Receptor-ligand interface
Also known as: ligand-binding interface, receptor binding site, binding interface
The residues on a receptor protein that make direct contact with its cognate ligand; on receptor-tyrosine-kinase ectodomains and other binding-driven receptors, these residues are conserved for ligand recognition rather than for substrate-pocket geometry.
Source: General structural biology concept; first pipeline catalog instance documented as Boundary #2b in axon_kb/product/esm2-boundary-conditions.md, derived from the INSR αCT helix result (Aslanzadeh et al. 2025, Nat Commun 16:9143).
The receptor-ligand interface is the set of residues on a receptor protein that make direct molecular contact with its cognate ligand. On insulin-receptor-family ectodomains, on growth-factor receptors, on cytokine receptors, and on most binding-driven (non-enzymatic) receptors, the interface residues are evolutionarily conserved because mutation breaks ligand recognition — the receptor’s primary function — rather than because they constitute a substrate pocket on a freely diversifying catalytic enzyme.
Why the Distinction Matters
A receptor-ligand interface is structurally and evolutionarily distinct from an enzyme active site, even though both classes show high conservation at functionally critical residues. The mechanisms are different:
-
Enzyme active site: residues are conserved because they participate in chemistry (substrate binding, catalysis, proton transfer). Mutations at active-site residues alter the catalytic mechanism or substrate preference. Engineers exploit this by deliberately mutating active-site residues to change substrate specificity, and the resulting variants are evolutionarily implausible (the model marks them as deleterious) but functionally desirable (the engineered enzyme has new substrate scope).
-
Receptor-ligand interface: residues are conserved because they participate in molecular recognition — the precise geometric and electrostatic complementarity that lets the receptor bind one ligand and not others. Mutations at the interface alter ligand affinity, kinetics, or specificity. The conservation tracks recognition fitness, not catalytic fitness.
ESM-2 and the Receptor-Ligand Interface
Unsupervised protein language models (ESM-2, ESM-1v, ESM-1b) read evolutionary conservation as a proxy for functional importance. At a receptor-ligand interface, this proxy can invert relative to in-vitro binding fitness measured in a cell-based deep mutational scanning assay: the residues most constrained by evolution for ligand recognition are not always the residues whose mutation most disrupts binding in the assay’s cellular context. The signals are orthogonal at the interface, and ESM-2’s rank ordering can anti-correlate with binding-fitness rank ordering.
This is catalogued in our methodology library as Boundary #2b (receptor-ligand interface), a refinement of Boundary #2 (active-site specificity in enzymes). The first pipeline instance is the αCT helix of the human insulin receptor (UniProt P06213 residues 716-746), where ESM-2’s Spearman ρ against the Aslanzadeh 2025 MAVE insulin-binding score set is −0.088 (n=507, p=0.046) — marginally anti-correlated, statistically borderline, but mechanistically distinct from zero-rank no-signal cases.
Implications for Variant Interpretation
For a clinical reader interpreting a variant of uncertain significance that sits at a receptor-ligand interface, an unsupervised conservation-based predictor’s rank should be treated as evidence about recognition function, not about binding fitness in a specific assay. Functional measurement (a binding MAVE, a cellular signaling assay) is the appropriate evidence source at the interface; computational rank is supplementary, not primary, in that region.