A plain-English primer on multi-specific antibodies

If you have followed antibody drugs for a while, you have watched the field move from molecules that grab one thing to molecules that grab two or three at once. This is a plain-English tour of what those molecules are, why pharma wants them, and why they are genuinely difficult to engineer.

Start with the ordinary antibody

A conventional antibody is a Y-shaped protein. The two tips of the Y bind a target — the same target on both arms — and the stem (the Fc region) talks to the rest of the immune system, recruiting cells and signals. One antibody, one target. That simplicity is why monoclonal antibodies became such a reliable drug class.

But biology is rarely a one-target problem. Disease usually involves several players at once. So the natural question became: what if one molecule could engage more than one target deliberately?

Bispecifics: two targets, one molecule

A bispecific antibody binds two different targets with a single molecule. The classic use is to physically bridge two things that would not otherwise meet.

The most cited example is a T-cell engager: one arm grabs a marker on a cancer cell, the other arm grabs a T cell, and the molecule drags the two together so the immune cell can do its job. Neither arm alone does much. The bridging is the mechanism.

That is the recurring logic of bispecifics — the value is in the geometry, in holding two things at a defined distance and orientation. You are not just blocking a target; you are building a bridge.

Trispecifics: three targets, more conditions

A trispecific adds a third binding arm. Now you can do things like engage a tumour marker, recruit an immune cell, and tune the strength or specificity of that engagement with a third interaction — for example, requiring two markers to be present before the molecule acts, which can sharpen targeting.

Every arm you add multiplies what the molecule can do and multiplies the ways it can fail. Three arms means three binding events that all have to behave, plus an architecture that holds them in the right spatial relationship without getting in its own way.

VHH shuttles: small binders for hard places

A VHH — sometimes called a single-domain antibody or nanobody — is a much smaller binding unit derived from a special class of antibodies. Because it is small and stable, it is a convenient building block, and it shows up often in multi-specific designs as one of the arms.

A shuttle is a particular trick built from these small binders. Some tissues are hard to reach — the brain is the headline example, walled off by the blood–brain barrier. A shuttle arm binds a transport receptor that naturally ferries cargo across that barrier, effectively hitching a ride for the rest of the molecule. The shuttle does not treat anything itself; it improves exposure — how much drug gets where it needs to be.

When we discuss brain-exposure figures for our own shuttle work, those are predicted values or design targets, not measured results. We say so wherever the numbers appear.

Why pharma wants them

The appeal is straightforward once you see the pattern:

• New mechanisms. Bridging, conditional activation, and shuttling are things a single-target antibody simply cannot do.
• Better specificity. Requiring two conditions at once can spare healthy tissue that shows only one of them.
• Reach. Shuttle formats open up compartments, like the central nervous system, that are otherwise hard to dose.
• Fewer molecules, more effect. One designed molecule can replace a combination, simplifying manufacturing and dosing.

Why they are hard

The same complexity that makes multi-specifics powerful makes them difficult to build.

• Chain pairing. With multiple different arms, the protein chains have to assemble in exactly the right combination. The wrong chains can pair up and produce useless or mismatched molecules.
• Geometry and valency. A bridge only works at the right length and angle. How many copies of each arm, and how they are arranged, changes everything about whether the mechanism fires.
• Developability. More parts means more chances for instability, aggregation, and manufacturing trouble.
• You cannot screen your way to it easily. Selection assays look at one binding event at a time, but a multi-specific has to satisfy several constraints simultaneously. The molecule that wins one arm's assay may be wrong for the whole.

That last point is the crux, and it is why we think these molecules call for design rather than pure discovery. You want to specify the geometry, valency, and arms up front and assemble candidates to meet that spec — then evaluate the whole molecule.

Where NOVA-3 fits

NOVA-3 is a computer-aided design suite built specifically for these formats — bispecifics, trispecifics, and VHH shuttles. It composes multi-arm candidates from parts and evaluates them using best-of-breed folding models (AlphaFold 3, Chai, ESMFold); it wraps those models rather than replacing them. The hard, simultaneous constraints that make multi-specifics tough to screen are exactly the constraints design handles well.

If you want the deeper version of the "design, not screen" argument, read Designed, not screened. To see the suite itself, visit the platform page.