Biologics

A small molecule works because it fits a pocket. Most of biology has no pocket: protein-protein interfaces are flat, wide and short of deep cavities. A biologic solves the problem the other way round — it is itself a large surface, shaped to complement the target. Every property of the class follows from that, including the awkward ones.

An antibody is two independent modules

Immunoglobulin G is two functionally distinct devices in one molecule. The Fab arms carry six hypervariable loops (the CDRs) that form the binding site; specificity lives here. The Fc sets almost everything else: it binds the Fc-gamma receptors of effector cells — above all FcγRIIIa on NK cells, the basis of antibody-dependent cellular cytotoxicity — it binds C1q and so triggers the classical complement pathway, and it binds FcRn.

FcRn is why an antibody survives in circulation for weeks rather than hours. A molecule taken up by an endothelial cell through pinocytosis meets an acidifying endosome at around pH 6.0, where residues at the CH2–CH3 interface engage FcRn; the complex escapes lysosomal degradation and returns to the cell surface, where pH 7.4 collapses the affinity and releases the antibody back into plasma. That pH-dependent rescue produces the characteristic half-life of roughly three weeks, and it is a direct engineering handle: mutations that strengthen binding at acidic pH without impairing release at neutral pH extend exposure.

Why a mammalian cell

Fc function depends on the N-linked glycan at Asn297 in each CH2 domain. The glycan is not decoration: it holds the two CH2 domains in the open conformation that FcγR requires. Its composition tunes the effect — removing core fucose raises FcγRIIIa affinity and markedly increases ADCC, while high-mannose species clear faster. Prokaryotic expression produces no such structures, which is why manufacturing runs in CHO and related lines, and why the glycan profile is a critical quality attribute that must be reproduced batch to batch. That, rather than the amino-acid sequence, is what makes a biosimilar difficult.

The ceiling on the class

Three constraints, all physical.

The first is target accessibility. A protein of roughly 150 kDa does not cross the plasma membrane and does not survive digestion, so it is given parenterally and acts on extracellular and membrane targets. The intracellular proteome is closed to the class; conjugates and cell therapies work around this only partially.

The second is aggregation. Partly unfolded molecules expose hydrophobic patches and stick together irreversibly. Aggregates are the best-characterised driver of immunogenicity, because they present a repetitive multivalent array that activates B cells efficiently. The resulting anti-drug antibodies either neutralise the molecule or accelerate its clearance, and this is a systemic property of protein therapeutics rather than an occasional mishap.

The third is viscosity. Subcutaneous injection is limited to roughly a millilitre, so a useful dose demands concentrations near 100–150 mg/mL. At those concentrations reversible self-association, driven by the distribution of charge and hydrophobic patches across the molecular surface, raises viscosity to the point where the solution will not pass through a needle. That limit is addressed by engineering the surface of the molecule, not by reformulating the buffer.