REDEFINING ANTIBODY SPECIFICITY PROFILING FOR IND

Rebekah Kuschmider, PhD, Fierce Biotech

Kris Raghavan, PhD, Senior Application Scientist, Integral Molecular

Dr. Raghavan provides an overview of the MPA platform, its role in drug development, and a typical project workflow. He highlights the MPA’s unique features—including screening native-conformation proteins on unfixed cells and quantitative analysis by flow cytometry—and presents several customer case studies.

Diana Norden, PhD, Senior Scientific Writer, Integral Molecular

Dr. Norden focuses on regulatory guidance for specificity testing (see below for links to the FDA guidance documents she mentions). She provides a detailed analysis of the limitations of traditional tissue cross-reactivity assays, highlights the additional information MPA screening can provide, and describes case studies comparing MPA and TCR data for clinical antibodies. She also gives an update on the MPA’s progress toward qualification as a drug development tool under the FDA’s ISTAND program.

Questions include:

• If you don’t do TCR, how do you know which tissues your antibody binds to?
• Can MPA replace TCR data only for CAR T-cells?
• How does MPA perform for screening lager molecules, such as bispecifics, or smaller molecules?
• How do you do validation for off-target hits?
• Is the initial screening also performed on unfixed cells?
• And more to help you determine if the MPA is right for you.

• Review article: The emergence of cell-based protein arrays to test for polyspecific off-target binding of antibody therapeutics. Norden et al. 2024 mAbs 16, 2393785
• Case study: MPA reveals off-target binding, allowing team to prioritize a highly specific lead for CAR-T cell development. Based on Bhoj et al. 2021, Molecular Therapy Oncolytics.
• Case study: The MPA enabled IND submission and helped progress clinical development of DSG3-CAART cell therapy for Cabaletta Bio. Based on Lee et al., 2020, Journal of Clinical Investigation.
• Book chapter: Testing for Off-target Binding, Norden & Doranz, 2021. Translational Medicine: Optimizing Preclinical Safety Evaluation of Biopharmaceuticals.

• Oh et al., 2023, Nature Biotechnology. Precision targeting of autoantigen-specific B cells in muscle-specific tyrosine kinase myasthenia gravis with chimeric autoantibody receptor T cells.

Rebekah Kuschmider (00:01):
Hello, everyone, and thank you for attending today's webinar, Redefining Antibody Specificity Profiling for IND. I'm Rebekah Kuschmider, I'll be hosting today's webinar. I'd like to start by introducing our speakers. We will be joined by Kris Raghavan, PhD, he's an Application Scientist at Integral Molecular, and Diana Norden, PhD, a Senior Scientific Writer at Integral Molecular. You can read their full bios on the left side of your window by selecting the Speakers tab.

(00:28): Just a few notes before we begin. To access additional resources from today's presentation, you can click the Handouts tab button on the left side of your screen. You can access closed captions from the bottom right corner of the video player. This webinar is being recorded. It will be available to watch on-demand within 24 hours. There will be three audience polls throughout the presentation today, we hope you will participate in all of them. And we would love to hear from you, of course. During the presentation you can submit any questions you have using the Q&A tab on the left side of your screen. All right, let's go ahead and begin.

Kris Raghavan (01:03): Thanks, Rebekah. Hello, everyone. I'm really happy to be here. Today, I, Kris Raghavan, and my colleague, Diana, are here representing Integral Molecular all the way from Philadelphia. Just to give you a little bit of intro about our company, a lot of people often hear small biotech and think startup, but no, in fact, we've been established in the industry for over 20 years. We started as a spin out from the University of Pennsylvania, and since then our expertise has centered around membrane proteins and antibodies, as I will, for sure, get into.

(01:37): Our talk will be broken into two main sections. The first, introducing you to the Membrane Proteome Array, or MPA, as you'll be hearing that acronym very frequently, our role in drug development, and a breakdown of our project pipeline. Later, Diana will touch on the MPA's relationship with regulatory bodies like the FDA, and some case studies from customers that have successfully used the MPA in IND submissions. We will, of course, have some Q&A at the end, but for now I will get into the presentation proper.

(02:09): As I'm sure you're all aware, the road from antibody discovery to clinical development is long, expensive and involves a great deal of risk, so we provide the MPA as a service to help de-risk the early development of your molecules. We have tested over a thousand molecules on the MPA and we have found that about one in three of these molecules are polyspecific, and of course, specificity screening is essential for identifying potential off-target protein interactions, which can be responsible for dangerous side effects, as we have observed from commercially available small molecules that have had major side effects or have been withdrawn from clinical trials. Of course, discovering that your lead molecule produces off-target effects so late in development can be a program killer. Thus, the MPA exists to prevent those types of outcomes and de-risk drug discovery campaigns early in pre-clinical development.

(03:07): So to address this global need for better pre-clinical tools, we developed the field's leading solution for specificity profiling. We are proud to have been trusted by over a hundred companies that have incorporated the MPA into their drug development programs, further validating our data in their studies published in high-impact journals taken into the clinic. You will hear more about these types of studies later on in this talk from Diana, but first I want to get into really dive into the technology that makes the MPA unique.

(03:39): The Membrane Proteome Array is a platform to screen roughly 6,000 natively expressed membrane proteins. We do this in a 384-well plate format, using transiently transfected HEK 293 cells. What we feel really separates us from our competitors is that we measure target binding on a highly sensitive flow cytometry platform, and more importantly, we are the only platform that screens against native epitopes in unfixed cells, so that we can be more representative to how these protein targets will be expressed in vivo. Now, using these methods, we're able to profile the specificity of your molecule for its intended target and identify any off-target binding that may exist.

(04:23): To get into what is included in our library, our protein array has about 94% of the human membrane proteome coverage. This includes transmembrane proteins, ion channels, GPCRs, transporters and more. We have also added a hetero complex library that has been highly curated to include proteins that may need to be expressed as a complex to facilitate binding that would be expected in vivo. So things like integrin complexes or TCR complexes, we have that as well. And please reach out, because we would be happy to share the full list of proteins included in our library.

(05:04): Now, our platform has been successful with a wide variety of molecules and modalities. We specialize in monoclonal antibodies, but we have also had a lot of experience screening scFvs, bispecific antibodies, nanobodies, and even peptides. Our platform is also appropriate for ADCs and CAR-T applications, and customers with successful IND experiences have also shown that, for CAR-T applications specifically, testing just the scFv portion can produce the most time and cost-efficient specificity profiling while still being in accordance with FDA guidelines.

(05:45): Now I'm going to move to highlight some of the advantages you will get when you're using our platform, the first being a diversity in cell lines to improve the resolution of our screens. This is important because some protein targets that are endogenously expressed in HEK 293 cells, in this case B7H3, can result in a high degree of background noise that makes the screening of this molecule unfeasible. As you can see here, there is just so much noise. By using QT6 cells, which come from quails and are evolutionarily divergent, we are able to significantly reduce the background signal and improve the sensitivity by which we can detect target binding. During the initial optimization stages of our projects, we will test your compounds in both cell lines to determine which is the optimal condition to proceed to screening with on the full array.

(06:36): Now, another key benefit to our platform that I touched on earlier is that we are the only technology conducting full library screens in unfixed conditions to really maintain the physiologically relevant binding conditions that you would expect in-vivo. I am sure you are all aware that fixation can alter the native conformation of certain epitopes. This type of fixation-induced changes to membrane proteins can cause false negatives by preventing antibody binding, or even false positives from artifactual binding. Understandably, this can be a huge issue when testing molecules for binding specificity.

(07:17): To further emphasize this, I will showcase this in-house testing that we did where we conducted parallel screens comparing unfixed and fixed binding conditions. We tested the same antibody, the target CD19, using our standard of unfixed conditions and found that it had a significant off-target binding interaction. However, when we tested the same antibody in fixed conditions, we did not identify this off-target, really proving to us that the screening for protein binding on fixed cells can really miss these critical interactions. This further supports the FDA's guidance against specificity profiling done in fixed tissues.

(08:00): That being the finer points of our technology, but what does an actual project look like? I'm going to walk you through very quickly what the pipeline of our projects look like, and explain how we go from receiving your molecule to performing the specificity profiling, sending results and more.

(08:17): To begin a project, we ask for three things from you: a purchase order, a project set up form with information about your molecules, and 500 micrograms of your sample. Once we have these things in our hands, we will begin the project's lead time, in which we will plan out the study and consider any special requirements or risks that may be relevant.

(08:34): After this lead time, we will move to our initial assay setup experiments that can take place, being our assay setup. This stage is set up to determine the optimal conditions to test the sample on the MPA screen. We will test the sample against the positive and negative controls using a titration for concentrations, and we will do this in both HEK 293 cells and QT6 cells, as I mentioned. Once we find an optimal cell line and concentration, we will advance to the MPA screen stage.

(09:05): At this stage, we will screen your sample against the full 6,000 protein library for potential binding interactions. Any hits identified are then passed on to the validation stage, which is crucial to rule out any false positive hits that may have arisen. For hit validation, we again titrate the samples against our positive and negative controls in addition to the hits identified from the MPA screen to determine what is a true binding interaction and what is not. Now, at this stage we can also include an isotype control to really rule out any nonspecific binding that may have resulted from the antibody backbone or isotype itself. After this validation stage, we will compile the data to be presented in our final report deliverable.

(09:49): Now, at the end of a project, say you find an off-target binding interaction, what should you do next? Well, we often get this question and we have several post-project services designed to follow up with these results. Our newest one is what we're calling Risk Analysis and it aims to provide a more thorough understanding of any off-target binding interactions that may have been identified in the original study. This package includes a more rigorous set of experiments, designed to better quantify the EC-50 for binding strength, the B-Max for protein expression, and even visualize the subcellular localization of your test molecule's interaction with the off-target protein. We will then put together our highly detailed risk analysis report, aimed to provide a comprehensive picture of your molecule's relationship to the off-target protein.

(10:45): In addition to the risk analysis, we also offer other services like plasmid transfer for clones of off-targets identified in the study, if you wanted to perform follow-up studies and follow-up assays in your own labs with consistency in data and reagents. In addition to that, Integral Molecular also has some more in-depth services such as epitope mapping and antibody engineering, which are designed to optimize your therapeutic compound. It is worth mentioning our referral program with LONZA, who can perform some more functional cell-based assays if you're really trying to determine whether an off-target identified in the study is benign or potentially cytotoxic.

(11:26): Now, we understand that clients are coming to us at really all different stages in their development pipelines, so we offer two main services to accommodate those different needs based on where you're at in your project. As I touched on the basics of, our standard MPA service is best suited for when you have several molecules and you are looking to select a lead candidate from a larger cohort. But if you already have one or two leads and you're looking to screen for off-targets before filing for an IND submission, we, specifically Diana, will author a customized report package ready-made for you to submit to the FDA. Now, this MPA plus IND package also has prioritized timelines, so you don't just get more value in your data and report, but we will also complete the project in half the time.

(12:18): To briefly get into some examples how our customers have used our data, I want to highlight this case study that was done by a lab in the University of Pennsylvania, that used us for lead selection in a CAR-T program. They had two antibodies being characterized for potential clinical development and they used the MPA to screen both of these candidates. We found that while both molecules hit their intended target, one of them had significant off-target binding. Using these results, they decided to deprioritize the antibody with off-targets, and in the end, the antibody with no off-targets was advanced to the clinic.

(12:55): Another Philadelphia neighbor of ours, Cabaletto Bio, has also been incorporating us into their development processes. With them, we have screened several novel modalities, and they have used our data in successful IND submissions, which we are especially proud of. Importantly, they were granted their IND without tissue cross-reactivity studies, tissue cross-reactivity being the previously established and inherently flawed method of specificity profiling, which Diana will definitely touch more on later in this talk.

(13:30): Additionally, we at Integral have also been able to endorse the power of the MPA firsthand as it has been critical to the development of a spin-out company, Cell Surface Bio, which specializes in producing high-quality recombinant antibody reagents for laboratory applications. We understand the reality that it's far too common for experimental progress to be limited by poor quality reagents. So to address this problem, Cell Surface Bio uses the MPA to screen and validate our VeRSaMAbs to produce clean, highly specific antibodies that bind their intended target and nothing else. So again, through the use of the MPA technology, you can trust this antibody and thus trust your data.

(14:16): Just to wrap up my section, there are, of course, many reasons why the MPA should be attractive to you in drug discovery and preclinical development. Our platform is compatible with a wide range of therapeutic modalities, and again, I cannot stress this enough, we are the only company screening against native unfixed proteins. Critically, our customers have proven successful in their IND submissions without tissue cross-reactivity, and Diana will talk more to that point, as well as dig into our track record with the FDA and regulatory bodies. I will also add that you can work with us directly, but our services are also available through the partnerships listed below.

(14:54): With that, I will hand off to my colleague, Diana Norden, as she gets into her portion of the talk.

Diana Norden (15:01): Hello, everyone. For the second part of this webinar, I will be discussing using MPA data for improved specificity profiling for regulatory applications. I've divided this talk into three sections. First, we will go over some regulatory guidance documents and talk about specificity profiling for IND applications. Then we will do a comparison of the MPA technology to what has traditionally been used, the tissue cross-reactivity study. And then finally, we will look at some case studies comparing MPA and TCR results. So, let's get started.

(15:36): FDA has two guidance documents that discuss specificity profiling of biotherapeutics. This is the Points to Consider in the Manufacture and Testing of Monoclonal Antibody Products for Human Use, as well as the ICH guidance, the Preclinical Safety Evaluation of Biotechnology-Derived Pharmaceuticals. Both of these guidance documents state that non-target tissue binding may have serious consequences and that cross-reactivity should be assessed.

(16:04): They further go on to say that assessment of cross-reactivity should be performed using human cells or tissues. This is where the tissue cross-reactivity study got its name and has since been used. I'd like to point out that the points to consider document has not been updated since 1997, this is still the current version used today. I'd also like to point out an interesting sentence. It does say that, "Appropriate newer technologies should be employed as they become available and validated." So already, back in 1997, there's this idea and thinking that the TCR assay might not be the best way to truly assess antigen specificity.

(16:46): Recently, there's been some new developments in how the FDA thinks about specificity profiling and the assays that they recommend. In January of this year, they have put out a new guidance document for development of CAR-T cell products. Again, it states that non-clinical evaluation of the antigen recognition domain should include assessment of specificity. It further goes on to say that unintended targeting of other antigens, which would be an off-target expressed on healthy or normal tissue, is a safety concern that may be evaluated using in-vitro or in-vivo studies. But this time it gives numerous examples. Again, it gives the assay TCR, but then includes other examples of assays like the protein array. So this guidance document now, for the first time, delineates specific alternative methods beyond the TCR study.

(17:40): Our own protein array, the MPA, that we are talking about today, is currently under review to become an FDA approved drug development tool. We submitted the MPA for inclusion into their ISTAND program, this is one of FDA's drug development tool qualification programs, and we are very excited that MPA actually became the first technology to be accepted into the program. Since then, we have now been working with CDER and CBER representatives to qualify the MPA as a new drug development tool. We've received feedback on the MPA technology and its uses. We have taken that into consideration in our product development and in the preparation of our qualification plan. We have since submitted our qualification plan for review. It was accepted for review and we hope to have full approval in 2025.

(18:32): Let's look at some real world data and how our customers are using the MPA right now. We surveyed our customers and found that about half of them, right around 50%, have used or planned to use the MPA data in their IND filings. We've had numerous successful submissions of MPA data that did not include TCR for IND applications of antibodies and CAR-T cells. We've had many successful submissions to the FDA, but also the EMA and NMPA. And about 25% of the customers that we surveyed reported that the FDA directly requested a newer array-based technology like the MPA.

(19:10): That was our customers. What is the general toxicology field as a whole, thinking about the TCR assay? This slide shows some results from a survey that the Biosafe Committee Working Group put out. They surveyed 26 different biotechnology and pharmaceutical companies and asked them about their experience in using the TCR assay and how they interpreted those results.

(19:33): Some key results were that 92% of respondents said that the TCR has not impacted the design of their toxicology study. Further, 81% said they have never seen a correlation of TCR data to preclinical or clinical toxicity. They were also asked about whether they had submitted an IND without TCR data, and here, 42% said yes and 31% said that they were considering doing so in the future. We think, based on our customer experience, and the fact that this survey was now completed six years ago, that this number now likely is very much higher, and that developers are more and more comfortable with submitting an IND that does not include TCR data. The general conclusion, or take-home message, from this survey was that the majority of companies believe that the TCR results really are not predictive of in-vivo toxicity.

(20:32): I've discussed TCR a little bit now, and Kris has also discussed the MPA, so let's do a little bit more of a comparison between the two technologies, or the two assays. First, what are tissue cross-reactivity studies, for those who may be aren't as familiar? They are immunohistochemical screening assays used to identify binding of test biologics in different human tissues. Over 30 different human tissues are tested for binding by the molecule, and that staining is evaluated by a pathologist. This assay can give you an idea of a target distribution, so it tells you what tissues, where the target is expressed. It can also give some indications of target epitope binding. And here, unexpected binding patterns could be due to off-target binding interaction, or it could also be due to an off-target tissue or organ reactivity. In these cases, the unexpected staining patterns really need to be investigated further to try to differentiate if the unexpected staining is due to an off-target or just a tissue where you weren't expecting to see staining.

(21:37): When looking at the two assays more head-to-head, first obvious difference is that the MPA is a protein array, it's a cell-based protein array that uses flow cytometry, versus a TCR study is an immunohistochemistry based assay to evaluate staining in tissues. This gives rise to the main differentiator. In the MPA, we will precisely tell you which proteins your molecule is binding to, versus in a TCR study you only find out which tissues that the molecule binds, and it's unclear which actual proteins gave rise to that staining.

(22:13): The analysis is also very different. We use quantitative analysis, and we can do statistical analysis as well, versus in a TCR, this is a more of a qualitative assessment. MPA has high sensitivity, as Kris already talked about, in the assay setup we ensure screening conditions that have high sensitivity with high signal and low background, versus in TCR, low sensitivity is often an issue, there can be high background and false positives due to native Fc receptors and IgG.

(22:43): The low sensitivity issue in TCR studies is really important, or has become even more of an issue for newer modalities, so the MPA is compatible to screen such things as VH molecules, nanobodies, scFv-Fc fusion proteins. And there are numerous instances where customers tell us, and there's also case studies showing that these types of modalities sometimes are not compatible with the TCR studies, where you really need to have high sensitivity to generate a robust assay. Again, in the MPA, we use native confirmation of targets, we never apply fixative in any of the steps of the process, versus in TCR, fixatives and freezing of tissue is often used to preserve tissue integrity, but this can lead to altered conformational epitopes on the protein level. The timeline is different as well. MPA is high throughput, we can complete studies in four weeks, versus the TCR is significantly longer to complete.

(23:44): If you're interested more in the technical details of these two assays comparing cell-based protein arrays to TCR, then I encourage you to read our chapter on testing for off-target binding. We were asked to contribute a chapter to the Translational Medicine book, Optimizing Preclinical Safety Evaluation of Biopharmaceuticals. This guidance book was published by two expert toxicologists, Joy Cavagnaro and Mary Ellen Cozensa, and our chapter is available online.

(24:15): One last comparison or contrast I would like to make is, what are the two assays really testing for or testing against? The MPA library was actually designed based on FDA guidance documents that discuss tissue cross-reactivity studies. So the tissues that are listed for this assay were selected and then we use an RNA-seq approach to identify what proteins are in there. We use online databases and topology servers to predict what membrane proteins are actually in there. This way, the MPA library and the proteins that are being evaluated should directly represent the tissues that are suggested to be used in a TCR assay. But here, the differentiator is that in TCR they use normal and healthy tissues, and here you have a lot of variability between donor to donor. There's also variability in endogenous expression levels of proteins, and proteins that are expressed in a more disease dependent matter. Versus in the MPA, we use a protein overexpression to ensure a more complete representation of all the proteins that can be found in these particular tissues.

(25:28): Last, let's look at some case studies that directly compare the two assays, the MPA versus a TCR. These case studies are coming from a larger study where we are testing clinical antibodies on the MPA. Some questions that we had going into this study, and when we were performing this study, is that, as Kris mentioned, off-target binding is commonly found in preclinical and early development MAbs. But what about antibodies that are actually in people right now? Do they also have off-target binding? Another question is, is MPA able to predict off-target binding similar or better than a TCR study? And can MPA data actually predict drug safety? This is a question that we are trying to address because it's needed for our clinical validation for our ISTAND proposal. Are there any instances where we find an off-target using the MPA where we can then link that off-target to a safety issue or adverse effects in people?

(26:25): Here, our approach on the experiments that we performed, we produced up to, or close to a hundred biosimilars of antibodies that are in all phases of clinical development, and also approved antibodies, and then we screened them on the MPA to see if they had any off-targets. This larger study is now prepared for publication, we just submitted it last week and it's under review. So if you're interested in this data in some of these discussions, then I encourage you to be on the lookout for our new paper.

(26:55): But back to our comparison to TCR. What we did was we identified a few approved antibodies that actually had off-targets on the MPA, and then we compared our results to the TCR results that are obtained from BLA applications that can be found online.

(27:14): This is the first case study. Here, the antibody targets plasma membrane protein that is primarily expressed on lymphocytes. The MPA identified the target but actually also identified two off-targets. The BLA application stated that the TCR staining was consistent with a known expression profile of the target. So this is a case where the TCR study simply missed two off-targets.

(27:41): In the second case study, the antibody targets a plasma membrane protein that is expressed on myeloid cells. The MPA correctly identified the target but also identified an off-target. Here, the off-target is actually a membrane protein expressed on the same cell type as the target. The BLA application, to no surprise, stated that staining was consistent with a known expression profile of the target. This is a great example where the TCR study really has a limited ability to identify some off-target proteins. The target and the off-target were expressed on the same cell type, and here it's basically impossible for the TCR to identify these proteins... To identify the off-target, I should say.

(29:03): In the third case study, this is a target that is expressed on plasma membrane and intracellular vesicles. This is an oncology target, it's very active in cancers, it is also upregulated in several cancers, but in normal tissue the target has a low expression but in several tissues... In healthy tissue, I should say. The MPA identified the target but also identified an off-target. The BLA applications stated that staining was consistent with a known expression profile of the target. This is then another example where the TCR really has a limited ability to identify some off-target proteins. The fact that the target is expressed both on the plasma membrane and intracellular vesicles, and throughout numerous healthy tissues, made it so that the TCR really was unable to pick up any staining from an off-target. Again, if you're interested in reading more about this, then you should be on the lookout for our publication, hopefully coming out soon.

(29:26): To wrap up my portion, today, we talked about regulatory agencies now starting to support these emerging cell-based protein arrays like the MPA. Protein arrays are now included in CAR-T guidance documents, and our own MPA has been for review into the ISTAND drug development qualification program. We have numerous successful IND applications that have used MPA data and not TCR data. We also discussed limitations of TCR studies and the advantages of the MPA. There are numerous technical limitations of the TCR study, and relying on TCR studies to assess specificity is really no longer supported by the majority of toxicologists. Then we looked at some case studies where the MPA was able to identify off-targets that the TCR missed.

(30:16): That wraps up this portion of the webinar. Thank you to everyone listening live and anyone listening to the recording. Here is the contact information or emails for Kris and myself, and you can also check us out at integralmolecular.com. Now I'll hand it back to our moderator for any questions. Thank you.

Rebekah Kuschmider (30:39): All right, well, it is time for Q&A now. You can still submit questions using the Q&A tab on the left side of your screen. I can see we've got a lot of great questions already, we'll try to get to as many of them as possible. All right, to begin, this is for Kris. How do you ensure proteins are expressed?

Kris Raghavan (31:00): Great question. Thank you. So yeah, a really important thing that we do is we put a V5 tag on the vast majority of our library. That allows us to take regular screenings of that V5 tag as an indirect measure of the expression of each one of these proteins.

Rebekah Kuschmider (31:19): All right, thank you. For the next question, this is for Diana. If you choose not to do TCR then how do you know which tissues your antibody binds to?

Diana Norden (31:31): Right, that's a great question and we get that quite a lot. I think the first thing, the easiest thing to do, and we do a lot, is to use a lot of available databases that are available online. There are both RNA and protein expression databases. We use Human Protein Atlas, there's also Expression Atlas, there's a website called Bgee that looks at different animals as well. So if a particular protein is fairly well characterized, these websites can be very useful. The Human Protein Atlas has both gene expression and protein expression data available, it's all consolidated in the website, so it gives you a good idea of, or information on where the target is expressed.

(32:20): Then for less well characterized proteins, say it's an off-target, comes up on the MPA and it's not really that well characterized, then what you can also do is what's called tissue microarrays. These are slides that have small sections of tissues on them and you can stain them to see which tissues express that target. The nice thing there is that you don't have to use your own test therapeutic sample, you can actually use a commercial antibody that's been validated as a detection reagent to see where this target is actually being expressed. So if you choose to not do the TCR, there's just other approaches and ways to think about understanding the tissue expression profiles of that particular target that you're interested in.

Rebekah Kuschmider (33:13): Thank you very much. Here's another one for you, Diana. Can MPA data only replace TCR data for CAR-T cells?

Diana Norden (33:22): Right. So the quick answer would be no, it can replace it for antibodies as well. Where we are right now, the guidance document that lists protein arrays is the CAR-T development guidance document, but I can speak a little bit about our customer experiences as well. We've had customers for antibody development submit a TCR for maybe different reasons. Maybe they tried TCR and they couldn't get a good assay to work, or maybe they did a TCR and they had inconclusive results and then chose to do the MPA instead. These are just some from what customers have told us. But there's also a lot of customers actually that do both, because they really want to really de-risk their therapeutic, and get a really complete understanding of their specificity, so there's also the option to do both.

(34:23): Then one last point to make is this is where we are right now and going forward things might change. In our ISTAND application to the FDA, we actually proposed the MPA to be a replacement to TCR as a complete alternative. It's under review right now, and if it gets accepted or approved as a DDT, then things might change in just the near future. It's always changing on how they're looking at the guidelines, but that's where we are right now, and we're seeing a shift in how people are using it and replacing the TCR.

Rebekah Kuschmider (35:04): All right, moving on. How does MPA perform for molecules larger than normal, maybe like a biospecific, or a smaller molecule like a cargo loaded EV?

Kris Raghavan (35:16): Yeah, great question. Essentially, we are going to screen the same way, but for biospecific antibodies, we actually get those fairly often, I think a good way to help identify where binding might be coming from, or off-target interactions might be coming from, would be to include the monospecific arms as controls in the validation stage. We would call those isotype controls, but really just a control to help identify where binding might be responsible, what area of the molecule binding might be coming from.

(35:54): For smaller molecules, that's a great question. We get a lot of peptides that are typically smaller than antibodies or even VHHs, I think we find that going with our standard titration is usually what we like to do upfront. But if there are ever issues with that, that's why our assay setup is so important, we can always modify the titration if the stoichiometry works a little bit better to start at lower concentrations if it's a smaller molecule. So there's a lot of flexibility there and that is really the most important part of early discussions in the project planning stage of our pipeline.

Rebekah Kuschmider (36:36): Okay, thank you. Is the cell surface expression in your system always at high levels, and do TCRs better reflect the physiological expression level of targets?

Kris Raghavan (36:51): It is an overexpression system. We are transiently transfecting our cells to overexpress one target. We're creating populations of cells that are all overexpressing one particular protein, so of course, there's a little bit of artificiality there. But with the overexpression approach, we do have very high levels of protein expression typically. As far as TCR goes, and the physiological relevance there, as we've said, you're assessing binding in tissues that have been fixed, and so that can really alter native epitopes, which is really the advantage of our platform, is that we are performing our binding on these cells and these proteins that have not been fixed in any way. So I would say the MPA is more physiologically representative compared to TCR approaches.

Rebekah Kuschmider (37:47): All right. Are your proteins fused to a tag like GFP?

Kris Raghavan (37:53): Our proteins, like I mentioned, are fused to a V5 tag, there's a V5 tag on the vast majority of these molecules, that's what we use for reporter molecule that we use in the actual screening that are on the proteins, we use a secondary antibody typically for detection.

Rebekah Kuschmider (38:21): Moving along. Do you see any use of doing MPA and TCRs in parallel in order to answer the questions regarding the tissue types that the AB of interest binds to?

Diana Norden (38:35): Yeah, definitely. I think our approach is to always do the MPA, gives you the most complete understanding of specificity, but the TCR could add to that as well, you could do the two in parallel. But the limitations of TCR are mostly because it is very costly and time-consuming. So I mean, if you have the resources to do both, that's great, but a lot of companies maybe don't want to do both. (39:07): Then also the compatibility. It's sometimes hard to get a really good signal on the TCR, and a clean signal. That's where a lot of people report the issues and tell us their issues, maybe why they tried it and they got some staining, but it's hard to interpret. So yeah, I mean, I think it's maybe great to do both, but you have to remember the limitations when you're analyzing the results, or looking at the results from the pathologists, that there might be things that are missed, like we saw in our case studies where we identified some of targets that the TCR didn't really pick up. Maybe there was some tissue staining, but the particular target wasn't actually revealed, or the cell that it was on wasn't revealed. So you can, of course, do both, but I think it is good to keep in mind the limitations of the TCR, that it won't actually tell you the protein that it's binding to, it just tells you the tissues.

Rebekah Kuschmider (40:12): Okay, if protein expression is checked by V5, do you just assume correct expression or folding?

Kris Raghavan (40:21): We are not checking for the proper confirmation of all 6,000 proteins. So yeah, that's relatively difficult to do. However, a big part of the assay setup is using the known target of your molecules to make sure that we're observing binding as expected. A lot of molecules that are given to us are conformational dependent, or can be at least, and so that's a very important step in our de-risking of the screen itself. So no, we are not confirming the actual protein folding, but there are a lot of checks that we do to make sure that we are observing binding as it should occur.

Diana Norden (41:08): I can also add that this is a little bit separate from our specificity profiling, but we use the MPA actually for research and development projects as well. So we do functional screens on the MPA on the research side, and we don't really see, or have any instances where we think that the proteins aren't being functional or folding correctly. We've done T-cell screens, dendritic cell screens on the MPA where the readout is a functional readout. So even though we haven't individually tested every single protein individually for folding and confirmation, all our data indicates that they are being expressed functionally.

Rebekah Kuschmider (41:57): How do you do validation of off-target hits?

Kris Raghavan (42:01): Great question. So our validation process is very similar to the assay setup in that we are doing a relatively straightforward binding assay using a titration of the known target, positive controls, our negative control is just empty vector, cells expressing empty vector, and then all of the targets that came up in the screen, so we are doing a straightforward binding assay there. What we would consider a passing validation, or indication of a true binding interaction, would be any protein that is titrating at two fold higher than the negative control at a series of concentrations. So it's a relatively low bar, I would say, but our policy is to report more potential interactions, just because it's so important to not miss any type of potential low binding interaction. So that's typically our cutoff is about two fold from the negative control.

Rebekah Kuschmider (43:08): Can you please clarify your approach for IND?

Diana Norden (43:10): Yes. We designed the library to be compatible, or to be a representation of the currently recommended assay, the TCR. Then we have updated all our reporting and all our data tracking to be compatible with GLP, which is often used for IND, GLP tox studies. So we have an IND ready report, and then all our data tracking and auditing is compatible with that as well. Kris, am I missing anything?

Kris Raghavan (43:51): No, I think that that about covers it.

Rebekah Kuschmider (43:56): All right. Have you evaluated post-translational modifications in your system compared with normal or diseased cells?

Kris Raghavan (44:05): That's a very interesting thought. We have not specifically addressed post-translational modifications at least. We know that some molecules are modified, like [inaudible 00:44:18] or cleavage events, things like that. Based on what data we have from, again, assessing expected binding for molecules that should be binding things that do get modified, we do see binding as expected. However, we are not doing specific assays to assess different types of post-translational modifications.

Rebekah Kuschmider (44:45): Okay. Moving on, I think this one is for Kris. Is the initial MPA screening also performed on unfixed cells?

Kris Raghavan (44:53): No. So we are strictly unfixed. It is performed on unfixed cells, sorry. We are strictly doing that. I showed that one comparison of the fixed versus non-fixed, but that was really just a proof of concept for underscoring the point that it really is important to us to assess binding in those native protein confirmations. So only unfixed cell conditions for us.

Rebekah Kuschmider (45:23): All right. Now a question for Diana. For the case studies where TCR missed an off-target, were there any safety concerns found from those off-targets?

Diana Norden (45:35): Right. We take a deeper discussion on off-target binding leading to safety in our paper, so definitely be on the lookout for that. We discussed some indications where the off-target might lead to a particular safety concern, and then we're doing more of those studies right now. Where we are right now is a little bit more speculative, but very reasonable speculation, based on the target biologies of what we have found and what we have found for those particular drugs. That's all discussed in the review. Then outside of our work there's also other published papers that show direct safety issues or toxicity from off-target binding in those drug programs. So we also discussed that in the review, if you're interested more about the off-targets directly causing safety issues.

Rebekah Kuschmider (46:38): Do you have any data correlating off-target identified by MPA screening and confirmation in functional assays, for example, from the client?

Kris Raghavan (46:50): Yeah, that's interesting. I mean, sorry, I can just jump in and then, Diana, you can add to it if I leave anything out.

Diana Norden (46:56): Yeah. Kris Raghavan (46:56): But typically, what customers share with us is what they feel comfortable sharing or what they feel is relevant. There have been instances where, in either troubleshooting conversations or project planning conversations, they have shared functional data with us. However, this is usually at the beginning of projects, not really at the end regarding off-target interaction. But yeah, again, that's not with every project and it's not with every customer. Diana?

Diana Norden (47:30): Yeah, I can add a few. We have quite a few case studies on our website that talk about different interesting findings that we have worked with clients on. I know there's one case study, for example, that Kris had in his slides, where they were doing lead selection. And the two antibodies, one was clean, one had an off target, and the one that had an off-target, they had seen bio-distribution issues in mice, and the off-target binding actually explained the bio-distribution assays, where they had found accumulation of the antibody where it wasn't supposed to be. (48:10): Then a lot of our CAR-T developers, what they will do is, we screen an scFv for them, and then tell them like, "Hey, we found this off-target." Maybe it was low, medium or whatnot. But what they will do then is take that information and do CAR-T testing in their own labs. So if we say binds is off-target on the scFv portion, the full CAR-T, they might test for binding. We have a few papers published on this too, where the CAR-Ts maybe bind, but maybe not become fully activated by that particular off-target. Then we have a list of papers on our website that point to some of these examples, where if they found an off-target, and then what to do next, assays to do after finding an off-target, and maybe how the off-target explained something that wasn't going right in their pipeline, or in their studies. A lot of that is on our website.

Rebekah Kuschmider (49:15): Okay. Here's another question for Kris. What does the MPA IND report entail? Is it sent with the FDA package as it is?

Kris Raghavan (49:24): Yeah, great question. The IND report, compared to the standard, is going to be almost three, I would say, times the size, length. That's going to include a much more detailed method section, things that really are required by regulatory bodies. So we're going to have a lot more reported data tracing, I would say a more detailed explanation of the results, an executive summary of the application of the test article as it is planning to be used in clinical trials, things like that. Our standard report is typically just a representation of the data and results that we found, but the IND is one that is tailor-made to be submitted directly to these regulatory institutions.

Rebekah Kuschmider (50:18): Another one for you Kris. Did you experience binding to off-target hits in your screen that did not translate into binding to off-target endogenous expressing cells? And if so, do you know why?

Kris Raghavan (50:30): Yeah, so we are not testing endogenous expression, or our molecules against cells that are just endogenously expressed. So I would say when we have identified off targets in the screen and they've passed our validation, again, we are using a low bar for validation and we are using an overexpression approach, so you're going to have a large degree of protein available for binding to occur. So there is a very good chance that when we identify an off-target in our screens, that if you go back and do some follow up assays, we always encourage our customers to do follow up assays with any results to confirm or further validate results that came out of our testing, if you're doing it in cells that are just endogenously expressing the particular protein or target that was identified as an off-target, there is a good chance that it might not be biologically risky or dangerous, that type of binding interaction. But again, we are here to provide just a high-level overview of what risks might be there for off-target binding that our drug developer customers are not aware of at the time.

Rebekah Kuschmider (51:59): All right. Moving back to Diana, are the proteins in the MPA all based on adult tissues or are there proteins that are expressed in different stages of development?

Diana Norden (52:09): Right. The library was designed based on adult tissues and adult proteins. We actually had this question ourselves, so we looked further into it and we did a bioinformatics analysis of it, and it does a good job representing placental and fetal proteins as well. Not as high, the adult percentage is 94% of adult membrane proteome, and then the fetal and placental proteins, I think, are around 90%, between 87% and 90%. So it does a good job representing all the developmental stages, but again, it was designed based on the guidance documents that [inaudible 00:52:51] test against adult tissues.

Rebekah Kuschmider (52:58): Do you express peptide MHC libraries too?

Kris Raghavan (53:03): That's an interesting question. Not at the moment. I mean, those are talks that we've had internally, but I would say not right now.

Rebekah Kuschmider (53:17): Does your library include secreted proteins?

Kris Raghavan (53:21): Another great question. It will very soon. Some of the proteins just at some point in their life cycle are secreted, so are in our normal library; but we are in the very final stages of developing a secreted membrane library to screen against itself. I think we are getting ready to launch that probably quarter three [2024]. So yeah, that's very exciting. And if that is something that is attractive to you for your drug development purposes, please reach out because we would love to get some early testing with customer samples done.

Rebekah Kuschmider (54:02): What stage of antibody discovery or development should you test for specificity?

Kris Raghavan (54:09): I would say probably as early as possible. I mean, we really, as we've emphasized, this is a de-risking experiment or study. So before you are pouring in lots and lots of money into a potential candidate that... We've seen in IND studies that we have customers that have a molecule that's all ready to go, and they're just wrapping up some last minute, or not last minute, but some late stage specificity testing, and we find a pretty significant off-target that can be worrisome, and that, of course, can really derail an entire project campaign. So I would say as early as possible, once you feel confident about your molecules. We have a lot of customers that come to us very, very early in their pipeline, and will give us up to 20 or more molecules just for us to tell them which ones are clean and which ones have off-target, so that they can just very quickly, very early on, go forward with the ones that are clean.

Rebekah Kuschmider (55:11): All right. For Diana, if MPA becomes an FDA-approved drug development tool, will all antibody drugs have to be screened by MPA for specificity or can other methods still be used?

Diana Norden (55:25): Well, right now there's no guidance saying that they have to be, but if it becomes qualified by FDA, I don't know why you wouldn't. There would be many advantages to doing an MPA screen. If FDA says that this is a tool that you should use for specificity testing, they will already know it, they will have evaluated it, they'll know that it gives you reliable results. And the regulatory review will be much simpler, both for the customer and for them. They'll know exactly what they're looking for, they won't have to reevaluate the assay. But unless they give a specific statement or update to the guidance documents, then you wouldn't have to, but I would recommend you do so.

Rebekah Kuschmider (56:14): And for Kris, what is the turnaround time for a standard project and an MPA plus IND project?

Kris Raghavan (56:22): Yeah, so the turnaround, I would say, from the point that the molecules arrive in our hands is about six to eight weeks for a standard project. Typically, we will give updates along the way. We like to share data as early as possible, so when we do our assay setup, our initial optimization assay, we like to give data as soon as possible, just so that if there is anything that would potentially require some troubleshooting that might push the project out a little bit longer, we can do that as early as possible. So yeah, 6 to 8 weeks for the standard project, and like I said, about half that amount of time, we try to aim for a tight four weeks, for IND projects, just because we know a lot of customers are coming to us with some pretty important deadlines. Especially if there are follow up assays that need to come out of that, we try to get that study done as soon as possible, so that also adds into the value of what you're getting too, I think.

Rebekah Kuschmider (57:25): All right. Well, we are coming to the end of our time, so we only have time for one more question. I'm sorry if we weren't able to get to all the questions that were submitted. We will do our best to get back to everyone after the webinar. For the last question, have applications been filed only to FDA using MPA data for specificity, or also other regulatory agencies like EMA and NMPA, et cetera? Were cross tissue reactivity studies also included?

Diana Norden (57:54): We surveyed our customers and I believe we have submitted to all of those you mentioned. Also, I believe for all of those you mentioned without TCR studies as well, actually. It's hard to keep track of everybody and all our customers, but I know we have many successes around the globe, and the different regulatory bodies have different standards including TCR or not. [inaudible 00:58:29] exact numbers, but I think the general consensus is that a lot of our customers have had success with just MPA, and especially for CAR-T, as we talked about.

Rebekah Kuschmider (58:41): All right. Well, thank you both so much. I'd like to thank all of you for attending this Fierce Healthcare webinar, for our speakers for participating. This webinar has been recorded. You can access the recording within 24 hours by using the same audience link that was sent to you earlier. Thank you again for joining this Fierce Biotech webinar today and we look forward to seeing you at future events.