Concept

Enzyme kinetics

T-094Home BU-306Threads structure · energy

Statement

An enzyme that binds one substrate reversibly at a single site and releases product irreversibly converts substrate at an initial rate that saturates hyperbolically with substrate concentration. Formally: for the mechanism \(\text{E}+\text{S}\rightleftharpoons\text{ES}\rightarrow\text{E}+\text{P}\) with mass-action rate constants \(k_1,k_{-1},k_{\text{cat}}\), total enzyme \([\text{E}]_T\) conserved, and the enzyme dilute in the sense \(\varepsilon=[\text{E}]_T/(K_M+[\text{S}]_0)\ll1\), the complex reaches a quasi-stationary concentration within a time \(t_c=1/\bigl(k_1([\text{S}]_0+K_M)\bigr)\), and for all later times the rate of product formation is \(v=k_{\text{cat}}[\text{E}]_T[\text{S}]/(K_M+[\text{S}])\) with \(K_M=(k_{-1}+k_{\text{cat}})/k_1\), the relative error being of order \(\varepsilon\).

Why it matters

An enzyme is described by its structure, but it is measured by two numbers: how tightly it engages its substrate and how fast it turns that substrate over once engaged. Enzymes and activation energy establishes why catalysis is possible at all — the active site stabilises the transition state and lowers the barrier — but says nothing about how the observed rate depends on how much substrate is present. Michaelis–Menten kinetics supplies exactly that missing dependence, and in doing so it defines \(K_M\) and \(k_{\text{cat}}\), the two parameters in which enzymology, pharmacology and systems biology are conducted. Every enzyme in a database is characterised by them; every enzyme-targeted drug is characterised by how it perturbs them.

The model also matters because it is the baseline against which deviations are diagnosed. A hyperbola is the null hypothesis: a sigmoidal saturation curve reports cooperativity (Allosteric regulation), a curve whose maximum rate falls at high substrate reports substrate inhibition, a curve whose \(K_M\) rises with an added compound but whose maximum is untouched reports competitive inhibition. None of those diagnoses can be made without first knowing precisely what shape a single active site working independently is obliged to produce, and precisely under which conditions it is obliged to produce it. Both halves of that — the shape and the conditions — are derived below, the second being the half that textbooks usually assert.

Hypotheses

The mechanism is elementary and mass-action: one substrate, one binding site, one kinetically significant complex, and an irreversible product-releasing step.Rates may be written as products of concentrations only for elementary steps. Enzymes with two substrates (\(\text{NAD}^{+}\)-dependent dehydrogenases, kinases) obey ternary-complex or ping-pong rate laws that reduce to the form below only when one substrate is held fixed, and the resulting \(K_M\) is then an apparent constant that depends on the concentration at which the other substrate was held.

The enzyme is dilute against the substrate scale: \(\varepsilon=[\text{E}]_T/(K_M+[\text{S}]_0)\ll1\).This, not the cruder \([\text{S}]_0\gg[\text{E}]_T\), is the correct condition, and Step 4 derives it. It guarantees that the substrate consumed while the complex is filling is a negligible fraction of the substrate present, so that the complex may be treated as tracking a substrate concentration it has not itself disturbed. When it fails — a nanomolar target and a nanomolar inhibitor, or an intracellular enzyme more abundant than its metabolite — the hyperbola is replaced by a quadratic (see Fails without).

The measurement is an initial rate, taken while \([\text{P}]\approx0\).Every enzyme catalyses both directions (Enzymes and activation energy), so writing the product step as irreversible is legitimate only before product has accumulated. Once \([\text{P}]\) is appreciable the reverse flux and any product inhibition both bite, and the measured \(v\) falls below the prediction — which is why assays are read over the first few percent of conversion.

Observation begins after the transient: \(t\gg t_c\).The quasi-stationary solution cannot satisfy the initial condition \([\text{ES}](0)=0\); it asserts a finite complex concentration at \(t=0\). The true solution needs a few multiples of \(t_c\) to climb onto that branch, and inside this boundary layer the model over-predicts the rate. Stopped-flow instruments are built precisely to resolve this layer, and the pre-steady-state “burst” they reveal is how multi-step mechanisms are distinguished from the single-step one assumed here.

Free substrate equals total substrate: \([\text{S}]\approx[\text{S}]_{\text{total}}\).Substrate sequestered in the complex is at most \([\text{E}]_T\), so this is the same smallness assumption again, but it is a separate arithmetic step and it is the one that fails first in tight-binding assays, where a substantial fraction of the added ligand is bound and the free concentration must be solved for rather than assumed.

The solution is homogeneous, well mixed, and held at fixed temperature, pH and ionic strength.All four rate constants are functions of these variables; \(k_{\text{cat}}\) carries an Arrhenius temperature dependence and \(K_M\) typically carries a bell-shaped pH dependence set by the ionisation states of active-site residues. A “\(K_M\) value” without stated assay conditions is not a measurement.

Proof

The derivation is done in two stages, because the model is a two-timescale statement and a one-stage argument cannot say when it is true. Stage one (Steps 3–4) solves the fast relaxation of the complex exactly, with the substrate held fixed, and extracts from it both the relaxation time and the small parameter. Stage two (Steps 5–8) uses that small parameter to justify the algebraic reduction and reads off the rate law and its parameters.

\[ \text{E}+\text{S}\ \underset{k_{-1}}{\overset{k_1}{\rightleftharpoons}}\ \text{ES}\ \overset{k_{\text{cat}}}{\longrightarrow}\ \text{E}+\text{P} \]

The mechanism. Association is bimolecular and reversible; catalysis and product release are lumped into one irreversible first-order step with constant \(k_{\text{cat}}\), legitimate while \([\text{P}]\approx0\) (Hypotheses). The enzyme is regenerated, so it is a catalyst and not a reagent. A

\[ \frac{d[\text{S}]}{dt}=-k_1[\text{E}][\text{S}]+k_{-1}[\text{ES}],\qquad \frac{d[\text{ES}]}{dt}=k_1[\text{E}][\text{S}]-(k_{-1}+k_{\text{cat}})[\text{ES}],\qquad [\text{E}]_T=[\text{E}]+[\text{ES}] \]

Summing each elementary step's contribution gives the exact system; the third relation is enzyme conservation, and it is what closes the problem, since the free enzyme \([\text{E}]\) is not independently observable. No approximation has been made. The system is nonlinear through the product \([\text{E}][\text{S}]\) and has no closed-form solution. A

\[ [\text{S}]\equiv[\text{S}]_0:\quad \frac{d[\text{ES}]}{dt}=k_1[\text{S}]_0[\text{E}]_T-k_1\bigl([\text{S}]_0+K_M\bigr)[\text{ES}] \ \Longrightarrow\ [\text{ES}](t)=\frac{[\text{E}]_T[\text{S}]_0}{[\text{S}]_0+K_M}\left(1-e^{-t/t_c}\right) \]

Freeze the substrate and the equation for the complex becomes linear, hence exactly solvable. Substituting \([\text{E}]=[\text{E}]_T-[\text{ES}]\) and collecting terms, the coefficient of \([\text{ES}]\) is \(k_1[\text{S}]_0+k_{-1}+k_{\text{cat}}=k_1([\text{S}]_0+K_M)\) once \(K_M\equiv(k_{-1}+k_{\text{cat}})/k_1\) is defined. The complex therefore relaxes exponentially, with relaxation time \(t_c=1/\bigl(k_1([\text{S}]_0+K_M)\bigr)\), onto a fixed point that is already the answer sought. B

\[ \left|\Delta[\text{S}]\right|_{\text{transient}}\approx\frac{[\text{E}]_T[\text{S}]_0}{[\text{S}]_0+K_M}\ \Longrightarrow\ \frac{\left|\Delta[\text{S}]\right|}{[\text{S}]_0}\approx\frac{[\text{E}]_T}{[\text{S}]_0+K_M}\equiv\varepsilon \]

The self-consistency test, and the whole content of the hypothesis. Freezing the substrate is legitimate only if the substrate barely moves while the complex fills. During the transient, substrate is removed almost entirely by sequestration into the complex — turnover contributes only \(k_{\text{cat}}[\text{ES}]t_c\), smaller by the factor \(k_{\text{cat}}/(k_1([\text{S}]_0+K_M))\le1\) — so the loss is the complex concentration itself. Dividing by \([\text{S}]_0\) produces the dimensionless group \(\varepsilon\), and Step 3 is self-consistent exactly when \(\varepsilon\ll1\). Note what \(\varepsilon\) is not: it is not \([\text{E}]_T/[\text{S}]_0\), and at low substrate it is the \(K_M\) in the denominator that rescues the approximation. C

\[ \frac{d[\text{ES}]}{dt}\approx0\ \Longrightarrow\ k_1\bigl([\text{E}]_T-[\text{ES}]\bigr)[\text{S}]=(k_{-1}+k_{\text{cat}})[\text{ES}]\ \Longrightarrow\ [\text{ES}]=\frac{[\text{E}]_T[\text{S}]}{K_M+[\text{S}]} \]

The quasi-steady-state approximation of Briggs and Haldane. For \(t\gg t_c\) the complex has relaxed onto the branch found in Step 3, and because the substrate now drifts on the much slower timescale \(([\text{S}]_0+K_M)/(k_{\text{cat}}[\text{E}]_T)\), the complex follows it adiabatically: at each instant it takes the fixed-point value with the current \([\text{S}]\). One differential equation has been replaced by one algebraic equation. B

\[ v=\frac{d[\text{P}]}{dt}=k_{\text{cat}}[\text{ES}]=\frac{k_{\text{cat}}[\text{E}]_T[\text{S}]}{K_M+[\text{S}]}\equiv\frac{V_{\max}[\text{S}]}{K_M+[\text{S}]},\qquad V_{\max}=k_{\text{cat}}[\text{E}]_T \]

Product appears only through the catalytic step, so substituting Step 5 into it eliminates the unobservable complex and leaves a rate law in measurable quantities alone. \(V_{\max}\) is the rate when every enzyme molecule is bound, an asymptote approached but never attained at finite substrate. Rearranged in symbols before any number enters. A

\[ [\text{S}]=K_M\ \Longleftrightarrow\ v=\tfrac12V_{\max}\ \Longleftrightarrow\ \frac{[\text{ES}]}{[\text{E}]_T}=\tfrac12;\qquad [\text{S}]\ll K_M:\ v\approx\frac{V_{\max}}{K_M}[\text{S}];\qquad [\text{S}]\gg K_M:\ v\approx V_{\max} \]

Three readings of the same hyperbola. \(K_M\) is by construction the substrate concentration at half-saturation, and equivalently the concentration at which half the enzyme population carries substrate — that is its operational meaning, prior to any interpretation as an affinity. The two limits are the two regimes an enzyme is ever observed in: first order in substrate when starved, zero order when flooded. The crossover is broad: going from \(10\%\) to \(90\%\) of \(V_{\max}\) requires an \(81\)-fold rise in \([\text{S}]\). A

\[ [\text{S}]\ll K_M:\quad v\approx\frac{k_{\text{cat}}}{K_M}[\text{E}]_T[\text{S}]\approx\frac{k_{\text{cat}}}{K_M}[\text{E}][\text{S}],\qquad \frac{k_{\text{cat}}}{K_M}=\frac{k_1k_{\text{cat}}}{k_{-1}+k_{\text{cat}}}\le k_1 \]

In the starved limit almost all enzyme is free, so \([\text{E}]_T\approx[\text{E}]\) and the rate law is second order overall: \(k_{\text{cat}}/K_M\) is the effective bimolecular rate constant for a free enzyme molecule meeting a free substrate molecule and going on to product. That is why it, and not \(k_{\text{cat}}\) or \(1/K_M\) separately, is the specificity constant that ranks substrates and measures catalytic efficiency. The inequality follows because \(k_{\text{cat}}/(k_{-1}+k_{\text{cat}})\le1\): the composite can never exceed the association constant, so it is capped by the diffusion-encounter rate, of order \(10^{8}\) to \(10^{9}\,\text{M}^{-1}\,\text{s}^{-1}\) for small metabolites in water. B

\[ K_M=\frac{k_{-1}+k_{\text{cat}}}{k_1}=K_S+\frac{k_{\text{cat}}}{k_1}\ \ge\ K_S=\frac{k_{-1}}{k_1};\qquad k_{\text{cat}}\ll k_{-1}\ \Longrightarrow\ K_M\to K_S \]

The relation between the kinetic constant and the thermodynamic one. \(K_S\) is the true dissociation constant of the complex; \(K_M\) exceeds it by \(k_{\text{cat}}/k_1\), because catalysis drains the complex and so shifts the apparent half-saturation point to higher substrate. Only in the rapid-equilibrium limit — the case Michaelis and Menten themselves assumed in 1913 — do the two coincide. The steady-state treatment of Briggs and Haldane (1925) removes that restriction and leaves the algebraic form of Step 6 untouched; only the meaning of the constant changes. B

\[ -\frac{d[\text{S}]}{dt}=\frac{V_{\max}[\text{S}]}{K_M+[\text{S}]}\ \Longrightarrow\ \int_{[\text{S}]}^{[\text{S}]_0}\left(\frac{K_M}{s}+1\right)ds=V_{\max}t \ \Longrightarrow\ K_M\ln\frac{[\text{S}]_0}{[\text{S}]}+\bigl([\text{S}]_0-[\text{S}]\bigr)=V_{\max}t \]

Integrating the quasi-stationary rate law gives the full progress curve, implicit in \([\text{S}]\) but exact within the approximation. The two terms are the two regimes of Step 7 made explicit in time: the linear term dominates while \([\text{S}]\gg K_M\), giving constant-rate (zero-order) depletion, and the logarithmic term takes over below \(K_M\), giving exponential decay with time constant \(K_M/V_{\max}\). This is the equation behind clinical elimination kinetics and behind fitting parameters to a single progress curve rather than to a series of initial rates. B

\[ \frac{1}{v}=\frac{K_M}{V_{\max}}\frac{1}{[\text{S}]}+\frac{1}{V_{\max}},\qquad v=V_{\max}-K_M\frac{v}{[\text{S}]},\qquad \frac{[\text{S}]}{v}=\frac{[\text{S}]}{V_{\max}}+\frac{K_M}{V_{\max}} \]

The three linearisations — Lineweaver–Burk, Eadie–Hofstee and Hanes–Woolf — obtained from Step 6 by inverting, by multiplying out, and by inverting after multiplying by \([\text{S}]\). They are algebraically equivalent to the hyperbola and to each other, but not statistically equivalent: reciprocal transformation maps the small, noisy velocities measured at low substrate onto the large-leverage end of the plot, so a least-squares line through a double-reciprocal plot is badly biased. They remain the standard diagnostic display, because inhibition mechanisms are distinguished by where the lines intersect. A

\[ \frac{d\mathbf{x}}{dt}=\mathbf{f}(\mathbf{x},\mathbf{y}),\qquad \varepsilon\frac{d\mathbf{y}}{dt}=\mathbf{g}(\mathbf{x},\mathbf{y}),\qquad \mathbf{g}(\mathbf{x},\mathbf{y}_{ss})=0 \]

Formally, Steps 3–5 are the leading order of a singular-perturbation expansion: the transient of Step 3 is the inner (boundary-layer) solution, the hyperbola of Step 6 the outer solution, and Tikhonov's theorem supplies convergence as \(\varepsilon\to0\) provided the fast subsystem's fixed point is isolated and asymptotically stable — here it is, since the linearised loss coefficient \(k_1([\text{S}]+K_M)\) is strictly positive. Recasting the same system in the total substrate \([\text{S}]+[\text{ES}]\) yields the total quasi-steady-state approximation, whose validity condition is far weaker and which remains accurate even when enzyme and substrate are comparable — at the cost of a quadratic, rather than hyperbolic, rate law. C

Result

\[ v=\frac{V_{\max}[\text{S}]}{K_M+[\text{S}]},\qquad V_{\max}=k_{\text{cat}}[\text{E}]_T,\qquad K_M=\frac{k_{-1}+k_{\text{cat}}}{k_1},\qquad \text{error}=O(\varepsilon),\ \ \varepsilon=\frac{[\text{E}]_T}{K_M+[\text{S}]_0} \]

Reading. Rate rises in proportion to substrate while the enzyme is mostly free, and levels off at \(V_{\max}\) once the enzyme is mostly occupied; \(K_M\) is the substrate concentration marking the halfway point of that transition, and \(k_{\text{cat}}\) is the number of substrate molecules a single enzyme molecule converts per second when saturated. Two numbers, one experimentally accessible only as the product \(k_{\text{cat}}[\text{E}]_T\), summarise the entire steady-state behaviour.

Scope. Valid for \(t\gg t_c=1/\bigl(k_1([\text{S}]_0+K_M)\bigr)\) and \(\varepsilon\ll1\), for a single substrate binding a single independent site, and while \([\text{P}]\approx0\). It predicts the shape of the saturation curve and the composite constants; it cannot resolve \(k_{-1}\) from \(k_{\text{cat}}\), since steady-state data depend on them only through \(K_M\) and \(k_{\text{cat}}\) — separating them requires a pre-steady-state experiment that observes the transient of Step 3 directly.

Corollaries & converses

Fractional saturation. \([\text{ES}]/[\text{E}]_T=[\text{S}]/(K_M+[\text{S}])\) is formally identical to a single-site binding isotherm (Receptor-ligand binding), with \(K_M\) in place of \(K_d\). The kinetics and the binding curve have the same shape for the same structural reason: one site, occupied or not.
The \(81\)-fold rule. \(v/V_{\max}=0.1\) at \([\text{S}]=K_M/9\) and \(0.9\) at \([\text{S}]=9K_M\). A hyperbola is a poor switch, which is precisely why regulatory enzymes are cooperative rather than Michaelian (Allosteric regulation).
Catalytic perfection. Since \(k_{\text{cat}}/K_M\le k_1\) and \(k_1\) is bounded by diffusive encounter at roughly \(10^{8}\) to \(10^{9}\,\text{M}^{-1}\,\text{s}^{-1}\), an enzyme with \(k_{\text{cat}}/K_M\) in that range converts essentially every molecule it meets, and no further improvement in the chemistry can raise its rate. Triosephosphate isomerase, carbonic anhydrase and acetylcholinesterase are the standard examples.
Competitive inhibition. An inhibitor binding only free enzyme multiplies \(K_M\) by \(\alpha=1+[\text{I}]/K_i\) and leaves \(V_{\max}\) unchanged: \(v=V_{\max}[\text{S}]/(\alpha K_M+[\text{S}])\). Saturating substrate therefore abolishes the inhibition, which is why competitive inhibitors make poor drugs against enzymes whose substrate is abundant.
Uncompetitive inhibition. An inhibitor binding only the complex divides both \(V_{\max}\) and \(K_M\) by \(\alpha'=1+[\text{I}]/K_i'\), leaving \(k_{\text{cat}}/K_M\) unchanged and giving parallel double-reciprocal lines. Substrate cannot outcompete it; more substrate makes it worse.
Haldane relation. Restoring reversibility, the forward and reverse steady-state parameters are tied to the equilibrium constant by \(K_{\text{eq}}=(k_{\text{cat}}^{f}K_M^{r})/(k_{\text{cat}}^{r}K_M^{f})\). A catalyst cannot alter \(K_{\text{eq}}\) (Enzymes and activation energy), so the four kinetic constants are not independent.
Converse fails. A hyperbolic saturation curve does not establish this mechanism. Mechanisms with several intermediates, and rapid-equilibrium mechanisms, produce the identical functional form with different composite constants; the hyperbola constrains the mechanism only weakly, and the transient of Step 3 is where the mechanisms differ observably.

Fails without

Drop \(\varepsilon\ll1\) — the tight-binding regime: when \([\text{E}]_T\) is comparable with \(K_M+[\text{S}]_0\), a substantial fraction of the substrate is sequestered in the complex, free substrate is no longer the amount added, and the conservation relations \([\text{S}]_{\text{total}}=[\text{S}]+[\text{ES}]\) and \([\text{E}]_T=[\text{E}]+[\text{ES}]\) must be solved simultaneously. The result is the Morrison quadratic, not a hyperbola, and its half-saturation point sits at \(K_M+[\text{E}]_T/2\) rather than \(K_M\). Fitting a hyperbola to such data returns a \(K_M\) that is an artefact of the enzyme concentration used, and repeating the assay at a different enzyme concentration exposes it. This is the routine regime of picomolar-affinity drug-target assays, and of many intracellular enzymes, where enzyme and metabolite concentrations are both micromolar.
Observe inside the boundary layer, \(t\lesssim t_c\) — the pre-steady-state regime: the model predicts a rate that is finite from \(t=0\), whereas the true rate starts at zero and climbs as the complex fills. For a fast enzyme with \(k_1\sim10^{7}\,\text{M}^{-1}\text{s}^{-1}\) and \(K_M+[\text{S}]_0\sim10^{-3}\,\text{M}\) this layer is about \(100\,\mu\text{s}\), invisible to a plate reader and central to a stopped-flow experiment. More than an error, the layer is informative: enzymes whose chemistry precedes a slow product-release step show a stoichiometric burst of product — one turnover per active site — before settling to the steady rate, and that burst is direct evidence that the single lumped step of Step 1 is a simplification.
Drop the single independent site — the cooperative regime: with coupled subunits, binding at one site changes affinity at another, the saturation curve becomes sigmoidal and is fitted by a Hill exponent greater than one (Allosteric regulation, The oxygen dissociation curve). Fitting a hyperbola to sigmoidal data does not merely give an imprecise \(K_M\); it gives a curve of the wrong shape, systematically over-predicting rate at low substrate and under-predicting the steepness of the physiological response.
Drop \([\text{P}]\approx0\): with product present, the reverse of the catalytic step and any product binding to free enzyme both operate. Product acting as a competitive inhibitor multiplies \(K_M\) by \(1+[\text{P}]/K_P\), so a progress curve decelerates faster than Step 10 predicts, and initial-rate parameters fitted to late-time data are biased low. Many pathway enzymes are strongly product-inhibited, which is a control mechanism rather than an experimental nuisance.
Drop the single kinetically significant complex at high substrate: if a second substrate molecule can bind the complex non-productively, the rate law acquires a term \([\text{S}]^2/K_i\) in the denominator and passes through a maximum before falling. A curve that turns down at high substrate is not a bad hyperbola; it is a different mechanism, and extrapolating \(V_{\max}\) from its descending limb is meaningless.

Common errors

“\(K_M\) is the dissociation constant, so a low \(K_M\) means high affinity.” \(K_M=K_S+k_{\text{cat}}/k_1\ge K_S\) (Step 9). The identification holds only in the rapid-equilibrium limit \(k_{\text{cat}}\ll k_{-1}\), which must be demonstrated, not assumed; for a diffusion-limited enzyme it is badly false.
“\(V_{\max}\) is a property of the enzyme.” \(V_{\max}=k_{\text{cat}}[\text{E}]_T\) is proportional to how much enzyme was pipetted. Only \(k_{\text{cat}}\) is intrinsic, and quoting \(V_{\max}\) without \([\text{E}]_T\) makes the number uncomparable between laboratories.
“The steady-state assumption says \([\text{ES}]\) is constant.” It says \(d[\text{ES}]/dt\) is negligible compared with the individual formation and loss rates. \([\text{ES}]\) in fact tracks the falling substrate throughout the assay (Step 5); it is quasi-stationary, not stationary.
“The assumption is that \([\text{S}]\) does not change.” The approximation is applied to the complex, not to the substrate. The substrate is allowed to change; that is what Step 10 integrates.
“The best substrate is the one with the highest \(k_{\text{cat}}\).” At the low substrate concentrations most enzymes actually see in a cell, the ranking is by \(k_{\text{cat}}/K_M\) (Step 8), and the two rankings routinely disagree.
“Fit the Lineweaver–Burk line by least squares.” The transformation inflates the weight of the least reliable points (Step 11). Parameters should be obtained by non-linear regression on the untransformed hyperbola; the reciprocal plot is for looking at, not for fitting.
“\(K_M\) was measured, so it is the enzyme's \(K_M\).” \(K_M\) is defined at a stated temperature, pH, ionic strength and co-substrate concentration, and for a two-substrate enzyme it is an apparent constant that moves when the fixed co-substrate is changed.

Discussion

Victor Henri gave the saturating rate law and its equilibrium derivation in 1903; Leonor Michaelis and Maud Menten's 1913 paper on invertase is remembered because it made the measurement trustworthy rather than because it first wrote the equation — they controlled pH with buffers, worked at initial rates, and thereby produced data the equation could actually be tested against. Their derivation assumed rapid equilibrium between enzyme and substrate. George Briggs and J. B. S. Haldane replaced that in 1925 with the steady-state argument used here, which requires nothing about the relative sizes of \(k_{-1}\) and \(k_{\text{cat}}\), leaves the algebra of Step 6 unchanged, and changes only the interpretation of the constant — a rare case of a generalisation costing nothing.

The reason the model has outlived a century of mechanistic refinement is that its two parameters answer the two questions that are actually asked. \(k_{\text{cat}}/K_M\) answers “how good is this enzyme on this substrate at physiological concentrations”, and its ceiling at the diffusion limit turns a kinetic measurement into a statement about how much room evolution had left. \(K_M\) compared with the physiological substrate concentration answers “where on the curve does this enzyme sit”: an enzyme operating with \([\text{S}]\approx K_M\) responds nearly proportionally to changes in substrate and so transmits flux control, whereas one operating far above \(K_M\) is saturated, insensitive, and a potential bottleneck (Metabolic flux analysis). The same comparison in pharmacology decides whether a drug is cleared with first-order or zero-order kinetics (Pharmacokinetics), which is the difference between a proportionate dose response and an accumulating one.

The uncomfortable modern observation is that the standard validity condition is often violated in vivo. Proteomic and metabolomic surveys repeatedly find enzyme and metabolite concentrations of the same order, so \(\varepsilon\) is not small and the hyperbola is not the correct reduction; the total quasi-steady-state approximation of Step 12, which is uniformly valid in enzyme concentration, is the appropriate replacement in whole-cell kinetic models, and it collapses back to the hyperbola exactly when \(\varepsilon\ll1\). Related, and more subtle: the derivation assumes a well-mixed continuum, whereas a bacterial cell may hold only tens of copies of a given enzyme, so the deterministic rate law describes an ensemble average about which individual cells fluctuate substantially. Neither observation makes Michaelis–Menten kinetics wrong; both make it a limit whose conditions have to be checked rather than a law.

Common misconceptions. That the model describes a reaction rather than a regime — the same enzyme obeys it in a dilute cuvette and violates it in a cell, because \(\varepsilon\) depends on concentrations rather than on chemistry. And that saturation is caused by the substrate somehow becoming harder to bind at high concentration; it is caused simply by there being no free enzyme left to bind it, which is why saturation sets in around \([\text{S}]\sim K_M\) whatever the substrate is.

Worked examples

Example 1. An enzyme is assayed at total concentration \([\text{E}]_T=5.0\,\text{nM}\). Two initial rates are measured: \(v=0.333\,\mu\text{M}\,\text{s}^{-1}\) at \([\text{S}]=0.25\,\text{mM}\) and \(v=0.667\,\mu\text{M}\,\text{s}^{-1}\) at \([\text{S}]=1.00\,\text{mM}\). Find \(K_M\), \(V_{\max}\), \(k_{\text{cat}}\) and \(k_{\text{cat}}/K_M\), judge how close the enzyme is to catalytic perfection, and verify that the assay lies in the regime where the model applies.

\[ \frac{[\text{S}]}{v}=\frac{1}{V_{\max}}[\text{S}]+\frac{K_M}{V_{\max}} \]

The Hanes–Woolf form of Step 11 is the right linearisation for a two-point determination, because both parameters appear as a slope and an intercept of a line in \([\text{S}]\), and neither measured velocity is inverted into a large-leverage position. Symbols first. A

\[ \frac{0.25\,\text{mM}}{0.333\,\mu\text{M}\,\text{s}^{-1}}=0.750\,\text{mM}\,\text{s}\,\mu\text{M}^{-1},\qquad \frac{1.00\,\text{mM}}{0.667\,\mu\text{M}\,\text{s}^{-1}}=1.500\,\text{mM}\,\text{s}\,\mu\text{M}^{-1} \]

Form the two ordinates. Keeping millimolar for substrate and micromolar per second for velocity is harmless provided the mixed unit is carried through and converted only at the end. A

\[ \frac{1}{V_{\max}}=\frac{1.500-0.750}{1.00-0.25}=1.00\,\text{s}\,\mu\text{M}^{-1}\ \Longrightarrow\ V_{\max}=1.00\,\mu\text{M}\,\text{s}^{-1} \]

The slope of the two-point line. The millimolar units cancel between numerator and denominator, leaving the reciprocal of a velocity, as it must. A

\[ \frac{K_M}{V_{\max}}=0.750-\left(1.00\right)\left(0.25\right)=0.500\,\text{mM}\,\text{s}\,\mu\text{M}^{-1}\ \Longrightarrow\ K_M=\left(0.500\right)\left(1.00\right)=0.500\,\text{mM} \]

The intercept, back-substituted into the first point. Check against the definition rather than the algebra: at \([\text{S}]=0.500\,\text{mM}\) the model gives \(v=1.00\times0.500/1.000=0.500\,\mu\text{M}\,\text{s}^{-1}\), exactly half of \(V_{\max}\). A

\[ k_{\text{cat}}=\frac{V_{\max}}{[\text{E}]_T}=\frac{1.00\times10^{-6}\,\text{M}\,\text{s}^{-1}}{5.0\times10^{-9}\,\text{M}}=2.0\times10^{2}\,\text{s}^{-1} \]

The turnover number: each enzyme molecule converts \(200\) substrate molecules per second when saturated. This is the step that requires \([\text{E}]_T\) to be known, which is why an enzyme preparation must be quantified before \(k_{\text{cat}}\) can be quoted (Common errors). A

\[ \frac{k_{\text{cat}}}{K_M}=\frac{2.0\times10^{2}\,\text{s}^{-1}}{5.00\times10^{-4}\,\text{M}}=4.0\times10^{5}\,\text{M}^{-1}\,\text{s}^{-1} \]

The specificity constant, with \(K_M\) converted to molar. Against the diffusion ceiling of \(10^{8}\) to \(10^{9}\,\text{M}^{-1}\,\text{s}^{-1}\) (Step 8) this enzyme captures roughly one encounter in a thousand: a perfectly ordinary enzyme, nowhere near catalytic perfection, with ample room for a better substrate or a better active site. B

\[ \varepsilon=\frac{[\text{E}]_T}{K_M+[\text{S}]_0}=\frac{5.0\times10^{-9}}{5.00\times10^{-4}+2.5\times10^{-4}}=6.7\times10^{-6} \]

The validity check of Step 4, evaluated at the lower of the two substrate concentrations because that is where \(\varepsilon\) is largest. A relative error of seven parts per million is far below any experimental uncertainty, so the hyperbola is not the limiting approximation here. B

\[ k_1\ge\frac{k_{\text{cat}}}{K_M}\ \Longrightarrow\ t_c=\frac{1}{k_1\bigl([\text{S}]_0+K_M\bigr)}\le\frac{1}{\left(4.0\times10^{5}\right)\left(7.5\times10^{-4}\right)}=3.3\times10^{-3}\,\text{s} \]

The transient cannot be timed without \(k_1\), but Step 8 of the Proof bounds it: \(k_1\ge k_{\text{cat}}/K_M\), so \(t_c\) is at most about \(3\,\text{ms}\). A manual assay read over \(30\,\text{s}\) therefore spends at most one part in \(10^{4}\) of its duration inside the boundary layer — the bound suffices, and the unknown constant never has to be measured. C

\[ K_M=0.50\,\text{mM},\qquad V_{\max}=1.00\,\mu\text{M}\,\text{s}^{-1},\qquad k_{\text{cat}}=2.0\times10^{2}\,\text{s}^{-1},\qquad \frac{k_{\text{cat}}}{K_M}=4.0\times10^{5}\,\text{M}^{-1}\,\text{s}^{-1} \]

Reading. Two rate measurements, one linearisation and one division by the enzyme concentration deliver the complete steady-state description of the enzyme. The specificity constant places it about three orders of magnitude below the diffusion limit.

Scope. Both assay points satisfy \(\varepsilon\lt10^{-5}\) and last far longer than the bounded transient, so the parameters are the model's parameters and not artefacts of the regime. Two points fix a line exactly and leave no residual, so this determination carries no estimate of its own uncertainty; a real assay uses six to eight concentrations spanning \(0.2K_M\) to \(5K_M\) and fits the hyperbola directly.

Example 2. Ethanol is cleared almost entirely by hepatic alcohol dehydrogenase, for which the class I isoenzymes have \(K_M\approx1.0\,\text{mM}\), while the clinically observed elimination rate is close to \(0.015\,\text{g}\) per \(100\,\text{mL}\) of blood per hour. A subject reaches a blood ethanol concentration of \(20.0\,\text{mM}\). Using the integrated rate law, find how long clearance to \(1.0\,\text{mM}\) takes, compare it with the purely zero-order estimate, and identify the time constant of the terminal phase. Take the molar mass of ethanol as \(46.07\,\text{g}\,\text{mol}^{-1}\).

\[ V_{\max}=\frac{0.015\,\text{g}\ \text{per}\ 0.100\,\text{L}}{46.07\,\text{g}\,\text{mol}^{-1}}\ \text{per hour}=\frac{0.15\,\text{g}\,\text{L}^{-1}}{46.07\,\text{g}\,\text{mol}^{-1}}\ \text{per hour}=3.26\times10^{-3}\,\text{mol}\,\text{L}^{-1}\,\text{h}^{-1}=3.3\,\text{mM}\,\text{h}^{-1} \]

Convert the clinical rate to molar units: \(0.015\,\text{g}\) per \(100\,\text{mL}\) is \(0.15\,\text{g}\,\text{L}^{-1}\), and dividing by the molar mass gives the concentration rate. The observed clearance is identified with \(V_{\max}\) precisely because it is measured in the flooded regime \([\text{S}]\gg K_M\), where Step 7 makes the rate equal to \(V_{\max}\) and independent of concentration. B

\[ \frac{[\text{S}]_0}{K_M}=\frac{20.0\,\text{mM}}{1.0\,\text{mM}}=20\ \gg1\ \Longrightarrow\ \frac{v}{V_{\max}}=\frac{20.0}{1.0+20.0}=0.952 \]

Check the regime before applying the result. At the starting concentration the enzyme is \(95\%\) saturated, so elimination is very nearly zero order — the reason blood alcohol falls along a straight line rather than an exponential, unlike almost every other drug (Pharmacokinetics). A

\[ V_{\max}t=\bigl([\text{S}]_0-[\text{S}]\bigr)+K_M\ln\frac{[\text{S}]_0}{[\text{S}]} \]

The integrated Michaelis–Menten equation of Step 10, rearranged for \(t\) but not yet evaluated. The first term is the zero-order contribution, the second the correction that becomes important only once \([\text{S}]\) approaches \(K_M\). A

\[ V_{\max}t=\left(20.0-1.0\right)\,\text{mM}+\left(1.0\,\text{mM}\right)\ln\left(20.0\right)=19.0\,\text{mM}+3.00\,\text{mM}=22.0\,\text{mM} \]

Numbers last. \(\ln20.0=2.996\), so the logarithmic term contributes the equivalent of \(3.0\,\text{mM}\) of extra substrate to clear — three times \(K_M\), which is the general size of this correction whenever the clearance runs down to about \(K_M\). A

\[ t=\frac{22.0\,\text{mM}}{3.3\,\text{mM}\,\text{h}^{-1}}=6.7\,\text{h}\qquad\text{versus}\qquad t_{\text{zero order}}=\frac{19.0\,\text{mM}}{3.3\,\text{mM}\,\text{h}^{-1}}=5.8\,\text{h} \]

Dividing by \(V_{\max}\) gives the elapsed time. The zero-order idealisation is short by \(0.9\,\text{h}\), about \(14\%\): a straight-line extrapolation of blood ethanol to zero systematically predicts sobriety too early, because the last few millimolar are cleared by an enzyme that is no longer saturated. B

\[ [\text{S}]\ll K_M:\ -\frac{d[\text{S}]}{dt}\approx\frac{V_{\max}}{K_M}[\text{S}]\ \Longrightarrow\ \tau=\frac{K_M}{V_{\max}}=\frac{1.0\,\text{mM}}{3.3\,\text{mM}\,\text{h}^{-1}}=0.30\,\text{h}=18\,\text{min} \]

Below \(K_M\) the rate law linearises (Step 7) and the decay becomes exponential with this time constant; the terminal tail is therefore short, and the elimination curve is a straight line with a small exponential foot rather than two comparable phases. B

\[ \varepsilon=\frac{[\text{E}]_T}{K_M+[\text{S}]_0}\approx\frac{1\times10^{-5}\,\text{M}}{2.1\times10^{-2}\,\text{M}}\approx5\times10^{-4} \]

A validity check with a deliberately generous assumption: even taking hepatic alcohol dehydrogenase to be present at \(10\,\mu\text{M}\) — a high estimate for an abundant liver enzyme — \(\varepsilon\) remains under \(10^{-3}\), so the quasi-steady-state reduction is safe throughout. The genuine uncertainties here are physiological (distribution volume, first-pass metabolism, isoenzyme variation), not kinetic. C

\[ t\left(20.0\to1.0\,\text{mM}\right)=6.7\,\text{h},\qquad t_{\text{zero order}}=5.8\,\text{h},\qquad \tau_{\text{terminal}}=\frac{K_M}{V_{\max}}=18\,\text{min} \]

Reading. Because the substrate starts twenty times above \(K_M\), ethanol elimination is a constant-rate process for almost its whole course, and the hyperbolic rate law shows up only as a \(0.9\,\text{h}\) tail once the concentration falls into the neighbourhood of \(K_M\).

Scope. A one-compartment description with clearance by a single enzyme; it ignores distribution, the contribution of the microsomal ethanol-oxidising system at high concentrations, and the large inter-individual variation in both \(V_{\max}\) and isoenzyme complement. The kinetic structure — zero order above \(K_M\), first order below — survives all of those.

Problems

An enzyme has \(K_M=4.0\,\text{mM}\). Find the substrate concentrations at which it runs at \(10\%\) and at \(90\%\) of \(V_{\max}\), and state the general rule. Comment on what this implies for an enzyme used as a regulatory switch.
Solution
From the Result, \(v/V_{\max}=[\text{S}]/(K_M+[\text{S}])\). Solving for \([\text{S}]\) in symbols first: \([\text{S}]=K_M\dfrac{v/V_{\max}}{1-v/V_{\max}}\).
At \(v/V_{\max}=0.10\): \([\text{S}]=K_M(0.10/0.90)=K_M/9=0.44\,\text{mM}\). At \(0.90\): \([\text{S}]=K_M(0.90/0.10)=9K_M=36\,\text{mM}\).
The general rule is that the span from \(10\%\) to \(90\%\) of maximal rate costs a factor of \(81\) in substrate concentration, independent of \(K_M\). A hyperbola is therefore an extremely gentle switch: no plausible physiological swing in metabolite concentration can move a Michaelian enzyme from off to on. Genuine metabolic switches are cooperative, where a Hill exponent \(n\) compresses the same span to a factor \(81^{1/n}\) — a factor of \(9\) for \(n=2\), and about \(3\) for \(n=4\).
Two initial rates are measured for an enzyme present at \([\text{E}]_T=2.0\,\text{nM}\): \(v=0.20\,\mu\text{M}\,\text{s}^{-1}\) at \([\text{S}]=0.50\,\text{mM}\), and \(v=0.50\,\mu\text{M}\,\text{s}^{-1}\) at \([\text{S}]=2.00\,\text{mM}\). Determine \(K_M\), \(V_{\max}\), \(k_{\text{cat}}\) and \(k_{\text{cat}}/K_M\).
Solution
Use the double-reciprocal form of Step 11, \(1/v=(K_M/V_{\max})(1/[\text{S}])+1/V_{\max}\). The ordinates are \(1/v=5.00\) and \(2.00\,\text{s}\,\mu\text{M}^{-1}\); the abscissae are \(1/[\text{S}]=2.00\) and \(0.500\,\text{mM}^{-1}\).
Slope: \(K_M/V_{\max}=(5.00-2.00)/(2.00-0.500)=2.00\,\text{mM}\,\text{s}\,\mu\text{M}^{-1}\). Intercept: \(1/V_{\max}=2.00-(2.00)(0.500)=1.00\,\text{s}\,\mu\text{M}^{-1}\), so \(V_{\max}=1.00\,\mu\text{M}\,\text{s}^{-1}\) and \(K_M=(2.00)(1.00)=2.00\,\text{mM}\).
Check: at \([\text{S}]=0.50\,\text{mM}\), \(v=1.00\times0.50/(2.00+0.50)=0.20\,\mu\text{M}\,\text{s}^{-1}\), as measured.
Then \(k_{\text{cat}}=V_{\max}/[\text{E}]_T=(1.00\times10^{-6}\,\text{M}\,\text{s}^{-1})/(2.0\times10^{-9}\,\text{M})=5.0\times10^{2}\,\text{s}^{-1}\), and \(k_{\text{cat}}/K_M=(5.0\times10^{2})/(2.00\times10^{-3}\,\text{M})=2.5\times10^{5}\,\text{M}^{-1}\,\text{s}^{-1}\). Both assay points have \(\varepsilon\le(2.0\times10^{-9})/(2.5\times10^{-3})=8\times10^{-7}\), so the model applies.
An enzyme acts on two substrates. For substrate A, \(k_{\text{cat}}=3.0\times10^{2}\,\text{s}^{-1}\) and \(K_M=0.10\,\text{mM}\); for substrate B, \(k_{\text{cat}}=1.2\times10^{3}\,\text{s}^{-1}\) and \(K_M=2.0\,\text{mM}\). Which substrate is preferred at low concentration, which at saturation, and at what concentration do the two rates cross if both are supplied at the same concentration?
Solution
At saturation the ranking is by \(k_{\text{cat}}\): B is faster by \(1.2\times10^{3}/3.0\times10^{2}=4.0\)-fold.
At low concentration the ranking is by the specificity constant (Step 8): \(k_{\text{cat}}/K_M=3.0\times10^{2}/1.0\times10^{-4}=3.0\times10^{6}\,\text{M}^{-1}\text{s}^{-1}\) for A, and \(1.2\times10^{3}/2.0\times10^{-3}=6.0\times10^{5}\,\text{M}^{-1}\text{s}^{-1}\) for B. A is preferred by \(5.0\)-fold. The two rankings disagree, which is the point.
Crossover: set \(\dfrac{300[\text{S}]}{0.10+[\text{S}]}=\dfrac{1200[\text{S}]}{2.0+[\text{S}]}\) (per unit enzyme, concentrations in mM). Cancelling \([\text{S}]\) and cross-multiplying: \(300(2.0+[\text{S}])=1200(0.10+[\text{S}])\), so \(600+300[\text{S}]=120+1200[\text{S}]\), giving \(900[\text{S}]=480\) and \([\text{S}]=0.53\,\text{mM}\).
Below \(0.53\,\text{mM}\) the enzyme turns over A faster; above it, B. If the physiological concentration of both is, say, \(50\,\mu\text{M}\), the enzyme is functionally an A-enzyme despite B's four-fold higher turnover number.
Without inhibitor an enzyme gives \(K_M=0.50\,\text{mM}\) and \(V_{\max}=1.00\,\mu\text{M}\,\text{s}^{-1}\). With \(50\,\mu\text{M}\) of a compound present, the fitted parameters become \(K_M^{\text{app}}=0.25\,\text{mM}\) and \(V_{\max}^{\text{app}}=0.50\,\mu\text{M}\,\text{s}^{-1}\). Identify the inhibition type, obtain the inhibition constant, and predict the fractional inhibition at \([\text{S}]=5.0\,\text{mM}\).
Solution
Both parameters fall by exactly the same factor of \(2\), so \(k_{\text{cat}}/K_M\) is unchanged and a double-reciprocal plot would show two parallel lines. That is the signature of uncompetitive inhibition (Corollaries): the compound binds only the ES complex, and both \(V_{\max}\) and \(K_M\) are divided by \(\alpha'=1+[\text{I}]/K_i'\).
Hence \(\alpha'=2.00\), so \(K_i'=[\text{I}]/(\alpha'-1)=50\,\mu\text{M}/1.00=50\,\mu\text{M}\).
At \([\text{S}]=5.0\,\text{mM}\): uninhibited \(v=1.00\times5.0/(0.50+5.0)=0.909\,\mu\text{M}\,\text{s}^{-1}\); inhibited \(v=0.50\times5.0/(0.25+5.0)=0.476\,\mu\text{M}\,\text{s}^{-1}\). The ratio is \(0.524\), i.e. \(48\%\) inhibition.
Contrast with a competitive inhibitor of the same \(K_i\), which at this substrate concentration would give \(v=1.00\times5.0/(2\times0.50+5.0)=0.833\), only \(8\%\) inhibition. Raising substrate rescues a competitively inhibited enzyme and does not rescue an uncompetitively inhibited one — in the limit \([\text{S}]\to\infty\) the uncompetitive inhibition tends to the full factor \(1/\alpha'\).
An enzyme has \(k_1=1.0\times10^{7}\,\text{M}^{-1}\text{s}^{-1}\), \(k_{-1}=1.0\times10^{3}\,\text{s}^{-1}\) and \(k_{\text{cat}}=1.0\times10^{2}\,\text{s}^{-1}\). (a) Compute \(K_M\), \(K_S\) and \(k_{\text{cat}}/K_M\), and say by how much treating \(K_M\) as a dissociation constant misstates the affinity. (b) For an in vitro assay with \([\text{E}]_T=1.0\,\text{nM}\) and \([\text{S}]_0=1.0\,\text{mM}\), evaluate \(\varepsilon\) and \(t_c\), and state the fraction of enzyme carrying substrate. (c) For an intracellular situation with \([\text{E}]_T=100\,\mu\text{M}\) and \([\text{S}]_0=10\,\mu\text{M}\), evaluate \(\varepsilon\) and say what must be done instead.
Solution
(a) \(K_M=(k_{-1}+k_{\text{cat}})/k_1=(1.0\times10^{3}+1.0\times10^{2})/(1.0\times10^{7})=1.10\times10^{-4}\,\text{M}=0.110\,\text{mM}\). \(K_S=k_{-1}/k_1=1.0\times10^{3}/1.0\times10^{7}=1.00\times10^{-4}\,\text{M}=0.100\,\text{mM}\). Treating \(K_M\) as \(K_S\) therefore understates the affinity by \(10\%\) — tolerable here only because \(k_{\text{cat}}\ll k_{-1}\); if \(k_{\text{cat}}\) were \(1.0\times10^{4}\,\text{s}^{-1}\) the error would be a factor of \(11\). And \(k_{\text{cat}}/K_M=1.0\times10^{2}/1.10\times10^{-4}=9.1\times10^{5}\,\text{M}^{-1}\text{s}^{-1}\), about \(9\%\) of \(k_1\), consistent with the bound of Step 8.
(b) \(\varepsilon=[\text{E}]_T/(K_M+[\text{S}]_0)=(1.0\times10^{-9})/(1.10\times10^{-4}+1.0\times10^{-3})=9.0\times10^{-7}\): the quasi-steady-state error is negligible. \(t_c=1/\bigl(k_1([\text{S}]_0+K_M)\bigr)=1/\bigl(1.0\times10^{7}\times1.11\times10^{-3}\bigr)=9.0\times10^{-5}\,\text{s}\), i.e. \(90\,\mu\text{s}\) — unresolvable in a conventional assay and irrelevant to it. Saturation: \([\text{ES}]/[\text{E}]_T=[\text{S}]/(K_M+[\text{S}])=1.0/1.11=0.90\), so \(90\%\) of the enzyme is occupied and the assay is running at \(90\%\) of \(V_{\max}\).
(c) \(\varepsilon=(1.0\times10^{-4})/(1.0\times10^{-5}+1.10\times10^{-4})=(1.0\times10^{-4})/(1.20\times10^{-4})=0.83\). The small parameter is not small; the enzyme outnumbers its substrate ten to one, and solving the two conservation relations simultaneously gives \([\text{ES}]=4.6\,\mu\text{M}\) — nearly half the substrate is sequestered in complex, so free substrate is nowhere near the total. The standard hyperbola is simply wrong here, and fitting it would return meaningless parameters. The correct treatments are either the exact conservation equations, solved as a quadratic for \([\text{ES}]\) in terms of total substrate and total enzyme, or the total quasi-steady-state approximation of Step 12, which is uniformly valid in enzyme concentration and reduces to the hyperbola when \(\varepsilon\ll1\).