Diagnosing Inhibition from Kinetic Plots

A pharmacologist hands you a table of initial velocities. These are measured at five substrate concentrations. Each is repeated with four different inhibitor concentrations. Nothing on the page says "competitive" or "uncompetitive." You must extract that diagnosis purely from the pattern the numbers trace. This requires plotting them the right way. This is the daily work of enzymology. It means converting a grid of $v_0$ values into a mechanistic story. That story reveals where a drug binds and how tightly. Linearized kinetic plots exist for a clear reason. These include Lineweaver–Burk, Eadie–Hofstee, Hanes–Woolf, and their secondary replots. The human eye judges straight lines far better than curves. It is much better at this than judging the curvature of a rectangular hyperbola. It is also better at spotting where lines cross. Each inhibition mechanism from Competitive Inhibition, Uncompetitive Inhibition, and Noncompetitive and Mixed Inhibition imprints a distinctive geometric signature. This signature is a specific pattern of shifted slopes and intercepts. It also includes a fixed or moving intersection point. These signatures appear on the plots described above. Learning to read that signature is one practical skill this article builds. The other is extracting $K_i$ (and $K_i'$ for mixed inhibition) from secondary replots. The big picture is straightforward. Every reversible inhibition mechanism modifies the Michaelis–Menten equation. It does this by rescaling $K_m$ , $V_{max}$ , or both. These rescaling factors depend linearly on inhibitor concentration $[I]$ . The Lineweaver–Burk transform is itself linear in $1/[S]$ . Because of this, these rescalings become linear shifts in slope and intercept. A "family" of lines at different $[I]$ then intersects at a point. Sometimes it intersects along an axis instead. That location is mechanistically diagnostic. This article develops the algebra behind each pattern. It shows how to build the primary and secondary plots. It also works through three full data sets. These range from clean textbook numbers to a messy real-world case with a mixed inhibitor. The goal is twofold. You should recognize the qualitative signature. You should also back out the actual $K_i$ values with confidence. By the end you will gain several skills. You will look at a set of Lineweaver–Burk lines and identify the mechanism before doing any algebra. You will distinguish competitive, uncompetitive, pure noncompetitive, and mixed inhibition this way. You will know which secondary replot isolates $K_i$ cleanly. This means either $K_{m,app}$ vs. $[I]$ , or $1/V_{max,app}$ vs. $[I]$ . You will also understand the traps that complicate this work. These include nonuniform error weighting in double-reciprocal plots. They include apparent noncompetitive behavior from substrate depletion. They also include the ambiguity of near-parallel lines. Together these make this diagnostic exercise harder in the lab than on the page. This sets up the transition to non-hyperbolic, allosteric kinetics. That is where Allosteric Enzymes and Sigmoidal Kinetics takes over.

Key takeaways

Each inhibition type rescales $K_m$ and/or $V_{max}$ by factors $\alpha = 1+[I]/K_i$ and $\alpha' = 1+[I]/K_i'$ .
On a Lineweaver–Burk plot, competitive inhibition changes slope only; lines meet at a common $y$ -intercept ( $1/V_{max}$ ).
Uncompetitive inhibition changes the intercept only; lines are parallel (same slope, shifted intercepts).
Mixed/noncompetitive inhibition changes both slope and intercept; lines meet left of the $y$ -axis (or on it, for pure noncompetitive, at $x = -1/K_m$ ).
Secondary replots (slope vs. $[I]$ , intercept vs. $[I]$ ) are straight lines whose $x$ -intercepts give $-K_i$ and $-K_i'$ directly.
Key idea: the geometry of a family of double-reciprocal lines — where they intersect — is a direct fingerprint of which rate constant(s) the inhibitor perturbs.

Where this lives today

Distinguishing inhibition type from kinetic plots is not a historical curiosity confined to problem sets — it is the first mechanistic filter applied to any new small-molecule enzyme inhibitor in drug discovery. When a hit compound from a high-throughput screen inhibits a kinase or protease, the very first kinetic characterization asks exactly the question this article answers: does increasing substrate concentration overcome the inhibition (competitive — often exploitable with ATP-competitive kinase inhibitors, though these are more often characterized by direct $K_i$ fitting than classic Lineweaver–Burk families), or does the inhibitor keep working regardless of substrate (noncompetitive/uncompetitive/allosteric — often a more selective and druggable mechanism because it does not require competing against high intracellular substrate concentrations). HIV protease inhibitors, many kinase inhibitors, and statins (competitive inhibitors of HMG-CoA reductase relative to its substrate HMG-CoA) were all characterized in part through exactly this kind of primary/secondary plot analysis before modern global nonlinear fitting software (GraphPad Prism, DynaFit, KinTek Explorer) took over the heavy numerical lifting. Even today, when the actual parameter estimates come from nonlinear least-squares fits to the full rate equation rather than from lines drawn by eye, the Lineweaver–Burk family and its intersection geometry remain the standard visual diagnostic reported in papers — a reviewer expects to see, at a glance, whether lines converge on the axis, share an intercept, or run parallel, as a sanity check on the fitted mechanism. The same logic — rescaling of $K_m$ and $V_{max}$ by inhibitor-dependent factors — generalizes to multisubstrate enzymes (ordered and ping-pong mechanisms) via Cleland's extended nomenclature, and it is the natural bridge into non-Michaelis-Menten, cooperative behavior, where the underlying binding polynomial becomes genuinely nonlinear and the clean $\alpha,\alpha'$ bookkeeping developed here is replaced by the sigmoidal analysis of Allosteric Enzymes and Sigmoidal Kinetics.

:::quiz[On a Lineweaver–Burk plot, a family of lines at increasing [I] all share the same y-intercept. What does this indicate?]

The inhibitor binds only the ES complex
The inhibitor binds both E and ES with equal affinity
The inhibitor is competitive: Vmax is unchanged, Km is inflated ✓
The enzyme has cooperative substrate binding

:::quiz[In the mixed-inhibition equation v0 = Vmax[S] / (αKm + α′[S]), what does the secondary replot of "slope vs [I]" isolate?]

Vmax directly
Ki, via the x-intercept at −Ki ✓
Ki′, via the x-intercept at −Ki′
Km alone, independent of any inhibitor constant

:::quiz[A data set shows Km,app increasing and Vmax,app decreasing as [I] increases, with neither held constant. What is the correct diagnosis?]

Pure competitive inhibition
Pure uncompetitive inhibition
Mixed inhibition, with Ki and Ki′ both finite and generally unequal ✓
No inhibition; this is normal substrate saturation behavior

Ask the tutor: How would this diagnostic framework change for a two-substrate, ping-pong enzyme mechanism instead of a single-substrate one?

The physical picture — thorough prose

The Michaelis–Menten equation, $v_0 = V_{max}[S]/(K_m+[S])$ , describes a rectangular hyperbola. Its two limiting parameters, $V_{max}$ (the horizontal asymptote) and $K_m$ (the substrate concentration at half-saturation), are estimated more reliably when the curve is transformed into a straight line, because linear regression on a hyperbola is visually deceptive — the eye cannot judge whether the top third of a saturating curve is "correctly" bending over. The Lineweaver–Burk double-reciprocal transform,

\frac{1}{v_0} = \frac{K_m}{V_{max}}\cdot\frac{1}{[S]} + \frac{1}{V_{max}}

turns the hyperbola into a line with slope $K_m/V_{max}$ and $y$ -intercept $1/V_{max}$ ; its $x$ -intercept is $-1/K_m$ . Each inhibitor type modifies this line's slope and/or intercept in a specific, mechanistically traceable way, because each modifies $K_m$ and $V_{max}$ differently:

Competitive: inhibitor binds only free enzyme E, competing with S for the active site. $K_m$ is inflated by $\alpha=1+[I]/K_i$ ; $V_{max}$ is unchanged (enough S always outcompetes I).
Uncompetitive: inhibitor binds only the ES complex, trapping it as ESI. Both $K_m$ and $V_{max}$ are deflated by the same factor $\alpha'=1+[I]/K_i'$ , so their ratio $K_m/V_{max}$ (the slope) is unchanged.
Mixed (general) / Noncompetitive: inhibitor binds both E and ES, generally with different affinities ( $K_i \neq K_i'$ ). $V_{max}$ is deflated by $\alpha'$ ; $K_m$ is rescaled by $\alpha/\alpha'$ . Pure noncompetitive is the special case $K_i=K_i'$ , so $\alpha=\alpha'$ and $K_m$ is unaffected while $V_{max}$ alone drops.

Because the Lineweaver–Burk axes are $1/v_0$ vs. $1/[S]$ , and because slope and intercept are each simple linear functions of $[I]$ , plotting a family of lines at increasing $[I]$ produces a fan whose common intersection point (real or, for uncompetitive, at infinity) is the fingerprint. Competitive lines share a $y$ -intercept because $V_{max}$ is untouched. Uncompetitive lines are parallel because the slope $K_m/V_{max}$ is untouched. Mixed lines intersect somewhere to the left of the $1/v_0$ axis, above, on, or below the $1/[S]$ axis depending on the relative sizes of $K_i$ and $K_i'$ .

Figure

Lineweaver–Burk fingerprints for the three canonical inhibition types

The second essential picture is the secondary replot. Because the slope of a Lineweaver–Burk line, $K_m/V_{max}$ , and its intercept, $1/V_{max}$ , are each linear functions of $[I]$ under every one of these mechanisms, you can extract the primary slope and intercept at each $[I]$ , then plot those against $[I]$ to get a second line whose $x$ -intercept is exactly $-K_i$ (from the slope replot) or $-K_i'$ (from the intercept replot). This two-stage linearization — line, then line-of-lines — is what lets you pull a genuine equilibrium dissociation constant out of what began as raw velocity data.

Figure

Secondary replots isolate Ki from slope and intercept

The core ideas

The unified rate law that generates every pattern above is the mixed-inhibition equation, from which competitive, uncompetitive, and pure noncompetitive are all special cases:

v_0 = \frac{V_{max}[S]}{\alpha K_m + \alpha' [S]}, \qquad \alpha = 1+\frac{[I]}{K_i}, \qquad \alpha' = 1+\frac{[I]}{K_i'}

where:

$K_i$ is the dissociation constant for I binding free enzyme E (the term that scales $K_m$ ),
$K_i'$ is the dissociation constant for I binding the ES complex (the term that scales $1/V_{max}$ and rescales $K_m$ again),
$\alpha, \alpha' \ge 1$ are dimensionless inflation factors, both $\to 1$ as $[I]\to 0$ .

Taking the reciprocal gives the double-reciprocal form actually used for plotting:

\frac{1}{v_0} = \frac{\alpha K_m}{V_{max}}\cdot\frac{1}{[S]} + \frac{\alpha'}{V_{max}}

Reading: this says the Lineweaver–Burk slope is $(\alpha K_m/V_{max})$ — proportional to $\alpha$ alone — and the intercept is $(\alpha'/V_{max})$ — proportional to $\alpha'$ alone. Since $\alpha$ depends only on $K_i$ and $\alpha'$ only on $K_i'$ , the slope replot isolates $K_i$ and the intercept replot isolates $K_i'$ , regardless of how tangled the two look in a raw $v_0$ vs. $[S]$ plot. The apparent Michaelis–Menten parameters at a fixed $[I]$ are:

K_{m,app} = K_m\cdot\frac{\alpha}{\alpha'}, \qquad V_{max,app} = \frac{V_{max}}{\alpha'}

where:

$K_{m,app}$ and $V_{max,app}$ are the parameters you would fit if you (wrongly) treated each $[I]$ -slice as an independent Michaelis–Menten data set,
the ratio $\alpha/\alpha'$ shows immediately why $K_{m,app}$ rises with $[I]$ in competitive inhibition ( $\alpha'=1$ ), falls in uncompetitive inhibition ( $\alpha=1$ ), and can rise, fall, or stay flat in mixed inhibition depending on whether $K_i \gtrless K_i'$ .

Setting the three mechanisms as limits of the general equation:

\text{Competitive: } K_i'\to\infty \ (\alpha'=1) \qquad \text{Uncompetitive: } K_i\to\infty \ (\alpha=1) \qquad \text{Noncompetitive: } K_i=K_i' \ (\alpha=\alpha')

Derivation AFrom the mixed-inhibition rate law to the Lineweaver–Burk slope/intercept replots

Goal. Starting from the general mixed-inhibition velocity equation, derive the Lineweaver–Burk line at fixed $[I]$ , then derive the two secondary-plot equations whose $x$ -intercepts give $-K_i$ and $-K_i'$ directly.

Step 1 — Write the rate law with explicit $\alpha,\alpha'$ . The steady-state derivation for E + S ⇌ ES → E + P, with I binding E (dissociation $K_i$ ) and I binding ES (dissociation $K_i'$ ), yields

v_0 = \frac{V_{max}[S]}{K_m\left(1+\dfrac{[I]}{K_i}\right) + [S]\left(1+\dfrac{[I]}{K_i'}\right)}

This is the same equation shown above with $\alpha, \alpha'$ substituted back in; it reduces correctly to plain Michaelis–Menten when $[I]=0$ (both parentheses become 1).

Step 2 — Invert to the double-reciprocal form. Divide numerator and denominator by $V_{max}[S]$ :

\frac{1}{v_0} = \frac{K_m\alpha}{V_{max}}\cdot\frac{1}{[S]} + \frac{\alpha'}{V_{max}}

This is a line in $1/[S]$ with

\text{slope}([I]) = \frac{K_m}{V_{max}}\left(1+\frac{[I]}{K_i}\right), \qquad \text{intercept}([I]) = \frac{1}{V_{max}}\left(1+\frac{[I]}{K_i'}\right)

Step 3 — Recognize both as linear functions of $[I]$ . Expand each:

\text{slope}([I]) = \frac{K_m}{V_{max}} + \frac{K_m}{V_{max}K_i}\,[I]

\text{intercept}([I]) = \frac{1}{V_{max}} + \frac{1}{V_{max}K_i'}\,[I]

Both are straight lines in $[I]$ : a fixed baseline value (the $[I]=0$ slope or intercept) plus a term linear in $[I]$ .

Step 4 — Find the $x$ -intercept of the slope replot. Set $\text{slope}([I])=0$ and solve for $[I]$ :

0 = \frac{K_m}{V_{max}} + \frac{K_m}{V_{max}K_i}[I] \implies [I] = -K_i

The $K_m/V_{max}$ prefactor cancels entirely — the $x$ -intercept depends on $K_i$ alone, never on $K_m$ or $V_{max}$ individually. This is why the slope replot is a clean, mechanism-specific readout of $K_i$ .

Step 5 — Find the $x$ -intercept of the intercept replot. Identically,

0 = \frac{1}{V_{max}} + \frac{1}{V_{max}K_i'}[I] \implies [I] = -K_i'

Again the $V_{max}$ prefactor cancels. The intercept replot's $x$ -intercept is $-K_i'$ exactly, independent of $V_{max}$ 's absolute value.

Step 6 — Specialize to the three named mechanisms. For competitive inhibition $K_i'\to\infty$ , so the intercept replot has zero slope (intercept is constant $=1/V_{max}$ for all $[I]$ ) — consistent with the common- $y$ -intercept fan. For uncompetitive inhibition $K_i\to\infty$ , so the slope replot has zero slope (slope constant $=K_m/V_{max}$ for all $[I]$ ) — consistent with parallel lines. For pure noncompetitive inhibition $K_i=K_i'$ , both replots have $x$ -intercept $-K_i$ , and substituting back into the primary equation shows the lines cross exactly on the $1/[S]$ axis at $-1/K_m$ (see the second derivation).

Go deeper · Derivation B · 1 minLocating the intersection point of a Lineweaver–Burk family algebraically

Derivation BLocating the intersection point of a Lineweaver–Burk family algebraically

Goal. Prove that all lines in a mixed-inhibition family cross at a single point, and derive its coordinates in terms of $K_m$ , $V_{max}$ , $K_i$ , $K_i'$ — explaining why pure noncompetitive lines cross exactly on the $x$ -axis while general mixed lines cross above or below it.

Step 1 — Write two lines from the family at $[I]=I_1$ and $[I]=I_2$ . Using the slope/intercept expressions from Step 2 above,

y = \left(\frac{K_m}{V_{max}}+\frac{K_m}{V_{max}K_i}I_1\right)x + \left(\frac{1}{V_{max}}+\frac{1}{V_{max}K_i'}I_1\right)

y = \left(\frac{K_m}{V_{max}}+\frac{K_m}{V_{max}K_i}I_2\right)x + \left(\frac{1}{V_{max}}+\frac{1}{V_{max}K_i'}I_2\right)

where $x=1/[S]$ , $y=1/v_0$ .

Step 2 — Set the two lines equal and solve for the crossing $x^*$ . Subtract the equations; all terms independent of $I$ cancel:

\frac{K_m}{V_{max}K_i}(I_1-I_2)\,x^* = -\frac{1}{V_{max}K_i'}(I_1-I_2)

Since $I_1\neq I_2$ , divide through by $(I_1-I_2)$ :

\frac{K_m}{K_i}x^* = -\frac{1}{K_i'} \implies x^* = -\frac{K_i}{K_m K_i'}

Because this result did not depend on which pair $(I_1,I_2)$ we chose, every line in the family passes through this same $x^*$ — a genuine common intersection, not just pairwise crossings.

Step 3 — Substitute back to get the crossing $y^*$ . Plug $x^*$ into the $[I]=0$ line (plain Lineweaver–Burk, $y = \frac{K_m}{V_{max}}x+\frac{1}{V_{max}}$ ):

y^* = \frac{K_m}{V_{max}}\left(-\frac{K_i}{K_mK_i'}\right)+\frac{1}{V_{max}} = \frac{1}{V_{max}}\left(1-\frac{K_i}{K_i'}\right)

Step 4 — Interpret the sign of $y^*$ and the special cases.

If $K_i = K_i'$ (pure noncompetitive): $y^*=0$ , and $x^* = -1/K_m$ . The lines cross exactly on the $x$ -axis at $-1/K_m$ — the textbook noncompetitive signature.
If $K_i > K_i'$ (inhibitor binds ES more tightly — "mixed, uncompetitive-leaning"): $y^*<0$ , lines cross below the $x$ -axis.
If $K_i < K_i'$ (inhibitor binds free E more tightly — "mixed, competitive-leaning"): $y^*>0$ , lines cross above the $x$ -axis, in the upper-left quadrant.
As $K_i'\to\infty$ (competitive limit): $x^*\to 0$ and $y^*\to 1/V_{max}$ — the crossing slides exactly onto the $y$ -axis, recovering the common-intercept fan.
As $K_i\to\infty$ (uncompetitive limit): $x^*\to -\infty$ — the "crossing point" recedes to infinity, which is exactly the algebraic statement that the lines never meet: they are parallel.

Worked examples

Example 1 — Foundational: confirming competitive inhibition and extracting $K_i$ from clean textbook data. An enzyme is assayed at five $[S]$ values, without inhibitor and with $2.0\ \text{mM}$ inhibitor. Uninhibited data give $K_m = 4.0\ \text{mM}$ , $V_{max} = 100\ \mu\text{mol min}^{-1}$ . With inhibitor, the fitted apparent parameters are $K_{m,app}=12.0\ \text{mM}$ , $V_{max,app}=100\ \mu\text{mol min}^{-1}$ (unchanged — the hallmark of competitive inhibition). Since $V_{max}$ is untouched, $\alpha'=1$ , confirming pure competitive behavior (this is the qualitative diagnosis: intercepts match, slopes differ). Now use

K_{m,app} = \alpha K_m \implies \alpha = \frac{K_{m,app}}{K_m} = \frac{12.0}{4.0} = 3.0

Solve for $K_i$ from $\alpha = 1+[I]/K_i$ :

3.0 = 1+\frac{2.0\ \text{mM}}{K_i} \implies 2.0 = \frac{2.0\ \text{mM}}{K_i} \implies K_i = 1.0\ \text{mM}

Limiting-case check: as $[I]\to 0$ , $\alpha\to 1$ and $K_{m,app}\to K_m=4.0\ \text{mM}$ — the inhibited line collapses onto the control line, as it must. As $[I]\to\infty$ at fixed $K_i$ , $\alpha\to\infty$ , so $K_{m,app}\to\infty$ : the apparent affinity for substrate is destroyed, matching the physical picture of a competitive inhibitor that can always be "diluted out" by enough substrate (since $V_{max}$ is untouched, full saturation is still reachable at high enough $[S]$ ).

Example 2 — Applied: diagnosing mixed inhibition from a real assay and extracting both $K_i$ and $K_i'$ . A drug candidate is tested against a metabolic enzyme at four inhibitor concentrations. From nonlinear (or Lineweaver–Burk) fits at each $[I]$ , the apparent parameters are:

$[I]$ (µM)	$K_{m,app}$ (mM)	$V_{max,app}$ (µmol/min)
0	2.00	50.0
10	2.60	41.7
25	3.50	33.3
50	5.00	25.0

Step 1: check the qualitative pattern. Both $K_{m,app}$ and $V_{max,app}$ change with $[I]$ — not pure competitive (would need constant $V_{max,app}$ ), not pure uncompetitive (would need constant $K_{m,app}$ ), not pure noncompetitive (would need constant $K_{m,app}$ too). This is mixed inhibition. Step 2: get $K_i'$ from the $V_{max}$ data, since $V_{max,app}=V_{max}/\alpha'$ , i.e. $1/V_{max,app} = \alpha'/V_{max}$ is linear in $[I]$ :

\frac{1}{V_{max,app}}: \quad 0.0200,\ 0.0240,\ 0.0300,\ 0.0400 \ \ (\text{min}\,\mu\text{mol}^{-1}) \text{ at } [I]=0,10,25,50\ \mu\text{M}

Fit a line to $(1/V_{max,app})$ vs. $[I]$ . Using the two extreme points for a quick slope estimate:

\text{slope} = \frac{0.0400-0.0200}{50-0} = 4.0\times10^{-4}\ \mu\text{M}^{-1}\,\text{min}\,\mu\text{mol}^{-1}

Intercept (at $[I]=0$ ) is $1/V_{max}=0.0200$ , so $V_{max}=50.0\ \mu\text{mol/min}$ (consistent with the uninhibited row, as required). From Step 5 of the first derivation, this line's slope equals $1/(V_{max}K_i')$ :

4.0\times10^{-4} = \frac{1}{50.0 \times K_i'} \implies K_i' = \frac{1}{50.0\times 4.0\times10^{-4}} = 50\ \mu\text{M}

Step 3: get $K_i$ from the $K_{m,app}$ data. Since $K_{m,app}=K_m\alpha/\alpha'$ , rearrange using the already-known $\alpha'=V_{max}/V_{max,app}$ :

\alpha = \frac{K_{m,app}\,\alpha'}{K_m} = \frac{K_{m,app}}{K_m}\cdot\frac{V_{max}}{V_{max,app}}

At $[I]=50\ \mu\text{M}$ : $\alpha' = 50.0/25.0 = 2.0$ , and

\alpha = \frac{5.00}{2.00}\times 2.0 = 5.0

Then $\alpha = 1+[I]/K_i \implies 5.0 = 1+50/K_i \implies K_i = 50/4.0 = 12.5\ \mu\text{M}$ . Check with the $[I]=25\ \mu\text{M}$ row: $\alpha' = 50.0/33.3=1.50$ ; $\alpha = (3.50/2.00)\times1.50 = 2.625$ ; then $K_i = 25/(2.625-1)=25/1.625=15.4\ \mu\text{M}$ — reasonably consistent with 12.5 µM given that this "quick two-point" method (rather than full least-squares) is being used for illustration; a proper linear regression across all four points would reconcile these into a single best-fit $K_i\approx 13$ –15 µM. Since $K_i (\approx 13\text{–}15\ \mu\text{M}) < K_i' (=50\ \mu\text{M})$ , the inhibitor binds free enzyme more tightly than it binds ES — a competitive-leaning mixed inhibitor. By Step 4 of the second derivation, this means the Lineweaver–Burk family crosses above the $x$ -axis.

Limiting-case check: if the assay had instead shown $K_i=K_i'$ exactly, the ratio $\alpha/\alpha'$ would equal 1 at every $[I]$ , forcing $K_{m,app}=K_m=2.00\ \text{mM}$ at all four rows — collapsing this mixed case onto pure noncompetitive inhibition, consistent with the $K_i=K_i'$ limit stated in the core equations.

Example 3 — Exam/competition level: distinguishing uncompetitive inhibition from substrate-depletion artifact, and computing $K_i'$ from a Cornish-Bowden-style secondary analysis. A kinetics problem gives the following $v_0$ (µmol/min) at $[S] = 1, 2, 5, 10\ \text{mM}$ , for $[I]=0$ and $[I]=4\ \text{mM}$ :

$[S]$ (mM)	$v_0$ , no I	$v_0$ , $[I]=4$ mM
1	16.7	9.09
2	28.6	15.4
5	45.5	24.0
10	58.8	30.8

Step 1 — fit the uninhibited data to Michaelis–Menten (or Lineweaver–Burk). Reciprocals: $1/[S] = 1.000, 0.500, 0.200, 0.100\ \text{mM}^{-1}$ ; $1/v_0 = 0.0599, 0.0350, 0.0220, 0.0170\ \text{min}/\mu\text{mol}$ . A linear fit through these four points (visually, slope from the extreme pair):

\text{slope} = \frac{0.0599-0.0170}{1.000-0.100} = \frac{0.0429}{0.900} = 0.0477\ \text{mM min}/\mu\text{mol} = K_m/V_{max}

\text{intercept} \approx 0.0170 - 0.0477(0.100) = 0.0123\ \text{min}/\mu\text{mol} = 1/V_{max}

So $V_{max} = 1/0.0123 = 81.3\ \mu\text{mol/min}$ , and $K_m = 0.0477\times81.3 = 3.88\ \text{mM}$ (rounding to $K_m\approx 4.0\ \text{mM}$ , $V_{max}\approx 80\ \mu\text{mol/min}$ for clean numbers, consistent with how such textbook tables are usually back-generated). Step 2 — fit the inhibited data. Reciprocals: $1/v_0 = 0.1100, 0.0649, 0.0417, 0.0325$ .

\text{slope} = \frac{0.1100-0.0325}{1.000-0.100} = \frac{0.0775}{0.900} = 0.0861\ \text{mM min}/\mu\text{mol}

\text{intercept} \approx 0.0325 - 0.0861(0.100) = 0.0239\ \text{min}/\mu\text{mol}

Step 3 — compare slopes and intercepts. Uninhibited slope $0.0477$ vs. inhibited slope $0.0861$ — these are not equal, so this is not clean uncompetitive inhibition (which requires identical slopes/parallel lines). The ratio of intercepts is $0.0239/0.0123 = 1.94$ ; the ratio of slopes is $0.0861/0.0477=1.80$ . These are close but not identical — a classic situation where a student might wrongly force an "uncompetitive" label just because both $K_{m,app}$ and $V_{max,app}$ decrease. Check what pure uncompetitive predicts: both slope and intercept should scale by the same factor ( $\alpha'$ ), since slope $=K_m\alpha'/(V_{max}\alpha')$ — wait, more carefully, uncompetitive has $\alpha=1$ , so slope stays at $K_m/V_{max}$ (unchanged!) while only the intercept scales by $\alpha'$ . Since here the slope clearly changed (0.0477 → 0.0861, a 1.80-fold increase), this data set is not uncompetitive — it is mixed, with $K_i$ finite (not infinite). Step 4 — extract $\alpha$ and $\alpha'$ properly. From $V_{max,app}=1/0.0239=41.8$ , so $\alpha' = V_{max}/V_{max,app}=81.3/41.8=1.945$ . From $K_{m,app}=\text{slope}_{inh}\times V_{max,app}=0.0861\times41.8=3.60\ \text{mM}$ , and $\alpha = K_{m,app}\alpha'/K_m = (3.60/3.88)\times1.945=1.804$ . Step 5 — solve for $K_i$ and $K_i'$ at $[I]=4\ \text{mM}$ :

\alpha = 1.804 = 1+\frac{4.0}{K_i} \implies K_i = \frac{4.0}{0.804}=4.98\ \text{mM}

\alpha' = 1.945 = 1+\frac{4.0}{K_i'} \implies K_i' = \frac{4.0}{0.945}=4.23\ \text{mM}

The two constants are close (4.98 vs. 4.23 mM) but not equal — this is mixed inhibition with a mild competitive character ( $K_i$ slightly larger than $K_i'$ means slightly weaker binding to free E than to ES, so the inhibitor leans uncompetitive, and by the intersection-point formula the Lineweaver–Burk lines should cross just below the $x$ -axis). This example is a deliberate trap: a quick glance at "both $K_m$ and $V_{max}$ dropped" invites a lazy "noncompetitive" label, but only the full slope/intercept comparison — and the near-but-not-quite equal $K_i, K_i'$ — reveals genuine mixed inhibition rather than the idealized pure case.

Limiting cases, scope & common traps

Situation	Lineweaver–Burk signature	What can go wrong
Competitive ( $K_i'\to\infty$ )	Common $y$ -intercept; slopes increase with $[I]$	High $[S]$ points are sparse/noisy → intercept looks like it "shifts" from bad extrapolation
Uncompetitive ( $K_i\to\infty$ )	Parallel lines (equal slope)	Only 2–3 $[S]$ points make "parallel" and "slightly convergent" visually indistinguishable
Pure noncompetitive ( $K_i=K_i'$ )	Lines cross exactly on the $x$ -axis at $-1/K_m$	Rare in practice; near-equal but unequal $K_i,K_i'$ (true mixed) is easily mistaken for this
Mixed, $K_i<K_i'$	Lines cross above the $x$ -axis	Easily confused with competitive if the crossing is close to the $y$ -axis
Mixed, $K_i>K_i'$	Lines cross below the $x$ -axis	Easily confused with uncompetitive if lines are nearly parallel
Substrate depletion / assay nonlinearity	Curvature mimics inhibition at high $[S]$	Always confirm initial-rate conditions ( $<10\%$ substrate consumed) before trusting any linearized plot

Common trapDouble-reciprocal plots distort experimental error

The Lineweaver–Burk transform takes reciprocals of both variables, so small, evenly-distributed errors in $v_0$ at low $[S]$ (where $v_0$ is itself small) become hugely magnified at large $1/[S]$ — a point measured with 5% relative error at $[S]=0.5K_m$ can dominate a naive unweighted least-squares fit and visibly bend the "line," creating spurious apparent changes in slope that can be misread as inhibition-type evidence. Modern practice fits raw $v_0$ vs. $[S]$ data directly to the Michaelis–Menten (or mixed-inhibition) equation by nonlinear regression, using Lineweaver–Burk, Eadie–Hofstee ( $v_0$ vs. $v_0/[S]$ ), or Hanes–Woolf ( $[S]/v_0$ vs. $[S]$ ) plots only for visual diagnosis of mechanism type, not for final parameter estimation.

The chemist behind it

Hans Lineweaver (1907–2009) and Dean Burk (1904–1988) were not primarily kinetics theorists; Lineweaver was a food and dairy chemist at the USDA, and Burk was a chemist at the National Cancer Institute whose broader career spanned enzyme kinetics, cancer metabolism, and — in his later years — increasingly controversial fringe science. Their 1934 paper, "The Determination of Enzyme Dissociation Constants" (Journal of the American Chemical Society), did something deceptively simple: it proposed plotting $1/v_0$ against $1/[S]$ so that the two Michaelis–Menten parameters could be read off as a slope and an intercept, using the (then labor-intensive) tool of straight-line graphical fitting rather than iterative curve-fitting by hand. The timing mattered enormously. Leonor Michaelis and Maud Menten's 1913 rate law was already two decades old, but extracting $K_m$ and $V_{max}$ reliably from a hyperbola — using slide rules and graph paper, with no computers — was genuinely difficult and error-prone. The double-reciprocal linearization made enzyme kinetics practical for an entire generation of biochemists, and it remains the default teaching tool today even though nonlinear regression has superseded it for actual parameter estimation. The extension of the Lineweaver–Burk framework to distinguish inhibition mechanisms — using families of lines at varying $[I]$ and their intersection geometry — was developed through the mid-20th century by kineticists including W. W. Cleland, whose systematic nomenclature and graphical criteria for competitive, uncompetitive, and mixed inhibition (and for more complex multi-substrate mechanisms) remain the modern standard reference for reading inhibition patterns off of kinetic plots.

Origins · 1 minHans Lineweaver and how this was discovered

Hans Lineweaver

1907–2009 · USDA food chemist, co-developer of the double-reciprocal plot

Lineweaver spent most of his working career in agricultural and food chemistry at the USDA's Western Regional Research Laboratory, far from the mainstream of enzymology after his early collaboration with Burk. The 1934 paper was, in a sense, a side project — a graphical convenience born of the practical need to fit rate data without modern computation — yet it became one of the most cited papers in the history of biochemistry, still appearing by name on the axes of plots in undergraduate labs nearly a century later.

History

- 1902 — Adrian Brown proposes that enzyme action proceeds through a bound enzyme–substrate complex.

- 1913 — Leonor Michaelis and Maud Menten derive the rate law for a single-substrate enzyme reaction.

- 1925 — G. E. Briggs and J. B. S. Haldane formalize the steady-state approximation underlying the modern Michaelis–Menten derivation.

- 1934 — Hans Lineweaver and Dean Burk introduce the double-reciprocal plot for graphical determination of

K_m

and

V_{max}

- 1963 — W. W. Cleland publishes systematic graphical and statistical criteria for distinguishing inhibition mechanisms from families of reciprocal plots.

Why does a competitive inhibitor leave Vmax unchanged?

Because a competitive inhibitor binds only the free enzyme E, in direct competition with substrate S for the same active site. At sufficiently high $[S]$ , substrate molecules statistically outcompete the inhibitor for every free active site, so the enzyme population can still be driven to full saturation — meaning $V_{max}$ , the rate at infinite $[S]$ , is reachable regardless of $[I]$ . What the inhibitor does change is $K_m$ : more substrate is needed to reach half-maximal velocity because some fraction of enzyme is tied up as EI at any given moment, so the apparent affinity for substrate looks weaker. This is exactly why $\alpha'=1$ ( $K_i'\to\infty$ ) in the general mixed-inhibition formula — I simply cannot bind ES, so it has zero effect on the maximal turnover rate itself.

How is uncompetitive inhibition different mechanistically, and why does it produce parallel lines?

An uncompetitive inhibitor binds only the ES complex, forming a dead-end ESI complex that cannot proceed to product. Because it cannot bind free E at all ( $\alpha=1$ , $K_i\to\infty$ ), it never competes with substrate binding — instead it removes ES from the productive pool after S has already bound. This lowers both $V_{max}$ and $K_m$ by the same factor $\alpha'$ (by Le Chatelier-like pulling of the E+S⇌ES equilibrium toward ES to replace what is lost to ESI), so their ratio, the Lineweaver–Burk slope $K_m/V_{max}$ , stays exactly constant — hence parallel lines. This mechanism is often seen with substrate analogs and certain drugs like lithium's inhibition of some phosphatases, and is diagnostically distinguishable precisely because it never manifests as a shifted slope.

What is the practical difference between "mixed" and "noncompetitive" inhibition in how they're diagnosed?

Noncompetitive inhibition is the special case of mixed inhibition where the inhibitor binds free E and ES with exactly equal affinity, $K_i=K_i'$ . In that case $K_m$ is completely unaffected (the $\alpha/\alpha'$ ratio equals 1 at all $[I]$ ) and only $V_{max}$ drops — producing lines that cross exactly on the $x$ -axis at $-1/K_m$ . General mixed inhibition, with $K_i\neq K_i'$ , changes both $K_m$ and $V_{max}$ , and the lines cross off the $x$ -axis, either above (if $K_i<K_i'$ ) or below (if $K_i>K_i'$ ). In practice, true noncompetitive inhibition is rare; most real "noncompetitive-looking" inhibitors are actually mixed with $K_i$ and $K_i'$ close but not identical, which is why careful secondary-plot analysis (rather than eyeballing the primary plot) matters.

Can you get Ki and Ki' from a single inhibitor concentration, or do you really need a family of lines?

A single $[I]$ gives you the apparent $K_{m,app}$ and $V_{max,app}$ , and hence the two ratios $\alpha=K_{m,app}\alpha'/K_m$ and $\alpha'=V_{max}/V_{max,app}$ — which, combined with the known uninhibited $K_m,V_{max}$ , does let you solve for $K_i$ and $K_i'$ algebraically at that one $[I]$ , as done in Example 2's single-row calculation. However, this approach has no internal redundancy: a single mismeasured point gives you a wrong answer with no way to detect it. Using multiple $[I]$ values and a genuine secondary-plot linear regression (or global nonlinear fitting) gives a much more robust estimate, because systematic deviations from linearity in the replot immediately flag either a wrong mechanism assignment or an experimental problem.

Why does the crossing point of a mixed-inhibition Lineweaver–Burk family not depend on Km or Vmax individually?

The algebra shows the crossing $x$ -coordinate is $x^* = -K_i/(K_m K_i')$ and the $y$ -coordinate is $y^*=(1/V_{max})(1-K_i/K_i')$ . While $K_m$ and $V_{max}$ do appear, notice that when you derive the crossing by equating two lines at different $[I]$ , the $[I]$ -independent parts of slope and intercept ( $K_m/V_{max}$ and $1/V_{max}$ , respectively) always cancel from the governing equation that determines whether a crossing exists — only the ratio $K_i/K_i'$ survives as the diagnostic quantity for whether the crossing sits above, on, or below the axis. So while the exact pixel location of the crossing point does scale with $K_m,V_{max}$ , its qualitative position (above/on/below the $x$ -axis) is governed purely by the ratio $K_i/K_i'$ , which is the mechanistically meaningful comparison.

How does substrate depletion during the assay create a false inhibition signature?

Initial-velocity kinetics assumes $[S]$ stays essentially constant during the measurement window — typically enforced by keeping substrate conversion below about 10%. If an assay runs too long, or if the enzyme is unusually active, $[S]$ measurably drops during the time course, so the "initial" velocity you calculate (total product formed divided by elapsed time) actually underestimates the true instantaneous $v_0$ , and does so more severely at low starting $[S]$ (where the fractional depletion is larger for a given amount of product formed). This artificially depresses the low- $[S]$ points more than the high- $[S]$ points, mimicking the intercept and slope shifts characteristic of noncompetitive-like inhibition even when no inhibitor is present. Always verify linearity of product formation vs. time before trusting any $v_0$ estimate used in these plots.

Why is Ki defined via the equilibrium E + I ⇌ EI rather than measured directly as a rate?

$K_i$ is defined as the dissociation constant of the EI complex, $K_i = [\text{E}][\text{I}]/[\text{EI}]$ , treating inhibitor binding as a rapid pre-equilibrium — which is the standard assumption for reversible, non-covalent inhibitors (this parallels how $K_m$ itself, under the Michaelis–Menten quasi-equilibrium limit, approximates a true dissociation constant for ES). This makes $K_i$ a thermodynamic quantity: the inhibitor concentration at which half the free enzyme is bound as EI at equilibrium, directly comparable across different inhibitors and different assays as a measure of binding affinity, independent of the catalytic rate constants $k_{cat}$ that separately determine $V_{max}$ . For slow, tight-binding, or covalent inhibitors this simple equilibrium picture breaks down and requires progress-curve or covalent-labeling kinetics instead of the steady-state plots discussed here.