Learn About SPR and Carterra’s LSA Platform

Posted by Noah T. Ditto

Rapid deployment of antibody discovery technologies capable of high-throughput and high-resolution screening is imperative for the timely development of therapeutics for global pandemics. Risk mitigation using multiple advanced technologies further increases the likelihood of identifying efficacious treatments in an aggressive timeline. In the first part of this webinar, Colby Souders, CSO of Abveris, will discuss the following:

  • An integrated in vivo approach incorporating two parallel workflows for accelerated discovery of lead antibody candidates targeting the SARS-CoV-2 spike protein
  • A single B-cell discovery workflow to directly interrogate antibodies secreted from plasma cells of human transgenic Alloy-GK mice and Abveris’s proprietary DiversimAb mouse for binding specificity and ACE2 receptor blocking activity
  • A concurrent accelerated hybridoma-based workflow utilizing the DiversimAb mouse model for increased diversity
  • Advanced high-throughput and high-content characterization of antibodies with the Carterra LSA in two assay formats (purified and crude samples) to rapidly determine binding affinity, blocking profiles, and epitopic coverage for lead candidate identification

The Carterra LSA is transforming the landscape of drug discovery and has now become an indispensable tool for cutting-edge research teams like Abveris. In the second part of the webinar, Noah T. Ditto, Technology Product Manager at Carterra, will provide an introduction to what makes the LSA critical for effective and modern drug discovery workflows:

  • Orders of magnitude more data than traditional platforms per experiment
  • Extremely low reagent requirements
  • Minimal hands-on time and unattended run times of up to a week
  • Characterization campaigns for hundreds of clones completed in just days

Gary, the director of marketing for Abveris

0:00:02.3 Gary: Alright, good morning, everyone. I’m Gary, the director of marketing for Abveris. It is my pleasure to be hosting this joint webinar brought to you by Abveris and Carterra. Today’s speakers are Colby Souders, the CSO of Abveris, and Noah Ditto, Technical Product Manager for Carterra. In just a minute, Noah and Colby will speak on the topics of accelerated antibody discovery using high-throughput and high-content characterization tools, including their Carterra LSA platform. Before we start, here’s a couple of very quick housekeeping items for the audience. The webinar is going to be split into three sessions, the first session will be a presentation by Colby on the recent Abveris discovery campaign targeting SARS-CoV-2 Spike Protein. In the second part, Noah will present the in-depth technical features and applications of the LSA platform. There will be a live Q&A at the end to address questions from the audience. Please feel free to type your questions in the Q&A box now and during the Q&A session, we’ll try to answer as many of them as possible given the scheduled time.

0:01:01.5 Gary: Abveris and Carterra will also follow-up for questions that we don’t have a chance to address today, we will make the webinar available on the website within 48 hours, typically. If you wanna rewatch it or wanna forward that to your colleagues for their reference, you are assured to have access to it very shortly. A little bit of introduction to the speakers. Colby is the CSO of Abveris, he provides and give leadership and develops innovative strategy for antibody discovery. He got his PhD in Cell and Molecular Biology from the Texas A&M University. Before Abveris, he was directing antibody discovery and engineering programs at MassBiologics and Kanyos Bio. Noah has been with Carterra since 2014. Before Carterra, Noah was with Bristol-Myers Squibb for almost a decade, supporting drug discovery and early clinical development work. Noah earned his Master’s of Science from the Penn State University where he developed chromatography-based methods to isolate biomarkers in dengue infection. He also holds an MBA from West Chester University. So without further ado, I would wanna pass the mic to Colby to start his presentation. Colby, whenever you’re ready.


Data that details our anti-COVID internal research campaign

0:02:17.7 Colby Souders: Thank you, Gary, and thanks to everyone for attending, and I’m happy to present the data that details our anti-COVID internal research campaign we concluded last year at Abveris. We took a very function-forward and rapid antibody discovery approach for this project, both are features that are key in the need for antibody discovery throughout the pandemic. It’s also an antibody discovery model we feel works very well across the broad range of target classes, not just to rapid antibody discovery response to infectious disease targets. And I’ll start by introducing the main factors that we feel are most important when designing an antibody discovery workflow, and these include antibody diversity, depth of screening and speed. When we refer to antibody diversity, it’s important not only to have large starting repertoire, but also to capture the full repertoire by eliciting a robust immune response and then to recover a diverse number of B cells in the downstream screening process. But at the same time, the screening resolution has to incorporate enough complexity to identify the most functional clones early in this discovery workflow. And both of these factors need to be completed in a rapid timeline, and this depends not only on the actual time that it takes, but also how many clones can be screened in this defined timeframe. In combination, this allows us to efficiently find the most relevant and rare functional hits.

0:03:39.5 CS: At Abveris, we combine advanced immunization techniques with our proprietary hyperimmune mouse platform to effectively generate a diverse starting repertoire, even to the most difficult target classes like high homology or self-targets, non-immunogenic epitopes or complex cell surface proteins. Then to mind the immune response, we developed specialized hybridoma and single B cell techniques paired with an optimized high-content characterization, and this will be the main focus of this talk to highlight how quickly and accurately we can triage a panel of hits to lead candidates. For the experimental design in this campaign, we leveraged both of a single B cell cloning and accelerated hybridoma workflow. Both include rapid immunization strategies followed by primary cell harvest from tissues of interest. For the single B cell workflow, screening is performed on the Berkeley Lights Beacon in just one day, followed by sequencing and recombinant expression. For hybridoma-based discovery, colonies are screened by ELISA and Octet BLI within just weeks after fusion to select top candidates for parallel sequencing and culture scale-up. Top candidates from both single B cell and hybridoma screening are then fully characterized by Carterra SPR for affinity, blocking activity and epitope mapping. Then combining the biophysical properties in silico sequence analysis, we can select fully validated lead candidates within just a couple of months after initiating immunizations.

0:05:07.4 CS: We also used two different mouse strains to take advantage of the benefits of each one. The Alloy GK mice provide a wild-type-like immune response, but with the advantages of producing fully human antibody variable regions, while the DiversimAb mice from Abveris produced murine antibody sequences that elicit an enhanced immune response to broaden epitopic diversity, and ensure valuable regions of the antigen are effectively targeted. With our immunization strategies, we’re specifically targeting S1 domain that contains the ACE2 receptor binding domain and both strains responded well with detectable serum counters even at a greater than 50,000 full dilution, giving us confidence that a robust anti-S protein B cell population was activated. In the single B cell workflow, plasma cells were loaded on to OptoSelect 14K chips using the Beacon, which uses opto-electric positioning to guide single cells into individual nanopens. Typically, we load more than 10,000 single cells per chip and can run multiple chips per day to have a total throughput of over 40,000 single B cells screened in a day.

0:06:14.5 CS: After cell loading, various capture and detection reagents are imported into the channels above the pen an antibody is secreted from the plasma cells diffuse out of these pens and into the channel forming blooms during the timelapse and imaging when they bind to the target reagents. And because the assay reagents are separate from the B cell, we can flush the reagents from the chip and import a second assay with all new reagents to repeat multiple sequential assays. Once B cells of interest are identified, we again use OEP to export the cell out of the well and into the channel, and then flush the channel into a well of a PCR plate for downstream sequencing of the heavy and light chain variable regions. For this particular study, we ran two sequential assays. In the first, we were looking to identify high affinity candidates in the low double-digit animal or better ranged, the monomeric S1 domain of the spike protein. By capturing secreted IgG on anti-Mouse polystyrene beads at the mouth of the pens, we detected total IgG with an anti-Mouse secondary antibody in the FITC channel and positive binding to the S1 protein that’s directly labeled with Alexa Fluor 647 in the Cy5 channel in a multiplex assay format.

Second mixture of assay reagents was imported

0:07:30.3 CS: Then after flushing the channel, a second mixture of assay reagents was imported, this time with the S1 protein pre-complex with a molar excess of ACE2 protein. So in this format, a lack of S1 binding single, indicated competition between the antibody and ACE2 for S1 binding, which identifies potential blocking candidates. Of course, it’s also possible to label the ACE2 protein and identify antibodies capable of finding the S1 protein and displacing ACE2, but that format wasn’t run in this example that’s shown here. On the hybridoma side, following a high-throughput and efficient electro-fusion process, thousands of colonies were screened by ELISA using various liquid handling platforms and clones showing reactivity to the S protein, but not to an irrelevant counter screening reagent, were pulled forward for secondary screening confirmation. Then following the ELISA screening, we performed a single point kinetic analysis against the trimeric S protein to determine avid affinity using the Octet. Recombinantly expressed antibodies from the single B cell workflow were already available for screening, so they follow the same characterization cascade. And out of the hundreds of binders that we discovered, the early triage criteria defined here identified dozens of top candidates to select for full validation, which included various SPR-based assays on the Carterra LSA.

0:08:56.9 CS: A representative subset of these clones is shown here, and full monomeric affinity to the S1 protein was determined in a six-point concentration titration series. Overall, the Alloy and Abveris mice have very similar affinities with clones derived from the single B cell workflow showing roughly a two-fold affinity advantage over those from a hybridoma workflow, and this is also in line with previous head-to-head studies we’ve performed where we select for average higher affinity on the Beacon. Next, a blocking assay was performed to identify clones with potential neutralizing activity. In this sequential format, an experimental antibody was pre-bound to the S1 protein and then exposed to the ACE2 protein. The amount of ACE2 binding was quantified and reported as a percent increase in binding. The candidate antibodies capable of blocking S1, ACE2 interaction showed negligible ACE2 binding like Clone 11955, shown in yellow, but in contrast, clones like 12637 S1 in green capture the S1 protein at a non-blocking epitope and allow the S1, ACE2 interaction to occur. Now, I’ll pause here and take a minute to talk about the important comparison that we made with respect to the SPR formats as a part of this campaign.

0:10:18.5 CS: For all the Carterra studies that we’ve performed, two different formats were used in head-to-head to validate the data from each one. The most simplistic version purified experimental antibodies were directly coupled to the chip surface prior to introducing analytes. In this case, I’m showing the S1 protein binding to one immobilized experimental antibody before introducing a second candidate antibody in a classical epitope binning experiment. While this format is straight forward to set up, it does require purified antibodies and a defined buffer, which takes additional time and effort. Recently, a second assay format’s been validated on the LSA, where antibodies in crude supernatant can be used, and this eliminates the time and cost associated with purification and buffer exchange. And this setup is slightly more complex because the captured antibody is immobilized on the chip surface to then bind the first experimental antibody, which then needs to be cross-linked followed by an additional blocking step that’s not actually shown in this diagram prior to then introducing the analyte and second experimental antibody. In this study we found the results from both assay formats were very similar, though, and an extra one to two weeks could be saved on the overall project timeline by using the crude sample format.

Analysis of the results from our epitope binning data

0:11:35.9 CS: Analysis of the results from our epitope binning data in this format reveals a conserved epitope of the ACE2 blocking region with some blocking candidates binding on the periphery of this core. All 12 are non-blocking candidates localized to different epitopes on the S1 protein, some adjacent and some distal to the blocking epitopes. An overlaying orthogonal data like avid monovalent affinity suggests there’s no clear correlation between affinity and epitope. Looking at another way to visualize the epitope binning data in the form of a community plot, we can clearly see six different distinct epitopes targeted by this subset of top candidates. And again, by overlaying orthogonal data onto the binning data, this time in the form of blocking clones in yellow and non-blocking clones in green, we can quickly establish which ones localized the receptor binding domain to select high-value lead candidates. Now, to better illustrate how the crude versus purified samples compared in the SPR analysis, I’ll highlight the binding profiles in more detail. So in this case, an example, Clone 16F2 has a nearly identical antigen-binding profile when either purified antibody or supernatant is used. And when taking the rate constants into account for all clones, we see a nice correlation between the measured value for supernatant samples on the Y-axis and purified material on the X-axis.

0:13:02.1 CS: Similar ranking and comparable values is evident for both the on and off rates and subsequently the final calculate of affinity, which validates the use of either workflow to acquire accurate results and perform under these optimized conditions. For us the option to use ZEUS means we can characterize the data earlier in the workflow and we have more flexibility in the sample requirements, because we can be largely agnostic to attributes like buffer composition and antibody concentrations. Finally, combining the in-depth biophysical characterization of detailed sequence analysis allows us to select the few lead candidates from this larger pool. And when taking sequence similarity into account to define clonotypes, three unique groups of blocking clones emerged. And for a group with multiple clones like the one with four similar candidates all binding to the core blocking epitope, we can incorporate in silico analysis of common sequence liabilities to further downselect. And in this case, clones 21C3 or 11955 have very few predicted liabilities making them excellent candidates for downstream development without the need for extensive engineering, which means that the time to clinic would be reduced for candidates that are like this.

0:14:17.5 CS: Interestingly, these two candidates do have the same sequence, but were each independently discovered with the hybridoma workflow for 21C3 and a single B cell workflow for 11955, which really validates both platforms for discovering high-value lead candidates. And to summarize the timeline for antibody discovery, the single B cell workflow can deliver sequences in as little as one month with validated lead selection of recombinantly expressed antibodies taking just one or two months later. On the hybridoma side, the timelines are roughly double with a total of about two to three months to sequence and three to four months from immunization start to recombinant expression and validation. But this can still be considered rapid when taking into account the large amount of data that’s collected in just a few months. And now I’d like to thank everyone who contributed to the work at Abveris and Carterra as well as Alloy Therapeutics who provided the mice to help accelerate global COVID therapeutic solutions. And now I’ll turn this presentation over to Noah who will speak in more detail about the capabilities of the Carterra LSA platform and all of its different applications.

0:15:33.4 Noah Ditto: Thank you, Colby, and thank you for everyone who’s online attending. My name is Noah Ditto, I’m Technical Product Manager at Carterra, and today I’ll be talking about how the Carterra technology can be used to raise the bar for biologic discovery and characterization.


Carterra has really redefined the real-time label-free screening area in drug discovery

0:15:55.6 ND: Carterra has really redefined the real-time label-free screening area in drug discovery and development. If we look at Carterra’s technology, what it affords is a hundred times more data in 10% of the time using only 1% of the sample versus traditional platforms. So this is a significant leap in terms of what customers can do with a real-time label-free device in their discovery workflows. Carterra’s technology has been adopted across a wide range of companies, listed here below is just an example. It really spans antibody developers, biotech, CROs, academics, as well as government research labs. The LSA is Carterra’s flagship technology. What the LSA brings to the label-free real-time binding market is novel microfluidics that take SPR to a new level. So in the LSA we have a multi-channel device, which is 96 channels that can dock on the sensor chip surface, and with that sensor chip fixed, the fluidics can switch over to single-channel mode where a single fluidic device docks on the surface and a single injection passes across all 384 locations simultaneously. So we can build arrays in the LSA of up to 384 proteins most commonly, but really any biomolecule can be attached to the surface. And then we’re testing in single-channel mode, one single injection across this array to really maximize the throughput on the system, so this really transforms what throughput looks like in terms of SPR.

What controls the LSA is our Navigator software?

0:17:34.9 ND: What controls the LSA is our Navigator software. The Navigator software platform is highly intuitive. Here’s a screenshot just showing you graphically how it lays out very simple descriptions of where sample locations are going, pictorials of how the binding sensorgram profiles are gonna look like and the time that the user can specify for those. Even the most complex experiments, say a 384 x 384 binning experiment which would last the better part of a week can all be set up in about five minutes. And the arrangement of the samples and the plates is fairly agnostic in terms of the system’s needs, so really the sample preparation steps are very simplified as well compared to other platforms that sometimes have quite elaborate and specific needs for where samples are placed. There’s two main applications on the Carterra LSA that are most sought after when customers need to do real-time label-free screening in a high throughput manner. The first of those is epitope characterization.

0:18:38.0 ND: So really, epitope characterization can span a number of different activities being epitope binning, where we’re competing antibodies to determine whether or not they share the same epitope on a given antigen, where we can compete antibody binding against a peptide array, and this is typically referred to as epitope or peptide mapping, so we’re looking for binding sites to specific peptide residues and regions. And the last is mutant mapping where we’re taking an array of antibodies and presenting injections of mutant forms of the antigen across that as another way to map more locally where the antibody epitope is located. Today, I’m gonna focus on what is our core epitope characterization workflow that virtually every customer uses on the system, because it is so high throughput compared to other platforms, and that is epitope binning. So just to step back and kind of look at the big picture, really what the LSA affords is faster and earlier epitope binning. This is just an example manuscript looking at over 100,000 interactions, 327 antibodies and only using five micrograms of each antibody.

0:19:52.5 ND: So this is the type of workflow that can be now placed very early on in the discovery process, where typically this would not be allowed with other platforms. So really transformative in the space of epitope binning. Our customers get a lot more detail about their clones, they’re not trying to take low information, low-resolution assays and use those to call clones out, they can carry those clones further with a high confidence of knowing where they bind and their interactions with the other candidates in the overall pool. Here is an example of how antibody epitope binning can be conducted. So, it all starts off with using that 96 Channel printing device that I mentioned earlier to array up to 384 antibodies onto our biosensing chip surface. We then switch to our single-channel mode and inject antibodies one at a time in a serial fashion to determine whether or not they can form a trimolecular complex in the presence of antigen.

0:20:50.2 ND: So in this case, this is a classical binning format I’m showing you here. We have a binding of an antigen and then we have an injection of two different antibodies. One of the antibody sandwiches, we have a clearly positive binding response, and the other fails to sandwich, which we have a no response, basically. And from this, we can gather whether or not the epitopes overlap or not for these respective antibodies. In the side of the software, we have proprietary technology that allows us to cluster the antibodies together in these epitope networks. So we show antibodies having relatedness by having these colored clustering or groupings, lines between antibody nodes represent blocking relationships, and then the lack of lines between them represents the ability to sandwich or no evidence of competition in this assay format. So these are… This is a very sophisticated and sort of groundbreaking way to visualize epitope relationships among antibody candidates.

0:21:49.7 ND: And then what our software also offers is the ability to bring in additional data in the context of epitope into the application. So the application itself can have additional data pasted in, and then it’s simple enough to display it in the context of all these networks. It was a very intuitive workflow, very graphical and really powerful in representing the antibody relationships and really what epitope means for the candidate overall. And just diving down a little bit more into those epitope networks, they’re very fascinating in that you can start at a sort of granular bin level, epitope network level, which is where we take the antibodies and we simply compete them. And in a very objective manner, we just say, “Are you identical or not?” In this case, we can see these shared groupings of antibodies. For example, this red group in the upper right here, these would all be identical in terms of their epitope, because they’re all competitive in the assay, so we’ve considered them identical for our purposes.

0:22:49.1 ND: And using hierarchical clustering, we can then say from these bin-level epitope networks, we can take this to a community level epitope network and say that although there might be a few small differences between some clones, we’re generally gonna say that these antibodies are probably very similar in their epitopic footprint. So we can see on the far left here, this initial group in the upper center being similar, but not necessarily… |They’re all clustering together even though there is some better relationships. Once we do some clustering analysis, we can see that by and large, they are virtually identical, and we can therefore cluster them. And this is a way to move out of the detail and the noise of the assay to a level that sometimes helps with decision making and driving distinct bins that can be used for downstream candidate selection.

What does epitope offer besides just simply clustering antibodies?

0:23:41.4 ND: So, I mentioned this a little bit previously, what does epitope offer besides just simply clustering antibodies? Well, there’s this ability then to bring in orthogonal data into the assay. And so, this is really powerful, a lot of customers love using this because any attribute of the antibody that you have on hand, affinity, which you maybe derived already, cause you have done a kinetics experiment, blocking neutralization, what type of model species antigens are reacted against by the antibodies, library sources, expression levels, even sequence nuances can all be taken into the software and displayed on top of the epitope clustering. But it’s really, really powerful to understand collectively what attributes maybe drive selection for certain epitopes, or possibly even within a group of very similar behaving antibodies, find distinguishing factors that allow you to pick a single representative clone out of that cluster overall.

0:24:41.9 ND: Now, moving on to our second major application on the LSA platform is kinetic characterization. So this would be assays looking primarily at binding kinetics, so we’re looking at the on-and-off rates of the antibodies against antigens. In some cases where necessary there may be a choice to do steady-state affinities where we’re not looking at the on rates themselves, but rather looking at as a function of response versus concentration. But by and large, kinetics via measuring on and off rate is the most typical workflow done on the system. So truly high-throughput kinetics, this is kind of a screenshot of a 384 unique epitope measurement, or excuse me, kinetic measurements. You can see here that we’re getting a wide range of affinities all measured in a single experiment. This entire experiment probably only takes about six hours, so we’re measuring 384 affinities, one run, seven micrograms of antigen consumed, and we get an eight-point titration. So really, really detailed characterization of clones.

0:25:47.1 ND: You can imagine that it’s easy enough to multiply this and get up to, close to a 1000 clones in a single experiment, because on the LSA, we have the 334 well plates, so there is the ability to take screening to the primary stage and really make sure that we get the most information per clone before starting to whittle down that list of candidates. So here’s an example using single-chain variable fragments raised against CARs to show how kinetics works. So we have 384 clones arrayed on the sensor chip. These are most commonly captured so that we get preferential orientation of the FABs towards the solution phase, and the subsequent antigen that comes in binding. Then we injected a titration series of antigen or in this case, tumor receptor across the array to see the binding on and off rates as a function of concentration. From there, we then take kinetic rate constants using a fitting model. So we have our kon or koff, and then our calculated affinities.

0:26:58.4 ND: And what’s really great about these assays is in addition to simply measuring and getting kinetic rate constants and calculated affinities, we can actually include additional receptors, in this case, SCFS for CAR purposes and look for specificity as well. So the same array that you’re using to measure affinity of your target antigen, you can also use by re-arraying it and testing over and over again to check that you have the specificity you need, which is particularly important in the application of CARs. And in the kinetic software, really everything again, is very much graphical and allows the user to explore the data, much like how epitope has it in the work plots. This is just an example from the kinetic software of an ISO-affinity plot that plots on rate versus off rate. And then we have, on our diagonal here, we have the affinities that are calculated from this and sort of distinguishing bars. And as you hover over in the software, you can see individual sensorgrams for each of these.

Easy way to visualize how your affinity distribution looks

0:28:00.5 ND: So again, a really easy way to visualize how your affinity distribution looks, and with the ability to label subsets of the data and the software maybe particular clones from a certain source or any other attribute about those clones, you can actually sort of see affinity in the context of other pieces of data as well. So again, really powerful, easy to visualize and all kind of baked into the software, is ready to use rather than having to move on to different applications to perform these analyses. So big questions that comes up is, “You can screen really fast on the LSA, that’s great. And it can be done very early and we’re getting real-time binding data, but is it accurate?” And the short answer to that is yes. There was an excellent paper published just last year by groups at Adam Hub and Engine comparing the landscape of label-free biosensors. And among other things they determined in that exercise was that on the LSA, we get excellent agreement with gold standard platforms in terms of our measure rate constants and calculated affinities.

0:29:04.3 ND: And then the great part about that is then, we’re not trading throughput for accuracy, which is usually a typical trade-off in a lot of technologies. So really, what that means in the big picture is that when measuring affinities very early on in the discovery process, you can have a high level of confidence that those values determined then will hold through throughout the transition to development, oftentimes where some of the early assays, historically, need to be repeated because we’re changing platforms and there’s discrepancies between low resolution screening and then the higher resolution screening towards the back end of discoveryas development kicks off. So again, lots of confidence in your measured affinities.

0:29:46.7 ND: So really, what does this mean for the researcher doing a study like this, where they’re looking both at affinity and epitope? It really drives decision-making. This is a manuscript from Lily published this past year looking at COVID therapeutics. And they effectively utilized the LSA to its full capabilities in this particular exercise, looking at affinity, understanding not only where, how tightly it binds, but destruction of ACE2 binding, which is particularly relevant for COVID therapeutics. And then also, looking at epitope, looking at the breadth of the unique epitopes that are getting elicited during this natural immune response to these cell-derived antibodies, then also looking at whether or not the neutralizing antibodies had relationships to the epitope. And really making… Painting the overall picture of what these antibody candidates look like in terms of where they bind and how tightly they bind. So this is wrapping it all together, and really the icing on the cake for this particular example is that all of these studies were done in under a month in total, including the LSA data, which was collected, everything you see here, in about a six-day span, which is incredible.

0:31:04.2 ND: I mean, when we’re talking traditionally that these activities take months to years, we’re condensing everything down into about one working week to collect a data set like this. And we have a huge breadth of information about the candidates’ excellent rationale and data-driven approach to select the candidates and all done in an extremely rapid and really record setting manner. So, one last thing I’d like to touch on is kind of an exciting new area that the LSA technology has been utilized for, again, in this fight against COVID. And this is looking at the ability of immunoglobulin immune response to bind RBD in the 96-parallel manner from patient’s serum. It’s a really exciting work here. There was a group… Academic group out in the Netherlands that partnered with us and was able to generate a really cool data set showing that we could track the strength of binding, so that’s the ability of the antibodies to bind and the rate of… Their off rate against the RBD binding.

0:32:16.2 ND: And then coming in after that, looking at whether or not there are certain isotype profiles that are unique among the infected individuals, so we’re looking at IgM, IgG, and IgA binding in addition to the strength and binding. It’s a really powerful data set, and instead of just saying whether or not there’s presence or absence of an antibody response, we’re getting that, plus the overall off rate effect of the antibodies as time progresses and as you would expect, the immune response to mature from an affinity perspective. Alright, with that, I’d like to, again, thank all the attendees for taking the time out of their day to listen to our presentation. Really, anything about Carterra’s technology, LSA or otherwise, software, hardware, what have you, applications, feel free to reach out to us. You can use the chat to send questions during this webinar or else contact me directly at ngido@carterrabio.com and I’d be happy to discuss further with you. Thanks again.


0:33:32.6 Gary: Alright. Thank you, Noah. It’s time for us to address questions from the audience now, and we have a lot of interesting questions coming in. The first one is for Colby. “You mentioned a lot of opportunities for improvement of the timeline, would you leverage what you learned in this campaign to respond, another viral outbreak in the fastest timeline possible?”

0:33:51.6 CS: Yeah, that’s a great question and a very important one too. So, a few things that I mentioned within the webinar would be how to leverage the Carterra using crude samples. So that’s something that you could do to save several weeks on a campaign, but still be able to capture the full value of that validation process to pick high value lead candidates, not just candidates, but it does save several weeks on that timeline. Another opportunity would be, if you did want to proceed with mouse immunizations, you could optimize that timeline, and again, use a rapid immunization approach to still elicit a diverse immune response and get high quality leads out of that. You’d also be able to use human B cells if the sample is available. So, definitely several things, but the Beacon really does help accelerate that too by compressing all of the screening into a single day to select candidates.

What are the pros and cons of both strategies in your opinion?

0:34:51.9 Gary: Alright. Thank you, Colby. The second question is also for you. “Do you believe that a mouse immunization strategy is better than a direct from human single B cells screening approach for viral outbreaks? What are the pros and cons of both strategies in your opinion?”

0:35:07.3 CS: Yeah, another good question. And again, kind of in response to the one that I answered for the previous one, both are absolutely valid approaches. So, the direct from human approach absolutely gives you fully human antibodies upfront. Often there are some difficulty in sourcing those samples though, so that would be pending availability of those samples. You’re also limited to what response that particular individual has, and usually you don’t have a large number of samples or the time to screen through a large number of different human samples that are available, and so you’re hoping that the one individual or the handful of individuals that you’re screening has the best response that you could get to find a lead candidate. When immunizing mice, you have that flexibility to elicit different immune responses based off of your immunization approach and your immunogen that you’re using, and then different screening strategies as well.

0:36:08.8 CS: So you have a little bit more flexibility in that, obviously with the disadvantage of a slightly longer timeline by a few weeks. And also in using the immunized mice, you’re also able to get a fully human antibody upfront, which will accelerate that downstream therapeutic discovery timeline, but murine antibody discovery is still a valid approach and again, can sometimes give you higher quality candidates and more diversity to find a better lead candidate as long you are able to afford the time to then do the humanization, the engineering that would be required to turn that into a therapeutic downstream.

0:36:50.3 Gary: It makes sense to me. The next one is for Noah. “How would you screen for Gamma receptor binding using the LSA platform?”

0:36:58.7 ND: Yeah, good question. I kind of highlighted in our discussion, epitope characterization and antigen kinetics, which obviously work through the FAB end of the molecule. But you could consider then neonatal receptor, FCRN or the EMO receptor studies, kind of in the same vein as the kinetics’ approach, a captured kinetics’ approach, and in that way, we would capture the antibodies via the FAB region in some manner. There’s a couple of different ways to approach it and then flow the receptors in solution across that array. So you can get quite a high throughput of screening, up to 384 antibodies against a single receptor at a time, so you get really good parallelized measurements, but it operates very similarly to the kinetic approach that I described in that presentation.

0:37:50.0 Gary: That sounds good to me. Here’s a follow-up question regarding Noah’s comment. “So the LSA can screen 384 samples, but is it critical to use the LSA if you don’t have that throughput requirement in some cases?”

0:38:05.3 ND: Yeah, yeah, definitely. There’s a diverse range, just of our customer base, some absolutely max out the capacity of the instrument and many don’t approach 384 unique species at a time that they have to screen, but one of the great things you can do, though, and this is kind of new to the SPR space, is you build in replicates in your assay or scout optimal conditions as you run the assay. So particularly for kinetics, for example, the amount of ligand you put down on the surface, the antibody that’s critical, you get that density right in order to get the confidence you need for the kinetic measurements you take.

0:38:38.6 ND: So in that example, you might have a subset of antibodies, but you could put them down, capture them, say, through different densities and run the experiment, then just use the signals that are most applicable to your kinetic measurement. Likewise, you have the ability to build in replicates, so you’re getting statistical confidence in the measures you’re making in a single experiment instead of having to run the experiment over and over again in order to build that full confidence. So yeah, the short answer is definitely, there’s a lot of power to be had, even if you don’t have 384 or more clones that are unique to screen at a time.

0:39:13.7 Gary: Yeah, thanks. The next question is for Colby. “So in your presentation, you used a recombinant protein assay, but is it possible to run an infectivity assay using a pseudovirus on a Beacon platform?”

0:39:26.9 CS: Yeah, absolutely. That one is something that has been developed. Really, the short answer is that it was availability of reagents. At the time that we were screening this, the reagents were not available to us, but that is something that you can develop. It does take a little bit more time, because you would like to run that and QC the assay and validate it. So if you are working as quickly as possible in a situation like this for a viral outbreak, then an assay, like a protein-blocking assay, is faster to validate and get up and running on the Beacon. And so you would be able to find high quality candidates, as we showed in this webinar, still using a recombinant protein-blocking assay and then validating those downstream off Beacon and pseudovirus assay, but definitely if the reagents are available and the extra few weeks for time to validate the assay can be afforded, then a pseudovirus assay is possible on the Beacon, as well as a number of other complex screening assays. That’s one of the big advantages, is that there is a lot of flexibility in the design and what you can run and the breadth of data that you can achieve, so definitely possible.

0:40:42.8 Gary: Yeah, thank you, Colby. The next one is actually for you as well. “So in this campaign, how large was the overall number of cells you screened and total number of cases identified? Did you show the full dataset or just a subset of what you have done?”

0:40:58.1 CS: Yeah. So in comparison to other campaigns we’ve run, this one isn’t quite as large, so we screened anywhere between 10,000 to 20000 total B cells. This was before we got a second Beacon, and so our throughput has doubled now, and this is actually… It was done in a single run, again, to kind of recapitulate the idea of being able to move as fast as possible with a single run and see what you could get on a per run basis. So definitely could be augmented in future outbreak situations or in standard campaigns. And this is not the full dataset that… We did actually find hundreds of total candidates, we’re just showing some of the best that performed downstream here in these assays. So the dozens that we show here is a subset of hundreds of total candidates.

0:41:51.0 Gary: Alright. So before we jump to Noah, I would wanna ask Colby one of these questions from the audience. “So you mentioned that you used multiple strains of mice to help with diversity. How did the transcript diversity play out across the different strains you used in your study?”

0:42:05.2 CS: Yeah, good question. Some of that is captured in the data. But in general, again, using the diverse amount of mice, we are able to see binding across multiple different epitopes, we do see broader diversity in the sequence that is recovered, so clonotype usage is broad. We still see a good usage within the humanized mice as well, but again, as the strain suggests, diversion map provides a larger diversity. But again, going back to, I think, one of the earlier questions that was asked, and different immunization strains and one… Sorry, immunization schemes. And one of the advantages of using mice direct from human B cells is the ability to use these different strains, elicit a different response and get broader diversity overall by using multiple different strains that did allow us to find the most critical epitopes, whereas with human B cells, you’re kind of limited in that response, and so we were able to find very good candidates by leveraging these different mice and finding hits to multiple different epitopes, which by the way, also can help if you’re making [0:43:23.4] ___ immunodominant cocktails. So for those to be really effective, you’d like to find antibodies, to find multiple different epitopes that all perhaps could be neutralizing epitopes as well. And so it’s important to get this diversity and it’s much easier to capture larger diversity when you use multiple mouse strains.

Reason to perform full kinetics versus single point when doing primary screening

0:43:47.3 Gary: Yeah, sounds good. So we have a question regarding the LSA platform for Noah. Is there any reason to perform full kinetics versus single point when doing primary screening?

0:43:58.6 ND: Oh yes. So I think that the question is really… Historically, if somebody was trying to gain throughput on a label-free device, they often do sometimes what is just a single point screen, they just measure a single concentration and try to estimate primarily from the off rate, a rank order that they can extrapolate for affinity, and that’s really just had been done historically, again, by the limitations of throughput of technologies. So if we think about the LSA and how fast it is and how… One of the examples I showed in my slide was having a full eight-point titration series for 384 ligands in only a six-hour run. You really don’t have to go through those exercises of trying to shortcut some of the assay processes. So really on the LSA, I would say that the most appropriate manner is to run a full kinetic titration, kind of that’s what Colby is showcasing, and in their workflow, there’s no point in trying to screen and get less data, it really doesn’t save you anything in the end. And the one point I made in my presentation about the high accuracy of the data is that once you’ve got that number, it’s good, it’s solid.

0:45:05.0 ND: It’s not like you need to go back and revisit it again and rerun that measurement, so you’ve got the number, you’ve got it screened early, and you’ve got as much candidate information as you can gather at that point to really make sure that decisions downstream are well-informed.

0:45:20.2 Gary: Alright, sounds good to me. The next question is for Noah as well. “When doing serum studies like cited with Colby’s serum work in your talk, are there any special requirements in assay design?

0:45:34.8 ND: Yeah, so generally, serum is not a problem. We do recommend 0.2 micron filtration just to make sure there’s no particles or anything in the sample type, but we typically do a dilution, I believe, in those particular studies. In that manuscript, it was around a 1 to 100 full dilution, so we’re fairly dilute. About the only thing you may see is a little bit of non-specific binding sometimes for serum. That’s usually just resolved by having PSA as part of the running buffer to mitigate any non-specific binding and serum components to the surface. But otherwise, generally, serum, given the level of dilution it has, and just some best practices and monitoring for non-specific binding really acts as any other analyte would down the system.

0:46:17.7 Gary: Alright, thank you, Noah. Since we’re running out of time, so I will ask one last question, and we are gonna wrap up on today’s webinar. This is for Noah. “Is it possible to combine binning and kinetics measurement into one single run?”

0:46:30.5 ND: Yes, it is possible to do that. There’s a little bit of balance of what you’re trying to achieve. So typically for kinetics, most commonly customers would capture the antibodies. The example Colby gave was capturing directly out of crude material, enriching them and doing your kinetics that way as opposed to having to then do a separate purification step. That’s usually pretty good, but depending on what you’re doing with the binning side of the equation, you might need different ligand densities for the two of them.

0:46:58.7 ND: So for kinetics, we obviously want a very low ligand density, a very low amount of antibody on the surface to get the highest quality kinetics that we could get, whereas in binning, we’re running a lot of repetitive cycles where we’re generating the surface and removing the antigen that’s found to test the next trimolecular complex formation in the next cycle, so we do have to increase the ligand capacity a bit.

0:47:23.6 ND: So you have to sort of balance the two of those. Sometimes customers may want to set them up separately and run them, but you can combine them if you’re willing to maybe make those trade-offs saying, I’m willing to get, maybe a little bit of non-ideal ligand density for kinetics in order to make sure that I can combine the binning maybe for your timing purposes, that’s a necessary or a reasonable trade-off. So that’s probably the only thing to consider, is the ligand density components of each assay.

0:47:53.4 Gary: Alright. So thank you Noah, thank you, Colby for answering the questions. And our time’s up, so we have to say goodbye, but please bear in mind that we’re gonna reach out to questions that we don’t have time to answer today. And as a reminder as well we’re gonna make this webinar available on the website shortly in 48 hours usually. So please, please visit our website to get a link to it if you wanna rewatch it or share with your colleagues afterward. And that’s it. Thanks for joining us today and bye bye. Look forward to seeing you in, in other webinar.