Mechanistic understanding of KCNQ1 activating polyunsaturated fatty acid analogs

Figure 1.. Previously described binding Sites I and II for LIN. (A) Sideview cartoon of KCNQ1 subunits S1–S6. (B) Top view of KCNQ1 with S1–S6 labeled and the voltage sensing domain (VSD) and pore domain (PD) circled. (C) Side and top view of binding site regions I and II for LIN. (D) Residues previously described in simulations to be important for LIN binding at Site I and II. Residues only suggested in simulation are shown in gray. Residues confirmed by mutation are shown in color. All simulation data and mutation data for L272, Y268, F269, A290, and L263 in this figure is from Yazdi et al. (2021). Mutation data for R218 and K316 is from Liin et al. (2018).

Figure 2.. LIN-Gly and LIN-Tyr have larger voltage shifts but smaller Gmax increases as compared to LIN. (A) Chemical structures of LIN, LIN-Gly, and LIN-Tyr. (B) Current of WT KCNQ1 with and without 2 µM of LIN-Gly, demonstrating the shift in voltage dependence toward negative voltages (ΔV0.5) and increase in maximal conductance (ΔGmax) effects by addition of PUFA. (C) Current traces of KCNQ1 before and after application of 7 µM of LIN (black), LIN-Gly (green), and LIN-Tyr (blue). (D) ΔV0.5 dose-response curve of LIN (n = 8), LIN-Gly (n = 10), and LIN-Tyr (n = 6). (E) Gmax dose-response curve of LIN (n = 8), LIN-Gly (n = 7), and LIN-Tyr (n = 6). (F) ΔV0.5 effect of LIN (−5.45 ± 0.69 mV; n = 8), LIN-Gly (−25.28 ± 1.69 mV; n = 10), and LIN-Tyr (−44.51 ± 5.05 mV; n = 6) at 7 µM. ANOVA with Tukey’s multiple comparison gave P < 0.0001 between all compounds. (G) Gmax effect of LIN (1.39 ± 0.13; n = 8), LIN-Gly (1.3 ± 0.06; n = 7) and LIN-Tyr (1.01 ± 0.09; n = 6) at 7 µM. ANOVA with Tukey’s multiple comparison gave P < 0.0001 between LIN and LIN-Tyr, P = 0.0005 between LIN-Gly and LIN-Tyr, and P = 0.25 between LIN and LIN-Gly. Data are given as mean ± SEM. LIN data from Yazdi et al. (2021). ***,…

Figure 3.. LIN-Gly and LIN-Tyr interact with many residues on the KCNQ1 channel. Contact analysis for AA LIN-Gly and LIN-Tyr simulations. The gray highlighted boxes in the background indicate the position of the transmembrane helices. (A) Histogram and heat map of total interaction time of each residue in KCNQ1 for LIN-Tyr and LIN-Gly in the Site I and Site II systems. (B) Histogram and heat map of longest lifetime interactions of each residue in KCNQ1 for LIN-Tyr and LIN-Gly in the Site I and Site II systems.

Figure S1.. Most likely lifetimes of LIN-Tyr interactions from Maximum-Likelihood fits. Most likely lifetimes of (top) long interaction events and (bottom) short interaction events for LIN-Tyr for each residue calculated from data of both the LIN-Tyr Site I and Site II simulations (Fig. 3). The starting configuration (i.e., a LIN-Tyr in Site I or Site II) had little effect of the distributions (see Fig. 3) as a number of LIN-Tyr molecules bind and unbind from the different residues. Therefore, the interaction events from the two simulations were combined to increase the number of events for better statistics. The total number of events in each case is also shown. Data given as mean ± SD.

Figure S2.. Most likely lifetimes of LIN-Gly interactions from Maximum-Likelihood fits. Most likely lifetimes of (top) long interaction events and (bottom) short interaction events for LIN-Gly for each residue calculated from data of both the LIN-Gly Site I and Site II simulations (Fig. 3). The starting configuration (i.e., a LIN-Gly in Site I or Site II) had little effect of the distributions (see Fig. 3) as a number of LIN-Gly molecules bind and unbind from the different residues. Therefore, the interaction events from the two simulations were combined to increase the number of events for better statistics. The total number of events in each case is also shown. Data given as mean ± SD.

Figure 4.. Representative poses of a LIN-Gly and LIN-Tyr constant binders at previously described Sites Iα and Iβ and Site II. Site I was split into Site Iα and β to better show the two major but different confirmations of PUFAs at Site I. We have previously published this separation of Site I into Site Iα and β in Yazdi et al. (2021). Red residues are making head group interactions with the PUFAs. (A) Side view of representative LIN-Gly constant binder at Sites Iα, Iβ, and II. (B) Overlay of all LIN-Gly constant binder poses at Sites Iα, Iβ, and II. The LIN-Gly head group is shown in red and the tail in green. (C) Side view of representative LIN-Tyr constant binder at Sites Iβ and II. (D) Overlay of all LIN-Tyr constant binder poses at Sites Iβ and II. The LIN-Tyr head group is shown in red and the tail in blue.

Figure 5.. LIN, LIN-Gly, and LIN-Tyr require residue R218 to maintain their voltage dependence shifting effect. Experiments were done on the human KCNQ1 channel with the mutation R228Q; the Xenopus numbering R218 is used here for ease of comparison with the simulations. (A) Model of PUFA interaction with S4 that causes the V0.5 shifting effect, with residue R218 in the voltage sensor marked. (B) Conductance versus voltage (G(V)) curve of the R218Q and WT KCNQ1 currents with and without 7 µM of LIN-Gly. (C–E) ΔV0.5 dose response curve of (C) LIN (WT n = 8 and R218Q n = 3), (D) LIN-Gly (WT n = 10 and R218Q n = 3), and (E) LIN-Tyr (WT n = 6 and R218Q n = 5) on WT versus R218Q mutated KCNQ1 channels. Data are shown as mean ± SEM. (F–H) Comparison of ΔV0.5 effect on WT KCNQ1 channels versus R218Q mutated channels at 7 µM of (F) LIN (WT: −5.45 ± 0.69 mV; n = 8 and R218Q: −1.04 ± 0.54 mV; n = 3); a t test gave P = 0.005; (G) LIN-Gly (WT: −25.28 ± 1.69 mV; n = 10 and R218Q: −6.03 ± 5.43 mV; n = 3); ANOVA with Tukey’s multiple comparison, also containing the mutations in Fig. 8, gave P = 0.0015; (H) LIN-Tyr (WT: −43.57 ± 1.03 mV; n = 6 and R218Q: −10.89 ± 4.61 mV; n = 5), ANOVA with Tukey’s multiple…

Figure 6.. LIN-Gly has a more condensed pattern of interaction than LIN-Tyr. Regions of LIN-Gly and LIN-Tyr binding on the KCNQ1 channel. (A) Map of residues with interaction times longer than 2 µs with LIN-Gly molecules. (B) Map of residues with interaction times longer than 2 µs with LIN-Tyr molecules.

Figure 7.. Larger head groups allow for more and simultaneous interactions. Each residue is shown with the percent of time it is interacting with the PUFA throughout the simulation. (A) Representation of constant binders of LIN-Gly in Site Iα. (B) Representation of constant binders of LIN-Gly and LIN-Tyr in Site Iβ. (C) Representation of constant binders of LIN-Gly and LIN-Tyr in Site II.

Figure 8.. Mutagenesis experiments show the importance of interacting residues from simulations. Experiments were done on human KCNQ1 channel with the mutations Y281F, Y278F, and K285Q; the Xenopus numberings Y271F, Y268F, and K275Q were used here for ease of comparison with the simulations. (A) ΔV0.5 dose response curve of LIN-Gly on WT and point mutation KCNQ1 channels (WT n = 10, Y271F n = 3, Y268F n = 4, K275Q n = 5). (B) ΔV0.5 dose response curve of LIN-Tyr on WT and point mutation KCNQ1 channels (WT n = 6, Y271F n = 4, Y268F n = 4, K275Q n = 3). (C) Comparison of ΔV0.5 effect on WT KCNQ1 channels versus point mutations at 7 µM of LIN-Gly (WT: −25.28 ± 1.69 mV; n = 10, Y271F: −25.49 ± 2.53 mV; n = 3, Y268F: −13.91 ± 1.7 mV; n = 4, K275Q: −7.66 ± 3.04 mV; n = 5). ANOVA with Tukey’s multiple comparison of WT to each point mutation gave ns for Y271F, P = 0.067 for Y268F and P = 0.0006 for K275Q. ***, P < 0.001. (D) Comparison of ΔV0.5 effect on WT KCNQ1 channels versus point mutations at 7 µM of LIN-Tyr (WT: −43.57 ± 1.03 mV; n = 3, Y271F: −52.28 ± 2.98 mV; n = 4, Y268F: −48.23 ± 1.15 mV; n = 4, K275Q: −39.47 ± 14.8 mV; n = 3). ANOVA with Tukey’s multiple comparison of WT to each point…