Evaluation of PIP2 binding sites on the TRPC6 channel
To investigate the relationship between PIP2 binding and TRPC6 channel activity, we evaluated three features of OAG-induced TRPC6 currents in HEK293 cells. First one was the kinetics of TRPC6 channel deactivation (decay, t1/2) induced by DrVSP-mediated PIP2 depletion. DrVSP was activated by membrane depolarization to + 100 mV for 700 ms. Second was the kinetics of reactivation (recovery, τ) mediated by replenishment of PIP2 after repolarization to − 50 mV. These deactivation and reactivation kinetics can be used to estimate the dissociation and binding processes in reversible first order reactions. For wild-type TRPC6 (TRPC6WT, C6WT), t1/2 (decay) was 229 ± 15 ms, while τ (recovery) was 2.01 ± 0.13 s (Fig. 1B, C, D and Fig. 2). The third feature examined was the functional impact of PIP2 dissociation (i.e. TRPC6 channel inhibition) elicited by DrVSP activation. This inhibition was expressed as the ratio of the current amplitudes after and before DrVSP activation (Ipost/Ipre). This ratio was ~ 0.5 for TRPC6WT (Fig. 1D, Fig. 2 bottom left). Co-expression of an inactive DrVSP mutant has no effect on TRPC6 currents (Fig. 1A,C,D). This indicates that TRPC6 channel activity led to no clear voltage-dependence, which was reported in previous for other TRP channels28 − 30.
However, the kinetics of both the deactivation and the reactivation should reflect several distinct processes including the kinetics of VSP activation/inactivation, the time course of PIP2 dephosphorylation, and the altered channel gating upon PIP2 dissociation and bindings. Among those processes, we measured the dynamics of PIP2 alternation upon activation of DrVSP. The time courses of the depletion (t1/2) and replenishment (τ) of PIP2, as measured by FRET from PIP2 sensor proteins, were very similar to that of the TRPC6 current decay and recovery (135 ± 24 ms and 5.66 ± 0.48 s, respectively, Fig. 1E). This indicates a close correlation between PIP2 level and TRPC6 channel activity, and the kinetics of OAG-induced currents should reflect PIP2-dependent processes that affect TRPC6 channel functionality.
Screening of PIP2 binding domains
TRPC6 channels possess nearly 90 positively charged (basic) residues in each subunit, among which we selected over 30 residues for the first screening based on following criteria: (1) residues were probably located on the intracellular side of the channel and (2) residues were clustered in the primary sequence with other positive residues. Mutant TRPC6 channels were co-expressed with DrVSP in HEK293 cells, and the membrane currents were measured by the whole-cell recordings. We found that following neutralization of basic residues in the distal N-terminal region (K75Q, R78Q, and R73Q/K75Q/R77Q), OAG no longer elicited any TRPC6-mediated currents (Fig. 2). This suggests that the N-terminal portion of the channel contributes to its translocation to membrane expression upon PIP2 binding, as was reported previously 20. The other 24 constructs tested carried OAG-induced currents, and 8 out of them exhibited a significantly faster decay upon PIP2 depletion than TRPC6WT (Fig. 2, upper, blue bars). Among those, the R437Q and R746Q mutants were the fastest (92 ± 10 ms (n = 6) and 118 ± 13 ms (n = 9), respectively).
On the other hand, the recovery of TRPC6 current from DrVSP-mediated inhibition, with restoration of PIP2 was significantly accelerated or delayed in 9 out of 26 constructs (Fig. 2, middle, green bars). The current recovery was markedly delayed in the R437Q and R865Q mutants. These mutations are respectively located in the pre-S1 domain and calmodulin/inositol-1,4,5-trisphosphate receptor binding domain (CIRB)31, the latter is also known as PIP2/Calmodulin binding domain21,32. Only the single mutation R758Q and double mutation K781Q/K782Q reduced the effect of PIP2 depletion, as indicated by the weak inhibition ratio (Ipost/Ipre), (Fig. 2, bottom, yellow bars). However, none of the mutations led to more than 50% current inhibition upon PIP2 depletion. This suggests the contribution of PIP2 binding to channel gating affects 50% of the total channel activity or the maximal efficacy of DrVSP to deplete endogenous PIP2 may not be 100%, possibly due to fast replenishment of PIP2 in some compartmentalized area.
Pre-S1 domain is a PIP2 binding domain
In contrast to the aforementioned constructs, the R629Q (in S4-S5 linker) and K748Q (within TRP box) mutants exhibited accelerated recovery from the inhibition, suggesting that these basic residues may be less contribution to rebinding of PIP2. Thus, among in the first screened mutants, the pre-S1 domain mutation R437Q exerted the most critical effect on both the decay and the recovery of TRPC6 channel currents (Fig. 2). This finding was further confirmed using the depolarizing step-pulse protocols over a wide range of membrane potentials (Fig. 3A-F). The results revealed that at every membrane potential tested, the R437Q mutation decays faster and recovers more slowly than TRPC6WT.
Because the pre-S1 domain is enriched in basic residues (Fig. 3G), this prompted us to conduct a second mutational screening focused on the pre-S1 domain, which revealed the importance of K442, in addition to K431 and K434. The time constants for the decay and recovery of K442Q were 54 ± 11 ms and 3.71 ± 0.53 s (n = 5), respectively (Fig. 3H). Amino acid sequence alignment of the pre-S1 domains of mammalian TRPC channels shows identical positively charged residues at sites equivalent to R437 and K442 (Fig. 3G). This indicates that R437 and K442 in the pre-S1 domain play an essential role in PIP2 binding to TRPC6 channel. Moreover, the contributions of basic residues within the TRP box were also unmistakable in that kinetic analysis, which showed R746Q to produce the fastest current decay among the tested mutations (Fig. 2). To understand whether the pre-S1 shoulder and the TRP box exert a cooperative effect on PIP2 binding, we examined the effect of R437Q/R746Q double mutations. The decay and recovery kinetics of this double mutant were indistinguishable to R437Q mutant (Fig. 4C, D), indicating the contributions of the R437 and R746 residues to PIP2 binding are cooperative rather than additive effect
To summarize the mutational screening results, for simplicity we estimated relative PIP2 dissociation constants from kinetics of the TRPC6 current decay and recovery. Based on this estimation, the R437Q and K442Q channels exhibited nearly five to eightfold less affinity for PIP2 than the TRPC6WT channel (Fig. 3I). Given that the dissociation constant for PIP2 binding to TRPC6WT was previously reported to be 2 µM15, we suggest that those for PIP2 binding to the pre-S1 domain mutants (R437Q and K442Q) are greater than 10 µM.
PIP2 binding to the pre-S1 domain, linker domain, and TRP box
Recent cryo-EM structures of TRPC3/6 channels showed that the pre-S1 domain is exposed to the outside of the channel complex33,34. Within the pre-S1 domain, positively charged residues are situated every three or five amino acids (K431, K434, R437, K442) to form an amphipathic helix termed the “pre-S1 shoulder” (Fig. 4A). Intriguingly, the pre-S1 shoulder is positioned at the inner surface of the membrane, where it encounters the residues H358 and R360. Neutralizing mutations of those residues slows the recovery, implying weaker PIP2 affinities for TRPC6 channel (Fig. 2). This suggests the basic residues in the pre-S1 shoulder as well as the H358/R360 residues are crucial for recognition of PIP2.
To confirm this idea, we performed a docking simulation with PIP2 and the pre-S1 shoulder using the cryo-EM structure of TRPC6. Within the simulation, the inositol head of PIP2 was positioned at a surface pocket surrounded by the pre-S1 shoulder and H358/R360 residues. The TRPC6WT model exhibited the highest docking energy (-6.0 Kcal/mol), while the R437Q and K442Q pre-S1 shoulder mutants exhibited somewhat lower docking energies (-5.4 and − 5.6 Kcal/mol, respectively). We also tested that point mutation at cysteine residue (C429) because it locates pre-S1 domain and near the center of PIP2-docking area. Although the docking energy of the C429S mutant was no lower than that of the wild-type channel (-6.0 Kcal/mol), the actual decay was significantly accelerated (121 ± 17 ms, Fig. 4D). These observations further confirm that the pre-S1 shoulder and H358/R360 residues contribute critically to PIP2 binding.
Functional role of the pre-S1 domain/shoulder
To evaluate the function of the PIP2 binding pocket identified in the docking simulation, TRPC6 channels were co-expressed with muscarinic receptors, which were then stimulated with a muscarinic receptor agonist carbachol (CCh). We found that the maximum current density upon receptor stimulation was significantly suppressed by single or double mutations, H358Y/R360Q, R339Q, K434Q, R437Q, and K442Q (Fig. 5A, B). Moreover, the extents of suppression elicited by these mutations as well as those by different combinations of double or triple mutations K434Q/R437Q, R437Q/K442Q, and K431Q/K434Q/R437Q were nearly identical. This confirms that the surface pocket for the PIP2 interaction is crucial for receptor-activated channel activity. To then assess the importance of the pre-S1 shoulder for channel localization in the cell membrane, we used confocal microscopy to compare the localizations of TRPC6WT and the R437Q and R437Q/K442Q mutants fused with cyan fluorescence protein (CFP) based on their co-localization with the yellow fluorescence protein (YFP)-fused PH-domain sensor. Co-localization of the channel and PH-domain after transfection into HEK293 cells did not statistically differ between the wild-type and mutant TRPC6 channels (Pearson’s correlation coefficient = 0.68 ± 0.02 for WT, 0.61± 0.07 for R437QK442Q, Fig. 5D). This suggests the pre-S1 domain/shoulder mainly contributes to channel functionality, but not to PIP2-dependent membrane localization.
PIP2 binding to TRPC channels allows the channels availability, and the less binding is possible to involve channel dysfunctionality vice versa. It has been shown that impairment of Ca2+-dependent inactivation of TRPC6 channels is a cause of FSGS32. We therefore measured the receptor-activated TRPC6 current inactivation of the pre-S1 mutants under low Ca2+ buffering conditions. As shown in Fig. 5C, R437Q, R437/K442Q mutants and TRPC6WT exhibited almost identical residual currents. Thus, binding of PIP2 to the pre-S1 domain/shoulder is not directly involved in Ca2+-dependent inactivation and may not account for the pathogenesis of FSGS.
In addition, here and previously, we demonstrated that reducing PIP2 inhibits channel activity, even in the presence of DAG8. We therefore tested how PIP2 affects the potency of DAG to activate TRPC6 channel by comparing the concentration-dependent effects of OAG on TRPC6WT and pre-S1 mutants. Figure 6A shows the concentration dependence of the responses of R437Q, R437Q/K442Q mutant and TRPC6WT channels to OAG. This dependency was done by ratiometric Fura-2 photometry35. The normalized dose-response data were best fit to the Hill equations with EC50 values of 48, 37 and 46 µm and the coefficients (n) of 1.4, 1.5 and 0.8 for R437Q, R437Q/K442Q, and TRPC6WT, respectively. Thus, mutations in the pre-S1 domain/shoulder cause practically no change in EC50 values. This suggests that PIP2 binding does not significantly affect DAG binding but instead may affect in allosteric activation by DAG.
Lastly, we used our previously described simulation model, to evaluate how PIP2 binding affects receptor-activated TRPC6 channels15 (Fig. 6B). By reducing the channel’s binding affinity for PIP2 while keeping its affinity to DAG, the amplitude of a simulated current was gradually decreased (Fig. 6C). When PIP2-channel affinity was reduced by one-fifth, the current amplitude was reduced by half (Fig. 6C, black vs blue traces). This result is consistent with those obtained with the R437Q and K442Q mutants, which exhibited reduced PIP2 affinities and current densities (Fig. 3I and 5A, B). Taken together, these findings suggest that reducing PIP2 binding affinity decreased TRPC6 channel activity without altering its cellular localization. Our study therefore uncover that the pre-S1 domain/shoulder of TRPC6 makes the critical contribution to PIP2 binding.
Mutations within the proximal TRP box domain alter the effect of PIP2 depletion
As shown in Fig. 2, upon the PIP2 depletion, the R758Q and K781Q/K782Q mutations significantly decrease the inhibition ratio (Ipost/Ipre) as compared to TRPC6WT. These residues are located within the proximal TRP box, the importance to the thermal sensitivity of which was shown in earlier studies of PIP2-mediated regulation of TRPV1 channels36,37 but lacking such importance for TRPC channels. We therefore questioned whether the other basic residues located between R758 and K781/K782 might affect the effects of PIP2 depletion. Unexpectedly, when K771, an evolutionary highly conserved residue, was mutated to glutamine, the resultant OAG-induced currents were potentiated after activation of DrVSP (Ipost/Ipre = 1.24 ± 0.13 n = 7, Fig. 7A-C). This potentiation was observed repeatedly during the recording period (Fig. 7B). By contrast, no potentiation of K771Q-mediated currents was seen when the inactive DrVSP was co-expressed (Fig. 7D). This eliminates the possibility of a gain-of-function effect on the voltage-dependent activation.
Moreover, using the rapamycin-inducible FKBP12-Inp54p system38,39, which also depletes PIP2 by the membrane recruiting of specific inositol 5-phosphatase, the potentiation of TRPC6 currents still occurred (Fig. 7E, 7G). To explore this reversed effect of PIP2 depletion can be generalized to other TRPC channels, we tested the effect of neutralization on the equivalent residue (K716) in TRPC7. As shown in Fig. 7F, the TRPC7 K716Q mutant exhibited potentiated the channel activity upon PIP2 depletion. The potentiation ratio for TRPC7K716Q was 1.28 ± 0.06, which is nearly identical to that of TRPC6K771Q. These results may imply that certain basic residues in the proximal C-terminal region (TRP box) act to control the polarity of PIP2-dependent effects on TRPC6/7.