Invenra Blog

How Are Bispecific Antibodies Made? A Quick Look Inside the Process

The Invenra Team — Thu, 20 Aug 2026 18:34:58 GMT

There’s no one way to make bispecific antibodies — it depends on which platform you use! The process is consistent, but the places where platforms differ are worth understanding, as they might influence your choice in one direction or another.

This quick guide explains that the biggest differences between platforms appear in the assembly step, where issues like light-chain mispairing can create downstream problems with yield, purity, scalability, and formulation. We then show how our B-Body platform avoids those issues structurally, enabling cleaner screening, more reliable characterization, and a more manufacturable bispecific.

If you want a deeper comparison of the platform families themselves, our guide to bispecific antibody platforms covers that ground. If this is the first you’re hearing of the B-Body platform for bispecifics, get the 101 here.

The four stages of bispecific antibody discovery

Here’s a quick rundown of how bispecific discovery works in general.

Stage 1: Get the target ready

Before anyone finds an antibody, they need the target in a form that can be screened. That means cell lines and purified protein that present the target the way it appears in the human body.

If the reagent doesn't represent the real thing, everything downstream inherits the error. Many teams looking for a bispecific have done part of this already while evaluating whether the target idea was viable.

Stage 2: Find binders

Two broad routes exist in this stage:

In vivo methods immunize an animal (usually a humanized mouse or a rabbit) and harvest what its immune system produces. Rabbits are popular because they generate antibodies unusually quickly, which significantly reduces the immunization wait time.

In vitro display methods skip animals entirely and screen enormous synthetic libraries in a dish. Phage display is the workhorse. Yeast display is often used alongside it.

Neither route is inherently better than the other. Display gives you more control over your starting material and avoids inheriting the quirks of an animal's immune response. Immunization can produce binders that are hard to find synthetically.

Most of the industry does something in this space, and the steps look broadly similar across discovery teams.

Stage 3: Assemble them into one molecule

Here is where the paths diverge, and this stage is why bispecifics are harder to produce than monoclonals.

An antibody is composed of four chains—two heavy and two light. In a normal monoclonal, both halves are identical, so there is only one correct way for them to come together.

A bispecific antibody has two different arms, so four different chains have to find the right partners. Two things can go wrong. The two heavy chains can pair with themselves instead of each other, and the light chains can attach to the wrong arm. The heavy chain problem is largely solved. Knobs-into-holes, a bump engineered on one heavy chain and a matching groove on the other, has been used across the industry for years and works well. (We use it too!)

The light chain problem is where platforms differ, and the approaches split into a few families: skip the problem with a shared common light chain, steer the chains with engineered charges, or change the architecture so the arms can't pair incorrectly. Fragment-based formats sidestep the question entirely by not building a full antibody.

Stage 4: Purify and characterize

You must separate the molecule you want from everything else the cells made, then find out whether it works. How hard that is depends on what you did in stage three.

Every IgG-like bispecific must address the same two fundamental challenges: pairing the correct heavy chains and preventing light-chain mispairing.

Bispecific formats like knobs-into-holes and common light chain work, but they can create manufacturing or compatibility compromises. Poor chain pairing leads to lower yields, higher impurities, and longer development timelines.

Combinatorial diversity of bispecific IgGs. Overview of possible combinations to arrange heavy and light chains from two different antibodies, including strategies to overcome incorrect heavy-light chain pairing.

How it works on the B-Body platform

Here’s the same process, run end-to-end, on Invenra’s B-Body platform.

Invenra’s Rapid Discovery services provide an efficient, industry-leading approach to discovering monoclonal and bispecific antibodies. With cutting-edge phage display technology and our streamlined workflow, we deliver high-quality antibody leads in record time.

Stage 1: Prepare the target

We evaluate available reagents and pick the ones that best present the target, so the libraries have something realistic to bind to. Cell lines, purified protein, whatever the target requires.

Stage 2: Screen the libraries

This is where Invenra expertise shines. Our primary workflow begins with phage display, using proprietary libraries holding so many possible sequences that you need scientific notation to describe the contents of a single tube.

Mechanically, it’s less mysterious than it sounds. The target is attached to a magnet, the candidate antibodies are added, and you see which ones stick. The craft is in the rest of it: washing away everything that binds non-specifically and getting as much sequence diversity as possible from a single shot.

The part that is genuinely different is the parallelization.

We separate our libraries based on physical characteristics, such as antibody size and parental sequence, and keep them separate rather than pooling everything.
Then we query them against multiple screening conditions at once, because nobody knows in advance which condition best represents what happens in a patient. That gives multiple shots on goal, both across libraries and across conditions.

The alternative, as it’s done elsewhere, is one tube at a time. Run it, see what you get, and if it isn't what you needed, go back to the beginning and repeat. Projects run that way take six to twelve months.

Stage 3: Add yeast display and sequencing

Material coming out of phage, ideally still millions of unique sequences, often moves into yeast display. Yeast gives more dynamic range and more dimensions along which to sort individual antibodies, so you can ask sharper questions than phage alone allows.

Everything from every selection and screen then goes into next-generation sequencing. The output tells us which sequences came out of which condition in which well, which turns a lot of data into a map.

Stage 4: Make the mAbs and characterize them

From that sequencing data, we choose around 400 sequences and express them as actual antibodies from mammalian culture.

Then the characterization phase begins, and this is where we would argue the difference shows up most. It’s not just "does it bind?”

It covers whether:

It binds anything it shouldn't (polyreactivity)
Its surface properties are amenable to manufacturing (hydrophobicity)
It’s assembling the way it should
It binds the analogous target in your safety toxicology species
The panel has epitope diversity, which we establish through binning experiments to see whether two antibodies stick to the same place or different ones

We spend time on this because we see issues when people bring us work done elsewhere. Frequently, what they received was a panel of antibodies and one experiment showing binding to the target, and that was it. That’s “tip-of-the-iceberg” characterization, and it can make results look better than they are, because they have not looked beneath the waterline.

Stage 5: Build the matrix

Now you have characterized monoclonals against target A and characterized monoclonals against target B, and you want bispecifics.

You pick the best monoclonal antibodies for each target, and the B-Body platform lets us build every possible permutation of those combinations rather than guessing at a few. A small-scale expression and purification pipeline isolates each permutation cleanly, in its final bispecific format.

Example functional data as a heat map representing the screen

Example SEC data in a 12x12x2 matrix

Stage 6: Characterize again as bispecifics

Then you go through the monoclonal characterization phase a second time, except now with bispecifics. Same assays, same readouts, same rigor.

That repetition is the whole point. We’re using tried-and-true methods that we can only run because of what the B-Body platform gives us. We don't have to worry about things like light chain mispairing. We can run assays that tell us how much impurity is present, because nothing is hiding underneath.

A platform that produces mispaired species muddies its own analytics. If a meaningful fraction of what you made is the wrong molecule and you can't cleanly resolve it, your purity number is essentially a guess. Removing that problem is what lets well-understood methods work on a multispecific.

Stage 7: Choose the lead

Function trumps everything. We start with the functional data, select the top performers, and then cross-reference them against developability assays. Of the top hundred, perhaps thirty carry no serious developability red flags, and those move forward.

This is a consultative review, not a simple data handover. You see the data, and we work through it together.

Stage 8: Scale up and confirm

Nobody goes from a one-milliliter high-throughput experiment to a conversation with regulators. Selected candidates are scaled up to 200 mL, supporting multiple purification steps and yielding many milligrams of protein.

At that scale, you can dig properly into the assays and validate what the screen suggested.

Check out slides 15-20 of our talk at AET for a map of the two entry points—our own mAb discovery and partner-supplied antibodies via B-Body Express—converging into bispecific discovery.

This deck also breaks out the discovery steps individually.

Where the platform choice pays off later

Everything above describes discovery. The decisions, whether they are good or bad, follow the molecule into manufacturing.

Light chain shuffling

This is the failure mode that causes the most trouble. If light chains mix and match, three things happen.

First, you may not have a competent molecule, because the wrong light chain paired with a given heavy chain may not bind properly.
Second, batched are not consistent, so you have to prove potency within guardrails, typically around +/-5%. If pieces of your molecule scramble, you risk failing batches of drug.
Third, and worst: a molecule where the light chains are flip-flopped is almost impossible to purify, because it is essentially identical to your drug. There is no clean separation method for something that similar.

On the B-Body and T-Body platforms, we don't see light chain shuffling, and we achieve that through the architecture rather than by adding charge engineering on top, which is also why the platform works with whatever antibody pair you choose.

There’s a second element worth mentioning. The knob-and-hole region can produce a small amount of knob-knob or hole-hole homodimer. On our platform, that isn't a problem, because those species differ enough in charge and other properties to be separated.

Yield

Yield is mostly a function of how complicated your purification is. Every chromatography step costs you molecule volume, and in early phases you can expect to lose around 20% per step even in a good case.

So the cost of a shuffling problem isn't only the shuffled material. It's the extra chromatography steps you need to clean it up.

Add one and you've cut another 20%.
Add two, and you may be down 40%. That comes directly out of what you can put in a bottle and ship to your Phase 1 trial.

With B-Body, we scale it up and process our bispecific molecules like monoclonal antibodies. B‑Body bispecifics antibodies typically deliver 6 to 11 g/L (the green bar in the graph below) from stable CHO cell lines. For comparison, a panel of other bispecific platforms averaged 2.4 g/L (grey). This is the highest‑expressing bispecific platform we're aware of, and it drops right into standard mAb‑like processes.

B-Body bispecific yield (green) against a panel of other bispecific platforms (grey). B-Body runs 6–11 g/L; the panel averages 2.4 g/L.

Purity

Purity and yield are the same conversation from different directions, because you can’t administer until you hit specification, generally 95% purity or better. If you can reach that in two chromatography steps, great! If you need a third, you have to run it, because until you hit spec, your batch has failed.

Those steps are not cheap. An additional step is not a scheduling inconvenience. It can be a seven-figure line item in the cost of goods.

B-Body reaches dosing-grade purity in two chromatography steps. We achieve >80% purity in a single purification step and >95% in a standard two‑column process. The figure below demonstrates purity >95% for all formats, including 1×1, 2×2, 2×1, one‑arm, and an IgG1 control.

That single-step purity holds across formats: after polishing, size-exclusion purity lands at 90 to 95% or better across 1×1, 2×1, and 2×2.

Size-exclusion purity by format, before polishing (grey) and after (green), across 1×1, 2×2, 2×1, trispecific, one-arm, and an IgG1 control. After polishing, purity lands at 90–95% or better.

Scalability

What kills scalability is having to do anything unusual. If you have to tell a CDMO they need a chromatography method they've never run, for example, that alone can kill a project.

Being able to treat B-Body molecules as if they were monoclonal antibodies and still achieve the purity and yield means they run as what the industry calls a platform process. That's roughly code for "we don't have to work very hard to give you what you want," and it's a compliment. Most facilities can turn a platform process to thousands of liters without breaking a sweat.

Subcutaneous compatibility

Concentration is the constraint here. To dose subcutaneously, a molecule must survive high concentrations without precipitating, and some formulations struggle badly.
Camelid-derived single-chain platforms are useful for comparison because they tend to be highly hydrophobic. Past a certain concentration, those molecules start sticking to each other, and once that begins, you have a snow globe in a tube and the product is finished.

There’s real engineering effort going into fixing this. We’ve seen presentations showing re-engineering camelid antibodies are less prone to aggregation, with reports reaching 10 mg/mL. When we make B-Body antibodies, ten to twenty times that concentration is within range.

We attribute this to the use of fully human antibody sequences and to the B-Body architecture itself. The payoff for patients is the difference between a subcutaneous injection and sitting in a chair for an hour or two on an IV.

The anti-CH1 handle, and where it actually helps

One structural detail deserves explaining because it does most of the heavy lifting.

On a B-Body antibody, only one arm carries a CH1 domain. That makes CH1 a purification handle: an anti-CH1 resin grabs only chains carrying it, and by expressing the other chains in excess, you stack the deck toward pulling out the correctly assembled molecule.

Here’s a look at the B-Body:

The B-Body scaffold. (1) knobs-into-holes Fc so the heavy chains pair with each other; (2) proprietary CH3 domains replacing CH1/CL in one Fab arm, so the right light chain pairs on its own; (3) plug-and-play variable domains from any source; (4) a sole CH1 domain that allows one-step anti-CH1 purification.

Where the anti-CH1 handle matters most is at the beginning, during screening. You can’t develop a bespoke purification method for every one of hundreds or thousands of candidate multispecifics.

It would take far too long. Anti-CH1 gets every one of them to "clean enough to run an assay" quickly and uniformly. That’s what makes matrix screening practical.

It can also be used at scale-up, added at the front to improve initial product quality and make later steps easier. We flag that as the more debatable use, because the industry norm is a standard starting path that doesn't include anti-CH1, and deviating from the norm is exactly the kind of unusual thing that complicates a CDMO relationship. It's an option, not a default.

A quick recap

Bispecific antibodies are made by finding good binders, combining them into one molecule, and then proving the result is manufacturable and does what you need. The industry broadly agrees on the first and last parts.

What differs is the middle, and the consequences of that difference arrive late, in purification steps you didn't budget for, batches that fail specification, and molecules you can't concentrate enough to dose the way you wanted.

Our approach with the B-Body platform is to eliminate the assembly problem structurally, so the rest of the process can be straightforward. Tried-and-true methods, run twice, on molecules that behave.

Talk to us

Have a target pair in mind? Tell us what you're trying to build and where you are, and we'll walk you through what the process would look like for your program.

Contact us »

Or start with the data: Our B-Body explainer covers the yield, purity, formulation, and matrix screening data behind the process described here. Learn more about the B-Body Platform »

Trispecific ADCs: Why Three Targets Can Beat Two

The Invenra Team — Wed, 19 Aug 2026 21:07:39 GMT

An ADC is a targeted drug delivery system. You attach a potent cytotoxic payload to an antibody; the antibody finds a tumor antigen; the therapy gets pulled inside the cell, and the payload does its work where it's supposed to rather than everywhere at once.

The approach works, but it also has a structural weakness that becomes more obvious the longer time goes on: most ADCs recognize a single antigen, and tumors are not uniform.

Within a single tumor, some cells express the target at high levels, some at low levels, and some not at all. A one-target ADC treats that variation as though it doesn't exist. The cells that don't carry the antigen survive, and they're the ones that come back.

Bispecific ADCs were the first real answer, and several are now producing encouraging clinical data. Here, we shed light on the next innovation — trispecific antibodies — and when it’s worth evaluating.

What a third arm actually does

The intuitive answer is breadth, and that part is true. Adding a third tumor antigen can expand the proportion of patients likely to respond within an indication and broaden the range of cancer types in which the molecule has a target.

It also guards against escape. When a tumor evolves and downregulates or sheds a target, a molecule with three arms still has something left to bind.

But breadth is only the surface of it. Three arms give you options in how you use them.

You can choose your logic

With three binding arms, you can build what amounts to “OR” logic, where binding any one of the three targets is enough to engage. That gives you the widest possible reach.

Or you can build “AND” logic, where the molecule only fully engages cells carrying two particular targets together. That narrows what the ADC acts on, which is a way to protect healthy tissue that happens to express one of your antigens.

You can also combine the two. As we'll come to, one of the more interesting mechanisms we're working on can effectively switch a molecule from “AND” behavior to “OR” behavior depending on where it is in the body.

You can fix internalization

This is the one that gets underappreciated. Binding a tumor antigen is necessary but not sufficient for an ADC. The complex must be pulled into the cell for the payload to be released. Plenty of targets, depending on which epitope you hit, internalize poorly.

Internalization is largely driven by clustering at the cell surface. A biparatopic antibody, meaning two arms that bind two different epitopes on the same target, creates daisy chains between adjacent target molecules and produces dense clusters, which substantially changes the internalization profile. That approach is already well established in bispecific ADCs.

Now, what a third arm adds. With a bispecific, going biparatopic costs you both arms; you've spent your whole molecule on internalization. With a trispecific, you keep the biparatopic pair and still have an arm left for breadth, escape prevention, or an entirely different mechanism.

You take a proven approach and put differentiation on top of it.

Why trispecific antibodies for ADCs?

Here’s what it boils down to, taken from what we presented at the recent PEGS conference.

Trispecific Advantage	Why?
Broader patient coverage within a cancer type	Targeting multiple targets increases the number of patients who are likely to respond in an indication
More cancer types covered	Targeting multiple targets increases the number of cancers that are likely to respond
Better tumor control	Targeting multiple targets limits the tumors' escape routes and attacks a larger portion of a heterogenous tumor
More mechanisms of action available	These include VEGF Activated Avidity Lock (VAAL), Biparatopic Plus, and Stroma & Tumor

Does a third target mean more toxicity?

The evidence so far says the payload usually drives the toxicity ceiling, not the number of antigens you're targeting.

An FDA analysis of eight vc-MMAE ADCs directed at entirely different antigens found that they all reached Phase II at a similar dose range of roughly 1.8 to 2.4 mg/kg. The antigen wasn't what set the maximum tolerated dose. The payload class was.

Clinical bispecific ADCs point in the same direction.

Izalontamab brengitecan, an EGFR × HER3 ADC with a topoisomerase-1 inhibitor payload, reached a recommended Phase II dose of 2.5 mg/kg with hematologic toxicities consistent with monospecific ADCs carrying the same payload class and received FDA Breakthrough Therapy Designation in August 2025.

JSKN016, a TROP2 × HER3 ADC, showed a manageable and predictable Phase I safety profile with class-consistent toxicities. Neither produced qualitatively new toxicity from the multi-targeting format itself.

That's a large part of the foundation the trispecific case rests on. If adding a second target didn't add toxicity, there's reason to expect a third arm can capture more tumor heterogeneity without a corresponding safety penalty. It isn't proof, but the direction of the clinical evidence matters.

You can find more on this on slides 20 and 21 of our PEGS presentation.

Four ways to spend the third arm

When we started designing trispecific ADCs, we organized the work around four mechanisms where a third arm could plausibly deliver something a bispecific can't. Those four became the structure for our forthcoming 52-program card deck initiative, and they're a useful way to think about the design space, so let me break it down.

52 programs, four mechanisms, one deck. Invenra is developing 52 T-Body ADCs programs across four strategic approaches to overcome resistance and improve tumor selectivity. Four suits with the example target combinations under each.

A closer look at the VEGF-activated avidity lock

Most ADC targets have a problem: they're also expressed on healthy tissue.

EGFR shows up in the skin, GI tract, liver, and kidney. TROP-2 in kidney, lung, liver, and GI epithelium.
HER3 in skin, GI tract, and reproductive tissues.
B7-H3 is broadly expressed across many normal tissues.

That's why on-target, off-tumor toxicity is a recurring problem in the field, and why EGFR-directed therapies in particular are known for severe skin toxicity.

The usual way out is to lower the affinity of your binding arm so it doesn't engage healthy tissue. The trouble is that a weak arm doesn't engage the tumor well either.

VEGF offers a way around that, and it works because of two facts about the molecule. First, VEGF is an obligate homodimer, meaning it exists as an identical pair, so one VEGF can bridge two antibodies. Second, it's often heavily overexpressed in the tumor microenvironment relative to systemic circulation.

Put those together and you can build a conditional switch. You deliberately make the tumor-antigen arms weak binders.

Outside the tumor, in tissue where VEGF is low, those weak arms interact briefly with healthy cells expressing the antigen but never reach the density needed to matter. Minimal off-tumor binding.
Inside the tumor microenvironment, where VEGF is abundant, the VEGF homodimer bridges two of your molecules and pulls them together. That clustering creates avidity, which is the total binding strength of the molecule as opposed to the affinity of any single arm. Suddenly the same weak arms are engaging strongly, and the clustering also triggers internalization, which is exactly what an ADC needs.

VAAL, the VEGF Activated Avidity Lock

The practical framing we use is that it makes “dirty” targets “clean.” It lets you use a low-affinity antibody against a target you couldn't otherwise safely address, and have it behave like a high-affinity one only where the tumor is.

Most ADC targets exhibit some degree of normal tissue expression. VEGF provides an avidity binding enhancement for bsAbs or tsAbs. Lower-affinity Abs to the TAA can be used to reduce normal tissue binding in locations where VEGF is not present.

Our data support the mechanism. Using A549 tumor cells expressing roughly 52,000 copies of EGFR per cell, an EGFR × VEGF bispecific with a low-affinity EGFR arm showed little engagement in the absence of VEGF. Add VEGF and you see strong binding and clear internalization, along with a large differential in killing in a piggyback ADC assay.

The A549 binding and internalization data

There's supporting evidence from outside our walls too. Our PEGS presentation shows that PD-1 × VEGF bispecifics have produced notable clinical results and substantial partnering activity, and that VEGF-driven avidity is one of the hypothesized mechanisms behind these results.

The caveat here is that this works well in vitro, and the open question is how strongly it translates in vivo. Is there enough VEGF in a real tumor to drive the effect at the level we see in a dish?

The clinical results from PD-1 × VEGF programs suggest there should be, but we're not going to overstate it before we have the in vivo data. (Follow us on LinkedIn for the latest updates on our progress) It's also worth noting that VEGF isn't the only obligate homodimer with tissue-selective expression, so the same principle could extend to other switches.

What the data shows

Two bodies of work sit behind this.

1. Discovery screen showed trispecifics killed more broadly

This multispecific ADC story is straight from what we presented at World ADC. Pop open the deck and follow along with the short version of the story below.
We built a 12×12 matrix of bispecific ADCs from six targets with two binding arms each, including biparatopics and one-armed controls, and screened every molecule for killing across three breast cancer profiles: HER2-high, HER2-low, and triple-negative.

Cytotoxicity was measured with a piggyback MMAE assay and scored alongside developability from the same plate data. The best-performing pairs were then rebuilt as trispecifics and compared head-to-head. The trispecifics killed more broadly and more potently than the bispecifics.

That matrix approach is worth a note because there's more than one way to run it. We explored combinations across six different targets to find pairs that work well together. The other version, which some partners come to us with, is to bring two targets and ask for twelve antibodies against each, then screen those combinations. That second design is especially useful for optimizing the biparatopic side.

2. Biparatopic increased internalization

From our recent PEGS talk:

For Biparatopic Plus, we measured the internalization rate for a HER2 × HER2 × HER3 trispecific across breast and gastric lines spanning a wide range of receptor density, from roughly 250,000 HER2 copies per cell down to 18,000, with HER3 varying independently. The biparatopic arrangement increased internalization, and the piggyback ADC killing data followed.

We've also run a T-Body trispecific ADC for breast cancer against benchmark comparators, conjugated at a drug-to-antibody ratio of 5.56 with 95% monomer.

Underneath all of it, the platform itself has to hold up, because none of this matters if the molecules are hard to make! T-Body trispecifics have been expressed at up to 1500 µg/mL transiently, can reach better than 90% purity in a single step, run on standard Protein A and ion exchange, support both kappa and lambda light chains, and show normal IgG-like pharmacokinetics in rats with a half-life of 6.6 days.

When two targets are enough

Not every program needs three arms. The question is always whether the added complexity buys you something you actually need.

A useful way to frame it is to suppose you're trying to cover three indications, and a well-chosen bispecific reaches 90% of the patients across them, while a trispecific gets you to 96%.

Is that six points worth the extra complexity? Sometimes clearly, yes. Sometimes, no. It depends on the indication, the competitive situation, and what you're willing to carry.

Potency shifts raise the same question. If a trispecific gives you a tenfold potency improvement, is that enough on its own? In ADCs, it might well be, because dosing matters so much. But it has to be argued with an expert read of the data in the context of your target, not assumed.

The hardest thing to quantify is escape. A trispecific and a bispecific can look roughly equivalent in a standard assay, and then diverge sharply once a tumor sheds one antigen, because the bispecific has far less to fall back on. That advantage is real but it doesn't show up unless you design an experiment that can see it.

Which leads to the point we most want a prospective partner to understand. The most challenging part of deciding between a bispecific and a trispecific is selecting the appropriate assays and controls for comparison. Pick the wrong ones and you'll either miss a genuine advantage or credit one that isn't there.

That's a big part of why we build monoclonals, bispecifics, and trispecifics on the same architecture and run them in the same campaign — so the comparison is fair, and so the simplest molecule that does the job is the one that wins.

52 Views of Cancer

Our initiative, 52 Views of Cancer, started as a way to organize our own thinking. We wanted a structure for the question of where a trispecific ADC could deliver something a bispecific couldn't, and the four mechanisms listed previously became that structure.

The number came from a deck of cards: four suits, around thirteen programs each. It’s a one-of-a-kind playing card deck that celebrates the creativity, resilience, and spirit of people whose lives have been touched by cancer. Each card in this deck will feature original artwork by a different cancer patient, survivor, or immediate family member, making every card a unique view of the human experience with this disease. A corresponding website offers an inspiring and poignant look at each artist’s story.

Target selection draws on the published literature, clinical trial outcomes, and conversations with academic groups and clinical experts.

How to engage with us for ADC projects

There are three ways to work with us on these programs.

For some, we take the molecule furthest ourselves. We select the linker and payload and generate an IND-enabling data package, including cell line development, toxicology, and PK. That suits groups who want an asset that's ready to move rather than a starting point.
We also offer pre-made ADCs with less data, allowing you to complete development prior to IND.
More often, partners bring their own linker-payload technology. Under a straightforward material transfer agreement, we provide material, you conjugate it with your own chemistry and generate your own package, and if the results are compelling, there's a path to licensing that molecule.
If you like a target combination but want it adapted, or you like two of the three arms, we can do custom builds adjacent to the programs we've described.

In every case, a partner taking one of these forward gets exclusivity on that target combination for a defined period. We're not interested in building competing molecules against the same combination.

Talk to us

Working on an ADC and wondering whether a third arm would help? Tell us about the targets and the tumor setting, and we'll give you a straight read on the best path forward.

Contact us »

Or start with the data: The full PEGS presentation deck covers the VEGF-activated avidity lock, the internalization data, and the four mechanisms in more detail. View the presentation »

The Guide to Bispecific Antibody Platforms (2026)

The Invenra Team — Wed, 19 Aug 2026 19:12:51 GMT

We talk to a lot of teams trying to pick a bispecific platform, and the conversation usually starts in the same place: they have a list of formats, some sense that the options are different, but need some help deciding which differences actually matter for their program.

That’s a fair place to be. Bispecific antibodies have been around for decades, and in that time the field has produced an enormous number of ways to make them. A widely cited review counted well over a hundred distinct formats, and that was in 2017. There’s only been more development since then.

So this is the high-level tour I’d give you if you were investigating your options. What the main methods are, what problem each was built to solve, what trade-offs exist, and where the various platforms sit in that picture.

At Invenra, we help teams use our B-Body® format, so you should read this knowing that. I’ve tried to be straightforward about what the other approaches do well and where they fall short. No format is universally best, so I want to present those nuances objectively.

What we mean by “bispecific”

I only touch on this because the term is broader than some people expect. A bispecific antibody is any molecule that can bind two separate epitopes. It is a broad category that includes molecules with different shapes, sizes, and engineering approaches.

Once antibody engineers started trying to combine two binders into one molecule, they came up with a lot of clever ideas, and most of them work at some level. The differences show up later, when you have to manufacture the molecule and put it in a human being.

Here’s a visual look at the explosion of bispecific formats over time:

Bispecific antibodies offer a wide array of architectures. Reproduced from Brinkmann & Kontermann (2017), mAbs 9:2, 182–212.

Let’s take just a minute to tour this graphic.

The fragment formats

The earliest good ideas were the simplest. Antibody engineers took the binding domains and stripped away most of the antibody, and stuck two of them together.

That produces formats built from single-chain pieces, the scFv and VHH shorthand you’ll see on any list. They’re small, easy to design, quick to make, and you can make a lot of them! Plenty of people know how to build them, and there are real drugs in the clinic made this way.

The problem is they don’t always work in the human body. A molecule built this way can look like an invader to the immune system.

Half-life can be short.
The molecule can be immunogenic.
These molecules often don’t have the inherent stability that a natural antibody has.

So, this route gives you something that binds great, but can also be an engineering nightmare.

The hybrid formats

There’s a “middle” group worth considering, and it's usually written on format lists as something like scFv-Fc or VHH-Fc. The name tells you exactly what it is: a fragment-based binder on one end and a full antibody Fc on the other. Some people call them hybrids here because that's what they are: part engineered fragment, part natural antibody.

The idea is to keep what's easy about the fragments while borrowing back what makes an antibody behave like one. You can take those fragment-based binders and attach them to an Fc, the vertical stem of the Y. The Fc matters because it’s how the molecule interacts with the immune system, and it embodies a lot of what makes an antibody behave like an antibody.

So these hybrids look more like the real thing. This architecture eliminates some challenges, but the binding ends are still assembled rather than natural, so they carry some of the same baggage.

The IgG-like formats

The other approach is to keep the natural antibody shape and solve the assembly problem.

The payoff is predictability. In the body, an IgG-like bispecific behaves much more like a normal antibody. There’s a lower immunogenicity risk. These antibodies deliver a half-life of roughly a week for a human IgG, which tells you something about dosing before you run a single study. You likely get better stability on the shelf and in the vial.

To get there, you have to manage how an antibody is built from its four chains: two heavy and two light. Of course, in a normal monoclonal, the two halves are identical, so the chains only have one correct way to come together. In a bispecific, they’re different, and now four chains are looking for partners.

This presents two engineering challenges. The graphic below shows Challenge 1 (heavy chain pairing) and Challenge 2 (light chain mispairing), with the various attempted solutions: knobs-into-holes, common light chain, and combined approaches.

I’ll briefly unpack each problem here.

Problem 1: Getting the heavy chains to pair

You need the two different heavy chains to pair with each other rather than with themselves. This one is largely solved using the knobs-into-holes design.

You engineer a bump on one heavy chain and a matching groove on the other, so they fit together preferentially.

It was originally developed at one of the large biotechs; it’s been used everywhere for years, and it works well. The B-Body platform uses this approach. It’s a conventional technique that most IgG-like platforms build on. It solves one of the two problems because the format is virtually error-proof.

Problem 2: Light chain mispairing

This is where platforms actually differ.

In nature, each light chain binds with the heavy chain it’s supposed to, because they evolved together. In a bispecific antibody, the two light chains are different, and the left one has to find the left arm while the right one finds the right arm.

When that goes wrong, you get mispaired molecules, impurities, and poor yield. At the kilogram scale in a steel tank, that’s expensive.

Three approaches are used to address light chain mispairing.

The common light chain approach

The simplest fix is to remove the problem. Use the same light chain on both arms. If there’s only one light chain, it can’t pair incorrectly.

It works, and molecules built this way are considerably easier to manufacture. The cost is your design space. You wanted the left arm to bind epitope A and the right arm to bind epitope B, and now both arms have to share a light chain that works well enough with both. You’re no longer choosing the best binder for each target. You’re choosing the best binder that happens to tolerate a shared partner.

For some programs, it’s a perfectly reasonable trade. If the binders you get are good enough, you have an easier molecule to make. The trade-off is you gave up options to get there, and you gave them up before you knew what you’d find.

The electrostatic steering approach

Another approach keeps two different light chains and engineers electrostatic charges into the interfaces, adding or removing positive and negative charge so each light chain is pushed toward the correct heavy chain and away from the wrong one.

These are viable. There are drugs in the clinic built this way. But it’s worth seeing them for what they are: a set of mutations designed to work around a problem on the underlying architecture.

You’re steering chains that would otherwise pair incorrectly, rather than making it structurally hard for them to fail.

The domain switching approach

Another approach changes the architecture. Instead of nudging the chains, you swap out one of the domains that drives pairing so the two arms are no longer alike, making them unlikely to pair incorrectly.

This is the closest approach to what we do at Invenra, and generally the strongest of the three. As you assess the options, from fragments, to common light chain, to charge steering, to domain switching, the molecules get more IgG-like and the platforms improve.

B-Body sits at the far end of that progression, so let’s dig deeper.

Where the B-Body platform fits

The B-Body platform uses a domain switch technique. The specific swap we made is what gives the platform its properties.

Here’s a look at the B-Body design:

The B-Body architecture features knobs-into-holes Fc for heavy chain pairing, proprietary CH3 domain pairs substituting for CH1/CL in one Fab arm, and the plug-and-play variable domains.

In one Fab arm, we remove the native CH1 and CL domains and replace them with a proprietary CH3 pair, along with some internal mutations. The result is two arms that no longer resemble each other at the pairing interface.

The light chains won’t go to the wrong side because the surfaces they need to bind with aren’t similar enough to be confused. On the other arm, the natural CH1 and CL still zip together the way they always have, with multiple contact points holding it stable.

This elegant design has two advantages, and this is what I stress the most when describing the format:

You keep your preferred binders. Because the engineering happens in the constant domains and not the variable ones, you aren’t constrained in which antibodies you can use. Any well-behaved mAb will work well. No common light chain, no sequence compromise, no giving up your best binder because it doesn’t tolerate a shared partner.
One arm carries a single CH1, and that turns out to be worth a lot. Because only one arm has a CH1 domain, we use it as a purification handle. An anti-CH1 resin grabs only chains carrying that domain. We deliberately express the other chains in excess, so when we purify, what comes off is overwhelmingly the correctly assembled molecule.

That matters twice!

In screening, it means a one-step, plate-based purification of hundreds of molecules at a time. We can build a large panel in the real final format, purify it quickly, and test what actually works.
Then in manufacturing, the same molecule switches to conventional purification: standard Protein A on the Fc, then ion exchange polishing. No custom process, no special resin required downstream. Most platforms optimize for one of them. B-Body serves both.

A good bispecific is more than two good binders

Here’s the thing that gets overlooked when people compare platforms on binding data alone:

Nature never had to solve the problems we have. An antibody made in your body doesn’t need to sit in liquid on a shelf for six months, or survive being made by the kilogram in a steel tank, or hold up through freeze-thaw. It gets made, and it gets used. When we make an antibody as a drug, we’re of course asking to survive a complex manufacturing and supply chain on top of binding well.

When you evaluate a platform, binding is the entry ticket, not the answer.

The questions that decide whether your molecule becomes a drug are about everything else: yield, purity, stability, viscosity, immunogenicity, and whether what you saw in screening delivers at scale. That’s the standard I’d hold any platform to, including ours.

Here’s where the B-Body platform stands out:

Purity across a lot of molecules (not a hand-picked few)

To date, we’ve made more than a thousand different B-Body bispecifics using variable domains from many sources.

Across that set, over 95% of 1×1 molecules hit their purity target without any optimization. That’s the number I’d point to first, because one clean molecule proves nothing about a platform. A thousand starts to tell you something about how the platform behaves on your molecules

The purity distribution figure from transient CHO expression. Grey shows the main peak purity after anti-CH1 capture. Green shows the same after polishing. Here, we’re illustrating consistency across diverse variable domain sources.

Yield, and whether early results predict success

With the B-Body platform, stable CHO clones run 6 to 11 g/L. For context, a typical bispecific results in 2 to 4 g/L. That difference shows up directly in manufacturing costs.

The part we find more useful, though, is the predictability. Expression tracks sensibly from transient to pool to clone, and early product quality is a good indicator of what you’ll see later. Late-stage manufacturing surprises are among the most expensive things that can happen to a program, and a platform where early data translates into manufacturability lowers that risk.

The progression figure shows expression yields of 0.1–1.0 g/L transient, 3–5 g/L pool, and 6–11 g/L stable clone, demonstrating that early product quality predicts manufacturing outcomes.

Formats without switching platforms

Not every program team wants the same bispecific shape.

A 1×1 binds each target once and suits most therapeutic applications, including tumor targeting and immune cell redirection.
A 2×1 gives you avidity on one side, which helps when one antigen is highly expressed and the other isn’t.
A 2×2 is bivalent on both and fits blocking or bridging work, including bi-paratopic designs.

With the B-Body platform, all three use the same discovery and manufacturing setup, so you can choose a format based on your biology rather than because it is what your platform can handle. (The more complex formats do give up some titer and purity relative to a 1×1, which is what you’d expect.)

Here’s a family portrait of the B-Body molecules:

The Invenra B-Body family. The B-Body platform supports three bispecific formats—1x1, 2x1, and 2x2—all with the same high expression and manufacturability. Most therapeutic programs start with 1x1 format for simplicity. If you need avidity or bi-paratopic binding, we can rapidly prototype 2x1 or 2x2 versions.

Titer and purity numbers for all three B-Body formats maintain excellent manufacturability with just one-step purification. Even our most complex 2x2 format achieves a transient titer of 170 mg/L and 85% purity without optimization (performance that typically requires extensive optimization on other platforms). Choose the format based on your biology and mechanism, not manufacturing constraints.

Getting it into a patient

A molecule that can’t be concentrated can’t be given as a subcutaneous injection, which limits how patients receive it.

B-Body bispecifics have reached up to 200 mg/mL while staying under the viscosity limit for subcutaneous delivery, and they hold up through the stress of UF/DF processing.

On the safety side, B-Body molecules show cytokine release profiles and T-cell proliferation comparable to an approved clinical antibody control, and pharmacokinetics in non-human primates that look like a normal human IgG.

Viscosity vs. concentration of BsAb A in excipient-free formulations showing the concentration reached while staying under the 15 cP subcutaneous injection limit. Most bispecific platforms struggle with aggregation above 100 mg/mL, limiting dosing options. B-Body's high solubility gives you strategic flexibility: Sub-Q delivery, smaller volumes, less frequent dosing.

What format fits?

The B-Body approach is strongest when what you want is a human IgG-like bispecific antibody that binds two different epitopes. That’s what it was built for, and that’s where it’s superior to other formats.

But there are good ideas that don’t require two epitopes. Say you want one Fab arm to find a tumor, and then you want to position a cytokine off the bottom of the molecule. Or you want one side to be an antibody arm and the other side to be a ligand that binds a receptor. Those are legitimate multispecific drugs, and people are building them.

At some point, though, the molecule stops being an antibody, and once you’re there, our platform isn’t adding much. Fragment-based and appended formats may serve you better.

If what you really want to make is a bispecific antibody, the B-Body platform has been the logical fit for most of the teams we talk to. If you’re building something that stopped being an antibody a while ago, pick a format accordingly.

A quick comparison worksheet

If you only take one thing from this guide, take the questions below. Use them when you are evaluating the B-Body platform or another approach.

Ask this	Because
Does this bispecific format stay IgG-like?	It sets your half-life, your immunogenicity risk, and how easy your antibody transfers to manufacturing.
Can I use my binders?	Some platforms require a shared light chain, which narrows what you’re allowed to pick before you’ve seen the data.
Do I screen the real molecule?	If you screen a simplified version and reformat later, behavior can change, and you find out late.
What happens at the CDMO?	A platform that needs a custom purification process moves cost and risk downstream.
Can you show me more than one molecule?	Anyone can show a good result once. Ask what the distribution looks like across many.
Can it be dosed the way I need?	Concentration and viscosity decide whether subcutaneous delivery is even on the table.

Most platforms answer two or three of these well. The reason we built the B-Body platform the way we did was to answer all of them, and the reason we put the comparison in writing is that the data is there to back it up.

Talk to us

Tell us what you’re building and what the molecule has to do. We’re happy to give you a straight read on whether the B-Body platform fits or when it doesn’t, and to evaluate your needs against our approach and other specific platform providers.

Contact us »

Or start with the data: The B-Body platform brochure has the yield, purity, formulation, and matrix screening data referenced above, in one file you can forward to a colleague. Download our B-Body data sheet »

Quick Path to POC

The Invenra Team — Tue, 28 Jul 2026 21:17:52 GMT

Anti-CH1–based purification, combined with an optimized DNA ratio, is the most efficient and cost-effective approach for early, transient expression of B-Body multispecific antibodies, while Protein A is better reserved for later, stable cell line–based manufacturing.

Multispecific antibodies are precision-engineered proteins requiring correct pairing of four or more distinct peptide chains. Unlike a standard monoclonal antibody, which pairs two identical heavy chains with two identical light chains, a multispecific must pair the right chains. Any single chain over- or under-expressed leads to aberrant homodimers or incomplete constructs.

Protein A complicates early POC work

During discovery and proof of concept stages, most programs rely on transient transfection, introducing plasmid DNA directly into a pool of CHO cells for rapid expression. Each cell randomly takes up the four plasmids, so while the pool on average expresses balanced amounts of each chain, individual cells are not optimized. During purification, Protein A captures both correctly and incorrectly assembled antibodies through conserved Fc and Fab regions leading to a higher fraction of misassembled protein, requiring additional purification steps and project-specific optimization of DNA chain ratios.

Anti-CH1 improves purification efficiency

For optimized transient expression, Invenra uses a two-part strategy.

1) Adjust input DNA ratios to push average chain expression toward a specific balance. 2) Purify using an anti-CH1 resin, which, in the B-Body platform, captures a specific assembled product rather than binding everything with an Fc domain.

This approach sidesteps the costly project-specific scouting work required by Protein A purification at the transient stage. Anti-CH1 purification enables high-throughput production of hundreds of bispecifics per month at suitable purity and yield for preclinical assays. It is also automation friendly and scalable across many projects, saving both time and money.

Cost, time, scalability and stage-appropriate tools

Anti-CH1 purification is the right strategy for rapid, cost-effective screening of B-Body multispecific antibodies during early development. The process delivers >95% purity after a 2-step process.
Once a program advances into cell line development and stable clones are selected, it naturally transitions to the lower-cost, platform-compatible Protein A approach. When used in a manufacturing environment with stable, balanced cell lines, correct assembly routinely exceeds 85%, enabling a standard two step Protein A + CEX purification to yield therapeutic quality material.

We hope this clarifies the tradeoffs so you can select the expression and purification workflow that best fits where your program is today.

A Crank of the Invenra Discovery Engine Produces Therapeutic Quality Antibodies

Paul Guyett, PhD — Tue, 28 Jul 2026 18:24:58 GMT

The Challenge: Speed Without Compromise

The traditional antibody discovery paradigm relies on animal immunization (WT or transgenic). It is a proven process but one with lengthy timelines and limited control over binder outcome and characteristics [1][2]. For multispecific antibody programs, where binder combinations develop emergent properties specific to functional success, having control of epitope, affinity, and functional drivers is imperative.

Invenra has established a robust and proven monoclonal antibody discovery engine focused on individual project requirements. Drawing on our experience delivering therapeutic-quality antibodies to partners in immunology and oncology, we refined and streamlined our selection and screening campaigns to meet the most critical performance metrics: affinity, diversity, developability, and function. Discovered sequences are well-suited to enter our B-Body® and T-Body™ multispecific antibody platforms for final format candidate assessments.

This paper presents unfiltered metrics from recent projects completed through the Invenra discovery pipeline, offering a clear view of the engine’s performance across key discovery outcomes.

Selections Suited to Project Success

With decades of experience developing multispecific antibodies, we perfected optimization of affinity, epitope specificity, developability, and pharmacokinetics. Our library and selection methodologies align to generate a diverse binder set specific to the project goals. By exploring affinity and epitope space, we develop multispecific therapeutics with the desired emergent properties, including AND-gate logic for improved tumor specificity or the regulation of signaling pathways as well as immunomodulation.

Each Invenra project generates dozens to hundreds of binders in the affinity ranges optimal to assess function, including sub-nanomolar affinities (KD < 1 nM) to 50% of targets and 85% or programs with binders in the single-digit nM (<10 nM) range. These data are from single selection campaigns directly from phage display library output. The number of high-quality binders and epitope diversity coverage rivals what can be obtained from immunized animal sources. However, results are available in weeks not months.

For projects requiring finely tuned binder features, Invenra a a yeast display module to the workflow (Figure 1). The power of sorting individual yeast cells based on the desired properties results in a highly enriched binder set for the feature of interest, whether it is steering toward a certain epitope, affinity, or biological/conditional function.

Figure 1. Selection strategy and outcome with an added yeast surface display (YSD) module added. A. Workflow starting with phage display to enrich the naïve library for target binders before reformatting and enrichment through YSD. B. Fewer binders coming through YSD, compared to phage display, but those binders are highly enriched for the feature of interest, an example of which is illustrated in panel C: Affinity profile of a binder panel, where each point represents a unique sequence. D Representative YSD sorting data steering selection toward a specific epitope using a blocking strategy. E Affinity tuning sorts, where gate is highly selective for tightest affinity binders. F Cell binding outcome from a YSD selection that used cell-panning with cancer cells to enrich yeast that functionally bind cells.

Germline Diversity Unlocks Epitope Coverage

The Invenra phage libraries present a variety of germlines across both heavy and light chain families, with parallelized selection strategies that give each germline equal opportunity to enrich based on functional superiority. The strategically selected germline frameworks convey diverse paratope architecture. It can access a correspondingly diverse set of epitopes, the currency of therapeutic antibody discovery [4][5]. Data from recent projects demonstrate target-specific germline enrichment patterns. For example, Cancer Target 3 selections favored IGHV1-46, Cancer Target 13 enriched IGHV1-69, while immune targets 1 and 2 recruited IGHV family 3 germlines (see Figure 2). Germline/target preferences emerge across projects for both heavy and light chains and illustrate the advantage of the Invenra library: multiple shots on goal. Discovery teams routinely advance hundreds of unique clones per target to functional screening, each exploring distinct epitopes and binding modes. Complex applications like bispecific AND-gate logic or biparatopics require testing a range of affinities (picomolar to low nanomolar) and diverse epitopes to avoid on-target toxicity in normal tissues. The Invenra Discovery Engine is an ideal solution for these applications.

Figure 2. Heavy and light chain germline diversity discovered per target. Annotated discovered sequences and identified germlines using PipeBio, an antibody-focused bioinformatics platform.

Designed-In Developability: Clinical-Grade Sequences by Default

Therapeutic antibody failure often occurs late—during developability assessment or clinical trials—when sequence liabilities trigger immunogenicity, aggregation, or manufacturing challenges. Invenra eliminates this risk at the source through liability-limiting library design. Every CDR, including the notoriously variable CDRH3, is engineered to minimize known liability motifs (deamidation sites, unpaired cysteines, isomerization-prone sequences, etc.). The platform requires little-to-no candidate sequence filtering or tuning.

Antibodies identified in Invenra discovery campaigns show liability profiles indistinguishable from approved clinical antibodies, building confidence in the later-stage candidacy of the diverse binder pools generated by the platform.
The analysis includes liability scores for all sequences found through the selection process, not just those cherry-picked from top performers, providing a representative view of developability across the full discovery output (Figure 3).

Figure 3. Distribution of liabilities discovered from the Invenra library (gold) versus antibody therapeutics in the clinic (blue). Sequences were scored dependent on the severity of the liability using PipeBio (www.pipebio.com).

Importantly, Invenra libraries are built entirely on human germline frameworks with CDR diversity constrained to naturally occurring patterns within those frameworks. Humanness analysis using HuMatch [6] confirms that binders from the library score better than clinical-stage antibodies for human identity, dramatically reducing the risk of anti-drug antibodies (ADAs) in patients (see Table 1). In contrast, mouse-derived antibodies require humanization—a process that risks affinity loss and introduces immunogenicity risk. The Invenra human germline framework eliminates this entirely [7].

Source	% Human VH (>0.95)	% Human VL (>0.95)
Invenra	90%	96%
Clinical	75%	58%

Table 1. Humanness scoring of antibody sequences discovered from the Invenra library versus sequences of clinical antibodies. Scoring was performed using HuMatch [6].

Function First: Integrating Cell-Based Screening into Discovery

Affinity alone does not predict therapeutic success. The most critical—and historically most delayed—readout is functional activity in disease-relevant cellular models. Invenra reduces project timelines by integrating functional assessment directly into early discovery.

Dedicated cell biology and immunology teams design project-specific assays: target binding on cancer cell lines, pathway activation/blockade in reporter systems, or even rapid progression to in vivo mouse efficacy models. By moving these powerful assays upstream, Invenra creates a highly effective purpose-built functional screening filter. The data demonstrate a diverse set of target-specific success profiles, underscoring the value-add in this early functional screen. Normalized activity signals consistently identify numerous lead candidates (see Figure 4).

Figure 4. Distribution of the binder’s project-specific functional activity. Cancer targets (CT) were tested for cell binding, while immune targets (IT) were assessed dependent on the project goals (cell binding, immune assay, or reporter assays). The data are normalized across assays using internal controls.

Complete Characterization Package Delivered in 8 Weeks

Contrast the Invenra integrated workflow with mouse immunization timelines:

Milestone	Immunization Timeline	Invenra Timeline
Antigen preparation	Weeks 1-2	Weeks 1-2
Immunization (multiple boosts)	Weeks 3-9	N/A
Serum titer testing	Week 10	N/A
Selection	Weeks 11-12	Weeks 3-4
Sequence triage & expression	Weeks 13-14	Weeks 4-5
Affinity characterization	Weeks 15	Week 6
Functional screening	Weeks 16-17	Weeks 6-7
Lead identification	Week 18	Week 8
Humanization and Recharacterization (if using WT mice)	Weeks 18-26+	N/A
Total Time	5-7+ months	8 weeks

Table 2: Timeline comparison between mouse immunization and synthetic library discovery approaches

By Week 8, Invenra delivers a complete characterization package:

Up to 336 unique sequences characterized by a battery of high throughput assays:

Binding properties and kinetics measurements
Functional activity in disease-relevant cellular assays
Developability assessment (liabilities, humanness, surface properties, stability measurements and predictions)
Expression and purification data in mammalian systems
Sequences formatted for immediate multispecific engineering

This acceleration isn't incremental. It's transformational. Programs move from antigen to lead antibody selection in months.

Key Differentiators: Why the Invenra Platform Stands Apart

1. Naïve Library with Immunized-Library Performance

Most naïve libraries struggle to generate high-affinity binders, often requiring additional affinity maturation. The Invenra library architecture combines controlled diversity in CDRH3 with data science-driven variability in the remaining VH and VL CDR regions. With multiple germline sources that drop seamlessly into a parallelized, automated workflow, we have proven that naïve selections access paratope space necessary to deliver affinities and target diversity with more exquisite control than somatic hypermutation [8][9][10].

2. Multiple Function-Validated Therapeutic-Grade Antibodies in 8 Weeks

Unlike discovery platforms that prioritize binding and defer function to later stages, at Invenra cell biology and immunology teams are embedded in the discovery workflow. Functional assays are designed in parallel with selections, ensuring that every lead candidate demonstrates project-relevant biological activity before characterization resources are committed.

3. Seamless Multispecific Integration

Discovery output feeds directly into the company's proprietary B-Body® and T-Body™ multispecific platforms. Variable domains discovered from naïve selections require no optimization for heterodimeric assembly—the constant domain architecture drives chain pairing automatically. This plug-and-play modularity eliminates the format-specific reengineering that plagues traditional multispecific programs.

4. Full-Service Workflow: Antigen to Candidate

Invenra controls the entire value chain: upstream antigen design and production all the way through discovery, functional assays and ultimately downstream scale ups and in vivo studies. This end-to-end integration eliminates handoff delays and ensures every stage is optimized for the project goal. Clients receive leads ready for preclinical development, not raw sequences requiring further characterization and optimization.

5. Continuous Improvement

The library and workflow are not static, they evolve. Insights from dozens of completed projects inform our library design. Which germline combinations deliver the best developability profiles? Which CDR3 design strategies maximize functional hit rates? This continuous improvement loop, powered by machine learning on proprietary selection data, ensures the platform does not stand still, using real data generated from real project success criteria to drive innovation.

Conclusion: The Future of Antibody Discovery is Synthetic, Functional, and Fast

If speed and biological function are paramount for your program, consider the Invenra Rapid Discovery Engine. The validated discovery engine delivers:

Speed: 8 weeks to characterized leads
Quality: Down to sub-nM affinities, clinical grade developability
Diversity: Hundreds of unique binders per target, exploring diverse epitopes
Function: Cell-based screening integrated into discovery, not deferred to development
Integration: Seamless progression from discovery to multispecific assembly to preclinical development

For therapeutic programs where time to clinic is measured in competitive advantage and patient lives, the Invenra synthetic discovery platform represents the leading edge of the post-immunization era of antibody therapeutics.

About Invenra

Invenra Inc. is a biotechnology company based in Madison, Wisconsin, specializing in rapid antibody discovery and multispecific therapeutic development. The company's integrated platform combines synthetic library technology, functional screening capabilities, and proprietary B-Body® bispecific and T-Body™ trispecific assembly platforms. Invenra partners with pharmaceutical and biotech companies globally to accelerate antibody programs from target validation to clinical candidate selection.

For more information, visit www.invenra.com | bd@invenra.com

References

[1] Köhler, G., & Milstein, C. (1975). Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, 256(5517), 495-497.

[2] Alfaleh, M. A., et al. (2020). Phage display derived monoclonal antibodies: From bench to bedside. Frontiers in Immunology, 11, 1986.

[3] Bertoglio, F., et al. (2021). A pandemic-enabled comparison of discovery platforms demonstrates a naïve antibody library can match the best immune sources. Nature Communications, 12, 1042.

[4] Shrock, E., et al. (2023). Germline-encoded amino acid-binding motifs drive immunodominant public antibody responses. Science, 380(6640), eadc9498.

[5] Briney, B., et al. (2012). Commonality despite exceptional diversity in the baseline human antibody repertoire. Nature, 566(7744), 393-397.

[6] Chinery, L., Jeliazkov, J. R., & Deane, C. M. (2024). HuMatch: fast, gene-specific joint humanisation of antibody heavy and light chains. mAbs, 16(1), 2434121.

[7] Gao, S. H., et al. (2013). Hybridoma technology revisited: Current status and future perspectives. Acta Pharmacologica Sinica, 34(10), 1273-1279.[8] Frenzel, A., et al. (2016). Phage display-derived human antibodies in clinical development and therapy. mAbs, 8(7), 1177-1194.

[9] Kügler, J., et al. (2015). Generation and analysis of the improved human HAL9/10 antibody phage display libraries. BMC Biotechnology, 15, 10.

[10] Xu, J. L., & Davis, M. M. (2000). Diversity in the CDR3 region of VH is sufficient for most antibody specificities. Immunity, 13(1), 37-45.

Trispecific Antibodies: What's Possible Right Now

Roland Green — Tue, 28 Jul 2026 13:16:25 GMT

Bispecific antibodies do something a single antibody can’t: bind two targets within one molecule. Enough of them have reached the clinic that the format is now an established drug class, and more programs start every year.

The concept is straightforward. However, making a bispecific that works as a drug is where some programs get stuck.

The hard part is execution. A bispecific has to assemble correctly, express at a yield you can manufacture, and hold together through the path to the clinic. The commercially available antibody platforms solve (or attempt to solve) those problems in very different ways. Those differences matter more than any vendor website makes clear.

This guide is for anyone deciding how to make a bispecific, or comparing one discovery platform against another. I’ll briefly cover what a bispecific actually is, the assembly problem every platform has to solve, how the molecules get discovered, and the specific numbers worth asking a provider for before you commit.

Backing up: what is a bispecific antibody?

Just to make sure we’re on the same page, a regular antibody has two identical arms, and both grab the same target. A bispecific antibody has two different arms, so one molecule can bind two targets at once.

That does things a single antibody cannot: pull an immune cell up against a tumor cell, block two signals at the same time, or grab two spots on one target to hold on more tightly.

The idea is decades old. Making bispecifics that behave like real drugs is the part that stayed hard. Most of the difficulty comes down to one thing: assembly.

Bispecific antibodies have emerged as one of the most promising therapeutic modalities, validated both clinically and commercially, with the ability to unlock new mechanisms of action. This is exemplified in that the development of this class of biological therapeutic is growing at an accelerated pace compared to more standard monospecific approaches,” said Emily M. Leproust, CEO and co-founder of Twist Bioscience.

The problem: two arms, wrong pairs

An antibody is built from four chains, two heavy and two light, and each arm is a heavy chain paired with its matching light chain. In a normal antibody, the two arms are identical, so there is only one correct way for the chains to come together.

In a bispecific, the two arms are different, and now the chains can mix and match. A heavy chain from one arm can grab the light chain meant for the other. Do that across a batch and you get a mess of mispaired, non-functional molecules instead of the one you designed. Add more variety and the number of wrong combinations climbs fast.

Platforms deal with this in different ways, and the choice has consequences.

Some give up the natural antibody shape. They fuse fragments together or build scaffolds that are not quite an IgG. That controls assembly, but you lose what a real antibody gives you for free: a long half-life in the body and a manufacturing process that facilities already know how to run.

Others keep the IgG shape but force every arm onto a single shared, or “common,” light chain. That removes the mispairing question and narrows your options at the same time. You can only use binders that happen to work with that one light chain, so you may have to leave your best antibody on the table.

The third route keeps a real IgG and engineers the parts of the antibody that do not touch the target, so the correct chains pair on their own while the binding regions stay untouched. That is the approach behind the Invenra B-Body platform, and the section below on the platform explains how it works.

Bispecific Antibody Discovery in 2026: A Practical Guide

The Invenra Team — Tue, 28 Jul 2026 13:02:59 GMT

The concept is straightforward. But making a bispecific that works as a drug is where some programs get stuck.

The hard part is execution on both the engineering and biological sides. A bispecific has to assemble correctly, express at a yield you can manufacture, and hold together through the path to the clinic. The commercially available antibody platforms solve (or attempt to solve) those problems in very different ways. Those differences matter more than any vendor website makes clear.

This guide is for anyone deciding how to make a bispecific or comparing one discovery platform against another. I’ll briefly cover the assembly problem every platform has to solve, how the molecules get discovered, and the specific numbers worth asking a provider for before you commit.

The challenge of bispecific discovery: Two arms, wrong pairs

Add more variety and the number of wrong combinations climbs fast. Platforms deal with this in different ways, and the choice has consequences:

Some give up the natural antibody shape. They fuse fragments together or build scaffolds that are not quite an IgG. That controls assembly, but you lose what a real antibody gives you for free: a long half-life in the body and a manufacturing process that facilities already know how to run.
Others keep the IgG shape but force every arm onto a single shared, or “common,” light chain. That removes the mispairing question and narrows your options at the same time. You can only use binders that happen to work with that one light chain, so you may have to leave your best antibody on the table.
The third route keeps a real IgG and engineers the parts of the antibody that do not touch the target, so the correct chains pair on their own while the binding regions stay untouched. That is the approach behind the Invenra B-Body platform, and the section below on the platform explains how it works.

How bispecifics get discovered

Before you can build a bispecific, you need binders: antibodies that bind each of your two targets. Some teams have binders on hand, while others need to discover them. When discovery is required, there are two methods to choose from.

In vivo methods immunize an animal and harvest the antibodies its immune system makes.
In vitro methods, such as phage display, screen large libraries of human antibody sequences in a dish, with no animals involved. In vitro methods give you more control over the starting material and avoid the developability surprises that animal-derived sequences can carry.

Here at Invenra, we run in vitro discovery on more than 30 proprietary phage libraries, each holding over a billion primary clones, and screen them in parallel across more than 90 conditions.

Screening the real molecule

A lot of discovery screens a simplified version of the molecule, picks a winner, then reformats that winner into the real bispecific at the end.

The trouble is that reformatting can change how the molecule behaves. A pair that looks strong as a simple construct can express poorly, aggregate, or lose activity once it’s built into the final format. You find that out late, after you have already committed to it.

The alternative is to build and test the actual final-format molecule from the start, and to do it at enough scale that you can compare many candidates side by side.

At Invenra, we do this with a matrix.

Instead of testing arm pairs one at a time, the matrix expresses a grid of bispecific molecules in their final format, in parallel, and characterizes all of them. A typical matrix runs a 12×12 grid in two orientations, which is more than 250 individual molecules built and screened in a single campaign.

The data you use to pick a lead comes from the real molecule, so nothing important changes between screening and manufacturing. The process goes from binder to lead in about four months and returns a ranked set of top B-Body leads for you to review.

The scale shows up in the published data. In one validation, we expressed a 15×15 matrix of clinical-stage antibodies in 1 mL cultures and purified them in a single step.

Over 99% of the combinations met both purity and yield criteria in at least one orientation. Purity was measured by capillary electrophoresis, binding by Octet BLI, and correct assembly by mass spectrometry.

Separately, we've published an expression dataset of 549 protein samples, roughly 493 unique sequences, across six antibody formats from standard IgG1 to trispecific.

A 15×15 matrix of clinical-stage antibodies, expressed in 1 mL cultures and purified in a single anti-CH1 step. Dot color shows purity, dot size shows yield; over 99% of combinations met purity and yield criteria in at least one orientation.

The more real molecules you can test, the better your odds of finding one that works, and the more the data can guide the lead selection instead of guesswork.

What good bispecific performance looks like

An antibody-based drug is more than just antigen binding. You have to make it economically and at scale, and it must be stable enough to get it to patients.

A good bispecific antibody drug must deliver three traits: biological function, manufacturability, and dosing/stability.

Biological function: Beyond good binders

A drug must deliver the right biology, not just stick to the right antigen. It must be specific to its target to avoid off‑target effects and engage the right mechanism of action, whether that means killing through immune cells, driving internalization to deliver a toxic payload, or orchestrating a complex immune response.

New drugs can’t be tested in humans until very late in the development process. We rely on an expert team of drug developers using panels of predictive assays to read the signals early:

Rat PK for exposure and clearance
In vitro immune assays for polyreactivity and effector function
High‑resolution binding analysis on platforms like Carterra (see graph below) to map affinity, kinetics, and epitope coverage

Manufacturability: Can you make it?

For a global supply, the molecule must express at thousand‑liter scale and be produceable in prebuilt manufacturing facilities that use platform processes.

Change is expensive here. If your bispecific needs a bespoke upstream or downstream process, you will feel it in cost, risk, and timelines. We built the B‑Body platform to acknowledge this reality. It accommodates standard monoclonal antibody manufacturing, so the facilities needed to make your B‑Body are readily available.

To determine if a bispecific platform is suitable for drug development, three numbers matter: yield, purity, and developability.

Yield

Yield is the amount of antibody you get per liter of culture. It's a critical component of manufacturability.

B‑Body bispecifics typically deliver 6 to 11 g/L (the green bar in the graph below) from stable CHO cell lines. For comparison, a panel of other bispecific platforms averaged 2.4 g/L (grey). This is the highest‑expressing bispecific platform we're aware of, and it drops right into standard mAb‑like processes.

B-Body bispecific yield (green) against a panel of other bispecific platforms (grey). B-Body runs 6–11 g/L; the panel averages 2.4 g/L.

Purity

Purity is how clean the molecule comes out, and it determines how easily you can isolate the final product.

The B‑Body delivers >80% purity in a single purification step and nearly 100% in a standard two‑column process. The figure below demonstrates purity >95% for all formats, including 1×1, 2×2, 2×1, one‑arm, and an IgG1 control.

That single-step purity holds across formats: after polishing, size-exclusion purity lands at 90 to 95% or better across 1×1, 2×1, and 2×2.

Size-exclusion purity by format, before polishing (grey) and after (green), across 1×1, 2×2, 2×1, trispecific, one-arm, and an IgG1 control. After polishing, purity lands at 90–95% or better.

Process‑focused developability

Manufacturability also depends on how the molecule behaves in the plant.

Highly specialized assays measure parameters like Tm, Tagg, thermostability, and aggregation potential. These are predictive of how a candidate will respond to shear, temperature shifts, and concentration steps in real manufacturing runs.

A molecule that scores well here moves smoothly into platform processes. One that doesn’t require special processes, which dramatically affect the cost of goods or timelines.

Storage, stability, and dosing: Getting to the patient

Being able to manufacture the drug isn’t good enough. You have to deliver it to the patient reliably using current processing and standard routes of administration.

Our B‑Body platform was built with this endpoint in mind. We can test solubility, viscosity, freeze–thaw behavior, long‑term storage, colloidal properties, post‑translational modifications, and more.

The goal is simple: make sure that the molecule you discovered can be formulated at practical concentrations and dosed in ways patients can live with.

The figure below shows viscosity versus concentration for two B‑Body molecules. Both stay within the subcutaneous dosing range (under the lower dashed line) at high concentration, which means they can likely be formulated for injection rather than for IV infusion, which is a real benefit for patient adherence and quality of life.

A drug that checks all three boxes—right biology, plug‑and‑play manufacturability, and patient‑friendly dosing and stability—is the one most likely to survive the long path from discovery to clinic.

Choosing a bispecific format

Not every bispecific has the same shape. The B-Body platform can build three different shapes, and the difference is how many times each arm binds.

The three B-Body bispecific formats. Each colored arm marks a binding site for one of the two targets.

A 1×1 binds each target once. It is the simplest and most IgG-like, with no avidity.
A 2×1 binds one target twice and the other once, which gives avid binding where you want it.
A 2×2 binds both targets twice. The higher valency formats bind more tightly and are more complex to make.

Which one fits depends on the biology:

How densely the target sits on the cell surface.
The therapeutic window you are aiming for.
What you can manufacture.

Because the platform tests formats in parallel in their final configuration, you can compare them on real functional data in weeks instead of committing to one on a hunch.

How the B-Body platform does it

This is a good time to dive into the B-Body platform itself. B-Body keeps the human IgG and moves the engineering into the constant domains, away from the binding regions.

Three changes do the work:

A knobs-into-holes Fc makes the two heavy chains pair with each other rather than with themselves.
Proprietary CH3 domains replace the CH1 and CL in one Fab arm, so the correct light chain pairs on its own with no common light chain required.
And that same arm keeps a single CH1 domain, which allows one-step purification with an anti-CH1 resin.

Here’s what the B-Body scaffold looks like:

The result is a molecule that behaves like a normal antibody, works with binders from almost any source, and scales from bispecific to trispecific on the same architecture.

Because the binding regions are never re-engineered, the antibodies you bring in are the antibodies you actually test.

What you should get at the end

A bispecific discovery program should deliver more than a sequence. The output should be a molecule in its final, manufacturable format, with the data package behind it: expression, purity, developability, binding, and function.

That’s the difference between a molecule and a development candidate that is ready for cell line development. It’s the standard a bispecific discovery service should be held to.

If you're comparing bispecific platforms, ask these questions:

Does it keep a real IgG?
Can you use your own binders, or are you locked to a common light chain?

Does it test the final molecule or a stand-in?
And can it show you yield, purity, and developability data on real molecules, not just a diagram of the mechanism?

A platform that answers all four lets the biology, rather than the method, decide your lead. B-Body lets biology decide, and we’d love to show you more.

Talk to us

If you have two targets and want to see how B-Body would handle them, let’s start the conversation. We’ll walk you through the approach, the formats worth considering, and what the data would look like for your program.

Contact us »

Or start with the data: The B-Body Platform data sheet has the full yield, purity, and matrix screening data referenced above, including the 15×15 clinical-antibody results and the developability profiles. Download our B-Body data sheet »