Zymeworks Inc. (ZYME) Earnings Call Transcript & Summary

Good morning, everyone. It's great to see you all today. My name is Neil Klompas, Zymeworks' President and Chief Operating Officer. I'd like to welcome you to Zymework's Early Research and Development Day. We appreciate you taking the time out of your busy schedules to join us, both in person here in New York and virtually via webcast. Before we start, just some housekeeping matters. In today's presentation, we will be making forward-looking statements. We rely upon standard safe harbor disclosures common to our industry and companies at our stage of development in our sector. For more information on risks and forward-looking statements, we encourage you to read our SEC filings available on our website. Additionally, for anyone watching the webcast live today, if you have any questions on today's material, please e-mail ir@zymeworks.com, and we'll try to answer your questions during the Q&A portion of this event at the end of today's prepared presentation. Today, we are very pleased to share this exciting look into our next-generation platforms and more importantly, a first look at some of the data on our preclinical programs. Before I hand it over to Paul and our research and development team leads, I want to say a few words in light of today's or yesterday's announcement. As you may have seen, yesterday we were thrilled to announce a global licensing and collaboration agreement for zanidatamab with Jazz Pharmaceuticals. I encourage everyone to view the webcast and read the press releases and other relevant disclosures, all of which are available on our investor website as we will not be spending any time today speaking to the agreement. The focus of today's event is on our early research and development pipeline. However, I would like to briefly frame what this opportunity means to Zymeworks as a company as I think it is pertinent to today's discussion. With our significantly improved financial position and our newly updated cash runway guidance, which enables Zymeworks to fund planned operations through at least 2026 and potentially beyond, we are now able to enact a focused R&D strategy with emphasis on development opportunities in difficult-to-treat cancers where we can utilize our in-house multi-specific antibody therapeutics and antibody drug conjugate platforms, which we will speak to today. Going forward, it's important to highlight the partnering and business development opportunities will continue to be core to our business. and to our long-term strategy of developing these and other novel antibody-based therapeutics, and we hope that this partnership provides an indication of our ability to form, execute and integrate important partnerships and collaborations that will be helpful to every aspect of our business. The agreement we announced yesterday will not be the end of what we intend to be a core piece of our go-forward strategy. Today, Paul and our research and development leadership team look forward to providing context on the pipeline you see before you on this slide. The team we have with us today was pivotal to the development of zanidatamab, zanidatamab zovodotin and the platforms that enable us to develop these and other therapeutics. We have an exciting first look today at our preclinical pipeline, and we hope you walk away with a better understanding of the rationale and data behind the therapeutic modalities and clinical candidates that will drive the next wave of growth here at Zymeworks. With that said, let me please introduce Dr. Paul Moore, our new Chief Scientific Officer. Paul brings with him significant experience in biologics, drug discovery and development, spanning a career of 25 years where he has led discovery and development efforts across a range of FDA-approved and clinical stage biologics. I know I speak for our entire leadership team and Board when I say we are thrilled to have him as part of our executive committee and broader team and look forward to the benefits that his experience brings to Zymeworks. As you will see from the presentation today, under his leadership, our research and development teams have been working diligently and with a renewed sense of excitement to meet our corporate goal of delivering 5 new Zymeworks developed candidates to the clinic in the next 5 years. Now let me hand the mic over to Paul and his team. And again, thank you all for attending and listening in online.

Okay. Well, thank you, Neil. That was a nice introduction, and I have to say, I'm delighted to be at Zymeworks at this stage in the company's development, working both with Neil and the other senior leadership team, but also very importantly, with the talented research team that is in place at Zymeworks. What we're really going to be focused on is the development of next wave of multifunctional antibody therapeutics in oncology. So since coming on Board, I've been, as I mentioned, extremely impressed with the talent and the tools in place at Zymeworks. And what we really have is a team that's capable of supporting both the development of multi-specifics and ADCs, which really kind of sets Zymeworks apart from other companies in the field. Importantly, the infrastructure and the experience to also take molecules from preclinical into development, into clinical studies is also in place at Zymeworks, which really provides us the runway to take our discoveries into the clinic. Key design works are the proprietary platform technologies that have been developed, that enable the design and generation of unique and differentiated functional -- multifunctional proteins. On this slide, I'd like to focus on the Azymetric and the drug conjugate platform technologies, because they're really the foundation of the multi-specific and the ADC molecules that we're developing. With Azymetric, we have the ability to generate bispecific molecules through the efficient heterodimerization of the Fc. We can then incorporate antibody-based binding specificities tailored to the Fc to generate bispecifics and tri-specifics. Importantly, these molecules maintain IgG1-like properties and can be purified as such and manufactured as such. Within our drug conjugate platform, we then also have proprietary technology and expertise to generate antibody or bispecific antibody drug conjugates. And we have a suite of payloads that can be matched -- can be plugged into that system. This includes our recently developed topoisomerase 1 inhibitor based payload technology that we'll be showcasing in the context of product candidates in the presentation today. In addition, we also have our EFECT and ProTECT platforms that we could also wire into our products. And there, with the EFECT, we can use that to enhance or diminish effector function, both in the context of the ADCs or bispecifics. And then we also have our ProTECT platform, which we could employ to limit functional activity of our molecules to the tumor macro environment. A common theme to our technologies are that they're all antibody-based, so they could all integrate with each other. And this really allows us to build molecules that are fit-for-purpose with the features that our technologies can endow on a molecule. So we can take an antibody and then we can have this choice in how we conjugate, which drug we conjugate to that and how we develop the ADC. But we can also, of course, plug in a second specificity, a second antibody into these platforms and make a bispecific. So this really provides us the opportunity to make either multi-specifics or ADCs with different features depending on the challenge of the target. I think what's really important, though, is that Zymeworks technology works clinically and has yielded therapeutics that have been clinically validated. The first program, of course, that you may be aware of is zanidatamab, which is a biparatopic antibody based on our isometric platform, which targets 2 independent episodes of HER2 that has shown clinical activity across multiple HER2 tumor types, and that's in clinic -- the pivotal clinical studies. We've then also developed ADCs, in this case, ZW49. So we also have that come from concept to clinic and was recently shown some clinical data on that. So we also have the experience of taking ADCs to the clinic. Our platform technologies have also been validated through external partnerships and other internal programs. So on the external partnerships, we've -- in the Azymetric, which is the foundation of our multispecifics, there we have T-cell engagers that are in the clinic by -- with Johnson & Johnson, and then additional other undisclosed programs that have been developed by our other corporate partners covering different therapeutic areas. And then on the antibody drug conjugate side, here we have also partnered those technology with companies. And so Exelixis are developing a tissue factor ADC and they will be presenting clinical data next -- this month at their triple meeting in Barcelona. While we are also developing an FAAC ADC targeting -- utilizing technology license from Zymeworks. In addition, the flexibility that can be shown in different examples of recently presented data or to be presented data from Zymeworks. So on the Azymetric side, we presented last year, we're showing conditional co-stimulation using a bispecific approach. And then we can also restrict cytokines to the tumor microenvironment using our Azymetric and then on the ADC side, we'll be presenting next month at SITC, work on our TLR7 conjugates. Before -- what it also indicates is that with our focus now moving forward with cancer therapeutics, we are going to focus on cancer indications where we consider the greatest unmet medical need, and that we can measure by patients with the poorest outcome, 5-year survival. And there, these are challenging diseases to treat. That's the reason they have poor survival. But we think that with the multifunctional proteins that we're developing, we now have the tools available to try and take on these challenges. And that will be a theme that you'll see during the presentation. So let me just -- maybe then summarizes our overview of our path from going from the tumors that we want to treat to integrating with our technologies, either ADCs or multi-specifics. And here, what we really like are the features, the optionality that we can do with the 2 foundational platforms modalities with an antibody drug conjugate platform. There, we can actually really customize the molecule through our experience in antibody engineering, looking at antibody properties with the proper format or the appropriate format for the challenge, with the payload, the linker conjugation in the dark. So they are all variables that we consider when we design a molecule and we can customize that antibody drug conjugate as needed. And then on the multispecific side, we likewise have the ability to really functionally fit that molecule fit for purpose. So there, if we need multiple mechanism of action or we need multiple specificities to get more precision targeting, you'll see that, then we can do that with the design of a molecule. So where are we? So I think it was 5 years ago here in 2017 that Zymeworks became a publicly traded company. And here -- and during that time, we've made a lot of progress. We've a select product pipeline. We've really developed a lot of good platform technologies. But what we really want to do is now pivot somewhat and really leverage those advantages in technology and infrastructure and accelerating in the next 5 years our in-house pipeline. And as Neil mentioned, we were -- our goal of developing 5 new products into the clinic in the next 5 years, and that will be a balance of ADCs and multi-specifics. But we -- and we see some of those initial programs that we're going to be bringing into the clinic in the presentations that the research leadership team are going to be giving shortly. So this is introduction to the team. They're here and Nina and Thomas, they really spearhead our multispecific team. and then they will be talking in a little bit of time. Stuart and Jamie will take the podium first and they're going to cover our ADC technologies. And just to give you a quick idea of the content, what we will be doing is we will be revealing for the first time data on our 2 next IND candidates, 191 and 171. And then behind that, we'll also be giving you a flavor of the other programs that we're working on and how we see us progressing towards this goal of having 5 INDs in the next 5 years. So with that, I'll hand it over to Stuart.

All right. Thank you, Paul. Good morning, everyone. My name is Stuart Barnscher, and I'm the Director of Preclinical Programs in the ADC Therapeutics Development Department, and I'll be introducing you to our thoughts on ADCs and their development. And I'll be also talking about how we're uniquely positioned based on our platforms, scientific expertise and our unique approach to ADC design to generate fit-for-purpose ADC product candidates. So delivering targeted chemotherapy through antibody-drug conjugates has emerged as an extremely effective therapeutic strategy for multiple types of cancers. In total, there are 11 FDA-approved ADCs and built on the learnings of the pioneering work from Seattle Genetics on ADCETRIS and Genentech on Kadcyla, we have seen an explosion of approvals in the last 3 years. Seven ADCs were approved in 2019, 2020 and 2021. There are a number of other promising ADCs on the cusp of approval, and we look forward to the continued success for this clinically impactful modality. The 11 approved ADCs represent success in both liquid and solid tumors, using a variety of payload mechanisms with a range of potency. The more potent DNA damaging agents like calicheamicin and the PBD dimers have been successful in liquid tumors targeting CD22, CD33 and CD19. While the slightly less potent microtubular inhibitors have been seeing success in liquid tumors targeting CD30, BCMA and CD79B and in solid tumors targeting HER2 tissue factor and Nectin-4. The relatively lower potency topoisomerase I inhibitors have seen success in solid tumors targeting HER2 and TOV2. Despite all of this recent success, the discontinuation rate remains high across multiple payload mechanisms, especially for the more established payload classes like the auristatins, the maytansinoids, the calicheamicinsand the PBD dimers. While still relatively early in its development cycle, the more nascent Camptothecin class is on an exciting trajectory with very few discontinued programs to date. Now the reason for each discontinued -- discontinuation can be categorized into 2 main themes: The first one being on-target toxicity due to target expression on normal tissue; and the second being lack of efficacy despite achieving the maximum tolerable dose dictated by the platform. Discontinuations due to on-target toxicity constitute a very small proportion of total discontinuations, and the majority of discontinuations occur because of lack of efficacy at the maximum -- at the platform maximum tolerated dose. We believe the key area for improving on past failures is by selecting the right design criteria for each ADC component. The target, the antibody, the linker conjugation and the payload, all need to be considered in order to drive robust efficacy at tolerable doses. As I just mentioned, building a successful ADC requires the right design, careful selection and combination of target antibody linker conjugation and payload are critical. For the target, one needs to consider the expression profile as this will dictate how to select the antibody binding properties and the payload mechanism and potency. The payload potency and metabolism will dictate the features of the linker and the conjugation method employed. Simply taking an old antibody and repurposing it for ADC use may not yield the best results. Similarly, taking a popular payload and just applying it to an antibody without considering the targets and the tumor biology may not yield a successful ADC. To realize the right design, we need the right tools. There's no one-size-fits-all solution to this, and employing a toolbox of technologies allows for careful tuning of each ADC design feature. And in order to bring all of this together, the right combination of design features and technologies, we need the right team. Having a fully integrated multidisciplinary team that eats, sleeps and breathes ADC's development will increase the likelihood that a program can be taken from design to discovery through to development and in the hands of physicians to treat patients. At Zymeworks, we're uniquely positioned to build fit-for-purpose ADCs due to the breadth of our internally generated ADC technologies. Our ZymeLink Auristatin technologies are our most mature and used in our anti-HER2 ADCs ZW49 and in XB002 and ATRC-301. Similar to our ZymeLink Auristatin technology, our ZymeLink Hemiasterlin technology is also a microtubular inhibitor and uses the same [indiscernible] spacer to link to payload core to a protease cleavable linker. Our topoisomerase 1 inhibitor platform utilizes a novel bystander active Camptothecin payload. And we'll be talking a lot about this technology as it is used in the 3 pipeline assets we'll be discussing today. We also have a site-specific conjugation technology that allows for homogenous conjugation at multiple sites. And when combined with our Azymetric technology, we can precisely control the DAR. We also have an immunostimulatory drug conjugate platform that Paul mentioned, and we'll be presenting a poster on this technology at the SITC Annual Meeting next month. So Zymeworks has been investing in ADC technologies for the last 8 years. The ZymeLink technologies were the very first ones to be developed, and the ZymeLink Auristatin technology has been applied in 3 programs now, 2 in clinical development and 1 in late-stage preclinical development. In the last 3 years, we've really focused on developing new technologies to fortify our ADC toolbox. This includes the ones I mentioned earlier, like the site-specific technology, the TLR7 immunostimulatory technology and the topoisomerase 1 inhibitor technology. But now what we're doing, as Paul mentioned, is really shifting our focus from developing technologies and tools to making these -- to building clinically and commercially relevant ADC assets. We've done it successfully once with ZW49. And we're really excited to unveil 3 new pipeline assets today in ZW191 ZW220 and ZW251. As I mentioned earlier, the combination of each ADC component is critical to designing a successful ADC. At Zymeworks today, our strategy is to focus on targets with evidence of clinical activity in indications with high unmet need. We believe that known targets are still actionable when armed with the right technology and the right design. The antibody is a far too often overlooked component of an ADC. Many ADCs resort to using repurpose antibodies. Here at Zymeworks, we focus on specifically discovering and developing optimal ADC antibodies because antibody specifically selected for ADC use are going to increase the likelihood of success of that ADC. For the linker conjugation, we've leveraged validated peptide cleavable linkers and stochastic conjugation for right now. The peptide cleavable linkers are used in more than half of the approved ADCs and provide the benefit of releasing the unmodified payload. Stochastic conjugation is still the gold standard and used in all 11 FDA-approved ADCs. For the payload, we're going to focus on our novel topoisomerase 1 inhibitor ADC platform because there is undeniable evidence right now that topoisomerase 1 ADCs are providing meaningful clinical benefits to patients. We believe that our topoisomerase 1 inhibitor platform is a critical component to building clinically impactful ADC assets at Zymeworks. And with that, I'm going to pass things over to my colleague, Jamie Rich, to discuss that novel topoisomerase 1 ADC platform.

Okay. Thanks, Stuart, and hi, everyone. I'm Jamie Rich. I'm the Director of Technology in the ADC Therapeutic Development group here. And I'm going to start off today by orienting you to Zymeworks topoisomerase 1 ADC platform, that's part of some of the pipeline that we'll be discussing a little bit later on. So you often hear that HER2 is transforming breast cancer. It's pretty common trend these days. And impressive patient responses in these waterfall plots from the DESTINY-Breast studies, seen in the upper left of the slide, really illustrate what's driving this change. Furthermore, HER2 is also showing really meaningful benefit to patients in several other solid tumor indications, as you can see both in this slide and in clinical data that we've presented -- or have been presented historically. I think perhaps most importantly for us though, there's 4 other DXd-based ADCs, specifically non-HER2-ADCs in solid tumor indications, and these are also starting to show real benefits for patients. So from my perspective, the key takeaway is that all of these data really illustrate just how successful antibody-based targeting of topoisomerase 1 inhibitors can be to the point where we don't really feel that, that can be ignored. So with this in mind, we've developed our own TOPO1 ADC platform and corresponding pipeline and indications that we feel will benefit the most from using this mechanism of action. So the clinical stage TOPO ADC landscape can be subdivided into a couple of different strategies. There's the repurposing of no one experimental drugs, and then there's the discovery of fit-for-purpose ADC payloads and linkers. Companies that are repurposing for ADC use experimental drugs that were previously in the clinic as normally as stand-alone chemotherapies are more common and they're grouped on the left-hand side of this slide. So I think there's a couple of points about these that bear consideration. First off, in many cases, the payloads are more potent compared to those that are in the approved drugs. So that would be SN-38, in [indiscernible] DXd and then HER2. Secondly, the biggest focus, both preclinically as well as clinically involves the repurposing of Exatecan. Exatecan is the most potent of the top payloads in the clinic. And it's among -- it's worth remembering that in order to use Exatecan for an ADC purpose, it needs to be engineered or at least we need a chemistry solution to enable its use. So a different approach is to search for and to develop novel payloads specifically for use in ADCs and a smaller number of companies are taking this tact. Daiichi-Sankyo and DXd are one of them. And this is also where Zymeworks has purposely decided to position ourselves, namely because the approach allows us to carefully tune the biophysical characteristics of the small molecule and thus the biophysical characteristics of the ADC. And also because it enables us to tailor not only potency, but other drug-like properties of a small molecule towards the intended use in an ADC. So being able to tailor the properties, these properties, is an imperative task. And ultimately, it's what led us to generate our own platform. So in designing a TOPO ADC platform, the 3 elements that require consideration are fairly straightforward. The payload, the linker and the conjugation chemistry. So together, these components make up the drug linker and they dictate in a large part, how an ADC will work, how it will function. And so we had specific criteria in line for our drug linker. And I can walk you through these criteria in the context of each component. So the payload. Historically, the payload is what's received the most attention in the ADC field. Our payload criteria were pretty simple. We were searching for a Camptothecin-based molecule with moderate potency, strong bystander activity and good drug-like properties. There isn't really time to go into depth about the drug-like properties comment, but I think solubility, permeability, metabolic stability, PK, transporter profile and so on. So the features of the linker, as I mentioned, are also very important. Our design criteria here included a cleavable peptide sequence that would release the payload unmodified, the so-called traceless linker. But what would -- we're also looking for a sequence that would be sufficiently stable in circulation to avoid release of the payload prematurely. In terms of chemistry, the [indiscernible] reaction was favored so that we could attach the drug linker to cysteine residues on the antibody. Not only is it simple chemistry that enables us to easily tune the DAR for each of our candidate molecules, it's also very well validated, and we feel it provides a good balance between safety on the one hand and antitumor activity on the other. So I'll shed just a little bit more light on our general thinking around topoisomerase ADC drug linker design. So ADCs are mechanistically complex, and I think it's naive to consider them as a means simply to deliver the payload directly to the tumor through a single mechanism. In fact, I think we know that the vast majority of payload number is not delivered directly to the tumor, but it's released elsewhere. And so, of course, that speaks to the need for tolerability. Further to this, we think that the circulating payload likely plays a role in the observed antitumor effects of topoisomerase 1 inhibitor based ADCs. And as such, it's important that the payload have those good small molecule drug-like properties that I referred to previously. So the effort that led to our TOPO ADC platform and its subsequent application to our pipeline is summarized on this slide. What we're trying to convey here is that our molecules have been put through rigorous screening process, and the features that we were looking for were evident not just in a single experiment, but really across a number of complementary experimental systems. The confidence that we have in this approach and in this data comes from, in no small part, the ADC team that we've assembled over the years and developed over the years at Zymeworks. It starts with medicinal chemistry. Of course, it includes antibody generation, bioconjugation and analytics. Of course, it captures in vitro and in vivo pharmacology groups as well as the translational sciences team and others. So we started on the effort to develop this platform by synthesizing and screening on the order of about 100 novel Camptothecin molecules as potential payloads. And we compared them to various benchmarks toxins, including things like DXd, SN-38, Exatecan, Belotecan and more. And the focus at this stage was primarily on potency as I mentioned previously, but also on reading out hydrophobicity because of its potential to lead to undesirable aggregation and off-target toxicities. So we identified a selection of payloads with appropriate potency, hydrophobicity and other preferred properties. We converted those payloads into drug linkers and then we conjugated them to different antibodies. The ADCs that we prepared were then put through our biophysical characterization workflow to get rid of anything that was poorly behaved in terms of a drug linker in ADC. And quite frankly, finding molecules in this class that don't aggregate is one of the biggest challenges. So the ADCs at this stage that remained were then assessed in a built-for-purpose tumor spheroid-based potency assay, and we gradually narrowed down the pool of ADCs by assessing in-vivo antitumor activity in CDX and PDX models. And then finally, by looking at the top ADCs in terms of tolerability in most mice and rats. Ultimately, a dedicated nonhuman primates toxicology and toxicokinetics study was conducted, and this yielded us with our lead payload and drug linker. So overall, the platform development effort was rigorous and it was comprehensive in scope. It employed 8 antibodies, 8 different tumor-associated antigen targets, dozens of cell line-derived and patient-derived xenograft models, several specific PK assessments and multiple toxicity studies. And ultimately, this enabled us to launch the 3 distinct pipeline programs that we'll be talking about today, ZW191, ZW251 and ZW220. And in addition, the technology has been applied to several early-stage assets. So I'm going to hand it back to Stuart now who's going to talk about ZW191.

All right. Thank you, Jamie. The first program I'd like to introduce you all to is ZW191. This is a potential best-in-class ADC targeting Folate Receptor Alpha. As I just mentioned, ZW191 targets the clinically validated target folate receptor alpha, the expression profile of FR alpha is compelling in ovarian cancer and other gynecological cancers and other solid tumors like non-small cell lung and triple negative breast. We've paired an internally discovered antibody with our novel topoisomerase 1 inhibitor platform to generate ZW191. The antibody itself was selected from a large panel to find a unique antibody with optimal internalization, payload delivery and tumor penetration. We're really excited by the preclinical data that we have on this program where we're able to dose nonhuman primates safely up to and including 30 milligrams per kilogram, and we see strong antitumor activities in models with a range of folate receptor alpha expression. Folate receptor alpha is a GPI-anchored membrane protein responsible for the transport of folate into cells. In normal tissue, expression is limited to the apical surface of the intestine, lung, fallopian tube, placenta, choroid plexus and the luminal surface of the kidney. In cancer, over-expression is observed in numerous gynecological cancers, like I mentioned, ovarian cancer, and in other solid tumors like non-small cell lung and triple negative breast. The strong over-expression profile in cancer and the restricted expression in normal tissues, makes FR alpha an ideal ADC target. So as I mentioned earlier, and I'll reiterate here, antibodies specifically selected for ADC use are very important to a successful ADC. Here, we show that the ZW191 antibody is superior to the antibody utilized in ImmunoGen's anti-folate ADC, mirvetuximab soravtansine in 2 important attributes: internalization and payload delivery. In the left-hand panel, we show the ZW191 internalizes to a greater extent compared to mirvetuximab in a 24-hour period in folate receptor alpha-expressing KB/HeLa cells. This is a fluorescence-based assay, and it's clear from the images that both the fluorescent intensity and the breadth of fluorescent distribution is greater for ZW191 compared to mirvetuximab. In the right-hand panel, we use a different assay. This assay utilizes mass spectrometry to quantify the concentration of intracellular payload present in 4 different folate receptor alpha-expressing tumor cell lines, after treatment with either the ZW191 mab or mirvetuximab conjugated to the same payload at the same drug antibody ratio. As we look at the blue bar, ZW191 in relation to the purple bar, mirvetuximab, we see that the concentration of payload found in all 4 cell lines is greater for ZW191. Now I'd like to draw your attention to the KLE3, HEC-1A and H2110 cell lines, which express moderate to low levels of folate receptor alpha and we see that 191 retains impressive payload delivery in these lower expressing cell lines. Internalization and payload delivery are important features that can drive antitumor activity. And the ZW191 antibody was selected to maximize both of those features. Another important attribute we assessed was the ability of ZW191 to penetrate 3-dimensional tumor spheroids. In vitro, we're moving more and more of our assays to a 3-dimensional format, as we believe this assay format better recapitulates the architecture of tumors. This is -- again, this is a fluorescence-based assay where we look at the fluorescent signal over a 96-hour time course in 3 different layers of a single tumor spheroid. From the images, you can see that the fluorescent signal for ZW191 is more intense in the outer layer, the middle layer and the inner layer compared to mirvetuximab. When we quantify this fluorescent signal in the graph on the right-hand side, we see that at every time point and in every layer of the tumor spheroid, the fluorescent intensity for ZW191 is greater compared to mirvetuximab. This means that ZW191 has the potential to penetrate tumors more deeply and to a greater extent compared to mirvetuximab, and this may contribute to a more robust antitumor response. When we designed ZW191, we wanted the ADC to have strong bystander activity, as we believe this is an important feature to target heterogeneous tumors and tumors with low antigen expression. Bystander activity is the ability of an ADC to kill antigen-negative cells in the presence of antigen positive cells. The antigen positive cells process the ADC and liberate the payload. The payload then permeates into neighboring antigen-negative cells and elicits its killing effect. The cartoon on the left depicts the bystander assay where we compare the viability of antigen negative cells in monoculture to the viability of antigen negative cells in coculture with antigen-positive cells. In the figure on the right, we see that ZW191 exhibits a strong bystander effect as the viability of antigen-negative cells in cold culture with antigen positive cells, the blue bar, is 30% lower compared to the antigen negative cells in monoculture, the green bar. To put this into perspective, we compared the bystander activity of ZW191 to the technology used in HER2, DXd but conjugated to our ZW191 antibody. We see that the folate receptor alpha, DXd-ADC elicit similar levels of bystander killing ZW191 in this assay. Now moving on to in vitro or in vivo experiments. We assess the ability of ZW191 to reduce tumor volume in 4 ovarian cancer xenograft models, with a range of folate receptor alpha expression. You can see that ZW191, the blue curve, is effective at reducing tumor volume in all 4 of these models. The left-hand panel is looking at 2 folate receptor alpha high-expressing models, OV90 and CTG2025, dosed at a single dose of 6 milligrams per kilogram. We see that ZW191 has similar antitumor activity to relevant controls like mirvetuximab in purple and MORAb-202 in red. The situation becomes more interesting on the right-hand panel when we look at 2 folate receptor alpha mid expressing models, OVCAR3, and CTG0958. In both models, we see that ZW191 is more effective in reducing tumor volume compared to these relevant benchmarks, suggesting that ZW191 has the potential to target lower levels of folate receptor alpha expression. This data is encouraging that across a range of folate receptor alpha expressing models, we see robust antitumor activity with ZW191. Finally, we assess the tolerability of ZW191 in both rodent and nonhuman primate toxicology studies. Nonhuman primates represent a relevant antigen binding species. And in this 2 dose, every 3-week study, we found that ZW191 was tolerated up to 30 milligrams per kilogram. In this study, we evaluated how we compared to the DXd technology, but conjugated to our ZW191 antibody at the same DAR, and we saw a similar maximum tolerated dose and similar target organs of toxicity with the folate receptor DXd-ADC and ZW191. From this study, we also see that ZW191 displays linear and stable toxicokinetic profile. Not shown on this slide, but worth mentioning, we also dosed ZW191 in mice and rats, and those represent non-antigen binding species, so that's going to be the platform talks that I mentioned earlier. And we found that ZW191 was extremely well tolerated up to 200 milligrams per kilogram. This exquisite tolerability in nonhuman primates and strong antitumor activity in models expressing a range of folate receptor alpha suggest a favorable therapeutic window for ZW191. We believe that ZW191 has significant commercial opportunity due to the distinguishing features of the antibody paired with our novel bystander active topoisomerase 1 inhibitor payload, and compelling preclinical -- the compelling preclinical data and the expression profile of FR alpha in cancers with unmet need. We have potential best-in-class opportunity in folate receptor alpha-high ovarian cancer and potential first and best-in-class opportunities in folate receptor alpha high endometrial, non-small cell lung and triple negative breast cancer and in those indications in folate receptor mid and low expression. So to summarize, FR alpha has a favorable expression profile in ovarian cancer and other solid tumors. And we think the mechanism of topoisomerase 1 inhibition is favorably positioned in ovarian cancer due to the sensitivity of ovarian cancer to topotecan chemotherapy. We're really excited about the ZW191 asset as we designed it to have optimal internalization, payload delivery, tumor penetration and antitumor activity. We also believe that our topoisomerase 1 inhibitor payload is likely to have a differentiated safety profile when compared to MERV and STRO-002. We're excited about the numerous best-in-class and first-in-class opportunities for ZW191 and we're moving full steam ahead with this program positioning it for 2024 IND. With that, I'm going to pass things back to Jamie Rich to talk about our next ADC asset, ZW251.

Okay. Thanks, Stuart. So now I'm going to spend some time as was just mentioned, to detail our second program, ZW251, which is potentially a first-in-class TOPO1 ADC targeting Glypican-3 or GPC3. So we're really excited about this program for a few reasons. First off, we're excited about the target. GPC3, Glypican-3, is an oncofetal protein that's very highly expressed in hepatocellular carcinoma or HCC, which is by far the most prevalent form of liver cancer. Liver cancer patients generally face a fairly poor prognosis. And we view an ADC as having the potential to make a significant difference here. Next, we're excited because we've developed a great antibody that we selected specifically to use as an ADC, and we're coupling with our bystander active TOPO technology. Finally, we're still excited about our data. We have strong in vivo pharmacology data supporting broad antitumor response of this ADC in PDX models and CDX models of hepatocellular carcinoma. And we've just initiated our pilot toxicology study in nonhuman primates. So let's start with a bit of background on the target. GPC3 is a GPI-anchored cell surface proteoglycan. It's expressed in the placenta. It's expressed in some fetal tissues, but its expression is downregulated in normal adult tissue. As I mentioned previously, it's highly expressed in some tumors, most notably in hepatocellular carcinoma in the liver. So there are a few important points that relate to the ADC mechanism for this target that I want to get into. First off, GPC3 expression in HCC is very common with GPC3 overexpression evident in about 70% of tumors. Furthermore, the expression is not only generally quite high, but it's also relatively homogeneous. And this is exemplified in the left most immunohistochemistry image below. It means that an ADC can target a high proportion of the tumor cells directly. Finally, the target is not appreciably expressed in normal adult tissues, and there's very little expression in other kinds of liver damage, which is frequently present alongside of hepatocellular carcinoma. So the takeaway is that GPC3 is not only a good marker for HCC, it's also a good target for the delivery of ADC payload, and we have an ADC antibody that's tailored for that purpose. So let's take a minute to go through some of the data that support ZW251 on the next couple of slides. So codrituzumab is an antibody from Chugai that demonstrated in the clinic that an antibody can engage the target, they can engage GPC3. The graph at the left-hand side of the slide here illustrates the strong and equivalent binding to an HCC cell line of ZW251, which is in blue and codrituzumab, which is in red. The second panel in from the left relates to our optimization of another key AAC specific property, namely internalization. So in this experiment, HEPG2, hepatocellular carcinoma cells were incubated with labeled antibody, the amount of internalized fluorescence was determined, and it's represented by the height of the bars in the bar graph. We actually identified the ZW251 antibody, that's the blue bar by screening a large panel of GPC3 targeting antibodies, some of which are shown here in black and by selecting the one with the greatest internalization. That internalization of the antibody or more specifically the ADC is also inferred or evident in the next panel over the third panel from the left, where the potent cytotoxicity of ZW251 ADC towards tumor steroids generated from the same HEPG2 cell line is shown. The blue line illustrates the cell killing and its concentration dependence. And the orange line corresponds to an isotype control and demonstrates that the cytotoxicity that we're seeing for ZW251 is driven by GPC3 targeting. So finally, at right, we measured strong bystander activity in an HCC coculture model. So here, you can see that when only antigen-negative cells, the light blue or purple colored bar are there, that the ADC doesn't kill them. But when you apply the ADC to a mixture of antigen-positive and antigen-negative cells, the dark blue data bar, the ADC eliminates about 80% of the cells that don't express the target. So we're seeing a bystander effect there. So taken together, these data highlight not just the ADC properties of the antibody, but also the compatibility of the ZW251 GPC3-targeting antibody and our TOPO platform. So we're also very pleased to note that ZW251 has strong antitumor effects across a number of cell line-derived and patient-derived xenograft models of hepatocellular carcinoma. Importantly, this strong tumor regression effect, which is illustrated by the blue data points in the left hand -- in the graph on the left hand of the slide, have been consistent across models that feature a range of high, medium and low GPC3 expression. Furthermore, you can see that the antitumor effects are targeted since the isotype control ADC, again in orange, have negligible impacts on tumor volume. So the PK or pharmacokinetics of the ADC is another important consideration. And the TG 32 humanized FcRn mouse model for PK is considered a pretty good way, an excellent way, in fact, to estimate the PK of antibodies at an early stage. So in this case, the PK observed are independent of the target, owing to the absence of a tumor in the animal and the lack of cross-reactivity between ZW251 antibody and mirrorring GPC3. It is, however, evident there are no PK liabilities for this antibody, the ZW251 antibody in blue, which overlaps more or less bang-on with the clinical benchmark codrituzumab. So together, these in vivo data here and multiple other studies that we haven't shown you today, emphasize the promise of ZW251, and we look forward to its continued development. So to summarize for ZW251, hepatocellular carcinoma is an indication with high unmet need, it could benefit from an ADC. And we believe that a GPC3 targeted TOPO ADC is the best approach owing to the strength of the target and the platform. We've identified a great ADC antibody that shows good specificity, strong internalization and when it's paired with our TOPO platform, it shows excellent antitumor activity in vitro and in vivo. So we're really excited about this first-in-class opportunity. We're looking forward to moving it on. So now I'm going to turn it back over to Stu and he's going to talk about ZW220.

All right. Thanks, Jamie. Okay. So the third ADC asset I'd like to introduce you to today, as Jamie mentioned, is ZW220, a potential best-in-class ADC targeting NaPi2b. So as I just mentioned, ZW220 targets NaPi2b, our solute carrier family 34 member 2. This target is overexpressed in ovarian cancer, non-small cell lung and other solid tumors, and ZW220 has been designed with an internally discovered novel IgG1 antibody conjugated to our topoisomerase 1 inhibitor platform. Again, we've identified the best antibody that is exquisitely suited for use as an ADC with optimal internalization properties. We have strong antitumor data supporting the activity of this molecule in patient-derived xenograft models of ovarian cancer and we've just initiated a pilot toxicology study in nonhuman primates. So just a little bit of information about the target. NaPi2b is a multi-pass transmembrane protein responsible for sodium phosphate transport and contributes to phosphate homeostasis. It consists of 4 extracellular loops. As you can see in the cartoon on the right-hand side of the slide, the second extracellular loop is the largest and provides the most epitope space for antibody binding. In normal healthy tissue, NaPi2b is expressed in the epithelial cells in lung tissue, the small intestine and the mammary gland. But in cancer, there are very high rates of over expression in ovarian cancer, non-small cell lung and endometrial cancers, as I just mentioned. So as I mentioned earlier, the NaPi -- the antibody used in ZW220 was specifically selected from a large panel for use as an ADC. Here we show that ZW220 has strong tumor cell binding and optimal internalization. We compare our antibody to 2 relevant NaPi2b antibodies, upifitamab. This is the antibody that's used in Mersana's upifitamab rilsodotin and lifastuzumab. This is the antibody that's used in Genentech's lifastuzumab vedotin. In the left-hand panel, we're looking at the dose-dependent binding of ZW220 in blue in the NaPi2b expressing ovarian cancer cell line, IGROV-1. Upifitamab, in purple, binds to similar levels as ZW220, while lifastuzumab, in red, binds at slightly lower levels. Now on the right-hand side, we see that ZW220 in the blue bars internalizes to a greater extent compared to competitor antibodies over a 24-hour period the NaPi2b-expressing ovarian -- in the NaPi2b expressing ovarian carcinoma cell line IGROV-1. Again, this is our fluorescence-based assay, and it's clear from the images that both the fluorescent intensity and breadth of fluorescence distribution is greater for ZW220 compared to upifitamab or lifastuzumab. Again, we believe that antibody internalization is a critical attribute and feature that contributes to the anti-tumor activity of ADCs. It's really important. That figure didn't turn out very well. But -- so similar to ZW191 and ZW251, the ZW220 ADC was designed with bystander activity in mind to target heterogeneous tumors and tumors with low antigen expression. In this assay, we determined the bystander activity by comparing the viability of antigen negative cells in monoculture to the viability of antigen negative cells in coculture with antigen-positive cells. We'll fix the figure on the right when the slides get distributed, but let's just imagine that there is a blue bar that's a lot higher than the green bar in both of these. But ZW220 exhibits a strong bystander effect. The viability of the antigen negative cells and monoculture -- or sorry, in coculture is much lower. It's about 20% lower compared to the antigen negative sales in monoculture. Again, we compared to the DXd technology, and we see similar levels of bystander when that DXd technology is conjugated to our NaPi2b antibody at the same DAR. So next, we're going to look at ZW220-mediated killing of 3 NaPi2b expressing tumor spheroids, IGROV-1, TOV21G and H441. As I mentioned earlier, we employ a 3-dimensional tumor spheroid assay, where a single tumor spheroid consisting of thousands of tightly packed cells has grown in a single well. The images on the left show the untreated tumor spheroids characterized by their large volume and visible proliferation. Conversely, when we treat these spheroids with ADC we see reduced volume, and there's visible evidence of apoptosis and necrosis. Now when we look at the figures on the right, we see that ZW220 in the blue squares exhibits potent dose-dependent killing in all 3 tumor spheroids. The NaPi2b high expressing tumors spheroid, IGROV-1, we see that lifastuzumab vedotin kills at similar levels to ZW220. But in moderate, expressing TOV21G and H441 spheroids, we see that ZW220 show superior tumor spheroid killing compared to the lipotuzumab vedotin benchmark. Now we wanted to see how this was going to translate into our in vivo studies, and we looked at the antitumor activity on PK of ZW220. So on the left-hand side, we evaluated ZW220 in 2 ovarian carcinoma patient drives xenografts. CTG0958 is a NaPi2b high-expressing model, and CTG2025 is a NaPi2b low-expressing model. In both models, we see that ZW220 shows robust antitumor activity when dosed at a single dose of 6 milligrams per kilogram. In both of these models, lifastuzumab vedotin is ineffective at reducing the tumor volume. On the right-hand side, we evaluated the PK of ZW220 in a TG32 mouse model. And as Jamie mentioned, we consider this model and many consider this model to be predictive of human PK. And in the blue circles, we see that ZW220 shows stable and linear PK over 21 days. The PK profile of ZW220 is comparable to that of the 220 naked antibody, which is in green. And ZW220's PK profile compares favorably with the relevant NaPi2b benchmarks I mentioned earlier, like lifastuzumab vedotin in red and the upifitamab antibody in yellow. So to summarize ZW220, NaPi2b is a target with a compelling expression profile in ovarian cancer and other solid tumors. And we think the mechanism of topoisomerase 1 inhibition, again, is favorably positioned in ovarian cancer due to the sensitivity of ovarian cancer to topotecan chemotherapy. ZW220 was designed to have strong internalization, and we believe this contributes to the compelling antitumor activity we saw in those preclinical models. We have numerous best-in-class and first-in-class opportunities, particularly in NaPi2b expressing ovarian cancers and lung carcinomas. And we're going to continue to drive ZW224 with a nonhuman primate toxicology study already initiated. So just to summarize here, we've presented 3 advanced preclinical ADC assets in ZW191, ZW251 and ZW220. In all 3 of these assets, we've carefully selected the antibody to have strong internalization to facilitate payload delivery and drive antitumor activity, and we've matched the tumor and target biology with a favorable payload mechanism by leveraging our novel bystander active topoisomerase 1 inhibitor payload technology. ZW191 is our most advanced asset. Strong antitumor activity in preclinical models and nonhuman primate tolerability suggest a favorable therapeutic index, and we're well underway in our IND-enabling studies for this asset. ZW220 and 251, for those programs, we've presented evidence of compelling preclinical antitumor activity, and we've initiated nonhuman primate toxicology studies for both of these programs. So we just told you how we're using our technology today to build a pipeline of exciting ADC assets. Now I'd just like to take a few minutes to tell you about where we might go in the future. So moving forward, we'll continue to reinforce our unique design philosophy around target antibody linker conjugation and payload to develop innovative technologies and use our existing technologies in new ways to develop fit-for-purpose ADC assets. On the target, we want to explore novel targets. As we continue to address difficult-to-treat cancers, we will likely need to explore targets with less validation. For the antibody, we want to continue our pursuit of optimal ADC antibodies, but now start to leverage our bispecific and biparatopic know-how to engage targets in unique ways to drive better internalization, specificity and potentially even lower the target expression threshold required to see antitumor activity. For the linker conjugation, we'll be exploring novel linkers as we believe the linker design is really dictated by the payload potency, solubility and metabolism. On the payload side, we're going to be developing novel payloads. We're going to be adopting MOAs with clinical validation to a novel ADC application. This is going to provide us more opportunities to match disease and target biology with payload mechanism. So with that, I want to thank you all for your attention for this section, and I'll pass things back to Paul to discuss our multi-specific antibody strategy.

Okay. Well, thanks, Stuart. So yes, we're going to switch now to the other foundational modality that we have, our multispecifics. And let me introduce that by maybe putting into perspective where the field of multispecifics are and the bispecifics. So similar to what we saw with the ADCs, bispecifics are now starting to accelerate the pace of approvals. And in this year alone, there has actually been 5 bispecifics that have been approved, 3 of which are CD3 engagers. And still, the CD3 engager space is really where there's a lot of focus and a lot of opportunity. And that really was heralded by the approval of BLINCYTO in 2014, the CD3/CD19, which was developed by Micromet and then Amgen. And that really ushered in a huge amount of excitement about CD3 engagers and their potential. But there's been, as you can see, a bit of a hiatus between that and then the more recent approvals. And I think that's really because it's become much aware that it's actually much more challenging to design a CD3 -- T cell engager than maybe was initially appreciated. Okay. So what we feel, though, is that those challenges still remain. There are still limitations with the existing science for T cell engagers, and we feel that there is opportunity to apply the Zymeworks technologies to actually advance those kind of molecules and overcome limitations that we'll briefly describe in the next few slides. So this table slide summarizes the technologies and the toolbox for Zymeworks on the multispecifics side. Again, the Azymetric technology really is the foundation of our bispecifics, the cornerstone. The most advanced bispecific, of course, is the -- called a -- is a biparatopic antibody, the zanidatamab. But again, as I mentioned earlier, we've also -- our partners, corporate partner, J&J, they've actually developed T-cell engagers using Azymetric platform. And what we're going to be talking about today is in the bispecific or the multi -- more specifically, the multispecifics, our T cell engagers where we've actually taken it away from just a 1 by 1, where we actually developed a 2 plus 1 T cell engager platform to sort of maximize therapeutic window. And that will be around the discussion of our 171 program that Nina will discuss. And then what we'll also do is then talk about our tri-T cell engagers, the TriTCEs, where we're actually incorporating a third specificity into the molecule to give advantages over the existing simple bispecifics. And that will be in 2 flavors. There will be the costimulation addition, and then there will also be a checkpoint inhibition into a trispecific. So here is a timeline of the advances made at Zymeworks. We've made a lot of progress internally not only in the design and development of bispecifics using Azymetric, but also in the manufacturing and clinical development. In parallel, our corporate partners have also been developing programs. And together, we can use these learnings to really accelerate the development of future bispecifics. So what we want to do, as I sort of alluded to, is moving forward is that we want to move forward with this knowledge base that's been developed on a platform focused -- platform focus and then move that into more of a pipeline focus. And again, more specifically in the -- we see real opportunity in the T cell engagers space. And there, what we want to do is we actually want to use a more trispecific strategy, and we'll talk about that in more detail. But that's going to be the 2 plus 1 TCE and the TriTCE platform. So why do we think that the Azymetric platform can -- is well suited for this challenge, okay? So something to bear in mind is that the flexibility of the Azymetric and the power of that heterodimerization really allows you to engineer and modify different components of the molecule. So we can really dial in the change in the CD3 paratope, the features of that, the tumor antigen paratope. And the valency in the geometry and the format could also be supported by this platform. And what we could also do is not only design different molecules, but importantly because, again, of the efficiency of the heterodimerization, we can screen in high throughput. And Nina will discuss this in a little bit more detail when we talk about the development of 171, but this together really allows us to power through screening different formats and be almost empirical in our screening format to pick the right molecule to push forward. So let's put this a little bit into the context of the T cell engagers. So here, blinatumomab, of course, was again a breakthrough for the CD3 bispecifics, did show benefit over chemotherapy. But there are challenges, of course, with that molecule. We know it's a short-acting -- a short half-life molecule. So that also necessitated continuous infusion, so subsequent molecules, including Fc, which would be -- we'd have with our Azymetric. And then of course, there's also, as you see on the clinical data, the durability of response is not optimal. It could -- would be nice to see responses that were more sustained and longer sustained. So overall then, with T cell engagers, there are also more global challenges that are associated with them. And probably the one that you may be most familiar with is cytokine release. So that's a key problem in that when you have T cell activation, T cell killing, it comes with cytokine release, which, to some degree, can be good, but you don't want too much of it. In some of the first-generation molecules, the balance between activity that you wanted with cytokine release was skewed. So there, you want to try and limit cytokine release while maintaining killing. What you also don't want to do is you want to -- you don't want to have too much killing of normal versus your tumor-expressing cell. And getting that difference -- differential activity also requires some thinking about design. And you'll see that we've factored that into the design of 171. Another challenge is that even with a -- many tumors don't have enough T cells or infiltration, but the T cells are energized. So the T cells are not strong enough to mount a response. So you need to help the T cells, and the signal 1 is not enough from the CD3. And then the third key problem is that in solid tumors, you often have other components that signal an immune suppressive activity or cell populations or pathways, and that could also then dampen the ability of a T cell engager to work. So what we've done, and we'll touch on these in more detail, and we'll get into these in more detail, but for the first biological problem, where there's a narrow therapeutic window in toxicity due to CRS, what we've done is we've engineered our 2 plus 1 T cell engager to have a CD3 binding, low-affinity CD3 binder, which maintains killing activity but has reduced cytokine response. And it's a novel CD3 binding domain. And then what we also have is then we've also engineered a 2 plus 1 format to sort of skew binding towards cells that have high level of tumor antigen expression. And then for the second challenge with the limited T cell activity, what we're doing is we're going to incorporate costimulatory signal into the trispecific to actually activate and give a second costimulatory signal to the T cell to enhance the fitness of those T cells. And then third, for -- to overcome the tumor microenvironment where you've got suppressive signals, we're going to incorporate a checkpoint inhibitor specificity into the trispecific T cell engager. So that's sort of just an introduction. Nina and Thomas are going to talk about these programs in more detail, so I'll hand over the podium to them.

Thank you, Paul, and good day, everyone. My name is Nina Weisser. I'm Zymeworks' Director of Multispecific Antibody Therapeutics. Today -- I need to change the slide first. Today, I will provide an overview of the data supporting our preclinical candidate, ZW171, which is a bispecific T cell-engaging anti-CD3 and anti-mesothelin antibody for the treatment of mesothelin-expressing cancers. ZW171 is a bispecific T cell engager that is built using Zymeworks' Azymetric platform. The illustration on the left shows the ZW171 antibody format, illustrating what is commonly referred to as a 2 plus 1 format, where there are 2 paratopes targeting mesothelin and a single paratope targeting CD3, shown in blue. Heterodimeric heavy chain assembly is facilitated by the Azymetric platform mutations in the Fc. ZW171 facilitates its antitumor activity by binding CD3 on T cells and redirecting T cell cytotoxicity to mesothelin-expressing cancer cells. The tumor target, mesothelin, is a glycoprotein that is highly expressed in several cancers indications, including pancreatic, mesothelioma, ovarian and many others, for which there is a high unmet medical need. ZW171 has a novel anti-CD3 paratope that was designed to widen the therapeutic window by having low T cell binding and cytokine release in potent tumor cell lysis. We applied to our engineering expertise and the flexibility of the Azymetric platform to modify and perform extensive assessment of various aspects, including valency, geometry and affinity. ZW171 was recently nominated for preclinical development with anticipated IND in 2020 (sic) [ 2024 ]. To speak further to the target, mesothelin is a GPI-anchored membrane glycoprotein that has normal tissue expression that is restricted to the mesothelial cells of the pleura, pericardium and peritoneum. Mesothelin is overexpressed, as I mentioned, in many cancer indications, including pancreatic, mesothelioma, ovarian and others. Mesothelin is known to bind MUC16, and preclinical data has shown that mesothelin plays a role in cell adhesion, tumor progression, metastasis, chemo resistance and the formation of cancer-associated fibroblasts that induce immunosuppressive environment. Mesothelin is a clinically validated target that has been targeted by different immunotherapeutic strategies, including bispecific T cell engagers, CAR-Ts and ADC. Recent clinical data has shown responses in cholangiocarcinoma, mesothelioma and ovarian cancers with a mesothelin-targeted autologous T cell therapy. The illustration on the right shows the structure of mesothelin. ZW171 binds the epitope at the N-terminal domain, as highlighted in cyan. This region is also the binding site for MUC16. The anti-mesothelin antibody is able to block this binding interaction with MUC16. Based on the restricted normal tissue expression and the abundant overexpression in cancer tissue, mesothelin is an ideal profile for bispecific immune cell targeting and the rationale for why we chose this target. Moving to the engineering and screening of ZW171. We have taken learnings from the clinical experience in the field and applied this to the engineering of 171. Specifically, we aim to engineer a molecule with a widened therapeutic window, with enhanced antitumor activity and improved safety profile for the treatment of mesothelin-expressing tumors. As Paul mentioned, protein engineering is a core competency at Zymeworks. And it is this deep experience, expertise, combined with the flexibility of the Azymetric platform, that allows us to screen multiple critical parameters in silico and in vitro in high-throughput fashion. The workflow starts with paratope screening and optimization in silico, affinity engineering, followed by the generation of an extensive panel of engineered antibodies based on valency, geometry and affinity. Based on the strength and flexibility of the Azymetric platform, we can easily make to high purity and yield a wide variety of antibody formats that is typically problematic to produce with other bispecific platforms. The Azymetric platform can be used for traditional expression and screening workflows but also in high-throughput screens where multiple parameters or antibody formats can be screened in parallel. We then screen the panel of antibody geometries, formats and affinities in a series of in vitro assays followed by confirmation of antitumor activity in vivo. Both the in vitro and in vivo screening are performed under high stringency conditions, meaning that we test in conditions that best represent the challenging tumor physiology, including screening in a large panel of cell lines and in established in vivo models which are less sensitive to antitumor inhibition. By screening under these conditions, we are able to identify ideal antibody properties and, if required, refine this in an iterative fashion with more rounds of engineering and screening. One of the first parameters we optimized was the anti-CD3 paratope. The goals around engineering the CD3 paratope were built off the learnings from the field, in particular, where generation 1 anti-CD3 paratopes based on the OKT3 and SP34 paratopes are associated with high-affinity CD3 binding and dose-limiting toxicity related to cytokine release syndrome or CRS. Thus, we engineered the antibody to have low-affinity CD3 binding that results in low cytokine release yet maintains potent tumor cell lysis in, in vitro assay to potentially avoid CRS. In the data presented, we have compared the ZW CD3_1 paratope in blue to a generation 1 high-affinity paratope in purple and generation 2 low-affinity paratope in orange. As illustrated, ZW CD3_1 has reduced binding to T cells, stimulates reduced cytokine release in the presence of tumor cells, yet elicits equivalent and potent T cell cytotoxicity. We believe this profile is ideal to maintain antitumor activity and avoid dose-limiting toxicity related to cytokine release syndrome in patients. Moreover, our CD3 paratope binds a unique and structured epitope compared to first-generation paratopes based on SP34. The structure illustrated is a CD3 crystal structure. Highlighted in yellow is the epitope for SP34, and the ZW CD3_1 paratope is highlighted in green. This distinct binding geometry may provide more optimized engagement and biology for mesothelin and other tumor targets. Additionally, the antibody is cross-reactive with cynomolgus CD3. We also evaluated aspects of the mesothelin paratope, including affinity as well as different antibody formats, and confirmed our lead properties and lead candidate by evaluating antitumor activity in the high stringency established tumor models that have reduced sensitivity compared to co-implantation models. In the established tumor model, the tumor, or tumor fragment in the case of OVCAR-3, is first engrafted subcutaneously and grown to a size of 100 to 200 cubic millimeters. And then the human effector cells or human PBMCs, which include T cells, are added, followed shortly by test article administration to evaluate antitumor activity. Established tumor models can better reflect the challenging tumor structure, environment and sensitivity for solid tumors. This contrast with co-implantation models where tumor and effector cells are administered simultaneously in mice, these models can be more sensitive to growth inhibition and may not best represent the challenging environment of solid tumors. Based on our in vitro screening, we have identified the 2 plus 1 format indicated at center and as the lead format. We then evaluated the impact of varying the affinity of the anti-mesothelin paratopes and identified that mesothelin paratope with affinity C mediated the greatest antitumor activity in the mesothelin high-expressing OVCAR-3 tumor model. We then compared our lead 2 plus 1 format to a different 2 plus 1 format, where the anti-mesothelin paratope with affinity C were Fabs instead of scFvs and observed greater antitumor activity with ZW 2 plus 1 lead candidate. We also compared the lead 2 plus 1 candidate to the Roche 2 plus 1 molecule, which has different affinities for both the CD3 and mesothelin and again showed greater antitumor activity with the lead 2 plus 1 format. This highlights an important point in that not only is a specific antibody format critical to antitumor activity, but even with 2 plus 1 formats that are -- but even with very similar 2 plus 1 formats, there are specific geometry requirements that likely impact the quality of T cell synapse formation and antitumor activity. This brings us to the preclinical candidate molecule, ZW171, that was engineered for enhanced antitumor activity and safety. To go a little more detail on the format, there are 2 anti-mesothelin single-chain variable fragments or scFvs with 1 on the N-terminal domain of heavy chain A of the CD3 paratope and a second scFv on heavy chain B. The reduced CD3 affinity anti-CD3 paratope is an Fab format or fragment antigen binding fragment attached to heavy chain A. A heterodimeric heavy chain assembly is facilitated by the Azymetric platform, as I mentioned, and with mutations in the Fc that also includes the effect IgG1 knockout mutations that aggregate Fc gamma receptor binding. A summary of the design criteria is indicated. Again, just to quickly review, we optimized the mesothelin paratope affinity for increased antitumor activity, reduced CD3 paratope binding to reduce the risk of CRS in the clinic. The antibody format and geometry were optimized for enhanced antitumor activity and safety. And the 2 plus 1 format facilitates avidity-driven, tumor-specific binding and cytotoxicity, thus mitigating the risk for on-target off-tumor toxicity. Additionally, the format was engineered for improved stability. Next, I will review the data that showcases Zymeworks' advantage that includes greater antitumor activity compared to benchmark and potentially improved safety profile. Speaking specifically to the advantage of the 2 plus 1 avidity-driven binding, we observed potent cell cytotoxicity with ZW171 in high mesothelin-expressing cells but do not observe antitumor activity in low mesothelin-expressing cell lines, which act as a surrogate for normal healthy tissue that express low levels of mesothelin. As illustrated in the graph at right, you can see that ZW171 mediates potent T cell cytotoxicity in high mesothelin-expressing cell line, but little to no activity is observed in low mesothelin-expressing cells, thus mitigating the risk of on-target off-tumor toxicities. Further in vitro evaluations in high and moderate mesothelin-expressing cell lines show that ZW171 exhibits mesothelin-dependent cytotoxicity in a panel of mesothelin-expressing cancer cell lines, including lung, ovarian, colon and mesothelia cancer. The W171 mediates high potency in both mesothelin-high and mesothelin-moderate cell lines. Additionally, ZW171 mediates cytokine release and T cell proliferation in the presence of mesothelin tumor cells and not in the absence of the mesothelin target, thus mitigating the risk of peripheral T cell activation and cytokine release syndrome. We then compared ZW171 to clinical benchmark as shown at right, HPN TriTAC or HPN536, that has a single anti-mesothelin single-domain antibody and a single anti-CD3 scFv paratope with an anti-albumin domain at the center for half-life extension. In vitro comparison showed equivalent cytotoxicity with ZW171 and HPN in a high mesothelin-expressing OVCAR-3. However, when we tested in an established OVCAR-3 model, ZW171 showed greater antitumor activity. Due to the different PK properties of the 2 antibodies, ZW171 was dosed once weekly IV, while the HPN TriTAC was dosed daily IP for 18 days. While the doses for ZW171 and HPN TriTAC were not exactly exposure matched, the serum exposure of HPN TriTAC is expected to be equivalent or up to 2x greater than ZW171. And even with these potential differences in serum exposure, ZW171 mediated greater antitumor activity. Similarly, we compared ZW171 in a mid mesothelin-expressing colon cancer tumor model, HCT116. As observed with the OVCAR-3 cell line, we observed equivalent in vitro cytotoxicity in HCT116. However, as with OVCAR-3, we observed greater antitumor activity following treatment with ZW171 compared to the HPN TriTAC at exposure matched doses. PK and exposure analysis from the study confirmed that the exposure of ZW171 and HPN TriTAC were comparable. These data highlight 2 important points: one, the importance of testing in vivo due to the lack of translation between in vitro and in vivo testing; and secondly, the importance of testing in more physiologically relevant established tumor models as the coadministration models may overestimate antitumor activity. We are excited about the opportunity of ZW171 and in treating mesothelin-expressing cancers. Mesothelin, as I mentioned, is expressed in many solid tumors, as shown at left. ZW171 has the potential to be first and/or best-in-class for mesothelin pancreatic, non-small cell lung, triple-negative breast, mesothelioma, ovarian and other mesothelin-expressing cancers. The estimated yearly incidence of mesothelin in patients ranges from 10,000 in mesothelioma to greater than 100,000 in pancreatic cancer. In summary, mesothelin is a clinically validated target with high expression in many solid tumor types that represent a high unmet medical need. Investigational mesothelin-targeted biologics have demonstrated clinical activity in mesothelin-expressing cancers. ZW171 has been engineered for optimal format, paratope affinity and stability with an improved safety profile and widened therapeutic index. This represents a first and best-in-class opportunity for the treatment of mesothelin-expressing cancers. ZW171 shows improved antitumor activity compared to clinical benchmark in mesothelin-expressing cancers. Our next milestones, as indicated, include a GLP toxicology study and IND submission anticipated in 2024. Up next, I'd like to introduce Thomas Spreter, who will present our data supporting our engineering strategies to further address the biological challenges with T cell engagers in solid tumors.

Thank you, Nina, and good morning, ladies and gentlemen. My name is Thomas Spreter Von Kreudenstein, Director of Protein Engineering at Zymeworks, and I would like to shift focus to our next-generation multispecifics and present to you 2 new project and technologies that we have been developing. And as Paul showed you on the slide earlier today, T cell engagers are a very promising class of therapeutics, but there are also a number of challenges that have limited their applications. And we're developing next-generation trispecific T cell engagers to target 2 key challenges in solid tumors, as outlined here as problem #2 and #3. And the first project I'd like to present is aiming to address what Paul discussed here as problem #2. But to better explain the problem and our approach, I'd like to briefly go back to basic piece of biology. So activation of T cells usually requires what is termed Signal 1 and Signal 2. And as shown here in this figure, Signal 1 is mediated by CD3 or the TCR and can also be termed the ignition. And Signal 2 is mediated by a CD28 and can also be termed the gas pedal. Both signals together are critical for T cell activation, sustained T cell proliferation and overall T cell fitness. Now in the case of a T cell activation by bispecific T cell engagers, the T cell engager will only target CD3 or Signal 1 and rely on natural B7 costimulatory ligands expressed on the tumor for optimal T cell activation. The problem is now that the number of solid tumors, particularly cold tumors with low T cell infiltration, are not going to express costimulatory ligands. And particularly in those settings, because of the lack of costimulation, bispecific T cell engagers are less active and are not able to induce sustained T cell activation and proliferation. And so to address those limitations, we're developing costimulatory trispecific molecules. They're able to target both CD3 and CD28 to induce better T cell activation. That didn't quite -- so as you are aware, CD28-targeting T cell engagers are truly also being developed by others. So I'd like to spend a minute to better explain our approach and why we believe we have a differentiated approach. And I'll ask you to focus on the left. There was supposed to be an animation. So I'm going to be talking on you on the left side first. So what you can see here in the figure on the left is that optimal T cell activation is not sufficient -- it's not sufficient to just provide Signal 1 and Signal 2, but what is critical is to balance Signal 1 and Signal 2. And if a T cell only -- as shown here on the left side of the left panel, if the T cells receive one signal, this leads to T cell energy, T cell death and lack of sustained T cell proliferation. At the same time, if the T cell receives 2 strong signals, as shown here in the middle, it leads to T cell overactivation, T cell dysfunction and excessive cytokine release. So for optimal T cell activation, the balance of Signal 1 and Signal 2 is critical, as shown here in the panel in the green in the middle. So to develop an effective and safe costimulatory T cell engager, we believe it will not be sufficient to just deliver 2 strong signals like, for example, in a combination approach, but it will be critical to balance CD3 and CD28 stimulation, or in other words, to balance Signal 1 and 2 in 1 molecule. So our aim here is to develop costimulatory trispecific with a balance of Signal 1 and 2 to be able to provide optimal T cell activation in 1 single molecule. Now I'm focusing on the right here. That was supposed to come up as an animation, as I said, so I'm going to talk about the right panel here now. So to achieve this, we went back to what's known about T cell engager engineering and our own experience. And we know from the field of T cell engager engineering and from what Nina has shown you from our 171 program, that both geometry and affinity are critical criteria to develop an effective T cell engager. And we believe that this will likely be equivalently important for the design of any trispecific molecule. So with this in mind, we have developed a novel screening approach, which allows us to interrogate and screen multiple different geometries and affinities, and we believe this will provide a differentiated approach to fine-tune T cell activation by CD3 and CD28 or Signal 1 and Signal 2 in a single molecule. So the first step, we've selected and engineered our own CD28 agonist, and we have specifically selected what's termed a conventional agonist. Conventional CD28 agonist binds to a different epitope than the known CD28 superagonist, which have had significant tox liabilities in the clinic. And we have further engineered our own CD28 agonist with a broad range of affinities to be able to fine-tune CD28 activation in the trispecific and to prevent potential CD28-mediated toxicities. And in addition, the CD28 agonist can be formatted at scFv or Fab, which allows for flexibility in the design of the trispecific. The animation works here. So after selecting the CD28 agonist, we designed the costimulatory trispecific, as shown here in the top panel. And this is just a representation of a subset of formats we're actually screening. But what you can see in this panel is that the CD3 and the CD28 Fabs or scFvs are attached at very different positions. And what is unique in our approach, we believe, is that by doing this, we are able to interrogate different CD3 and CD28 but also CD28 TAA and CD3 TAA geometries, which ultimately might result in a better balance of Signal 1 and Signal 2 in a more optimal T cell activation. The figure on the bottom left shows the design space of CD3 and CD28 affinities we are able to screen with the CD3 affinity on the y-axis and CD28 affinity on the x-axis, and each of the dots represent one format. In the second step, we picked the lead formats and further fine-tuned CD3 and CD28 affinities. And this is illustrated in the insert in the graph where we show various CD3 and CD28 affinities in one format. So in this illustration, I should say, we're screening one TAA, but we routinely test different paratopes targeting the same TAA in the same screening. So the different variants are expressed in a high-throughput workflow and screened for activity in a T cell-dependent cytotoxicity assay on different tumor cell lines. And the figure on the right shows a representation of an activity screen. And as you can see, the different variants show significant differences in potency and a number of them being significantly more potent than the benchmark bispecific. At the same time, our aim was not to just make more potent T cell engagers, but what we wanted to do is to develop trispecifics that are active under conditions where bispecifics have shown limitations. And as I discussed in the slide earlier, those limitations particularly include cold tumors with low T cell infiltration or low effector to tumor cell ratio and also conditions that reflect more longer-term T cell proliferation. So to do this, we developed a screening assay at low E to T ratios with longer-term incubation, which would better reflect conditions in solid tumors where CD3 bispecifics generally show low activity. And as you can see in the left graph for day 3, under those more stringent conditions, the bispecific benchmarked. But actually also, trispecifics were not active. But when we take this assay out to 7 days, we see activity of the trispecific but no activity of the bispecific control. In addition, what you can see on the right, we also looked at T cell proliferation and the same assay. And the trispecific is significantly better at long-term T cell expansion, which is in -- overall in line with the mechanism of costimulation driving enhanced T cell proliferation. So overall, this shows the potential of our costimulatory trispecifics to be active under conditions that are more relevant for low E to T and cold tumors. In the same assay, we also looked at cytokine release. And as you can see in the top panel, the trispecific induces increased levels of cytokines. And in particular, IL-2 levels are significantly increased for the trispecific, which really is a sign for enhanced CD28 costimulation and is in line with the mechanism of action for a trispecific. As a further step, a critical criteria in our design of the trispecific is obviously safety. And in particular, we want to avoid nonspecific activation of peripheral T cells. So just this, we used cultures of T cells only and tested cytokine release as a sensitive readout for T cell activation. And as shown in the graph on the bottom, the lead trispecific does not induce any nonspecific T cell activation without the presence of tumor cells. So to summarize, a key limitation of bispecific T cell engagers in solid tumors is lack of durability of responses and low activity in cold tumors. And costimulatory trispecifics targeting both Signal 1 and Signal 2 have the potential to reinvigorate cold tumors and potentially show prolonged responses. We are developing a novel and differentiated approach of modular geometry and avidity screening of trispecifics, which allows us to optimize and balance T cell activation by Signal 1 and Signal 2 or CD3 and CD28. Further, we have shown that the trispecifics have superior activity to bispecific benchmarks and low effector to target ratio and induce enhanced T cell proliferation. In addition, the lead costimulatory trial has been showing no activation of T cells without presence of tumor cells as a critical criteria for safety. And currently, we have different programs at lead selection stage. So the second program I'd like to share is our trispecific T cell engager with checkpoint inhibition. And this program and technology is attempting to address problem #1, as highlighted here about by Paul earlier. And to explain this observation -- the observed clinical limitations, I'd like to briefly go back to basic T cell biology again. And what you can see here, what happens after a T cell activation or what happens after T cells have been activated is rapid upregulation of PD-1 and PD-L1. And that PD-1/PD-L1 interaction reduces and inhibits further T cell activation. This is a natural mechanism that prevents autoimmune reaction and is illustrated on the left panel. In tumors, as shown on the right, the same mechanism will inhibit T cell engager-mediated tumor cell killing and T cell activation. And in the clinic, it has been shown that in inflamed tumors with a high content of dysfunctional and exhausted PD-1 positive T cells, bispecific T cell engagers have reduced or no activity. In addition, after dosing, T cell engager activity leads to a rapid upregulation of PD-1/PD-L1 on tumors and T cells as part of the mechanism of action. And this can lead to secondary resistance mechanisms of T cell engagers in the clinic. So to address these clinical limitations, we're developing trispecific T cell engagers with checkpoint modulation, one molecule. So we've designed a trispecific T cell engager that engages CD3 and the tumor-associated antigen but can also bind and block PD-L1. And the molecule here shown is just one representation of the structure. We have a number of different geometries that I'll show you in a later slide. But there's a few things that we believe make this molecule and the mechanism of action unique and differentiated, and I want to point them out in a bit more detail before I walk you through the data. So the first mechanism of action of this trispecific is blocking the PD-1/PD-L1 action in the synapse, which is shown here on the left. And this restores the original T cell activation, but you could argue this can certainly be achieved by just a combination of the T cell engager and an anti-PD-1/PD-L1 antibody. The unique mechanism of action of the molecule is that in addition to checkpoint modulation, it also creates further avidity by engagement of PD-L1, which is expressed on the tumor cell, but is -- which is also upregulated on T cell after activation. So PD-L1 engagement functions like an additional adhesion molecule, strengthening and stabilizing the T cell tumor cell synapse interaction. And this leads to enhanced T cell activation and tumor cell killing. So overall, we believe that the molecule will function by dual MOA, which makes this unique and has the potential to be superior and significantly different to the combination therapy. So for the design of the trispecific, we used the natural PD-1 domain as an anti-PD-L1 binder and PD-1/PD-L1 antagonist. The figure on the left shows a crystal structure of natural PD-1/PD-L1 with PD-L1 on the left and PD-1 on the right. And we have used the PD-1 domain shown in the box here and engineered it for higher affinity to PD-L1. So it functions as a natural PD-1/PD-L1 inhibitor equivalently to an anti-PD-L1 blocking antibody. As shown on the bottom graph, we further engineered this PD-1 domain with a range of affinities to be able to fine-tune efficacy and tolerability of the trispecific. And using this engineered higher affinity PD-1 domain as a natural PD-L1 antagonist, it allows us -- allows for significant flexibility in format and geometry design of the trispecific. And the domain overall has a modular -- a more modular nature than scFvs or Fabs and can be easily fused to different termini of a T cell engager. So this flexibility allows us to design multiple different trispecific formats, as shown here on the right. And as I discussed previously, we believe that geometry and affinity will be critical criteria for the dual mechanism of action to achieve both high avidity and functional PD-1/PD-L1 blockade. So to select the best trispecific, we used the screening approach to fine-tune both mechanism of action. In the screening approach, we are interrogating different formats and geometries, but we're also testing different PD-L1 affinities. And as illustrated here on the left, the trispecifics are screened in a T cell-directed cytotoxicity assay. And again, as you can see, the different variants exhibited significant -- different activities. And for the lead format shown here on the 2 graphs on the right, we consistently see a large gain in potency over the bispecific control with easily over a hundredfold improved potency for the 2 leads selected here. In addition, you can see the lead molecules also show superior potency to benchmark competitor bispecifics. So to better understand the reason and the mechanism of actions for this large potency gain and to also ask the question if the same activity could just be achieved by a combination of a T cell engager and an anti-PD-L1 antibody, we ran a series of more mechanistic assays. And the first assay we looked at, shown here on the left, looks at the ability of the trispecific to just block PD-1/PD-L1 interaction and restore effective T cell activation. In this assay, we have a high expression of PD-1 and PD-L1, which results -- which reduces the T cell activation. And we measured the extent of T cell activation under those suppressive conditions. As you can see on the left, the bispecific is able to induce some T cell activation, but overall has fairly low activity. And then as you can see in the middle, adding the anti-PD-L1 antibody restores a lot of the signal and leads to higher T cell activation. But as you can see on the right, the trispecific has significantly higher activity. And interestingly, it is able to not only restore T cell activation by blocking the PD-1/PD-L1 action, but it's also able to further increase the T cell activation. In the second assay, we look at the ability of the trispecific versus the bispecific to bridge T cells and tumor cells, and this can be viewed as a surrogate measure of how strong or stable the T cell-tumor cell interaction mediated by the trispecific is. And what you can see here on the right panel is the amount of cross-linked T cells to tumor cells. And the bispecific, as expected, is able to cross-link T cells and tumor cells, but the trispecific is significantly more potent in cross-linking, which again suggests that it's able to induce a more effective and more stable T cell-tumor cell interaction. So those 2 assays taken together support our hypothesis that the lead trispecifics are functioning by a differentiated dual mechanism of action of both blocking the PD-1 and PD-L1 action, but also enhanced avidity, which leads to superior T cell activation. So we wanted to better understand if the trispecific can be active under more relevant conditions of suppressive solid tumor microenvironments. And as discussed previously, one key limitation of T cell engagers is lack of activity in settings of exhausted or dysfunctional T cells. And to better reflect this condition, we established assays with exhausted T cells and testing the first step, just binding of bispecific and trispecific to naive T cells versus binding to exhausted T cells. So the graph on the left shows binding of the trispecific and a benchmark bispecific to naive T cells. And as expected, it shows equivalent binding of the trispecific and the bispecific benchmark to T cells. And binding for both is then expected around medium 2-digit nanomolar range. One thing to note here on the graph on the left is that we also tested binding of an anti-PD-L1 antibody, and it shows no binding to those T cells. And as expected, naive T cells do not express PD-L1. So the graph on the right shows binding to exhausted T cells, and it interestingly shows significant differences between the bispecific and the trispecific that I'd like to call out in a bit more detail because we believe this is very relevant for the differences -- differentiated activity and mechanism of action of the trispecific. So you'll first notice that the clinical anti-PD-1 antibody, atezo, shows binding to the exhausted T cells and really confirming that exhausted T cells are expressed in PD-L1. This has been described previously but not -- might not be as widely appreciated. Second, as you can see, the trispecific shows significantly higher binding to exhausted T cells than the bispecific control. And this really suggests that trispecific is able to bind both CD3 and PD-L1 on T cells and achieve a higher binding affinity and decoration than the bispecific. Lastly, what I would like to point out is that the bispecific control seems to overall show reduced binding of -- to exhausted T cells when compared to naive T cells. This is overall in line with previous observations by others that activated T cells actually downregulate the TCR and CD3 as a mechanism to reduce T cell activation. And overall, the trispecific seems to have the ability to better bind to exhausted T cells, which could be critical to retain activity in settings of suppressive tumor microenvironment with exhausted T cells. So the next step, obviously, we wanted to understand if the differences in binding will translate into cytotoxic activity. And as you can see in this graph, the assay uses -- this assay uses exhausted T cells as effector cells as opposed to naive T cells. And we observed a significant increase in potency for the trispecific versus the bispecific control. And in addition, we also observed that the combination of the bispecific control and the clinical anti-PD-L1 antibody, atezo, did not show improved potency. So in summary, the observations support our hypothesis that the trispecific has the potential to be active on conditions of suppressive tumor microenvironment where bispecific T cell engagers have shown limited activity. In addition, the superior activity of the trispecific supports the unique activity through a dual mechanism of action of both checkpoint blockade and also enhanced T cell tumor cell avidity. So we were very encouraged by this activity and the unique properties of the trispecific, but one obvious question is how broadly applicable the superior activity and enhanced potency actually is. So to better evaluate the broader application of the TriTCE, we screened a larger panel of cell lines with different levels of tumor-associated antigen and PD-L1 expression. And as you can see here on the left, the cell lines have varied expression levels of the tumor-associated antigen from medium to high and similarly for the PD-L1 expression. On the right, you can see the activity in the T cell-dependent cytotoxicity assay, and we observed a similarly large potency gain across all cell lines tested, fairly independent of expression levels. What we observed in addition to the data that I'm showing here is that after incubation with a T cell engager, PD-L1 is rapidly upregulated on most of the cell lines, and this rapid PD-L1 upregulation might contribute to the broad activity across multiple different cell lines. We believe this is a unique mechanism of our trispecific, and it suggests that the trispecific mechanism of action might be broadly applicable and potentially be less susceptible to PD-1/PD L1 mediated secondary resistance mechanisms. As a next step, we wanted to understand if the superior in vitro activity translates into in vivo, and we tested the activity of the trispecific in a humanized mouse model. The model used an established tumor model with human PBMCs injected and a dosing start once tumors are established, so very similar to the models that Nina has shown you for our 171 development. And as you can see here, the bispecific in this established, more difficult model shows only very little tumor control, whereas the trispecific is able to control tumor growth. And we observed 6 complete responses out of 8 mice. And interestingly in this model, the combination of bispecific control and the anti-PD-L1 antibody in T cell did not show improved activity, which again suggests that trispecific has a differentiated activity to the combination therapy and supports our hypothesis of the dual mechanism of action. So to summarize, we know that PD-1/PD-L1 upregulation in tumor macro environment is a key clinical limitation of bispecific T cell engagers. And to overcome these limitations, we're developing novel, differentiated trispecific T cell engagers with a dual mechanism of action, our full checkpoint inhibition and avidity-driven T cell activation. We have demonstrated potent and differentiated activity in vitro and in vivo, and high activity with exhausted T cells. And the activity supports our dual -- novel dual mechanism of action of both checkpoint inhibition and avidity of the trispecifics. In addition, the data suggests superior activity of the trispecific to combination therapy of bispecific with checkpoint inhibitor. So overall, we believe that the trispecific T cell engager due to the differentiated dual MOA may provide unique opportunities for targeting microenvironments with exhausted T cells, and to -- and potentially also provide more dual responses. And we have a number of lead molecules in -- number of lead molecules in at least selection stage. And with that, I thank you for -- and I'll hand it over to Paul.

So I'm going to just wrap up with like a 10 minutes sort of wrap-up summary. And what you've hopefully heard is the significant progress that we have made with our ADC and multi-specific platforms, with various product candidate opportunities. But before I recap on those, I wanted to take a step back and just again, frame in context our overall focus of -- which is to address unmet medical need. So we're really developing -- so that we're developing modules that will make a difference. So we're going to really focus on indications with the worst patient prognosis. We then intend to apply our -- either our multi-specific platform, our ADC platform, against targets from those indications. And we'll apply the technology that best serves that target, the biology associated with that disease, and that tumor type and the disease complexity. From there, we pick molecules that with the most favored profile for development and with the attention that they will ultimately enter into the -- into earlier lines of treatment and have the potential for accelerated approval. So that's always at the back of our mind is thinking about the tumor target in the disease. As we shared, as the team shared with you, we're really uniquely placed at Zymeworks to leverage either world-class ADC or world-class bispecifics or multispecifics. Our strategy, as you heard on the antibody drug conjugates, is to think of the molecule totally. Think about the antibody properties in format, payload, linker conjugation and DAR. And then likewise, on the multispecifics, we really can fine-tune and engineer the molecule to get the right features, multispecific features ingrained into that molecule to address what the challenges is. Now some of the diseases that tumor types that you heard is described are listed on the left, and so are consistent with our thinking of trying to treat patients with the poorest outcome, poorest overall survival. Now getting back to the programs that we talked about today and how this integrates with our planning and vision in the near term and the longer term. So in the context of the ADC programs, our next -- our next IND will be 191, the folate receptor targeted topoisomerase program, topoisomerase ADC. And that targeting folate receptor is a target, as you heard, is validated, as responsive to ADCs clinically. But we believe with an antibody optimized for internalization features and the use of our propriety TOPO payload, we can develop a best-in-class molecule for folate receptor alpha expression tumors that can also expand into the middle range expressing middle-range folate receptor expression tumors as our in vivo modeling demonstrated to you. In the context of multispecifics, our next IND will be the 171 program, a 2 by 1 or 2 plus 1 mesothelin CD3 trivalent T cell engager. As Nina described, we feel mesothelin is an excellent target for a bispecific T cell target approach. The target is highly expressed in a range of tumor types, including those hard to treat with -- hard to treat and with poor overall survival. A lot of engineering has gone into the design of that molecule to really optimize all components, the CD3 affinity and epitope, to limit cytokine release together with bivalent targeting of mesothelin and the optimal geometry to increase therapeutic window of activity versus normal tissue. So that you get a good window of activity between tumors and normal tissue. Importantly, we maintain excellent antitumor activity of that molecule. That was a real high bar for us to make sure that we maintained activity in established tumor models, and we see that. We then look behind that, like in our research engine, what we've got coming behind that, Stuart and Jamie described additional TOPO ADC-based product opportunities. So they are the 2 programs we decide to describe were the 251 targeting Glypican-3 and 220 targeting NaPi2b. And these were again engineered through careful selection of antibody payload, and we're thoroughly really encouraged by the profile of each molecule and the cancer indications they serve. And these molecules are now completing preclinical development and could serve as future IND candidates. And as Stuart alluded to, we have other programs behind that, that we've also identified tumor antigens for in cases of deploying the biparatopic also with the TOPO platform. Moving to the multispecific side, there's an area that we think is still great opportunity is in the redirected T cell killing. Major challenges remain for many tumors that are not expected to respond to just a CD3 bispecific, either due to the intrinsic fitness of T cells in the tumor microenvironment or the interplay of immune inhibitor molecules. And where this has led to -- this in the field has led to combination approaches where people are thinking about adding co-simulation along with primary CD3 engagement through a combination of antibodies. We think we can do it in the one molecule and the benefit of the design and the flexibility of Azymetric allows you, as Thomas explained, to really fine tune both signals so you can get that right balance. In addition, we also have employed this also for checkpoint inhibitor, where we were able to, again, demonstrate proof of concept, where we could actually see enhanced activity and synergistic biology by incorporating PD-1 binding. So moving forward from here, what we -- we're really are well placed to sort of utilize the technologies for more targets. And we really actually have to think about -- also we will think more about more novel targets. We've initially focused on targets somewhat validated. But as we go into other tumor indications, we will select more novel targets. And potentially, we'll select them in combination to further potentially advance therapeutic window there. And this may include targeting that enables gated activation to increase therapeutic window or studies that combine simultaneous tumor cell killing and reversal of immune cell population repression, either through a single blocking or perhaps direct license. To this end, we intend to continue involving our payload strategies as well and the competitive edge there within our ADC platform. And likewise, in the context of multispecifics, we can also consider additional effector cell populations or more specific effector cell populations through avidity-driven selectivity. As Thomas showed with the tri-TCD checkpoint where we could select the PD-L1 population or really drive that targeting of that population. But that could also be extended to other immune subsets through a similar approach that we described. In the near term, then we have 4 exciting molecules, 2 of which have already been identified, targeted for IND filing. And then behind that, as we mentioned, we've got a machine or an engine that's working on additional molecules. And we think that these will continue to fuel our clinical pipeline. And also, we will continue to partner opportunities moving forward. And so this is just to sort of reiterate that partnering will remain a part of Zymeworks. I think our focus moving forward will be deriving more value from product candidates than platform, not necessarily just purely platform capabilities but we will continue to also leverage opportunities, partnering opportunities through that as well. And then to finalize, I'd just like to finish with this slide that really captures our 5x5 vision for Zymeworks, building on the company's proven success in leveraging bispecific and ADC technology, ADC technology to develop therapeutics and the advances we've made in the platform, technologies and tools based on the team's capability and infrastructure. We envision transitioning from a company focused on a select product pipeline and platform technology advancement to one focused now on accelerating more program products to the client. We have one -- we have a goal of 5 new programs in the clinic in 5 years. We shared with you 2 of those programs that have already been identified as INDs, one in ADC and the other a novel bispecific, each fit for purpose designed to prove an existing treatment options for patients suffered from hard-to-treat cancers. We're uniquely placed to have both world-class ADC and multi-specific platforms to leverage and are well placed to meet our goal with truly differentiated molecules. With that, I'll let Neil close out.

Great. Thanks, Paul. And thank you, Nina, Thomas, Stu, Jamie for presenting the exciting work that's coming out of our early research and development groups. Additionally, before we start Q&A, I'd like to thank everyone who tuned into the webcast and participated live and share some closing thoughts. With yesterday's announcement of our partnership with Jazz Pharmaceuticals and a host of potentially first and best-in-class ADCs and multispecific therapeutics based on our cutting-edge platform technologies, these are exciting times for Zymeworks. And it's fair to say that the combination of exciting and dynamic partnerships, coupled with world-class science truly brings Zymeworks back to its roots. By leveraging our core competencies in antibody engineering, to create the right antibodies as you've heard today, and coupling these antibodies with our leading and proprietary platforms, we are focused on developing meaningful, novel therapeutics that we believe will deliver meaningful clinical benefits, when these programs enter the clinic and deliver real shareholder value. We are focused on our 2 key research areas, ADCs and multispecific antibodies to drive pipeline expansion. And as I noted at the start of the session, going forward we are fully committed to the use of partnerships and collaborations as a key strategy to bring these novel therapeutics to patients, setting up success for years to come. With that, I'll open up the floor to Q&A, where we will have Paul and our research team leads join us upfront and available to answer questions on today's presentation. As a reminder for those listening into the webcast, if you have questions, please e-mail ir@zymeworks.com and we will do our best to answer your questions. Thank you, everyone.

Excellent. So we'll open it up to questions from the floor here in New York.

Good morning, everyone, and thanks again for taking the time to host this event. Charles Zhu from Guggenheim Securities. Maybe one scientific philosophy-type of question. When you're designing some of these next-gen molecules, whether it's an ADC or a T cell engager, how should we think about the relative contribution of potential clinical activity from the various components? Or specifically from, let's say, the antibody itself? Where I'm coming from with this is when we look at zanidatamab versus ZW49 as an example, zani, has without the payload, does have activity. So from that perspective, how should we think about that with your emerging ADCs? And similarly for ZW171, given where you're binding on mesothelin, is there contribution of activity through the blockade of MUC16 activity as well?

Thanks for that question. So I understand what you're getting at is that -- I think a lot of that depends on the dose that you're going to be dosing your molecule, okay? So there -- we do believe that there's going to be certain features that are additive to the sort of primary mechanism of action of a T cell engager. So the potential that the second the specificity against the target could have in and itself biology that you could also be getting a benefit from, right? But that -- in the case of -- and you asked the question specifically about 171 in mesothelin. And there, we do believe that we're binding an epitope and because we're doing it with bivalent strategy there, there actually is a potential that we could interfere with binding there with it's ligand, okay. Now we're optimistic that we will produce this molecule quite high, because it has the reduced CD3 activity. But that will -- we'll have to -- we see how that transpires as we clinically develop it. So there is a potential there with that design in that particular molecule because it is a block that antibody does block -- bind to an epitope that blocks engagement and it's bivalent that we may actually, in that case, have the opportunity to see that. That may not always feature in all the programs that we have. It may be that, that mechanism is not necessary. If you can get the power of the other modality that you're trying to exploit in the design of your molecule. So, yes, I don't know maybe on the 171, if there's anything else?

Yes, you said well. Thanks.

Akash from Jefferies. So it's kind of interest -- like ZW49 was a really elegant molecule, right? Like -- but you were going for something very different. You're going at a DAR2 higher internalization, nongreasy. And it's -- you've completely flipped the switch with these next-gen ADCs. Just conceptually speaking, like you're trying to avoid the eye toxicity, but you've avoided the liver, you've avoided the lung toxicity, which is a big deal. Why not take, let's say, ZW49 and then just put a TOPO toxin and put a DAR2? Why go to a DAR8 and making it greasy, given like you probably are still increasing net exposure either way?

Yes. No, we think a lot about the design of our ADCs. And -- with our programs going forward, we do think totally about the ADC, right? We think about the antibody, we think about the payload and how we attach it and how that total molecule will give us the effect that we want. And I think at the time, depending on the time frame of when we've got our knowledge base, we will design the molecule to the best of our knowledge at that time and what information we have. And so I think moving forward, we're attracted with the TOPO platform for these targets. We think ZW49 is a great program itself with the design of that molecule, but we kind of consider them independently when we're looking at different targets. I guess I'll let -- I mean I don't know if Stuart or Jamie, want to elaborate on that?

Yes. I think we talked about this maybe before Akash, too, and there is this dichotomy exists in kind of the hydrophobicity and the greasiness. So on antibody, you probably want something that's a lot less greasy, so you have a lot of more antibody-like properties of your ADC. And we really do think we have that with ZW49. And that was infused to the topoisomerase 1 inhibitor platform. On antibody, we wanted to have as antibody-like properties of our final ADC product, but now it's the payload when it's released that happens to be hydrophobic, right? And that can then permeate neighboring cells and have a bystander effect. So I would call them kind of like tweaks as opposed to like wholesale changes in philosophy. It might seem like that a little bit. But I think the mechanism of action of TOPO1 versus [ MTI ] really will dictate kind of like where we apply this. And we think that the 191, the 251 and the 220 are really good places to apply TOPO.

Right. So just to follow-up on that. So ZW49 and then the Exelixis tissue factor ADC, both of them showed eye toxicity, right? And you could say maybe that's -- to me, it seems like it's probably the toxin, but it could theoretically be the linker. And it also could be an on-target effect, right, like [indiscernible], even with the antibody itself was showing eye toxicity. But you're going to -- if you're using TOPO as a toxin, how much do you think that was because of the [ ZLA ] toxin being, I think, MMAF derived? And so if you look at the ImmunoGen compound that you're competing against, that also has eye toxicity as its most common AE, and it seems like it's an on-target effect. So when are you guys going to -- like do you think you're going to have lower eye toxicity with the compounds you're taking forward because of the Topo1 toxin, or we just don't really know what's driving this right now. And I think specifically outside of the linker, higher internalization might also be connected to higher eye toxicity. So why was it important for you guys to incorporate high internalization here?

Yes. I think the high internalization is something that was attractive for these targets due to their target expression profile. With mirvetuximab soravtansine, that soravtansine payload DM4 has prevalent ocular toxicity across regardless of what antibody you conjugate it to, regardless of what target. So we really think that, that is a platform effect of that payload. So I don't think it will be a -- a unique feature of any ADC that targets folate receptor alpha. The 2 other kind of clinical molecules that have data for folate receptor alpha are MORAb-202 which doesn't have ocular toxin as other toxicity profile. And then the STRO molecule, STRO-002 and they don't observe ocular tox. They have quite a bit of neutropenia. So different tox profiles. I don't think fully receptor alpha is going to be driving ocular tox. And also the Camptothecin class of molecules, whether it's DXd in HER2 or any of the other clinical programs or Trodelvy. There hasn't been observations of ocular tox. So there is a class effect, but there's also some of these design features that we're really conscious of that can contribute to that. Because we also don't want to totally shift the tox profile to something that's unmanageable and problematic. Hopefully, that helps.

I have a related question. David Martin from Bloom Burton. When you were looking at the efficacy of 191 compared to mirvetuximab, you're comparing head on to that antibody. But when you went to the toxicity comparison, you compare it to your antibody with DXd toxin on it, why not compare to mirvetuximab on the toxicity front?

I think, Stuart, you're best placed.

Yes. The mirvetuximab soravtansine kind of like tox profile is a pretty known entity, right? We know how it's performing in clinical trials, and we know how it performed preclinically. We know the doses where they see toxicity, and we know the tox profile. So that was kind of a known entity.

So would you say comparing across you are less toxic or equal toxic or...

I think it's -- we're in preclinical models. So what we can say is that we think that in our preclinical models, we compete really well with the DXd technology, which is in the same class of molecule. So we thought that was a much more important comparison than saying -- the other thing, too, is that mirvetuximab soravtansine sees ocular tox. And there's really, really poor models of ocular tox preclinically. So we could have dosed the monkeys with mirvetuximab soravtansine, saw a tox profile that actually isn't consistent with the clinical tox profile and said, well, we're somewhere in the range sort of thing. So we thought the comparison we made was relevant.

Okay. Second question is more high-level platform related. With the ADCs, you talk a lot about bystander effect. Is there any bystander effect you get with the T cell engagers or is it something that will be missing from that therapeutic class?

Yes, I can take that question. So I think that, that is one -- the feature of T cell engagement, an attraction of that technology in one principle is that you will only hit the cell that expresses the target and you will share the cell that doesn't have it, right? So I think on direct engagement that you will have the -- you are focused on the target cell that you co-engage the T cell with that cell, right? But I think what really attracts me to the platform technologies that Zymeworks is, I think by adding in other features into your molecule, they may lend you some benefit of cells that are in the tumor microenvironment. So if you can actually start resurrecting T cells, PD-1 antigen positive T cells that recognize tumor antigen TILs. They may -- by expanding that, they may have actually intrinsic ability to take out more of the tumor. That's somewhat hypothesis in print, but in principle, that could be an extension of the sort of primary mechanism of action of T cell engagement and activation.

This is Ken Shields from SVB Securities. Congrats on these early programs. I'd say, relative to these early developments, how committed are you to ZW49? And if you think the clinical profile you've seen with the drug is competitive and compelling.

Paul, why don't I take that one? Great question. Thank you. We are fully committed to ZW49 going forward. We believe we have an exciting compound, and we look forward to advancing it as we move forward. I think the additional capital and resources that we have following yesterday's announcement open up some great new opportunities to advance our emerging pipeline and that definitely includes ZW49.

This is James Shin from Wells Fargo. Just wanted to go back to 191 real quick. You mentioned the cysteine conjugation. Is that a Camptothecin linker? And then in your nonhuman primate studies, can you clarify and state whether or not you saw any ocular neutropenia? And I ask because, I believe Camptothecin linkers are sort of related to some of the ILD and neutropenia issues. So I just want to make sure that they're not. And then what is your assay that you've determined for FR alpha high? Because right now, I believe the 2 comparisons you made to 002 and mirv you're using different IHC based assays. So I just want to get your thoughts there.

Yes. There's quite a few questions there. I think most will fall on the ADC team. Yes. I think we think about a lot of those features about the IHC. That's something you mentioned about IHC, I think, obviously, we're using -- I mean, I'll let Stuart and Jamie can elaborate on that. But I think there were -- obviously, we're just losing side by side, and we're using relatively assays that give us a sense on the relative expression of those targets in those preclinical models. As we move forward into clinical development, we will be thinking about a select and a specific biomarker strategy to support that, that we'll wire into our Phase I study when we get there. But I think regarding the more specific questions on the technology and details, I'll hand that to -- I think Stuart or Jamie should take that.

Sure. I mean I can take the previous question, Stuart may be want to take IHC. So yes, the linker, as we discussed is a peptide-based linker. It is created by proteases, and I would include Camptothecin among those. I sort of shy away from just focusing on Camptothecin, that's accurate, I think. In terms of whether or not we saw it in ocular tox, we did not see any ocular tox in the NHP study.

And then just to be really clear about the IHC, we're not using the Roche Ventana assay and PSQ scoring in internal kind of research-level IHC scoring. So it's -- think of it as relatively high, right? We're not using the strict scoring that ImmunoGen uses.

Stephen Willey from Stifel. So I guess with the T cell engaging bispecifics, you kind of talked about some of the problems you're trying to overcome in the solid tumor space. I guess it would seem conceptually that kind of problem 2 and problem 3, those concentric circles are kind of overlapping. So I guess, how do you think about whether or not you want to go with costim feature or with a checkpoint inhibitor. And then maybe just on the TCE side again, how comfortable are you guys with some of the pharmacologic modeling for selecting an appropriate starting dose specifically for solid tumors? I know some of this has been trying to work out in and a lot of heme malignancies, but yet we've seen a lot of companies kind of get locked into these multiyear dose escalation Phase I protocol. So how complicated is some of the modeling with respect to selecting a starting dose for solid tumors?

Sure. So I think your first question was where would you go with the costim or would you go with the checkpoint inhibitor? And I think there -- we're encouraged with both the preclinical profile we're seeing for both. We are going to be continuing to do studies with those modules. We've actually got tumor associated antigens that we've -- that we think are attractive for both of those approaches. And so that data will kind of drive -- will be data-driven on which one we take forward. But we're very excited with both. I think in particular, the PD-1, just it's got some striking in vivo data that we've already seen and that profile across multiple cell lines is really eye-catching. But we think the CD28 program as well could be very beneficial. And of course, we can learn from what others are doing in combination and see as there are certain tumor types. And again, it may really come down to that. There's going to be certain tumor types that will benefit from one over the other, on that one. So maybe we'll take that question and elaborate on that, Nina and Thomas anything?

Yes. One thing to add on the target, I think obviously, it has to fit with the biology and the safety profile, particularly. Particularly if we think of the CD28 costimulatory, so that's an important feature. But as we mentioned, we are focused on indications with high unmet medical need and maybe safe for a few indications that are cold or hot. Most tumors actually have a continuum. And so I think as long as you pick the target wisely, these approaches can be used really for any indication. And then it comes down to patient selection and setting up your clinical trial appropriately.

Yes. And then on your second question, I mean, that's something I've been working on T cell engagers in my career and seeing that evolve, how you design the starting dose. And again, there's a lot we can learn. And I think nothing too much. But we can think about those novel approaches, and it tends to be more of a [ novel ] approach that you use. But there's increasing, I think, models and in vitro and assays that you can deploy to sort of get a logical starting dose. Of course, that has to be -- you have to run that through the FDA. But I think we will really be as safe as possible. But designer study is a way that we can get into the therapeutic doses as soon as possible.

Yigal Nochomovitz from Citi. I just had a few fairly specific questions on some of the data that you presented. So I was intrigued by the experiment where you compared 171 to the HPN536 molecule where you really didn't see any difference on cytotoxicity, but a substantial difference on tumor reduction in vivo. So I was wondering, if you could expand a little bit on what you think is different about 171 that is providing that interesting signal in vivo but not cytotoxically?

Yes. I mean, I -- we've been really obviously excited with that data and also trying to also put our -- think about it, because it is quite startling the difference that we see there. I think if you looked as we develop that design of that molecule earlier in the slide deck, Nina shared that, that geometry, that format is actually really important. So the way that we've got the 2 binding domains for mesothelin in that is actually also important. So although we didn't see activity with the HPN molecule, we actually didn't see activity with other formats as well. And I think what it might be is what we like -- we like to think is that we've actually struck just the right molecule structure for mesothelin, and that's reflected then in the functional -- the really strong anti-tumor activity we see in those models. I think that's -- it's more -- we just hit the right format that we believe. And that whether it's the strength of the synapse and the ability to get into the tumor because it's an established tumor model, the right balance of the T cells, it just seems to have really remarkable activity. And as you observed, as we shared, we see it in more than one model. So it's not just an out of pipe of one model, we see it repeated. So I think it's just, again, we like to say -- we believe that with the Azymetric platform and that ability to screen different formats, we could plug in FEBs, single chains, different positions, and that allowed us to find molecules that had a good activity. And then we also relied on the in vivo model to select selector molecule. And it was only then that we -- we also compare some other molecules and that we saw that this one worked really well there. Nina Or Thomas?

Yes. Yes, I think exactly like the format had a strong impact on the synapse formation, number one, which is what Paul was highlighting. And I think also the avidity-driven binding and its ability to bind mesothelin high but also maintain binding and maintain the T cell synapse formation is likely a component. And then there's the potential for less effects due to mesothelin shedding as kind of reducing antitumor activity, also because of that avidity-driven binding. So there's a number of kind of design features that are distinctly different between the 2 molecules.

Okay. And the other thing I noticed in the data, which admittedly it's a bit of a subtle observation, but I noticed that for ZW220, you still had some killing of the antigen-negative cells. But for 251, and I know it didn't show up on the slide, but I saw it on the download. For 251, there was actually no killing of the antigen negative. So I'm just wondering what was going on there, if there's an explanation for that subtle difference?

Yes, that one. Stuart will take that one.

Yes. So -- and that was in the bystander assays, right?

I believe so.

Yes. So oftentimes what we see, depending on the tumor cell line, is that there is a phenomenon called macropinocytosis. So that's where kind of you can think of cell lines as being a little bit thirsty and kind of taking up nonspecifically things in the dish. So sometimes, we will see with antigen negative cell lines if we were to run 30 antigen negative cell lines with 1 of our antibody-drug conjugates, some of those cell lines would be flat, no change, no reduction in viability and some of them would have this kind of modest reduction in viability. It's not anything that we read into too much and it's, like you said, it's a very subtle observation and one that I'm surprised you made, but it's kind of -- it's very interesting to us and like we care about it. So it's cool that you noticed it too. It's nothing to read too much into.

Okay. And then just really fast, just one housekeeping question. I think you said on the slides that for 191, it was DAR8. And then in the cartoons, you did have DAR8 for the other 2, but you didn't specify. So can you just clarify if the 251 and 220 are also DAR8. And for the TOPO1, are they all the same molecular structure? They just all TOPO1 inhibitors but slightly different structures for the 3?