general have been really difficult to drug. That being said, when we've successfully drugged them, whether it's historically with our first efforts against nuclear hormone receptors, which are a special kind of transcription factor that have a druggable pocket, or more recently with the IMiDs, which degrade transcription factors in myeloma, the benefit to patients has been immense. There's a real clear need and understanding of why we should drug transcription factors. Why haven't we been able to do it? It's not for lack of trying. Folks have been trying since the early 2000s, and to date there's really not been a successful systematic effort to really access this target class. Kronos hopes to change this, and we recognize all the work that's been done in the past and are hoping to change this by bringing forward new approaches to solve the problem. For us, fundamentally, this starts with recognizing what exactly is the problem we're trying to solve. At Kronos, we really break it down into these three context-dependent challenges. We think overcoming these challenges is critical to any effort to drug transcription factors. First and foremost, transcription factors have context-dependent activity. What this means simply is you can't just recapitulate their activity in isolation on a test tube. They're not like a kinase, where you can develop a biochemical kinase assay. Rather, transcription factors bind to discrete spots in the genome, and there they regulate nearby genes, and they do so in a very programmatic fashion. You have to really understand and measure that program if you really want to capture the true activity of a transcription factor. Second, transcription factors are the classic case of what we call an intrinsic disordered protein. They lack a defined structure in vitro or in a test tube. It's only in the nucleus of the cell with their appropriate co-factors that do adopt a highly dynamic and context-dependent structure. What this really means is that, again, you cannot put a transcription factor in a test tube and use the traditional drug discovery playbook. The classic biophysical assays or structure-based assays are just not going to work. Finally, I always liken transcription factors to a general contractor. They'll show up to a job site and hold a clipboard, but unless you know the machines that are there, you don't know if you're building a house or a highway. These are transcription factors in a nutshell. They bind to the genome, but they fundamentally function through the recruitment of complexes, other molecular machines that will actually act on and drive the transcription cycle. Here's an example to really hammer this home. Here I want to contrast a classic druggable protein, ALK, on the left, and our favorite transcription factor, MYC. With ALK, there's a clear structure function relationship. It's a kinase. It has an ATP binding pocket that's ligandable, and there are established in vitro activity assays. You can make biochemical kinase assays. Finally, because it's a kinase and because you're trying to drug a kinase, you can understand your off-targets, your selectivity profile, by simply profiling other kinases. MYC, on the other hand, a classic transcription factor, and many of the same problems that occur with MYC occur with other transcription factors. That structure I showed you on the previous slide, that only accounts for about 20% of the MYC protein. The remaining 80% is intrinsically disordered. We can't get a crystal structure of it. Unfortunately, it's this 80% that's really the business end of MYC. It's this part of the protein that acts by recruiting numerous co-factors like CDK9 to the MYC genomic binding sites. Now, all of this happens without any sort of active site. Again, this is not like a kinase or receptor. There's no intuitive place on the protein where we can stick a small molecule and expect a drastic change in MYC function. As a result of this, there really aren't great in vitro activity assays. We can't put MYC in a test tube and recapitulate what it'll do in a cancer cell. When we do work on it in a cancer cell context, the assays are confounded by the difficulty in distinguishing MYC specific versus global effects. MYC is tied up into so many processes in the cell that it's often hard to know if you're hitting one of those processes directly or you're hitting MYC. This is just one example of why it's so critically important to solve these context-dependent challenges if you're actually going to systematically approach transcription factors as a target class. At Kronos, our product engine really seeks to solve these challenges and ultimately to translate these therapies in patients with dysregulated transcription factors in their cancers. We break this up into four main parts: map, screen, optimize, and validate. I'll touch on how each of these parts really serves to address one or more of the challenges that we face when thinking about translating transcription factor therapies. The first is our ability and the importance of mapping oncogenic transcription regulatory networks. What this really says is, look, we need to take a step back when it comes to transcription factors. We can't just drug them in isolation because we have to understand their context. We call this the transcription regulatory network approach. Fundamentally what it means is that we understand a transcription factor's place in the cell. We understand how signaling communicates to it. We understand its co-factors, how it's post-translationally modified. We understand where it sits on the genome with chromatin modifications and other transcription factors. Ultimately, by perturbing these networks, we understand how this transcription factor and others form a regulatory logic that reprograms the gene expression landscape of a normal cell to a tumor. By taking a step back and mapping TRNs, we really have a much better understanding of what the true activity of a transcription factor is and what its complex is, who it's bringing to different parts of the genome. I want to make this more concrete, and for the rest of my talk, I'll try to give you two clear examples of how this mapping is really informing and driving our discovery platform and product engine here at Kronos. The first is in thinking about targeting transcription factor dysregulation in small cell lung cancer. Small cell lung cancer is a deadly disease with a huge unmet patient burden. What I think makes it so unique is that it's almost 100% a cancer driven by transcriptional deregulation. Almost all cases are characterized by loss of P53, a transcription factor, and RB1, a transcriptional co-regulator. Within small cell lung cancers, we've since learned that the heterogeneity of small cell lung cancers is also defined by transcription factor deregulation. In fact, there are four really clear molecular subtypes of small cell lung cancer, each of them driven by a distinct oncogenic TRN named after the master transcription factor that's really at the heart of the TRN in each case. ASCL1, NEUROD1, POU2F3, and YAP1. With small cell lung cancer and other cancers, what we've since learned is that this heterogeneity is critically important. We know that across tumors, there's heterogeneity of TRN subtypes. We also know that within tumors, there's heterogeneity. This is important both for patient selection and understanding exactly what kind of cancer a patient has, and also for more practical things that power our drug discovery, like making sure our cancer cell lines and other model systems reflect and align to one of these TRN states. From a therapeutic perspective, it's clear and easy to understand why understanding and assessing your ability to target different TRN states might have an impact. Just take a look in this case of patient 10, a small cell lung cancer tumor that's defined by both cells of the ASCL1 or A subtype and cells of the NEUROD1 or N subtype. It's easy to imagine that a therapy that only targets one of these cell types is not going to be sufficient to really drive a deep response in these patients. This is especially going to be true if one of these subtypes is in general more therapy resistant or underlies and drives relapse. We think this is the case for this A subtype. It's named A after ASCL1, a key transcription factor that drives its identity. In general, these tumors have a very neuroendocrine-like identity. Sure enough, when we look at how compounds fare against this A subtype, you can see here that they uniformly perform worse than against the other cell types. Here the circles represent the potency of the drug on cell lines of a given type, with greener, yellow, bigger circles indicating a bigger effect. Across these 20 inhibitors, the A subtype is uniformly more resistant. We actually see this with our own compounds. Here is our CDK9 inhibitor KB-0742. Similar to the previous slide, although we're seeing really great efficacy in the N, P and Y subtypes, we don't get much of an effect on the A subtype. This is important because we actually really think these kinds of compounds will work. In vivo preclinical models, our CDK9 inhibitor actually does really great against a P and an N subtype. Although this is encouraging and motivates our exploration of KB-0742 in small cell lung cancer, it also leaves the A subtype as a really critical tumor subtype that we're going to have to deal with if we want to drive deep responses in these patients. How do we actually do this and how does mapping of TRNs really drive this forward? First and foremost, if we can map the TRNs, then we can know the heterogeneity of patient tumor samples as well as the model systems that we have in the lab. Now we're able to do this at both the population level and the single-cell level. By mapping and assigning different populations to TRN states, we can also understand the differences between TRN states. This is an approach called pseudotime that allows us to explore the transition states between the A and the N subtypes. We can figure out which cells are closer to one versus other and what are the molecular signatures that actually drive that transition. I'll show this to you here on this slide where we can actually now identify the key genes that drive this transition as well as the broader on the right signatures that are associated with each of these states. This is really important for two reasons. One, it allows us again to better classify and identify patient samples. Two, from a drug discovery perspective, it actually arms us with the molecular signatures that we need to know in order to, for instance, see if we're targeting the A versus the N subtype. To summarize this section, we know that small cell lung cancer tumors exhibit intra and intra-tumoral heterogeneity, and this is really driven by a number of dysregulated TRN subtypes. We know that we have to think about and account for some of these difficult subtypes, and in this case in small cell lung cancer, the A subtype driven by ASCL1. Our CDK9 inhibitor KB-0742 does show great preclinical efficacy in the N, P and Y TRN subtypes and we're exploring this further in preclinical models. Going back to our ability to discover and to attack these subtypes, it's really critical that mapping of TRNs power this. By mapping TRN subtypes, we can better identify sensitive or resistant tumor types. We can understand what happens when you perturb a tumor cell or treat a tumor cell. We can go beyond the very simple measurements of viability or apoptosis to now really understanding how we're affecting distinct cell states. Ultimately, we can use that in a campaign to prioritize small molecules that exhibit specificity towards targeting one cell state versus the other. This exact concept is something that I want to focus on in our next section. Here, I'd like to tell you a little bit about our efforts to identify selective modulators of the AR-driven transcription program in castration-resistant prostate cancer. This is also a great place to introduce our hit discovery platform, the Small Molecule Microarray. This is the platform that drives hit discovery at Kronos. It was developed by Angela Koehler at the MIT and Broad Institute, and she spent her career in academia really pioneering this platform, validating it, and demonstrating its ability to find really interesting differentiated hits. Angela's our academic founder, and at Kronos, we've taken this platform and really continuously developed it into a product engine that we're really proud of. Fundamentally, it's a small molecule binding platform. It looks for small molecules that stick to proteins. Where it's differentiated and uniquely suited to the transcription factor problem is its ability to operate in the right context. Again, as we said in the beginning, context is everything when it comes to transcription factors. In this case, the context is the tumor cell nucleus. We can take tumor cell nuclei, carefully lyse them, and present the protein complexes within to the Small Molecule Microarray. When it comes to transcription factors, this means that we're presenting the transcription factor in the appropriate context-dependent structure. We're also presenting it with its appropriate complex. It's still stuck to all of its cofactors. We look for binders to this transcription factor complex. What this fundamentally means is that this is a more unbiased approach, and it allows us to find a variety of distinct hits, each of which might be attractive for follow-on. Some of these are described on the right. We can find direct interactors to a transcription factor when it's properly structured. We can find protein-protein interaction modulators. In some cases, we can find small molecules that hit and inhibit more classically druggable cofactors, like CDK9. When we do so, this actually gives us a really rapid path for forward development since we know the playbook of how to go after a druggable target. Because we can actually hit the transcription factor complex through all these distinct mechanisms, it creates a next challenge for us. How do we distinguish the hits we want from the hits we don't want? Fundamentally, what this means is how do we find the hits that are hitting the transcription factor in a selective way versus those hits that are hitting the basal transcription machinery? We know we don't want to go down the path of globally drugging transcription. We think that's likely going to be toxic. I'll show you one example of this in the context of castration-resistant prostate cancer. Now, prostate cancer is really one of the great success stories for drugging a transcription factor. The androgen receptor is itself a transcription factor. It just happens to have a ligand binding domain on it, and this is where the generation of drugs like enzalutamide have really engaged and taken advantage of. Unfortunately, a lot of these cases, cancers will relapse, and they'll relapse in a way where they become insensitive to our current generation of androgen receptor-targeting drugs. One of the ways they do this is actually by losing that druggable ligand binding domain, creating these so-called AR variants that have all of the same capabilities as the androgen receptor but are no longer druggable. At Kronos, we've said, well, what if we use the Small Molecule Microarray approach to target these androgen receptor variants as a way to get at androgen receptor-dependent transcription independent of the ligand binding domain? We know these cancers are still utterly addicted to the activity of the androgen receptor, and we're trying to find a new path to the waterfall. To do so, we again went back to this idea of mapping a Transcriptional Regulatory Network. We used proteomics to define the protein-protein interaction network of the androgen receptor variant. We integrated this with multi-omic mapping, especially chromatin mapping, to figure out where these complexes likely sit on the genome. This helps us define what genes and what parts of the genome are regulated by the androgen receptor variant. By perturbing this network and degrading different factors, we can actually figure out a response signature, a set of genes that are very selective to androgen receptor activity. This allows us to figure out if we're on target or not. Now, I want to show you how this actually works in practice by looking at some of the hits that came from our androgen receptor variant SMM screen. On the left, I'm showing you a very classic traditional way of prioritizing hits. In this case, with a reporter assay that really prioritizes hits just based off of their potency, in this case, their ability to downregulate PSA. I want to call out three hits, X, Y, and Z. X and Y are very potent, and you'd probably prioritize these in a normal campaign. Z, not so much. If you look at these same three hits, now through a richer lens of gene expression signature profile, you see a much clearer picture. Hit X looks pretty good. It moves the genes up that we expect to move up, it moves the genes down that we expect to go down. Hit Y, which is equipotent, you can clearly see is non-selective. It's not really changing gene expression the way we would imagine from degrading the androgen receptor. What I think is even more important is our ability to actually prioritize things like hit Z in the first place, something that would've been lost from a traditional campaign. Hit Z isn't very potent in the reporter assay on the left, as you can see by our signature profiling, it's actually exquisitely selective. It moves all the genes in the right way. This is really exciting to us because in general, we think it's easier to improve potency than it is to improve selectivity. You can look at these three hits, each with very different characteristics, and wonder how is it that you're targeting the same protein. The short answer is you're not. Again, our SMM really looks for hits to the transcription factor complex. Now that we have this ability to prioritize hits by how selectively they affect transcription, it actually gives us a really powerful product engine, one that can actually spawn multiple distinct programs. Take an example here where we have compounds hitting X and Y in the basal machinery. If we know that cofactors X or the protein interaction with Y are really selective to a transcription factor, then we can prioritize these as distinct lead programs. On the other hand, the kinds of hits that are hitting the basal transcription machinery, we can really quickly triage out. This actually gives us the ability to take oncogenic TRN like AR and really come at it through a variety of distinct and orthogonal approaches, each potentially with a good likelihood of success. Now I want to hand it over to our CSO, Chris Dinsmore, who's going to tell you even more about our SMM platform and how we've continued to develop and refine it over the last year and a half. Hi, I'm Chris Dinsmore, CSO at Kronos Bio. In this section, I'll continue the discussion of our discovery pipeline by briefly touching on platform development and ending with an outline of our current pipeline focus. Charles has just reviewed the challenges of targeting transcription factors. He also described how we address those challenges by systematically targeting Transcriptional Regulatory Networks or TRNs. Our approaches in mapping and screening open up the targetable space beyond just the transcription factor itself and can lead to multiple projects that ultimately target transcription factor activity. Our Small Molecule Microarray screening platform, or SMM, is a key part of this product engine. We've considered it essential to continuously optimize its capabilities. I'll expand on that in the next several slides. We focus on several technical aspects of the SMM platform for continuous development, each contributing to what in the end amounts to high screening efficiency, excellent data quality, and rapid workflow. This gives us more freedom to use the platform iteratively and in multiple relevant cellular contexts simultaneously for a given TRN. A crucial part of this is the preparation and characterization of the sample being screened, as described on the left side of the slide. Nuclear lysates are extracted and isolated to preserve intact transcriptional complexes. Getting conditions right enables us to avoid non-relevant cytosolic complexes, thereby reducing background effects and providing results with greater signal-to-noise. Further, we characterize the TRN complex within that nuclear extract, both computationally and experimentally to further define the actual screening context. This can give us an idea of the actual set of proteins and interactions that exist in the sample to be screened with SMM. As an important quality check, pre-screen validation is done using positive control agents, ensuring the integrity of some of the expected molecular interactions and complexes within the lysate before the screen is actually conducted. Certain key features of the microarray itself are described on the right. The microarray currently features a diversity library of almost 250,000 lead compounds. This compound collection was sourced and computationally curated to ensure an attractive range of physicochemical properties and molecular features so that each screening hit is a good starting point for a project. In other words, each member of the library is a molecule we would want to work on as a lead for a medicinal chemistry campaign. The library is printed on slides in a spatially defined array through covalent tethering from the microarray slide to each library compound at various types of functional groups and positions. As for the microarray slides themselves, we've adjusted the process for manufacture and implemented automation, which optimizes for more consistent and regular surface chemistry. Because of this, we get good signal-to-noise for the detection of binding events. Which is done by the use of fluorophore-labeled transcription factor targeting antibodies and image analysis. Finally, machine learning and informatics enhancements have enabled us to handle the image and hit picking data much more easily, getting us away from what used to be a time-consuming visual inspection. Now overall data analysis turnaround time is rapid. With each of these platform variables having been under recent development, our overall screening efficiency and quality have been enhanced substantially. One of the key features of this lysate-based TRN screening approach using SMM is the potential broad opportunity that the hit compounds can provide. As Charles mentioned earlier, the collection of hits that are deemed attractive based on their TRN gene expression signature profiles may be bound to multiple different proteins associated with transcription factor, but not just the transcription factor itself. Because of this, target deconvolution is needed to identify the mechanism of a given hit. We employ various follow-up approaches for target deconvolution, including activity profiling against panels of relevant protein target classes. Also a key aspect of the SMM is that chemoproteomic probe approaches are easily enabled by this platform. Here's why. Because we know that the compound was bound to the target while covalently linked to the SMM slide, inspection of the hit's parent chemical structure indicates the likely site of a covalent linker attachment, and thereby an exit vector from the compound while it was bound to its protein target. Knowing the placement of this exit vector, we can now easily design various chemoproteomic probe molecules that are configured for pull-down experiments, as shown in the lower left of the box, for photoaffinity experiments, and probes that cause induced proximity, such as for targeted protein degradation, as shown on the lower right. Each of these chemical probes can enable deconvolution of the TRN target for an SMM hit. In a new direction that we're currently developing, we believe the SMM can be leveraged for direct on-slide target deconvolution, as described here. In this new approach, a screening hit of interest is reprinted multiple times on the SMM slide, and this slide is used to recapture the target from the lysate that was used in the screen, enriching for higher target protein concentration. After capture, elution, trypsin digestion, and mass spectrometry-based protein ID, this should efficiently help us identify the TRN target of the hit. In doing this, we hope to compress further the time from hit selection to target deconvolution, and in turn, this will shorten the overall turnaround time for moving from mapping and screening onto the optimization phase of a project. To summarize our platform, the mapping and SMM screening components we've reviewed are essential for us to capture binders to proteins within a TRN, and to then select those that are appropriately functional as potential therapeutic agents. Screening of cell lysates accommodates a transcription factor's contextual dependencies around activity, structure, and protein complexation. We use post-screen workflows that are designed to identify modulators that are selectively functional, as opposed to being non-selective basal transcriptional modulators. We deploy chemoproteomic-enabled variants of the hits for target deconvolution. All of these elements combine into a differentiated product engine that addresses the historical challenges of drugging dysregulated transcription factors in oncology. These approaches as a foundation, our current pipeline focus is as shown on the last slide. We're pursuing four main areas in which we're screening various TRNs. In heme malignancies, we've set course on the HOXA9/MEIS1 TRN led by the SYK inhibitor program, as reviewed earlier. Now we're following that up with other TRNs such as MYB and IRF4. We're pursuing neuroendocrine tumors such as small cell lung cancer, as described earlier with the ASCL1 TRN, and we're also pursuing androgen receptor for prostate cancer. Right now, working up interesting SMM hits. Finally, we're screening the MYC TRN, following behind the CDK9 inhibitor KB-0742 program, with projects directed at alternative mechanisms for inhibiting MYC. With that, I'll now hand it over to Stephanie to moderate our next Q&A session. All right. Great. Thanks, everyone. We are headed into our second Q&A section. Just as a reminder for those who are with me, if you have a question, please type it into the Q&A chat box to be read. I'll start off. We've been getting some questions about selectivity of KB-0742 compared to maybe other CDK9 inhibitors. Maybe we can turn it over to Chris and Charles to just comment on different ways for measurement and just, again, selectivity that we're seeing with KB-0742. Yeah, sure. I can comment. The data we've shown measures the activity against the CDK kinases at approximate KM for ATP, and in this case, for CDK9, it's 10 micromolar of ATP. One would expect an IC50 shift upon much higher ATP concentrations versus physiological concentrations. That would be not unexpected. We've measured the selectivity relative to other kinases, other CDKs, at that lower ATP concentration, and for most kinases, most CDKs, was greater than 200-fold shift. If one measures the same selectivity at a higher concentration, your ability to assess that delta can collapse if you're pressing up against the limits of the assay. I think I wouldn't perceive the data as indicating a lesser selectivity. Greater than 20x selectivity is quite consistent with greater than 200-fold selectivity measured at lower concentration of ATP. Okay, great. We'll move on. I think we had a question from Andrea Tan at Goldman Sachs. A couple actually. How frequently are solid tumor patients screened for MYC amplification? When in the treatment paradigm would this happen? What would we consider manageable or acceptable adverse events that we would expect to see with our phase I trial? Yeah, I can take that. In terms of the MYC screen, I don't think anybody's screened explicitly for MYC amplification. What's becoming more and more common in practice now is for physicians to send out samples of their patient's tumor for sequencing, these are sequencing panels that look at a few dozen or up to a couple of hundred different target genes, with the idea of trying to identify whether there are actionable mutations that can result in treatment options. These platforms will commonly report out MYC amplification if it's detected and if it's within their algorithm to detect that. That is what I meant in terms of it is reported by the common sequencing panels that are out there right now. Depending on how rare the patient's tumor type is or where they're being seen, this may be done at initial diagnosis or at the time of relapse. In terms of the AE frequency, really the key one that I think has been a limitation for earlier programs has been neutropenia. Many of these programs have encountered Grade IV neutropenia lasting more than seven days. I think as a general rule of thumb, it's about both the depth and duration of neutropenia. I would think anything Grade II or better would be acceptable or, perhaps Grade III with a more limited duration, perhaps less than seven days. If you're seeing very frequent, more than a third of the patients with Grade IV or more severe neutropenia, that would probably give us some pause about advancing further. Great. On the other side, we're getting some questions about anticipated therapeutic window for KB-0742, and just maybe we can expand a little bit more about on that. How far apart are the threshold exposures for efficacy versus the exposures that cause neutropenia and GI tox? Have we seen any variability in the size of the therapeutic window along the various preclinical models that we've studied? That's always the very important question, and it's one that in practice is really difficult to address, right? We do all our efficacy modeling in mouse xenografts. The tox species are dog and rat, so the exposure toxicity relationship is figured out in different species than the efficacy exposures. Mice can tolerate lots of KB-0742, right? Whether we give it daily or three days on, four days off, the mice don't really seem to care. It's the tox species that respond to the schedule. With all those caveats, I will also add that we did not define in our GLP tox studies an HNSTD or highest non-severe toxicity dose or an STD 10. What we defined was a no adverse effect level, and we have, as I mentioned, at least a seven or eightfold margin, exposure margin over our starting dose. That means we could increase, dose escalate to seven or eightfold higher than that and still stay within what's considered not severely toxic, if things translated perfectly from the animal models. Essentially, that gets us to a starting dose. We feel like we've got a lot of room where we can escalate over the coming cohorts and we'll basically be defining that therapeutic index as we approach it. That's kind of the best answer I can give right now. Great. Along the lines of that, it sounds like we'll have around six patients worth of data to report when we share out in Q4. Can you describe what we'd like to see in terms of PK/PD? Maybe Jorge go through, again, some of the questions that we're seeing in cancer. In terms of PK, again, we want to see a half-life that will lead to accumulation. That would be anything on the order of 15, 20 hours plus as a terminal plasma half-life allowing for accumulation from day one to day two to day three. That's sort of what we're looking for in terms of the PK. Again, if the moderate to high volume of distribution holds up, I think that's also going to be a plus. That means the drug is getting into tissues preferentially and not just staying in the plasma compartment. Of course, that's exactly what we want in a drug that's being directed at treating solid tumors. In terms of the PD targets, again, we've sort of indicated that at efficacious exposures, we want to see something on the order of 12 hours of target suppression to around 50% or better. I think because these first two dose cohorts are relatively low doses, we may not necessarily be achieving that threshold yet after the first two dose levels. However, we're looking for consistent movement of the PD markers at those exposures, indicating that we're engaging the target. Great. A question about the specific tumor types that we're planning to study in the expansion cohort. Why did we select triple-negative breast cancer, non-small cell, and small cell lung cancer, and gastrointestinal cancer for the dependent cohort? No, again, it was sort of a combination of factors. Certainly, we're looking for tumor types that, based on the analysis of the TCGA data set, have a relatively high incidence of MYC copy number gain. This also partially overlaps with the models in which we've seen the greatest sensitivity. Small cell and triple negative have proven to be very sensitive models, both in patient-derived organoids and patient-derived xenograft models in association with MYC overexpression. Finally, also looking at the intersect, it's a sort of a Venn diagram, if you will, of high unmet need conditions where there's not a lot of different treatment options and where patients definitely need improved outcomes. A combination of those things. Great. How are we thinking about what we're hoping to see with KB-0742 in comparison to maybe other CDK9 inhibitors that have responses, for example, CRs in lymphoma? Is lymphoma something that we would branch into? Yeah. Lymphoma patients are eligible for the dose escalation part of the study. I'm not sure how many we're going to get. We're going to major centers. At some centers, the lymphoma patients are referred for solid tumor phase I trials. At other centers, they're not. That part of it is kind of an unknown. I think the biology around lymphoma is very compelling for MYC dependencies, certainly. I think the challenge there for the near term continues to be the ability to enroll substantial numbers of these patients into clinical trials. I think the advent of novel therapies such as cell therapies are really starting to limit the number of eligible patients with lymphoma that are available for trials. In terms of what we're expecting to see in comparison with those agents, as I said, I'm not aware of any other agents for which there's public information that combine both the selectivity profile of KB-0 742 with the favorable PK that we think is critical for achieving a therapeutic exposure in one molecule. Really what we're hoping to see by the end of the year is clinical data supporting that hypothesis. Great. All right, moving on to some questions about our platform. Maybe, Charles and Chris, if you can go over again the importance of mapping a TRN and why it's central to our discovery research and how our understanding of a TRN is important in terms of generating hits and potential programs. Absolutely. I always go back to this idea that transcription factors don't act alone, and you really have to understand the broader context. That really is what a TRN mapping exercise is all about, figuring out who's upstream, who's interacting with the transcription factor, and what's downstream. Now, from a therapeutic and discovery perspective, this has really two key advantages. One is it allows you to power better assays. We can understand, for instance, an example I gave, are we on target for a transcription factor, or are we just generally hitting transcription? The second, and I think this is really critical, is that when we talk about a campaign against a TRN, like an oncogene receptor variant, like MYC, what we're really trying to understand is how can we hit this network in ways that selectively block or inhibit that transcription factor's activity. This means that because we understand the structure of this network, there's probably not just a single interaction or a single target. There are multiple paths for me to follow, so to speak. I think this is really critical because this gives us a wider net to cast than simply going after the transcription factor directly. It allows us to quickly ascertain which nodes within a TRN are, so to speak, critical, especially not only in the cell-based context that we study and do drug discovery in but ultimately in the patient population. Chris, did you have anything you wanted to add to that? Sorry. Yeah. No, I think that's the power of this approach. It's to widen the scope of how we drug TRNs, and the platform is very well suited for that through the use of the SMM. We think it rolls into the possibility of affording novel starting points for druggable targets. Another question for Charles regarding MYC and whether or not the MYC IDR forms condensates. You want to address that, Charles? Yeah, absolutely. This is such an exciting area of research right now. It's not just MYC. Most of these transcription factors have large intrinsically disordered regions. It's long been sort of a conundrum, well, what are these doing? We know that they're recruiting specific co-factors, why are they so disordered? I think one of the emerging concepts that came out of some of the work from my old mentor's lab at MIT, Rick Young, is this idea that these transcription factors can effectively nucleate these transcriptional condensates. What's so unique about these transcriptional condensates is they become areas within the nucleus where essentially the transcriptional machinery is asymmetrically concentrated. We know that MYC does this, we know that Brachyury does this, we know that many transcription factors do this, not only that, we know that a lot of the key co-factors we're interested in, the CDK9s of the world, the bromodomains, are also parts of these transcriptional condensates. We think that the ability to form these condensates is really critical to drive the aberrant expression of these oncogenes in cancer. Certainly, there are numerous cases where there are prominent transcriptional condensates, such as the MYC amplicon itself, that are driving high levels of oncogene transcription. We think this is actually something that's unique to cancer in certain cases, or something that can be targeted, and it's something that we're definitely actively exploring. It does, though, present new ways of looking at the problem. Obviously, we can't look at the structure here. Even if we knew what was inside these condensates, we wouldn't know the structure because by definition they're very dynamic. We have to think carefully about how we connect genomic data, looking at, for instance, large enhancer loci, to the more classic [inaudible] times. It's certainly an area where we're trying to build the tools to be able to figure out how we can selectively target the transcription coming from condensates, as well as the formation of condensates themselves. Great. We have another question from there. In Chris's presentation, he noted the ability to potentially develop degraders. Is that something an area of interest for Kronos Bio, and what are your thoughts about degraders and the capabilities of the SMM platform? Yeah, that's a good question. The platform really holds the potential to address multiple different kinds of modalities for targeting any E3 ligase target, such as inhibition or activation or degradation. We see degraders not only as useful probes, which is the way I described them in this piece, but we also see them as a piece of potential therapeutics, especially where the targets are druggable in some other way as well. Since the SMM platform is very well behaved to identify components of a bifunctional molecule in the ligand and the linker from that which becomes ultimately a portion of the bifunctional molecule, it's a very advantageous approach for this. We actually have been working actively in this area with some of our targets within the categories that I showed at the end for our portfolio. These are not at the stage of being able to describe them just as yet, but we certainly do look forward to describing more specifically what we're doing in that space. Good. All right. A question about BE, noting that our platform has many capabilities, how are we thinking about fully leveraging that and in what kind of partnering? Thank you, Stephanie. It's a great question. As you clearly heard from both Charles and Chris, transcription factors are implicated in a wide variety of different diseases. Given this versatility, not just in terms of diseases, but also in terms of modality, as we just heard from Chris in the last question, we view the potential for therapeutic benefit to be substantial. As we think about partnering on the platform end, we're solving really for one key thing, and that's patient impact. Can a partner bring capabilities that we otherwise would not have in order to get therapeutics to patients quicker and more effectively than we would have otherwise done ourselves? That's how we're thinking about partnering on the platform end. All right, great. With that, I will turn it over to Yasir to close out our event. Thanks, Stephanie. Well, we've certainly covered a lot of ground here today, and I hope that you'll come away just as enthused as we are about all of the exciting science that we are undertaking here at Kronos Bio. As we close, I wanted to reiterate some of the key points from today, as well as highlighting some of the information that we are sharing for the very first time. With regards to SYK, while we had previously discussed our pivotal trial with ENTO in combination with 7+3 in the frontline setting, today, for the first time, we have communicated our overall SYK portfolio strategy and more specifically, how we will be deploying LANRA. We shared preclinical data that shows LANRA to have equivalent anti-leukemic activity as that of ENTO. We also shared preclinical data that illustrates that the combinations of LANRA and gilteritinib, as well as LANRA and venetoclax, are synergistic. This provides rationale for these combinations being explored in the clinic. With ENTO positioned in the curative potential setting and LANRA in the setting where continuous disease suppression is needed, our SYK portfolio has the potential to address mutations present in more than 2/3 of AML. With regards to KB-0742, our CDK9 inhibitor, Jorge highlighted the unique properties of selectivity, long plasma half-life, and oral bioavailability that we believe make KB-0742 uniquely positioned for success in the clinic. Stage I of our ongoing phase I/II trial is designed to quickly establish the appropriate dose and safety in an unselected patient population to allow us to move quickly into our expansion cohorts next year. With that in mind, in Q4, we hope to be able to report out initial data from Stage I with a view to confirming safety, a long plasma half-life, as well as initial signs of target engagement. We also shared today, for the first time, preclinical anti-tumor activity observed with KB-0742 in small cell lung cancer, non-small cell lung cancer, as well as triple-negative breast cancer. As a result, have provided further detail around these tumor types and others that we will be specifically targeting in our expansion cohorts. Last, and certainly not least, we discussed some of the exciting work we have been doing in discovery. This is a part of the organization that in and of itself can fill the entirety of an R&D Day. What we have covered today just scratches upon the surface of a platform that we believe will drive substantial value in years to come. To showcase some of this potential, Charles walked us through our differentiated approach to TRN mapping with a couple of vignettes in both small cell lung cancer as well as castration-resistant prostate cancer or CRPC. As an example, in androgen receptor-driven CRPC, by optimizing for selectivity, we have been able to identify multiple attractive hits and are currently at the stage of identifying their mechanism of action. Chris highlighted how our platform has and is continuously optimized for both throughput and quality. As a result, SMM is a versatile product engine that is designed to yield binders that have multiple modalities, which may allow us to finally drug these historically undruggable targets. Its first pivotal trial. All of this great progress sets us up with a rich stream of news flow that positions us well over the coming years. On the SYK side, our pivotal trial with ENTO in fit NPM1 AML is on track to start in the middle of this year, with MRD-negative CR data expected in the second half of 2023. We also intend to initiate our trial of LANRA in the relapsed/refractory FLT3 AML setting later this year, with initial data in the second half of 2022 and clinical proof of concept data in the second half of 2023. Our trial of LANRA in unfit AML is scheduled to start early next year, with initial data in the first half of 2023 and proof of concept data towards the end of 2023. For our CDK9 inhibitor, we look forward to sharing initial safety, PK, and QD data from at least six patients in Q4, with initial efficacy data in the second half of next year. We're actively working towards an IND filing for our next clinical program in the coming year also. Finally, thank you all for joining us today. We hope that you'll continue to join us on our journey to making a difference to the patients that we serve.