Size-Dependent Target Engagement of Covalent Probes

Abstract

Labeling proteins with covalent ligands is finding increasing use in proteomics applications, including identifying nucleophilic residues amenable for labeling and in the development of targeted covalent inhibitors (TCIs). Labeling efficiency is measured by the covalent occupancy of the target or by biochemical activity. Here, we investigate how these observed quantities relate to the intrinsic parameters of complex formation, namely, noncovalent affinity and covalent reactivity, and to experimental conditions, including incubation time and ligand concentration. It is shown that target engagement is beneficially driven by noncovalent recognition for lead-like compounds, which are appropriate starting points for targeted covalent inhibitors owing to their easily detectable occupancy and fixed binding mode, facilitating optimization. In contrast, labeling by fragment-sized compounds is inevitably reactivity-driven as their small size limits noncovalent affinity. They are well-suited for exploring ligandable nucleophilic residues, while small fragments are less appropriate starting points for TCI development.

Figure 1

Two steps of covalent ligand binding and its schematic free energy profile.

Figure 2

Simulation of occupancy % versus ligand concentration ([L]) or incubation time (t): (a) Occupancy % versus ligand concentration ([L]) for two-step mechanism with k_inact = 10^–3 s^–1, t = 16 h, and several K_I (M). (b) Occupancy % versus ligand concentration ([L]) for two-step mechanism with K_I = 10^–5 M, t = 16 h, and several k_inact (s^–1). (c) Occupancy % versus ligand concentration ([L]) for three-step mechanism where ligand binding precedes deprotonation of the reactive residue. Curves calculated for several pK_a values using k_inact = 5·10^–4 s^–1, K₁ = 10^–5 M, t = 16 h, and pH = 7. (d) Occupancy % versus incubation time (t) for two-step mechanism with k_inact = 10^–3 M, [L] = 10^–7 M, and several K_I (M).

Figure 3

Simulation of occupancy % versus ligand concentration ([L]) for fragment (K_I = 10^–3 M, k_inact = 10^–3 s^–1, and t = 24 h), lead-like (K_I = 10^–5 M, k_inact = 10^–4 s^–1, and t = 24 h) and three drug (acalabrutinib K_I = 1.8·10^–7 M, k_inact = 5.6·10^–3 s^–1, and t = 30 min; ibrutinib K_I = 5.4·10^–8 M, k_inact = 2.7·10^–2 s^–1, and t = 30 min; adagrasib K_I = 3.7·10^–6 M, k_inact = 1.3·10^–1 s^–1, and t = 30 min) compounds.

Figure 4

Simulation of occupancy % versus ligand concentration ([L]) for KRAS^G12C labeling by ARS-853 at several pH values using the three-step model of labeling (K_I = 5.62·10^–6 M, pK_a = 9.2, k_inact = 0.80 s^–1, and t = 1 h).

Figure 5

K_I and k_inact are for compounds in Tables 1–3. Lead-like compounds are shown with black squares. Drugs are shown with green squares with K_I and k_inact taken from refs (34) and (35). Occupancy increases from bottom right to top left. 50% occupancies calculated with parameters [L] = 2 μM, t = 2 h, and [L] = 2 μM, t = 16 h applied for determining occupancies in Tables 1–3 are indicated.