Direct detection of the chloride release and uptake reactions of Natronomonas pharaonis halorhodopsin

Membrane transport proteins undergo multistep conformational changes to fulfill the transport of substrates across biological membranes. Substrate release and uptake are the most important events of these multistep reactions that accompany significant conformational changes. Thus, their relevant structural intermediates should be identified to better understand the molecular mechanism. However, their identifications have not been achieved for most transporters due to the difficulty of detecting the intermediates. Herein, we report the success of these identifications for a light-driven chloride transporter halorhodopsin (HR). We compared the time course of two flash-induced signals during a single transport cycle. One is a potential change of Cl−-selective membrane, which enabled us to detect tiny Cl−-concentration changes due to the Cl− release and the subsequent Cl−-uptake reactions by HR. The other is the absorbance change of HR reflecting the sequential formations and decays of structural intermediates. Their comparison revealed not only the intermediates associated with the key reactions but also the presence of two additional Cl−-binding sites on the Cl−-transport pathways. The subsequent mutation studies identified one of the sites locating the protein surface on the releasing side. Thus, this determination also clarified the Cl−-transport pathway from the initial binding site until the release to the medium.


1-1. Detection of the H + -transfer reactions by NpHR
As shown in Fig. 2B, the potential of the Cl − -selective membrane has a positive slope against pH.Thus, the negative potential changes in upper panels of Fig. 4 might reflect the decrease in pH, that is, the photolyzed NpHR might exhibit H + release first followed by the uptake.We tested this hypothesis using the ITO electrode, which is a transparent and pH-sensitive electrode with a fast response (24).Herein, we used an electrochemical cell in Fig. S1B.The lipid-reconstituted NpHR was deposited on the working electrode and then activated by the laser pulse.Time-dependent voltage changes are plotted in Fig. S4A.
In this experiment, the positive signal indicates the pH decrease around the working electrode (24), that is, the photolyzed NpHR surely exhibits H + release first, followed by uptake.The detected signals are small but appear around 2 ms after the flash, which is almost at the same time as the negative potential change of the Cl − -selective membrane (Fig. 4).
The H + release from the photolyzed NpHR was previously suggested by an FTIR study, where the H + source was assigned to Glu234 residue around the EC surface (Fig. 1A) (35).Thus, we also performed ITO experiments for the E234Q NpHR mutant.The results are shown in Fig. S4B, where the H + -release signals still appear.Thus, Glu234 is not the source of H + .

1-2. Removal of the contribution of "nonpumping photocycle" from the measured absorbance changes
Figure S5A shows the three-wavelength data of the nonpumping photocycle, which are measured at 0 mM Cl − .The initial absorbance change at 500 nm is almost constant until 0.1 ms.This value gradually increases with the increase in the Cl − concentration, reflecting the increase of Cl − -bound NpHR ("500 nm" traces in Fig. 4).Thus, using the signals at 500 nm, we estimated the fractions of Cl − -bound NpHR.In Fig. S5B, the net increases in signal amplitudes (ΔΔA500) are plotted against Cl − concentration with the filled circles.
The solid line is the best-fit curve using the following equation: where Amax and [Cl − ] stand for the maximum amplitude of the plot and Cl − concentration, respectively, and fCl represents the fraction of the Cl − -bound NpHR.The determined Kd is 7.4 mM, whose value roughly matches the Kd value determined from the Cl − -induced spectral shift of the unphotolyzed state.This spectral shift is shown in Fig. S5C.Upon binding of Cl − , the λ max shifts to a shorter wavelength and finally reaches 580 nm.The absorbance increase at 580 nm (ΔA580) was calculated and then plotted in Fig. S5B S4 with the open circles.The best-fit result (broken line) was obtained with a Kd of 3.6 mM.Thus, both plots surely reflect the Cl − -dependent increase of f Cl .Herein, we used the K d from the ΔΔA 500 to calculate the fraction of Cl − -free NpHR (ffree) as follows: Next, we removed the contribution of the nonpumping photocycle from the measured absorbance changes.Here, we simply calculated the following subtractions for all wavelength data: where ΔACl(t, λ) and ΔAfree(t, λ) are the measured flash-induced absorbance changes in the presence and where k is the proportionality coefficient.According to Eq. 2, the corresponding potential change, ΔΨCl(t), is expressed as follows: where Eq.S4 is used.The potential change of ΔΨCl(t) corresponds to f4⋅P4(t).Therefore, the deformation of Eq.S5 takes the following form: Thus, f4 is described as follows: where k′ is also the proportionality coefficient.Consequently, f 4 is inversely proportional to the Cl − concentration.This relationship actually appears in Fig. S9.The value of 54.5 included in Eq.S7 comes from the slope of the plot in Fig. 2A (54.5 mV/decade).However, this slope was smaller (19.1 mV/decade) below 3 mM Cl − .Thus, f4 at 1 mM Cl − cannot be directly compared to f4 values at higher Cl − concentrations.
In Fig. S9, f 4 at 1 mM Cl − is plotted but is not involved in the fitting analysis by Eq.S7.

2-2. Discrepancy in the dark state structure
Regarding Site IV, we failed to identify its position through the mutation experiments.Two mutants of Arg22 and Arg176 could not be prepared due to their negligible expression in E. coli.Thus, either Arg sidechain might form Site IV.In contrast, the presence of Site IV is not consistent with the crystal structure of the dark state because, on its EC side, Cl − is only observed at Site I (26).As shown in Fig. 4, the Cl − -S6 transfer timings are maintained even at a low Cl − concentration of 1 mM, indicating that both Sites I and IV bind Cl − at 1-mM Cl − .Thus, the Cl − -binding affinity of Site IV is at least comparable with that of Site I. Why is Cl − binding at Site IV not observed in the crystal structure despite the binding at Site I being observed?One explanation might be Cl − binding after photoexcitation.If this uptake by Site IV occurs at a very fast rate, the phase of potential decrease cannot be observed.This fast Cl − uptake seems to correspond to the negative potential change observed at earlier time ranges (Fig. 4, upper panels).However, as mentioned above, the early potential change does not disappear even at 100-mM Cl − .Thus, at present, we could not assign this potential change to the Cl − -transfer reaction.Thus, the location of Site IV should be further investigated in future studies.S7

Figure S2 .S9Figure S3 .Figure S4 .Figure S5 .Figure S6 .Figure S7 .Figure S8 .Figure S9 .Figure S10 .
Figure S2.Constant light-induced potential changes of the Cl − -selective membrane.We deposited the lipid-reconstituted NpHR on the membrane surface facing the sample chamber.The horizontal bar indicates the duration of illumination.The data in panels C and G are the same as those in Fig. 3A and B, respectively.

Table S1 .
Comparison among HRs for residues corresponding to K203 and K215 in NpHR