Rescuing a potent lead from hERG liability without losing target potency
A discovery programme has a lead compound, A, that inhibits its peripheral G-protein-coupled receptor target with an IC50 of 20 nM. A is a tertiary amine with pKa 9.1 and logD7.4 of 4.6. It inhibits the hERG channel with an IC50 of 0.3 micrometre in patch-clamp, shows time-dependent inhibition of CYP3A4, has a mouse plasma half-life of 0.8 h, is 99.5% protein bound, and achieves brain-to-plasma ratios above 3, which is undesirable because the target is not expressed in the CNS. Two analogues exist. B (amine replaced by a morpholine amide, logD 2.4) has hERG IC50 of 12 micrometres but target potency falls to 400 nM. C (the amine replaced by a carboxylic acid, logD 1.8) has hERG IC50 above 30 micrometres, target potency of 60 nM, but Caco-2 permeability of 0.5 x 10^-6 cm/s and oral bioavailability of 3% in mouse. Management wants a molecule with hERG IC50 above 10 micrometres, target IC50 below 50 nM, and measurable oral exposure.
1. Explain how the physicochemical properties of A drive the hERG, CYP3A4 and CNS liabilities, and what structural features you would target for change.
Read the worked answer
hERG blockade is promoted by three structural properties, all present in A: a protonatable basic centre (pKa above about 8) that forms a cation-pi interaction with Tyr652 and Phe656 in the pore, high lipophilicity (logD above 3) that drives membrane partitioning to the channel, and an aromatic-rich scaffold that stacks against the pore residues. At logD 4.6 and pKa 9.1, A is more than 99% protonated at physiological pH, which is precisely the combination that gives submicromolar hERG block.
The same lipophilicity and basicity explain the CYP3A4 time-dependent inhibition: the basic nitrogen can be oxidised to a nitroso or iminium intermediate that coordinates the haem iron, so the inhibition is mechanism-based and will produce drug-drug interactions even at low exposure. High logD plus a protonated amine also explains the brain-to-plasma ratio above 3: passive permeability is high and efflux is not limiting, so the compound distributes into CNS tissue.
Structural direction: reduce the pKa of the basic centre below 7 (add an electron-withdrawing substituent beta to the amine, or replace the amine with an amide, sulfonamide or oxetane-containing amine), lower logD by 1.5-2 units, remove one of the aromatic rings in the hERG pharmacophore if it is not needed for potency, and block the metabolic soft spot that generates the reactive intermediate. Reducing basicity and lipophilicity simultaneously should also reduce CNS penetration through increased polar surface area.
2. Using analogues B and C, propose a design strategy that could satisfy all three criteria, and justify why neither B nor C alone is acceptable.
Read the worked answer
B fails on potency: replacing the basic amine with a morpholine amide removes the key ionic or hydrogen-bond interaction with the target, so the 20-fold drop from 20 nM to 400 nM reflects a lost pharmacophore contact, not a tolerable trade-off. C confirms that the hERG problem is fixable by removing basicity, but the carboxylic acid converts the molecule into a low-permeability entity: at physiological pH the carboxylate is ionised, the polar surface area rises, and Caco-2 permeability of 0.5 x 10^-6 cm/s plus 3% oral bioavailability means no usable exposure.
A workable design combines both lessons. First, keep the pharmacophore that gives 20 nM potency but relocate it away from a basic centre: replace the tertiary amine of A with a weakly basic or neutral isostere such as a fluoro-substituted pyrrolidine, an oxetane, or a sulfoximine, targeting a pKa of 6.0-6.5 so the molecule is largely neutral at pH 7.4 while retaining a weak hydrogen-bond acceptor. Second, lower logD to 2.5-3.0 by adding a polar substituent (hydroxyl, primary amide, or ring nitrogen) at a position shown by the B analogue to tolerate change. Third, if a carboxylic acid is needed for the potency and hERG profile, make it a prodrug: the ethyl or isopropyl ester of C would restore passive permeability and be hydrolysed systemically, though this adds esterase-dependent variability and a stability programme.
A practical starting point is C with the acid masked as an ester or replaced by a bioisosteric tetrazole or acyl sulfonamide that is less acidic, then screening for hERG and permeability in parallel.
3. What in vitro and in vivo package would you require before selecting a candidate, and how would you interpret a negative patch-clamp result?
Read the worked answer
In vitro, use manual or automated patch clamp on hERG-expressing HEK or CHO cells at 37 degrees C with a minimum of three concentrations plus a positive control, and report IC50 with the Hill slope rather than a single-point percentage. Run it alongside a dofetilide displacement binding assay only as a screening filter, not a substitute, because binding assays miss state-dependent block. Pair this with a Nav1.5 and Cav1.2 screen to detect mixed ion-channel effects, a Caco-2 or MDCK permeability assay with efflux ratio, microsomal and hepatocyte stability across species, a CYP3A4 time-dependent inhibition assay with a pre-incubation step and dilution, and plasma protein binding by equilibrium dialysis.
In vivo, a mouse or rat pharmacokinetic study with intravenous and oral arms to obtain clearance and bioavailability, then a telemetered dog or non-rodent cardiovascular study with free-drug concentration measurement, since the interpretation depends on the unbound fraction: at 99.5% protein binding, a total plasma concentration of 10 micrometres corresponds to only 50 nM free, which may be at or below the hERG IC50.
A negative patch-clamp result must be interpreted against the unbound exposure achieved at therapeutic doses: if the free Cmax is 100-fold below the hERG IC50, the negative result means little and a higher-exposure or metabolite scenario should be checked. Conversely, a compound with a hERG IC50 of 12 micrometres and a free Cmax of 0.1 micrometres gives a 120-fold margin, which is a defensible starting point. Finally, check whether the metabolite rather than the parent carries the hERG liability, using metabolite identification from hepatocyte incubations.