Bispecific and multispecific formats
A bispecific binds two different targets, or two different epitopes on one target. That buys mechanisms a single specificity cannot reach: bridging two cells, blocking two pathways at once, forcing proximity between an enzyme and a substrate, or gaining selectivity by requiring both targets to be present.
The architectures
IgG-like with two different Fabs. Looks and behaves like an antibody, with a long half-life and the usual effector options. The problem is chain pairing: two heavy chains and two light chains can assemble many ways. Engineering solutions include complementary mutations in the heavy chain interface, domain crossovers that make each light chain pair only with its own heavy chain, and charge pairs at the interfaces.
Appended formats. An scFv, a VHH or another binding module fused to an IgG at a terminus. Easy to assemble because the second specificity travels on one chain, at the cost of a less native architecture and sometimes reduced stability.
Fragment-based formats. Tandem scFvs, diabodies, VHH tandems and similar small constructs. Compact, easy to make, short half-life unless an albumin binder or an Fc is added, and often the right choice for cell engagers where small size helps.
Multispecifics. Three or more binding modules, usually built from single-domain blocks, where the geometry between modules is part of the design rather than an afterthought.
Geometry is a design parameter
For anything that works by bringing two things together, the distance and flexibility between the binding sites change potency, sometimes more than affinity does. Linker length, module order and the position of the attachment point are all variables to test rather than to assume, which usually means building a small matrix of constructs rather than one.
Affinity balance
A bispecific rarely wants maximum affinity on both arms. For cell engagers, too much affinity on the ubiquitous target can sequester the molecule away from the rare one. For transport constructs, an arm that binds too tightly to the transport receptor does not release. For selectivity by coincidence, the individual arms are often deliberately weak so that only cells carrying both targets are engaged.
That means affinity engineering on a bispecific is a tuning exercise with a target window, not a race to the tightest binder.
Manufacturing and characterization
Every format has a purity problem specific to its architecture: mispaired species, half molecules, homodimers and aggregates. The analytical package has to be able to see them, which usually means mass spectrometry at the intact and subunit level plus a size-based method, and the purification has to be designed to remove them.
Stability also needs re-checking. Fusing modules creates new interfaces and new liabilities, and a construct assembled from two well-behaved parts is not guaranteed to be well behaved.