Channel gating in bacteriorhodopsin-like channelrhodopsins with contrasting K⁺ and Na⁺ selectivity

Abstract

Two closely homologous channelrhodopsins from Hyphochytrium catenoides, known as HcKCR1 (kalium channelrhodopsin 1) and HcCCR (cation channelrhodopsin), exhibit >100 times different K⁺/Na⁺ relative permeabilities and are emerging optogenetic tools for controlling neurons and cardiomyocytes. Key residue motifs and trimeric organization relate them structurally to bacteriorhodopsin-like cation channelrhodopsins (BCCRs). Here, we demonstrate that they utilize a specific gating mechanism previously suggested for the earlier-discovered BCCR from cryptophytes. Using a comparative analysis of transient absorption changes and photocurrents under single-turnover conditions, we identified an early, far-UV-absorbing photocycle intermediate that precedes channel opening in wild-type HcKCR1 and HcCCR. The UV-absorbing product is a spectrally distinct M1 state, characterized by a deprotonated Schiff base that facilitates cation passage through the channel. It is subsequently converted into an M2 state that absorbs at ~400 nm. These photocycle steps appear to be a common feature of the entire BCCR family, as they were also found in Guillardia theta CCR2 (GtCCR2) and Rhodomonas abbreviata CCR1 (RaCCR1). Channel gating involves the transfer of a proton from Asp116, the conserved residue in the position of bacteriorhodopsin’s proton donor to the Schiff base, to an unidentified residue. Here, we show that this residue is located on the cytoplasmic side of the molecule. The time course of this deprotonation correlated with the opening of the channel. The D116N mutation completely abolished HcCCR channel activity and converted HcKCR1 into a weak Na⁺ channel. Furthermore, HcCCR did not exhibit outwardly directed active proton transfer, as observed during the M2 state in HcKCR1. Finally, M2 rise coincided with the slow phase of M1 formation and channel opening in Na⁺-selective variants, but was delayed in HcKCR1 and its mutants. Our results contribute to a deeper mechanistic understanding of light-gated cation conductance in BCCRs, facilitating the further development of optogenetic tools.

Introduction

Channelrhodopsins (ChRs) are retinylidene proteins that, upon photoexcitation, generate passive ionic currents across the membrane (1). In an approach known as optogenetics (2), ChRs are used as molecular tools to control cellular excitability with light (3). They are proposed as gene therapy for vision restoration and curing neurological, psychiatric, and cardiac disorders (4–6). Upon photoexcitation, cation channelrhodopsins (CCRs) generate H⁺ and Na⁺ influx, depolarize the membrane, and activate neurons and cardiomyocytes (7). Anion channelrhodopsins (ACRs) (8) and kalium channelrhodopsins (KCRs) (9) generate, respectively, Cl^- influx or K⁺ efflux, which hyperpolarizes the membrane and inhibits the generation of action potentials.

The genome of the stramenopile Hyphochytrium catenoides encodes three closely related ChRs we named HcKCR1, HcKCR2, and HcCCR (9,10). HcKCR1 and HcCCR have 74/87% sequence identity/similarity in their seven-transmembrane (7TM) domains at the protein level, but differ >100 times in their K⁺/Na⁺ relative permeability. Atomic-resolution structures of all three H. catenoides ChRs have been obtained by cryo-electron microscopy (Fig. S1A,B) (11–13). These proteins and their homologs from other stramenopiles, cryptophytes, and alveolates (14–16) form a structurally and functionally distinct class of ChRs, known as “bacteriorhodopsin-like cation channelrhodopsins” (BCCRs) (17). BCCRs share the highly conserved DTD residue motif (corresponding to Asp85, Thr89, and Asp96 in Halobacterium salinarum bacteriorhodopsin, HsBR) and trimeric assembly with haloarchaeal proton-pumping rhodopsins (17,18). HcKCR1 has been used for photoinhibition of mouse neurons in brain slices and in vivo (9,12,19,20), and for photocontrol of behavior in zebrafish, worms, and flies (21). HcCCR (also known as HcNCR1) enabled photocontrol of aquaporin-mediated water flux in Xenopus oocytes (22). Low relative permeability for protons compared to Na⁺ and weak desensitization under continuous illumination are advantages of HcCCR over better-known excitatory optogenetic tools, such as Chlamydomonas reinhardtii channelrhodopsin 2 (CrChR2) (23) and Rhodomonas lens CCR1 (ChRmine) (10,24).

In all ChRs, the retinal chromophore is attached via a protonated retinylidene Schiff base (RSB) linkage to a conserved Lys residue in the seventh transmembrane helix (TM7). The RSB is located in the middle of the putative channel pore. A key question in ChR research is how photon energy creates an ion conduction pathway within the protein molecule. Photoexcitation induces all-trans to 13-cis isomerization of the chromophore, followed by thermal relaxation to the initial unphotolyzed state via a series of spectrally and/or kinetically distinguishable photocycle intermediates. Deprotonation and subsequent reprotonation of the RSB are manifest as the formation and decay of a blue-shifted M intermediate. In canonical chlorophyte CCRs such as CrChR2, the RSB deprotonation (M formation) precedes channel opening (25), whereas in ACRs, it occurs ~50 times slower than channel opening (26). In HcKCR1, we reported a biphasic rise of absorption at 395 nm, the main component of which was ~13 times slower than the channel opening (9). The presence of the positively charged RSB in the open K⁺ channel was a puzzle that needed to be resolved. The photochemical conversions and single-turnover photocurrents of HcCCR have not yet been investigated.

In the photocycles of rapidly desensitizing cryptophyte BCCRs, we identified a strongly blue-shifted, UV-absorbing intermediate that exhibited fine structure with peaks at 318, 330, and 346 nm, typical of deprotonated RSB (27). In this study, we report similar products in the photocycles of HcKCR1 and HcCCR, their mutants, and two wild-type cryptophyte BCCRs. We show that accumulation of the UV-absorbing photocycle intermediate precedes channel opening in these proteins. We interpret this intermediate as a spectrally shifted M1 state with a deprotonated RSB, likely a common characteristic of all BCCRs. Comparative analysis of transient absorption changes and laser flash-evoked photocurrents, as well as mutational analysis, revealed both a common channel gating mechanism and a difference between K⁺- and Na⁺-selective channels.

Results

Fast, far blue-shifted UV intermediates

To monitor the formation of the UV-absorbing products, we recorded transient absorption changes in HcKCR1, its mutants, and HcCCR in the UV region of the spectrum (Fig. 1A-D). A fast increase in absorption at ~340 nm was observed before an increase at ~400 nm, which was previously interpreted as the formation of an M intermediate (9). In HcKCR1, the rise of absorption at 340 nm exhibited at least three components with the time constants (τ) 3 ± 0.17 μs, 155 ± 23 μs, 2.7 ± 0.13 ms; n = 4 cells (Fig. 1A). The fastest μs component of the 340-nm absorption rise appeared concomitantly with formation of the L intermediate monitored at 430 nm (Fig. 2A). It could be interpreted as a contribution of L absorption in the UV range, also detected in HsBR (28). Alternatively, these components may reflect a rapid conversion of the L intermediate to a strongly blue-shifted M1, in contrast to HsBR, where the M1 and M2 spectra differ only slightly. They could only be distinguished kinetically (29), energetically (30), and structurally (31), but not spectrally.

Figure 1.

The kinetics of transient absorption changes corresponding to the M1 state (black line), M2 state (blue line), and photocurrents recorded under single-turnover conditions (red lines) in HcKCR1 (A), HcKCR1_Y106A (B), HcKCR1_C110A (C), HcCCR (D), GtCCR2 (E), and RaCCR1 (F). The holding voltage at the amplifier output for the photocurrent recordings was 0 mV in A-C and −60 mV in D-F. The photocurrent sign in D-F was inverted for easier comparison with the absorption changes.

Figure 2.

The kinetics of transient absorption changes corresponding to the L (blue) and M1 (black) intermediates in HcKCR1 (A) and HcCCR (B).

In wild-type HcKCR1, photocurrents under single-turnover conditions developed well before the formation of M2, monitored at 395 nm but after the fast rise of M1 (Fig. 1A). Such temporal correlation between M1, M2, and photocurrent rise was also observed in the mutants with altered channel kinetics. In the fast HcKCR1_Y106A mutant, photocurrents developed in a time window when absorption at 400 nm practically did not change (Fig. 1B). Moreover, channel closing occurred in parallel with the M2 rise. This means that the deprotonation of the RSB occurred well before the generation of M2, leading to the formation of a short-wavelength-absorbing UV product (M1) and allowing channel opening. The slow component of the UV signal can be explained by the reversibility of the M1 ⬄ M2 transition, in contrast to HsBR, in which the M1 => M2 transition at room temperature is practically irreversible and plays a key role in the vectorial proton transport (32). A similar temporal correlation between M1, M2, and photocurrent was observed in the slow HcKCR1_C110A mutant (33), although all these processes were slower than in the WT (Fig. 1C).

A UV-absorbing intermediate was also detected in the HcCCR photocycle (Fig. 1D and 2B). However, the relationship between M1, M2, and photocurrent differed from that in HcKCR1. The relative amplitude of the fast component of the UV absorption rise was almost twice as small as in HcKCR1 and its mutants. The delay between the absorption rise at 400 nm and that at 330 nm was also smaller. Finally, all four components of the UV absorption rise (8 ± 0.4 μs, 300 ± 150 μs, 1.1 ± 0.24 ms, and 8.7 ± 4.4 ms; mean ± SEM, n = 8 measurements in independent preparations of each protein) were slower than in HcKCR1. The biphasic rise and fast decay of M2 were also slower in HcCCR (0.47 ± 0.05, 7.97 ± 1.49, and 51.6 ± 4.6 ms) than in HcKCR1 (0.26 ± 0.04, 6.43 ± 1.57, and 27.0 ± 2.8 ms). Nevertheless, the photocurrent kinetics correlated with the changes in UV absorbance rather than those at 400 nm. A stronger reversibility of the M1 ⇔ M2 transition can explain a slight delay between M2 and M1 formation.

To test whether the relationship between M1, M2, and photocurrent also holds in other Na⁺- selective BCCRs from cryptophytes, we analyzed Guillardia theta CCR2 (GtCCR2), for which a high-resolution molecular structure has been obtained (Fig. S1C; (34)), and Rhodomonas abbreviata CCR1 (RaCCR1). In GtCCR2, this relationship was similar to that in Na⁺-selective HcCCR (Fig. 1E). In the RaCCR1 photocycle, a UV-absorbing product (the M1 intermediate) was also detected. Absorption at 340 nm (M1) rose in parallel with absorption at 400 nm (M2) up to 1 ms. It continued to rise after the decay of M2 (Fig. 1F). This unusual behavior correlated with the accumulation of a long-lived, structured UV-absorbing product (27). This can be explained by a substantial shift of the M1 ⬄ M2 equilibrium toward M1 during the photocycle, or by the transformation of initial M1 into a spectrally identical but non-conductive long-lived product after ~1 ms, leading to photocurrent decay.

Slow photocycle components

Complete sets of absorption changes and results of their global fit analysis are shown in Fig. S2A and C. Global fit analysis of flash-induced transient absorption changes in the visible range revealed the M2, N, R (recovered), and O intermediates, typical of retinal proteins (Fig. S2B and D). The slow photocycles of HcKCR1 and HcCCR were basically similar. However, the slowest component of the unphotolyzed state recovery was faster in HcKCR1 than in HcCCR (3.7 and 12 s, respectively). We also found a difference in the reversibility of the spectral transitions between the two H. catenoides ChRs, which was manifested by different relative amplitudes of N, R, and O intermediates in the main closing component spectra. A more detailed analysis of the slow photocycle components was beyond the scope of this study.

The long-lived UV-absorbing products

The maximal absorption of detergent-purified HcKCR1 and HcCCR was 15 and 9 nm shorter than the respective maxima of the photocurrent action spectra (Fig. 3A, B). This shift can be explained by the presence of non-electrogenic short-wavelength-absorbing fractions, probably cis-retinal-bound forms, ~20% of which have been detected in HcKCR1 by HPLC analysis (12,13). To estimate the absorption maxima of two fractions and their relative concentrations, we deconvolved the absorption spectra into two Gaussian components (Fig. 3C, D). The results showed that the peak absorption wavelength (λ_max) of the long-wavelength (trans) form equaled that of the corresponding action spectrum. The contribution of the inactive cis form was larger in HcKCR1, which explained the larger difference between the λ_max of the absorption and action spectra in this protein (15 nm) than in HcCCR (8 nm) (Fig. 3A, B).

Figure 3.

Stationary absorption spectroscopy of purified HcKCR1 and HcCCR. (A, B) The absorption spectra (solid lines) and photocurrent action spectra (symbols and dashed lines) of HcKCR1 (A) and HcCCR (B). (C, D) The absorption spectra recorded before and after photoexcitation in HcKCR1 (C) and HcCCR (D) with a series of laser flashes. (E, F) Deconvolution of the absorption spectra into Gaussian components. (G, H) The difference light minus dark spectra in HcKCR1 (G) and HcCCR (H).

Our flash photolysis setup does not allow the determination of the M1-state spectral properties. Therefore, we employed an alternative approach and measured the accumulation of a practically irreversible state formed upon excitation with multiple laser flashes (530 nm, 6 ns, 5 mJ). Such excitation of purified HcKCR1 and HcCCR reduced absorption in the main visible band at the rate of ~0.04–0.06% per flash (Fig. 3E, F). The bleaching was slowly reversible in the darkness; only <20% of the bleached protein recovered after 24 hours. Remarkably, the difference spectra showed the maximum bleaching at 537 nm in HcKCR1 and 530 nm in HcCCR, exactly matching the respective maxima of the photocurrent action spectra. We conclude that only long-wavelength forms are functionally active in purified protein samples and HEK cells. The λ_max of the remaining protein fraction was 4–5 nm shorter than that of the initial sample, as expected of cis-retinal-bound forms. This suggests that only the red-shifted, all-trans-retinal-bound electrogenic forms were photoactive.

In both HcKCR1 and HcCCR, the long-lived bleaching products showed a main structured absorption band at 337 nm, as revealed by the difference spectra (Fig. 3G, H). One possibility is that it reflects a small fraction of a slightly modified M1 state that has been trapped from further conversion. Earlier, we identified similar but reversible UV-absorbing products in RaCCR1 and Rhodomonas salina CCR1 (RsCCR1) (27). The second band at ~400 nm, corresponding to the M2 intermediate, was much weaker. The ratio of the absorption changes of the M2 intermediate (monitored at ~400 nm) to that of M1 was ~0.36 in HcKCR1 and 0.77 in HcCCR. HcKCR1 photocurrent recordings showed much stronger desensitization under continuous illumination than HcCCR, which correlated with a smaller contribution of M2 to the photostationary mixture. Complete desensitization was observed in RaCCR1, where practically no M2 was detected (Fig. S3). We conclude that the shift of reversibility from M2 to the nonconductive N state is the primary cause of desensitization in BCCRs.

Laser flash-induced photocurrents in wild-type HcKCR1 and HcCCR

Earlier, we reported an analysis of photocurrents in HcKCR1 under single-turnover conditions (9). Here, we compare its laser-flash-induced photocurrents with those from the highly homologous but Na⁺-selective HcCCR (Fig. 4A, B). Under standard ionic conditions (130 mM KCl in the pipette, 130 mM NaCl in the bath), channel opening was biphasic in both proteins. The fast opening occurred with a similar τ, but the slow opening was faster in HcCCR than in HcKCR1. In HcKCR1, the contribution of the slow phase to the total opening was almost equal to that of the fast component, whereas in HcKCR, it did not exceed 10%. Channel closing in HcCCR was biphasic, unlike the monoexponential closing in HcKCR1. Photocurrent in HcCCR but not in HcKCR1 showed a slight inward rectification. The reversal potential (V_rev) of all exponential kinetic components was around −95 mV in HcKCR1 and 45 mV in HcCCR (Fig. 4C and D), indicating that ionic selectivity did not change during the single-turnover photocycle.

Figure 4.

Photocurrents generated by HcKCR1 and HcCCR under single-turnover conditions. (A and B) Photocurrent traces recorded from HcKCR1 (A) and HcCCR (B) upon incremental voltages applied in 20-mV steps from −100 to 20 mV using standard solutions (130 mM K⁺ in the pipette, 130 mM Na⁺ in the bath). The solid lines represent experimental data, while the dashed lines represent multiexponential computer approximations. (C and D) The voltage dependencies of the photocurrent kinetic components in HcKCR1 (C) and HcCCR (D).

Photocurrent recording under single-turnover conditions enables analyzing intramolecular charge displacements, such as active proton transfers, in addition to channel activity (35). We probed for intramolecular proton transfers in HcKCR1 and HcCCR by recording photocurrents upon substituting non-permeant N-methyl-D-glucamine (NMDG⁺) for K⁺ and Na⁺ in the bath and the pipette. Relatively large multiexponential signals were observed in HcKCR1 in the absence of monovalent metal cations (Fig. 5A). The rates of both components of the current rise were slower at low pH (Fig. 5B), indicating that they reflected outwardly directed active proton transfer that predominantly occurred during the M2 state. The photocurrents detected under these conditions in HcCCR were negligible at 0 mV holding potential (Fig. 5C). Moreover, the V_r values for these currents differed > 170 mV (-180 mV for HcKCR1 and −6 mV for HcCCR) (Fig. 5D). This indicates that practically no active proton movement occurs in HcCCR in contrast to HcKCR1. The rise of M1 and M2 intermediates in HcCCR occurred at similar rates. Therefore, the absence of active photocurrent in HcCCR is most likely due to the high reversibility between deprotonation and back-reprotonation of the Schiff base.

Figure 5.

(A and C) Photocurrent traces recorded upon incremental voltages in the absence of permeant monovalent metal cations, pH 7.4 from HcKCR1 (A) and HcCCR (C). (B) Photocurrent traces recorded at 0 mV from HcKCR1 at pH 5.4 and 9.4. (D) The voltage dependence of the peak current in HcKCR1 (squares) and HcCCR (circles).

Photocurrents in the donor and acceptor mutants of HcKCR1 and HcCCR

The first BCCR studied in detail (GtCCR2) exhibited an unusual gating mechanism (17). The open state of the channel requires the RSB donor residue (Asp98) to be deprotonated. Reprotonation of the RSB during the photocycle takes place from an earlier protonated acceptor (Asp87), whereas a proton from the donor moves to an unknown residue. We analyzed photocurrents of the corresponding HcCCR and HcKCR1 mutants to test whether this gating mechanism also operates in these proteins.

As in GtCCR2, neutralization of the residue in the proton donor position in the HcCCR_D116N mutant eliminated channel current. To probe for intramolecular charge transfers, we tested this mutant using solutions free of monovalent metal cations. A clear fast current with a rise time of ~13 μs and decay of 80 μs was observed (Fig. 6A). The V_r value for this current was −150 mV (Fig. 6B). Fast kinetics and a very low V_r indicate that it reflects proton movement from the RSB to the acceptor. We think that fast proton transfer from the RSB to the acceptor also occurred in the wild type. However, it overlapped with slower proton transfer in the opposite direction, which could be resolved in the D105N mutant (Fig. 8). The traces recorded in the wild type result from the superposition of these two transfers.

Figure 6.

Photocurrents and their voltage dependencies of the HcCCR_D116N donor mutant in the absence of monovalent metal cations.

Figure 8.

Photocurrents and their voltage dependencies of the acceptor mutants HcCCR_D105N (A, B) and HcKCR1_D105N (C, D) in the absence of monovalent metal cations.

The D116N mutation also dramatically suppressed photocurrent in HcKCR1, but the properties of the bidirectional residual current (Fig. 7A) were different. The strong voltage dependence of its second peak showed inward rectification with V_r ~34 mV (Fig. 7B), similar to that of the HcCCR current. Replacement of metal cations with NMDG⁺ abolished this second peak (Fig. 7C), consistent with its interpretation as channel activity. These observations confirmed that the proton donor residue plays a key role in K⁺ selectivity of HcKCR1, as deduced from experiments with continuous light pulses (12,15,16). The fast positive component with a decay τ of 280 μs and V_r −170 mV (Fig. 7D) was similar to that in HcCCR_D116N. It persisted when monovalent metal cations were replaced with NMDG⁺ (Fig. 7B) and likely reflected proton transfer from the RSB to an outward acceptor.

Figure 7.

Photocurrents and their voltage dependencies of the HcKCR1_D116N donor mutant in standard solutions (A, B) and in the absence of monovalent metal cations (C, D).

Mutation of the proton acceptor (D105N) eliminated channel current in HcCCR. The remaining fast currents in the monovalent cation-free solutions exhibited the fast negative peak, likely reflecting unresolved charge movement due to retinal isomerization reported in other ChRs (35). A second peak followed it, with a rise time τ of 0.22 ms and a decay time τ of 2.1 ms (Fig. 8A). A large positive V_r (90 mV) for this current indicated that it reflects inwardly directed charge movement (Fig. 8B). Photocurrents of the HcKCR1_D105N mutant in the monovalent cation-free solutions (Fig. 8C) were very similar to those of HcCCR_D105N, with a fast negative isomerization peak and a later negative wave, which was more rapid than in HcCCR_D105N, and a large positive V_r (Fig. 8D). This wave cannot be attributed to RSB deprotonation, as M formation was an order of magnitude slower in HcKCR1_D105N than in the wild type (12). The rise time of the second wave preceded channel opening, for which Asp116 must be deprotonated. Therefore, we interpret this current component in HcKCR1_D105N and HcCCR_D105N resulting from proton transfer from Asp116 to an unknown inwardly located acceptor, which we hypothesized earlier in GtCCR2 (17).

Discussion

The light-gating mechanisms of ChRs are now better understood, although not entirely at the molecular level, despite their widespread use as optogenetic tools. The convergent evolution of ion conductance in different ChR families limits the interpretation of the mechanisms' details. We analyzed transient absorption changes and photocurrents in response to nanosecond laser flashes in HcKCR1 and HcCCR, two ChRs from the hyphochytriomycete H. catenoides with contrasting cation selectivities. Structurally, these proteins are members of the BCCR family, characterized by conservation of the Schiff base proton acceptor and proton donor residues (corresponding, respectively, to Asp85 and Asp96 in HsBR) and trimeric tertiary structure (36–38) (11–13). Here, we demonstrated that they are BCCRs, also functioning similarly, utilizing the same specific gating mechanism as the first BCCR from a cryptophyte (17).

We identified a UV-absorbing intermediate with λ_max ~330–340 nm and a fine structure typical of the deprotonated RSB in the photocycles of HcKCR1, HcCCR, their fast and slow mutants, as well as in two wild-type cryptophyte BCCRs, which suggests that it is a common trait of all BCCRs. We interpret this product as the spectrally distinct M1 intermediate, and the product absorbing at ~400 nm as the M2 state. A large spectral shift between the M1 and M2 states in BCCRs distinguishes them from HsBR, in which M1 and M2 are spectrally identical (29).

The first description of a strongly UV-shifted intermediate, which we designate here as the M1 state, was reported earlier in the cryptophyte BCCRs, RaCCR and RsCCR (27). An intermediate with the structured spectrum and a peak at ~330 nm appeared at the expense of absorption in the visible range and was shown to represent a deprotonated RSB (Figs. 2 and 3 in (27)). In these proteins, the UV-absorbing intermediate is long-lived, which enabled us to investigate it in detail. In other tested BCCRs, including HcKCR1 and HcCCR, the lifetime of the main M1 fraction is in the ms range (Fig. 1). However, antiparallel absorption changes in the short UV range and in the range of maximal absorption of the L-like intermediate strongly suggest that the M1 intermediate appeared upon the L to M1 transition also in these proteins (Fig. 2). Moreover, the absorption difference spectra of the small long-lived fraction of these intermediates (Fig. 3G, H) were very similar to those reported (Fig. 3E, F in (27)). These arguments make it unlikely that the beta band of the L intermediate contributes to the structured UV absorption in any studied BCCRs.

The formation of M1 in BCCRs reflects deprotonation of the RSB preceding channel opening, which becomes possible because the deprotonated RSB no longer prevents passive cation flow. We observed a delay between the M2 rise and the slow phase of M1 formation in HcKCR1 and its mutants. However, in Na⁺-selective HcCCR and GtCCR2, all three processes (slow absorption changes at 330 nm, M2 formation, and photocurrent) developed within the same time window. An earlier study of the HcKCR1 photocycle (12) attributed absorption changes at 384 nm to the M1 state and those at 404 nm to the M2 state. However, the initial parts of the absorption rise at these wavelengths coincided and were slower than the opening of the channel, which suggests that they both reflected M2 accumulation.

Our results revealed that H. catenoides ChRs share a common channel-gating mechanism with GtCCR2, the first characterized cryptophyte BCCR (17). As in the latter, the mutation of the donor residue (D116N) strongly suppressed channel current in both H. catenoides ChRs. Whereas this mutation completely abolished the passive current in Na⁺-selective HcCCR, residual channel activity remained in the corresponding mutant of K⁺-selective HcKCR1 (12,15,16). However, ion selectivity was either not analyzed (12) or continuous light stimulation was used, which complicated the results due to the potential effect of second-quantum absorption (15,16). Here, for the first time, we analyzed the ion selectivity of the HcKCR1_D116N mutant under a single turnover cycle and found that the reversal potential is more positive than under continuous light (15). This implies that the mutation switched the potassium selectivity to sodium selectivity, and Asp116 is a key residue for potassium selectivity. Notably, the side chains of Asp105 and Asp116 undergo the largest reorientations of all residues upon illumination, as revealed by cryo-EM microscopy of HcKCR1 (13). As demonstrated by patch-clamp electrophysiology, neutralizing mutations of these residues (D105N and D116N) strongly inhibit photocurrents (11,12,15,16), further confirming their role in channel gating.

Most importantly, we detected fast inwardly directed currents reflecting active proton transfer from the proton donor Asp116 to an unidentified residue in the cytoplasmic part of the molecule in the proton acceptor (Asp105) mutants of both HcKCR1 and HcCCR. The appearance of this current correlated with or preceded channel opening, suggesting their functional relation. In our earlier GtCCR2 study (17), we postulated that a proton is transferred from the Asp116 homolog to an identified residue, X. Here, we experimentally verified this hypothesis in H. catenoides ChRs and clarified that this residue is located in the cytoplasmic part of the molecule.

Methods

Molecular biology

The HcKCR1 and HcCCR expression constructs encoding the amino acid residues 1–265 (Genbank accession numbers MZ826861 and OL692497, respectively), GtCCR2 expression construct encoding residues 1–300 (Genbank KU761992), and RaCCR1 expression construct encoding residues 1–331 (Genbank MN585300) were fused with the mCherry or C-terminal EYFP tag and cloned in the pcDNA3.1(+) vector (Cat. #V79020, Invitrogen, Carlsbad, CA) for expression in HEK293 (human embryonic kidney) cells, or fused with the C-terminal 8His-tag and cloned in the pPICZalpha-A vector (Cat. # V19520, Invitrogen) for expression in Pichia pastoris. Point mutations were introduced using a QuikChange XL site-directed mutagenesis kit (Cat. #200516, Agilent Technologies, Santa Clara, CA) and verified by DNA sequencing.

HEK293 cell culture, transfection, and patch clamp recording

No cell lines from the list of known misidentified cell lines maintained by the International Cell Line Authentication Committee were used in this study. HEK293 cells were obtained from the American Type Culture Collection (ATCC; catalog #CRL-1573). The absence of mycoplasma contamination was verified by the Visual-PCR mycoplasma detection kit (GM Biosciences, Frederick, MD). The cells were plated on 2-cm-diameter plastic dishes 48–72 hours before the experiments, grown for 24 hours, and transfected with 10 μL of Lipofectamine LTX with Plus Reagent (Cat. #15338100, Thermo Fisher Scientific, Waltham, MA) using 3 μg of DNA per dish. All-trans-retinal (Cat. # 116–31-4, Millipore-Sigma, Burlington, MA) was added immediately after transfection at the final concentration of 5 μM. Plasmids encoding different ChR variants were randomly assigned to identical cell batches for transfection.

Manual patch-clamp recordings were performed using an Axopatch 200B amplifier (Molecular Devices, San Jose, CA). The cells were selected for patching by inspecting their tag fluorescence; non-fluorescent cells and cells in which no GΩ seal was established or lost during recording were excluded from the analysis. The pipette solution contained (in mM) KCl 130, MgCl₂ 2, HEPES 10, pH 7.4. The bath solution contained (in mM): NaCl 130, CaCl₂ 2, MgCl₂ 2, glucose 10, HEPES 10, pH 7.4. The signals were digitized using a Digidata 1440A (Molecular Devices) at a sampling rate of 4 μs per point, using pClamp 10.7 (Molecular Devices). Patch pipettes with resistances of 2–3 MΩ were fabricated from borosilicate glass. The current-voltage dependencies were corrected for liquid junction potentials, which were calculated using the built-in pClamp calculator. Laser excitation was provided by a Minilite Nd:YAG laser (532 nm, pulse width 6 ns, energy 5 mJ; Continuum). The time resolution of our patch-clamp measurements was ~15 μs, as estimated by recording the ultrafast charge displacement associated with the trans-cis retinal isomerization (35). The current traces were logarithmically filtered using LogPro software (39). Curve fitting was performed using OriginPro 2016 (OriginLab Corporation, Northampton, MA). Continuous light pulses were provided by a Polychrome V light source (T.I.L.L. Photonics GMBH, Martinsried, Germany) in combination with a mechanical shutter (Uniblitz Model LS6, Vincent Associates, Rochester, NY; half-opening time 0.5 ms). The action spectra of photocurrents were constructed by calculating the initial slope of photocurrent recorded in response to 15-ms light pulses at the intensity <25 μW mm^-2, corrected for the quantum density measured at each wavelength, and normalized to the maximal value.

Descriptive statistical analysis of the patch clamp data was performed using OriginPro 2016 software. The photocurrent traces recorded from different cells transfected with the same construct were considered biological replicates (reported as n values). These values indicate the number of independent experiments. No statistical methods were used to pre-determine sample sizes, but our sample sizes were similar to those reported in the previous publications (9,11).

Expression and purification of H. catenoides ChRs from Pichia pastoris

The plasmids encoding wild-type H. catenoides ChRs and HcKCR1 mutants were linearized with SacI and used to transform P. pastoris strain SMD1168 (his4, pep4) by electroporation according to the manufacturer’s instructions. Resistant transformants were selected on 0.5 mg ml^-1 zeocin (Cat. # R25001, Gibco, Grand Island, NY). The transgene expression in the presence of 5 μM all-trans-retinal (Cat. # 116–31-4, Millipore-Sigma) was initiated by the addition of 0.5% methanol. After 24–30 h growth, the cells were harvested and disrupted in a bead beater (Cat. #607, BioSpec Products, Bartlesville, OK) or French press as described previously (9). Membrane fragments were collected by ultracentrifugation and solubilized by incubation with 1% dodecyl maltoside (DDM, Anatrace, Cleveland, OH) for 1.5 h at 4°C. The supernatant was mixed with nickel-nitrilotriacetic acid agarose beads (Cat. # R90101, Thermo Fisher) and loaded onto a column. The protein was eluted with buffer containing 20 mM Hepes, pH 7.5, 300 mM NaCl, 300 mM imidazole, and 0.05% DDM. The proteins were further purified by size exclusion chromatography in an elution buffer containing 20 mM Hepes, pH 7.5, 300 mM NaCl, 1 mM sodium azide, 5% glycerol, and 0.05% DDM. The addition of 1 mM sodium azide to the elution buffer helped prevent bacterial growth and thus ensure the stability of the purified protein during purification and storage.

Absorption spectroscopy and flash photolysis

Absorption spectra of detergent-purified protein samples were recorded using a Cary 4000 spectrophotometer (Varian Medical Systems, Palo Alto, CA). Photoinduced absorption changes were measured with a laboratory-constructed crossbeam apparatus. Excitation flashes were provided by a Minilite II Nd:YAG laser (532 nm, pulse width six ns, energy five mJ; Continuum Electro-Optics, San Jose, CA). Measurement light at wavelengths >390 nm was from a 250-W tungsten incandescent lamp and a McPherson monochromator (model 272, Teledyne Acton Optics, Acton, MA). Absorption changes were detected using a Hamamatsu Photonics photomultiplier tube (model R928; Shizuoka, Japan) in conjunction with a second monochromator of the same type. For measurements at lower wavelengths, a protein sample was placed in a UV-transparent cuvette. A xenon lamp, combined with interference filters (half-bandwidth, 10 nm), or a 340 nm LED was used as the source of measuring light. Absorption changes were detected by a UV-enhanced photodiode in combination with a UV (310–360 nm) bandpass glass filter. The laser artifact was measured separately and digitally subtracted from the UV absorption changes. The signals were amplified by a low-noise current amplifier (model SR445A; Stanford Research Systems, Sunnyvale, CA) and digitized with a GaGe Octopus digitizer board (model CS8327, DynamicSignals LLC, Lockport, IL), аt a maximum sampling rate of 50 MHz. Logarithmic data filtration was performed using the GageCon program (40). Deconvolution of the absorption spectra into Gaussian components was performed using Origin software.

Statement of Significance.

Channelrhodopsins (ChRs) are light-gated ion channels from protists. Expression of their genes in neurons and cardiomyocytes enables the control of these cells with light (optogenetics). However, this approach requires a greater understanding of ChRs’ gating mechanisms. Here we applied optical and electrophysiological techniques and successfully elucidated the relationship between photon absorption, intramolecular proton transfers, and ion conductance in a ChR family with K⁺- and Na⁺-selective members. Our findings add to the foundation for molecular engineering of more efficient optogenetic tools.

Data availability

The data is provided in the main manuscript and supplemental figures. The authors will provide additional information upon a reasonable request.