A high throughput assay for simultaneously measuring the concentration and binding of IgG, IgM, and IgA antibodies to the Receptor Binding Domain (RBD) of SARS-CoV-2 is now available.
This new label-free assay provides, for the first time, detailed resolution of the immune response of patients recovering from COVID-19 in a single assay.
A peer-reviewed paper describing this assay is pending publication in Nature Communications and a preprint of this groundbreaking research can be accessed on Research Square.
Posted by Chris Silva
- So you will see later that in Leon's presentation actually that the printhead, which is typically used to generate ligand array in early discovery, as shown here, can be also used for the characterization of sample, and in the case of the collaboration with University of Twente, we use it to characterize serum sample. So a maximum of 384 serum sample has been characterized in one single experiment.
0:04:18.6 JP: So now I jump into the assays themselves, so not the assays that Leon will present later, but the kinetic assay to measure real-time binding kinetics of typically antibodies and antigen. In this exercise, up to 1152 antibodies can be screened in a single experiment using three 384 well plates of antibodies. So the antibodies themselves are captured on the surface via the Fc part, then we come with the antigen titration to measure the kinetic rate constants. So the advantage of this assay is that it's highly parallel. So we're measuring in a single injection the interaction of the antigen with 384 antibodies simultaneously.
0:05:02.7 JP: So there's no disparity of changes of the quality of the antigen over time. Also, the array capacity are low if there's no 384 unique clones to build in replicate, which is until now, not performing SPR at this stage of the project. So in early discovery. So the LSA change is the data quality you can get early on and the same array capacity can be leveraged to titrate the ligand on the surface, or trying different surface densities within a single experiment, then choosing the best data automatically, ultimately for kinetic analysis.
0:05:43.1 JP: One more time, it's something that's never been done historically, as most biosensor require a day or two of assay optimization, then running the actual experiment. So optimization, in the case of LSA, is a one and done process. And really, that what we are chasing in the end, high throughput kinetics. So here are shown 384 ligand kinetics and within each individual tile is eight analyte concentration titration. So all is done in a single run and the picture is actually a screenshot of our software and the software is able to flag data that may be borderline. So in gray are shown data with a low signal, in yellow, data with a complex signal. So for, to generate a 384 dataset shown here, it takes about six hours, so it is quite fast to obtain 384 affinities measurements.
0:06:57.9 JP: And I skimmed over it previously but there is an ability to really replicate to ensure the assay and we found a really good reproducibility in the system. So in this case, I'm shown... On this slide, we're looking at 12 replicates, the same antibody or the same ligand and we're looking at the mean of rate constant calculated from these 12 replicate and the standard deviation from this replicate as well, giving a high confidence in determined kinetic parameters. So we have good agreement around the data just obtained from the, generating from these 12 replicates, showing that you can read, replicate and get statistical confidence in your assay.
0:07:44.3 JP: And to build upon this, there is always a question if we have good reproducibility within the assay and how is this data translate outside this individual assay? So from an accuracy perspective, we can say they agree very well with the Biacore. For example, what established platforms that most people just reference when looking at level 3 binding. So there's been an excellent paper put out last year by folks at Adimab and Amgen, looking at benchmarking different SPR platform. And the LSA was right in step of what is gold standard SPR, the Biacore, showing a really good agreement with individual rate constant and affinities. So here are shown the affinity of 36 clones and the difference is that the LSA is obviously faster and require much less sample, and can finish these types of assay in a day versus multiple day and run to be achieved on other platform. So only, you have speed, but also you have confidence in the data you are generated. And I will add that in this assay, we measure the 36 clone in replicates since we have the capacity to do so.
0:09:01.4 JP: With that, I shift to epitope binning, which is the other main application with kinetic on the LSA. So epitope binning is a pairwise exercise, where we take antibody and compete them with each other to determine if they can form a three molecular complex with the antigen. And the goal of this is to group antibodies is what is called bins of community. So it's very, very advantageous to perform epitope binning in early discovery, in order to identify diversity into set of sample you characterize and to maintain that diversity going forward as candidate progress in the therapeutic discovery, 'cause there is so much information from this data that changes the mindset of drug discovery, where historically, it's kind of funnel process, where we try to get rid of the high number of sample to have a number of sample that is manageable. And now, we are looking at a sampling strategy using epitope details. So a nice thing about epitope binning as well is it allow to have the test about sandwiching pair, and that sound very well for diagnostic and also assay development.
0:10:25.5 JP: So the main user requirement for epitope binning like for kinetic in early discovery is low sample concentration and throughput that's completely transformative. As I mentioned previously, the LSA can airway 384 ligand on the surface of the chip. And then in a typical binning experiment, we will bring an injection of the antigen in each cycle, then an injection of one of this 384 antibody. So for each injection, we measure competition again 384 species. And in total, we generate 147,000 interaction measured in a single experiment, it's a huge number of data point and details about the candidates.
0:11:06.2 JP: Great things about it, obviously, is a low antibody requirement of concentration that fit very well with very early discovery, where the amount of material is pretty limited. And even the antigen level are very low, considering we are handling a 200 microgram of antigen used for an epitope binning in its largest format of 384 x 384. And at backend of it, we have our proprietary software package, which is very interesting, integrating visualization tools that really allows you to take this huge dataset and make meaningful inference of them. That's something that's tangible for an investigator to understand.
0:11:58.3 JP: So the other exciting things in the epitope binning software is the ability to pull in all the antibody attributes onto our proprietary network plot. What I show here is from a publication of Pasture and we can see different community network for antibodies and each antibody is represented. I note on this plot and a line between antibody, are blocking relationship, so indicating just antibody overlap. So we can see, we have very discrete communities, we do not have any communication with each other.
0:12:42.5 JP: So what we can also do with the data in the software is bringing piece of information that describes this antibody, but still in the context of epitope. So for example, library source, expression level, affinity, cross reactivity. There's no really, no limit how you can layer data on the top of this network plot. So it is a hugely powerful tool, that for not only to visualize the epitope landscape but also to visualize attributes of this antibody that are meaningful for the selection criteria.
Carterra in the fight against COVID-19
0:13:22.0 JP: So that was my introduction to the LSA system, its major application, epitope binning and kinetics. I'd like to switch gear a little bit and before to hand back over to Leon, I would like to quickly present the status of Carterra in the fight against COVID-19. So we have a lot of customers that leverage LSA to meet the needs of COVID-19 research, and they are listed here on the right of this slide. So there are academic and public institutions that are involved in this fight, but we have also industry customer like Eli Lilly or Twist. Additionally, the LSA has been selected by the Gate Foundation that put together an antibody consortium screening arrangement, where the device will be a key platform to perform epitope binning and kinetic energy to select potential therapeutics.
0:14:12.1 JP: So that's really encouraged us at Carterra that we can contribute to this effort and be part of the solution. And in order actually just to accelerate the different discovery workflow, we develop and now offer biosensor pre-functionalized with the receptor-binding domain of the spike protein, as its chip will carry the RBD protein on its surface, and it is a really solution to characterize anti-RBD antibodies.
0:14:49.9 JP: So now the last thing I would like to touch is a kind of an expansion of the LSA application, which is a serum-based assay. So I don't want to say much about this assay, but Leon will, but this assay we highlight as LSA platform is not only limited to discovery, through the selection of therapeutics, but could be potentially used for medical diagnostics. And now I hand back over to Leon, who is going to go ahead and show you nice looking data generating during the development of assay to monitor human response to infectious agent. Thank you.
0:15:28.2 Leon Terstappen: Welcome to this webinar. I am Professor Leon Terstappen, chairholder of the Medical Cell BioPhysics group of the University of Twente. I will present this webinar for Carterra on behalf of our COVID-19 team. At the time, the Netherlands went into a COVID-19 lockdown earlier this year, research henceforth from our group moved our SPR equipment to the local hospital to start the development of an assay to profile the immunity for COVID-19. Here, we will present the results.
0:16:04.1 LT: Here, we show the COVID-19 virus. The two important immunogenic proteins are the nucleocapsid protein and the spike protein. The spike protein contains the important receptor-binding domain or RBD, and on the bottom, you'll see that the SARS-CoV that appeared in 2002 and the SARS-CoV-2 that appeared in 2019 are quite similar.
0:16:32.9 LT: Here you can see the amino acids sequence of the spike protein of the coronavirus. It contains the S1 and the S2 subunits. In the S1 subunit, the receptor-binding domain RBD is located. The RBD binds with the antigen, turns and convert an enzyme 2 or ACE2 of cells. It is important that the virus is blocked by neutralizing antibodies to prevent binding to the ACE2. To protect and cure COVID-19, an active immunization with a vaccine that generates neutralizing antibodies is needed, as well as a passive immunization strategy to diseased COVID-19 patients, to treat with anti-COVID-19 antibodies.
0:17:19.5 LT: For COVID-19 immunity profiling, it is important to measure the total immune response of the patient. The patient generates the isotypes IgM, IgA and IgG, and it's essential to find out how the polyclonal response bind to the various regions, such as the nucleocapsid, the S1, the S2 and the RB2. The affinity of the antibodies for their target, likely is important for an adequate response to the COVID-19 infection and might provide important insights for the severity of the disease of COVID-19 infected patients. The hypothesis is, do we get long-lasting, higher antibody response and strong antibody affinity, or a lower antibody response and weak antibody binding?
0:18:23.0 LT: How did we perform patient sample spotting on the SPRi sensor surface? The sample of corona patients are flown over the sensor surface with the corona protein immobilized on it, only the COVID-19 specific IgM, IgG and IgA antibodies will bind into the surface. The sensogram shown shows an example of the IgM, IgG and IgA antibody detection by injecting respectively anti-IgM in the beginning, followed by anti-IgG, followed by anti-IgA. And you can see clearly the sequence of responses. So the principle of detection of the IgM and IgG and IgA concentrations of COVID-19 patients is shown on this slide.
0:19:20.0 LT: So first, we start with the sensor prism is coupled with the receptor-binding domain of the SARS-CoV-2 to the sensor prism. And the sera of 96 patients is spotted to the sensor for 15 minutes. Then we place the sensor, the spotted sensor in the SPRi imager, then we inject affinity purified anti-human IgM followed by anti-IgG and IgA. Each interaction is five minutes and the total analysis time is 26 minutes, and then we calculate the R-values for each patient as shown in these pictures.
0:20:04.0 JM: This slide shows the tile view of over all 96 serial plots, so to illustrate the difference observed responses of the various cases of IgM, IgG and IgA we highlight a couple of examples. So for all patients, we can measure this isotope concept patients simultaneously. So here you see no response, then you see slight response, and then a little bit more. So this patient is completely different so you don't see something in the beginning, but then a very rapid response. Then this one is again difference, here you see again difference, and here you see a little difference. So tou already can see here is that the responses of the patients are quiet different and you can actually cluster them in different groups of responses, and of course, this measuring in time will become very important. Here we plot for all 96 patients, the IgG, IgA and IgM, and the total antibody response, we have however also want to measure the affinity when offline printed issues, then we do not know how much was bounce during the printing process. This is needed to measure the on and off rate of the antibodies of the patients, which we feel could be of important.
0:21:34.1 JM: So how did we perform affinity profile? Importantly, we can follow the binding process in Carterra's LSA. It turned out that the on rate of the mixture of isotopes cannot actually be measured due to the enormous mixture and concentration of polyclonal antibodies, however, the total response can be measured in the LSR for an accurate off rate determination. And you see here the on-rate and off-rate of the sets. So to perform this experiment Richard Schasfoort travelled to Munchen where they had an LSR machine and here he is standing next to the LSA. And on the screen, the IgM and IgG and IgA measured, performed on the LSR is shown. So how to get the data for the strengths of binding? So during the association process, we measure the response R in resonance units. In the dissociation phase, we measure the slope the DRUs, and then the off-rate constants can be calculated simply by dividing the slope with R. We applied an additional trick by adding the RBD to the dissociation buffer to prevent rebinding and getting a pure off-rate for all sera at once.
0:23:06.9 JM: By zeroing the transition between the association and the dissociation phase, the slope of dissociating molecules can be measured, the weaker the binding the faster the off-rate. So if there is maturation in the immune response then the slope for longitudinal samples will go up. The response in the association phase is already measured and the LSR generates the off-rate constants KDs. Then we plot the off-rate as a function of the days after symptoms onset for all the patients, we observe a relative large mixtures of all kinds of antibody responses, but clearly we see maturation effects that the off-rate is going down. So the strengths of binding is increasing during the development of the disease. This was an important conclusion which is shown in our pre-printed paper available at the Research Square pre-printed platform. And of course, you can already see here that we can measure in time and see clear differences in the immune response. What it means, of course, will have to be determined in a bit more patients and more longitudinal studies. So, in conclusion, we developed a rapid test for profiling COVID-19 immunity for 96 times four patients simultaneously.
0:24:39.3 JM: We can measure three ISO types, the IgM, IgG and IgA and the off-rates of the anti-RBD antibodies for strengths of binding. Maturation of the IgG response is shown in convalescence patients and research is progressing. With that, I want to conclude. So, the results I've shown you was initiated by Richard Schasfoort at the University of Twente, who got together members of the University of Twente, Saxion, Carterra, VySens and MST/Medlon which is basically the hospital and the pathology department to form our Twente COVID-I9 team. I hope this was informative and Richard Schasfoort of the University of Twente can answer any questions. Thank you.
0:25:31.1 JM: Thank you, Leon and Judicaël. We'll now start the Q&A discussion. Please submit your questions with the box on the right side of the screen. There's a link to the pre-print of the paper we presented on today at the bottom of this slide.
0:25:43.9 Speaker 4: Judicaël can crude samples be used on the LSA?
0:25:52.9 JP: Thanks John. My short response would be, yes, they can use on the LSA actually, LSA has been developed just to bring SPR in early discovery, so we don't expect... Or our customer just bring some verified sample, and actually we have a number of customer using the LSA for primary screening, using, for example, bacterial or periplasmic extract to screen single channel fragment, and some are using the MTV5 tag to capture this single SCRV. It's a kind of ideal set-up because tv5 tag has a affinity for the tag.
0:26:40.2 JP: And the LSA is able also just to cycle the sample just during the capture of of the SCRV or just any ligand on the surface for a very long time. So even if you are just a low limited amount of extract or low expressing clone, by increasing the contact time between your sample and the MTV5 surface, you can easily just capture and reach the surface with this ligand and then get just a full kinetic characterization of your clone. And actually the same is true for hybridoma and just recently did some screening campaign for one of our customer using... For COVID for... Using hybridoma and we also use some culture supernatant so, there's no no limit in using a good sample with a with LSA, we just recommend... Since we have a microfidic system, so we recommend just to or to filtrate the sample or just simply just centrifuge them before just to apply to them on the device.
0:28:07.2 JM: Thank you, Judicaël. We have a question about binning. What is the total amount of antigen required per clone and what is the maximum binding affinity, that can be measured with the LSA?
0:28:21.4 JP: So to answer first the second question, what's the max bind... The binding affinity is the maximum, so we are basically in the same range of affinity of the other platforms so, I'm not just give you a KD. But in term of rate of association, we just can measure up to 10 to the eight and in term on KR, we are 10 to the minus five to 0.5. That should give you just an idea in term of just a kinetic parameter we can measure on the device. Now for epitome binning, it will depend on the affinity of your antibody, but if you're just thinking running classical sandwich binning in a large format, so 284 by 284 just to make some calculations. Just standard condition would be just... So in such a classical sandwich binning, we inject for each cycle, we inject one concentration of one injection of antigen, which is followed by one injection of the 284 antibody you would like to characterize so meaning you have about 284 injection of antigen, plus of course some reference injection so, let's say 400 injection of antigen, and typically, we work at antigen concentration of one to 200 nanomolar solution. So, you can calculate just the 400 injection with 200 nanomolar of antigen will give you an idea of what the conscription of antigen you will need, but I can say where we've seen some... We just perform a large binning with some antigen just staying under 200 microgram of antigen.
0:30:25.8 JM: Thank You Judicaël. Richard, what is the off-rate of samples in the absence of RBD?
0:30:37.7 Richard: Yeah, that's a good question because that's the trick we applied for better off-rate yeah determination. So perhaps it's even be better if we could optimize the RBD concentration, but a nice thing of measuring 384 off-rates only you spoil I think 200 microliters I think Judicaël for...
0:31:15.0 JP: Yeah.
0:31:15.3 Richard: Having this off-rate, but the concentration of the RBD should be optimized for the real 100% rebinding to present rebinding event but yeah, it is definitely a way of doing it, and sometimes it can be 25-30% has an effect on the off-rate. It depends also definitely on the ligand density if the ligand sort of patient spotting on the RBD surface, if it is highly dense then the RBD is not... The rebinding effect is less affected by the RPD that's ligand density effect.
0:32:16.4 JP: And just to add to Richard the presence of RBD is necessary since we measure the dissociation of species which are multivalent, and we got the IgM which is dicavalent. We got the IgA which tetravalent, IgG are ambivalent so just to avoid, to have a huge rebinding effect and to be able just to discriminate the chaos and the different samples the presence of our RBD is necessary in the... During the dissociation of the sample from the surface.
0:32:56.6 Richard: Yeah, definitely, you will get best a parameter if you add RBD, and the... Yeah. The whole assay is strengths of binding is based on the off-rate, there is also... We found a good correlation but then we tested patient by patient in different concentration, etcetera, etcetera. That there is an... And then also you don't know the concentration but then you reference the concentration etcetera. But we found a good correlation between the affinity constant and the off-rate, so now we can say strength of binding is correlated to the affinity or avidity, what you can say because it's consistent. A lot of antibodies all together who are binding to the full RBD, the IgM, IgA, in a mixture spot, so that is really there. But we think the strengths of binding is now very good, and a nice thing is that when you measure off-rate you're not depending on the concentration because the concentration of your buffer is zero, you know the response on your spot, so it is concentration in there, how much it was captured, we observe then the off-rate. And that's the nice thing, because in an on-rate, yeah, you need to add concentration, but what is the concentration? And which one you should take? IgG, IgM or whatever? So it's quite difficult. So this is an affinity parameter that's very effective to be measured and also with this trick of RBD we can measure this simply.
0:35:04.6 JP: And actually just to... For the calculation of difference, we don't need just to have the K... We don't need to have the KD or the KM. I think the K-off is more simpleton parameter here, just defining the different strength of the binding of each serum. As well as after isotipization which is allowed to determine the quantity of each species in the serum. Yeah.
0:35:41.2 Richard: Yeah, I agree.
0:35:46.4 JM: Here's a question about detection anti IgM, G and A antibodies causing crosslinking affecting the affinity measurements?
0:36:01.9 Richard: No, we are not measuring in that phase, we're not measuring the affinity of the binding of anti IgM and then measure the off-rate, etcetera, no. This is a polyclonal injection of anti IgM and anti IgG and anti IGA, it's only for getting the ratios of concentrations, what's bound from the serum, we measure the off-rate earlier, directly after binding. So we are not measuring in these first anti IgM binding event.
0:36:55.6 JM: Thank you Richard. Judicaël.
0:37:00.6 JP: Yeah.
0:37:01.8 JM: Can you perform kinetics and epitope binning assays on the same chip?
0:37:12.4 JP: Yes, yes, but there is bit of nuance, how you set up your assays, so particularly when you want to measure kinetic in SPR you'll actually have just to optimize density of ligand on the surface just to avoid mass trans polymerization or binding effects, during the association phase and the dissociation phase, while when you're running epitope binning, you're not concerned by this artifact which is mass trans polymerization, what you want to see in epitope binning, is the binary assays when you see additional signal, or no additional signal. But actually the two assays can be... Could be just combined on the same surface. And we have different costumer just binding just to running now on using our new navigator software we just installed recently by our costumer just to allowing to queue different samples... Queue different assays in the same... Two different method in the same assay, running... So first the kinetic on the surface and then the epitope binning on the same chip if you just have the same immobilization solution. So in principle, yes you can just do both assay on the same surface, and... Which is great because you can have not only the affinity of the antibody for the antigen, but also where it does bind on the surface of the antigen, so there's plenty of information in two queued assays and just to put all your sample on your device and you can walk away.
0:39:10.0 JM: Thank you, Judicaël. We have one more question for Richard. Did you compare SPR with ELISA, lateral flow assays or other techniques?
0:39:25.9 Richard: Yes we compared isotype measurement as a concentration test, where SPR brought the RBD antigen, the ELISA that is one antigen, and a Roche platform, the COBAS, they applied to nucleocapsid response. And as you can see in the pre-print on the research grant platform we show the data and found some correlation with S1 and showed a quite scattered correlation with nucleocapsid measured by Roche. So first people said, your SPR instruments, etcetera, is not the same as Roche so yeah it's not a good... I said, "No, the data of SPR is okay." So, this is... Was really remarkable that we could see such different levels that the protein part of the virus generates. And later we injected the serum of the single patient over various proteins so typical multiplex test, and we observed also a scattered level in SPR, so against S1, S2, RBD, nucleocapsid, a variation per patient different stages of disease, etcetera.
0:41:02.3 Richard: And I'm really excited, how is this... How this response in COVID-19 refer? Why does a patient generate such a scattered immune response to COVID-19? And I think more subsidy for universities is necessary to find out this for COVID-19. So, I think many years of research is needed before getting really the clinical relevant SPR test. And I think also academia and institutes and industry in Europe should cooperate, perhaps also in the US, on an international level to solve the many scientific questions because there are so many things you can ask for, what really is the case? Is this patient? Etcetera, and why is binding is... In relation to the angiotensin receptor binding etcetera? And I am daydreaming about a breakthrough with... Where SPR as unique technology can generate the clinical data, so it tells something about the status of the patient. And then definitely we need then in for the future future.
0:42:34.3 Richard: And my institute is really optimistic, but as you may know, not any clinical test is based yet on SPR, so it costs years for getting such instrument or whatever. And the LSA, in my opinion, as a very early discovery tool is needed for this kind of research and I hope also there's many, many more researchers all over the world will step into this patient immunity testing and see... Cooperating, finding epitope binding, etcetera. And yeah, this is my answer, and yeah. I thank also Carterra for getting me the opportunity to do this. I also am very... We will miss Leon, Leon did this presentation on the data, etcetera. But I'm... And we will see each other, I think, I hope in the future. Thank you.
0:43:52.1 JP: And I also share the excitement of Richard is bringing SPR in medical diagnostics at the moment where we use it more for early discovery, and just to give an idea, of course, when you go to diagnostics, you think you need just speed and to have some assay, which is pretty cheap. So just to give you an idea how long it takes just to analyze the 284 sera on the LSA it'd take between three and three and half hours, and I just to... Just to make some calculation how much would it cost to... The analysis of one single serum sample, and we just under $2, just per serum. So we just perfectly in line what... With the cost of diagnostics.
0:44:45.9 Richard: Yeah, I totally agree, Judicaël.
0:44:54.1 JM: Well, thank you so much, Leon and Richard from the University of Twente and Judicaël from Carterra. And especially want to thank you, the audience, for attending today's webinar. If you have any additional questions, please reach out to Richard or Judicaël with their email addresses listed on the screen. Thank you all for attending. I hope you have a great day.