Posted by Noah T. Ditto

Antibody Discovery, Selection & Screening


In biotherapeutics candidate discovery and development, innovation during the COVID-19 pandemic has been led by a diverse set of strategies used to identify candidates. Within this highly competitive space traditional hybridoma and phage approaches now both contend with and are complemented by NGS and synthetic strategies to meet candidate timelines measured in weeks rather than years. Despite substantial market pressures driving more sophisticated and differentiated approaches, there are stages across these varied workflows where turnkey solutions can be implemented to eliminate bottlenecks. In this talk the role of HT-SPR and its successful enablement of COVID-19 therapeutics will be explored as a technology in biotherapeutics discovery that is deployable in any workflow. Using case studies this talk will highlight touchpoints common across otherwise distinct workflows where leveraging HT-SPR can highlight critical attributes for biotherapeutic candidate selection.

Welcome to Antibody Discovery


0:00:00.1 Michael Keenan: Hello, everyone. Welcome to Antibody Discovery, Selection and Screening Digital Week, brought to you by the organizers of the Global Antibody Engineering & Therapeutics Events and Content series. My name is Michael Keenan. I'll be your host for today's session, the transcendence of HT-SPR technology across discovery platforms and its impact on time-to-clinic. First, I'll cover some quick housekeeping announcements. If you experience difficulties with the audio or advancing slides, refresh your screen with the F5 button. If you're experiencing other issues, hit the question mark button to receive assistance. At any time during the presentation, you can submit your questions into the Q and A window on the left-hand side of your screen. In 24 hours, you'll receive a link to watch the recording of this session. You can also download a few featured articles in the resource list box to the right side of your screen. Let's now begin by introducing our speaker for the session, Noah Ditto, Technical Product Manager, Carterra. Thank you for joining us today, Noah. Now, I'll hand it over to you to begin the presentation.


0:01:16.9 Noah Ditto: Great, thank you, Michael, and thank you everyone who's online for attending today. As the title of my slide indicates, I'll be talking about high throughput surface plasmon resonance technology. It really had such a broad impact across discovery platforms, particularly antibody/biotherapeutic platform, in really accelerating time-to-clinic. We've had such a unique past 18 to 20 months in terms of how the drug discovery landscape has changed, and big part of this, an enabler, as you'll see in my talk, has been Carterra's HT-SPR technology at the forefront of those efforts to overcome the pandemic.


 Brief Introduction to Biotherapeutic Discovery


0:02:00.3 ND: So in my talk today, I'll give a brief introduction to biotherapeutic discovery, just outlining it and basically setting up the stage for the follow-up points in my talk, moving on then to Carterra's HT-SPR technology in relation and characterization of biotherapeutics, and lastly, finishing off with how HT-SPR can be applied to biotherapeutic discovery and have a meaningful impact in that exercise.


0:02:26.2 ND: The ongoing evolution of biotherapeutic discovery... When we kind of step back and look at all the different ways there are to develop and discover biotherapeutic, there's a number of sources out there, and then really interestingly, none of these have sort of fallen by the wayside as completely obsolete. It really has been an interesting development over the past decade or so, where next-gen sequencing has really changed the landscape and enabled a couple approaches that maybe have seemed like they've been around a long time, are gonna get left by more recent techniques and brought them back to the forefront.


0:03:01.9 ND: So I'm listing out here a number of approaches that would be used to generate biotherapeutic candidates, and really all of them are in play at this point, and you'll see from my talk that we have a number of different research groups using all these different approaches, really exciting stuff, and again, next-gen sequencing really enables a lot of this and keeps it all very much current and valuable in those efforts.


0:03:27.8 ND: And one thing I'd like to start off in discussing is the funnel concept that we think of so much in drug discovery. Traditionally, the concept was drug discovery was an exercise in whittling down a large list of candidates as rapidly as possible in order to meet timelines and also capabilities and downstream processes. I would say strongly that this has become an inaccurate portrayal of modern biotherapeutic discovery, primarily because we've seen a number of instances of really advanced discovery teams really using the collective knowledge that comes from these drug discovery efforts and feeding it back in a much more iterative and quick process.


0:04:06.6 ND: So frankly, they're not necessarily going through an exercise in simply trying to cut down the candidate pool, but rather really learning from that candidate pool and using that information to iterate and further improve the candidates going forward. And these workflows aside from just to try to do what bench approaches actually lean heavily on informatics, next-gen sequencing like I just mentioned, and even machine learning and artificial intelligence are really becoming mainstream for a lot of teams in order to maximize the knowledge and not simply cut down the numbers per se. And then really again, this last bullet highlights... The point here is that the discovery then becomes to adapt and iterate rather than trying to simply find the one candidate right away. There's more of an opportunity to learn and improve the internal processes, so the next round of candidates or even within that current round can improve.


 Drug Discovery Process


0:05:07.3 ND: And this is just conceptually what I just described in the previous slide. So on the left here is maybe a grossly over-simplified drug discovery process, but for the sake of sharing the information here, we have our discovery, development, preclinical and ultimately clinical stages in the process. So if we looked at this more traditional funnel approach where we just focused on whittling down candidates and getting to clinic, historically, there's been numerous details in the literature about late stage failures where we're trying to feed back information into discovery from failures happening in the clinic where there's high cost and other issues and lost opportunities there if we wait that long in the process to use information in a meaningful way to inform early discovery, whereas on the right, if we kind of take this more sampling style approach, more of the information indicated by these arrows is flowing back earlier in the process to inform discovery and development sooner in the process rather than learning at the clinic whether or not the candidate was suitable and had the desired developability characteristics. So really more information sooner in this process is what's valuable, and that's what really, I think, we're seeing a transformation taking place now in drug discovery.


0:06:27.4 ND: And then kind of in the backdrop of all these exciting changes that are going on in bio-therapeutic discovery is the COVID-19, elephant in the room. Its, on the one side of it, sort of the silver lining is it has driven substantial innovation in the pursuit of medicines to end the pandemic, that we've got these great opportunities which I'll showcase where we've done things that had seemed impossible. Not that are in the past. And really, yes, new opportunities to develop vaccines and in particular therapeutic, using approaches that really are novel and really kind of change things going forward. We're not... The bar has been changed and set higher, and we're expecting that to become the norm rather than the exception just in this pandemic phase that we're in.


0:07:16.1 ND: So I really like this quote from Elon Musk. Obviously, all these commercial space flight information is in the news and very well-known to everyone these days. But really, I think this also captures kind of where we're at with some of these opportunities, I'll call them, in drug discovery of... There are some pretty big challenges out there that we need to get drugs out sooner, particularly in the pandemic, there's lives on the line. It's critical that we identify therapies faster and get them out to patients sooner. So really, this kind of showcases it in this quote, that if something is important enough, you should try even if the probable outcome is failure, and I think that summarizes what we've... If you ask somebody prior to the pandemic, could we develop a bio-therapeutic in a matter of months and get it to the commercially viable stage rather than in years, which traditionally had been the case, most people would have said, "That's probably not likely to work. It's gonna fail." And I think it's been proven now that that isn't necessarily true. Certainly there is a risk of failure, but the severity of not trying is much higher in this case. So really a great way to shift our perspective to the opportunities and the ability for bio-therapeutic discovery to really impact human health in a much faster way than it has historically.


Kind of Layered into This Whole Discovery Process


0:08:37.8 ND: And kind of layered into this whole discovery process, I've indicated a really awesome early-end discovery platforms, libraries and all sorts of other means of developing, discovering antibodies is characterization. So understanding what are the molecules coming out of these different processes, how are they differing from each other. Obviously there's a mechanism action component, but there's also developability questions and other attributes of the molecules. And again, getting this information well-understood and understanding molecules very much to the sequence level and how they behave as a molecule interacting with, say, their target, it's critical towards continuous improvement, like I've highlighted, and obviously if longer discovery and development go on, any failure that occurs comes at a higher cost, both in outright costs for that program, but also opportunity costs from other programs that were put aside in favor of that one that ultimately didn't pay off. And then there's external pressures that are constantly must be accounted for. So we've got first-in-class, best-in-class considerations of any molecule making it to market. Then obviously, intellectual property, certainly knowing exactly the intellectual space that you're dealing with at a sooner stage just avoids any complications down the road and just sort of shores up the value in those molecules that are being identified.


0:10:02.4 ND: Characterization of biotherapeutic attributes. If we look at these, there's some just generalized big picture questions that were asked about any molecule no matter what it's designed for, or what the plans are for it in the development process, but really who will it bind. So what do we consider to be the target and most importantly, what are potential off-targets, homologous proteins potentially that may be hit by this molecule, usually with antibodies we see with specificity, but there's always concern that there's something that's similar but not necessarily identical that it may or may not bind. It may be in the case of mutations, maybe this is of value. We know in the COVID space, certainly the receptor bonding domain protein with opportunity to mutate has presented some sort of challenges and questions about, "Does the antibody retain strong binding to that particular protein even if there is a slight change?" Then obviously model species are critical for drug discovery. So we've got to have a clear understanding of whether it binds model species antigens. If there's a good reactivity to Sino and mouse and rat, for example, then that makes that whole discovery and development process that much easier as we move to in-vivo assays.


0:11:23.0 ND: And then the question becomes where will it bind. So we know it's binding to the target, but where does it actually bind? And this gets into the intellectual property question, and also some other questions about mechanism of action. So where is the epitope? What domains are engaged? Do we even have residue level descriptions of these engagements? And these are all questions that are identified for, by and large, every bio-therapeutic discovered, but I would argue that it's not always discovered at the earliest stage or the stage where it can make the most impact in driving the program through effective decision-making.


 How Tightly Will it Bind?


0:11:56.2 ND: And then we have, how tightly will it bind? So again, it's engaging a location. We now have a pretty good idea of where it's binding to, but the binding, the off-rate, on-rate and overall affinity become critical for the ability to effectively dose and get a long half-life in-vivo. So again, knowing this in the earlier stage is important because there's lots of strategies to improve this if necessary like affinity maturation, but it's nothing that's great to wait until the last stage of discovery to do. We wanna obviously get the affinity tuned to the correct window as soon as possible on the right set of candidates.


0:12:34.0 ND: And the last point I'll make is what is the effect of this binding, so we obviously have antagonism agonism. So we can measure things like just blockade, if you will. But even looking at the non-variable region side of the molecule, neonatal receptor binding and effector function, just understanding collectively overall, what does this molecule do and how does it interact with the targets and also just other components in-vivo that are gonna be relevant for its ability to effectively overcome some sort of disease.


0:13:12.1 ND: We're left with this one big question, "Which one or ones are to ultimately get taken forward?" Just to summarize what I've just said in these handful of slides, that biotherapeutic generation has really made substantial leaps in the last decade or so. Next gen sequencing is really taking things to the next level, and the competitive landscape is really, really growing fast. There's a lot of companies doing so much innovative stuff out there that it's not... There is no low-hanging fruit, per se. Everybody has the same target they're going after, they're going after it faster with better tools and more information, so it's obviously a competitive landscape.


0:13:53.1 ND: And really, what we'll segue to now in my discussion is the analytical tools on the backend must adapt to these modern scales, so this is particularly from Carterra's perspective that characterization techniques used to determine how tightly does a molecule bind to its target, what's the epitope, all those other attributes that I just listed in previous slides. Historically, these have been very lagging in terms of... There's traditionally great ways to discover thousands of antibodies and not so many great ways to characterize thousands of antibodies, so I'll showcase where things are shifting in the field. And HT-SPR is becoming more and more embedded as the key link of carrying candidates forward further in the process and really getting as much information out of those before the attrition or whittling process starts to begin and we start to lose candidates and lose information as they're culled from the list.


 Carterra LSA Enables High-Throughput Characterization


0:14:53.5 ND: Yeah. So if we think about it really, characterization is a risk-reduction exercise. The Carterra LSA enables high-throughput characterization, so we use surface plasmon-based resonance detection, real-time label-free detection of two molecules binding each other, and most commonly in this case it would be biotherapeutics interacting with their antigen, but could also be a neonatal receptor or other SC gamma receptors, for example. So really any sort of combination of molecules interacting can be measured in a label-free format rapidly. SPR itself has been around for a number of years, probably 30... I guess we're going on 30 years now or so as a commercially available technology, but historically, like I had said, it's been quite a low-throughput technology. So really, with the LSA, there's a whole new level of throughput that comes to play now at the disposal of drug discovery researchers.


0:15:52.3 ND: The real main benefits of the LSA are the ability to screen more clones simultaneously in a single experiment. I'll have a slide giving those numbers in just a bit. The results are in substantially less time, so experiments and sets of experiments that would take weeks to months can get done in a matter of days, and even less than a day in some cases, and then really, really small amounts of sample. So again, moving thoughtful characterization earlier in the process means they are also coming into portions of the process, where there's not been significant scale of the material most commonly. So we do need to work with very small amounts of material, both the antigen in some cases, and certainly the biotherapeutic candidates, antibodies, SCFDs, whatever they are in that format at that point in the stage, so critical that we use small amount of sample. So really the needs and the abilities to fill those needs are all checked off by the Carterra LSA.


0:16:48.9 ND: And the LSA itself has two microfluidic modes that really take SPR to a whole new level. On the left here, we have our... Hopefully, everybody can see my pointer. We have our print head or multi-channel mode, where it takes 96 samples at a time, draws them into the instrument and presents them to the chip surface, and we build an array, shown here now on the right, on our gold-sensing chip surface. And in total, we can do 96 blocks four times, equaling 384 unique address surfaces on the sensor chip. And then we come in with a single channel fluid device that encompasses that its entire array of, let's say, biotherapeutic candidates on the surface and flows the sample across all 384 simultaneously. So with one injection of, say, one antigen, we measure 384 interactions simultaneously, and there's also additional 48 reference inner spots in there that are used to do typical data correction in SPR as well.


0:17:50.0 ND: So it's quite a significant amount of data that comes through per injection, and you can imagine, if you just do multiple injections, that that number scales tremendously. So really we then start to see these metrics that play out in the bottom half of the slide here where we're getting 100 times the data in about 10% of the time and about 1% of the sample requirements in really any of that platform that's out there. So again, this is a transformative leap in biotherapeutic characterization, which ultimately allows for a tremendous leap in biotherapeutic discovery, because traditionally this was sort of a bottleneck process... In the process.


0:18:28.3 ND: So the types of samples that you can bring onto the LSA measure really are quite broad-reaching antibodies, biosimilars, samples from different workflows like CAR-T and crude extracts, even B-cell, supernatants, all can be brought into the system. It doesn't matter if the proteins were common or native. And even there's some groups doing work with membrane proteins at this point and sort of non-traditional antibody formats as well, so quite a breath of molecules that can be used in the system.


0:19:02.2 ND: And then if we look at the core characterization approaches, so really these three assays, kinetics and affinity, epitope characterization, and quantitation, are the main assay formats that customers gravitate to on the system. It's really... The workflows are really purpose-built to address these quickly and easily. And if we look, starting at kinetics, we have the ability to determine on rates and off rates, as well as steady state affinities in the same experiment, attachment strategies of the biotherapeutics most commonly to the surface or covalent, they can be non-covalent as well, and you have the ability to use crude or purified sources. In terms of epitope characterization…


Most Commonly is Epitope Binning Assays


0:19:46.0 ND: Most commonly is epitope binning assays, competitive epitope binning assays, so these were sandwiching type assays, but there's also peptide mapping and mute mapping assays where we can look at residue or domain level kind of mapping those interactions and even blockade type assays that give us information on where we get, for example, antigen engagement, and it really helps to understand mechanisms action, we have a really, really good detailed description of where binding is occurring, and again, this is coming very early in the process 'cause again, you could go out of crude material, without even having to do a purification and get this level of detail that would give a strong argument to how mechanisms action is working based on where you're mapping to specific residues or domains on a protein. The last sort of assay that dovetails into all this is quantitation, so obviously working in crude material particularly requires an understanding of what titers are present. If we're using a standard curve and a known protein, we can do a label-free quantitation on the LSA, get broad dynamic range, and again, the sample type does not really matter to the instrument, it can be crude or purified, both are amenable.


0:21:00.4 ND: So when we kind of step back and look at the numbers of throughput here, they're tremendous, so Capture Kinetics in a single experiment, we can screen 1152 clones and unique clones, so that's 1152 affinities that about a day's runs. So huge, huge throughput there, Epitope binning. You can do up to a 384 X 384 competitive matrix. The 384 clones fully competed, both in solution and on the surface against each other, totaling about 150,000 data points, so a tremendous amount of information here and giving you a really, really detailed epitope resolution. Similarly is epitope mapping or where we take peptides and we can put down a 384 peptide array, maybe this is an overlapping library from an antigen and then present up to 384 maps across that, so we can definitely map to the residue level, any antibodies that are binding to linear epitopes in this exercise, and it's a very straightforward assay to set up and run, and then just running through some other assay types, so quantitation type assays looking at concentrations, again, we can do 1152 in a single experiment unattended, and then general blocking assays and multiplexing assays, we have about a 384 capacity as well.


0:22:16.7 ND: So really you're in the hundreds, if not thousands, for any assay you wanna be run on the system, so that really starts to match well, with the needs of hundreds, if not thousands of potential candidates early in the discovery process. And just a quick note on the set-up of the system, so it sounds like the system might be highly complex and take a lot of effort and complexity to really set up really it's... It's quite intuitive. This is kind of a screenshot of the control software to set up the instrument, really we just put in the plate of reagents that we wanna array on the surface, plate of reagents that we wanna inject across the surface, and using a very friendly, sort of graphically intuitive interface, choose the amount of time for each portion of that cycle for the injection, so it really only takes even for the most complicated experiments, about five minutes to write any method and the actual kind of sample prep and where we put the samples, the system is fairly agnostic too. So it's really, really friendly for setting up and not requiring lots of complex plate layouts or anything to prepare the sample.


Kind of Horsepower in Terms of Throughput


0:23:25.2 ND: And then on the back end, obviously, with all this kind of horsepower in terms of throughput and the ability to generate data, there's a need to analyze that data. So we have two software packages, Kinetics and Epitope, which kind of by their names sakes Kinetics is heavily focused on doing kinetic characterization of interactions, and Epitope is more focused on competitive binning and mapping type assays, looking at where something binds or where something binds in relation to other candidates in the assay. And really the system, despite it being sort of Ultra High end in terms of its throughput compared to other platforms that are out there, gets wide adoption across all sorts of research groups, so obviously, we have huge major pharma companies adopting the platform, but even we've got academic groups and government groups as well, and even smaller kind of research CRO teams grabbing the instrument because really, this is what you need to make the most out of your library or whatever, again, your source of antibodies is you have the ability to get the most information as soon as possible.


0:24:31.1 ND: So it kind of is a necessity more than just a luxury in the drug discovery space. So I've kind of given an introduction to where biotherapeutic discovery is today at a high level, kind of some of the changes in the discovery landscape and where the LSA is at a high level, and how that kind of fits in, but I'll maybe marry those two points together in this last third of my talk to really go through how HT-SPR is enabling unique discovery workflows for a number of different research teams because these are all, what I'm gonna highlight in the next few slides are all published data sets that everybody on this call can go ahead and check out on their own, but all manuscripts highlighting how HT-SPR in very unique ways and in different ways is enabling discovery workflows via its Ultra High throughput and minimal sample consumption and wealth of information that it generates.


0:25:33.3 ND: We will start off with Twist, this is a manuscript from 2020. It came out of Twist which is a contract research organization but really specializes in their DNA technology, that's really head and shoulder and kind of cutting edge above what historically has been done in that space. So this particular manuscript highlights how in just a matter of about one month, they took Ebola Survivor B cells, extracted antibodies, sequence and did a number of really rapid steps to get first pass HT-SPR epitope binning identified some core, if you will, sentinel antibodies from that step that they were linked to major epitopes, scale those up and actually did the second pass really drill down and understand these epitopes further, but that what's kind of shocking when you look at this is how fast they've taken...


0:26:36.3 ND: You know now survivor and convalescent serum E-cells and put them through this process in just under a month, and really, you can see that in this process two major stopping points along the way are to do HT-SPR binning in here. Really highlighting that, if you wanna do this fast, you wanna get as much information as you can from the sequences and act on that, in order to understand an antigen, this would be the way to do it.


0:27:04.3 ND: In this particular case, the sequences were just derived from the patient and they were publicly available, so really there was no information in this that they needed to do this other than just purchasing the particular Ebola antigen and using the sequences that were out in the literature. So it really shows you how minimal kind of tools that you need to actually get this level of data and how quickly you can do that, and so really powerful.


Checking the Manuscript Showing the Heat Map


0:27:34.3 ND: And on this next slide is a little bit of a nice figure from that, checking the manuscript showing the heat map we have here on the right. So they competed a number of antibodies kinda shown as ligands, which are surface bound species. Antibodies on the surface in the assay and against the analyte, so in this case it was 52, the total antibodies injected across the array to look for competition. And the red cells in this heat map indicate competition, the green cells indicate sandwiching and highlighted in the figure are communities, basically of antibodies that are similar, if not identical in their epitope recognition in the assay. And then kind of colored onto this structure, the Ebola antigen protein is the different communities themselves shown here in different colors, and even some novel communities which are kind of down at the bottom of the slide here, showcased as well, but not, I don't believe necessarily all those are indicated here. So really, we're taking a quick level characterization data and instantly going back and applying a structural context to that. And again, this was done in less than a month, we have this level of data, so this really speaks to the speed that you can get from running assay that just simply tell you how antibodies localize based on competition and understanding structurally where they bind and what they actually mean, so really, really powerful stuff.


0:28:58.0 ND: Really exciting, kind of to see how fast these types of assays progress, and obviously this is put out around the time frame of when COVID was really wreaking havoc on the world, so the authors obviously highlighted that this is... The test case here was Ebola, but it's easy enough to apply this to COVID or any other kind of infectious disease outbreak where time is of the essence. But we can get a huge amount of data with a minimal amount of knowledge beforehand... And then another manuscript that came from Abveris, this was looking at Abveris particular technology, they use the diversimAb strategy a particular humanized mouse, that gives them some advantages in terms of the antibody breath that it develops, and really comparing here, in this case, they were looking at single base number of workflows versus hybrid number of workflows, kinda the time, time it took to generate these clones, or these candidates, if you will.


0:30:03.2 ND: As well as you can see highlighted here is the Carterra characterization that fell in place during this process. So again, this particular group uses the Carterra LSA for both kinetics and epitope characterization, and again, owing to its rapid nature, but this is a great manuscript because it did highlight the ability to try different mouse types and kinda compare those, and when you do this level of characterization you do get the ability to see what is the epitope diversity you are getting out of these... And again, going back to one of my earlier figures that I presented on, you have information from flowing back rapidly into the process because within a matter of less than two months, we already understand the epitope landscape and the binding affinities we've gotten from one approach versus another, so this is quite a powerful strategy and a strategy we've seen utilized in a number of different research groups who wanna really sharpen the edge of their discovery processes and understand what is the approach to getting the best diversity. And particularly against many targets, there's multiple strategies taken to maximize chance of success, and this obviously gives a tremendous amount of feedback into the process as it goes along in terms of epitope characterization and kinetic characterization.


0:31:20.1 ND: And the next slide here, highlight just a little bit of that affinity data, so kind of on the right here, we have some Sensogram showing some actual data from the study, looking in particular a clone against the antigen that we measure on-rate off-rate, and ultimately affinity from this interaction, and this is done for hundreds of samples. And then on the left-hand side here, we kinda have a layout of affinity, so we're really looking at two different mouse strains, kinda understanding which mouse strain might be delivering higher affinity and what that distribution is. So again, lots of great information doesn't take... It really only takes microgram quantities, a few microgram quantities of each clone in order to generate all this information and you can be doing it within weeks of when the clones were isolated and discovered.


0:32:14.2 ND: So shifting gears a little bit to looking at BI-specific discovery, this is a manuscript from Ligand, showcasing their OmniClic technology. When they're doing this particular assay and their chicken models, they wanna understand whether or not they're getting diversity as well and kind of how it matches up to standards. So in this particular manuscript, they looked at, again, high affinity and brought up a top coverage, generating these heavy and light chains in order to get bi-specific capabilities out of them. So in this particular figure, I'm showing here, we've got a dendrogram showcasing the different communities.


Progranulin Binding Domain


0:32:57.1 ND: In relation to the progranulin binding domain, so progranulin must be antigen test case used in this particular study, they mapped several domains, which overlaps very well with the particular epitope clusters they were finding... So these clusters on the right are out of the Carterra analysis software, proprietary clustering tools. And the great part about this, is using some standard known antibodies that are mapped. We see really good diversity, they're getting good coverage across all the major domains and distinct groupings, which means that just from the perspective of developing broad epitope coverage and discovery campaign, they're really checking the boxes... We go to our next slide... In addition to epitope, they're comparing kinetics of the clones with the same DH but native or germline B chain, so we've got iso-affinity plot here on the left, showing on rates plotted against off rates, and with the affinities indicated diagonally here, for the different domains of progranulin, and then on the right-hand side, just the distribution based on affinity of the VK... From meta germline sources. So really, really cool stuff in the sense that they're developing by therapeutics, excuse me, by specifics in this approach, and again, leveraging all the LSA to really give them the information they need to make these decisions.


0:34:27.7 ND: So similar to the previous set of slides of that various manuscripts, in this case, they're also showcasing getting more information soon or understanding antibody generation, how it's being done, and ways to improve it. And then another case for by specifics that possibly... Is a little bit of a different twist is this paper that came out of the NIH, actually just this past year, so looking at by specifics driving from plasmablasts and Memory B-cells focused on overcoming SARS-CoV-2 mutation. So, basically having to... Means of engaging source code 2 instead of one, to reduce the likelihood of mutation and viral escape. So we're plotting some affinity data here, that is the figures out of that manuscript. On-rate versus off-rate. Again with diagonal affinities shown in here from the different sources, MBC versus plasmablast, this is against the RBD, and this is the N-terminal domain. Shown here on the right as well, same kind of plot though. So again, there's of lots of great information here about where affinity is really coming from? Do we have high affinity at the get go? Or is there some differences here we can tease out?


 Speaks to the Power of the Throughput of the System


0:35:43.9 MK: So again, really speaks to the power of the throughput of the system and not needing much protein, so we're just taking these plasma B-cells plasmablast, Do we cell-derived sources and doing some really great assays with them. And then kind of going back again to using the epitope structural relationship to inform the whole process, so these are epitope communities to find in the experiments and displayed in the analysis software and understanding really mechanism action where we're binding to on these particular proteins. So you can see here, the N-terminal domain, receptor binding domains are both highlighted structurally and the coloration shown here, is indicative of the particular binding domains in the community plots we have purple communities and a blue community indicated on the structure. So again, really early on with not a lot of investment here, we simply had candidates and the antigen. We were able to take all that and make a structural inference from it, using the Carterra LSA.


0:36:50.4 ND: And then my last sort of piece I'll highlight on, and we had a great talk a little earlier today from Valentin from AbCellera, was on this Bamlanivimab exercise where they discovered an antibody in about 90 days, which is really the fastest time to date for getting a bio-therapeutic in a clinic, and really the quote here, I highlight that and really calls out how is this possible? What was the change that occurred here together? Human treatment in only 90 days and really advanced discovery and characterization platforms were called out in that manuscript of being key enablers in this process. So again, this is setting the benchmark for everybody in the field that it's no longer years to discover biotherapeutics and get them to market. It's a matter of really months.


0:37:43.8 ND: Certainly, this is under extreme circumstances, but it shows that the process is possible and going back to that Elon Musk quote: "Even though there is the risk of failure, going after it and reaching for that high goal is definitely feasible." So this publication came out this past year, highlighting a discovery Bamlanivimab. And if we go to the next slide, we'll see a little bit of data from that manuscript, so we've got kinetic and epitome profile, and this was the cool part about this, this is just one week of experiments, so basically the patient-drived antibodies get scaled up a bit and within one week, we have all this data in our hands understanding where it's binding to, so these are antibodies that are mapped and showing competition, the different regions of the proteins, the affinity breaths from this, going back to those questions of "Where does it bind?" and "How tightly does it bind?" please check the boxes. We have this information. We have it in seven days effectively of where to drive the discovery process rather than waiting months and months for the assays to catch up once the antibodies have been generated.


0:38:53.6 ND: So, really speaks to the tremendous power of the platform. Really what it can enable and really this is just for Carterra's sake, we're just so happy to be contributing on this... The pandemic is obviously has been life-changing for so many people and difficult, and our ability to contribute something that has the immediate impact on human health and really hopefully get this issue just really exciting. So we're very proud from the Carterra's side, everyone for being able to enable this great work at AbCellera and Lily have showcased here.


0:39:30.0 ND: Alright, so with that, maybe I'll just wrap up my key thoughts from my presentation, so one biotherapeutic discovery really is a highly competitive space, and the competition drives the search for more and more sophisticated discovery strategies, there's no way around this but there's tons of really great research groups out there, they're all driving to hit kind of a similar set of targets, some people have some more ability to get to more unique targets than others, but frankly, really it is the discovery platforms that are differentiating highly in this process. But what is one necessity, no matter what your process is to get to these lead candidates, you have to just do a detailed characterization of as many clones as possible, really this not only helps reduce risks, which is critical, so you're ideally identifying robust development candidates, you're also ensuring that in a commercial space that you're able to operate with solid intellectual property, and it just means that when we get into the hundreds of millions of dollars in the clinic that we're not surprised by something we did not uncover earlier in the process.


 Kind of this one Shared Step Amongst all of  These Different Discovery Workflows


0:40:38.9 ND: So really a characterization is kind of this one shared step amongst all of these different discovery workflows, it's necessary. And really, as I've shown in the last few slides, essential to kind of accelerating things using the LSA, and then the last little point is Carterra's HT-SPR characterization approach via the LSA platform, really gives critical insights and unlocks the full potential of biotherapeutic discovery, as you've seen in a number of those different example cases, there's just a wide breadth of discovery workflows, doing different strategies, but it all kind of boils down to the same things that's asked of the LSA, where is it binding? How tightly is it binding? Am I seeing blockade? Am I getting off-target binding? All these questions can be answered on the LSA, so it's become a huge enabler for these research groups, and we're just, again, really excited to be enabling this and kind of at the forefront of cool new things happening in biotherapeutic discovery. So with that, I'd just like to say a big thank you for everybody into attendance and for anything else that we can follow up on that we haven't covered here or we don't get to maybe in the Q&A that's coming up, please just don't hesitate to reach out to us, and we'll be happy to discuss more with you. Thank you.


0:41:57.2 MK: Great, thank you Noah for an excellent presentation. That was really interesting. I see that we've received a few questions already in the chat box, but we'll give the audience a few moments to go ahead and answer enter your questions in there on the Q&A box to the left of the slides, and we'll pick up on those in just a moment. But before we do begin the Q&A, I'll just run through some real quick announcements. First, I'd like to thank all of our sponsors for sponsoring this digital week, next I'd like to mention that we have several antibody and engineering therapeutic digital weeks and events planned for 2022, so please keep an eye out for more details, be sure also to take a look at our upcoming antibody engineering and therapeutics US event, taking place as a hybrid event this December in San Diego, it's registering very well, really solid in-person attendance. So we're excited about that. Also, be sure to check out the resources list to the right of your screen where you can download some featured articles. Now, let's get back to Noah, for some Q&A and the first question I see in the chat is... Noah, could you describe how the antibodies are attached to the sensor surface on the LSA?


0:43:16.6 ND: Yeah, so really, there's kind of two main strategies, I sort of mentioned this briefly, so it's great we have this question to allow me to follow up in more detail, but effectively you can do a covalent attachment strategy, but commonly that's a mean coupling through the primary means on the protein, or there is the ability to capture, in the case of an antibody where we have an Fc region that capture is usually done on the Fc region maybe an anti-Fc antibody or protein A or protein AG, so those are pretty much the two approaches covalent or non-covalent and using a capture or direct coupling VME coupling, the two approaches you would use.


When Performing Experiments from Crude Sources


0:43:58.2 MK: Perfect. Next question, When performing experiments from crude sources, what are the minimal concentrations of antibody required?


0:44:09.0 ND: Yeah, so this gets interesting, so on the LSA crude samples aren't a problem, I think this is... [chuckle] kind of a lot of people, I guess, 'cause surface plasmon resonance most commonly relies on microfluidics that there's the tendency to think that crude samples are not amenable to the platform but really, they're not a problem, and I would say that the majority of LSA users actually run crude samples in some capacity in addition to maybe purified as well. But going back to the question itself, it kind of is I would say dependent on what you're trying to do, if you're doing kinetics, you can get pretty low down to... I've seen some datasets where even 50 ng/ml was giving reasonable enough signals. It depends on the size of the antigen as well. So on the kinetic side, you can get to maybe 0.1 to 0.05 µg/ml. If you were looking for doing something like thinning, you probably need to have a little bit of a higher concentration, so if you're getting maybe 10 µg/ml that's better, we need a little more robust surface density to really make sure those assays have great signals on them, so those are the two ranges about 10 µg/ml minimum for thinning and maybe about 0.05 µg/ml for kinetics.


0:45:23.9 MK: Perfect. Perfect. Just for attendees just reminding you, please go ahead and answer enter your questions in the chat box, we do have about 15 more minutes for questions, I see a couple more in there, but just if you have any questions please enter them now. Next question Noah, is the LSA practical if the number of clones I have is typically less than 384?


0:45:48.3 ND: Yeah, yeah, so I think the numbers that we show as our maximum throughputs are sometimes daunting to people. They say, well we're not at that stage yet or just for whatever reason, that's quite a bit higher than they're planning to use. So the short answer is yes, the LSA has max throughputs of 384, and even you can go screen like 1152 in some experiment types, but realistically, if you want to do something like really get the most out of your experiments, you can build in things like replicates or test surface densities, kind of in the same experiment.


0:46:21.1 ND: So you might not have 384, maybe you even only have, let's say 48, but that 48, you can print them at multiple densities, and if you're doing kinetic screening, this is a great way to do a one and done where you basically get multiple densities and then when you do the analysis, you can figure out where the optimal density was, so you both optimize and run the actual experiment at the same time, and you also have the ability then to build in replicates... So replicates in SPR are really unheard of, 'cause most systems just don't have the capacity to make it practical to run replicates, but it's easy on the LSA to do triplicate measures of kinetic interactions, for example, and really get confidence in the measurement all in the same experiment. So definitely, there's some customers that'll go... You use 384, you need to screen those from the outset, but there's many that will do much less than that and just take advantage of the replicates, for example.


0:47:18.9 MK: Perfect. Great, thank you. Attendees, this is a last call for questions for Noah, I see one more in the chat, so you have a little more time to enter a question if you have one, so Noah, here's the last question I see for the moment from crude samples, Can you comment on the non-specific binding signal?


0:47:40.2 ND: Yeah, so I think this question is probably looking at when you have a crude sample going across the surface, there's your protein of interest in there, but there's probably some other components in there, maybe from cell lysates or something like that, that can interact from the surface, so there's probably nothing on the LSA that's unique compared to any other kind of label-free platform historically that's been around all the methods for addressing non-specific binding are fairly tried and true, so you would look at the surface chemistry, how you're attaching the molecule, is there any sort of non-specific interaction that you could reduce with changing the tip chemistry so that would be an obvious place to start and how you're attaching it.


0:48:21.4 ND: It might also be from something you add to the buffer to improve that, so sometimes adding protein components like BSA or KC into the buffer might help with a non-specific binding, so yeah, there certainly is a chance that crude samples could give non-specific binding... The one kind of advantage though, in the LSA is we often array our captured... Our crude samples have captured onto the surface, and between capturing and then going on to running the antigen across the system, inherently washes itself out so you typically wash away some of that crude material in that process, so even though they start off crude, you actually sort of do an on-chip purification, and it's usually not a problem for the downstream measurements that you make, but in some cases, like I said, if you do have to measure the crude sample directly, considering the chip surface, the chemistry and potentially the buffer added, those are usually the main areas to improve the process.


What is the Size Limit on the Molecules?


0:49:20.0 MK: Perfect, next question, what is the size limit on the molecules that can be measured in this system on both the high end and the low end?


0:49:30.3 ND: Yeah, good question. So really on the low end, we don't recommend anything smaller than about 1,000 daltons, so pretty small peptides are about the limit, we don't want to be... I think customers potentially could push a little below that, but we really don't recommend it because the signal to noise just really doesn't supported it as a robust assay. So yeah, about 1,000 daltons I'd definitely say is the limit. On the upper end, there really isn't a practical limit necessarily, I mean these proteins that are huge, 500,000 kilodaltons or something like that, or 500 kilodalton can be run on the system. I don't think there's necessarily a size limit there. Or a limit. Just for size itself.


Is the LSA Best Used by Therapeutics Companies


0:50:15.9 MK: Okay, great, next question. Is the LSA best used by therapeutics companies, CROs, reagent companies. Basically, what's the ideal investigator that Carterra is looking for or working with?


0:50:32.2 ND: Yeah, good question. So one of my earlier slides kind of highlighted it, it's really everyone... They all find utility, if you're discovering biotherapeutics, antibodies to stay in the game and be competitive, you need to be able to characterize hundreds, if not thousands of clones, for example, and even reagent companies know that. We have a number of customers that are reagent companies, so I didn't maybe bring those out as much in this discussion since I'm talking more about biotherapeutic discovery, but really reagent companies, it's critical when developing an amino acids reagents that you find great pairs, you understand affinities. So we have a number of companies that have really latched on to the system and are really leveraging that to get improved reagent generation and characterization capabilities, so they're not... Most of them aren't developing as therapeutics, maybe some of them are in a service capacity for outside parties, but for the most part, they're just looking as a way to generate the best amino acids reagents in the quickest way they can, and just generally for any other reagents it gives the customer more data on them, so kind of the same process as you would find in a traditional pharma is happening at reagent companies, but just different end goals.


0:51:47.2 MK: Got it, cool. Now, the last question I see in here is, can we measure non-antibody types such as nanobodies and what special methods do we have to enable to do so?


0:52:01.5 ND: Yeah, so that's sort of why in the title of my talk, I kind of called it biotherapeutics, 'cause there is a huge, lots of different molecule types that folks are generating out there, they're all protein or peptide-based, and they don't fall into the classical antibody description. For the most part, there are some, maybe portions of antibodies in the case of nanobodies or other scaffold protein biotherapeutics for example, that are out there, but the short answer is really, yes. Really anything larger than about 1,000 daltons, like I mentioned, can be analyzed on our system, most commonly, we array the biotherapeutics on the surface, so in this case, like in nanobodies, we would array them on the surface for smaller molecules, it's best probably to use a capture strategy 'cause we can... If there's a tag or something on there, we can grab them with good specificity and make sure the orientation is proper for the assay, but it's really nothing necessarily special, it's just a matter of having a V5 tag or some other means of grabbing onto the molecule that's amenable to the discovery and antibody generation process.


0:53:15.7 MK: Perfect. See another question came in, in what format are voltage-gated channels used on the Carterra? Is it solubilized in detergent?


0:53:27.8 ND: Yeah, I think one of the examples we had recently worked with a customer or became aware from a customer, I believe they had been solubilized in detergent, I believe, but I think they were captured onto the surface in that particular instance, so I believe they were used as ligands in the assay, if that's what the question kind of is.


0:53:47.7 MK: Okay, perfect. Alright, well, I'm looking at the chat, it looks like we have exhausted all the questions in the chat, so... Noah I'd like to thank you for a great session. For attendees still online, I'd like to... Please take a moment to complete the feedback form so we can continue to improve your digital week experience. And on behalf of Informa Connect Life Sciences, Noah, and the audience, I hope you have a wonderful day. Thank you.