AlphaLISA and HT-SPR together can help you better understand FcRn biology

0:00:02.4 Jen Carlstrom: Hi, my name is Jen Carlstrom. I'm a senior scientist, project manager for PerkinElmer, and I will be giving the first part of the today's talk, Improving Therapeutic Discovery by Leveraging HT-SPR to Complement AlphaLISA. And then Noah Ditto, a technical product manager from Carterra, who is there in person, will be giving the second half of the talk. Next slide. FcRn is known as the neonatal Fc receptor. It got its name because it was first discovered to be responsible for the transport of IgGs from mother to offspring. FcRn binds to the Fc portion of antibodies. Later, it was found that FcRn binds antibodies in acidic endosomes, protecting them from lysosomal degradation. The binding of antibodies to FcRn is highly pH dependent, with binding occurring at pHs less than 6.5. The antibodies are then recycled to the cell surface where the increased pH weakens the interaction with FcRn and allows for antibody release, therefore, how well an antibody binds to FcRn determines the antibody's serum half-life. Next slide.

0:01:24.1 JC: Since the binding to FcRn determines serum half-life, a lot of research has been done to see if modifications to the Fc region of therapeutic antibodies can be used to increase or decrease binding to FcRn. In a review by Liu et al, they described the history of Fc engineering from mapping the FcRn binding sites in 1999 to rationally designing libraries and screening for modifications that modulate FcRn binding. Listed here are a few examples from the literature of Fc modifications within antibodies that can increase the serum half-life of IgGs. Next slide. Techniques that can easily measure binding of an antibody therapeutic to FcRn are highly desirable. FcRn binding assays can be used to measure relative FcRn binding after engineering mutations in an antibody. Additionally, FcRn binding is one of the techniques used to show functional similarity between a biosimilar and the original therapeutic. Using more than one technology to determine the relative potencies of antibodies binding to FcRn can build confidence in the data. Here, we are presenting two orthogonal FcRn binding techniques that can be used to confirm relative potencies of therapeutic antibodies. Next slide.

0:02:57.7 JC: And so for this first part of the talk, I will be focusing on AlphaLISA assays we've developed to detect FcRn binding. Next slide. So what is AlphaLISA? Alpha stands for Amplified Luminescent Proximity Homogenous Assay. It is a bead-based technology. The Alpha beads are hydrogels that minimize non-specific binding and self-aggregation. They're functional groups that can conjugate different biomolecules. The donor beads have a photosensitizer that produces singlet oxygen upon illumination at 680 nanometers. The singlet oxygen can diffuse approximately 200 nanometers in solution, making it ideal for looking at large complexes. The acceptor bead is europium-based. It's excited by the singlet oxygen. It emits light at 615 nanometers. Some of the benefits of AlphaLISA assays are their homogeneous technology, mix and read, which means there's no wash steps; they have high sensitivity because of the amplified signal and a very large assay window; there's broad dynamic range, you can measure high and low analyte concentrations; it's highly versatile, compatible with small and large binding partners; it detects a large range of binding affinities from picomolar to millimolar; and it's automatable. Next slide, please.

0:04:27.5 JC: So how did we develop an FcRn detect binding assay with AlphaLISA? So we used a biotinylated FcRn that bound to streptavidin-coated donor bead, and we used a human IgG conjugated AlphaLISA acceptor bead, so when we mix the beads and the FcRn together, we get high signal. Then in the presence of your antibody test sample, the antibody, your antibody would bind to the FcRn and compete off the IgG conjugated acceptor bead resulting in a decrease in signal. From this, you can measure IC50 values of your test sample. The protocol is very easy. It's a 40 microliter reaction where you add 10 microliters of your test sample, 10 microliters of biotinylated FcRn, and 20 microliters of a mixture of the donor and acceptor beads. Then the mixture is incubated for 90 minutes at room temperature in the dark because our donor beads are light sensitive. Then the reaction is read on an EnVision alpha plate reader. Next slide, please. So here we validated this FcRn binding assay using a variety of different human IgG isotypes. We tested both mixture of IgGs, as well as the individual isotypes as shown on the left. We saw a nice IC50 curves, and the values were all within two-fold of each other, which was consistent with literature references. On the right, we also tested four different therapeutic antibodies, adalimumab, etanercept, pembrolizumab, and trastuzumab, and, again, all four of these had within two-fold differences in potencies. Next slide.

0:06:27.4 JC: So you'd say can the AlphaLISA assay actually distinguish between different potencies since we saw all the same potencies? So we validated this by looking at a oxidized therapeutic. So oxidation of methionine 252 and 428 in the Fc portion of IgGs has been shown to decrease binding to FcRn. So we incubated adalimumab with 0.3% hydrogen peroxide for one hour, three hours or overnight. And as you can see, the FcRn binding assay is able to detect a shift in binding potency after oxidation of the Fc portion of the antibody. Next slide. So in summary, AlphaLISA assays are suitable for screening for modifications of the Fc portion of a therapeutic antibody for potency shifts in binding FcRn or comparing a biosimilar to a therapeutic drug. The speed and high sensitivity of the AlphaLISA FcRn kit and it's noteworthy signal to background robustness enables easy assessment of precious IgG samples. Using more than one technology to determine relative potencies of antibodies binding to FcRn can build confidence in the data. So with that, I will hand it over to Noah, who's gonna tell you about a second orthogonal technique for measuring binding to FcRn.

0:07:56.0 Noah Ditto: So my name's Noah Ditto, I'm Technical Product Manager at Carterra, and I'm gonna talk a little bit about assays that we ran on our end to corroborate and dig a little deeper into the results found in the AlphaLISA assay. Again with Jen's point she made about being orthogonal in our approaches, our technology is surface plasmon resonance-based, so we're measuring real-time binding interactions of non-labeled partners. And so it's a slightly different format than AlphaLISA and a great way to kind of understand results from two different platforms and ensure that we're getting a result that's meaningful and can make decisions, drive decisions. So Carterra's technology is the LSA, it's a high throughput surface plasmon resonance device. What it effectively does is we array 384 binding species on the surface of our sensing chip, and we can introduce a binding partner in solution across that array. So what you get out of this is a tremendous throughput gain over historical biosensors that are out there. In a typical assay format, in about a day, maybe even a little less, you can screen 1152 affinities, complete affinities with a well-defined KA and KD.

0:09:08.1 ND: You can also go ahead and do something like 150,000 competitive interactions in a single experiment to do things like epitope characterization. So the throughput is massive, the data coming off the platform is massive, and it really enables some types of assays that really just weren't explored on these types of systems historically. Shown here in the snapshot is a very probably difficult-to-see plot of different sensograms in the upper right hand corner. This is 384 sensogram, this would be just maybe a six-hour run, something like that, so huge wealth of data. And in the bottom half is one of our publications highlighting a heat map and network plot showing epitope relationships of antibodies. So really, again just the power of the system is in the throughput and the ability to drill down and generate lots of data. And I think that was really what drove us to put this technology with the AlphaLISA in comparing this assay, 'cause while we run the assay and generate data, we have the ability to also kind of look at variables in the assay simultaneously given the amount of space or real estate on our biosensing chip.

0:10:17.6 ND: So we went ahead in the assay doing that. And in the first step here, you can kinda see the step number one is we took the antibodies as well as four drugs which Jen just highlighted and I'll show in subsequent slides along with controls, Fc fragment, Fab fragment and all these done in triplicate and at three different concentrations will translate into three different densities array that on our sensing chip via immune coupling. So you've got 96 discrete locations on your sensing chip, and this is only a quarter of the capacity of the system, but we're getting a huge amount of data, again, from one small set of experiments. So first step was arraying that sensor chip with the 96 samples that I just mentioned, then we moved on to the actual steps of binding. So we went ahead in the first kind of step and did a titration of FcRn, so this is bound molecules on the surface being probed for binding against a non-labeled FcRn molecule. So we're strictly looking at binding as a function of mass at the surface. And that titration was done, and then we moved on to a second experiment where we took the same FcRn at fixed concentration and then progressively introduced injections with increasing amounts of drug to determine potency or IC50s from those interactions and develop response curves.

0:11:38.6 ND: So the outputs really are three different things we did with the data. One, we did a kinetic fitting of the data to get on and off rates. We also did a steady state fit as well, and that's another way to kind of understand affinity and kind of check your work, especially for interactions where there's faster rate constants, sometimes it's good to verify that the kinetic fits aren't perturbed by the speed of the interactions, so we did a steady state analysis on the data as well. And then for the portion where we were looking for inhibition, we went ahead and did an IC50 inhibition curve plot, and I'll describe that more in the next slide. So FcRn steady state and kinetic affinities. One of the first things that jumped out at us was these molecules are very similar, and that's kind of what the AlphaLISA had just suggested to us, so they're very similar. Interestingly in our assay that the surface densities didn't have a large impact on the reported affinities.

0:12:33.2 ND: And in the top half, you have kinetic affinity, so this is a measure of affinity determined from the rate constance. Bottom half chart is a steady state affinity. Those response fit to a steady state model. And you can see that they're in excellent agreement, we really have very similar results from the two models, as we kind of would expect. So we don't see any issues here with one, the surface densities used, we can kind of compare a higher density medium in the lower density surface, we don't see effects of that. And even within that, you can see across the molecules that they were quite similar, the exception being etanercept, which generally had a slightly weaker affinity in this particular format than the other molecules. Etanercept, the lowest concentration, we didn't have sufficient signal really with that molecule to show here, so we're only showing the two higher density surfaces. But yeah, in the end, a huge amount of data, triplicate measures on everything, and then the two different approaches to characterizing affinity, both were in really excellent agreement.

0:13:36.9 ND: We went ahead then also in the other experiment and looked at IC50 potencies. So these are plots of those curves. In the assay we had four different sub-classes on the surface, so we could take each drug and look at whether or not there was a sub-class difference in competition across the molecules. I'm just showing here a subset of the data, which is the 5 microgram per ml concentration, but we did have those other data sets that didn't show, again, an effective surface density influencing the outcomes of the IC50 calculations. Yeah, but in a nutshell, the four drugs did show high comparability in all the conditions tested, consistent, again, with the AlphaLISA. We did notice interestingly that etanercept does appear to be modestly potent, I'll show the values in the next slide for that. Which was kind of curious, 'cause we did see a slightly weaker estimated affinity. So that's kind of a curious behavior, but again, everything we're looking at here is within two-fold. So, interesting, but not probably enough to say that these are markedly different molecules in terms of the FcRn potency.

0:14:44.7 ND: And then looking at these values, so this is the same data I just showed you, but actually plotting the actual IC50 values. So I should make a point that the IC50s, we report in the LSA are kind of in a different dynamic range 'cause the instrument functions in a different dynamic range than AlphaLISA does. So again, these are relative comparisons and mostly comparisons within themselves of molecules across the board. So we can see that, for example, adalimumab, very consistent around an 800 nanomolar IC50 in this particular assay format. Even across the sub-classes, there's only a modest difference among them. Trastuzumab, again, very similar overall. Pembrolizumab was curious in that the IG2 actually was slightly less potent. And we kinda saw that in the data, if you happen to catch that, where it was, in fact, kind of had a different profile that curved into. So that was an interesting behavior as well, so there's some subtle differences. But again, back to that main point that frankly, the data is remaining in two-fold. So we have both this level of assurance that these molecules are similar, but we do see subtle differences here, which are interesting, suggesting that there's some dissimilarities among them, but minor at best.

0:16:06.4 ND: So really, from these assays, the key takeaways were, we went ahead and we want to understand, does AlphaLISA compare with HT-SPR in terms of the outcomes? Are we seeing similar results? The short answer to that is, yes, we do see similar outcomes as we would expect, and this is consistent with the literature. But the great part about this is that these are really different types of assays that we're exploring here. AlphaLISA with labeled reagents and components versus HT-SPR, which is unlabeled and really just your bare binding partners directly interacting. And the great part about the LSA setup was that we had so many variables we can incorporate into the test. So we only used a quarter of, frankly the capacity of the system to test this, but we were able to test multiple surface densities, concentrations, IgG sub-classes. I didn't even have time here today to really get into the controls listed, but lots of other conditions you can build into assays to better inform the outcomes.

0:17:05.8 ND: And we'd like to think of this really is when you run assays, because we have such tremendous real estate on the surface where we can introduce so many species, you can both optimize the assay and get your final data on the same go. You really run the assay and find that optimal condition at the end, and that's the data that you progress with. So yeah, there was slight differences that we detected by SPR in this particular format amongst the different drugs tested, but frankly, they didn't rise to the level of suggesting there were marked differences between these molecules. And again, that's consistent with AlphaLISA in literature. So yeah, we really like to just emphasize that this particular set of studies corroborates AlphaLISA and importantly highlights that using orthogonal technologies is valuable in order to really understand your biomolecules. In this case, FcRn/IgG binding, because we can run what effectively is one assay and take three different measures of the data to understand the interactions there.

0:18:04.9 ND: So, with that, I'll wrap things up and just if there's any questions after the talk, feel free to shoot me an email and we'd love to talk to you. And with that, I'll go ahead and a big thanks to all our PerkinElmer folks who helped support this, both the assays as well as just the execution here. And then Carterra's folks as well, for enabling all this to come to fruition. And I thank you for your time.

[applause]