Presentation at Antibody Engineering & Therapeutics Europe 2023 by Judicaël Parisot, PhD

The Carterra LSA platform is the leading platform for high-throughput kinetics and epitope analysis (HT-SPR technology) of mAbs and protein therapeutics. The new LSAXT platform has enhanced sensitivity, further extending the application space of HT-SPR to include molecule classes such as kinase inhibitors, PROTAC®S, and transient interactions like Fc Receptors.  Selection of bi and tri-specific binders often requires the screening of large combinatorial sample sets derived from panels of monospecific binders.  The Carterra LSA and LSAXT make analyzing the binding properties of these combinatorial panels straightforward and requires minimal amounts of mAb sample and antigen. The affinity of the binders to the targets can be measured in several assay formats and used to verify the independence and activity of each binding site for hundreds of clones. Along with the binding kinetics and specificity measurements, the LSA and LSAXT enable large scale epitope binning to ensure diverse sets of clones are being carried forward to functional evaluation.

Posted by Judicaël Parisot, PhD

0:00:05.4 Speaker 1: So thank you very much for giving me the possibility to present the Carterra's technology. During my presentation, I will address the high-throughput SPR for the characterization of multi-specific antibodies, and in particularly, I will present our new platform, which is LSA-XT. So here is the outline for my presentations. I will just first talk about the need of SPR or high-throughput SPR in early discovery and how it can actually accelerate early discovery. I will present our technology. Then I will present the high-throughput epitope binning and its role in bispecific antibody discovery. So this part of my talk will be based on a paper published by the National Institute for Allergy and Infectious Diseases, and they work on the, actually, just on generation of bispecific and biparatopic anti-COVID antibody. And then I will just present a different SA format we developed on our platform just to characterize this multi-specific construct.

0:01:19.1 S1: So as we know, the speed of biotherapeutic discovery is constantly increasing. We have access of different platforms, which include B-cell screening, hybridoma, or we hear a lot of, at the moment, from AI-ML. And all these technologies allow to generate a lot of different binders, again, a target. And all these methods also can be enhanced by the use of high-throughput SPR. And not only for kinetic, because usually, just like traditionally, when you hear about SPR, we just think about kinetic, but epitope binning. We developed a high-throughput platform, SPR, and we just can bring now epitope binning in early discovery. So it's not only about how tight are my binders bind to the target, but what's the epitopic landscape of this panel of antibody has. And that's very nice information where you want to actually select some monospecific antibody to just produce multi-specific one, for example, again, COVID. And finally, as industry just go, as new tool, adopt new tool, just develop new workflow. So the time of development dropping, we need this kind of high-throughput platform.

0:02:41.7 S1: So talking about bispecific and multi-specific antibody, we hear about bispecific antibody for a long time, but actually, there's only just nine of them on the market. And five have been just approved in 2022. So we are just probably at the very beginning of exponential growth for this molecular technology. There's more than 100 of them, actually, in clinic and pre-clinic. So for the characterization of this multi-specific antibody, we need additional step. We just first select the monospecific one. We create a multi-specific construct, and then it needs to be tested and validated. So we test it, of course, the ability still to bind the target has been just engineered to. But also, we need just to assess if this molecule is able just to engage both targets simultaneously. We have just a bunch of SA for that.

0:03:41.9 S1: So talking about the Carterra SPR technology, we are developing array-based SPR technologies where we have two instruments on the market, the LSA and the LSA-XT. We put on the market very recently. And I will just show you what's the difference between these two devices. It comes with a software package, one dedicated to run the SA, so the navigator, and two dedicated to the analysis of kinetic data and epitope-binning data. We have also a line of chips and also some regions to run this data, to run these SAs. So here, describe this machine has been, this device has been built around a proprietary microfluidic system. Historically, Carterra is a microfluidic company. And this microfluidic system is divided in two. We got the multi-channel or 96-channel printed and the single flow cell. So the printed, we refer to it as a printed. When docked on the surface of biosensor, just form 96 little flow cell. And we use this little flow cell to depose, underflow your ligands. It can be antibody, it can capture them, or just coherently bound on the surface. But it does it 96 at a time.

0:05:00.1 S1: And this printed is able to address four different nested locations. So we're able just to produce a 384 array with this device. And what's pretty cool about the printed, it does not just like throw you a sample away. Actually, when the capture's been done, it just bring by your sample into the plate. So you don't lose your sample. You can just like do single SA for, for example, cross-reactivity or check a counter target, but you don't lose them. So this array has been generated, the printed just undocked from the surface and comes as a single flow cell, which is a large flow cell, which is docked over this array of molecule. You can imagine a big channel with a 384 molecule. And we inject our protein of interest. So the antigen, the target. And the system is monitoring really the binding of this target to 300 up to 384 ligands simultaneously.

0:06:02.8 S1: So here are the core application. When we talk about SPR, as mentioned, we talk about kinetic and affinity. So it's just a machine just done for screening antibody, but also epitope binning. And there, I just mentioned that this machine bring epitope binning in early discoveries as we can do epitope binning in a format 384 by 384. So we can use now epitope binning as a selection tool. It can be totally integrated in selection process of antibody. It's very nice in early discovery, really just to have the epitopic landscape of the panel of antibody you were characterizing.

0:06:39.8 S1: Then the third application is mapping. So it can be peptide mapping or mutant mapping. And finally, we got a quantitation. So as mentioned, we have two different instruments on the market. So both instrument has exactly the same microfluidic system, has the same throughput. So you can just do some kinetic, of course, epitope binning with purified or non-purified sample. And what we did just while developing the LSA-XT, just improve the optics. So now we still just bring, so we keep the same throughput, but this machine, this device can also use a small molecule as analyte. And I'm gonna show you an example quite soon. And when we just think about small molecule, we think about fast kinetics. So this machine is also just able just to analyze quick kinetics. So quick on rate and quick off rate.

0:07:43.6 S1: And the differences in time of metrics, so this one, compared to the LSA, has a better signal to noise. So it allows us to work with a small molecule, a better signal uniformity across the chip, because it's a pretty large chip. And also, we increase the data collection rate. So we can just really work with the fast kinetic. So to illustrate what I say about the sensitivity of the device, I know it's just antibodies conference, but here, what we did, we immobilized six different kinases on the surface. So it's kinases that just immobilize and duplicate, and we titrate the staunch point, so which is under 500 daltons. So now we just keeping the throughput of the SPR, but we bring this machine in the field of small molecule.

0:08:33.9 S1: And as mentioned, fast kinetic, when you think fast kinetic, it's not only for small molecule, but Fc gamma receptors. That's not our work. That's the work of our colleague of DragonFlight. And what they did, they just immobilized a panel of Fc gamma receptor on the surface of our chip. So they did it because we have a throughput of 384, so immobilized eight replicates of each Fc gamma receptor, Sino, and human. They could have also changed the densities, they didn't do it, but it could have been good to do that if I had to actually just improve this SA. But they just injected this antibody. So you have one single experiment, you have also the characterization of binding a few antibody against a bunch of Fc gamma receptor, which is a fast kinetic as well for some of them.

0:09:26.6 S1: So talking about high-throughput epitope binning, high-throughput epitope binning is just mentioned. I mean, the epitope is very important for antibody. We know that it's linked to their functionality, to their mode of action. And really understanding the competition profile of a panel of antibody, a set of antibody in the context of wanting just to do some cocktail or some multi-specific molecule is very, very important. And here, so how we do this epitope binning on machine, we can immobilize up to 384 antibody on surface of the chip. And we use a classical sandwich setup or premix setup to run this SA. So we immobilize our antibody on the chip, we inject one concentration of the antigen, and then we inject the rest of, I mean, one by one, but the rest of the different antibody. And everything, every signal in the green zone are non-competitive antibody. So they're not competing with the one on the spot. And the one in the red zone, or pinkish red zone, are the one which is competing with the one immobilized on surface.

0:10:40.7 S1: And we got just this visualization panel just to present the results. So we know the heat map. But also, we have this network plot, which is kind of a Facebook of antibody, which antibody just is connected to which antibody. And usually, just antibody under the same envelope are the one which is competing for the same binding site. So how is this epitope binning are used for the generation of multi-specific or bispecific antibody? In this situation, they're not only bispecific, they're also biparatopic. That's the work of the group of Josh Tan at the NIH. And what it is, they work, of course, like a lot of people on the anti-COVID antibody. And they isolated B-cell from patients. And they tested the B-cell against the potency affinity, but also epitopic diversity.

0:11:29.9 S1: And they ran two large binning, the first one against RBD, and the second one against the N-terminal domain. And for each epitope binning, so the antibody just fell in four different non-overlapping binding site. Then they selected antibody from the non-overlapping epitope bins. And they tested the cocktail of antibody. And they wanted to test to find not only a cocktail with a better potency, but also for mutation resistance. So they tested the cocktail of antibody against a different variant of the virus. But what they found is just while we have a better resistance to mutation, we don't have a better potency. We have just additive effect and not synergetic effect. So they decided just like, let's try to generate some bispecific antibody or biparatopic antibody.

0:12:34.5 S1: And here, so they based their format on the DVD-Ig format. And what is, first, they have to demonstrate the bispecificity or the dual specificity of this molecule. So what they did, here is an example. So what they did, they ran some sandwich SA, competitive sandwich SA. So as an example, we took two monospecific antibody, which is a parental of this bispecific. And they immobilized this, so 5, 4, 3 on the surface, and 6, 6, 4 on the surface. Inject the RBD and inject, actually, all the bispecific antibody. So the antibody itself to demonstrate is able to compete with itself. And the other one, so the monovalent one, is not competing, of course, it's just targeting another bin.

0:13:23.4 S1: So all the different bispecific or biparatopic antibody just show us an additional signal, meaning they are also just able to sandwich. So they are bispecific. Then they tested this bispecific or biparatopic antibody for their potency and also for the resistance to mutation. And one of them, which is made, actually, which has a dual specificity N-terminal and RBD, had a potency which is 100 times higher than the cocktail of its constituent. So here, we just lack not only just resistance to mutation, but also we got a better potency. So why we have a better potency, they studied that or they investigated that using electron microscopy. And as I just heard earlier also, so this bispecific antibody were able to cross-link as well different spike protein. So we had a better potency through this cross-linking. This kind of mode of action, actually, is not available for normal antibody or cocktail of antibodies.

0:14:34.8 S1: So finally, how do we just analyze this antibody and this bispecific antibody in early discovery? When you reformat your monospecific antibody in a bispecific construct, first you want to just test if both specificity of both side of the antibody is able just still to target the two different antigens. So like for the monospecific one, we immobilize antibody on surface, capture them on surface, titrate antigen one, titrate antigen two. So that works pretty nicely. And here, what we did, and to give you how long does it take, actually, just to run such a measurement on our device. So it just took us about 48 hours for full kinetic, about three 384 well plates against two antigen. And we ran such an SA for some customers. So they identified also 12 sequences against antigen A, 12 sequences against antigen B. And they just developed just 144 antibodies, and they first characterized them against antigen A, antigen B.

0:15:47.3 S1: So how we do it, we first capture all this 144 antibody on the surface of the chip. So it's one single SA, and we titrate antigen A. So that takes about eight hours. We do the same for antigen B. We just take the same sample, because the device does not just throw your sample away. We recapture your bispecific antibody and titrate antigen B. So you have that in 16 hours. You've got all your set of bispecific antibody characterized. Finally, what we have to demonstrate as well, as I mentioned before, is just the ability of this bispecific construct to engage both targets simultaneously. In another just word, we want just to show the independence of target binding. And for that, we just have to do just an SA, but a bit different from the previous one. You're still capturing antibody on the surface, blocking one binding site with one antigen, and titrating the second one. And you can just, that's just here, you capture visually both SA, just done in presence and in absence of the second antigen.

0:16:56.4 S1: And the last SA, also just to characterize also the ability of the antibody just to engage both targets simultaneously, is a bridging SA where we have just one antigen captured on the chip, injecting the antibody, so up to 384 antibody on the surface of the chip, and titrating the second antigen.

0:17:25.6 S1: So this bring me to the end of my presentation. I hope I convinced you that high-throughput epitope binning and high-throughput kinetics just add significant value for the characterization, not only of bispecific antibody actually, but also monospecific antibody. And in particular, you're just convinced that high-throughput binning, just giving information about the epitopic landscape of a set of antibodies, very, very invaluable information. Bispecific antibody just allow you to just have access to, in this case, additional mode of action. In such a case, it's able just to, in the case of a spike protein, it's able to cross-link spike protein, just having just better potency.

0:18:07.6 S1: And finally, we develop, really, just on our device, our platform, a set of different SA for the characterization of this multi-specific format. So we're always interested, and we are technology developers, so we're always interested just in collaborating on different projects, so just approaching us. We have a booth, booth 13. And I would like just to thank now my colleague, which is Rebecca Rich, Noah Ditto, Daniel Bedinger, and Tim Germann, which is here in the room, and Josh Tan for his work. And thank you for your attention.


0:18:53.1 Speaker 2: Any questions from the audience?

0:19:00.5 Speaker 3: I've got one actually. I can probably say it very loudly without using the microphone.


0:19:03.9 S3: Okay. That was me told. So just to be clear, so for your 4 times 96 format, so you can spot different analytes onto each individual spot. The user can do that themselves via the microfluidics, so you buy the chips that are already done. I just wasn't quite clear on that. Maybe it's just...

0:19:24.3 S1: We have just constant chip, so if we just talk about immobilizing antibody, usually we just capture them through the [0:19:34.1] ____ FRAB or through the Fc. So yes, we can have some chip with a full surface with cover with anti-Fc antibody, for example. And we spot each sample on a discrete location. So imagine.

0:19:46.9 S3: So you do that when you buy the chips that are pre-coated, or people coat their own plates? And that's what I'm not quite clear on.

0:19:53.8 S1: You can just coat your own chip. So you can just stereotype your own chip.

0:19:58.3 S3: We have got a question over there.

0:20:08.7 Ali Asli: Thank you. Thank you. Ali Asli from Lonza. I want to ask about if we have sample that it's not 100% bispecific, and you put it in this device, how can effect of the affinity and the KDs, and if you have any solution about if it can bind one antigen, but the second, we are not sure if this, the same sample can bind the two antigens or not. So if you have any solution for that. Thank you.

0:20:43.1 S1: So for non-fully bispecific sample, we use a bridging SA. So knowing how much antibody is capturing on surface, we expect to have saturation with the second antibody at certain level. And you will notice that you're not 100% bispecific.

0:21:08.5 S3: Any other questions?

0:21:10.5 S1: But it does not never happen. It never happened.

0:21:14.8 S3: Any other questions? So with the newer model, the XT, you say it's more sensitive for things like weak affinity interactions like Fc-gamma receptors, for example. So it's the optics that are more sensitive, or what's giving you that extra sensitivity? And what's the limit of the kind of affinities you can look at with the first machine?

0:21:40.6 S1: For the first machine, just really the first machine has been developed for protein-protein interaction. So everything, just like you think about antibody binding, from micromolar to, I would say not picomolar, because usually biosensor systems are not done just to characterize picomolar. But now we can go, I mean, with small molecule, we can use a 10 to 100 micromolar in lower affinity. In higher affinity, lower affinity, it stay in picomolar. The problem of characterizing high affinity on biosensor system, you need very, very long dissociation time, so stable signal. So in general, across the board, biosensor systems are not done for high affinities. That's why also, Andrew just presented characterization of double-digit picomolar binder as a user KinExA system.

0:22:35.8 S3: Cool. Brilliant. If there are no more questions, I think, as you said, you're on Booth 13.

0:22:43.6 S1: 13, yes.

0:22:45.1 S3: If there's more questions. I think we have got... There is another question. My work is never done.

0:22:57.0 Speaker 5: Just a question about a couple of practicalities. So with high-throughput, I didn't really catch what the system does in terms of auto-sampler. So how many analytes and ligands can be stored in the auto-sampler as a temperature control? And the other part of the question is about the output. So obviously, you do the analysis in the proprietary Carterra software, but how easy is that to export into a LIMS system or to analyze offline?

0:23:21.5 S1: Sorry, I didn't get your first question.

0:23:24.7 S5: It's about the auto-sampler capability. So is there an auto-sampler where you can put analytes and ligands, and is the temperature controlled, etcetera?

0:23:32.6 S1: Actually, just the system can just accommodate about 1,152. So three well plates, but 96 and 384 well plates. Of course, we just have always in mind just to improve this range of sampler.