Part 1: Identification of Monoclonal Antibodies that Block Ion Channel Activity
Learn how Tetragenetics has developed a strategy to generate and identify monoclonal antibodies targeting ion channels based on high-level expression of recombinant channels in Tetrahymena thermophila, enabling the development of potent immunogens and tools that allow deep mining of the immune repertoire.
Part 2: Carterra’s High Throughput mAb Discovery and Characterization Platform
Carterra’s LSA can analyze up to 384 binding interactions simultaneously, providing mAb characterization throughput to match the output of the current generation of mAb expression platforms; all antibodies can be rapidly and comprehensively screened so that unique epitopes and potential novel therapeutic candidates are not missed while expanding IP coverage.
Combined with application-focused analytical software, the LSA facilitates high throughput:
Paul Colussi, PhD
Vice President, Research
Yasmina Noubia Abdiche, PhD
Chief Scientific Officer
0:00:00.1 Speaker 1: Let's now begin by introducing today's speakers, Dr. Paul Colussi, Vice President of Research at TetraGenetics. Dr. Yasmina Noubia Abdiche, Chief Scientific Officer of Carterra. Thank you for joining us today Paul and Yasmina and thank you to Carterra for sponsoring the event.
0:00:18.1 Dr. Paul Colussi: Thanks for the introduction. We're gonna be talking today about our strategy for identifying antibodies against human ion channels using tetrahymena thermophila, the system we use to make the recombinant ion channels. So, ion channels, of course, are at the core of many diseases. They are a very large potential therapeutic target list. A lot of targets in oncology, pain especially, autoimmune amongst a host of others are, but really under-represented in terms of therapeutic development of molecules in particular, biologics. One of the challenges with ion channels is that in terms of drug discovery, in particular antibodies discovery, is the lack of a robust, recombinant source of ion channel, where you get enough of the protein to really develop and generate the antibody screening tools that lets you get deep into the immune repertoire and pull out the rare antibodies that will actually modulate the function of the ion channels.
0:01:42.7 DC: Now, with recombinant expression in general, typically there's always a balancing act between the quantity of the ion channel or the recombinant protein, in this case, the ion channel and the quality of the recombinant protein. So, the ion channels, certainly probably the gold standard recombinant expression platform is mammalian cells, insect cells also do very well. Alternative systems like bacteria and yeast will generally make a lot more of a recombinant protein, but you'll suffer on quality. Typically for mammalian cell produced ion channels, the quality is very high. It's just the quantity is so low to make drug discovery in general, and antibody discovery in particular very difficult. So that's just one part of the challenge. The other part are the ion channels themselves, there's a lot of technical challenges with targeting channels with antibodies.
0:02:56.0 DC: To begin with, although a lot of the channels are relatively large molecules, the target surface area, the exposed, extracellular loops tend to be pretty small, and there's a high degree of conservation. And typically, you require again, going back to the expression platform, you typically require quite a lot of the protein, either for reimmunization or various screening strategies. So that's the overall challenge. We have really focused on the recombinant expression, it doesn't really solve the problem of small epitopes or not particularly immunogenic epitopes, but it does allow you... If you can crack a robust production of the ion channels, it does allow you the ability to develop the kind of screening tools that will find those rare antibodies.
0:04:04.7 DC: Our production platform for the ion channels is based on tetrahymena thermophila, this is ciliated produce that you find in ponds, fresh water ponds. It's a true microbe, it grows rapidly, it has a doubling time of two to three hours depending on the strain. You can get it to high cell density. It grows on a lot of basic inexpensive peptone based media. The genetics have been worked out and it's easily frozen for long-term storage, and you can also grow it at scale, not that we do for our use. Importantly though, despite having most of the attractive features of a microbial expression system, it's still a very sophisticated eukaryote. It has about 20,000 or so annotated genes, so getting close to mammalian genome numbers. Mammalian like PTMs. It has no cell wall, so it's easy enough to get proteins out of membranes. It has a number of enhanced features for production of membrane and secreted proteins. And probably most interestingly, the bug has really developed or evolved to spend a lot of its metabolism on producing the kinds of membrane proteins like ion channel that we're interested in generating recombinantly.
0:05:56.6 DC: So, if you look at this table on the right-hand side, this gives you a comparison of the major membrane protein families, ABC transporters, MFS, voltage-gated ion channels and P-type ATPases. It kinda gives you a comparison of the genes that encode those proteins in tetrahymena compared to what's encoded in human cells. And for each of the major classes, you're looking at three times as many of those family of proteins that are encoded by Tetrahymena. So the bug is really evolved to produce those. It spends a lot of its metabolism producing these proteins, and so we have really... Using that as a base, we've spent quite a bit of time developing the bug through standard molecular biology, our vectors, our culturing conditions to get the bug to produce recombinant human ion channels in particular. And that's basically where we start all of our ion channel antibody discovery. So, if you look at it from a... If you stand back and look at our complete strategy, the first step is always a customized expression of a human ion channel in Tetrahymena.
0:07:22.8 DC: The goal of course, is to get as high yield and preferably in a purified format and high quality amount of protein. Once we've achieved that, we'll typically split the program in two. We'll spend time developing antibody screening tools. We actually spent quite a bit of time developing solid state tools that really work well with various antibody screening platforms, like Phage display and what have you, things that will allow us to get really deep into the repertoire. And the other part of what we do is because of the conservation of a lot of these ion channel family members, we try to get into multiple and diverse antibody platforms, so for any one program typically we go into two or more animals. We will do mice as well, but we like to go into morphology genetically distinct species like chickens for example where we've done quite well, llamas as well and cows, that have their own... Particularly for ion channels, their own advantages in cow bodies. And we basically do that to mitigate tolerance, and as long as we get antibodies from each platform, then we'll go through the process of screening them to discover... To screen out the ones which functionally modulate the channel.
0:08:53.5 DC: So, let me give you a case study, which is our Kv1.3 antibody discovery program. So Kv1.3 is a voltage-gated potassium channel. The output subunit is made of a tetramer, you can see the diagram on the right-hand side there. This is a very common structural architectural element of the voltage-gated ion channels. It's actually a tetramer of those... Made up of four of those six-transmembrane domains. It's a very well-validated target for autoimmune disorders including psoriasis, multiple sclerosis, lupus and Type 1D. And the reason of that is, Kv1.3 is one of two potassium channels that regulates membrane potential and cell volume on T cells, especially effector memory T cells. In the case of auto-immune disorders, Kv1.3 is significantly up-regulated in auto-reactive effector memory T cells. And there's been a number of studies that have shown that if you block Kv1.3 activity, you get therapeutic benefit, and that's actually just been shown in the clinic with one of the Kv1.3 peptide toxins.
0:10:23.0 DC: So again, we start here with expression in Tetrahymena. On the left hand side, you'll see a Western at the top. This is kind of the... That shows you the kind of optimization that we're looking for just in the culture. This is really culture conditions that are some simple things like the type of media, the temperature control, the kinetics of the expression. And there's also additives that tend to be ion channel specific, in this case we found one that really boost the expression of Kv1.3 which is, you see on the right-hand side. Where possible if they are reagents, we like to make sure that the channel is being made correctly in Tetrahymena and makes it to the cell surface. That's a pretty good early indication that the bug is making the protein correctly.
0:11:21.6 DC: And so in this case, the Tetrahymena that are producing the Kv1.3, if you look at the image on the bottom left, here we've labeled the cells with a toxin that only binds the correctly-assembled tetramer. You can see the night cells surface staining so we know at this point that the bug is putting together the channel correctly. And so, from this point we proceed to the purification. So here we've spent some time with the typical detergent-based extraction, affinity purified. Typically with the proteins, with any of the membrane proteins we work with, especially the ion channels, we will reconstitute the purified protein, take away the detergent, reconstitute it into a lipid bilayer. So that maintains the stability of the protein, it also allows us to freeze it for long-term storage, as we prepare for the antibody discovery program.
0:12:28.1 DC: On the right-hand side, again the Kv1.3, we were able to test the binding of our purified protein on toxin ligands that we know should bind the protein, and which it did. And so we were confident going ahead that the material was of sufficient quality and was followed correctly. So once we've gotten to this stage, we really spend some time thinking about how we want the platform that we are discovering the antibodies from, whether it's animal immunization followed by some kind of screening platform, whether it's phage displayed directly, cell cloning. If we're going into animals, we spend some time developing the immunogen, which is to say the formulation that will be immunized into the animal, this can be pretty specific. We tend to incorporate adjuvants into the liposome prep so we can easily add lipid-based adjuvants, which we tend to do. We also can add soluble adjuvants so various TLR, CpG agonist.
0:13:44.6 DC: Again, this is gonna be customized based on the animal we're going into, particularly with the TLR agonist that had their own various specific sequences. On the antibody screening tool side, typically we will opt for going on to some kind of solid support. Magnetic beads work pretty well for us. Certainly, they did for the Kv1.3. On the right-hand side, you'll see the basic design. If you look at the diagram in the middle there. That shows the Kv1.3 monomer that we attached to the magnetic bead through engineered tags of the C terminal end. After that, we actually incorporate or we build a lipid bilayer around the entire bead with the attached ion channels so that we can stabilize it.
0:14:45.0 DC: In this case, we spiked the lipid bilayer with some phosphoethanolamine that's being labeled with rhodamine so we can look at the lipid forming around the bead and we check all of this by immunofluorescence. So if you look at the top left-hand side, under rhodamine CE, that's just the lipid around the individual particles. In the middle is an immunofluorescence of these beads done with an antibody that recognizes an extracellular loop where you see the little star on the diagram. So, we can see we're getting exposure of those surface loops, which is what we want. We wanna preferentially pull down antibodies that bind those extracellular loops. And on the right-hand side, you see the merge.
0:15:41.3 DC: The Kv1.3, we actually tested this before we did the antibody screening and just with the... We tested pull-downs using these beads with commercial antibodies that recognize either the engineered tags, either FLAG, or ERs. That should be intracellular with respect to the lipid bilayer compared to the pull-down with the antibody that binds to the extracellular epitope, and majority of the antibody we pull-down was from that external epitope-recognizing antibody. So we've done this for... We've actually done it for a few others but this is a nice way, particularly good for phage display.
0:16:36.1 DC: So, we actually went into two platforms with the immunogen. So, we went into chickens. So we did this in collaboration with... They were previously Crystal Biosciences but they are now part of Ligand Pharmaceuticals, where the chickens were immunized with our immunogens, our Kv1.3 liposomes, and then the actual antibody screening was done with their proprietary GEM assay, which is essentially a straight out B cell cloning assay. The other platform we went into was llama. This was in collaboration with argenx. There, the animals were actually immunized with a combination of DNA and they are proteoliposomes, and then the screening was done with phage display. And so at the bottom there, you see a handful of the antibodies we pulled out that were specific, these are just ELISA, showing that the antibodies are those responsive to the Kv1.3 but not to the control proteoliposomes. For our controls, we typically use other ion channels that we make but we just think that's a better control than just the empty liposomes themselves.
0:18:00.8 DC: So after that, we went into... The next important screen is to set the binders for those that are able to block ion channel activity. Again, we do all of this in transfected mammalian cells at this point. We don't do any of the actual antibody characterization and screening on the tetrahymena-based material. And so what you see there is basically the traces from an antibody that gave no inhibition on the left hand-side to an antibody that blocks the current, the Kv1.3 current, on the right-hand side. And so when all was said and done, when the output from this program from the two antibody platforms identified about 69 binders, most of those were from the chicken, 50 were from the chicken, 19 from the phage library made from the llamas. And from those, we identified 10 that were functional. That was a pretty big hit, it was that 14% functional antibodies to binders.
0:19:13.0 DC: So then, we did some more characterization to get a handle on the selectivity, the potency. Some of these antibodies are showing really quite strong potency, at least below 10 nanomolar for one of them, showing good dose response curves. The one in red here, the C1H6, that's a chicken antibody. The green dose response is from the llama, so it's a little bit less potent. Both of these antibodies or actually all of the functional antibodies show a time-dependent block. So, not as fast as is typical for the peptide biologics, but over about a 10-minute period, we're getting nearly complete block of the ion channel. And certainly not like some of the other antibodies against ion channels, where the mechanism of action is actually a multi-hour internalization of the ion channel antibody complex, this looks like it's a bonafide activity block. Importantly and probably not unexpectedly, this is the major reason people are interested in isolating ion channels for antibodies or isolating antibodies that target ion channels, is the cell activity. So, for the various family members we checked against for other KV family members, we didn't really see that either antibody that we're interested in, is showing any activity.
0:21:01.0 DC: So, the cell activity looks really quite good. So, finally, this is the work we did with Yasmina at Carterra. Because we produce so many functional antibodies, particularly from the chickens, we're interested in seeing what the epitope binning was, whether there were distinct epitopes that were really driving the generation of these, particularly the functional blocking antibodies. And essentially, we divided the antibodies into three groups, one was the functional antibodies, the ones that could block the Kv1.3 function, those that couldn't. And then those that could block, and were also shown to bind Jurkat cells. So they're the red ones in the network probably you'll see below. So, Yasmina did this on 36 antibodies using an array-based SPR imaging. The target was a detergent solubilized Kv1.3, and as you can see from the heat map is shown on the top, but if you look at the networking plot down below, it was interesting, most of the antibodies you could segregate to one bin, bin one.
0:22:27.4 DC: Interestingly, we did this on the chicken antibodies, but we included the llama antibody as well as a reference, and that one also segregated to that first bin. So although, a big majority of the functional proteins went to that bin, suggesting that there was one dominant epitope that was driving the generation of functional antibodies, they were antibodies that segregated to other bins. So you see two here so F1, F8, which is the blue one in bin two and 1DA which is in bin five and 1DA especially in bin five, really didn't find nothing else blocked that particular antibody, so it really stands by itself. I will say there's actually another antibody that wasn't tested because... Which ends up being the most potent by AsES, it wasn't tested here because it really didn't bind to the purified protein strongly enough to allow to be analyzed by this assay.
0:23:39.3 DC: That one actually probably falls in its own bin as well. So, this was really telling us there is probably a couple of ways to kill the activity of these channels, and telling us that at least our preps are allowing different epitopes a chance to generate antibodies that will eventually block the channel. It's also interesting, because I think it really opens up a possibility, if you wanted to really work with a potency. It may be developing bispecifics, where you're picking different bins, as the separate arms of your antibody. So, it was a really nice, very interesting study here. So in summary, let me just say, our approach here has always been foremost to generate enough of the target ion channel so that we can have enough protein where we can spend some time, and after developing the tools, be they immunogens or screening tools, that we can apply to whatever antibody drug discovery program we do.
0:25:02.8 DC: Like I said, we like to use multiple and diverse antibody platforms to at least mitigate some of the challenges associated with generating the ion channel antibodies, particularly mitigate the tolerance from the conservation that's prevalent in a lot of these channels. And so, with our KV 1.3 program, we were able to make a Kv1.3, we showed that it was correctly folded. We were able to develop the customized immunogen and screening tools that preferentially pulled down surface binding antibodies. We went into multiple platforms, and we're able to do functional analysis and then did functional analysis on those binders on human cells.
0:25:54.9 DC: And the result of all of that was about 40% functionality blocking hit rate with that group of antibodies, some of which certainly have the kind of potency that we feel will allow us to develop this as clinical candidates. So, let me just end there by thanking all the people that worked on this. Our crew at TetraGenetics, Yasmina at Carterra, Heike and Hai at UC Davis, they're our Kv1.3 AsES experts. The guys at Crystal; Bill, Ellen, Ryan and Darlene, that pulled out the antibodies from the chickens and the guys at argenx, Bas, Lori and Hans, who did all the work on generating and isolating the llama antibodies. So, that's the end of my presentation. So, let's hand it over to Yasmina.
0:26:55.7 Dr. Yasmina Noubia Abdiche: Hello, my name is Yasmina Abdiche. Thank you so much, Paul, for that great presentation and showcasing how our array SPR technology has helped guide your research program. I'm gonna be explaining a little bit more about how the Carterra LSA platform is being used throughout the antibody discovery to really help characterize different properties of antibodies. In addition to epitope that we heard from Paul, I'm also going to be describing some kinetic applications. So first, I'm just going to show an overview of what I'm gonna talk about. First of all, I will go into a little bit of background about Carterra and who we are. And then I will introduce the array SPR platform that we're calling the LSA. And then I will go into three examples of some applications that are really key to antibody discovery. And they are the epitope binning that we heard from Paul, the mapping, and then binding kinetics and affinity.
0:28:07.1 DA: So, a little about Carterra. So, we're not a new company, we actually have just changed our name, and really developed our applications. We were founded in 2005 and called Wasatch Microfluidics. And we have a patented technology for continuous flow microspotting, which allows proteins or other molecules to be printed onto slides, or gold chips for various applications. And we've really harnessed that ability to pair it with SPR to enable interaction analysis. So, in late 2016, we received some significant funding that allowed us to really move forward in a very application-driven way. And we married the CFM, continuous flow microspotting technology with our own SPR imaging. And we rebranded the company Carterra is a play on words, cartography or mapping and Terra, world. We are literally trying to map your world. So, help you to navigate your antibody discovery.
0:29:27.1 DA: And with this significant funding, we were able to really expand the management team, the r&d and customer support teams, with people who really have a deep knowledge of antibody discovery, and the technical services needed to really make a successful instrument. And we have expanded the facilities, we have the headquarters in Salt Lake City, Utah. And we also have demo labs on the East Coast and West Coast. So, now I'm going to talk about the LSA. So, we're really pleased to announce that in early March, we had already sold four instruments. And these were to a range of initial early adopter customers, including large pharma, biotech, CRO and academic institutions.
0:30:24.1 DA: And we're really excited to have this endorsement of our high throughput array SPR platform. So, the motivation for our high throughput array based platform was really throughput. In antibody discovery, a lot of money is spent on generating antibodies, but analytical tools to characterize them really fall orders of magnitude behind in terms of their throughput. So, while you're making many, many antibodies, you're only really sampling a very small percentage of them and characterizing just a very small subset of your total library. And this means that you could maybe overlook a really great clone, a nuanced binder and possibly throw away a blockbuster candidate.
0:31:19.5 DA: So, LSA is short for Lodestar array, and we like to think of our instrument as really helping researchers as a point of reference to navigate through their antibody discovery campaigns. And the three main applications that we feel we can have the most impact in antibody characterization are kinetics, epitope binning, and epitope mapping. And we have taken a core based approach in our software using icon to detect the assay configuration that you want to run. You would simply click on the icon, and then we take you through a series of questions, and you would be guided through how to set up your experiment. In addition to the experimental layout, we have supporting software to help analyze the data. As you generate a lot of data, you need very powerful, sophisticated and intuitive tools to help you understand what the data means. So, what you can see on your top right are kinetic fits.
0:32:31.6 DA: We have kinetic software that allows for batch mode fitting of kinetics, and we have an epitope tool software that allows for heat map sorting, dendrogram analysis and network plots. We like to think of these network plots or community plots, like the Facebook of antibodies. It shows you which antibodies are clustering with which other antibodies. Also for epitope mapping, where you have a peptide library that maybe you want to array onto your surface and then characterize the epitopes of antibodies and solution, and again, we have software to support this. Our epitope and kinetic software packages, we like to think of as siblings, brother and sister, and so we have skinned them similarly and features in one are often found in the other. So, if you can understand how to use one of them, the other one is very intuitive to use.
0:33:34.3 DA: So, for maximum automated capacities, these are kind of what the capabilities are for the LSA. We enable captured kinetics, and unattended throughput of three 384 well plates, which would be 1152 ligands or antibodies, and we allow for coupled kinetics on a 384 array. We also allow for epitope binning on a 384 array for really unprecedented throughput, and epitope mapping on a 384 peptide array and a variety of other applications that will become quite obvious. Our chips are supplied through Samtec Chemistries, this is a company in Germany who are experts at creating surface chemistry, and we have fashioned them into a cassette that is compatible with our instruments. And we're supplying a variety of chip types that are most appropriate for various applications that you would use for antibody discovery. So, we would have the carboxymethyl dextran surfaces, which are commonly used in Biacore, for example, a similar matrix.
0:34:56.1 DA: We would also have streptavidin surfaces, plainer surfaces, and also more of a two-dimensional hydrogel or poly carboxylate non branched hydrogel. So, a little bit about how the LSA works. So, on the sensor surface, which is a gold chip derivatized in various ways, with different types of surface chemistry, we use two different microfluidic modules. A multi-channel module, which is a 96-channel print head that can dock perpendicular to the chip surface and deliver 96 proteins or 96 ligands in parallel. It stamps onto the surface, docks and undock, and you can do that four different times to nest four times to create a 384 array. Then the print head can move out the way, and a single channel will dock over the entire array to form a single big flow cell. And this really allows for minimal sample consumption because you're now creating a one-on-many assay format. And we also use a bi-directional delivery of sample. The sample is shuttled back and forth across the surface, that has many passages back and forth to interact with the surface.
0:36:26.1 DA: So, this automated flow cell switching between multi and single channel mode allows for a variety of applications, it allows for creating a 384 array with referencing to spots, it allows for reloading an array and it supports the capture formats and standard and mean couplings that you would be used to for SPR. This is a little closer look at our 96-channel print head. So as I mentioned, the print head docks perpendicular to the chip, but it is comprised of these very tiny microfluidic channels, there are 96 in parallel, and each one is discrete and this forms its own spot on the surface. And this is a cartoon or image of what a 96 print head dock would look like. So, the pink rectangles indicate each of the discrete spots and the blue rectangles indicate the inter-spots or local referencing. And as I mentioned, the print head would dock, undock and shift across, and you can do that four different times to nest four 96 spot arrays to create the 384 spotted array.
0:37:50.1 DA: Once you've created your array, either your 96 array or a 192 or 288 or a 384, depending upon how many times you dock, you can then flow your analyte in a single channel mode across the entire array, and that would flow back and forth using a single volume, this is very, very efficient on your analyte consumption. So, with the LSA sample deck on the left-hand side, we have dedicated two bays that can accommodate a 96 well plate or a 384 well plate, and the bay one can additionally accommodate a sample block, which would house Eppendorf tubes or conical tubes upto 50 ml quantity. And this would service the single-channel mode, this large flow cell. Then on the right-hand side of the sample deck, we have dedicated three plate positions, bay three through five. These can accommodate 96 well plates or 384 well plates for a maximum capacity of three, 384 well plates. The whole deck is also chilled and your samples obviously would have covers on them.
0:39:08.7 DA: So, let's talk about epitope binning. So, when I talk about epitope binning, I'm referring to a competitive assay where we're asking the question, can two antibodies bind at the same time or not to their specific antigen? And if they can bind at the same time, we infer that the two antibodies in question are binding non-overlapping epitopes, and if they can't bind at the same time, we infer that they're blocking one another's epitope, in other words, they are overlapping. And in order to do this in a high throughput mode, you can really take advantage of the one-on-many mode. So, you can create your 384 ligands on the surface, and then come in each time with a different antibody in solution and explore a full 384 by 384 matrix. And I wanted to draw your attention to some literature from 2016.
0:40:08.7 DA: So, we first demonstrated this with a publication, and this publication goes into all the nuances that you can see in high definition when you have this very granular way of looking at how antibodies compete with one another. And in the same year, this is very representative of the throughputs of other label-free instruments such as Bio-layer interferometry based ones; where much, much smaller, antibody panels are being explored here only a 13 x 13. And I'd like to just make the analogy here that you're really looking at black and white TV versus color TV, you just see more definition, see more granularity, the more antibodies you have in your panel. As I mentioned, we have sophisticated an analysis software tool to allow you to navigate through the data that you're generating, and this is an example of a screenshot from our epitope tool.
0:41:11.8 DA: So, once you pull in your data from an experiment, three panels will appear, you're always connected to the primary data in terms of the sensogram, and then you can use that view to set your threshold settings of what you consider to be a block or non-block, and then a heat map will automatically populate and it will sort automatically and network plots or community plots will also populate. And if you click anywhere on any of those panels, the relevant data in the other panel will show. So, in this example on the right, if you click on a cord in the network blocking plot, the relevant cells in the heat map will be highlighted as well as the relevant sensogram. So, you can inspect just visually whether you agree with an assignment or not.
0:42:10.8 DA: And what's really nice about the network blocking plot, is that we have also got a lot of granular detail in there, such that a solid cord would mean that a blocking interaction was observed in both orders of addition, where the ligand blocks the analyte and the analyte blocks the ligand, and a dotted cord shows an asymmetric block as shown here. We have other visualization tools in our epitope software. As I mentioned, the networks, they show the... They represent the most granular of binning, and if you're thinking about a dendrogram plot, this would be at the tip of the branches. But if you wanted to have a coarser, fewer bins, you can actually move the cut height up the dendrogram and show a coarser binning, in terms of community plots. This is an example of a recently published binning that we did to support a study, where it was actually Ligand Pharmaceuticals, were looking at different types of chickens and wanted to compare their epitope outputs, but this was on a 192 array.
0:43:27.6 DA: What's nice about our epitope tool, is that when you create the network plot, you can also pull in data from orthogonal assays and color the plot by those data. So in this example, the binning data were colored by some epitope mapping in terms of sub-domain of protein, they were also colored in terms of library. We had some humanized antibodies versus some wild type chicken antibodies, and also mouth cross-reactivities. This is a really nice way of showing multiple different parameters within the context of the epitope landscape. I'm gonna touch briefly upon the peptide array mapping application. So, if you can imagine, you have a peptide library, you could array that if it was biotinylated, for example, you could array that as 384 array on a streptavidin chip, and then you could use the single flow cell to sample each of your antibodies.
0:44:31.0 DA: And you can do this with more than 384 antibodies on a 384 array, and again, we have a software that will support this application. Here is a snapshot of the software, the data again, would always be linked to all the other data in the other panels, so in addition to the heat map and the dendrograms, we also have a stacked plot version of data representation, so you can see which peptides are mapping to which antibodies and cluster them accordingly. This is a very facile way of grouping antibodies in a very quick assay. Here you can see an example where we took 96 antibodies and injected them over 384 array, and you can see the epitope bins or the mapped epitope groups just falling out of the data.
0:45:30.5 DA: So, now I want to turn to binding kinetics and affinity. So, this example that I'm showing is very typical where you have un-purified antibodies and you're going to capture them or inline purify them using an appropriate capture reagent. So, for this type of experiment, you would use the single flow cell to coat your chip with the appropriate capture reagent. In this example, an anti-human Fc lawn was used, and then you would go to the print head and stamp out your antibodies, and you can use crude superlatives for this, because this is really an inline purification step. Once you've created the array, you go back to the single flow cell and titrate the antigen. And this is very conservative on the sample, so you use very, very little by employing the one-on-many mode. This is what the high throughput kinetics look like.
0:46:32.6 DA: So here, what you're looking at is, I like to think of it as a stamp collection or a tile view, where each one of those tiles represents an antigen being titrated over a single spot of the array and all spots are being analyzed in real-time at the same time. So, it takes the same amount of time to analyze one interaction as it does 384 interactions. And our software enables a really quick batch mode analysis of these types of data. Here poor data had been flagged with gray boxes, this shows you that those spots had either no binding or barely bound. The yellow highlighted panels show very poor fit, unacceptably high chi square values, for example, and the purple panels show binding that didn't have enough binding to determine a good on-rate, in other words, the Rmax that was fit was two-folds above the highest value of the RU that was recorded.
0:47:47.4 DA: So, our software really allows you very quickly to navigate and create the data to flag the good, bad, and the ugly. So, if you didn't have 384 anti-bodies, you had a smaller panel, for example, you could still use the full capacity of the 384 arrays and print antibodies on replicate spots, and this is a really nice way of increasing the N or increasing the statistical confidence of your kinetics. So, here what you're looking at are 16 discrete spots each printed with the same antibody, and then the reported association rate constant, dissociation rate constant and affinity, represent the mean and standard deviation of each of these global fits. This example shows that even in a capture experiment because you are titrating the antigen from very low to very high of the entire array, you can really have a very broad affinity range captured in a single experiment.
0:48:50.8 DA: Here you see a very stable interaction with a picomolar affinity, and then at the bottom, you see a transient interaction with a higher nanomolar affinity. And our kinetic tool also generate iso affinity plots automatically. This is a really nice way of seeing how the kinetics cluster and then picking antibodies, perhaps with the same affinity but reached through different kinetics. We have various QCs that you can apply as well. For example, you can apply the 5% rule to the off rate, such that anything that is below one each, say minus five for example, can just have a cut-off value. So, if you do apply the 384 array concept to a very small panel of antibodies, you can get this really good statistical significance of each of the parameters. This is a really nice example showing that while off rate can correlate with affinity, you're actually losing information about the on rate if you were only to screen clients by off rate.
0:50:05.2 DA: And so, this example really shows you that because all of the clones were analyzed in the same experiment, you can have very high confidence in differences between them, so it really gives you an exquisite discriminating tool to really separate antibodies from one another. So, a little bit about the kinetic software, I mentioned it's very robust and it supports batch mode fitting, and we also have input some QCs as default, and these are user adjustable, and we have the various visualization tools that can really aid in communicating the data with your say non-biosensor colleagues, as the results are really conveniently exported as Excel files. And here's an example of what the kinetic tool looks like when you input your data from a kinetic experiment. It would populate the ligand table, which would represent the array and then it would populate the analyte table, which would represent the antigen that you have to titrated over the array.
0:51:10.2 DA: And then you would see the data on the right-hand side. As I mentioned, it supports batch mode fitting, so just with a simple click of a button, you would be able to apply unique kinetic parameters to each of the spots and create different tile views. Our kinetic program also supports steady-state analysis. So, if you had clones that reached an equilibrium binding response during the allowed association phase, you can choose to apply an isotherm instead. And again, the same rules apply. You can look at it in tile view, overlay view, you can sort the tiles in various ways by affinity, by spot number, by antibody name, and as I mentioned, the epitope tool and the kinetic tool, we like to think of as siblings, so color options, and sorting, and advanced tools are similar across the two programs.
0:52:15.6 DA: So, I hope I've shown you that by allowing you to expand your analysis to look at a much broader panel of antibodies in a single run, you're really able to get kinetics and specificity and epitope information up front, which allows you to have a more informed decision earlier and allows you to prioritize your resources in a more streamlined way. In summary, the LSA really is an integrated premiere antibody characterization platform, really with unparalleled capability that it harnesses array SPR, you have their microfluidics that are really excellent at depositing very low concentrations of antibodies onto the surface of chips, and we have dedicated software to enable you to really navigate the results in a meaningful and quick and easy way. And this really allows you a more comprehensive characterization of all your antibodies. So, I'd now like to proceed to Q and A, so I'm going to pass this back to our host.
0:53:32.5 Speaker 1: We've received a few questions already, but we'll get the rest to you in a moment and your questions in the Q and A box to the left of the slides. Before we begin the Q and A session, first, I'd like to again thank Carterra for sponsoring digital week. With that being said, let's begin the Q and A, Paul and Yasmina, are you ready?
0:53:50.0 DC: Yep.
0:53:50.1 DA: Yeah, ready.
0:53:52.6 S1: Alright. We have a question here for Yasmina, I believe. How long does it take to process and analyze the data?
0:54:03.3 DA: That's a great question, thank you for that. So, if you're talking about the binding kinetics, we have a very sophisticated and powerful analysis tool, and we support batch mode fitting, so just a few minutes of inputting the data, a few clicks of the software, and then you can see the really nice binding kinetic fits just within minutes. So it's pretty facile, and the same goes for the epitope tool. We have two dedicated programs that we like to think of siblings, the epitope tool and the kinetic tool. And with the epitope tool, again, just within a few minutes, you can generate the heat map and those network plots, and then you can groom the data to your satisfaction there after.
0:54:50.5 S1: Great, thank you, Yasmina. Alright, we have another question here, I believe, for Paul. What's the amount of purified protein you use per antibody campaigns?
0:55:04.5 DC: So typically, we use for the chicken antibody campaign, it was about five mgs of purified protein, for the alginate campaign, it was a little bit less for the llamas because some of those immunizations were done with DNA. So typically, what our antibody partners are looking for is five to 10 mgs, of purified protein.
0:55:35.4 S1: Great, thank you. We have another question for Paul. Have you used whole cells to immunize animals?
0:55:50.0 DC: Well, we've used membrane preps. So, not necessarily the whole cells, but the fractionated membranes that contain the target antigens. We've been successful in generating antibodies to membrane proteins doing that, I think that's pretty standard. And it's what's being used by other labs, where they haven't had the opportunity to purify the proteins, but typically now we're pretty much exclusively purified protein, we just get a much better response and we have much more flexibility in developing the antibody screening tools, so we can really mind the immune repertoire with the purified proteins that we make. Though we have done it, but our preference is always to use purified protein.
0:56:54.8 S1: Great, thank you. Alright, we have a question for Yasmina now. What's the auto-sampler capacity of LSA? The second part of the question is, how many ligands and analytes does it support?
0:57:11.8 DA: Okay, so the LSA has five bays, two of them are on the left-hand side of the instrument, and they are for your single flow cell, so that is two bays, each of which could accommodate a 96 or a 384 well plate. Additionally on that side, one of those positions can support various sizes of tubes and Eppendorfs. So, from the analyte side, you have two bays on the ligands side, for the printer, we have three bays, those can accommodate either a 96 or 384 well plates. So, in capture kinetic mode for example, if at full capacity, you could have three 384 well plates, so that would be 1152 ligands per unattended run.
0:58:07.7 S1: Thank you. We have another question here. They didn't specify if it was to Paul or Yasmina, But it is, what are the advantages of your system over Xenopus oocyte stably transfected with a specific ion channel messenger RNA?
0:58:30.6 DC: I think that's probably mine, Yasmina.
0:58:34.1 DA: Absolutely.
0:58:35.9 DC: Alright. Yeah. So, the main advantage is its scalability and manufacturability. And recovery of the recombinant ion channel, though Xenopus, although a great system and shown to generate really high-quality ion channels, it's not a system that you would be able to use to generate the amounts of protein that we purify, to do the drug discovery campaigns. So, that's basically why.
0:59:20.3 S1: Great. Okay, we have another question for Paul. Have you applied the strategy to other membrane proteins, GPCRs, transporters?
0:59:30.9 DC: We have, as most of our focus is on ion channels, we have done other membrane proteins, we've actually done quite a lot of work with vaccine antigens. We haven't, although GPCRs are obviously a very, very important therapeutic class of target proteins, we've done most of our work with the ion channels so far. We're starting to branch out now beyond the ion channels, and really apply what we've learned in generating the large amounts of ion channel in our system to other classes of membrane proteins. So, it's coming, you know, but certainly the ion channels has been the focus up until this point.
1:00:29.7 S1: And we have another question that isn't assigned to a specific presenter. How did the ion channel binding data look, do you see much NSB?
1:00:44.8 DA: I think that's probably for me partly, but also Paul, I'll pass to you after I respond. So, for this particular project, we were lucky enough to have a really nice recombinant source of the ion channel. And maybe Paul can elaborate a little bit more. And so, under the conditions of the experiment, we didn't see any significant nonspecific binding in the binding assays. The ones that we show here are for the binning. And we did that in a premix assay format. And so, we didn't really see any unspecific or nonspecific binding that interfered with the assay. Maybe, Paul, maybe you'd like to comment in the context of other assays where you used this protein.
1:01:30.4 DC: Yeah. So, thanks, Yasmina. So, the only thing I will add to that is, for the assays, the other assays, where we attach the protein to a chip to do SPR based assays. Actually, we... Our first attempts were unsuccessful, because we had the protein in the live design, which is actually probably a good sign because if the thing's oriented correctly, the engineered tags wouldn't be readily available for the attachment chemistry. And so, as Yasmina has said, we started using the purified protein that was just in a mixed micelle detergent and that seemed to work really well for the binning assays and some of the other SPR assays we used.
1:02:34.4 S1: Great. I'm gonna go ahead and close us out. Thank you both for presenting today.