Two-step irreversible inactivation — reversible encounter complex (Ki) followed by irreversible covalent bond formation (kinact) — and how progress-curve kinetics extracts both constants
Covalent irreversible inhibition is formally a two-step mechanism: E + I ⇌ EI (governed by Ki, a true dissociation constant) followed by EI → E–I (governed by kinact, an irreversible first-order rate constant). The first step is mechanistically identical to reversible inhibitor binding — shape complementarity, hydrogen bonding, and hydrophobic packing dock the molecule into the pocket with no chemistry yet. Ki sets the occupancy of this pre-reactive complex and, critically, positions the electrophile close enough to a nucleophilic residue for the second step to occur at all.
The two-step covalent mechanism was formalized by Kitz and Wilson in their 1962 study of acetylcholinesterase inactivation by organophosphates, and it remains the operative framework for every modern targeted covalent inhibitor (TCI) program:
E + I ⇌(Ki) EI →(kinact) E–I
Ki here is NOT an IC50 and is not necessarily comparable across assays run at different incubation times — it is the dissociation constant of the purely non-covalent complex, exactly analogous to a Kd or a competitive Ki for a reversible inhibitor. Structurally, this recognition step is what medicinal chemists optimize first: a reversible "recognition fragment" (e.g., the aminopyrimidine core of a kinase-hinge binder) is elaborated with substituent vectors that reach a nearby nucleophile, most commonly a non-catalytic cysteine thiol positioned 4–6 Å from the ATP pocket.
Measuring Ki in isolation is difficult because the reaction proceeds continuously once binding occurs. Three practical strategies are used:
1. Rapid-mix / stopped-flow with a non-covalent warhead surrogate: replace the electrophile with a saturated (unreactive) analog and measure binding directly by SPR (SPR sensorgram kon/koff → Kd ≈ Ki) or isothermal titration calorimetry.
2. Global fitting of full progress curves: modern nonlinear regression (e.g., KinTek Explorer, DynaFit) fits the entire two-step differential equation system directly to raw fluorescence traces, extracting Ki and kinact simultaneously without requiring the hyperbolic approximation of stage 4.
3. Mass-spectrometry-based competition: incubate enzyme with covalent probe and increasing concentrations of unlabeled reversible fragment; IC50 for blocking labeling approximates Ki.
For structure-based design, Ki is what a co-crystal structure of the non-covalent complex (often captured using a reduced-reactivity warhead mimic) reports on: hinge hydrogen bonds, hydrophobic back-pocket occupancy, and — most importantly — the geometry (distance and Bürgi–Dunitz trajectory angle) between the electrophilic carbon and the target nucleophile's sulfur. A Ki in the low micromolar range with poor warhead geometry will show negligible kinact; conversely, a modest Ki (10–20 µM) with ideal trajectory can still yield a highly efficient inhibitor because kinact/Ki depends on both terms multiplicatively in the productive regime.
Once the EI complex is formed, the electrophilic warhead — typically a Michael acceptor (acrylamide, propiolamide), an SN2 electrophile (chloroacetamide, epoxide), or a reversible-covalent electrophile (nitrile, boronic acid, aldehyde) — reacts intramolecularly with a proximal nucleophilic side chain. For an irreversible inhibitor this bond, once formed, does not break on any physiologically relevant timescale. kinact is a unimolecular first-order rate constant (units: time⁻¹) describing how fast EI converts to E–I; it is independent of [I] because the inhibitor is already bound.
kinact is governed by three coupled factors: intrinsic electrophile reactivity, the microenvironment-tuned nucleophilicity of the target residue, and the geometric alignment locked in by the Ki complex.
Warhead classes, roughly ordered by intrinsic reactivity toward thiols: • Chloroacetamide / bromoacetamide — SN2 alkylation; fast but often insufficiently selective (reacts with free cysteine and glutathione readily) unless heavily geometrically constrained. • Acrylamide (α,β-unsaturated amide) — the dominant clinical warhead (afatinib, osimertinib, ibrutinib, neratinib all use acrylamides against EGFR-Cys797 or BTK-Cys481); Michael addition kinetics tunable over 100-fold by substituting the acrylamide β-carbon. • Propiolamide / alkynyl amide — more electrophilic Michael acceptor, faster kinact, used when a shallow, fast-exchanging pocket limits achievable Ki. • Nitrile / boronic acid / α-ketoamide — form reversible covalent adducts (thioimidate, boronate ester) with measurable off-rates; used when full irreversibility is undesirable (e.g., nirmatrelvir against SARS-CoV-2 3CLpro uses a nitrile forming a reversible thioimidate). • Vinyl sulfonamide / vinyl sulfone — highly electrophilic, fast kinact, used in activity-based protein profiling (ABPP) probes more often than in drugs due to selectivity liabilities.
Nucleophile activation: the reactive cysteine's thiol (pKa ~8.3 free in solution) is frequently perturbed by the local active-site electrostatic environment — proximity to a backbone amide dipole or a basic residue can lower the effective pKa toward 6–7, increasing the fraction present as the highly nucleophilic thiolate (S⁻) at physiological pH and directly accelerating kinact. This is why the identical acrylamide warhead can show a 50-fold difference in kinact against two different cysteines depending on local environment (KRAS G12C's Cys12 versus a generic surface cysteine, for example).
Geometric alignment: the Bürgi–Dunitz trajectory (the ~105° angle of nucleophilic attack on an sp² electrophilic carbon) must be satisfiable within the conformational ensemble sampled by the bound EI complex. Molecular dynamics and QM/MM transition-state modeling are increasingly used during lead optimization to predict which vector off the recognition scaffold will deliver the warhead along a productive trajectory, since a mispositioned but bound warhead can leave Ki excellent while kinact remains vanishingly small.
Because covalent inactivation removes active enzyme permanently, product-formation progress curves recorded at fixed [I] bend away from linearity over time — the instantaneous velocity decays exponentially as more enzyme is trapped in the E–I state. Fitting each trace to Y = (v0/kobs)·(1−e^(−kobs·t)) + offset extracts a single observed first-order rate constant, kobs, for that specific inhibitor concentration. Repeating this across a concentration series is the experimental backbone of kinact/Ki determination.
Practical protocol for a kinact/Ki determination (continuous, "Method A" per Copeland's Evaluation of Enzyme Inhibitors, 2nd ed.):
1. Pre-equilibrate enzyme with substrate at Km or below (to avoid substrate protection artifacts where the substrate competes with inhibitor for active-site occupancy and confounds apparent Ki).
2. Initiate reactions by adding inhibitor across a concentration series spanning roughly 0.3×–10× the anticipated Ki (typical series: 0, 0.5, 1, 2.5, 5, 10, 25, 50 µM), each in triplicate, alongside a DMSO-only (vehicle) control defining v0.
3. Record continuous fluorescence or absorbance for 60–120 minutes at 25–37°C on a plate reader capable of kinetic mode (e.g., BMG PHERAstar, Molecular Devices SpectraMax).
4. Fit each concentration's raw trace individually to the one-phase exponential association/decay equation above using nonlinear least squares (GraphPad Prism, KaleidaGraph, or Python scipy.optimize.curve_fit); extract kobs and its 95% confidence interval per concentration.
5. Confirm the control trace is linear (no time-dependent inhibition in the absence of compound) and that vehicle DMSO concentration is held constant (typically ≤1% v/v) across all wells to avoid solvent-dependent kobs artifacts.
A critical control unique to covalent inhibitor kinetics is the jump-dilution (or rapid-dilution) assay: pre-incubate enzyme with a saturating concentration of inhibitor, then dilute 100-fold into substrate-containing buffer and monitor whether activity recovers. True irreversible inhibitors show no recovery of activity over hours, confirming the bond is covalent and not merely a very tight reversible complex (which would slowly re-equilibrate and show partial activity recovery). This distinguishes genuine kinact/Ki inactivators from ultra-high-affinity reversible binders that can superficially mimic time-dependent inhibition in a standard progress-curve assay.
A single kobs value cannot separate binding affinity from chemical reactivity — both are folded together. The resolving step is a secondary replot: kobs measured at each [I] in stage 3 is plotted against [I] itself, and the entire dataset is fit to the saturating hyperbolic relationship kobs = kinact·[I] / (Ki + [I]). At low [I] the relationship is nearly linear with slope kinact/Ki (second-order regime); at high, saturating [I], kobs asymptotes to kinact itself (the pocket is always occupied, so the rate-limiting step becomes purely chemical).
Two equivalent analytical routes recover kinact and Ki from the kobs-versus-[I] dataset:
Direct nonlinear regression (preferred with modern software): fit kobs = kinact·[I]/(Ki+[I]) directly by nonlinear least squares across all concentrations simultaneously. This is statistically preferable because it does not distort the error structure of the data the way linear transforms do, and it naturally weights each point according to its actual experimental variance rather than an inverted, compressed scale.
Kitz–Wilson double-reciprocal linearization (historical, still widely reported): taking the reciprocal of both sides yields 1/kobs = (Ki/kinact)·(1/[I]) + 1/kinact — a Lineweaver–Burk-style straight line. Plotting 1/kobs against 1/[I] gives a y-intercept of 1/kinact and a slope of Ki/kinact, from which both constants are recovered algebraically. This method remains popular because it is easy to present in a figure and was the original 1962 Kitz–Wilson approach, but it disproportionately weights low-concentration (high 1/[I]) points, which are often the noisiest, and can bias parameter estimates. Modern best practice runs the double-reciprocal plot as a visual sanity check while reporting parameters from direct nonlinear fitting.
A well-designed concentration series is essential: at least two concentrations should approach saturation (kobs approaching kinact plateau) and at least two should be well below Ki (where kobs ≈ (kinact/Ki)·[I], the pure second-order regime). If every concentration tested falls far below Ki, the data can only constrain the ratio kinact/Ki and not the individual constants — a common pitfall when compound solubility caps the achievable [I] below several multiples of Ki. In such solubility-limited cases, laboratories often report only kinact/Ki (obtained from the linear low-concentration regime) rather than over-interpreting an unresolved hyperbola, explicitly flagging that Ki and kinact individually carry wide confidence intervals.
A 2019 J. Med. Chem. case study on a KRAS G12C inhibitor series reported an apparent kinact/Ki of 4,200 M⁻¹s⁻¹ when only three sub-saturating concentrations were tested, but re-measurement with a wider concentration series revealed Ki was actually >200 µM with a fast kinact of 0.6 min⁻¹ — the compound was chemically reactive but poorly recognized, a profile invisible from kinact/Ki alone and only resolved by properly saturating the secondary replot.
Under the sub-saturating conditions relevant to most cellular and in vivo exposures ([I] << Ki), the second-order composite constant kinact/Ki — not Ki alone and not kinact alone — is the number that predicts how fast target inactivation proceeds, exactly as kcat/Km (the specificity constant) predicts substrate turnover efficiency for an enzyme. Because the covalent bond is permanent, a covalent inhibitor's residence time is formally infinite: pharmacodynamic recovery depends entirely on target resynthesis (protein turnover half-life), not on drug washout — a fundamentally different pharmacology from reversible inhibitors.
kinact/Ki has units of M⁻¹s⁻¹ (or M⁻¹min⁻¹), identical in form to a bimolecular rate constant, and it is the appropriate metric for comparing covalent inhibitors across chemical series or against reversible competitors at pharmacologically relevant, sub-saturating exposures. Two compounds with identical kinact/Ki can have wildly different individual Ki and kinact values — one might achieve potency through tight, well-optimized non-covalent recognition (low Ki, modest kinact), while another achieves the same net rate through a highly reactive warhead compensating for weak recognition (high Ki, fast kinact). These profiles are pharmacologically distinct: the first is generally safer, since target engagement requires genuine molecular recognition; the second risks broad off-target covalent modification because a sufficiently reactive electrophile will eventually react with any solvent-accessible cysteine it samples, recognition or not.
Because inactivation is irreversible, pharmacodynamic duration of action decouples from plasma drug exposure — once a target protein is covalently modified, functional recovery requires new protein synthesis. This is why covalent kinase inhibitors like ibrutinib (BTK, kinact/Ki ≈ 35,000 M⁻¹s⁻¹ at Cys481) or KRAS G12C inhibitors like sotorasib and adagrasib (Cys12 within the switch-II pocket) can be dosed once or twice daily despite plasma half-lives of only a few hours: sustained target suppression is maintained by residual covalent occupancy from the prior dose while new protein resynthesizes over its natural turnover half-life (often 12–30 hours for many signaling kinases).
Selectivity triage for covalent chemotypes requires assays beyond the standard kinase or enzyme panel:
• Glutathione (GSH) reactivity assay: incubate the free warhead (or full compound) with excess reduced glutathione and monitor depletion by LC-MS/MS or a colorimetric Ellman's-reagent-based readout; a GSH half-life under ~2 hours flags a warhead as intrinsically over-reactive regardless of target-specific kinact/Ki.
• Chemoproteomic profiling (isoTOP-ABPP, competitive activity-based protein profiling): a cysteine-reactive alkyne probe is used to map compound engagement across thousands of endogenous cysteines in a cell lysate or intact-cell system by quantitative LC-MS/MS, directly measuring selectivity across the proteome rather than a curated kinase panel alone.
• Paralog/isoform panel kinact/Ki comparison: measuring kinact/Ki against the nearest structural paralogs (e.g., BTK versus ITK, TEC, EGFR versus HER2/HER4) quantifies achieved selectivity ratios directly in the same kinetic units as potency, avoiding the ambiguity of comparing IC50 values measured at different incubation times across targets with different intrinsic kinact.