Moderated by Drug Discovery News, this webinar features discussions from Carterra’s Dan Bedinger, PhD and Aviva Systems Biology President Kevin J. Harvey, PhD regarding “High-Throughput Epitope and Antibody Characterization.” The DDN webinar discusses the utility of high-throughput surface plasmon resonance (HT-SPR) for measuring antibody kinetics and high-resolution competition-based epitope binning for rapidly characterizing diverse monoclonal antibody panels.


Dan Bedinger, PhD
Field Applications Science Manager Carterra
Kevin Harvey, PhD
Aviva Systems Biology


Posted by Daniel Bedinger, PhD


0:00:01.4 Luisa Torres: Hello everyone and welcome to today's DDN Webinar. I'm Luisa Torres, assistant science editor for DDN, and I will be moderating our discussion. We have an exciting webinar planned for you, our speakers, Kevin Harvey and Daniel Bedinger will discuss the high throughput measurement of antibody kinetics and high resolution epitope binning for rapidly characterizing antibody panels.

0:00:27.6 LT: After the talk, Dr. Harvey and Dr. Bedinger will participate in a live Q&A session. To submit your questions or comments, use the Q&A portal to the right of your screen, and we will try to get to as many questions as possible at the end of the webinar. I'd like to take this opportunity to thank our webinar sponsor, Aviva Systems Biology. Aviva Systems Biology support scientists in their basic research, preclinical, drug development and diagnostics workflows. With an extensive portfolio of nearly 500,000 products for antibodies, proteins, and immunoassays, and over 20 years of expertise in antibody and protein production, scientists can find the unique reagents and support they need for success in their work.

0:01:12.0 LT: Our sponsor has provided us with some helpful handouts related to today's webinar. You can access those in our handout section located on the right side of your screen where you can also find your certificate of attendance for participating in today's live event. And with that, let me introduce our first speaker, Kevin J. Harvey. Kevin Harvey leads Aviva Systems Biology and Genway Biotech operations with a focus on delivering profitable growth initiatives across the company. He has over 19 years of biotechnology experience and a proven track record of delivering profitable growth within the antibody product segments. Most recently, Harvey served as a general manager for Thermo Fisher Scientific's antibody business, where he had financial and product development responsibility.

0:02:00.4 LT: Harvey has held leadership roles at PYMETRICS and EMD Millipore, where he was responsible for research and development, product management, business development, and commercial strategy. Harvey has a PhD in microbiology and immunology from the University of Illinois in Chicago, and held postdoctoral fellowship at the University of California in San Diego. Dr. Harvey, we can see your slides now, so please take it away.

0:02:31.9 Kevin J. Harvey: Great. Thank you, Luisa. I really appreciate the introduction and excited to present this webinar today with my colleague from Carterra, Dan Bedinger, we'd like to... I would like to introduce Aviva. Aviva has been around for quite some time, but many of you may not be familiar with the company. Also gonna tell you why we've chosen to use the Carterra LSA to do high throughput SPR both for our internal customers and product development here at Aviva, but also for external customers as a service we're providing to biotech, pharma and diagnostic companies and then talk about a couple of case studies and Dan will expand on that after me, and look forward to your comments and questions. So let's get started.

0:03:27.9 KH: So, as I said before, Aviva has actually been around for over 20 years with an antibody focus established in 2002. Many of you may know that, we made our name in Polyclonal Antibody Development beginning, in 2010 with over 24,000 polyclonal antibodies developed against human mouse and rodent targets with a focus on transcription factors, we've expanded the portfolio over the last 10 years, and importantly acquired a custom service company called Genway Biotech a little over five years now. As part of that team, we've expanded our lab operations here in San Diego, providing not only custom service offerings around antibody development, but also GMP manufacturings of proteins in general. We've expanded that team and under our new leadership here, beginning in 2021, are offering new services and products as well. And I will talk a little bit about the Carterra LSA platform, the SPR measurements we're doing both for our internal pipeline, but also for some external customers.

0:04:49.7 KH: So roughly 60 employees headquartered here in San Diego, California, serving global customers both on the catalog side, but also custom services side where we provide, not only antibody characterization services, as I said before, but also GMP manufacturing for a number of diagnostic companies as well as research organizations, but today I'm really gonna focus on our high throughput antibody characterization offering. As I said before, we're leveraging this for internal research and diagnostic antibody development for our own products, but also as a service.

0:05:30.7 KH: And this is really related to an investment we made in an instrument, which I think is a game changer and is really important for the careful and quick development of antibody therapeutics, but also I think ultimately for diagnostics and research grade antibodies. So we chose the Carterra for a number of reasons. So we are switching our business, the focus moving away from polyclonal antibody development to a recombinant or renewable source of antibodies. So the market has changed in the last 15 years as it relates to research grade antibodies and our pivot away from polyclonal antibodies to recombinant antibodies, requires us to better characterize the recombinants that are coming out of our pipeline, which will allow us to select fit for-purpose antibodies, not only for typical research applications, but also for our customers that are demanding this now.

0:06:43.9 KH: One of the important things that the Carterra LSA allows us to do for our internal pipeline is epitope binning. So Dan will talk a lot about this in his talk, but it's really important for us to identify early those antibodies that we can pair in a sandwich immunoassay, both for our internal catalog, but also for some of our customers. And the platform provides a powerful hands-off approach to do that, which allows you to put antibody candidates in bins depending on competition assays, which is fully automated with a nice software that allows you to put those into families and classes, allowing you to quickly identify pairs of antibodies that'll work in a typical ELISA or multiplex assay like Luminex and other platforms.

0:07:37.1 KH: We like the Carterra because it's robust. It also allows you, which is I think, different from previous platforms that measured affinity using different biophysical measurements. It allows you to do many at a time. So it's robust. There are low error rates, it's reproducible, but you can also do duplicates and triplicates and get real data with CV and error rates, which is important for any kind of characterization you're doing at scale.

0:08:12.7 KH: We also find it ease of use. So it's not for, I would say, an expert in the field. Certainly, we have some of those people in-house right now, but it really is amenable to scientists of various skill levels, whether they're RAs or senior scientists. So ease of use of the instrument we've found, which we've had in place for a little over a year and a half right now.

0:08:39.0 KH: We're one of the first adopters within the research market to incorporate this into our own pipeline, but also to offer as a service. One of the most important things is the quick turnaround time, both for our internal and external customers. We can complete roughly 100 to 384 full kinetic profiling within a day. We typically turn this around and give customers information in a week. So we receive materials, set up, potentially do scouting experiments and turn that data around to our customers within a week. So this is really important for not only our internal pipeline, but our customers as well.

0:09:30.0 KH: I just wanna highlight a paper that some of you may have seen. It came out last month in nature and it's really related to the topic of understanding your binding kinetics, especially your on and off rates. Mark Cragg at University of South Hampton, published this, and I think it's a great example of showing the value of not just selecting the highest affinity binders, but when you're looking at developing antibodies, that you would like to be agonistic, Mark calls out a number of examples, CD40, some CD137 candidates, and actually converting high affinity PD-1 binder into a lower or medium affinity binder, and actually changing that from an antagonistic to an agonistic antibody. In general, therapeutic antibodies are considered high affinity if those KD values are under one nanomolars. So people are generally looking for picomolar binders.

0:10:36.3 KH: This paper describes that the optimal dissociation constant for these agonistic antibodies actually are in the nanomolar range. So it's a really important paper, I think, that starts to describe different ways antibodies are binding and the different factor or functions or ultimate outcomes. And it's certainly not surprising because models of PCR signaling and fast Ligon signaling already demonstrated that binders of lower affinity actually are better signalers and Mark and team, support this with both some In Vitro cell based models and some In Vivo mouse models. So really interesting paper and really shows the value of understanding not just the high affinity binders that you may have selected in a campaign, but being able to go broad across hundreds of candidates and identify different types of binders which display different on rates and off rates.

0:11:48.0 KH: So we, as I said before, we brought the Carterra in-house, the LSA in-house, a year and a half ago. We are leveraging this for our internal programs, but we're also providing high throughput screening, full kinetic analysis as well as epitope binning, which will allow you to do antibody pairing generally in the 96 to 384 Well Format. This is kind of a sweet spot for customers to begin to select candidates that they wanna bring forward based on additional information, whether it's the epitope that they're associating with or displaying different types of on rates and off rates.

0:12:42.2 KH: So, as I said before, some of the characteristics of Carterra LSA really fit our needs. It's high throughput, it's fast turnaround time. Let me go back one slide. I don't think I highlighted this. What ultimately what Carterra allows us to do is get 100 times the data in a fraction of the time. So within one week we can analyze thousands of antibodies, whether those are, purified antibodies or culture supernatant candidates. Be able to assign affinities to those, and actually do epitope binning.

0:13:21.2 KH: The sweet spot as I said before is really doing a one versus many candidates whether it's 96 or 384 on a 384 chip using one antigen to flow across that chip. Dan will go through kind of the mechanisms and microfluidics of how the instrument actually works, but this allows you to interrogate hundreds of antibodies at a time. If you're able to develop a regeneration protocol which we have done for most of our projects, you can reuse that chip and interrogate it against multiple antigens whether those are mutants or different other antigens that you wanna run across the Carterra LSA.

0:14:06.3 KH: So it provides high throughput which is important for our internal program, a fast turnaround time and really important is small sample requirements so we typically require our customers to give us 300 microlitres of the interacting so that typically is about a microgram per mil, at least of an antibody or 100 picomoles of... Excuse me, 200 nanograms of antibody. That's enough for us to do a full kinetic analysis. So very little materials required and the other thing is not a lot of antigen is required as well so roughly three micrograms for a typical 30 kilo Dalton antigen. Certainly, I'm gonna show some peptide data where you would require maybe 15 times that amount, but small amounts of antigens, small amounts of ligand, which allows you to do things at scale as well.

0:15:17.2 KH: One of the things that I'll mention that's important for our services is that there are chips for this instrument which allow you to capture and interrogate a number of different formats of antibodies or other large biomolecules. So a number of chips which allow you to do covalent detachment of your antibodies or targets of interest, but what we find even more valuable are some of the streptavidin chips or protein A/G chips which allow you to quickly capture antibody candidates and allow you to interrogate them quickly and I think Dan will talk about some of the chemistry associated with the chips that are available and we've leveraged all of these for both our internal customers as well as external customers.

0:16:13.2 KH: The first thing we did since we are a polyclonal antibody company, we wanted to make this switch to recombinant antibody development, but we had this portfolio of 24,000 affinity purified polyclonal antibodies that have been in the market for a while. We had some good data around kind of typical research grade applications. What we wanted to do is say, can we take our current polyclonal antibody portfolio and can you leverage something like the Carterra to provide more better, deeper knowledge around those antibodies? And this is one of the experiments that convinced us into kinda heading headlong into high throughput SPR.

0:17:01.3 KH: We took roughly 96 of our antibodies versus 96 different antigens whether those are full length proteins, partial length proteins or peptides. In this case we have both peptides and proteins. What was interesting is that we can actually generate one-to-one fits with these polyclonal antibodies that are affinity purified against the antigen that they're raised against and it actually allows us to generate KDs against polyclonal antibodies.

0:17:35.0 KH: And I'm just showing a couple of examples here for some important cancer signaling molecules, K-Ras. This is 48 different... Excuse me, 96 different sensor grams so provide you visualization of the on-rate and off-rate of replicate here so we did this in duplicate for this protein versus either the K-Ras peptide or K-Ras protein and we also did this against... Excuse me, 48 other antigens as well. So what this shows is the high specificity of this specific antibody, but also the binning kinetics which are in the picomolar range, the high picomolar range for this K-Ras.

0:18:22.0 KH: If we look at notch one example, we see similar things where we see high specificity and a lower calculated kD which is a measure of both the on-rate and the off-rate. What we like about this is that it is robust, we can generate CVs with replicates and it is allowing us to characterize our complete polyclonal antibody portfolio using this high throughput LSA.

0:18:54.3 KH: But what's probably more important and Dan will talk about some campaigns around monoclonal therapeutic antibodies, but in the development of new antibodies, I'm gonna talk at a high level about a use case for the Carterra around generating new recombinant antibodies to, I would, say complex or combinatorial post-translational modification targets.

0:19:28.6 KH: This is something that a number of companies have approached in different ways, and we thought that Carterra might allow us to quickly identify candidates during a typical antibody development campaign against PTM, whether it's phosphorylation, acetylation, et cetera. And the case that we're going to talk about is related to a target that I'm not going to name now because this is still in development, but we're excited that a combinatorial PTM, which is shown here in the fourth peptide down there that we call BOTH, PTM1 on the left and PTM2 on the right, comprise this combination of PTMs. We call one proximal and one distal. What we did was generate recombinant antibody candidates to the combination of PTMs labeled BOTH.

0:20:33.3 KH: We did this through a typical single B cell cloning, amplifying and cloning those heavy and light chains into HEK293 cells. And then even at small scale levels, we're able to generate enough antibody in crude cell lysates to allow us to evaluate candidates directly on the Carterra after that.

0:20:56.0 KH: And we evaluated that binding versus the peptide without any PTMs, the peptide with one of the PTM or the other PTM or BOTH. And we were pretty excited to find very different candidates within the campaign, which are recognizing different combinations of the PTMs and giving us very distinct on-rates and off-rates. We think we are going to be able to leverage this into development of not only research tool reagents, potentially diagnostic reagents, which are going to be more specific and more meaningful in interpreting biological mechanisms with antibodies like this.

0:21:46.0 KH: So we presented this actually at AACR last year, got good feedback on the posters, but this first one is versus PTM1. So we have 48 candidates and duplicate against PTM1. And what you can see is we get these typical sensorgrams. So if you look in the two boxes that are highlighted here, those are replicates. So this suggests that those antibodies are recognizing that first PTM with a measurable off rate. I think these off rates are in the nanomolar range or high picomolar range. But what's interesting is when you replicate this, right? So these are the exact same 48 supernatants versus peptide PTM2. And what you see is there are different candidates that are coming up that are recognizing PTM2 in the absence of PTM1 modification.

0:22:47.4 KH: And what's interesting is that the on-rates are very high for all of them, but the off-rates in most of them are also very high. So what this allows you to do is to identify candidates with different kinetic characteristics that are important for your application downstream. So for example, if you look in the second to last low on the right here, there are candidates here that are recognizing PTM2 very differently than some of the other candidates. But what's even more interesting is when you put both PTMs together, and this is what you probably would expect during an antibody campaign.

0:23:33.3 KH: This is using the antigen that actually was used as the immunization reagent. What we discover is very different candidates with very high KD, excuse me, low KD. So really, really low off-rates, really good on-rates, and it allows for stable kinetics for antibodies recognizing these PTMs in combination.

0:24:06.9 KH: And you can see that here when we overlay the PTM1, PTM2, and then both PTMs together, you can see that there are many different kinds of candidates in here which are recognizing those PTMs in different ways. We're now taking these with a partner and evaluating these functions in different applications. The red here is PTM1, the green here is PTM2, and you can see that there are many candidates with very high off-rates. But when you add the second PTM, you turn those candidates from a high off-rate to a very low off-rate and get very stable binding. And here are some of the examples. We've done this in duplicate also at a number of concentrations, and it allows you to robustly differentiate different binding characteristics, whether it's high affinity clones, which are in the picomolar range or what we're calling medium affinity clones, which are in the nanomolar range.

0:25:16.1 KH: The big difference that we've seen in most of our work is that a lot of the difference in overall KD is related to that off-rate, and you can see that here graphically. So, as I said before, we presented this last year at AACR, and what it allows us to do is in one campaign, identifying pan binders, but also binders that are either dependent on PTM or PTM1 or PTM2, or having PTM1 positive and PTM2 negative. So there's a number of reagents and research tools that can be developed out of this to dissect those different PTMs as it relates to biological response. And those are in development now.

0:26:15.8 KH: So Dan, I think this is my last slide, but what I'd just like the audience to know is that we found the Carterra LSA extremely flexible, so we're not just doing it for antibodies, though that is our main application. We have successfully executed a number of campaigns, not only around full length antibodies, whether they're human mouse or rabbit, and including polyclonal antibodies. I showed you the example before, but also scFvs, small molecule libraries and even DNA as ligands.

0:26:54.3 KH: As I said before, what's attractive to us is that the formulations for these ligands or antibodies are flexible. So we've used purified antibodies in a number of our applications, including the polyclonal ones, but we've also used crude cell culture supernatants as long as they're high quality in concentration. But we've also used a number of DNA conjugates as formulations.

0:27:29.6 KH: As I said before, some of the analytes tested against those ligands include not only peptides, protein fragments or full-length protein fragments, but also DNA as well as, a good example of antibody conjugated to nanoparticles which is resulting in some cooperative binding that we've been able to dissect. All of this is, really helpful for developing our catalog. We are upping our game as it relates to antibody characterization for the existing catalog and any new antibodies that come out. So not only testing antibodies in typical applications against different cells and tissues, but also running typical binding assays using Carterra SPR against, either purified proteins or antigens against those same antibodies. And then working with some open science consortiums that are generating knockout cell lines for all of the antibody targets that we're going after, and ensuring that our antibodies are specific as well.

0:28:38.8 KH: And we think that this combination of increased characterization, as it relates to binding the specificity requirements that we're showing, by evaluating knockout cell lines as well as leverage these in different types of tissues or cell lysates is going to help us develop much better antibodies as we go forward and convert our polyclonal antibody portfolio into a recombinant antibody portfolio.

0:29:13.5 KH: So I'd like to express we're open to working. We already offered this as a service. Carterra has been a tremendous partner with us setting this up. We are in San Diego. We're executing Carterra experiments every day, both for our internal pipeline, but also for number of biotech and pharma customers. Every day we learn about... A little more about the capabilities of these instruments and doing some really interesting, scouting experiments for not only antibodies, but other unique biomolecules and look forward to more from Carterra and partnering with them as we go forward. So I'll stop there and turn it over to Dan.

0:30:11.0 LT: Thank you, Dr. Harvey. Before I introduce the next speaker, just a quick reminder to everyone that you can submit your questions on the Q&A portal to the right of your screen and we'll answer them at the end. So let me introduce our second speaker, Dr. Daniel Bedinger. Daniel Bedinger helped launch Carterra's LSA platform and now leads the company's global application science team. He has over two decades of experience in the generation and characterization of therapeutic monoclonal antibodies, most notably at Xoma and Abgenix. Bedinger earned his PhD from the University of California Davis in cellular and molecular physiology. We can see your slides now. So Dr. Bedinger, please take it away.

0:30:57.4 Dr. Daniel Bedinger: All right, thank you very much. I wanna say I was impressed by Kevin and his team at the breadth of applications they've been able to work on, since they got the platform. It's a significant chunk of sort of the application space I think our customer base broadly is working on. So to see it all at one site in such a short amount of time is really impressive. So today I'm gonna talk a little bit about sort of the Carterra and the LSA platform in general, as well as I'll talk about high throughput epitope binning, and cover some of the applications we've been working on in infectious disease with it or... And our customer base has. Let's see if I can... So in early 2020 when the SARS-CoV-2 pandemic really started to become at the front of everybody's mind, guess that's a way to say it, the LSA platform had been on the market for about two years, and our base was really growing.

0:31:55.7 DB: And so at that point, all of these drug discovery companies working with the platform had to either make a decision, kind of stop working or do COVID two antibody discovery research or drug discovery research. And so pretty much all of them decided to do that. And I think the silver lining there was that there was a huge emphasis on speed, both in terms of the drug development and getting things into the clinic and the market, but also in publication.

0:32:28.1 DB: So we had a bit of a rash of really good publications that showed the importance and the utility of showing high throughput epitope binning early in the discovery process. And so I'm gonna go over some of those examples at the beginning of this talk then I'm gonna introduce the CoVIC consortium work that I was a part of which we published in science at the end of 2021.

0:32:54.8 DB: And we'll also mention our new version of the LSA, the LSA XT in this talk. And just as Kevin made clear, Aviva is a great customer of ours, and they offer all of these applications I'm gonna talk about today as a service, so hit them up. So why do we do competitive epitope binning in the first place? So the functionality and mechanism of action of an antibody is really linked to its epitope. It's relatively straightforward nowadays to assess the affinity of the antibody, and it can be engineered and optimized in a variety of ways, but that epitope is really innate and can't, at least as of now, be rationally designed. So it needs to be screened or selected. So by doing early epitope characterization, you can ensure that you're carrying meaningful diversity forward in your funnel.

0:33:49.9 DB: Also, you can assure that your discovery process is generating a diverse set of antibodies, not just in sequence space, but in epitope and functional recognition at their earliest points, sometimes certain immunization or phage selection methods can lead to biased output with predominant epitopes. And so this is a very good way to ensure that doesn't happen. It can also be used to inform large sequence sets. So with the integration of NextGen sequencing and some of these higher throughput antibody discovery workflows, people end up with large numbers of sequences to putative antibody binders, and they really can't characterize them all in a deep way. So if you can do sort of an in silico analysis and assess a representative subset of your antibodies to screen in a high throughput kinetics and binning, you can then sort of backfill a large understanding of your sequence space based on that.

0:34:54.3 DB: We have customers that will express 200 or 300 clones out of several thousand that they have, but they're able, usually if they do that selectively and pick representatives, they can learn a lot about their entire sequence space and know where to go back and mine for things with properties they're interested in. It's also a powerful tool, as Kevin mentioned, to identify sandwiching pairs, in 24 or 48 hours you can find which clone is a good capture, which is a good detection for the development of amino assays. Also you can establish IP and freedom to operate. Sometimes there are certain epitopes or competition profiles that are claimed in patents, especially some little bit older patents and if you have a comprehensive epitope binning, you can often find differentiating factors in that data set.

0:35:45.8 DB: So the first example from the literature I'm gonna talk about today is the collaborative work between Eli Lilly and AbCellera, both are Carterra LSA customers, and they, in 90 days, well, it was actually 94 days into patients, from when they received blood from one of the first COVID recovered patients in Canada, they were able to screen spike binders, they expressed 187 of them in very, I think, less than two weeks, at which point they pushed them into a detailed characterization process on the LSA, where they looked at affinity, premix, epitope binning versus the spike protein ACE-2 Neutralization. And they were able to down select to the highest affinity of 24 antibodies from a diverse set of epitopes to take forward into subsequent characterization and developability assays.

0:36:50.0 DB: So they thought enabled them to file an IND 90 days later and be in patients in 94 days. And they actually were able to get emergency youth authorization for Bamlanivimab which was the first anti COVID-19 biologic on the market. Another example is the group, Josh Tan's group at NIAID, which is a division of the US NIH, they made, using also human B-cell derived antibodies, panels of SARS-COVID-2 neutralizing antibodies. And they did, epitope binning of their set. They found four non-overlapping RBD epitopes and one NTD epitopes that contain potent neutralizers and they wanted to find mutational resistance and also synergistic potency by combining those antibodies into cocktails. What they found was that with cocktails, really the potency advantage was additive, it wasn't synergistic. So they had the idea of making a double variable domain IgGs with these using a selection of epitopes that they knew would be non-competitive.

0:38:04.8 DB: So they were able to construct a relatively limited number of those and screen them for potency. And what they found was that in some cases they had up to 100 fold increase of potency of this DVD-Ig construct biparatopic, it's like a bispecific, but targets the same protein, antibody versus the cocktail it's shown here, so that the black curve is the combined of the two monoclonal antibodies and then the biparatopic antibodies are much more potent. They went on to do some cryo-EM studies of those and found there was really interesting mechanism for the most potent clones where they were getting this, inter spike cross-linking that was forcing the protein to be in a really inactive confirmation.

0:38:55.5 DB: Another example, this is another publication from the Lilly and AbCellera work, is actually a continued screening of antibodies from that original patient source. And they realized that as... So the original Bamlanivimab was found so early that there wasn't a whole lot known about all the mutations that were going to occur and I mean, people knew there was going to be mutations, but they didn't know what they would be.

0:39:23.8 DB: So by few months later, it was obvious they were occurring rapidly. So they used additional screening to try to identify clones that bound to epitopes that were highly conserved across all the variants, and they found Bebtelovimab. And this was actually also approved by an EUA by the FDA on February 11th, of 2022, and lasted until, well, the Omicron BQ 1.1, I believe, in the end of November of 2022, where the mutational escape occurred, but it was sort of the last man standing in terms of monoclonals on the market. Two more examples shown here. One is the group at Twist in collaboration with La Jolla Institute for Immunotherapy and the Crowe Lab at Vanderbilt used four libraries using both human CDR sequence and llama VHH CDR sequences to find neutralizing antibodies and characterize them in shared epitope space.

0:40:35.9 DB: And then they used those to test in rodents cocktails of antibodies that would be neutralizing against broad COVID strain specificity. And the last one of these I'll go over here is the group at Abveris also was using high-throughput epitope binning and kinetic characterization to compare the output of two different types of antibody discovery, where they were looking at clones derived from B-cell selection and hybridoma and they found... Using their transgenic mouse strains and they found that they got actually slightly different epitope recognition between the two approaches. So that was a really interesting data set for them and also a heavily utilized epitope binning in that discovery process.

0:41:31.2 DB: I will talk more about this publication near the end of the talk. This is the science paper from the CoVIC Consortium, which I was a part of where we... And I think this is an interesting paper in that it really is probably the richest example of combining competition based epitope binning analysis with structure. They have a cryo-EM, they're at LJI and they... I believe at least 40 antibodies have been analyzed by structure analysis and overlaid into this Epitope binning mapping in a sense. So we'll come back to that. And then the last one I'll talk about is not just COVID. The group at Twist did a really interesting study where they took a bunch of already published, I think it was over 200 antibody sequences from patients that had recovered from Ebola. And they wanted to see if they could leverage their synthetic DNA technology to rapidly express, characterize and identify unique clones in this set.

0:42:35.4 DB: So they did that. They cloned these antibodies and then very rapidly did affinity and epitope binning of them. And in the epitope binning assay, they actually included some, they called them pathfinder antibodies. These are antibodies that had already been published of known structural binning epitope in there. And so they were able to use that to inform this larger binning set of all of the patient-derived antibodies, and found that sure enough many of the clones did cluster to these known epitopes, but they were also able to find antibodies that bound to unique epitopes that hadn't previously been described. So if you were going to pursue development efforts, those would potentially be some of the most interesting clones to include in that analysis. Again, this was a rapid process in a couple of weeks they did this whole workflow.

0:43:30.0 DB: Okay, so now that I've hopefully shown why we think this is interesting, we can talk a little bit about our platform. So this is what I call the LSA ecosystem. So we have the LSA instrument and then our new LSA XT version of the platform. We have three software packages that people use using the platform. One is called Navigator and it controls the instrument, collects data, use it to set up the assays. Then there's two analytical software packages. One is called kinetics and one is called epitope. It's relatively straightforward what they focus on. And then we have a whole line of biosensor chips and consumables.

0:44:08.5 DB: So the LSA is an array based SPR platform, and it's really ideally suited for many of the core applications in antibody discovery and characterization or generally biologics, discovery and characterization, those being kinetics and affinity analysis, competition based epitope, binning, peptide mutant mapping and quantitation.

0:44:30.9 DB: So in addition to having a hardware architecture that makes these assays very high throughput fast and minimally sample consumptive, we put a huge amount of effort into having analytical software packages that make processing these large data sets. So both fast, relatively easy, but also very visually rich with lots of tools of ways to look at things, including in our epitope tool, the sorted heat maps and network plots. So the way the system works, you can think of it as having a sensor chip that docks into basically the center of the instrument. And then we have two robotic semi-independent fluidic systems that address the chip. One is a 96 channel side that can be used to immobilize up to 384 proteins onto the array surface at a time. And then the other is a single flow salt that flows one sample over the entire array area.

0:45:23.3 DB: And this is where we get what called one on many binning studies and is responsible for the really highly parallel nature of the analysis. So just a little bit more resolution on this. So this is looking at the right hand side sample deck of the instrument. You can see there's 96 needles positioned above a microplate. There's three plate bays on the instrument that can hold up to 384 well plates. So you get 11 or 1,152 ligands possible in a single unattended run.

0:45:51.6 DB: So the needles descend into the plate and aspirate sample, which then flows to the chip surface and cycles back and forth to allow you plenty of contact time for immobilization or capture. If you're capturing from crude samples, the needles can then go wash, go pick up more sample and print additional or capture additional locations onto the array, creating up to a 384 spot array. Another cool feature of this is that the samples are actually returned to the plate when using the 96 channel side, so you can get them back to spot them again or to use them at a different assay. It's just a map of how this layout works. So the pink vertical rectangles are 96 spots that are flow cells that are created when the print head device, that's what we call that, is docked. And then the blue are 48 inter spot references used for real time referencing during single flow cell injections. And then we can create a high density 384 region of interest or spot array as shown on the bottom by interlacing four of those prints.

0:47:00.2 DB: The other fluidic system is the single flow cell. So this is where one 270 microliter sample volume is injected over that entire array area. So this can be used for preparative steps if you're making what we call a capture lawn or doing regeneration or chemistries on the chip. It also can be used for kinetics and epitope competition studies. In this case say we're injecting a single concentration of antigen, you get parallel data from all 384 active spots at the same time.

0:47:29.3 DB: So it's highly sample efficient and allows doing these large scale assays very quickly. Again, one sample in this case is being flowed over everything we've immobilized on the array, including the references in the active spots with the single flow cell. As Kevin mentioned, there's a variety of chip chemistries available on the platform. The green and orange ones are different forms of carboxylic acid functional groups. So these are sort of general purpose chips, that you can immobilize essentially anything you'd want to it, using well understood chemistries. Then we also have some more ready to use chips like nickel NTA for capture of His-tag proteins. Several different thicknesses of streptavidin. We have protein A and protein A/G for capturing Fc.

0:48:22.7 DB: So typically, with the set of chips shown here, we're able to achieve almost all of the applications. Our customers come to us. If there are things that are needed that are not on this list, there is some capability for us to access some custom surfaces as well. So there's always that option. All right, so I'm gonna briefly talk about high-throughput kinetics on the system before we go into the epitope binning. It is one of the real common applications people use on the platform. So a typical assay would be shown up... Set up as shown here on the right where you would immobilize some sort of capture molecule. In this example, it's an anti-human IgG Fc polyclonal. And then you would capture the antibodies of interest to the array. So this is the ligand step and uses very little material. And then we would inject a titration series of antigen over the entire array.

0:49:15.7 DB: And in this assay format, you can do up to 1,152 unique ligands because we can repeatedly build the array during the course of an experiment. And the antibody source can be either crude or purified because we're essentially doing this affinity capture step. It's basically like we're purifying the antibodies onto the chip. So this is an example of what the data from the experiment, I just outlined looks like. This is 384 interactions. It's actually about 38 clones spotted in eight to 12 replicates each kind of spread around the array. This entire analysis set up in an afternoon run overnight used seven micrograms of antigen. So this is PD-1, a 17-kilodalton protein where the titration series started at one micromolar and in a threefold cereal down. Yeah. So it only used seven micrograms of protein for this entire analysis.

0:50:11.5 DB: Also shown here are some of the data highlighting features in the analysis software. Pink is kinetic limitation of slow off-rate beyond a limit that you set to prevent reporting ridiculous results. There's non binders, there's things with higher error. So there's some great visualization tools and sort of data curation tools built into the software as well. A lot of times people don't have 384 unique antibodies. The array is still very powerful for analyzing that in that you can actually generate a significant end of the analysis.

0:50:51.9 DB: So this is the same clone spotted 12 times three slightly different densities, but it's all fit together. Shows great kinetic equivalency between all of the replicates. And there's a page in the software where it actually will calculate the mean and standard deviation of the rate constants based on the replicates as you name the samples. So there's some visualization tools as well. This is what we call an Iso-affinity plot. So the off-rate is the x-axis, the on-rate is the y-axis and the diagonal blue lines are single affinities. So you would have 100 picomolar or one nanomolar or 10 nanomolar.

0:51:28.9 DB: So it's a great way to just visualize kinetic diversity within a set. And in the actual analysis software, there's some cool interactive features where if you hover over a dot, a little window will pop up and show you the actual Sensor grams then sample name and the rate constants for that.

0:51:47.9 DB: Another example I wanna show of a kinetics assay is Fc gamma receptor binding. So this experiment was actually run on LSA XT, but it would work equivalently on the LSA. So 11 human and cynomolgus Fc receptors were immobilized onto the array in eight replicates at four different concentrations, each on a streptavidin chip. And then an antibody was injected over a broad two-fold concentration series. And this is some of the data from that run, you can see some of the replicates go across the rows. We're able to get very good kinetic description of both high and low affinity interactions and very consistent results across the replicates.

0:52:34.3 DB: So this is... This assay requires very little sample to run. You can do it with a high end and it's quick. This was set up in an afternoon run over an evening. So these assays can now be scaled in a robust way to include many different Fc receptors, interrogate many different analytes or proteins of interest. And do so quite easily. Just zoom in on some of the data, both a high and a low affinity interaction. So we're able to get well resolved kinetics for both micromolar binders and, well, in this case it's a high picomolar binder, but.

0:53:14.3 DB: Okay, so I mentioned the two different platforms. So the LSA is a platform we've been selling since 2018. It's really established itself as a market leader in high throughput antibody characterization and it's excellent for detailed antibody and biologics kinetics, especially if you're trying to incorporate this early in your discovery process when you have high throughput and often crude samples. And it can do epitope binning at a really meaningful scale.

0:53:39.7 DB: So the XT has all of the benefits that we've talked about for the LSA, but it has more optimized optics and detectors. It makes it more suitable for doing things like small formats such as very small peptides, things like small molecule drugs like targeted protein degraders and molecular glues. Also it has some advantages for things with rapid interactions such as Fc gamma receptors and cytokine panels. The same stuff is true for the XT is the LSA. We like to say a hundred times the data, 10% of the time, 1% of the sample versus competing platforms. But now you get three times better signal to no noise, two times better signal uniformity and a two and a half times faster data collection rate.

0:54:28.9 DB: So just one more example of the XT and some assays we're really excited to show for it. This is a kinase, so receptor tyrosine kinase is immobilized onto an HC30 chip and we're injecting small molecule kinase inhibitors starting about three micromolar. And this was done in the presence of 3% DMSO. So the kinetics look really nice. This is staurosporine, a 466 dalton small molecule. We're able to get well resolved correct kinetics to all of these receptors. Obviously, it's a very pleiotropic kinase inhibitor. But these kinetics generally match what we see in the literature. If we look at a different one, sunitinib, it's 530 daltons. We get a very different profile. It's more selective and binds with weaker affinity to things like IGF1R, but this again is consistent with what you'd expect.

0:55:25.0 DB: So we're happy that we can incorporate small molecule analysis to some extent on the platform. And in particular, this assay could be really useful for people interested in kinase profile, where you could capture a large number of different kinases to the chip and generate kinetics of an inhibitor to the entire panel in a single assay. That used to be a quite a bit of work using other means. Okay, so now we're gonna focus more on the epitope binning application.

0:55:55.6 DB: So the LSA is really the only label-free biosensor that allows you to scale these assays in a symmetrical way to a large scope. So the LSA can do up to a 384 by 384 Epitope binning. A lot of people operate it in the 96 by 96 or 192 by 192 range, and there's lots of interactions there. You have a 96 by 96 assay is over 9,000. A 384 by 384 is almost 150,000 independent interactions you're analyzing.

0:56:28.8 DB: This is important in that each unique interaction can be thought of as a probe. So the more clones you include in your Epitope binning assay, the more likely you are to see meaningful differentiation among clones within your sample set. So the less you interrogate with, the less resolution you have on the output. So the way we approach Epitope binning on the LSA is we create an array of antibodies and then inject antigen followed by a sandwiching antibody, or if the antigen is multivalent, you would inject that antigen in the presence of a saturating concentration of the potential competitor antibody.

0:57:12.0 DB: We then analyze that data in our epitope tool which has tools for doing normalizations and report point setting, as well as cutoff setting of those binning events. And that will automatically populate these sorted heat maps and network plots. This is kind of a view of the working screen in the software where you see these three panels.

0:57:37.7 DB: There's a sensorgram view, a heat map view, and then the heat map, the immobilized antibodies are the rows, the injected antibodies are the columns. And the self versus self interactions are the dark bold, and black outlined. And if something is a red intersection, that is a blocking relationship and if it's green it's sandwiching. So this can be sorted. And then the software will also automatically make these network plots that show the relationship between clones. So a connecting line means that the clones are competitive with each other and if they're contained within a colored shaded region they're part of the same epitope bin.

0:58:18.4 DB: And what's really exciting about using this platform is that the three panels are interactive. So if you click on a cord in your network plot, it will actually highlight the cells in the heat map and show you those sensorgrams. So it allows you to really analyze the data in a deep level and really pay attention to subtle differences and investigate any complexities or behaviors that seem interesting in all the way from the raw data to the fully interpreted data. And it works in all the orientation, so if you click a sensorgram in the sensorgram view, it will highlight it in the heat map and in the network plot. So it works in all directions.

0:59:00.8 DB: So I mentioned these network plots. So the base level network plot can be very granular. A single difference in competition profile will break something out of what we call the epitope bin or the bin cluster. One thing the software does when you sort that heat map is it automatically generates a dendrogram, which shows the extent of relatedness of the competition profile of the clones. You can use this dendrogram to set a custom cutoff height and build what we call community level plots. So these are ones where it's more generalized clusterings, essentially sort of for giving some differences, between them to create an interpretable or a more easily interpretable map. Another cool feature in the analysis software has to do with visualization and that's where once you have a network or a community plot that you like, you can load in tables of orthogonal data and color it by that.

1:00:03.1 DB: So here's three examples. Whether or not the antibodies are mouse cross reactive, whether they came from a chicken, a mouse, or a commercial source, what sub-domain they bind to. So these are all categorical examples of coloration. You can also import numerical values like potencies or affinities in color on a gradient. And so these are all visualization tools built into the Epitope software that you can just toggle between the different options to create different images for sharing data.

1:00:37.5 DB: So, Kevin mentioned this, and I'm gonna reiterate here, this is a really powerful tool if you're trying to develop immuno-sandwich assays for ELISA, Luminex, MSD, lateral flow. And typical the people in biomarker groups or PKPD groups will make anti-idiotype things or even people monitoring infectious disease need these reagents. So if you have a panel of antibodies that bind to your target in about 24 hours, depending on the sample size, maybe 48, you can do a rapid kinetic assessment to understand how well each of these clones bind to your target, as well as the full binning competition assay where you get to see in a pairwise interaction every antibody behaving as both a ligand and an analyte or a capture and a detection as sandwich amino assay speak. And so you can really see directly which clones bind rapidly to antigen and hold onto it stably and which ones pair well with others and make stable high signal complexes.

1:01:42.4 DB: So it can really inform the downstream, final development of those assays in a rapid fashion. Okay, so now I'm gonna talk a little bit about the CoVIC Consortium and the work we did there. So this was a global nonprofit collaboration organized by the Bill & Melinda Gates Foundation and LJI, where they asked the industry to submit antibodies that targeted SARS-CoV-2, spike protein.

1:02:12.2 DB: And they anonymized them and distributed them to a number of labs for various characterization aspects to try to see what are the best antibodies that could target this protein and what can we learn about that in the process. So there were two groups involved in this that had the LSA, so both, Carterra, ourselves and Duke University, the Georgia Tomaras Lab. They largely focused on kinetic characterization of the antibodies. And we did the RBD Epitope binning.

1:02:45.6 DB: So we can go to the next slide. So of the 397 antibodies entered into this collaboration, about 75% bound to the RBD as shown here as it's a portion of the spike molecule. Some of the early mutations are shown here, there are many more now. And so we did full Epitope binning of this panel of RBD binders against that. As mentioned, the group at Duke used LSA to do kinetics. You can see a comparison here of the RBD, which was a monomeric construct versus the spike which was a trimeric construct. So you can see lots of kinetic diversity when binding to the RBD and still some kinetic diversity, but clear avidity effects binding to the spike. So both of these assays were done and recorded along with all of the other data into the COVID database. This is an example of a representative heat map of an experiment I did fairly late in the collaboration.

1:03:45.2 DB: This was 170 ligands by 188 analytes, or about 32,000 interactions total in the map. And this is what you would like an Epitope binning experiment to look like, where you get self versus self competition all throughout, lots of sandwiching and competition and clear clusters of shared behaviors. And so when we apply the community analysis to this, you can see we find communities like this yellow one here where all of the clones are competitive with each other and they share many blocking and sandwiching relationships.

1:04:21.8 DB: Some of the value of doing this at this scale can be highlighted here by the neighboring community. So this is a similar community there. It's competitive with the yellow, the purple, and it shares many blocking and sandwiching relationships. But you can see we have these two fairly rare communities that are shown as the blue and the yellow where these clones sandwich and the other ones compete.

1:04:49.6 DB: So this is a clearly meaningful differentiation between these two clones. And it shows that these clones bind to an epitope that has shifted, relative to the other. There's a space that these are binding that is not accessible in the other one. So this is the kind of resolution where the more clones you include in the assay, the better your understanding of the epitope diversity and the more likely you are to be able to elucidate these nuances.

1:05:16.6 DB: So this was done to this panel. This was published at the point, there was about a 100 antibodies in this when the science publication went out. But we were able to apply the clustering and then they sampled into this for cryo-EM, and we were able to map the binding epitopes that is recognized by each of these communities onto the Spiker B, which is shown in these images below the ACE-2 binding site. The Orthosteric site is shown in the dotted line, and then the colored region shows the epitope for these different communities. And you can see that we have everything from outside of the ACE-2 binding domain to ones that are essentially Orthosteric and then shifting or across various phases of the molecule. So I think this was really exciting and it was really great to see how closely the dendrogram communities and the Epitope binning informed all of these structures that were be generated by LGI and the Cryo-EM.

1:06:16.0 DB: So this was then overlaid with analysis to resistance to mutations and up through about Delta. This was really exciting. We could see that there were clear sort of branches on the dendrogram or discrete epitope clusters that were just not binding the locations that were being affected by these mutations. And so we had maintained neutralization across the clones in that cluster. Then as you can see here, the Omicron strains emerged and the mutational space went wild. Many more mutations were accumulated and the amount of sort of pre-existing epitope space that wasn't being modified became a lot less. And what we found then is that it became a bit more discreet, little pockets of clones and space would still be resistant and bind to and neutralize the Omicron. But so much of the available epitope was being affected by those mutations. It was a little bit less discrete in terms of large branches of the dendrogram showing resistance.

1:07:27.7 DB: And this was actually summarized more recently in the Cell Reports paper where they actually realized that there were two mechanisms to have broad neutralization, especially against the Omicron strains. Those being that the antibodies bound bivalently within a spike or the antibodies bound to sort of very small, highly conserved epitopes, the avidity effect of this bivalent binding seemed to overcome some loss of binding affinity by the clones. And that, so I'll, to refer this paper came out in January of this year.

1:08:08.8 DB: So with that, hopefully I convinced everyone that high throughput Epitope binning is broadly becoming an integral part of early antibody discovery projects. And the epitope competition data can really be a crucial aspect for selecting antibodies particularly if you're interested in making cocktails and bispecific, but just in ensuring clonal diversity going forward in screening campaigns as well.

1:08:35.2 DB: The LSA allows kinetic and Epitope binning to be done at an un-rivaled scale with very modest sample requirements. And so with that, I wanna thank you for attending. I wanna thank Aviva for the invitation to participate in this. And again, I'll remind everyone that Aviva is an early adopter of the Carterra LSA platform and can provide this type of data for your projects. So, again, I'll just leave this up, but we can end here. I wanna thank all the people involved in all the work I've been talking about today, both at Carterra, the Josh Tan's group, the NIH, and the CoVIC Consortium. So thank you. Hopefully we can take some questions.

1:09:17.7 LT: Thank you very much, Dr. Bedinger. Yes. In the time that we have left now, we will get to our listeners questions. So the first question is for Dr. Harvey. What are the requirements for sample preparation for a characterization run?

1:09:36.7 KH: Sure. This common question actually, we get this question from every one of our external customers, and Dan, feel free to chime in, but generally it just needs to be physiologically relevant buffers. It can contain protein, whether it's BSA or some protein that is coming from cell culture soups, for example. But sample preparation we don't do much of. They don't need to be purified. We kinda highlighted the minimum requirements, so we prefer at least 300 microliters. We could probably do less than that. With, as I said before, concentrations for typical full length antibody of at least one microgram per mil. And on the analyte side, if you're using typical kinds of proteins they just need to be at that minimum concentration.

1:10:37.7 KH: We do, do dilutions both for the ligand and the analyte to both for scouting experiments, but also to generate the centigrams. So in general, when we work with our customers, the more is always better. But we can go very low as Dan highlighted down to the kinda nanogram and picogram levels of total analyte and ligand. Dan, anything else you wanna add? Kind of our customers generally are either giving us typically purified antibodies or CFPs or culture supernatant.

1:11:22.2 DB: Yeah. Limits are always pretty assay specific based on what you're doing. For things like capture kinetics, we find with human IgGs, usually 50 nanogram per mil concentrations can capture to the chip surface and still give a meaningful kinetic signal. Obviously for Epitope binning we can do that from soups, especially things like HEK 293 soups that are relatively high expressing. It becomes a matter of whether the analyte is being injected at a concentration high enough to drive association in the five minutes we wanna do that injection for something. There are some assay specific requirements, but generally it's pretty minimal as we showed for the PD-1 example, if you had seven micrograms of antigen of that 17 kilodalton protein, you could do a full titration series starting at one micromolar. So, yeah.

1:12:22.7 LT: This next question is for Dr. Bedinger. What is the advantage to performing epitope characterization by SPR instead of other methods like ELISA?

1:12:38.0 DB: So I think it's really throughput and simplicity of that. The array allows for the highly parallel analysis to occur so from that one sample injection, you get data to everything that's immobilized on the array. So things scale linearly. If you wanna do, say a 96 by 96 Epitope binning, you have 196 well plate of antibodies to immobilize and another 196 well plate of analytes to inject to perform the sandwiching and a tube with your antigen and regen solution in the assay. And that's really all the setup you need. If you were to try to do a 96 by 96 matrix on an ELISA, you would need to one, probably modify half of your antibody so that you have a biotinylated form or labeled form, but you'd also have to have basically 196 well plate per clone to generate symmetrical data. So instead of being a two plate setup, it would be a 96 plate setup for a full symmetrical matrix.

1:13:44.8 LT: Okay. Another question for Dr. Harvey. So once you have identified the antibodies with different binding characteristics, how do you use that info for follow up experiments?

1:14:05.0 KH: Certainly for our internal projects, depending on what the application is downstream, we tend to go straight to that application. For example, if we're trying to develop a clone that's gonna be used in flow cytometry, we'll go straight to that application, we tend not to do follow up affinity experiments. The data that we're getting out of the Carterra is sufficient for us as really a screening tool, but basically directing us to the clones that are most likely going to perform in the application that we're trying to develop. For our customers, we basically generate a report which will list KD sensograms. If we're doing epitope binning, we'll share all those maps with them. Normally our customers will take that data and then go off and do their next step whatever that may be. Our follow up experiments, there's none required we immediately move to the downstream application for a researcher diagnostic tool.

1:15:20.2 LT: Okay. This next question is for Dr. Bedinger. Neil would like to know using your methods, are you able to distinguish competition between monoclonal antibodies due to Steric interference versus allosteric effects?

1:15:34.9 DB: Yeah, that's a really good question and I guess the answer is sometimes. So one of the benefits of doing sandwiching assay analysis with real time label free SPR is that you get to see the data while it's being generated and not just rely on an endpoint value. A common thing we see in especially classical sandwich epitope binning is that if you have an antibody that is sort of allosterically interfering or maybe Sterically hindering a sandwiching interaction you actually see a fairly rapid formation of tri molecular complex, but then an accelerated dissociation of the antigen and the sandwiching antibody from the capture surface from the immobilized ligand antibody.

1:16:25.0 DB: We call that kickoff effect. And so you can cluster clones into different behaviors or at least, identify them where you say, okay, well, clearly this is sandwiching, but then by the time I'm taking my report point it's either below control or it's negative value and so that is elucidated it's not exactly a specific assay format to ID that and some things like steric interference, there's not a lot of binding that can distinguish between a blocker and a steric inhibitor. But there's definitely quite a few cases where you get this allosteric modulation or change in kinetics of the binding that becomes quite evident in the data.

1:17:10.3 LT: A related question also, do you ever encounter problems with self associating monoclonal antibodies?

1:17:21.5 DB: Not frequently, sometimes you find antibodies that will self sandwich but that's usually an issue with the antigen preparation. It'll have a certain percentage of dimer that's allowing that to occur. I have seen some antibodies that are just sticky. Usually, they're non-traditional constructs like engineered human VAHHS or things coming from synthetic sources that have strange modifications. A lot of the more naturally derived style antibodies don't have that kind of behavior at least that the concentrations that we're working with them usually we're injecting them in tens to hundreds of nanomolar, so not too much self association.

1:18:11.6 LT: And this next one is for Dr. Harvey. What is the typical timeline with this antibody characterization services?

1:18:21.3 KH: Yeah, from the service perspective we generally give a one week timeline. That the experiment can be executed usually in one day. There is some setup and then as we evaluate data and generate a report, we need a day on each end we'd like to do, so if we receive samples on a Monday absolutely we turn things around within a week, especially if it's just typical affinity measurements. Sometimes for larger epitope binning experiments we need to do some initial scouting experiments to determine kind of how stable and how we regenerate, depending on what the ligand and analyte are. So epitope binning can stretch to a week and a half to two weeks because we need to do some additional experiments, but general turnaround is a week for our service.

1:19:22.4 LT: Another question for Dr. Bedinger. Can the Carterra platform be used to study RNA nucleic acids and RNA small molecule interactions?

1:19:36.8 DB: It should be able to. I'm trying to think if I have any specific examples that come to mind for the RNA nucleic acids. So we've done a number of Aptamer studies where we bind RNA by based binding proteins to the chip and look at their association with proteins. I believe Kevin, you've done some DNA binding studies and I have not yet done RNA to small molecule, but I don't see why you couldn't. It's important when designing LSA assays at least if you intend them to be of any large scale that you have the ability to get your diverse species onto the array.

1:20:16.2 DB: So if you have quite a few RNAs, you could capture them or immobilize them to the chip surface then, especially on the XT inject, the binder or the targets to that, or your, I guess, maybe the drugs, whatever your RNAs are targeting over that and collect that kinetic data in parallel just like we showed for the kinase inhibitors.

1:20:40.7 KH: Yeah. Dan, I agree. It's an interesting question, and we've actually discussed this with a number of customers. We haven't executed yet on an RNA project we don't see any reason why the LSA could be used for this. As you said, we have done DNA versus antibodies, DNA versus nanoparticles. So we think we should be able to do that given the right kind of screen.

1:21:08.8 DB: Mm-hmm. Thanks Kevin.

1:21:14.0 LT: Going back to Dr. Harvey's talk, if the analysis identifies a large number of candidate pairs, what would be the selection strategy for what binning pairs to pursue?

1:21:26.4 KH: Yeah, and I think Dan kind of highlighted the fact that you can create communities. This is kind of a higher level binning as we call it. And if we for example, generate too many bins for us to evaluate in the next step, we tend to start with the communities. So we'll select candidates from different communities to pair together. That's generally the first step after that. That's how we choose. Dan, any other insights there?

1:22:00.4 DB: Yeah, I think when you're doing those pairing assays you really get to see how rapidly the clone and stably it binds the antigen, how rapidly and then stably the sandwich forms. So you can usually even among the different sets where they sort of share competition profiles, pick the most likely clones to give you the highest signal in those other assays. And for larger epitope binning in drug discovery, it's all about ensuring diversity or evaluating the most interesting or highest affinity or most potent clones from the various epitope clusters to ensure that your downstream screening assay is not biased.

1:22:48.1 LT: Okay. Wesley has a question for Dr. Bedinger. So related to the question about experiment requirements, do the small molecule kinetics on the XT require high immobilized light and densities?

1:23:02.7 DB: Yeah, so on the LSA like with the most other SPR platforms, we see sort of Stoichiometric recovery of binning signal typically. So in the case of that small molecule binding, we are getting 30 RU signals or so of those small molecules. So there was a lot more of the, say, 50 to 70 kilodalton in protein immobilized than you were seeing in those sensorgrams. But it wasn't what I would say extremely high level. The chip type that was used for that assay was just our standard HC-30M, which is sort of our medium low capacity surface.

1:23:40.5 DB: So there was probably... I don't specifically remember off the head exactly how much RU, but it's probably three 3,000 to 4,000, maybe slightly more than that RU of immobilized protein. But it wasn't like we were using ultra dense chips or pushing really hard to get very, very high immobilization levels to get that data. It's sort of well within the sort of normal operational range of the instrument and our ideal chip types.

1:24:08.5 LT: Okay. Another question for you, Dr. Bedinger, how does the epitope mapping handle intra-assay precision? Is there a plus or minus standard deviation thresholding layered onto to, to know how significantly a ligand affinity has been impacted?

1:24:31.0 DB: So our peptide mapping or the mapping application where you would immobilize a bunch of different species than inject analytes and see which ones they bind into well. It uses analyte normalization approach. So it looks at like peak binding versus relative binding. That is a relatively simplistic interpretation and in terms of replication, you can include as many replicates as you want and it will really analyze them independently. But that just enables you to compare the results between those. If you're actually curious about the effect on the affinity of mutations or different like say peptides, you can design the experiment and actually run it more as a kinetics experiment, at which case you can generate a specific N that would calculate mean and standard deviation and actually understand the difference in the rate constant between those.

1:25:30.7 DB: So I would say typically like our epitope mapping application is a bit more... It's supposed to guide you to where the protein binds, it's not designed exclusively to quantitate it. That would be more of a kinetics experiment where you would probably go back and run kinetics on the relevant peptides of interest or mutants of interest for the mapping. Hopefully, I answered that question. Okay.

1:26:02.4 LT: Thank you. Yes. For Dr. Harvey, what range of temperatures can you run the assay at?

1:26:11.3 KH: I think the LSA, Dan, is we've done a number of different temperatures depending on what the analyte ligand requirements are. Generally, this isn't a big issue, but some may be required and Dan could talk to that. But I think the instrument can be controlled at least the platform between 15 and 32 degrees.

1:26:33.8 DB: It's, I think, technically 15 and 40 is the analytical range. And then the chip temperature and the analysis temperature is controlled independently from the sample deck. So the samples are also kept, well, typically cold. I guess you could warm them up if you for some reason wanted to, but typically people will set a chilled sample deck and then run the analysis at something between 15 and 40 degrees C.

1:27:01.2 LT: Yeah. Another question related to the instrument. Does it use strong lasers and is the temperature on the surface well controlled?

1:27:14.1 DB: So yes the temperature on the interaction surface is well controlled, you know, it's heated both from above and below within the instrument by regulated thermal system. I don't know if I would say it uses strong lasers. I don't think it technically uses a laser at all. It has a fairly bright LED light source that's highly polarized and collimated to illuminate the chip at a very specific wavelength. But it's not certainly not like a dangerous laser and it's not a true laser, it's just a very bright LED based, highly coherent light. It doesn't heat the chip up though. It's actually the light source is far away from the chip and the light that gets to the surface. Well, that's reflected onto the underside of the prism and doesn't interact with the sample really at all.

1:28:19.8 LT: Okay we have time for one last question. So this one is for Dr. Bedinger. My lab doesn't have the capabilities to purify all of our antibody candidates. Can I still get kinetics and epitope binning information on the LSA using crude samples? How much supernatant should be required?

1:28:43.0 DB: Yeah, so absolutely a large number of our customers do kinetic characterization from soups or even bacterial extracts and you don't need much. The amount depends somewhat on the expression level, like for recombinant soups sometimes we'll dilute them 300 fold. So you would need a microliter at that point of soup to run that analysis. If they're good expressing some... But I think with things like B-cell supernatants that are very low antibody content usually we'll dilute those like one to one in running buffer. So you would need a minimum of about 100 microliters. Well, I think some people kind of bias that. So say 75 microliters of soup.

1:29:24.1 DB: For epitope binning you can do it and we have a couple different strategies for doing that from soups especially on the analyte side. So when you're injecting the soups for to look at sandwiching interactions, it's important that those antibodies are relatively well expressing and that you're able to achieve analyte concentrations that will show rapid binding. So that can be a little bit of a limitation, but typically like a good hybridoma soup or definitely a transient soup will have plenty of antibody in it to give a very reliable epitope binning data and you don't need much. If you had 200 microliters of soup, that would probably be enough to run an epitope binning assay in both 'cause you can actually use the same soup for the immobilization as you use for the analyte injection. So it's quite efficient.

1:30:20.5 LT: Okay. Well, thank you both for answering all those questions and taking the time. Unfortunately, that's all the time we have for today. If you have any further questions please consider reaching out to the speakers directly. Their emails are shown on the screen. As a reminder, the webinar will be archived on the DDN website and you will receive an email notifying you when the webinar is available on demand.

1:30:45.6 LT: On behalf of DDN, I would like to thank our speakers, Kevin Harvey and Daniel Bedinger, as well as our sponsor Aviva Systems Biology. And of course, thank you to everyone who showed up today to listen. Have a great day. Goodbye.

1:31:00.0 DB: Thank you.

1:31:02.9 KH: Thank you Luisa.