Channelopsin sequences from C. augustae, C. yellowstonensis and C. raudensis
We cloned a channelopsin homolog from each of C. augustae (715 amino acid residues; nucleotide Acc. No. JN596951), C. yellowstonensis (717 residues, Acc. No. JN596948) and C. raudensis (635 residues, Acc. No. JN596949). As in all so far known channelopsins, the new proteins consist of a predicted 7TM (rhodopsin) domain responsible for light-gated channel activity and a C-terminal domain, the function of which is yet unknown. CrChR1/VcChR1 and CrChR2/VcChR2 form two distinct branches on the phylogenetic tree of their 7TM domains (Fig. 1a). The 7TM domains of the new Chlamydomonas sequences do not show closer homology with either the CrChR1/VcChR1 branch or CrChR2/VcChR2 branch, when overall sequence homology of their 7TM domains is concerned (Fig. 1a). The 7TM domains from C. augustae and C. yellowstonensis are however very close to each other.
Figure 1. Phylogenetic trees of the 7TM domains (a) and the C-terminal domains (b) of the so far known channelopsins constructed by the neighbor-joining method. CrChR1 = channelrhodopsin 1 from Chlamydomonas reinhardtii; CrChR2 = channelrhodopsin 2 from C. reinhardtii; VcChR1 = channelrhodopsin 1 from Volvox carteri; VcChR2 = channelrhodopsin 2 from V. carteri; MvChR1 = channelrhodopsin 1 from Mesostigma viride; CaChR1 = channelrhodopsin 1 from C. augustae; CyChR1 = channelrhodopsin 1 from C. yellowstonensis; CraChR2 = channelrhodopsin 2 from C. raudensis; HpChR1 = channelrhodopsin 1 from Haematococcus pluvialis.
Download figure to PowerPoint
Two molecular determinants are conserved in CrChR1/VcChR1 and CrChR2/VcChR2 sequences, respectively, and shown to determine their different properties: (1) Glu87 (CrChR1 numbering) in the predicted first helix is responsible for pH-dependent color tuning and fast channel inactivation of CrChR1, as compared to CrChR2 (20); (2) Tyr226 (CrChR1)/Asn187 (CrChR2) in the predicted fifth helix confers differences in spectral sensitivity, inactivation and kinetics between CrChR1 and CrChR2 (21). According to these criteria, the sequences identified in C. augustae and C. yellowstonensis (CaChR1 and CyChR1, respectively) belong to the CrChR1/VcChR1 class (Fig. 2). This placement is confirmed by their redshifted spectra (see below), characteristic of the CrChR1/VcChR1 class, as compared to the CrChR2/VcChR2 class (13,20). In contrast, the sequence from C. raudensis (CraChR2) belongs to the CrChR2/VcChR2 class according to the two above-mentioned molecular determinants. However, since this sequence failed to generate photocurrents in HEK cells (see below), the lack of functional data makes this placement only tentative.
Figure 2. Partial alignment of Chlamydomonas channelopsin and bacteriorhodopsin (BR) sequences. Black background indicates conserved identical residues. Turquoise background indicates positions of the residues that form the retinal-binding pocket in BR. Green background indicates conserved Glu residues in the predicted second helix. Magenta background indicates molecular determinants that differentiate CrChR1/VcChR1 from CrChR2/VcChR2. Red background indicates residues in the position of the proton donor in BR. Blue background indicates residues in the positions of Glu194 and Glu204 in BR. Yellow background indicates predicted glycosylation sites. Olive background indicates conserved residues known to be phosphorylated in CrChR1 or CrChR2. Underlined characters show the regions that form transmembrane helices in BR.
Download figure to PowerPoint
In the 7TM domain, residues at the critical sites characteristic of other known channelrhodopsin apoproteins (channelopsins) are conserved in all three new homologs. These include: (1) Glu in the position of the Schiff base proton acceptor (Asp85 according to bacteriorhodopsin (BR) numbering); (2) His in the position of the Schiff base proton donor (Asp96 in BR); (3) five Glu residues in or near the predicted second helix; (4) Cys128 and Asp156 (CrChR2 numbering) that form a predicted hydrogen bond between the third and fourth helices (28–30). Out of other residues known to form the retinal-binding pocket in BR, Tyr57, Gly122, Trp182, Asp212 and Lys216 (BR numbering) are conserved in new channelopsins, as they are in previously known ones. Positions of Tyr185 and Trp189 (BR numbering) are occupied by Phe residues, as in previously known channelopsins.
Phylogeny of the C-terminal domain does not match that of the 7TM domain, although CaChR1 and CyChR1 again show close similarity (Fig. 1b). No helices are predicted in the C-terminal domains of either of the new channelopsin sequences, in contrast to CrChR1 (5). As in previously known channelrhodopsins, the C-terminal domains of the new channelopsins contain several highly conserved regions with no homology to any other so far known protein interspersed with repeats that vary in length and amino acid composition among different channelopsin variants. Long stretches of Gly–Met repeats and Met-rich regions are found in CaChR1 and CyChR1, and Gln repeats are found in all three new channelopsins. Such repeats, known as homopolymeric tracts (31), occur in many eukaryotic proteins and have been associated with protein–protein (32) or protein–membrane (33) interactions. A highly conserved region of about 40 residues at the very end of the C-terminal domains present in all so far known channelrhodopsins but MvChR1 shows homology to domains in fibrinogen and ABC transporters that are responsible for protein multimerization and protein–protein interaction. In algal cells, channelrhodopsins are confined to the membrane area above the eyespot (7,11) and are associated with acetylated microtubules of the daughter four-membered flagellar rootlet (34). It is plausible that the function of the C-terminal domains is to control appropriate subcellular positioning of channelrhodopsins.
The extracellular N-terminal regions of CaChR1 and CyChR1 contain a predicted conserved N-glycosylation site. Such sites, although at different positions, are also predicted in the N-termini of CrChR1 and several other channelopsins, but not in CraChR2 (Fig. 2). Another such site conserved in all so far known channelopsins is located at the cytoplasmic end of the third predicted transmembrane helix. The existence of predicted N-glycosylation sites in the cytoplasmic domains of proteins has been mostly ignored, although it was confirmed biochemically in at least one example, the α-subunit of mammalian Na+K+-ATPase (35). Occurrence of glycosylation in heterologously expressed CrChR2 was demonstrated by treatment with peptidyl N-glycosidase F, but its site(s) were not determined (36). A requirement for glycosylation for correct folding and targeting of channelrhodopsins may explain why no functional channelrhodopsin has been produced by expression in Escherichia coli, despite many attempts. All three new channelopsin sequences lack an additional α-helix predicted in the N-terminus of CrChR1 as a signal peptide, as do CrChR2 and both VcChR1 and VcChR2. However, such a helix is predicted in the sequence from Haematococcus pluvialis (nucleotide Acc. No. JN596950). The putative role of this helix in protein targeting needs further investigation.
In CrChR1 and CrChR2, three and one phosphorylated residues, respectively, were identified by phosphoproteomics of the eyespot fractions (37). These residues are found in the cytoplasmic loop next to the 7TM domain that is highly conserved in all so far known channelopsin sequences, with the exception of MvChR1. Out of the three phosphorylated residues of CrChR1, Ser359 is conserved in all five Chlamydomonas channelopsins (Fig. 2), and the corresponding residue (Ser321) is the only phosphorylated site detected in CrChR2 (37). Thr374 is unique for CrChR1 (with Val found at this site in other sequences), and Ser377 is conserved in four sequences with Thr conservative substitution in CraChR2 (Fig. 2). Possibly channelrhodopsin phosphorylation plays a role in the adaptation processes which enable the large range of light intensities over which Chlamydomonas algae exhibit phototaxis (38).
Proteins from psychrophilic organisms show characteristic biases in amino acid composition, compared to their meso- and thermophilic homologs that are believed to increase flexibility at low temperatures, among these are decreased percentages of Pro, Arg and Ala residues, and an increased percentage of Ile residues (39,40). The same trends are observed in Chlamydomonas channelopsins: the combined percentages of Pro, Arg and Ala residues in the sequences of CrChR1, CaChR1 and CyChR1 are 20%, 15.7% and 16%, respectively, whereas the percentages of Ile are 4.9%, 5.9% and 5.2%, respectively. Other trends, such as an increased percentage of Gly, were not detected in channelopsins.
Functional characterization of new channelrhodopsins
The 7TM domains of all three new channelopsins showed robust expression in the plasma membrane of HEK293 cells, as indicated by fluorescence of their EYFP-tags. CaChR1 and CyChR1 exhibited light-gated channel activity in this system, but no currents could be recorded upon expression of CraChR2. The kinetics of the currents generated by CaChR1 and CyChR1 were however quite different from that generated by CrChR1. Figure 3 shows typical signals recorded at the maximal light intensity under our standard conditions, i.e. bath pH 7.4, and holding potential (Vhold) −60 mV (for more details see Materials and Methods section). Upon switching on the light, currents generated by CrChR1 underwent a rapid initial rise with a time constant (τ) ca 1 ms, reached a peak and rapidly (τ ca 5 ms) decreased under sustained illumination, i.e. inactivated to a lower level (Fig. 3a, black line). In many cells the currents showed a subsequent slight increase with τ ca 200 ms. This behavior closely resembled the results reported for CrChR1 earlier by others (41). In contrast, the rise of CaChR1- and CyChR1-generated currents was biphasic. The first rapid phase was similar to that of CrChR1-generated currents, but it was followed by a slower rising phase with τ ca 20 ms (Fig. 3a, red and green lines). The relative contributions of these two components varied from cell to cell. After reaching a peak, CaChR1- and CyChR1-generated currents exhibited very slow inactivation (τ ca 500 ms).
Figure 3. (a) Typical kinetics of light-induced currents generated in HEK293 cells by CrChR1 (black dots), CaChR1 (red dots) and CyChR1 (green dots). (b) Decay of the same currents after 2 s illumination. The currents in (a) were normalized to the peak amplitude, and the currents in (b), to the plateau level and fitted with three (a) or two (b) exponential functions (solid lines). The excitation wavelength was 520 nm for CaChR1 and CyChR1, and 480 nm for CrChR1, which corresponded to their spectral maxima (see below). The traces are average signals measured in response to a series of light pulses delivered with 30 s time intervals. Cells expressing ChRs were selected for EYFP fluorescence before measurements. Bath pH was 7.4, Vhold was −60 mV. For complete ionic conditions, see Materials and Methods section.
Download figure to PowerPoint
For all three tested channelrhodopsin variants the response to the first flash showed a larger peak relative to the plateau level (measured at the end of the light pulse), which was not fully recovered even after 30 min dark interval, suggesting a contribution of a very slow adaptation process, or irreversible bleaching of an unstable fraction of the pigment. However, no difference was observed between responses to the second and all subsequent flashes recorded with 30 s dark intervals. Under these conditions, the peak to plateau ratio at the maximal light intensity was 1.7 ± 0.2 (mean ± SEM, n = 8) for CrChR1, close to the earlier reported results (42). For both CaChR1 and CyChR1 this ratio was significantly smaller: 1.2 ± 0.1 (mean ± SEM, n = 12 and n = 6, respectively). The absolute plateau amplitude was 101 ± 25 pA (mean ± SEM, n = 8) for CrChR1, 64 ± 9 pA (mean ± SEM, n = 12) for CaChR1 and 49 ± 13 pA (mean ± SEM, n = 6) for CyChR1. After switching off the light, the currents decayed biexponentially with τ ca 15 and 120 ms for CaChR1, ca 13 and 150 ms for CyChR1, which was slower than ca 4 and 18 ms measured for CrChR1 (Fig. 3b), but close to that for CrChR2 (20).
The most widely used channelrhodopsin variant, CrChR2, generates large enough currents in HEK cells even without the addition of exogenous retinal ( and E. G. Govorunova and J. L. Spudich, unpublished observations), indicating that its trace amount present in these cells is sufficient for reconstitution of functional protein. However, currents in cells transfected with the new channelrhodopsins or CrChR1 were considerably smaller if no exogenous retinal was added: their plateau levels were only ca 8% for CaChR1 and ca 26% for CrChR1 (data for CyChR1 are not shown), relative to the results obtained with the respective channelrhodopsins in the presence of 2.5 μm exogenous retinal (our standard conditions).
For CrChR1-generated currents, the dependence of the plateau amplitude on the stimulus intensity saturated earlier than that of the peak (Fig. 4a), as was shown previously (41). In fact, the curve for the peak amplitude was biphasic and showed two levels of saturation, the first of which corresponded to that of the plateau level, whereas the second was at more than 10-fold higher light intensity. Therefore, the magnitude of light inactivation, calculated as the difference between the peak and plateau amplitudes relative to the peak amplitude, increased with light intensity and did not saturate even at the highest available intensities (Fig. 4c, solid triangles). Similar results were reported earlier also for CrChR2 (43) and VcChR1/VcChR2 hybrid (44). In contrast, the curves for both peak and plateau amplitudes of CaChR1-generated currents consisted of two phases (Fig. 4b), so that the magnitude of light inactivation reached the maximum at 10% maximal light intensity and then declined (Fig. 4c, open triangles). Similar behavior was observed for CyChR1-generated currents (data not shown).
Figure 4. (a, b) The dependence of peak (solid squares) and plateau (open circles) amplitudes on the stimulus intensity for currents generated by CrChR1 (a) or CaChR1 (b). Data points are the mean normalized values ± SEM (n = 3 [a] and 5 [b]). For complete ionic conditions, see Materials and Methods section. (c) The dependence of light inactivation (calculated as the difference between the peak and plateau amplitudes shown in panels [a] and [b], relative to the peak amplitude) on the stimulus intensity for currents generated by CrChR1 (solid triangles) and CaChR1 (open triangles).
Download figure to PowerPoint
CaChR1- and CyChR1-generated currents showed a typical dependence on the holding potential (Vhold), as described earlier for CrChR1 and other channelrhodopsins (Fig. 5a,c,e). The reversal potentials (Vr) were similar to that for CrChR1 and close to zero under our experimental conditions. Since CrChR1 is known to be a highly proton-selective channel (9), we tested CaChR1 and CyChR1 for proton permeability by measuring current–voltage relationships (I–V curves) under variable external pH. Acidification of the external medium caused an increase in the current amplitude at a given voltage and a shift of the reversal potential to more positive values. The magnitude of this shift for CaChR1 and CyChR1 was similar to that for CrChR1 (Fig. 5a,c,e). Therefore, it can be concluded that both CaChR1 and CyChR1 are highly selective for protons, as is CrChR1 (7,9). It has been reported for CrChR1 that the rate of current decay after switching off the light decreases at acidic bath pH (20), which we also observed in our experiments (Fig. 5b). In contrast, for both CaChR1 and CyChR1 the decay rate slightly accelerated upon a change of the bath pH from 7.4 to 5.4 (Fig. 5d,f).
Figure 5. (a, c and e) Typical current-voltage relationships (I–V curves) for the plateau level measured at the end of a 2 s excitation light pulse upon an increase of Vhold in 20 mV steps from −60 mV at the bath pH 7.4 (solid squares) and 5.4 (open circles) in HEK293 cells transfected with CrChR1, CaChR1 or CyChR1. The wavelength was 520 nm for CaChR1 and CyChR1, and 480 nm for CrChR1, which corresponded to their spectral maxima (see below). (b, d and e) Normalized current decay traces recorded from cells transfected with CrChR1, CaChR1 or CyChR1 at holding potential (Vhold) −60 mV. Traces at the bath pH 7.4 or 5.4 (indicated in the panels) were recorded from the same cell. Note the opposite effects of pH changes on the decay kinetics in CrChR1 and the new channelrhodopsins. Zero time corresponds to the end of a 2 s excitation light pulse. Excitation light was as in a, c and e. Experimental data (dots) were fitted with two exponential functions (solid lines).
Download figure to PowerPoint
The spectral sensitivity of photocurrents was analyzed under low intensity light, as described earlier for MvChR1-generated currents (14). The spectral maxima for the new Chlamydomonas channelrhodopsins, CaChR1 and CyChR1, were at 520 nm at pH 7.4 (Fig. 6a,b, black squares), which is 40 nm longer than that of the action spectrum of CrChR1-generated currents in oocytes, and the absorption spectrum of purified CrChR1 at neutral pH (7,20). To rule out a possible influence of a different expression system and/or a different algorithm of construction of the spectra, we also measured the action spectrum of currents generated by CrChR1 in HEK cells. It showed a maximum at ca 480 nm at pH 7.4 (Fig. 6a, blue open triangles), which confirmed the earlier published results obtained in oocytes (7,20). The spectrum had a broad shape with significant absorption above 500 nm, indicating that the redshifted protonated form of the pigment dominating the spectrum at low pH (7,20) also contributes to the spectrum at pH 7.4. For both CaChR1 and CyChR1, the spectra measured at neutral (7.4) and acidic (5.4) pH were identical (Fig. 6a,b, black and red symbols), in contrast to the spectrum of CrChR1, which showed a significant redshift upon acidification of the medium (7,20). However, a small ca 10 nm blue shift was observed upon the pH change from 7.4 to 9 for both CaChR1 and CyChR1 (Fig. 6a,b, green solid triangles).
Figure 6. (a, b) The action spectra of photoelectric currents generated in HEK293 cells by CaChR1 (a) or CyChR1 (b) at the bath pH 7.4 (black squares), 5.4 (red circles) or 9.0 (green solid triangles). For comparison, the action spectrum of ChR1 from Chlamydomonas reinhardtii measured at pH 7.4 is shown in Panel a (blue open triangles, dashed line).
Download figure to PowerPoint
The maximum of the action spectrum of photocurrents generated by CrChR1 in native C. reinhardtii cells is 505 nm (5), whereas that of currents generated by CrChR1 in heterologous systems at neutral pH is blueshifted by at least 25 nm (Fig. 6a; [7,20]). Since most studies on heterologously expressed CrChR1 were carried out with its 7TM domain, one possible explanation for this difference is that in native cells the spectral properties of CrChR1 are altered by the C-terminal domain. However, the spectrum of the currents generated by the full-length CrChR1 in HEK cells was also blueshifted (data not shown), which rules out that the C-terminal domain presence is sufficient for the redshift at neutral pH.
Expression of CaChR1 in P. pastoris in the presence of all-trans-retinal yielded photoactive pigment, in contrast to the previously reported unsuccessful attempts to produce functional CrChR1 in this system (7). The absorption spectrum of CaChR1 partially purified from Pichia exactly matched the action spectrum of photocurrents generated by this pigment in HEK cells, which indicated that its native state was essentially preserved in detergent (Fig. 7). Spectroscopic characterization of purified CaChR1 will be reported separately.
Figure 7. Comparison of the absorption spectrum of partially purified CaChR1 in detergent (black line) with the action spectrum of photocurrents (open circles), pH 7.4.
Download figure to PowerPoint