Ligand binding outside the orthosteric pocket — conformational stabilization, cooperativity, and PAM/NAM dose-response shifts
Classical pharmacology textbooks depict a receptor as a lock waiting for the right key. Modern structural and single-molecule biophysics paints a very different picture: a receptor is a conformational ensemble, constantly fluctuating between inactive and active-like states on a microsecond-to-millisecond timescale, even in the complete absence of ligand.
Two conceptual models describe how ligands engage a fluctuating receptor. In induced fit, the ligand binds and then forces a conformational change. In conformational selection (now supported by most single-molecule and NMR relaxation-dispersion data for GPCRs), the receptor already visits the active conformation transiently, even without ligand — an agonist simply binds with higher affinity to R* than to R, and by mass action shifts the population equilibrium toward R* by trapping it there once it forms.
This reframing matters enormously for allosteric pharmacology: because R* already pre-exists at low populated levels, any ligand — orthosteric or allosteric — that has differential affinity for R vs. R* will shift the equilibrium, whether or not it occupies the "functional" binding site used by the endogenous hormone or neurotransmitter.
Constitutively active receptor mutants (found in some endocrine disorders) prove the ensemble model directly: a single point mutation can shift the R⇌R* equilibrium toward R* even with zero ligand bound, producing hormone-independent signaling.
The orthosteric site is defined functionally, not structurally: it is wherever the endogenous ligand binds. Because it evolved to recognize a specific natural molecule, it is usually deeply conserved across receptor subtypes — which is exactly why orthosteric drugs so often lack subtype selectivity. Allosteric sites face no such evolutionary constraint.
Muscarinic acetylcholine receptors M1–M5 all bind acetylcholine at a nearly identical orthosteric pocket, making subtype-selective orthosteric agonists extraordinarily difficult to design — an M1-selective drug for Alzheimer's cognitive symptoms risks cross-reacting with M2 (cardiac) and M3 (smooth muscle/salivary), causing dose-limiting side effects.
Allosteric sites, often formed at extracellular loop interfaces or transmembrane helix grooves not involved in the primary recognition event, diverge much more between subtypes because they were never under the same functional selection pressure. This is the central pharmaceutical rationale for allosteric drug discovery programs across GPCRs (mGluRs, muscarinics, GABA-B), ligand-gated ion channels (GABA-A benzodiazepine site, NMDA glycine site) and enzymes (allosteric kinase inhibitors).
A positive allosteric modulator does not activate the receptor on its own (in most cases) — it has little or no intrinsic efficacy alone, but dramatically potentiates the response to the orthosteric agonist by increasing either its binding affinity (cooperativity factor α), its signaling efficacy (cooperativity factor β), or both.
PAM pharmacology is quantitatively described by an allosteric ternary complex model: the receptor (R) can simultaneously bind orthosteric agonist (A) and allosteric modulator (B), forming an ARB ternary complex. Two cooperativity terms govern the system: α (binding cooperativity — does B binding change A's affinity?) and β (activation/efficacy cooperativity — does B binding change how efficiently the ARB complex signals once formed?).
For a PAM, α>1 and/or β>1: modulator-bound receptor binds agonist more tightly and/or transduces a bigger signal per bound agonist. On a functional dose-response curve, this appears as a leftward shift in EC50 (from α) and/or an increase in Emax (from β) — critically, benzodiazepines exemplify a nearly pure α-effect PAM: they have no efficacy of their own and cannot open GABA-A channels without GABA present, which is exactly why benzodiazepine overdose alone is rarely fatal by respiratory depression the way barbiturates (which have direct channel efficacy) can be.
The safety profile of benzodiazepines versus barbiturates is a direct pharmacological consequence of PAM (cooperativity-only) versus direct agonist mechanism — the allosteric ceiling effect is not incidental, it is the reason millions of benzodiazepine prescriptions are relatively safe in overdose.
A negative allosteric modulator produces the mirror-image effect: it stabilizes the inactive conformation, reducing agonist affinity (α<1) and/or efficacy (β<1). Because the NAM never touches the orthosteric pocket, its inhibition is non-competitive — it cannot be fully surmounted by adding more agonist, unlike a classical competitive antagonist.
A defining and clinically important feature of NAM pharmacology is the inhibition ceiling: because the effect operates through cooperativity with a fixed α/β multiplier rather than direct competition for the same site, once the allosteric site is saturated, adding more NAM produces no further inhibition — response is suppressed to a floor set by α and β, not driven to zero.
This contrasts sharply with competitive orthosteric antagonism, where sufficiently high antagonist concentration can drive response arbitrarily close to zero (until receptor reserve is exhausted). The NAM inhibition ceiling is therapeutically valuable when complete blockade of a physiological pathway would be dangerous (e.g., completely silencing NMDA receptor signaling is neurotoxic) — a NAM can dial signaling down without ever risking total shutdown, a mechanism-level safety margin unavailable to classical antagonists.
The mathematical structure of allosteric cooperativity — multiplicative α/β factors acting on a pre-existing agonist signal rather than an independent dose-response of their own — produces a self-limiting pharmacology that is one of the most attractive features of allosteric drugs for chronic, fine-tuned indications.
Because a PAM cannot push signaling beyond the ceiling set by its β factor, and cannot activate the receptor at all in the complete absence of endogenous agonist (for most "pure" PAMs), the pharmacological effect is inherently self-limiting and tracks physiological ligand tone — quieter when endogenous signaling is low, capped when it is high. This is fundamentally different from a direct agonist, whose effect scales with dose until receptor saturation or toxicity, independent of the body's own signaling state.
This property underlies why allosteric modulators are increasingly favored for indications requiring fine, context-dependent tuning rather than blunt on/off control: cinacalcet (CaSR PAM) sensitizes the parathyroid calcium sensor rather than fixing calcium at an artificial setpoint; mGluR NAMs in development for fragile X syndrome aim to dial down excessive glutamatergic tone without abolishing normal synaptic transmission; GABA-A PAMs (benzodiazepines, z-drugs) potentiate inhibitory tone proportional to endogenous GABA release rather than driving the channel open unconditionally.
The pharmaceutical industry's growing allosteric drug pipeline — spanning mGluRs, muscarinics, chemokine receptors and kinases — is driven precisely by this combination of subtype selectivity (Stage 2) and mechanistic ceiling/floor safety margins (this stage), properties that orthosteric drugs structurally cannot offer.