The molecular determinants of R-roscovitine block of hERG channels

Abstract

Human ether-à-go-go-related gene (Kv11.1, or hERG) is a potassium channel that conducts the delayed rectifier potassium current (IKr) during the repolarization phase of cardiac action potentials. hERG channels have a larger pore than other K+channels and can trap many unintended drugs, often resulting in acquired LQTS (aLQTS). R-roscovitine is a cyclin-dependent kinase (CDK) inhibitor that induces apoptosis in colorectal, breast, prostate, multiple myeloma, other cancer cell lines, and tumor xenografts, in micromolar concentrations. It is well tolerated in phase II clinical trials. R-roscovitine inhibits open hERG channels but does not become trapped in the pore. Two-electrode voltage clamp recordings from Xenopus oocytes expressing wild-type (WT) or hERG pore mutant channels (T623A, S624A, Y652A, F656A) demonstrated that compared to WT hERG, T623A, Y652A, and F656A inhibition by 200 μM R-roscovitine was ~ 48%, 29%, and 73% weaker, respectively. In contrast, S624A hERG was inhibited more potently than WT hERG, with a ~ 34% stronger inhibition. These findings were further supported by the IC50 values, which were increased for T623A, Y652A and F656A (by ~5.5, 2.75, and 42 fold respectively) and reduced 1.3 fold for the S624A mutant. Our data suggest that while T623, Y652, and F656 are critical for R-roscovitine-mediated inhibition, S624 may not be. Docking studies further support our findings. Thus, R-roscovitine's relatively unique features, coupled with its tolerance in clinical trials, could guide future drug screens.

Fig 1. Dose-response curve for R-roscovitine inhibition of WT hERG channels.

A) Skeletal formula of R-roscovitine, 2-(1-ethyl-2-hydroxyethylamino)-6-benzylamino-9-isopropylpurine. B) Pulse protocol (top) along with representative traces from a WT cell (bottom) exposed to the indicated R-roscovitine concentrations. Channels were activated with a 1-second step to +40 mV from -80 mV, and tail currents (◊) were elicited by a 1-second repolarization to -50 mV. The single 2-second episodic pulse was repeated for 3.44 minutes as the various concentrations were washed in and out of the gravity-fed perfusion system. C) Time-course of tail current inhibition (measured from ◊, in B) from a representative cell expressing WT hERG in the presence of the indicated R-roscovitine concentrations. D) Fractional block data were fitted with a Hill equation (see methods) to obtain the IC₅₀ for WT hERG (196 ± 12 μM, n = 11). Error bars here and elsewhere represent standard errors.

Fig 2. Characteristics of R-roscovitine inhibition of hERG channels.

A) Voltage protocol and current traces from a representative cell expressing hERG channels before (Control) and during 200 μM R-roscovitine application. From a holding potential of -80 mV, outward currents were elicited at two phases: depolarization (from -50 mV to +60 mV) and repolarization (to -50 mV). Mean step currents were measured at the end of the 2-second step (▽) and tail currents were measured from their peak (◊). B) Normalized I-V curves for Control and 200 μM R-roscovitine from mean step currents (▽, n = 14). Inhibition was strongest between -30 mV and +30 mV, a range where the open state is more prevalent than either the closed or inactivated state. C) Average percent step current inhibition by 200 μM R-roscovitine at various step voltages. Inhibition was weakest at voltages where channels are closed (-50 mV) or inactivated (+50 mV; p = 0.0052 for 0 vs. -50 mV, p < 0.0001 for 0 vs. +50 mV, n = 14, one-way ANOVA). D) Tail I-V curves for Control and 200 μM R-roscovitine, with normalized peak tail currents (◊, n = 14). Smooth lines are Boltzmann fits to the activation curves, which produced significantly different V_0.5 (Ctrl: -16.5 ± 1.1 mV, Rosc: -19.4 ± 1.1 mV; p = 0.0105, n = 14, paired t-test) and slopes (Ctrl: 11.7 ± 0.5, Rosc: 10.6 ± 0.5; p = 0.0114; n = 14, paired t-test). The dashed line shows a scaled activation curve in the presence of R-roscovitine. E) Average percent tail current inhibition by 200 μM R-roscovitine at various step voltages. The inhibition of tail current significantly increased with rising levels of depolarized potentials (p = 0.0134 for -40 vs. 0 mV, p = 0.0004 for -40 vs. +60 mV, n = 14, one-way ANOVA). Error bars represent standard errors; when not visible, error bars are smaller than the symbols. * = P < 0.05, ** = P < 0.01, and *** = P < 0.001.

Fig 3. Comparison of unaffected step and tail currents from WT hERG, S624A, and Y652A hERG mutants.

A) Step I-V curves for WT (○, blue), S624A (□, red), and Y652A (△, green) from a step depolarization protocol (top; currents measured at ▽). Rectification of S624A currents was reduced in comparison to WT current, with more than twice the amount of normalized current remaining for S624A (0.77 ± 0.06) compared to WT (0.28 ± 0.04) at +60 mV. B) WT, S624A, and Y652A tail I-V curves from a step depolarization protocol (top; currents measured at ◊). Boltzmann fits generated the following V_0.5 of activation (neither of which were different from WT; p > 0.8 for WT-mutant comparisons): WT V_0.5 = -16.5 ± 1.1 mV, S624A V_0.5 = -17.7 ± 3.2 mV, and Y652A V_0.5 = -16.5 ± 1.0 mV. The slopes for the activation curves were 11.7 ± 0.45 for WT, 16.3 ± 1.62 for S624A (p = 0.0033 compared to WT), and 12.5 ± 0.84 for Y652A (p = 0.7913 compared to WT); n_WT = 14, n_S624A = 9, n_Y652A = 9, one-way ANOVAs.

Fig 4. S624A hERG channels have an increased sensitivity to R-roscovitine.

A) Voltage protocol and representative current traces, from a cell expressing S624A hERG, before and during 200 μM R-roscovitine application. Note that tail currents were greatly diminished upon R-roscovitine application. B) S624A step I-V curves before and during 200 μM R-roscovitine application (currents measured at ▽ in A). The strongest inhibition occurred between -30 mV and +30 mV. C) Percent step inhibition of WT and S624A at varying step voltages. S624A was inhibited more potently than WT at intermediate voltages, where open probability is high (p < 0.05 for WT vs. S624A between -20 mV and 0 mV, n_WT = 14, n_S624A = 9, one-way ANOVA). D) S624A tail I-V curves measured at ◊. Smooth lines are Boltzmann fits that generated V_0.5 of activation in Control: -17.7 ± 3.2 mV and in R-roscovitine: -3.3 ± 6.1 mV; ns, p = 0.09. The slopes were 16.3 ± 1.6 in Control, and 44.5 ± 3.3 in R-roscovitine (p < 0.0001; n = 9, paired t-tests). Dashed line indicates the fit to normalized currents in R-roscovitine. E) Percent tail inhibition of WT and S624A. When compared to WT over a range of voltages, levels of S624A tail current inhibition were much larger (p < 0.05 for WT vs. S624A between -20 mV and +50 mV, n_WT = 14, n_S624A = 9, one-way ANOVA). F) Concentration-response relationship for S624A. The S624A IC₅₀ (152 ± 22 μM) was not significantly lower than the WT IC₅₀ (196 ± 12 μM;p = 0.8323 for WT vs. S624A) but the dose-response slope for S624A was statistically smaller (0.84 ± 0.08 vs 1.5 ± 0.1 for WT; p = 0.0003, n_WT = 11, n_S624A = 9, one-way ANOVA). ** = P < 0.01, and *** = P < 0.001.

Fig 5. The Y652A mutation attenuates hERG inhibition by R-roscovitine.

A) Voltage protocol and representative current traces of a cell expressing Y652A hERG before and during 200 μM R-roscovitine application. B) Y652A step I-V curves before and during 200 μM R-roscovitine inhibition (currents measured at ▽). The curves were bell-shaped and displayed inhibition at intermediate voltages. C) Percent step inhibition of WT currents were much stronger than that of Y652A between -20 mV and +10 mV (p < 0.05 for WT vs. Y652A, n_WT = 14, n_Y652A = 9, one-way ANOVA). D) Boltzmann fits of the Y652A tail I-V curves from the step depolarization protocol (currents measured at ◊) showed a significant shift in V_0.5 and slopes for the Ctrl vs Rosc activation curves. Ctrl V_0.5 = -16.5 ± 1.0 mV, and a Rosc V_0.5 = -19.6 ± 0.5 mV (p = 0.0094), and activation slopes were 12.5 ± 0.8 for Ctrl and 9.3 ± 0.3 for Rosc. (p = 0.0025, n = 11, paired t-tests). Dashed line indicates the fit to normalized currents in R-roscovitine. E) Percent tail inhibition of WT and Y652A at their respective step voltages. Levels of Y652A tail current inhibition were significantly reduced compared to WT between -20 mV and +60 mV (p < 0.05 for WT vs. Y652A, n_WT = 14, n_Y652A = 9, one-way ANOVA). F) Concentration-response relationship for Y652A. When compared to WT IC₅₀ (196 ± 12 μM), Y652A had a ~ 2.9-fold increase in IC₅₀ (567 ± 122 μM; p = 0.0005). The slope of the dose response curve significantly changed from 1.522 ± 0.1 for WT to 1.0 ± 0.14 for Y652A (p = 0.0052; n_WT = 11, n_Y652A = 8, one-way ANOVAs). * = P < 0.05, ** = P < 0.01, and *** = P < 0.001.

Fig 6. Weaker hERG current inhibition in the high K⁺ solution.

A) Representative WT tail currents in high K⁺ prior to (Control) and during the indicated R-roscovitine concentrations, elicited by the protocol above. Tail currents from the dashed boxed region are expanded for clarity. B) Time course of inhibition from the cell in A. Peak tail currents were measured at ▽ during which the indicated R-roscovitine concentrations were applied and washed off. Peak tail currents were plotted over time. C) The IC₅₀ for WT hERG in high K⁺ was 513 ± 43 μM (n = 9), which was significantly different from WT in low K⁺ (p <0.0001); the slopes (high K⁺ = 1.1 ± 0.2) were not (low K⁺ = 1.5 ± 0.1; p = 0.091, unpaired t-tests). D) Voltage protocol and representative current traces of a cell before (Ctrl) and during 500 μM R-roscovitine application. E) WT tail I-V curves from currents measured at ◊ in D. The smooth quadratic fits yielded a reversal potential (E_rev) of -11.1 ± 3.3 mV in control, and -14.6 ± 2.4 mV in 500 μM R-roscovitine (p = 0.0237, n = 10, paired t-test). F) Percent tail inhibition calculated from E. Inhibition was significantly stronger at lower voltages than at high voltages (p = 0.0004 for +20 mV vs. -120 mV, p = 0.0002 for +20 mV vs. -80 mV, n = 10, one-way ANOVA). *** = P < 0.001.

Fig 7. T623A hERG reduces R-roscovitine mediated inhibition.

A) Voltage protocol and representative current traces of a cell expressing T623A before (Control) and during 500 μM R-roscovitine application. B) T623A tail I-V curves during the step repolarization protocol (currents measured at ◊). Quadratic fits generated E_rev values (Ctrl = -39.7 ± 2.0 mV, Rosc = -40.1 ± 2.1 mV) that did not show a significant shift occurring with 500 μM R-roscovitine (p = 0.5995 for Ctrl vs. Rosc, n = 10, paired t-test). C) Percent tail inhibition of T623A hERG was significantly weaker over a range of voltages than WT inhibition (p < 0.05 for WT vs. T623A between +20 mV and -120 mV (excluding voltages close to the reversal potential; n_WT = 10, n_T623A = 10, one-way ANOVA). D) Compared to the WT IC₅₀ (513 ± 43 μM), T623A IC₅₀ was ~ 5.5-fold larger (2786 ± 488 μM; p = 0.0092) while the slope factor had a non-significant change from 1.1± 0.22 for WT to 0.7 ± 0.09 for T623A (p = 0.4901; n_WT = 9, n_T623A = 7, Kruskal-Wallis tests). * = P < 0.05, ** = P < 0.01, and *** = P < 0.001.

Fig 8. R-roscovitine inhibition is almost completely abolished in F656A hERG.

A) Voltage protocol and representative current traces of a cell expressing F656A before (Control) and during 500 μM R-roscovitine application. B) F656A tail I-V curves during the step repolarization protocol (currents measured at ◊). Quadratic fits generated to obtain E_rev values (Ctrl = -45.1 ± 1.1 mV, Rosc = -42.1 ± 1.7 mV) did in fact show a shift with drug block (p = 0.02, n = 9, paired t-test). C) F656A tail current inhibition was almost non-existent for most repolarized voltages, which was a substantial difference from WT inhibition (p < 0.01 for WT vs. F656A between +20 mV and -120 mV (excluding points close to the reversal potential), n_WT = 10, n_F656A = 9, one-way ANOVA). D) Concentration-response relationship for F656A hERG IC₅₀ (21.5 ± 10.6 mM) shows a ~ 42-fold increase from WT IC₅₀ (513 ± 43 μM; p = 0.0004); and a significant reduction in the slope of the Hill equation from 1.1 ± 0.22 for WT to 0.42 ± 0.04 for F656; p < 0.005; n_WT = 9, n_F656A = 5, Kruskal-Wallis tests. ** = P < 0.01, and *** = P < 0.001.

Fig 9. Deactivation time constants for hERG and its mutants.

A) Representative traces from the indicated hERG constucts at -100 and -120 mV repolarization voltages. Tail currents (from the dash-boxed region of the voltage protocol) were fitted with a standard biexponential equation (red dashed line) to measure deactivation time constants (see methods). B) Average τ_fast for S624A, Y652A and T623A constructs at -120 mV were all significantly smaller than their respective WT controls (p = 0.0082, p = 0.0134, and p < 0.0001 respectively). At -100 mV, Y652A and T623A were significantly different from their respective WT controls (p = 0.0054 and p = 0.0009), while S624A was not (p = 0.0953). F656A was not different from WT_{High K+} (p = 0.4973 at -120 mV and p = 0.9430 at -100 mV). N = 6–10; one-way ANOVAs. C) τ_fast before (Control) and during R-roscovitine application at -120 mV. R-roscovitine sped up deactivation for WT (p = 0.0034) and T623A (p = 0.0009), but not for S624A (p = 0.312), Y652A (p = 0.2739), or F656A (p = 0.1708). N = 6–10; paired t-tests. R-roscovitine concentration was 200 μM for low K⁺ and 500 μM for high K⁺ solutions. * = P < 0.05, ** = P < 0.01, and *** = P < 0.001.

Fig 10. Molecular docking predicts R-roscovitine binding to the hERG selectivity filter and S6 helix residues.

A) Structure of an open WT hERG homology model [43] based on the KvAP structure [44], with the selectivity filter and S6 helices highlighted in red and green, respectively (left = side view, right = intracellular view). 624–628 are selectivity filter residues and 635–666 are S6 residues. B) Binding site of a lowest-energy R-roscovitine conformation (ball & stick) shown in side-view and an inside-out view through the channel (see docking methods). Colored residues (stick) interacted with R-roscovitine using van der Waals interactions. Hydrogen bonding (red dashed-line) occurred between the backbone carbonyl oxygen of T623 and R-roscovitine. hERG subunit D, which did not interact with R-roscovitine, was removed for clarity in the zoomed-in side view (top panel). The residue color scheme is the same in both views.