Spotlight: Single B Cell Discovery Essentials

The Biologics Brief: Spotlight | Single B Cell Discovery: Platforms, Trade-offs, and How to Actually Choose

Single B cell discovery has become one of the most powerful tools in antibody drug discovery, but with so many platforms on the market, how do you know which one is right for your campaign?

In this special Spotlight episode of The Biologics Brief, Mosaic Biosciences CSO Tracey Mullen sits down with Director of Antibody Discovery Christina Palmer to break down everything you need to know about single B cell discovery platforms, without the vendor spin.

What we cover:

• What single B cell discovery is and why adoption has surged in recent years
• The core trade-off between resolution and throughput, and why theoretical throughput numbers can be misleading
• How to compare the major platform categories: high-content optofluidic screening (Beacon), microfluidic droplet systems, sequencing-first workflows, and FACS-based platforms
• Why genotype-phenotype linkage matters more than most people realize
• How AI is changing the way antibody candidates are prioritized
• A practical 5-question framework for choosing the right single B cell workflow for your program

Whether you're a biotech team evaluating platforms for the first time or a seasoned drug discovery scientist looking to pressure-test your current approach, this episode gives you the tools to make a more informed decision.

About The Biologics Brief
The Biologics Brief is a podcast from Mosaic Biosciences, a biologics CRO specializing in antibody discovery and engineering. Each episode features conversations with scientists and industry experts on the topics that matter most in biologics drug development.

🎧 Listen now and subscribe wherever you get your podcasts.

mosaicbio.com

Transcript

Welcome And Today’s Roadmap

Tracey Mullen 0:04

Welcome back to the Biologic Sbrief from Mosaic Biosciences. I'm Tracy Mullen, Chief Strategy Officer at Mosaic. Today we're going to be talking about single B cell antibody discovery, which has become a very important way to access immune repertoires directly and recover native heavy and light chain pairs without relying on hybridoma fusion or display library construction. There are a lot of different workflows in this space, high-content optophic platforms like the Beacon, microphylitic systems, microcapillary and microwell-based systems, droplet and sequence-first approaches, even fax-based workflows. And almost every platform provider will tell you that their approach is the right answer. But the honest answer is that there's no single platform that's best for every campaign. And so today we're going to talk through how to actually think about these technologies. And we'll answer some questions like what exactly is single B cell discovery? How does it compare to hybridoma and display-based discovery? Where are different classes of platforms shine, where they fall short, and ultimately how to choose the right workflow based on the biology of the target, the goals of the campaign, and of course your organization's downstream capacity. I'm joined today by Christina Palmer, Mosaic's Director of Antibody Discovery and Optimization. Christina is one of our internal experts in antibody discovery strategy, including in vitro discovery, library design, screening selection, and of course optimization. Christina, thanks for joining me.

Christina Palmer 1:24

Thanks for having me, Tracy. It's such a treat to be here with you today. All right, so let's start with the basics. So

What Single B Cell Discovery Means

Tracey Mullen 1:30

we're going to start by talking about what exactly is single B cell discovery. When people talk about single B cell antibody discovery, they're generally talking about direct interrogation of antibody-producing B cells from an immunized animal or from a donor rather than going through something like a hybridoma fusion or building a display library. And I'd say, you know, the major appeal here is that you can recover native heavy light chain pairs from individual B cells, which means that you're actually capturing antibodies much closer to how they exist in the immune repertoire. So, Christina, from your perspective, you know, how do you explain the value of single B cell discovery to someone who might be, you know, more used to thinking about phage display or other in vitro display approaches?

Christina Palmer 2:10

Yeah, that's a great question. I think the biggest conceptual difference is that single B cell discovery starts from an immune response. You have an animal or donor that has already done a lot of the biological work for you. You have B cells that have gone through not just antigen exposure, but clonal expansion, somatic hypermutation, and affinity maturation. So you're starting at a very different point than you are with a synthetic or a need library and applying selection pressures in vitro. You're sampling from a repertoire that has already been shaped through an in vivo process. And that can be really powerful, especially when native heavy and light pairing is really important or when you want antibodies that reflect an actual immune response rather than what, you know, survives a panning or display selection. You know, I'd say on the in vitro side, I obviously have a love for display platforms and I see display as incredibly valuable, but it's a very different tool than you know a single B cell platform. Display platforms give you a lot of control about what comes out and a scale that is really unparalleled. You can screen through really large libraries, you can build really large libraries and tune them however you like. You can, you know, really pressure the system to select for certain criteria, be that affinity, specificity, cross-reactivity, developability, you know, whatever it is that's important to your project. But single B cell discovery really gives you a more direct window into an actual immune repertoire. And I think that's really cool.

Tracey Mullen 3:39

Yeah, yeah. And that distinction is important, right? I mean, I think single B cell discovery, it's not just like faster hybridoma or, you know, another another display format, right? It's a different way of accessing the biology. And I think also in the context of in vivo discovery, the immunization strategy becomes really important. The quality of the single B cell output really depends heavily on what happened upstream. You know, the animal model, the antigen format, the immunization schedule, the tissue you collect, the timing of harvest, of course, the B cell population you choose to interrogate. All these things are really important when it comes to single B cell discovery. And I think actually that last point maybe matters more than people sometimes realize. You know, some workflows, we'll talk about this today, are focused primarily on antibody secreting plasma cells or plasma blasts. Others are interrogating antigen-specific memory B cells. You know, others might even be looking more broadly at just repertoire-level paired heavy light chain sequencing. And, you know, these aren't necessarily interchangeable populations.

Christina Palmer 4:37

Yeah, I think that's something that, you know, I've learned in my recent times working with you and working to set up our internal platforms is really appreciating those differences and the, you know, those different populations can tell you different things. And so you have, and please correct me if I get any of this wrong, but you have the plasma cells and the plasma blasts that are actively secreting antibody, which makes them very attractive for secretion based screens, but they may represent a really different slice of the immune response than something like a memory B cell. But memory B cells can be lower frequency and not actively secreting. So they might be harder to isolate and capture. So when someone says like it a platform is a single B cell platform, there's still so much that you have to ask about that, including what B cells are you actually screening for?

Tracey Mullen 5:24

Exactly. Yeah, yeah, the platform matters, but the upstream immune biology matters just as much as the platform that you're using downstream. And also, you know, the downstream workflow matters too. Recovering a B cell or a sequence is just not the same as having a validated antibody. You still need to recover the paired heavy light chain sequence, clone or synthesize it, express it recombinantly, confirm binding, confirm that specificity, test function, you know, evaluate developability. So the platform itself, it can move you faster to sequences, but you know, the downstream burden doesn't necessarily disappear.

Christina Palmer 5:58

Yeah. And I think that's one of the themes we'll probably keep coming back to is that you know, single B cell platforms are not like a magic box where you just put in your repertoire and get out your antibody. They can reduce some bottlenecks, but they can also kind of create or expose other bottlenecks. You know, you can have your bottleneck upstream at the hyperbotoma fusion and clonal expansion stage, or you can push it downstream to the aspects of the workflow like sequence recovery or common expression, validation, and triaging large numbers of candidates. So it's really just a shift of where you place that burden.

Tracey Mullen 6:33

Definitely. And certainly depends on the platform too. And we're going to get into that and more today during today's discussion.

Why These Workflows Took Off

Tracey Mullen 6:40

So maybe we'll start at the beginning and talk a little bit about why single B cell discovery really took off recently within the antibody discovery space. Because, you know, single B cell discovery, it's been around conceptually for a pretty long time. But it really just became more visible, I think, more recently as platforms like the Beacon came into the market, made it possible to start to interrogate these B cells in sort of a more controlled, very assay-rich environment. And I'd say now the category is just it's broader, it's growing, and it's it's exciting to see. You know, we we have optophilic platforms, of course, but we also have microphilitic systems, microcapillary arrays, micro well, micro RAF systems. You see droplet approaches, uh, sequencing first workflows, and, you know, of course, even fax-based sorting. So the list goes on, right? I mean, why do you think this area has gained so much attention?

Christina Palmer 7:31

Yeah, I think it's a great point. And I think there are a couple of reasons. You know, I've been in the antibody discovery field for quite some time, but really seen a lot of exploding platforms in the last five years or so. And I think there are a couple of reasons. So, you know, the first being speed, right? Hybridoma, tried and true, can be very powerful, but does take time. The fusion process, cloning, outgrowth, screening, self-cloning, like we we know the steps. It is a labor-intensive process. So single B cell workflows can shorten that path from immune response to antibody sequence and some more than others. You know, the second thing is diversity. You are really able to capture a much larger swath of potentially interesting diversity compared to a hybridoma, where you know, the fusion process and outgrowth can bias what you recover. If you're a really interesting B cell, does it fuse well or does it grow well? You may not detect it. And so single B cell approaches can help to capture a broader snapshot of the repertoire, depending, of course, on the platform. The third thing that I think single B cell platforms are really strong for is really maintaining that native pairing, depending on the platform. You know, in display platforms, sometimes you are not necessarily maintaining a natural heavy and light chain pairing. You're often looking at just random combinations of heavy and light chains, and that can be suboptimal. In a single B cell workflow, you're really working hard to preserve that pairing, which can be really important for some antibodies in terms of affinity, specificity, you know, biophysical behavior. And the fourth thing I've seen a lot of growth in is the assay resolution, right? Depending on the platform, and I'm sure we'll talk through this, some B cell platforms can allow you to ask more questions at the screening stage beyond just can you bind antigen A? Can you also not bind antigen B or C? Can you block ligand? Can you bind to protein expressed on a cell surface? You know, and the list goes on. So the quality and depth of data you can collect can actually be incredibly powerful and is a real advantage over other platforms.

Tracey Mullen 9:42

Yeah, I think to your point, we will be talking a lot more about this. But I think that last point is really where the platform differences become really important and a bit more obvious. I think it's one thing to say, you know, we can recover it into specific B cells. It's a whole different ballgame to say, you know, we can identify ultra-rare B cells that are secreting antibodies with the right functional profile against a difficult cell surface target. And those aren't just fundamentally not the same problem. Absolutely.

Christina Palmer 10:10

You know, from an in vitro discovery perspective, this is where I get interested in how single B cell discovery complements display, because that's the holy grail, right? Being able to put function as early in your screening funnel can be incredibly powerful. You know, a display platform is excellent when you can define your selection pressure well, you want to have a large library coverage, but a single B cell platform can be really compelling when, you know, the immune system is doing something useful that you don't want to lose. For example, you know, specifically if you have a confirmational target, a membrane-associated target, a multimeric target, or something that's really difficult to recapitulate in a display system, allowing the animal to take on that burden of work and then adding that to a single B cell recovery platform can be a really powerful approach.

Tracey Mullen 11:01

Yeah. And, you know, I think of single B cell as especially valuable when the biology of the target benefits from an in vivo context. But you know, you don't want to pay the full time or in some cases attrition costs associated with hybridoma workflows. So it really gives you access to the immune repertoire with kind of a more direct path to sequence.

Comparing Hybridoma, Display, Single B Cell

Tracey Mullen 11:21

And I guess, you know, while we're on the topic, maybe we should just talk a little bit about kind of comparing major discovery modalities. You know, let's talk about single B cell versus hyperdoma versus display. Because I think if we simple oversimplify the antibody discovery options, you know, we have hyperdoma discovery, we have display-based discovery, and we have single B cell discovery. And all three can really generate good antibodies, but you know, they're all good at different things. So I guess, Christina, maybe to start, when would you still want an in vitro display campaign?

Christina Palmer 11:52

Yeah, no, I am excited for this discussion because while display is close to my heart, I fully take a platform agnostic approach when I'm looking at a target and evaluating like what is going to be the best platform for that project. And so, you know, display is still a very strong answer when you want scale, right? When you want to screen a large library, you know what you're after in terms of maybe epitope or domain or something along those lines that you can control and you wanna be able to tune your selections. So, you know, if you have a well-behaved antigen and you want to screen a large sequence space, display is super efficient. You can enrich binders over multiple rounds, you can change the stringency, you know, select for cross-reactivity or species reactivity, you can include negative selections to try to eliminate things that you're not after and really push that campaign toward just those specific binders that you want. You know, it's really useful when you want to work from a defined library architecture or when an immunization isn't ideal. Maybe the target is toxic or highly conserved, and so you don't imagine you're gonna get a great immune response. Or maybe you, you know, start with a human framework diversity from the start. You don't want to have to go through any sort of humanization process. You know, those are still and continue to be strong advantages of display. And you really get to define the parameters of your library and your selections very intentionally.

Tracey Mullen 13:17

Yeah. Yeah. So I guess if I were to massively oversimplify that, display gives you control over the selection environment, if you will. Absolutely.

Christina Palmer 13:25

You know, but that comes with trade-offs, right? The selection happens in a tube, most likely. And so, you know, you may enrich for clones that behave well in display, but when you make those proteins as soluble antibodies and you're looking at a cell-based context, the binding is different or something about that is the system is different. You don't necessarily maintain a native heavy and light pairing. You've maybe just kind of put things together, sort of like building blocks. And you typically do need a lot of downstream screening when you're running a display platform, you know, whether that's to enrich for binders that are actually functionable, uh functional or looking for developability, or, you know, really testing things in a biological context. Yeah. Display platforms are really powerful, but they tend to benefit from more of an iterative process where you can learn and then feed that back into the platform.

Tracey Mullen 14:17

Yeah, exactly. This is gonna be, I think, one of like our key takeaways from this entire broadcast today is you know, it's it's it's wrong to think about this as single B cell is better than hybridoma or single B cell is better than display. The right question really to ask here is, you know, what's the target? What does success look like? What kind of antibody do we actually need? And where are the biggest risks in the campaign so you can make a more educated decision? I think, you know, if the risk is that the immune response will be weak and rare, and you need a workflow that gives you enough repertoire coverage, you know, if the risk is that many binders will be non-functional, you're gonna need screening resolution, you know, if if the risk is that downstream validation is limited, you might need to front load that triage so you don't send hundreds or thousands of low-value sequences downstream. So there's just there's a bunch of real trade-offs there. And I think that that's a great segue and kind of what I'd love to chat through next with you, which is what are the trade-offs? And so I promise we'll get into the platform landscape, but I'd love to just quickly touch on you know the attributes that people kind of look for when they look at different platforms in the single B cell space, because I think that that really ultimately answers the question of what platform should I work with.

Trade-Offs: Resolution, Throughput, Linkage

Tracey Mullen 15:31

So, you know, most single B cell platforms, they're attractive because they're fast. I think that's true for pretty much everything. It's and it's also just a shared value prop within the industry, right? So compared to traditional hybridoma workflows, you can move from immunized animal or donor donor sample or whatever it might be to that paired heavy and light chain sequence much more quickly. But you know, once you get past speed, I think when it comes to the different platforms out there, the real trade-off becomes resolution versus throughput. So, you know, how many cells can you screen? How deeply can you interrogate each cell? How many characteristics can you measure at the same time? Can you can you look only at energy and binding, or can you also bring specificity, blocking, cell binding, maybe even function forward into the initial screen? Uh and importantly, can you do all that while maintaining genotype-phenotype linkage so that we can ultimately connect that functional or binding phenotype to the actual paired heavy and light chain sequence that produced it?

Christina Palmer 16:30

Yeah. And that phenotype linkage piece is critical, right? It's one of the reasons single B cell workflows are so attractive in the first place. You know, similar for a display platform, you're trying to connect an individual B cell to the antibody it produces, to the phenotype you observe, and then the sequence you recover. But not every workflow preserves that equally well. Some methods are very good at preserving the native heavy and light pairing. Others may generate a lot of sequence data, but depending on how the downstream sequencing is done, you can lose confidence in the native heavy and light pairing or lose that really nice and important connection between phenotype and exact A to body sequence.

Tracey Mullen 17:07

Right. Yeah, exactly. And I think this is where people sometimes tend to focus too much on theoretical platform throughput as the most important aspect when you know evaluating different platforms out there. Because I think, you know, a platform may say that it can screen a certain number of cells, wells, droplets, chambers, whatever it might be. But it's important to keep in mind that the realized throughput can actually be a fraction of that, depending on how the cells are loaded. And all of that ultimately tiles back to how many sequences you're ultimately recovered and whether or not you deeply understand that genotype, phenotype linkage. So, you know, unless you're placing cells one by one, I think we all know, you know, we're dealing with just statistics, right? With limiting dilution or random loading, you're you're going to have empty wells. You're going to have empty droplets or chambers or, you know, you might even have doublets, multiplets, right? So the number of physical compartments is not the same as the number of single cells that you're actually screening. Right.

Christina Palmer 18:06

That poissant distribution really matters. If you load too lightly, you might get more single cells, but then you have a lot of empty wells. If you load too heavily, you fill more compartments, but you increase the frequency of those doublets. And doublets can be a real problem because then you compromise the interpretation of the phenotype as well as sequence recovery.

Tracey Mullen 18:26

Right. Yeah. So I mean, so when we're comparing platforms, we're not just asking, you know, what's the theoretical max throughput? We also want to ask, you know, what's the realized single cell throughput at the occupancy rate required to preserve that clean genotype, phenotype linkage that is, you know, inherent and core to the single B cell workflow? And I think that's a much more meaningful question when looking at these different platforms.

Christina Palmer 18:48

Yeah, absolutely. And that really connects then to the downstream burden, right? If your platform screens fewer true single cells than advertised or a meaningful fraction of the hits are ambiguous because of doublets or mixed compartments, the downstream validation burden increases.

Tracey Mullen 19:05

Exactly. Yep. And the other important point, too, that I think is worth discussing is functional screening. You know, in theory, you can imagine adapting many fluorescent-space readout assays to a single cell platform. And in fact, you'll probably hear that a lot as well. You know, if the assay produces a fluorescent signal, it can be moved upstream into a single B cell screen. But I, you know, I'd say in practice, many functional assays are more complicated than that. They, you know, they may require multiple wash steps, sequential addition of reagents, incubation periods, media changes, you know, reporter cells, target cells, whatever it might be. And so if the platform can't exchange buffer or can't wash efficiency, then you're largely limited to some homogeneous, no-wash functional assays. That list is quite a bit shorter, right? And while they can be powerful, you know, they're not universal. Yeah.

Christina Palmer 19:55

No, and this has been my experience too, that this is non-trivial. And this is where platform architecture really matters. If the biology can be captured in a simple homogeneous fluorescent assay, then many platforms are probably able to support it. But if the functional readout requires, like you said, like wash steps, sequential reagent addition, or more complex self-handling, then only certain platforms are going to be compatible with that. And that really matters when the functional cone is rare, right? If you've gone to the effort to screen function upstream, you really want to be able to do that in a clean way. And so if you expect many binders and many functional antibodies, maybe it's fine to do binding first discovery and push function downstream. But if you're looking for that rare functional antibody and your downstream functional assay is low throughput, then you want to bring that function earlier in the screen. And that can make a huge difference to success.

Tracey Mullen 20:47

Yeah, yeah, that's definitely the central trade-off. Yeah. If functional hits are common, you can afford to recover those binders first and test function later. But you know, if it's difficult, functional phenotype is rare. You might actually need a platform that lets you enrich for function up front. And that's really where screening resolution, you know, starts to become worth paying attention to and worth paying for too. Okay, so let's get into the meat of our discussion today.

Platform Tour From Beacon To FACS

Tracey Mullen 21:12

Platform discussion. And ultimately, what I'd like to land on, hopefully, is that there's really no single tool that that wins out there. So I'll spoil the conclusion. But so, you know, with that, with that framework in mind, speed, realized throughput, resolution, genotype, phenotype linkage. But let's talk about the platform landscape. So there's several broad categories. I think the first is high content optophilitic screening with the beacon as the best known example. Then there are microphilitic and sort of compartmentalized secretion based platforms that kind of try to balance assay depth with broader throughput. There are microcapillary, micro well, sort of micro RAF systems that offer different combinations of secretion, binding, recovery, trade off with cost there. And then there's droplets. And sequence-burst workflows that really emphasize repertoire coverage, impaired light chain, heavy chain recovery at scale. And then, of course, there's, you know, tried and true old school facts-based single-cell sorting workflows, which are certainly the oldest and simplest version of this idea, but still valuable in their own right. So the truth is that I think all of these platforms, they solve different problems. And those differences only really become obvious when you ask what trade-off you're willing to make.

Christina Palmer 22:28

Yeah. And the problem you're trying to solve should really drive platform choice, not just what you think is the coolest or you know, most interesting platform.

Tracey Mullen 22:38

Or loudest. Yeah. So let's uh let's start with the beacon. So the beacon, I think most people kind of view this as the most standard platform for single-B cell discovery nowadays. It's for those of you that don't know, it's an optophilitic platform. It isolates cells in nanopens. The strength is not just in cell isolation, though. It's actually that you can run fairly sophisticated assays on chips. So you can look at secretion, binding, multiplex binding. You can do blocking screens, you can do functional or cell-based readouts. And you know, from my perspective, I think the beacon is most compelling when you have a high-value, very difficult target where screening resolution matters most as compared to things like raw throughput.

Christina Palmer 23:17

Yeah, no, that's definitely how I think about it too. The beacon is likely not the cheapest or highest throughput way to recover antigen-specific B cells. But if your campaign requires you to distinguish between binders, blockers, cross-reactive clones, nonspecific clones, et cetera, then the ability to do more of that screening upstream can be really valuable. It's, you know, it's a way of front-loading biological resolution. Right. Yeah, exactly.

Tracey Mullen 23:43

And you know, you may look, you may look at the cost and think this is expensive. But the question is like, compared to what, right? If it prevents you from expressing and validating hundreds of low-value clones, or if it helps you recover that rare functional clone in a lower resolution workflow that it would miss otherwise, right? Then I think that might be worth it. But of course, you know, if the target's simple, all you need is the low-cost antigen-specific B cell recovery, the beacon might be overkill.

Christina Palmer 24:07

Yeah. And I think that's the key point, right? That resolution is valuable, but only if you actually need it, right? If you can just isolate binders and get what you need, you probably don't need that investment. Exactly. Yeah.

Tracey Mullen 24:19

And so then there's there's other platforms out there, right? I mean, the next category I think we should be worth talking about, put on the spectrum here is microphylitic or sort of compartmentalized secretion-based technology screens. You know, these platforms are they're trying to create information-rich single-cell workflows without necessarily using the same architecture as the beacon, right? So they aim to preserve cell level information, detect secreted antibody, of course, and enable binding and in some cases functional interrogation in a controlled micro-scale environment. And I'd say, you know, from my perspective, these are interesting because they offer more assay context than a simple sort, while potentially giving more flexibility or throughput than some of the high content optophluid workflows out there.

Christina Palmer 25:03

Yeah, yeah. And I think that is the right way to talk about those platforms, right? The question is not whether the technology is interesting. Many of them are really interesting, and I've enjoyed learning about them. The question is really whether the platform gives you the right type of information at the right stage of your campaign. If a platform gives you more assay content than a simple sort, but more scale or flexibility than a very high content lower throughput platform, that could potentially be a great fit. But we still need to understand the fundamentals, right? What cell population is being screened, what asset, what the assay actually measures, how strong that genotech phenotype linkage is, how clean the signal is, and what the paired recovery rate is. As well as, you know, you want to think about how reproducible that binding data is when you have recombinant protein downstream.

Tracey Mullen 25:51

Yeah, exactly. Yeah, I think I think this category shows that the field is not binary, right? It's not just, it's not just speaking versus fax or high content versus low throughput. Like there is a spectrum here. You know, there's also the microcapillary, micro well, sort of micro-raft-based approaches. These these can offer different combinations of secretion screening, you know, binding readout, cell recovery, imaging, cost, you know, they kind of sit somewhere in the middle depending on the specific architecture. So, you know, for example, some of these platforms, and they could be attractive if you want maybe moderate throughput secretion plus antigen binding, and if the sequence recovery and recombinant validation data will support that, right? But you know, I would want to understand the assay timing, the signal quality, the actual, the actual realized single-cell occupancy, because as we talked about earlier, that could vary. You know, the sequence recovery rate. And as you said, you know, whether the platform can support on-cell or functional assays in a way that's actually meaningful.

Christina Palmer 26:46

Yeah. And I would want to know how much downstream work it creates, right? If the platform gives us a manageable number of high confidence hits, I think that's sort of ideal, right? But if it gives us a lot of ambiguous positives that then require a lot of follow-up, the efficiency sort of may disappear once you get downstream.

Tracey Mullen 27:04

Exactly. Yeah, yeah. And then last but not least, or maybe I think a couple more that I listed there, but let's talk about the droplet-based, you know, the sequencing first approaches. I think, you know, these workflows they are really enticing because they give massive repertoire coverage. You can profile huge numbers of cells, or you know, you get back paired heavy and light chain sequences. That's very powerful if your goal is to understand, you know, the entire immune repertoire or mind super rare events. But if you generate a huge sequence space without functional triage, there's a trade-off there, right? You're going to need very strong downstream expression and validation. You're really just pushing the discovery burden downstream to that expression and validation space.

Christina Palmer 27:45

Totally. This is, I think, where people sometimes underestimate the operational burden of getting a million sequences. So not the same as getting a million useful antibodies, right? You need a strategy for selecting which sequences you're going to synthesize, express, and test. That may involve looking at things like clonal expansion, mutation patterns, lineage analysis, antigen specificity, applying different filters around predictions like expression or developability, you know, or some other prioritization criteria. And if you don't have that, now you're just drowning in data. Yeah, yeah, definitely.

Tracey Mullen 28:19

And I think, yeah, I mean, ultimately, I think sync synchronous workflows, they are incredibly powerful, but only if they're connected to that very thoughtful triage strategy downstream. And so lastly, this is the last one we were going to talk about today is the fax-based single B cell workflows. I know I think these are, again, the oldest, most straightforward version of single cell recovery, right? You can stay in your B cells with labeled antigen, sort antigen positive cells into plates, recover those sequences, express those antibodies. I think the advantage here is that fax is it's pretty fast. It's also very accessible and often relatively inexpensive. You know, many organizations they already have access to sorters. There may not be any major CapEx necessary to access them. And it can be very efficient baseline, if you will, for simple binding campaigns. But the limitation, of course, is that you usually have little to no functional context at that screening step. So you're often sorting on just surface binding or an engine standing alone, not getting additional information like secretion, blocking, function, whatever it might be. And depending on that downstream sequencing workflow, you may also have to be careful about how you preserve that native heavy and light chain pairing and ultimately enable that clean genotype-phenotype linkage.

Christina Palmer 29:33

Yeah, definitely. This is the same issue that display platforms often face. You know, if you have a sticky reagent or, you know, something about your reagent is suboptimal, it's aggregated, incorrectly folded, doesn't representative the native target in context, then your sort is only as good as that reagent. And so you, you know, are you know not really able to use the power of the platform. And like, you know, like you said before, you know, it is really a binding first workflow. So whatever you sort for it, that's what you get. You may get throughput and speed, but you push a lot of that functional triage downstream and it does come with potential risks.

Tracey Mullen 30:10

Yeah, yeah, that's that's a great way to say it. I think, yeah, fax gives you efficient recovery, but lower front-end biological resolution. So I think you know, for some campaigns, that's that's totally fine. For others, it does create that downstream burden.

Christina Palmer 30:23

Right. And so, you know, like any high throughput discovery method has this same issue, right? The more broadly you screen, the more downstream triage you need, right? You just end up with more things you have to sift through. So, you know, adding throughput at the front end can be a gift or problem, depending on if you have that infrastructure to then validate what comes out or make those selections during the workflow.

Which B Cell Population To Screen

Tracey Mullen 30:46

So let me put on the in vivo discovery hat for a hot minute here. So single B cell discovery, you know, it depends heavily on what B cell population you're sampling. And I want to get into this a little bit more because we kind of sampled this a little bit at the beginning, but I think this is an important topic. So, you know, plasma cells, plasma blasts, memory B cells, antigen-positive B cells, you know, these aren't interchangeable cells, cell population. So if you're looking at antibody secretion, you're often biased toward antibody secreting cells. That can be useful because secreting antibody is what you ultimately care about, but it can also, you know, bias toward high secreting or robust cells depending on the assay condition. And if you're sorting antigen-positive memory B cells by FEX, for example, you may get a different slice of that repertoire. And if you're doing, you know, sequencing first repertoire analysis, you may capture breath, of course, but then you'd have to infer which sequences matter. So, you know, I think the platform is not just a piece of hardware, if you will. It really defines the biology that you're you're able to see in the end.

Christina Palmer 31:48

Yeah, no, that's a really important point. And it also affects how we compare single B cell to in vitro discovery, right? With a display platform, you decide what library you're screening, what selection pressures you apply. With single B cell, the animal and the immune response defines the starting repertoire. And that platform then decides defines which part of the repertoire that you can recover.

Tracey Mullen 32:14

Right. Yeah. And I think that that's why, you know, I don't think any organization should treat single B cell as a single capability. It's more of like a family of workflows. You know, a secretion-based workflow interrogating plasma cells is different from an antigen-specific memory B cell sort. And, you know, similarly, a high content opto fluidic workflow is different from a high throughput sequence-first workflow. And a workflow optimized for, you know, active secretion is obviously different from one that's optimized for repertoire breath.

Christina Palmer 32:46

Yeah, which is why platform selection should start with the target, the campaign goal, and the biology that you're trying to sample, not just the instrument that's available.

Tracey Mullen 32:56

Yep. Yeah. That's the headline. That's the headline right there. All right.

AI Triage And Data-Connected Discovery

Tracey Mullen 33:00

So I think, you know, we have to, we can't go through a podcast without talking about AI. We have to talk about AI. You know, it's really changing how people think about these workflows. And so I think it's important for us to kind of touch on this at least a little bit. Because, you know, historically, one of the biggest challenges with high throughput single B cell or, you know, these sequencing first approaches was that you generate more data than what you knew what to do with it. You know, you might recover tens of thousands or even hundreds of thousands of paired sequences. But the question becomes like, okay, now which ones do we actually express? And AI and computational modeling, you know, these are starting to change that conversation significantly. So, Christina, from your perspective, you know, where do AI and computational tools fit into antibody discovery workflows like this?

Christina Palmer 33:43

Yeah, no, I mean, we have to talk about it, right? I think, you know, I think AI honestly becomes most useful when it helps you prioritize a large search space, right? So if a platform gives you thousands of sequences, AI can potentially help you rank candidates based on predicted properties like developability, liability risk, expression, and also things like sequence diversity, clonotype families, human likeness, or you know, other, you know, other characteristics of interest. But it doesn't replace experimental validation, um, you know, but it can help decide which molecules are worth spending experimental resources on.

Tracey Mullen 34:24

Yeah, that's the key point, right? AI doesn't eliminate the need for wet lab validation. It just changes how we choose what to validate. Uh so if you have a sequencing first workflow, AI can obviously help make that massive repertoire more actionable. You know, it can identify families, prioritize diverse representatives, you know, flag liabilities, suggest which candidates might be most likely to express or or behave well. But if you know the upstream assay also gives you very rich functional data, AI can can probably help into integrate that information with the sequence clonotype and developability and downstream characterization data as well.

Christina Palmer 35:01

Yeah, absolutely. I mean, I think the highest value version of AI is not isolation. It's AI connected with good experimental data. So if you have, you know, if your input data are poor or disconnected, what you can get out of that from a model is pretty limited. But if you have sequence, binding, functional data, et cetera, the company computational tools can start to really help make more informed decisions.

Tracey Mullen 35:26

Yeah. And that's where I think the landscape gets interesting, right? I mean, the future may not be choosing between high throughput sequencing first and lower throughput functional screening. You know, AI could actually make it easier to work with large sequence spaces, but also increase the value of high quality, you know, well-annotated experimental data. So I think the question becomes you know, can we design workflows where the data coming off of the single B cell platform are not just a list of hits, but a structured data set that in turn helps us learn? Right. Isn't that the dream?

Christina Palmer 36:00

And that learning can like feed back into future campaigns. So if you know which sequence features or assay signatures or clonotype patterns tend to produce good recombinant antibodies or good functional activity, that can then help improve future campaigns.

Tracey Mullen 36:15

Yeah. Yeah. So so AI doesn't really make platform selection irrelevant, to be clear. You know, it may make platform selection even more important, right? Because the quality and structure of the upstream data kind of determine what you can learn downstream. Absolutely. Yeah, absolutely, absolutely. All right, so let's let's pivot here and talk a little bit about in vitro discovery because I really want to pull more on the, you know, your in-vitro discovery perspective in particular, Christina. So when you hear people talk about single B cell discovery, it can sometimes sound like it's meant to replace display, but I don't think that that's the right way to think about it. So, how do you see the two platforms kind of working together?

Christina Palmer 36:53

Yeah, I absolutely don't love that framing of them being opposing forces, right? I see these as really complimentary platforms, each with, as we've been talking about, their own pros and cons and proper use cases. You know, single B cell discovery, as we've been talking about, is really strong when you want to capture an immune response, preserve that native pairing, and recover antibodies that have been shaped by an in vivo affinity maturation. Display is stronger when you want to control that selection process, you want very large library coverage, or you maybe want the ability to engineer selection pressure around a specific property, like affinity or epitope or developability. In some campaigns, you know, single B cell discovery may be the front-end discovery engine. And in others, display may be the primary engine for a variety of reasons. And my favorite approach is using both if you know the things make sense, right? Multiple shots on goal is always the way to cover, you know, hedger bats and make sure that you are getting to that ultimate, you know, ideal antibody. You know, you can use in vivo discovery and single B cell platforms paired to identify antibodies that have gone through natural affinity and epitope diversity, and then use display for, you know, affinity maturation or developability optimization or something like pH engineering or you know, any other optimization you can think of, right? So it's definitely not an either or. We have to be all in on one or another, but there are always going to be advantages to capture.

Tracey Mullen 38:27

Yeah, that's how I like to think about it too. It's not just, you know, single B cell versus display. It's what combination of technologies gives you the best probability of finding the right molecule. And it's funny because, you know, if you and I have both been in the antibody discovery industry for quite some time. And I remember, you know, way back when I was in the lab doing hybridoma discovery, I moved from that particular role over uh into a large pharma where we started using in vitro display. And I kept getting asked the question of which one's better, which one would you prefer? Because back then it really was an either-or, and it's that's just not the case anymore. And now I think it really is, you know, how do you combine technologies? How do you ultimately find the best molecule with the tools that we have at our disposal?

Christina Palmer 39:09

Yeah, absolutely. You know, in vitro display can be really valuable when you've found a great hit out of a single B cell campaign, but it's not your final, final, final, right? Like it's it's in the world we are in today of super ultra high affinity finders and you know, very clean developability profiles, those things need to be engineered. They don't just happen naturally. And so when we compare in vitro display platforms and single B cell platforms, we can really achieve the you know very high bar that we're looking for for a lot of platforms of projects.

Tracey Mullen 39:46

Yeah, it's a really great point. It's a really great point. And yeah, I mean, ultimately in vivo discovery gives you the immune solution, but in vitro optimization can really help turn that into a therapeutic solution. You said that way more nicely than I did. If you are a biotech team trying to choose a single B cell discovery workflow, I personally would start with five questions. So the first would be what's the biology of the target? You know, is it soluble? Is it membrane associated? Is it conformationally complex? You know, do you need cell binding, blocking, agonism, antagonism, internalization, maybe some other functional property? So just need to know that up front. Second, you know, what B cell population do you want to interrogate? You know, are you trying to capture plasma cell or plasma blast? Are you interested in antigen-specific memory cells? Are you trying to just profile broad repertoire diversity and then infer which sequences are most valuable? You know, those are all going to play into the actual instrument that you'll end up working with. Third is what does success look like? You know, do you need maximum repertoire diversity, those rare functional binders, you know, epitopic coverage, species cross-reactivity, or do you just need a small number of highly validated leads? Those are important questions to know, the answers to. Fourth would be, you know, where do you want to put that burden of discovery? Do you want to push it downstream to recombinant expression or do you want to do more upfront? And then lastly, what does your organization have the capacity to do? It's kind of tied to the fourth question, right? Like, do you actually fund it doesn't really matter if you want to push the burden of discovery downstream if you just don't have the infrastructure to express it, right? So do you have a workflow that actually reduces that downstream burden? Christina, what am I missing? Any any other questions that you think we'd add to that?

Christina Palmer 41:27

I mean, I think based on what we've talked about today, I would add that you know, you should think about how you plan to use computational tools or AI in the workflow before you have thousands of minutes, right? If you're generating a large sequence data set, you need a strategy for prioritization. And you know, we talked about some of the things that that could include clonotypes, sequence diversity, developability projections, et cetera. But AI is never going to substitute for knowing what question you're trying to answer, right? If when we use AI, it's most useful when the experimental workflow is designed to produce interpretable data and that can get then tied to individual clones.

Tracey Mullen 42:06

Yeah, that's a great point. You know, AI can help triage, it can help prioritize, but it doesn't fix a poorly designed campaign. Garbage in, garbage out is still true to this day.

Christina Palmer 42:16

100%, 100%. And if you only collect sequence data and have no phenotype linkage, what are you gonna like there's not a whole lot you can do with that, right? Like you can get those sequences, but there's not a ton you can do. If you collect functional data, but don't collect it cleanly back to sequence and recombinant value validation, then you lose a lot of the value. So just like thinking about those things ahead of time, I think is is really important.

Tracey Mullen 42:41

Yeah, yeah. So yeah, if I were to summarize that, you know, the best workflows are not just high throughput, right? They're information dense. They, you know, connect immune repertoire, sequence, phenotype, function, developability, recombinant validation, all in a way that ultimately helps teams make the right decisions for their programs. So this kind of brings us around to Mosaic's point of view here. And I'd love to just spend a little bit of time talking about what I previewed at the beginning, which is a platform agnostic but data connected point of view. So I don't think that the future of single B cell discovery is that every campaign goes through one platform. I think that the future is platform agnostic. It is data connected, and it's it's importantly use case driven. You know, the right workflow depends on the target, it depends on the desired antibody profile, it depends on which B cell population you want to interrogate, how much functional resolution you need up front, you know, how much downstream validation capacity you have, all these things that we have been talking about. And I guess increasingly depends on how you plan to use computational tools as well. No, I agree with all of that.

Christina Palmer 43:42

And I think that's where we at Mosaic are, you know, our goals is to be differentiated, right? We're not just asking which platform can we access. We are really trying to pair the right discovery strategy for a particular target, a particular client, and you know, the downstream decision. Right. So that we're really taking a holistic view.

Tracey Mullen 44:03

Yeah, yeah, exactly. Yeah. Mosaics Value prop here is it's not just running a single B-cell platform, right? It's helping clients decide where to place the work. For one client, the right answer may be broad repertoire capture and you know AI-guided prioritization before the high throughput IG validation. But for another, the right answer may be lower throughput but higher content upfront screening so that only the most compelling candidates move downstream.

Christina Palmer 44:27

Yeah. And for another, the right answer may be single B cell plus display. Recover that natural immune-derived antibody that has the function you want, but requires some in-bitro optimization to improve, you know, affinity or potency or developability or even in format.

Tracey Mullen 44:44

Yeah, yeah, exactly. And, you know, this is really a consultative opportunity, if you will. You know, a vendor says, here's our platform. It's great. Use it. But a partner says, hey, let's understand your target. Let's understand your B cell biology. Let's understand your organization, your downstream capacity, and your tolerance for risk, you know, and let's build a workflow around that.

Christina Palmer 45:08

Yeah. No, this is what I love about what we do is having these conversations with customers. And I think that's exactly what you need to have successful campaigns.

Tracey Mullen 45:17

Yeah, yeah, totally. We are ultimately, you know, scientists talking to scientists over here. We we love talking, as you could tell. All right. So maybe

Rapid Fire Picks And Key Takeaways

Tracey Mullen 45:25

we'll do a fun wrap-up here. I I have a fun game I'd like to do. Christina didn't prepare you for this, so it's slightly unfair, but it's a rapid fire round that I'd like to do. So if you had, I'm gonna ask basically a series of questions, and I just want you to answer rapid fire. All right. Okay, so if you had a simple, soluble antigen and you wanted fast, low cost, binder recovery, what platform would you want to work with? All right, easy one, right?

Christina Palmer 45:49

Softwall, max, single cell sorting, or even like a more traditional workflow, right? Depending on the antigen and goals, but like keep it simple, don't over-engineer it. No need for fancy upfront platforms.

Tracey Mullen 46:02

Totally. Yep, agree. All right, what if you had a difficult cell surface target where function really matters?

Christina Palmer 46:07

Yeah, I mean, function, it's really clear, right? You need a platform with higher resolution, something like the beacon or another high resolution functional platform.

Tracey Mullen 46:16

Yeah, I agree. Yep. But what about if you wanted a balance of, you know, throughput and assay depth?

Christina Palmer 46:21

Right. So I mean, we talked about this earlier. This is where I would look more closely at those sort of middle ground microfluidic, microcapillary, compartmentalized single cell platforms, right? They can be a good fit if the assay readouts are meaningful and the recombinant validation supports that platform. Yeah, yep. Okay.

Tracey Mullen 46:38

Well, what if you wanted maximum immune repertoire coverage?

Christina Palmer 46:42

Well, that sounds a lot like saying sequencing first or droplet-based workflow. But right, the caveat there is that you have to have a strong plan for um how you prioritize and select sequences to then triage and validate.

Tracey Mullen 46:54

Yeah. Okay. What about if a platform advertises very high theoretical throughput? What's the first question you're gonna ask them?

Christina Palmer 47:00

I would definitely want to know what the actual, like realized single cell throughput is, right? How many true single cells are actually screened once you account for empty compartments, doublets, all of those things that we talked about?

Tracey Mullen 47:12

Okay. What if the campaign depends heavily on preserving native heavy and light chain pairing?

Christina Palmer 47:19

Right. So genotype, phenotype linkage, right? That is central to a lot of what we're talking about. I would want to know how we connect that observed phenotype back to the sequence and how confident they are that you know they can get the right heavy and light chain pairing from that particular cell.

Tracey Mullen 47:36

Okay. Okay. Doing great. All right. What if a customer comes to you? What if a customer comes to you and says, you know, functional assay requires multiple wash steps and sequential reagations, but I gotta add this functional screen at the primary screening stage? Yeah.

Christina Palmer 47:52

Yeah. Then I think that's really gonna narrow what platform that you can work with, right? You need to understand whether the platform supports those like buffer exchanges, wash stabs, more complex assay workflows, or you know, whether you're more limited to something much more straightforward and streamlined.

Tracey Mullen 48:09

Well, what about if the functional hit's gonna be rare in that set?

Christina Palmer 48:13

Then I think we would absolutely go the function forward route. Otherwise, you know, you're relying on downstream functional assay that may or may not be able to handle that kind of throughput to find what is considered a rare clone.

Tracey Mullen 48:27

Okay. Okay. What if you, Christina, had a company that could express and validate thousands of IgGs downstream? Wouldn't that be great?

Christina Palmer 48:35

Then I would be definitely more comfortable taking that broader swath of sequence pool and pushing that burden downstream where we can really get that true, you know, validated recombinant protein data.

Tracey Mullen 48:48

And what if it's the flip side, right? What if you had a company with limited downstream capacity?

Christina Palmer 48:52

Yeah, then you're gonna want to do the opposite, right? You're gonna want to front load as much as possible. You're probably not going for a sequence first platform, you're going for something that gives you a lot richer data set for each of those clones to not push that burden downstream.

Tracey Mullen 49:06

Okay. And what if you have a massive sequence data set and wanted to use AI?

Christina Palmer 49:10

Yeah, I mean, that's where we operate now, right? But I would make sure that the AI is connected to meaningful experimental data, right? Sequence alone is useful, but sequence, genotype, function, expression, all of those make it so much more powerful.

Tracey Mullen 49:27

Okay, I got one more question for you. If you had a hit from a single B cell that looked biologically interesting but needed optimization, what would you do?

Christina Palmer 49:35

You would call me. That is where in vitro discovery and engineering tools become incredibly powerful, right? Designing libraries, you know, running selections for affinity maturation, developability engineering, any sort of like format optimization, that's where an in vitro platform really shines.

Tracey Mullen 49:54

So I I'm sensing a trend here, right? It seems like the answer to every question is like, it depends. Welcome to science. Yeah, yeah, that might be the title of the episode. Single B cell discovery, it depends. Yep. I love it. All right. Well, I think that's actually the perfect place to end. It was a ton of fun chatting with you today. Yeah, I think the main takeaway from this conversation is that you know, single B-cell discovery is not just one thing, right? It's a it's a family of workflows. Each platform gives you a different balance of speed, throughput, resolution, cost, sequence recovery, you know, that fundamental genotype, phenotype linkage that we talked about so many times today, functional context, downstream validation. You know, the right answer just depends on more than the target, right? It depends on so many other factors that go into that. And it depends on the B cell population you want to interrogate and the optimization that's required for that, right? So some teams are built to handle very large downstream workflows. You know, they may want broad repertoire coverage, large sequence space, AI-assisted prioritization before expressing thousands of IgGs. And you know, other teams, they may just want to deeply interrogate cells earlier and advance a smaller, you know, more highly curated set of antibodies into recombinant validation. And I think what we what we learned today is that neither strategy is inherently right or wrong. You know, the question is which strategy matches the target biology, you know, the immune repertoire, the campaign goal, the downstream infrastructure, ultimately the decision that the team needs to make next, right? So for Mosaic, that opportunity is really to be thoughtful. You know, we're we're platform agnostic, we're data connected. We like to start with the target, define that campaign goal, understand the biology, understand the organization's bottlenecks, and then decide from there what work should happen and then choose ultimately the right workflow that gives the highest probability of success. So importantly, you know, I think single B cell discovery shouldn't be viewed in isolation. I think it works best when it's connected to the rest of the discovery engine, immunization strategy upstream, you know, maybe in vitro discovery, expression, recombinant validation, functional screening downstream. Um, and of course, you know, AI-enabled prioritization as that continues to unfold. So, Christina, thank you so much for joining me. Today was a lot of fun. Really enjoyed our conversation. Yeah, well, thanks for having me. And thanks for everyone listening to the Biologics Brief. If this raise any questions about which antibody discovery workflow is right for you and your target, please reach out to mosaic team at mosaicbio.com. Thank you so much.