Allosteric Enzymes and Sigmoidal Kinetics

Hemoglobin is not an enzyme. But it taught biochemists how to think about enzymes that refuse to obey The Michaelis-Menten Model. When oxygen binds the first heme of hemoglobin, the affinity of the remaining three hemes for oxygen rises. This is a molecular whisper passed between subunits through shifts in protein quaternary structure. Aspartate transcarbamoylase (ATCase), phosphofructokinase-1 (PFK-1), and glycogen phosphorylase behave the same way. Their velocity-versus-substrate plots are S-shaped, not hyperbolic. Their activity can be switched on or off by molecules binding far from the active site. This sigmoidal behavior is the kinetic signature of cooperativity. Understanding it mechanistically requires abandoning the single-active-site picture of Michaelis-Menten. Instead, we adopt models built on multi-subunit conformational equilibria. Why does this matter beyond a curve's shape? Sigmoidal kinetics converts an enzyme into a switch. A hyperbolic enzyme responds gradually to substrate concentration change. A highly cooperative enzyme can flip from 10% to 90% activity over a narrow concentration window. This gives cells sharp, digital-like control over flux through a pathway. PFK-1 sitting at the committed step of glycolysis needs exactly this kind of switch. It is tuned further by ATP, AMP, citrate, and fructose-2,6-bisphosphate acting as allosteric effectors. Understanding cooperativity is understanding how metabolism computes decisions using thermodynamics rather than transistors. The physical picture rests on two founding frameworks proposed in 1965: the Monod-Wyman-Changeux (MWC) concerted model and the Koshland-Némethy-Filmer (KNF) sequential model. Both explain how ligand binding at one subunit can influence affinity at another. But they differ sharply in their mechanistic commitments. The question is whether subunits move as a single rigid unit or hinge open one at a time. They make distinguishable, testable predictions about negative cooperativity, the shape of Hill plots, and the effect of allosteric activators versus inhibitors. This article builds the Hill equation from first principles. It derives the concerted (MWC) partition function and the sequential (KNF) alternative. It connects both to measurable kinetic parameters ( $K_{0.5}$ , the Hill coefficient $n_H$ ). It works through PFK-1 and hemoglobin-style oxygen binding as concrete numerical cases. It closes with the biological logic of allosteric regulation, which sets up direct study of Regulation of Enzyme Activity.

Key takeaways

Sigmoidal (S-shaped) saturation curves arise from cooperative ligand binding among multiple subunits, not from a single active site.
The Hill equation is an empirical fit; the Hill coefficient $n_H$ measures the degree of cooperativity, not literally the number of binding sites.
The MWC (concerted) model: all subunits switch together between a low-affinity T state and high-affinity R state; this model cannot produce negative cooperativity.
The KNF (sequential) model: subunits change conformation one at a time upon binding, propagating strain to neighbors; this model can produce negative cooperativity.
Allosteric effectors shift $K_{0.5}$ (K-type) or $V_{max}$ /apparent $L$ (V-type) without touching the chemical mechanism at the active site.
Key idea: Cooperativity converts gradual substrate binding into a steep, switch-like response by coupling the conformational states of multiple subunits through their shared quaternary structure.

Where this lives today

Allosteric regulation is the molecular basis of nearly every major metabolic control point. PFK-1's sigmoidal response to fructose-6-phosphate, tuned by ATP, AMP, citrate, and fructose-2,6-bisphosphate, sets the pace of glycolytic flux in every tissue and is dysregulated in some cancers that upregulate glycolysis (the Warburg effect) partly through altered PFK-1 isoform expression. Aspartate transcarbamoylase, the textbook MWC enzyme, controls the first committed step of pyrimidine biosynthesis and is feedback-inhibited by CTP, the pathway's end product — a mechanism visualized directly in Lipscomb's crystal structures showing the T-to-R quaternary rearrangement upon substrate analog binding. Hemoglobin's cooperative oxygen binding, while not enzymatic, is taught alongside these enzymes because it is the cleanest experimental system for distinguishing MWC from KNF behavior, and its Bohr effect (pH-dependent oxygen affinity) is itself a heterotropic allosteric phenomenon of direct clinical relevance in acid-base physiology. Pharmaceutically, allosteric sites are now prized drug targets precisely because they are less conserved across species and paralogs than active sites, offering more selective inhibition. Allosteric kinase inhibitors, allosteric modulators of G-protein-coupled receptors, and positive allosteric modulators of ion channels (used in some anticonvulsant and anxiolytic drugs) all exploit the same MWC/KNF logic: shifting a pre-existing conformational equilibrium rather than competing directly at a catalytic or orthosteric site. Structural biology techniques — cryo-EM, time-resolved crystallography, and NMR relaxation dispersion — now directly visualize the T⇌R conformational ensembles predicted by these fifty-year-old kinetic models, confirming that sigmoidal kinetics is the macroscopic fingerprint of a genuinely microscopic conformational switch, not merely a convenient curve-fitting trick.

Ask the tutor: How would the Hill coefficient and $K_{0.5}$ change if an allosteric enzyme's oligomeric state shifted from a tetramer to a dimer under a given physiological condition?

The physical picture — thorough prose

Michaelis-Menten kinetics describes a single active site cycling through free enzyme (E), enzyme-substrate complex (ES), and product release, giving the familiar hyperbola

v_0 = \frac{V_{max}[S]}{K_M + [S]}

A hyperbola has a defining geometric property: it rises fastest at $[S]=0$ and its slope only decreases thereafter. An allosteric enzyme's curve does the opposite at low $[S]$ — it starts nearly flat, then steepens, then flattens again as it approaches $V_{max}$ . This inflection is impossible for any single, independent binding site obeying simple mass action; it requires communication between sites. The physical vehicle for that communication is the enzyme's quaternary structure: multiple subunits (protomers), each with its own active site, arranged so that a conformational change in one subunit is mechanically or electrostatically transmitted to its neighbors via the subunit-subunit interfaces. Two conformational states are typically invoked for each subunit: a tense (T) state, with low substrate affinity, and a relaxed (R) state, with high substrate affinity. In an isolated monomer this distinction would be kinetically invisible — you'd only see one hyperbola. In an oligomer, the T↔R equilibrium of one subunit is coupled, through the quaternary contacts, to the T↔R equilibria of its partners. Substrate binding is a ligand that shifts an equilibrium, exactly like a Le Chatelier perturbation: because substrate binds R much more tightly than T, each binding event pulls the population of other subunits toward R too, raising their affinity. This is positive cooperativity, and it produces the sigmoidal curve. The MWC and KNF models formalize two different mechanical assumptions about how that coupling happens, sketched below.

Figure

Sigmoidal vs. hyperbolic saturation curves

Figure

MWC concerted vs. KNF sequential transitions in a tetramer

The core ideas

The empirical description of sigmoidal binding, valid regardless of mechanism, is the Hill equation:

\theta = \frac{[S]^{n_H}}{K'+[S]^{n_H}}, \qquad v_0 = \frac{V_{max}[S]^{n_H}}{K_{0.5}^{n_H}+[S]^{n_H}}

where:

$\theta$ = fractional saturation of binding sites,
$[S]$ = free ligand (substrate) concentration,
$n_H$ = Hill coefficient, an empirical measure of cooperativity (not necessarily an integer),
$K_{0.5}$ (or $K'^{1/n_H}$ ) = substrate concentration giving half-maximal velocity/saturation,
$V_{max}$ = limiting velocity at saturating substrate.

The linearized (Hill plot) form is

\log\!\left(\frac{\theta}{1-\theta}\right) = n_H \log[S] - n_H \log K_{0.5}

Reading: plotting $\log[\theta/(1-\theta)]$ against $\log[S]$ gives a straight line of slope $n_H$ ; $n_H=1$ recovers Michaelis-Menten (no cooperativity), $n_H>1$ signals positive cooperativity, $n_H<1$ signals negative cooperativity. The maximum possible value of $n_H$ equals the number of interacting sites $n$ ; in practice $n_H<n$ because binding is never infinitely cooperative. The MWC concerted model describes an oligomer of $n$ identical subunits existing in only two global quaternary states, T and R, in equilibrium described by the allosteric constant

L = \frac{[T_0]}{[R_0]}

where:

$T_0, R_0$ = unliganded T and R state populations,
$L$ = equilibrium constant for the T⇌R transition with no ligand bound (typically $L \gg 1$ , favoring T),
$c = K_R/K_T$ = ratio of the microscopic dissociation constants for substrate binding to R versus T (typically $c \ll 1$ , since R binds substrate more tightly).

The fractional saturation predicted by MWC is

\bar{Y} = \frac{L c \alpha (1+c\alpha)^{n-1} + \alpha(1+\alpha)^{n-1}}{L(1+c\alpha)^n + (1+\alpha)^n}

where $\alpha = [S]/K_R$ . This reduces to a hyperbola when $L=0$ (no T state) or $c=1$ (T and R have equal affinity, no cooperativity), and it can never produce negative cooperativity because all subunits are locked into the same global conformation at every instant — there is no way for one subunit to become a worse binder than baseline. The KNF sequential model instead allows each subunit to flip T→R individually upon binding, with an interaction (coupling) factor $K_{ij}$ describing how a neighbor's conformational change alters the local subunit's intrinsic affinity. Because subunits can adopt mixed conformations, KNF can generate either positive or negative cooperativity depending on the sign of the coupling — a genuine mechanistic distinction from MWC.

Derivation AMWC partition function and the tetramer saturation curve

Goal: derive the MWC binding polynomial for an $n$ -subunit oligomer and show it collapses correctly to the Michaelis-Menten hyperbola in the appropriate limits.

Step 1 — State the postulates. MWC assumes (i) the oligomer has $n$ identical, symmetric subunits; (ii) only two conformations exist for the whole molecule, T and R (no hybrids); (iii) ligand S binds independently to each site within a given conformation, with dissociation constants $K_T$ for T-state sites and $K_R$ for R-state sites, $K_R < K_T$ ; (iv) the T⇌R equilibrium is preserved regardless of how many sites are occupied — occupancy only shifts the population between the two pre-existing conformations (a "conformational selection" flavor), not create new states.

Step 2 — Build the partition function for each conformation. For $i$ ligands bound out of $n$ sites in the T state, the statistical weight (relative to unliganded T) is a binomial-type term $\binom{n}{i}(\alpha/c_T)^i$ where we define $\alpha = [S]/K_R$ and let $K_T = K_R/c$ , so binding to T scales as $(c\alpha)^i$ . Summing over all occupancy states of the T ensemble gives, via the binomial theorem,

Z_T = \sum_{i=0}^{n}\binom{n}{i}(c\alpha)^i = (1+c\alpha)^n

Similarly for the R ensemble,

Z_R = \sum_{i=0}^{n}\binom{n}{i}\alpha^i = (1+\alpha)^n

Step 3 — Weight by the T/R equilibrium. The unliganded T state is more populated than unliganded R by factor $L$ , so the total partition function (relative to unliganded R = 1) is

Z = L\,Z_T + Z_R = L(1+c\alpha)^n + (1+\alpha)^n

Step 4 — Compute average sites occupied. The average number of ligands bound is obtained by differentiating $\ln Z$ with respect to $\ln[S]$ (equivalently, weighting each term by its occupancy $i$ and dividing by $Z$ ). Carrying this out term by term reproduces the standard MWC result:

\bar{Y} = \frac{1}{n}\frac{d\ln Z}{d\ln \alpha} = \frac{Lc\alpha(1+c\alpha)^{n-1} + \alpha(1+\alpha)^{n-1}}{L(1+c\alpha)^n + (1+\alpha)^n}

Step 5 — Check limiting cases. If $c=1$ (T and R bind identically), $Z = (L+1)(1+\alpha)^n$ and $\bar{Y} = \alpha/(1+\alpha)$ : a pure hyperbola — cooperativity vanishes because there is no differential affinity to propagate, exactly as physically required. If $L=0$ (protein permanently in R), $\bar Y = \alpha/(1+\alpha)$ again, a hyperbola, because there is only one conformation present. Sigmoidicity requires both $L\gg1$ and $c\ll1$ simultaneously — the T state must dominate at zero ligand, and it must bind far more weakly than R. This is the essential MWC insight: cooperativity is not a new chemical interaction, it is the thermodynamic signature of a pre-existing conformational equilibrium being dragged by unequal ligand affinities.

Go deeper · Derivation B · 2 minHill coefficient as a limiting slope of the MWC curve

Derivation BHill coefficient as a limiting slope of the MWC curve

Goal: show that the maximum Hill coefficient predicted by MWC never exceeds $n$ , and derive the exact relationship between $n_H$ , $L$ , $c$ , and $n$ near $\bar Y = 0.5$ , connecting the phenomenological Hill equation to the mechanistic MWC model.

Step 1 — Define the Hill coefficient operationally. From the Hill equation, the coefficient is defined as the slope of the Hill plot at half-saturation:

n_H \equiv \left.\frac{d\log[\theta/(1-\theta)]}{d\log[S]}\right|_{\theta=0.5}

This is a purely operational definition — it applies to any saturation curve, hyperbolic or sigmoidal, MWC or KNF or neither.

Step 2 — Relate this slope to the saturation function generally. Using $\theta = \bar Y$ and $x=\ln[S]$ , a standard identity from binding theory gives

n_H = 4\,\bar Y(1-\bar Y)\,\frac{d\bar Y}{d\ln[S]}

This identity is exact for any single-valued saturation function $\bar Y([S])$ and does not assume any particular mechanism; it simply restates the Hill-plot slope in terms of the local sensitivity of $\bar Y$ to $\ln[S]$ .

Step 3 — Evaluate for MWC at low and high $c$ . In the limit where $c \to 0$ (T state essentially cannot bind substrate) and $L$ is large, the transition between T-dominant and R-dominant populations becomes sharp, and the maximum slope of the Hill plot approaches $n_H \to n$ — the theoretical ceiling, achieved only in the idealized limit of "all-or-none" switching between conformations with zero residual T-state affinity. For finite $c>0$ , some ligand always binds T-state subunits directly (non-cooperatively), which "fills in" the sigmoidal curve and reduces the maximal slope below $n$ . Explicitly, for a tetramer ( $n=4$ ) MWC model, typical experimental values fall in the range $n_H \approx 2$ – $3$ , never exactly 4, because $c$ is never exactly zero in real proteins.

Step 4 — Show KNF cannot exceed this bound either, but can go below 1. In the KNF framework, each pairwise coupling constant $K_{ij}$ multiplies the intrinsic affinity of a neighboring, as-yet-unliganded subunit. If $K_{ij}>1$ (positive cooperativity), the sequential build-up of affinity mimics the MWC sigmoid, bounded by the same $n_H \le n$ ceiling. If $K_{ij}<1$ , binding of the first ligand lowers the affinity of remaining sites — negative cooperativity, giving $n_H < 1$ . Because MWC's concerted symmetry preserves conformational homogeneity, it structurally forbids $K_{ij}<1$ behavior: this is the sharp, falsifiable distinction between the two models. Observing an enzyme with $n_H < 1$ (as in yeast glyceraldehyde-3-phosphate dehydrogenase binding NAD⁺, historically one of the motivating cases for KNF) rules out pure MWC and requires sequential-type induced strain.

Step 5 — Numerical sanity check. Consider a tetramer with $L = 200$ and $c = 0.02$ — realistic PFK-1-like values. Solving $\bar Y=0.5$ numerically for $\alpha$ and evaluating the local slope gives $n_H \approx 2.9$ , well below the ceiling of 4 and consistent with the experimentally observed Hill coefficients for PFK-1 with respect to fructose-6-phosphate ( $n_H \approx 2.5$ – $3.5$ depending on species and effector concentration).

Worked examples

Example 1 — Foundational: extracting $n_H$ and $K_{0.5}$ from raw velocity data. An allosteric enzyme is assayed at five substrate concentrations, giving the following initial velocities (as a fraction of $V_{max}$ ):

$[S]$ (mM)	0.5	1.0	2.0	4.0	8.0
$v_0/V_{max}$	0.037	0.138	0.407	0.759	0.941

Construct the Hill plot. First compute $\theta/(1-\theta)$ at each point:

\frac{0.037}{0.963}=0.0384,\quad \frac{0.138}{0.862}=0.1601,\quad \frac{0.407}{0.593}=0.6863,\quad \frac{0.759}{0.241}=3.1495,\quad \frac{0.941}{0.059}=15.95

Taking $\log_{10}$ of both $[S]$ and these ratios:

$\log[S]$	$-0.301$	$0.000$	$0.301$	$0.602$	$0.903$
$\log(\theta/(1-\theta))$	$-1.416$	$-0.796$	$-0.163$	$0.498$	$1.203$

The slope between the extreme points (a reasonable estimate of $n_H$ for this near-linear stretch):

n_H \approx \frac{1.203-(-1.416)}{0.903-(-0.301)} = \frac{2.619}{1.204} = 2.18

This is close to the value chosen to generate the data ( $n_H=2.2$ ), confirming self-consistency. At $\theta=0.5$ , $\log(\theta/(1-\theta))=0$ , which by inspection of the table occurs almost exactly at $[S]=1.0$ mM ( $\log[S]=0$ , giving $\log(\theta/(1-\theta))=-0.796$ , close to but not exactly zero — the true $K_{0.5}$ is very slightly above 1.0 mM). Solving $n_H\log K_{0.5} = 0.796 + n_H\cdot 0$ using the local slope near that point (≈2.19, consistent with the global fit) gives

\log K_{0.5} = \frac{0.796}{2.19}\times(-1)\ \text{correction} \implies K_{0.5} \approx 1.09\ \text{mM}

Limiting-case check: if the data instead gave $n_H=1.00$ exactly at every pair of points, the Hill plot would be perfectly linear with unit slope, and the underlying $v_0$ vs. $[S]$ curve would be exactly the Michaelis-Menten hyperbola — confirming the Hill equation's correct reduction to non-cooperative kinetics when $n_H \to 1$ .

Example 2 — Applied: PFK-1 regulation by ATP and AMP. Phosphofructokinase-1 catalyzes

\text{fructose-6-phosphate} + \text{ATP} \rightarrow \text{fructose-1,6-bisphosphate} + \text{ADP}

and is the primary flux-control point of glycolysis. ATP is both substrate and allosteric inhibitor: at high concentrations, ATP binds an allosteric site (distinct from the catalytic site) and stabilizes the low-affinity T state, increasing $K_{0.5}$ for fructose-6-phosphate (F6P) without lowering $V_{max}$ — a classic K-type (heterotropic) allosteric effect. Suppose at low ATP (physiological "fed" conditions) PFK-1 shows $K_{0.5}^{F6P} = 0.10$ mM with $n_H=3.0$ , but as ATP accumulates (signaling abundant energy charge), the enzyme is driven toward T, raising $K_{0.5}^{F6P}$ to $1.0$ mM (same $n_H$ , same $V_{max}$ ). At a fixed physiological F6P concentration of $[S]=0.30$ mM, compute the fold-change in velocity between the two ATP states using the Hill equation:

\frac{v_{low\ ATP}}{V_{max}} = \frac{(0.30)^3}{(0.10)^3+(0.30)^3} = \frac{0.027}{0.001+0.027}=\frac{0.027}{0.028}=0.964

\frac{v_{high\ ATP}}{V_{max}} = \frac{(0.30)^3}{(1.0)^3+(0.30)^3} = \frac{0.027}{1.000+0.027}=\frac{0.027}{1.027}=0.0263

\text{fold-change} = \frac{0.964}{0.0263} \approx 36.7

A roughly 10-fold shift in $K_{0.5}$ (driven by ATP occupancy of the allosteric site, an entirely separate binding event from catalysis) produces an almost 37-fold change in flux at a fixed substrate concentration — this is the amplification power of a high Hill coefficient combined with allosteric $K_{0.5}$ modulation, and it is precisely why PFK-1 sits at the committed, rate-limiting step of glycolysis rather than an uncontrolled hyperbolic enzyme. AMP (signaling low energy charge) acts as a heterotropic activator, binding a different allosteric site and stabilizing R, pulling $K_{0.5}^{F6P}$ back down even in the presence of high ATP — the physiological basis of the adenylate energy-charge sensor.

Example 3 — Exam/competition level: distinguishing MWC from KNF using a mixed hybrid experiment. A tetrameric enzyme is engineered so that exactly two of its four subunits carry an active-site mutation abolishing substrate binding (call this the "2:2 hybrid"). The wild-type tetramer shows $n_H=3.2$ . Two hypotheses are proposed: (a) MWC concerted model: the hybrid tetramer still flips entirely between T₄ and R₄ as a whole (only two of the four sites can ever be occupied, but the global conformational switch still occurs in an all-or-none fashion governed by the same $L$ and $c$ ). (b) KNF sequential model: each subunit converts individually upon local ligand binding, and a mutant subunit that cannot bind ligand also cannot be induced into R, so it never contributes a "step" to the sequential wave of conformational change; cooperativity should collapse toward that of an isolated dimer of functional subunits. Predict $n_H$ for the hybrid under each hypothesis, and describe the experiment that discriminates. Under MWC (hypothesis a), the two functional sites still experience the identical global T⇌R switch experienced by wild-type, because the switch is a property of the whole oligomer, not of individual site occupancy. Using the general formula with $n=2$ functional binding sites but the same $L$ and $c$ inherited from the tetramer's quaternary architecture (the two dead subunits still participate structurally in the conformational transition, they just cannot bind ligand):

\bar{Y}_{2:2} = \frac{Lc\alpha(1+c\alpha)^{1} + \alpha(1+\alpha)^{1}}{L(1+c\alpha)^2+(1+\alpha)^2}

Numerically solving with the wild-type-derived $L=200$ , $c=0.02$ (values consistent with wild-type $n_H=3.2$ for $n=4$ ) gives a Hill coefficient for the hybrid of approximately $n_H \approx 1.7$ – $1.9$ — reduced from 4-subunit cooperativity but still clearly greater than 1, because the concerted switch persists even though only two sites report on it. Under KNF (hypothesis b), the two dead subunits break the propagation chain; the two functional subunits, now effectively isolated from productive cooperative coupling (each surrounded by non-responsive neighbors), should behave almost independently, giving $n_H \to 1.0$ – $1.2$ , close to non-cooperative Michaelis-Menten behavior. The discriminating experiment is exactly this hybrid-tetramer construction (a classic strategy used experimentally on aspartate transcarbamoylase and hemoglobin-like model systems): if the measured $n_H$ for the 2:2 hybrid remains substantially above 1 (≈1.7–1.9), the concerted MWC mechanism is supported; if it collapses to ≈1, sequential KNF-type propagation is supported. Historically, such hybrid experiments on hemoglobin (using valency hybrids with one or two subunits permanently liganded) showed intermediate cooperativity consistent with elements of both models — hemoglobin's real behavior is now understood as MWC-like at the coarse quaternary level with KNF-like tertiary strain contributing at the subunit level, illustrating that the two idealized models are limiting cases of a more general two-tier allosteric mechanism.

Limiting cases, scope & common traps

Situation	Behavior	Why
$n_H = 1$	Pure hyperbola (Michaelis-Menten)	No cooperative coupling between sites; Hill equation reduces exactly
$n_H > 1$	Positive cooperativity, sigmoidal curve	Ligand binding at one site raises affinity at others
$n_H < 1$	Negative cooperativity	First ligand binding lowers affinity of remaining sites; MWC forbids this, KNF permits it
$c \to 1$ in MWC	Sigmoidicity vanishes even if $L\gg1$	T and R bind substrate equally; no differential pull on the equilibrium
$L \to 0$ in MWC	Sigmoidicity vanishes	Protein is always in R; nothing to switch
$n \to 1$ (monomeric "oligomer")	$n_H=1$ identically	Cooperativity is fundamentally a multi-subunit phenomenon
K-type effector (e.g., ATP on PFK-1)	Shifts $K_{0.5}$ , $V_{max}$ unchanged	Effector shifts T⇌R equilibrium without altering the chemical step at saturation
V-type effector	Shifts $V_{max}$ , $K_{0.5}$ roughly unchanged	Effector alters catalytic efficiency of R state itself, not just its population

Common trapCommon misconceptions

The Hill coefficient is not a literal count of substrate molecules binding simultaneously or in a single step; it is almost always non-integer and represents the degree of coupling, bounded above by the true number of interacting sites $n$ . Do not confuse $n_H = 2.8$ for a tetramer with "2.8 substrates bind per catalytic event" — the reaction stoichiometry is unchanged; only the sensitivity of the binding curve is being described. A second common trap is treating MWC and KNF as competing chemical mechanisms in the sense of alternate reaction pathways. They are competing hypotheses about the topology of conformational states available to the oligomer — concerted (only two global states) versus sequential (many mixed states). Real enzymes, including hemoglobin, often show mixed behavior, and neither idealized model should be treated as the unique correct description without direct structural or kinetic discrimination (for example, via hybrid-subunit experiments as in Worked Example 3).

The chemist behind it

Jacques Monod, working at the Institut Pasteur alongside Jean-Pierre Changeux and Jeffries Wyman, proposed the concerted allosteric model in 1965 after years of studying feedback inhibition in bacterial metabolic pathways — most famously the lac operon system, for which Monod shared the 1965 Nobel Prize in Physiology or Medicine with François Jacob and André Lwoff (awarded for gene regulation work, not allostery itself). Monod's insight, distilled in the phrase he used to describe allosteric proteins as "molecular transducers," was that a protein's quaternary structure could serve as a physical medium for information transfer — a ligand binding at one site changing the probability of a conformational state that affects a distant site, with no direct chemical contact between the two events. This reframed enzyme regulation from a purely chemical question (what bonds form at the active site) into a statistical-mechanical one (what is the free-energy landscape of the whole oligomer's conformational states), directly borrowing the logic of ensemble partition functions from physical chemistry into protein science. Daniel Koshland, working independently, had already proposed induced-fit binding for individual proteins in 1958, and extended it with Némethy and Filmer in 1966 to explain how induced-fit conformational changes could propagate sequentially through an oligomer's subunit interfaces, providing a mechanistic account for the negative cooperativity that concerted models structurally cannot explain.

Origins · 1 minJacques Monod and how this was discovered

Jacques Monod

1910–1976 · Institut Pasteur, Paris; developed the concerted (MWC) allosteric model with Jeffries Wyman and Jean-Pierre Changeux in 1965, building on his earlier work on bacterial enzyme induction and feedback regulation

History

1958

— Daniel Koshland proposes induced fit as an alternative to the rigid lock-and-key model of substrate binding.

1963

— Monod, Changeux, and Jacob introduce the term "allosteric" and the general concept of regulatory sites distinct from active sites.

1965

— Monod, Wyman, and Changeux publish the concerted (MWC) two-state model of allosteric transitions.

1966

— Koshland, Némethy, and Filmer publish the sequential (KNF) model incorporating induced-fit propagation and negative cooperativity.

1970

s–1980s — X-ray structures of hemoglobin (Perutz) and aspartate transcarbamoylase (Lipscomb) provide direct structural confirmation of T and R quaternary states predicted by MWC.

Why can't a single-subunit enzyme show sigmoidal kinetics?

Cooperativity, in the classical allosteric sense, requires communication between multiple binding events, and that communication is physically mediated by protein quaternary structure — shared subunit interfaces that transmit conformational information. A monomeric enzyme with one active site can only exist in a single ligation state at a time (bound or unbound), so there is no "other site" whose affinity could be modulated. Some monomeric enzymes show apparent kinetic complexity (e.g., due to slow conformational isomerization or half-of-sites reactivity in a different sense), but true substrate-binding cooperativity as described by the Hill equation with $n_H>1$ requires at least two interacting sites, almost always achieved through oligomerization.

Is the Hill coefficient ever exactly equal to the number of subunits?

Only in the idealized limit of infinitely tight coupling — for MWC, this requires $c\to 0$ (T state cannot bind ligand at all) combined with $L\to\infty$ in a particular balanced way, an idealization never fully reached by real proteins. Real enzymes always have some residual affinity in the low-affinity state and imperfect coupling, so $n_H$ is always strictly less than $n$ . For hemoglobin ( $n=4$ ), $n_H$ is experimentally about 2.8–3.0, not 4; for PFK-1 tetramers, values of 2.5–3.5 are typical.

How do MWC and KNF differ in their prediction for allosteric activators and inhibitors?

Both models can accommodate heterotropic effectors, but MWC gives a particularly clean picture: an allosteric inhibitor binds preferentially to the T state (shifting the T⇌R equilibrium constant $L$ upward), and an allosteric activator binds preferentially to R (shifting $L$ downward), with the substrate-binding machinery itself unchanged. KNF instead frames effectors as altering the local coupling constants $K_{ij}$ between neighboring subunits, which can produce more varied, position-dependent effects since propagation is directional and stepwise rather than global. Experimentally, K-type effects (changing $K_{0.5}$ without changing $V_{max}$ ) are the classic MWC-style signature seen in PFK-1's response to ATP and AMP.

Does a high Hill coefficient mean the enzyme is more efficient?

Not in the sense of catalytic efficiency ( $k_{cat}/K_M$ ) — a high Hill coefficient is about sensitivity, not speed. It means the enzyme's activity changes very sharply over a narrow range of substrate concentration, functioning as a switch rather than a graded rheostat. This is metabolically valuable when a pathway needs to be decisively "on" or "off" depending on a signal (like PFK-1 responding to energy charge), but it does not imply a faster turnover number or lower Michaelis constant in the ordinary sense.

Can an enzyme be allosterically regulated without showing sigmoidal kinetics?

Yes. Allostery, strictly, refers to regulation via a site distinct from the active site (from Greek allos, "other," and stereos, "shape/site"), and this is a broader category than cooperative substrate binding. Some allosteric enzymes are monomeric or show hyperbolic substrate kinetics ( $n_H \approx 1$ ) yet are still regulated by effectors that raise or lower $K_M$ or $V_{max}$ through a distinct binding site — for example, some monomeric or dimeric regulatory enzymes bind effectors that shift conformational equilibria without generating a homotropic (substrate-substrate) cooperative signal at all.

Why does hemoglobin get used to teach enzyme cooperativity when it isn't an enzyme?

Hemoglobin is the best-characterized system, structurally and thermodynamically, for cooperative ligand binding, and the same MWC/KNF mathematical machinery applies directly to it, since both frameworks were originally developed with hemoglobin in mind alongside allosteric enzymes like ATCase. Its T and R quaternary states have been solved to atomic resolution, its oxygen-binding curve is one of the most precisely measured sigmoidal curves in biochemistry, and its regulation by 2,3-bisphosphoglycerate, CO₂, H⁺ (the Bohr effect), and chloride ions gives a rich, well-understood parallel to heterotropic allosteric effectors in true enzymes.

How is the Hill coefficient measured experimentally in practice?

Initial velocities (or ligand-binding fractional saturation) are measured across a range of substrate concentrations spanning well below and above the apparent $K_{0.5}$ , then plotted as $\log[\theta/(1-\theta)]$ against $\log[S]$ — the Hill plot. The slope of the central, most linear region of this plot (near $\theta=0.5$ ) gives $n_H$ . Modern practice more often fits the full untransformed $v_0$ vs. $[S]$ data directly to the Hill equation by nonlinear regression, which avoids the statistical distortions introduced by the log-log transformation of noisy data, but the conceptual meaning of $n_H$ as a local slope at half-saturation remains the same either way.