Molecular Determinants Differentiating Photocurrent Properties of Two Channelrhodopsins from Chlamydomonas*

A light signal is converted into an electrical one in a single molecule named channelrhodopsin, one of the archaea-type rhodopsins in unicellular green algae. Although highly homologous, two molecules of this family, channelrhodopsin-1 (ChR1) and -2 (ChR2), are distinct in photocurrent properties such as the wavelength sensitivity, desensitization, and turning-on and -off kinetics. However, the structures regulating these properties have not been completely identified. Photocurrents were analyzed for several chimera molecules made by replacing N-terminal segments of ChR2 with the homologous counterparts of ChR1. We found that the wavelength sensitivity of the photocurrent was red-shifted with negligible desensitization and slowed turning-on and -off kinetics when replacement was made with the segment containing the fifth transmembrane helix of ChR1. Therefore, this segment is involved in the determination of photocurrent properties, the wavelength sensitivity, and the kinetics characterizing ChR1 and ChR2. Eight amino acid residues differentiating this segment were exchanged one-by-one, and the photocurrent properties of each targeted mutant ChR2 were further analyzed. Among them, position Tyr226(ChR1)/Asn187(ChR2) is one of the molecular determinants involved in the wavelength sensitivity, desensitization, and turning-on and -off kinetics. It is suggested that these amino acid residues directly or indirectly interact with the chromophore as well as with the protein structure determining the photocurrent kinetics. Some of the chimera channelrhodopsins are suggested to have several advantages over the wild-type ChR2 in the introduction of light-induced membrane depolarization for the purpose of artificial stimulation of neurons in vivo and visual prosthesis for photoreceptor degeneration.

Light is perceived by many living organisms on the earth as vital information. In the case of vertebrates, including human beings, the rhodopsins are the molecules involved in the light perception of the photoreceptor cells in the retina (1)(2)(3). Each rhodopsin is a seven-pass transmembrane molecule homologous to G-protein-linked receptors and activated by a photoisomerization of a covalently attached chromophore, 11-cis-retinal, to all-trans configuration. The signal is then transmitted to cyclic GMP phosphodiesterase and reduces the intracellular level of cGMP, which opens the cyclic nucleotidegated cation channels (4,5). A light signal is thus converted into an electrical one through a cascade of at least four molecules. On the other hand, during phototactic and photophobic movements of unicellular green algae, light is perceived by archaeatype rhodopsins that are localized in small regions of the plasmalemma covering the eyespot (6 -9). Two rhodopsins named channelrhodopsin-1 (ChR1) and -2 (ChR2) were identified in a green alga Chlamydomonas reinhardtii and extensively studied (7,8,10,11). Each channelrhodopsin consists of a seven-pass transmembrane apoprotein, channelopsin, and a retinal which covalently binds to the apoprotein. The photoisomerization of all-trans-retinal to 13-cis configuration is coupled to conformational changes in the protein and causes the permeation of ions. A light signal is thus converted into an electrical one in a single molecule (12). When exogenously expressed in Xenopus oocyte, the ChR1 photocurrent was maximally activated at 500 nm (10). Previously, the ChR1 photocurrent was thought to be primarily carried by H ϩ , but a recent study noted that it was also dependent on other cations (13). On the other hand, the ChR2 photocurrent was preferentially activated at 460 nm and carried by cations like Na ϩ , K ϩ , Ca 2ϩ , as well as H ϩ (11). The photocurrents are also kinetically distinct. The ChR1 photocurrent was hardly desensitized during bright light illumination, although that of ChR2 was rapidly desensitized (10,11,14,18). In this study, we replaced the N-terminal segments of ChR2 with the homologous counterparts of ChR1 and generated several chimeras. These chimeras generated photocurrents showing intermediate properties between ChR1 and ChR2. A possible molecular determinant was identified in the fifth transmembrane helix, which is involved in both the light absorbance and the photocurrent kinetics. Some chimera molecules may be more optimal than ChR2 to depolarize exogenously expressed cells by light.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Expression-The cloning methods of channelopsin1 (chop1) have been described previously (8). Chimeric channelopsins (chops) 2 between chop1 (amino acids 1-345; GenBank TM accession numbers, AB058890/AF385748) and chop2 (amino acids 1-315) (GenBank TM accession number, AB058891/AF461397) with 5Ј-EcoRI and 3Ј-BamHI restriction sites were constructed by overlap extension PCR as described previously (15) using KOD plus DNA polymerase (Toyobo, Osaka, Japan). A chimeric chop fragment was obtained, purified, digested by EcoRI and BamHI, and subcloned in-frame into the plasmid pVenus-N1 that has the Venus construct (14). Single amino acid mutants of chop2-(1-315)-Venus were produced using KOD plus mutagenesis kit (Toyobo). Primers used in overlap extension and site-directed mutagenesis are listed in supplemental data 1. Coding regions in all constructed plasmids were fully sequenced to verify that no undesired mutations had been introduced by PCR. HEK293 cells were cultured at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum and transfected using Effectene transfection reagent (Qiagen, Tokyo, Japan) according to the manufacturer's instructions. Twenty four hours post-transfection, the cells were replated onto the collagen-coated glass coverslips and served for the electrophysiology. We did not supplement the culture and experimental media with retinal, but we observed enough large photocurrents for the following experiments.
Measurements of Fluorescence-Cells were prepared for electrophysiological recordings 48 h after transfection. Under conventional confocal microscopy (LSM 510 META, Carl Zeiss, Oberkochen, Germany), Venus was excited at 488 nm, and the fluorescence was obtained with an LP505 emission filter. All the images were captured under fixed conditions such as the laser power and the photomultiplier gain so as to compare the fluorescence distribution among the ChR2 variants. The fluorescence derived from the contour of a cell was automatically detected by an algorithm (supplemental data 2) and then measured in arbitrary fluorescent units. This value was normalized by the number of pixels in the region, and the fluorescence density was obtained.
Light Illumination-Three modes of light illumination were used according to the experiments. For the wavelength-response relationship monochromatic light at 400 -560 nm (width, 10 nm) of 1-s duration was applied every 20 s under a conventional epifluorescence system equipped with a xenon lamp and electromagnetic shutter (CAM-230, Jasco, Tokyo, Japan). Usually the photocurrent was measured by 2-3 repeats of a protocol, in which the wavelength was changed in the order of 460, 480, 500, 520, 540, 560, 560, 540, 520, 500, 480, 460, 440, 420, 400, 400, 420, 440, and 460 and was averaged for each wavelength. The power density at each wavelength was directly measured by a thermopile (MIR-100Q, Mitsubishi Oil Chemicals, Tokyo, Japan), and was (in mW mm Ϫ2 ) 0.021 (400 nm), 0.018 (420 nm), 0.027 (440 nm), 0.027 (460 nm), 0.018 (480 nm), 0.021 (500 nm), 0.015 (520 nm), 0.014 (540 nm), or 0.012 (560 nm), respectively. To investigate the wavelength-response relationship, the amplitude of the peak photocurrent at each wavelength was divided by the power density because there was assumed to be a linear relationship between these values (14). For the I-V relationship analysis the xenon lamp light was filtered at 477 Ϯ 10 nm (power density, 1.6 mW mm Ϫ2 ) and was applied for 0.2 s duration every 10 s under a conventional epifluorescence system equipped with an electromagnetic shutter (OSP-3, Olympus). For the analysis of the photocurrent kinetics, we used a blue LED (470 Ϯ 25 nm, LXHL-NB98, Lumileds Lighting Inc., San Jose, CA) regulated by a pulse generator (SEN-7203, Nihon Kohden, Tokyo, Japan) and computer software (pCLAMP 9, Molecular Devices Co.). The maximal power density of LED light was 0.077 mW mm Ϫ2 .
Data Analysis-The desensitization was quantified as the difference between the peak and the steady-state amplitudes divided by the peak amplitude. The time constants of turning-on (ON) transition (ON time constant, ON ) were obtained by fitting the photocurrent using the simplex method of nonlinear least squares protocol of the appropriate software (Clampfit 9.2 and 10.1, Molecular Devices Co.). Because the photocurrent was better fitted by two exponential functions after light-off, the effective turning-off (OFF) time constant ( OFF ) was measured as the time to reach e Ϫ1 (37%) of the steady-state amplitude. All
Membrane Expression-These channelrhodopsins were tagged with one of the green fluorescent protein derivatives, Venus, at their C-terminal ends and expressed in HEK293 cells. Under confocal microscopy, ChR2-(1-315)-Venus fluorescence was distributed preferentially at the contour of the cell, suggesting that it is distributed in the plasma membrane ( Fig.  2A). Although the fluorescence was weak in the cases of ChR1-(1-345), it was again preferentially distributed at the contour of the cell. The expression was obviously enhanced in any case of chimera channelrhodopsin. As shown in Fig. 2B, the fluorescence density in the plasma membrane region was quantified and compared among ChR2-(1-315), ChR1-(1-345), and chimera channelrhodopsins. The fluorescence density of ChR-(Abcdefdg) was similar to ChR2, whereas those of ChR(ABcdefg), ChR(ABCdefg), ChR(ABCDefg), and ChR(ABCDEfg) were 150 -200% of ChR2. The further "f"-to-"F" exchange significantly reduced the fluorescence density but to a level as great as that of ChR2-(1-315). In contrast, the fluorescence density of ChR1-(1-345) was very small. This result was generally consistent with the immunoblot quantification of membrane-targeted proteins, although the immunoblot density of ChR1-(1-345) was similar to that of ChR(ABCDEFg) (supplemental data 3). Therefore, the fluorescence density at the contour of the cell would well represent the density of the molecule at the plasma membrane.
Reevaluation of ChR1 Photocurrent at pH 7.4-Because the ChR1 photocurrent has been studied at pH 4 -6 in previous studies (10,18), its properties at pH 7.4 were re-investigated. Under whole-cell patch clamp at Ϫ40 mV, a 1-s light pulse evoked a negative ChR1-(1-345) current with negligible desensitization at wavelengths of 460 and 520 nm (Fig. 3A), whereas the same light pulse evoked a ChR2-(1-315) current with a peak and a plateau, both of which were dependent on the wavelength (Fig. 3B). For each, the photocurrents were evoked at wavelengths of 400 -560 nm. Because the light power was dependent on the wavelength, it was measured directly and was in the range of 0.012-0.027 mW mm Ϫ2 . The peak photocurrent was almost proportional to the light power density in this small value range (14). The photocurrent sensitivity to each wavelength was thus obtained by dividing the peak photocurrent amplitude by the light power density and normalized to the value at 480 nm. The sensitivity was maximal between 480 and 520 nm in the case of ChR1-(1-345) (Fig. 3C). This sensitivitywavelength relationship was almost identical to the sensitivitywavelength relationship of ChR1 at pH 7.5 (13). On the other FIGURE 1. Primary structure of channelopsin chimeras. A, sequence alignment of two channelrhodopsin apoproteins, channelopsin1 (chop1, amino acids 1-345) and channelopsin2 (chop2, amino acids 1-315). The identical amino acids are indicated with an asterisk. The putative seven transmembrane domains are indicated by blue cylinders (16,17) or green cylinders (8). Yellow background indicates the amino acid residues predicted to lie close to retinal. Crimson letters indicate the consensus retinal-binding Lys. Blue letters indicate the consensus residues defining the Schiff base counterion complex. Magenta letters indicate the amino acid residues involved in the retinal-Schiff base binding pocket. Green letters indicate the amino acid residues affecting the electron distribution of retinal-Schiff base. Tyr 226 ("E")/Asn 187 ("e"), which are involved in both the wavelength preference and channel gating, are shown by white letters with black background. Vertical red double lines indicate the segment boundaries. The segments are named from the N-terminal as "A" to "G" of chop1 and as "a" to "g" of chop2. B, primary constructs of the chimera channelrhodopsins replacing the N-terminal segments of chan-nelrhodopsin2 (ChR2-(1-315)) with homologous counterparts of channel-rhodopsin1 (ChR1-(1-345)).
The ChR1-(1-345) photocurrents were usually small in comparison with the ChR2-(1-315) as noted in Fig. 3, A and B. The current-voltage (I-V) relationship was investigated for each peak photocurrent evoked by blue light (477 Ϯ 10 nm), which is the wavelength at the maximal sensitivity for both ChR1-(1-345) and ChR2-(1-315), with a pipette solution containing 120 mM Na ϩ . When the extracellular solution was the standard Tyrode solution containing 142 mM Na ϩ , the I-V relationships of ChR1 and ChR2 were slightly rectified inwardly as shown in Fig. 3, E and F, i.e. the slope conductance at positive membrane potentials was smaller than that at negative membrane potentials. Under close inspection, the ChR2 photocurrent was more rectified than the ChR1 one (supplemental data 4). To evaluate the size of the photocurrent, we estimated the effective conductance as a slope of the I-V relationship between Ϫ40 and 40 mV divided by the input capacitance of the cell. The mean effective conductance of ChR1-(1-345) was 0.054 Ϯ 0.009 S/pF (n ϭ 20), whereas that of ChR2-(1-315) was 0.228 Ϯ 0.037 S/pF (n ϭ 20) with a significant difference (p Ͻ 0.0001). The conductance was one of the differences between ChR1-(1-345) and ChR2-(1-315) photocurrent as noted previously (16). Its difference would result partly from the difference of expres-  sion in the plasma membrane (Fig. 2B). When the extracellular Na ϩ was reduced from 142 to 20 mM, the reversal potential (E rev ) of the ChR1-(1-345) peak photocurrent was shifted from 12.6 Ϯ 2.5 to Ϫ24.9 Ϯ 2.3 mV (Fig. 3E), whereas that of ChR2-(1-315) was shifted from 13.9 Ϯ 1.8 to Ϫ21.1 Ϯ 2.0 mV (Fig.  3F). As a result the Na ϩ -dependent shift of E rev , ⌬E rev (Na ϩ ), of ChR1-(1-345) was Ϫ37.5 Ϯ 2.2 mV (n ϭ 11), which was similar to that of ChR2-(1-315) (Ϫ35.0 Ϯ 1.1 mV, n ϭ 11), and the difference was insignificant. Similar results were obtained when the plateau photocurrents were compared. Therefore, we concluded that ChR1-(1-345) and ChR2-(1-315) are similar in terms of the relative Na ϩ permeability to other cations (K ϩ , H ϩ , Ca 2ϩ , Mg 2ϩ , and N-methyl-D-glucamine ϩ ) at pH 7.4. At pH 6.0, as noted previously (10), the ChR1 photocurrent was large in effective conductance with very positive E rev and showed little dependence on Na ϩ (supplemental data 5) as the relative Na ϩ permeability of ChR1 is dependent on the pH and is small at low pH (13).
Wavelength-Photocurrent Relationships of the Chimera Channelrhodopsins-The wavelength sensitivity of the photocurrent was similarly investigated for each chimera channelrhodopsin expressed in HEK293 cells. As shown in Fig. 4A, each wavelength relationship of the chimera photocurrent was intermediate between ChR1-(1-345) and ChR2-(1-315). For example, in the case of ChR(ABCDEfg), the light sensitivity was maximal at 480 nm, but the photocurrent was more preferably activated at wavelengths 500 -540 nm than the wild-type ChR2. The wavelength sensitivity was well quantified by the ratio of the mean photocurrent at 500 and 520 nm to the mean at 440 and 460 nm (G/B ratio) (Fig. 4B). Although the significant enhancement of G/B ratio was limited to the exchange of segment "e" of ChR2-(1-315) with "E" ("e"-to-"E" exchange) or "g"-to-"G" exchange, it was significantly greater than that of ChR2-(1-315) but significantly smaller than that of ChR1-(1-345) in chimeras ChR(ABcdefg), ChR(ABCdefg), ChR(ABCDEfg), and ChR(ABCDEFg) (p Ͻ 0.005).
Ion Permeability of Chimera Channelrhodopsins-The ion permeability was again investigated for each chimera channelrhodopsin by comparing the I-V relationships at high Na ϩ (142 mM) and low Na ϩ (20 mM) extracellular environments (Fig. 5A). As summarized in Fig. 5B, the ⌬E rev (Na ϩ ) was similar to those of ChR1-(1-345) and ChR2-(1-315) in the cases of ChR(ABcdefg), ChR(ABCdefg), ChR(ABCDefg), ChR(ABCDEfg), and ChR(ABCDEFg). Therefore, these chimera channelrhodopsins are similar to ChR2-(1-315) in terms of the relative Na ϩ permeability to other cations (K ϩ , H ϩ , Ca 2ϩ , Mg 2ϩ , and N-methyl-D-glucamine ϩ ) at pH 7.4. However, the absolute ⌬E rev (Na ϩ ) of ChR(Abcdefg) was significantly smaller than those of the others. It is possible that the permeability of this chimera is less selective to Na ϩ and permeable to large cations such as N-methyl-D-glucamine ϩ (11).
The photocurrents of ChR(Abcdefg) and other chimera channelrhodopsins were also less rectified than that of ChR2-(1-315) in their I-V relationships ( Fig. 5A; supplemental data  4). On the other hand as shown in Fig. 5C, the effective conductance was variable among chimera channelrhodopsins. It was significantly enhanced by the "b"-to-"B" or "e"-to-"E" exchange, whereas significantly reduced by the "d"-to-"D" or "f"-to-"F" exchange. The effective conductance is dependent on the single channel conductance, the probability of a channel being open and the channel density in the plasma membrane. The results of Fig. 5C were consistent with the high fluorescence density at the plasma membrane (Fig. 2B) in the cases of ChR(ABcdefg), ChR(ABCdefg), and ChR(ABCDEfg). Therefore, the large effective conductance was partly attributable to the high channel density, although there remains a possibility that other factors are different. Because the effective conductance was smaller than expected from the fluorescence density in the cases of ChR(ABCDefg) and ChR(ABCDEFg), either the single channel conductance or the probability of a channel being open was possibly small in these cases.
Photocurrent Profiles of Chimera Channelrhodopsins-When an LED light was applied as a square pulse, the ChR2-(1-315) photocurrent peaked almost instantaneously, desensitized rapidly to a steady state within 1 s, and turned off rapidly. Because the photocurrent kinetics are dependent on the light intensity, the holding potential as well as pH, the photocurrents of chimera channelrhodopsins were measured at the maximal LED power, at the holding potential of Ϫ20 mV and at pH 7.2 inside and pH 7.4 outside of membrane. Although the photocurrents of ChR(Abcdefg), ChR(ABcdefg), ChR(ABCdefg), and ChR(ABCDefg) showed characteristic peak-and-plateau time courses like ChR2-(1-315), they were found to be diverse in their kinetics (Fig. 6A). The desensitization of the ChR(Abcdefg) photocurrent was smaller than that of ChR2-(1-315) but larger than that of ChR(ABcdefg) with the maximal LED power. With the progressing replacement of N-terminal segments, the desensitization had a tendency to be further reduced, e.g. ChR(ABCDEfg). The magnitude of desensitization was normalized by the peak current and compared among the channelrhodopsins (Fig. 7A). The desensitization was significantly reduced by the "a"-to-"A" exchange and was further by "b"-to-"B" exchange. Although the contributions to the desensitization of segments "C/c" and "D/d" were small, "e"-to-"E" exchange significantly reduced the desensitization. The "f"-to-"F" exchange further reduced the desensitization, but to a significantly larger level than that of ChR1. The reduced desensitization of ChR(ABCDEfg), ChR(ABCDEFg), and ChR1-(1-345) may come from the shift of the sensitivity-wavelength relationship. To test this, using the data of Fig.  4A, the photocurrents measured at each wavelength from 400 to 560 nm are integrated (Fig. 6B), and the desensitization was evaluated on this integrated photocurrent. As shown in Fig. 7B, the desensitization of the integrated photocurrent was well correlated to that evoked by the maximal LED power (r ϭ 0.96, p Ͻ 0.0005, paired t test). Therefore, the reduced desensitization cannot be explained by the shift of the sensitivity-wavelength relationship. However, the difference between ChR(ABCDEFg) and ChR1-(1-345) in Fig. 7A may be attributed to the different wavelength sensitivity because no obvious difference was observed in the desensitization of the integrated photocurrents (Fig. 7B).
These photocurrents were also variable in their turning-on (ON) and turning-off (OFF) kinetics. As shown in Fig. 6, C and D, where the LED light pulse was applied at its maximal power, the photocurrent of ChR(Abcdefg) was faster than that of ChR2-(1-315) in the ON kinetics, whereas the two were similar in the OFF kinetics. The photocurrent of ChR(ABcdefg) was faster than that of ChR(Abcdefg) in both ON and OFF, whereas that of ChR(ABCDEfg) was similar to ChR2-(1-315) in both ON and OFF. The ON time constant ( ON ) of ChR2 is a function of the LED power, whereas the effective OFF time constant ( OFF ) is not (14). Therefore, the ON and OFF were both compared at the maximum LED power (Fig. 7, C and D). The contribution of segment "A/a" to the OFF was negligible, whereas both the OFF and ON became significantly small by replacing  segment "b" of ChR(Abcdefg) with "B." Although both the OFF and ON were significantly enlarged by "e"-to-"E" exchange, they were significantly reduced by "f"-to-"F" exchange.
Significance of E-segment as a Determinant of the Photocurrent Properties-The above results indicate that each photocurrent property was variably dependent on the segment exchanges. As shown in Fig. 8, the significance score of the Mann-Whitney U test was compared with the photocurrent properties of the G/B ratio, desensitization, and ON and OFF , and the effects of each single segment exchange were evaluated as follows: from ChR2-(1-315) to ChR(Abcdefg), from ChR(Abcdefg) to ChR(ABcdefg), from ChR(ABcdefg) to ChR(ABCdefg), from ChR(ABCdefg) to ChR(ABCDefg), from ChR(ABCDefg) to ChR(ABCDEfg), from ChR(ABC-DEfg) to ChR(ABCDEFg), or from ChR(ABCDEFg) to ChR1-(1-345). Because the "b"-to-"B" and "f"-to-"F" exchanges definitely affected the photocurrent kinetics as well as the effective conductance, these structures may be related to the channel properties. On the other hand, the "g"-to-"G" exchange more definitely influenced the G/B ratio than photocurrent kinetics. We found that all four properties were definitely affected by the "e"-to-"E" exchange. Therefore, it is suggested that segment "E/e" is one of the key determinants of photocurrent properties such as the wavelength sensitivity and the kinetics.
To evaluate the effects of the amino acid exchanges, the statistical significance was scored according to the p value of the Mann-Whitney U test as done previously. Fig. 11A summarizes the significance scores of the single segment and the single amino acid exchanges from "e" to "E." We found that all four photocurrent properties, the G/B ratio, desensitization, ON , and OFF , were definitely affected by the single segment exchange from ChR2-(1-315) to ChR(abcdEfg). These properties were also definitely changed in ChR2(N187Y). The ChR2(N187Y) resembled ChR(abcdEfg) in all four properties.

Molecular Determinants Differentiating ChR2 from ChR1-
In this study we divided ChR2-(1-315) into seven segments, "a" to "g," so that each segment contained one of the seven putative transmembrane domains. By replacing the N-terminal segments of ChR2-(1-315) with the homologous counterparts of ChR1-(1-345), "A" to "F," we identified the structure differentiating ChR2 from ChR1. The G/B ratio of the photocurrent,  used to evaluate the wavelength preference, was gradually increased with the replacements of the N-terminal segments, "A/a" to "C/c," and further by the "e"-to-"E" replacement. But the effect of the "g"-to-"G" replacement was the most definitive, suggesting that the seventh transmembrane helix has structures involved in the color tuning of the molecule. Previously, ChR1 and ChR2 have been differentiated in the relative permeability of Na ϩ (10, 11). However, as the cation selectivity of ChR1 photocurrent is dependent on the pH, the photocurrent is less dependent on Na ϩ at low pH (13). Consistent with this notion, we did not find any difference in the ion selectivity at pH 7.4. On the other hand, they were differentiated by the effective conductance as noted previously (16). The effective conductance was large even when the first five segments "a" to "e" of ChR2-(1-315) were replaced by "A" to "E" of ChR1-(1-345). The large effective conductance of this chimera channelrhodopsin was in part attributable to the expression level in the plasma membrane. However, further "f"-to-"F" replacement reduced the effective conductance to as small as that of ChR1-(1-345). This is somewhat inconsistent with the fluorescence density of these molecules in the plasma membrane. The "F/f" segment itself or its interaction with other segments is likely to be involved in the regulation of channel conductance or probability of a channel being open. We profiled the photocurrent kinetics by three quantities, the desensitization, ON , and OFF . Although variable, the chimeras ChR(Abcdefg), ChR(ABcdefg), ChR(ABCdefg), ChR(ABCDefg), and ChR(ABCDEfg) generated photocurrents with properties kinetically intermediate between ChR1-(1-345) and ChR2-(1-315). On the other hand, the chimera ChR(ABCDEFg) was kinetically more similar to ChR1-(1-345). Taken together, the properties differentiating the photocurrent kinetics of ChR1 from those of ChR2 are not confined to a specific structure, but are variably influenced by the exchange of each segment. Even the exchanges of the "C/c" and "D/d" segments definitely changed the photocurrent properties, although the third and the fourth transmembrane domains are highly conserved among channelrhodopsins (16,17). This is consistent with the notion that these domains are also involved in the channel properties of ChR2-(1-315), and indeed, the single amino acid mutation H134R enhanced the photocurrent amplitude (19). In this study we further investigated the "E/e" segment because it definitely changed the photocurrent properties without reducing the effective conductance.
Fifth Transmembrane Helix as the Determinant of Wavelength Sensitivity-In relation to other archaea-type rhodopsins, interactions between the retinal chromophore and its protein environment are the determinants of the wavelength preferences. Three factors are suggested to be involved in these interactions (20,21) as follows: (i) the strength of the protein environment forcing the retinal to be a coplanar 6s-trans conformation; (ii) the protein environment electrically interacting with the chromophore with its counter ion or hydrogen bond acceptor; and (iii) a less well defined long range coupling, which   FEBRUARY 27, 2009 • VOLUME 284 • NUMBER 9 is thought to involve the interaction of polar or polarizable amino acid groups with the chromophore, influencing the stability of the ground state. The amino acid residues assumed to be related to these interactions are mostly present from the third to the seventh transmembrane domains (22,23) and are almost conserved between ChR1 and ChR2 (Fig. 1), e.g. the consensus retinal-binding residue, Lys 296 ("G" segment)/ Lys 257 ("g" segment) in the seventh transmembrane domain; the consensus residues defining the Schiff base counterion complex, Arg 159 ("B")/Arg 120 ("b") and Glu 162 ("B")/Glu 123 ("b") in the third transmembrane domain; and Asp 292 ("G")/Asp 253 ("g") in the seventh transmembrane domain, which are critical for the wavelength preference (17,24,25). The nonpolar amino acid residues involved in the retinal-Schiff base binding pocket, Gly 220 ("E")/Gly 181 ("e"), Leu 221 ("E")/Leu 182 ("e"), and Cys 222 ("E")/Cys 183 ("e") in the fifth transmembrane domain, which are predicted to be closely located to the ␤-ionone ring of retinal, are all conserved between ChR1 and ChR2 (17). Other amino acid residues that affect the electron distribution of the retinal-Schiff base are Asn 153 ("B")/His 114 ("b"), Glu 274 ("F")/ Glu 235 ("f"), and Ser 284 ("G")/Ser 245 ("g"). However, the wavelength sensitivity of the chimera was only minimally changed by the "b"-to-"B" exchange. On the other hand, the single "e"-to-"E" exchange of ChR2-(1-315) red-shifted the wavelength sensitivity of the photocurrent, and the single "E"-to-"e" exchange of ChR(ABCDEfg) blue-shifted it. This indicates the presence of another structure either regulating the electron distribution of the retinal-Schiff base or interacting with it through long range coupling. The results of our experiments using ChR2(N187Y) and ChR1(Y226N) indicate that one-by-one amino acid replacement of Asn 187 ("e") to the corresponding Tyr red-shifted the sensitivity-wavelength relationship, whereas the reverse replacement of Tyr 226 ("E") to Asn blueshifted it. The position of Tyr 226 ("E")/Asn 187 ("e") would lie adjacent to the nonpolar retinal binding pocket if the ␣-helix structure is predicted for the fifth transmembrane domain. It is possible that these amino acid residues have some additional interaction with retinal. The hydroxyl-bearing Tyr 226 would probably interact preferentially with the excited-state charge distribution around the ␤-ionone ring of retinal to lower the excitation energy and red-shift the absorption spectrum (26,27). Alternatively, these residues may interact with retinal-Schiff base with long range coupling.

Determinants Differentiating Channelrhodopsin Photocurrents
An alignment of the motif corresponding to the fifth transmembrane helix domain for ChR1 and ChR2 is shown in Fig.  11B, in comparison with the other phylogenic relatives. The recently identified channelrhodopsins from Volvox carteri, VChR1 (17) and VChR1/2 (28), are highly homologous to both ChR1 and ChR2. Of 18 amino acids conserved between ChR1 and ChR2, 12 are found in both VChR1 and VChR1/2, 4 are found only in VChR1/2, and 1 is found only in VChR1. The counterpart of Tyr 226 (ChR1)/Asn 187 (ChR2) is Asn in the case of blue-absorbing VChR1/2, whereas it is Tyr in the case of more red-shifted VChR1. Although this is consistent with the present result that Asn-to-Tyr replacement red-shifted the absorption spectrum, the involvement of other amino acid residues should not be excluded. On the other hand, the sequences of sensory rhodopsin II from Natronobacterium pharaonis (the maximum absorption at 500 nm) and bacteriorhodopsin from Halobacterium salinarum (the maximum absorption at 550 nm) are less homologous in this region. Of 18 amino acids conserved between ChR1 and ChR2, 3 are found only in sensory rhodopsin II and 0 in bacteriorhodopsin. It is thus possible that the color-tuning contribution of the fifth transmembrane domain is different in these prokaryotic rhodopsins.
Fifth Transmembrane Helix as the Determinant of Photocurrent Kinetics-The "E/e" segment, which contains the fifth transmembrane helix, was also one of the structural determinants of the photocurrent kinetics. When the "e"-to-"E" exchange was made on chimera ChR(ABCDefg), the desensitization was reduced, whereas both the ON and OFF were definitely enlarged. It was also the case when the same exchange was made on ChR2-(1-315). Therefore, the photocurrent kinetics such as desensitization, ON , and OFF are dependent on the structure intrinsic to the "E/e" segment. In the "E/e" segment the amino acid residue Tyr 226 ("E")/Asn 187 ("e") again was central in determining the photocurrent kinetics. The desensitization was strongly reduced in ChR2(N187Y) and ChR2(H191N), whereas it was enhanced by Y226N exchange of ChR1-(1-345). However, it was also dependent on other structural components such as the "A/a" and "B/b" segments. The OFF was definitely enhanced in ChR2(N187Y) but reduced in ChR2(L180M) and ChR2(A186I). The Y226N exchange of ChR1-(1-345) also reduced OFF . It is thus suggested that the position Tyr 226 ("E")/Asn 187 ("e") is involved in the channel closure but that the interaction with other key residues is important. However, both the N187Y exchange of ChR2-(1-315) and the Y226N exchange of ChR1-(1-345) elongated ON . Because ON is dependent on the light intensity (14) but is also expected to be dependent on OFF as a consequence of the photocycle response (18,28,29), further studies are necessary to reveal the variability of this value.
Conclusion-This work revealed the significance of the fifth transmembrane helix, particularly its residue Tyr 226 (ChR1)/ Asn 187 (ChR2), as a molecular determinant differentiating the photocurrent properties between ChR1 and ChR2. Chimeric mutation by a segment exchange has been a powerful tool for identifying the molecular determinants involved in a certain function. For the first time, this study made a series of chimeric mutations by replacing the molecular segments containing each transmembrane helix of ChR2 with the homologous counterparts of ChR1. Based on systematic and quantitative studies of the distribution of the molecules, wavelength sensitivity, reversal potential, and kinetic profile, it is suggested that the fifth transmembrane helix is involved in both the wavelength sensitivity and the kinetic profile of the photocurrent. Moreover, all these results strongly suggest that the residue Tyr 226 ("E")/Asn 187 ("e") lies at a key position enabling it to interact with the retinal-Schiff base. It also lies at a position interacting with other residues involved in the photocurrent kinetic profiles such as desensitization, ON , and OFF . The significance of Tyr 226 ("E")/Asn 187 ("e") has to be further investigated by mutations replacing Asn 187 of ChR2-(1-315) with various amino acid residues and/or by x-ray crystallographic analysis (20,30).
Recently the ChR2-mediated photostimulation of neurons has been applied to investigate the function of neural networks in vivo (31)(32)(33)(34). Moreover, ChR2-expressing transgenic animals have been generated that are successfully used to study the neural basis of behavioral responses in Caenorhabditis elegans (19), zebrafish (35), and mammals (36). Given its superiority in spatio-temporal resolution, ChR2 has become a powerful tool for the investigation of neural networks.
ChR2 is also a potential tool as a visual prosthesis for photoreceptor degeneration (37). In retinal degenerative diseases such as retinitis pigmentosa, photoreceptor cells are lost while preserving the inner retinal neurons such as retinal ganglion cells and bipolar cells. Exogenous expression of ChR2 in these neurons using viral vectors (38,39) or in vivo electroporation methods (40) restores visually evoked responses in the visual cortex of rodents.
However, some further technological developments are necessary to achieve the full potential of ChR2. First, photosensitive channels with various wavelength sensitivities are desirable. One of the Volvox-derived channelrhodopsins, VChR1, may be a candidate because it has a peak absorbance at 540 nm (17). Second, the small single channel conductance of ChR2 (in the magnitude of fS) (12,37) should be compensated by facilitating the membrane expression. Alternatively, it would be useful to develop a ChR2 with increased single channel conductance. Third, the prominent desensitization of ChR2 photocurrent limits its application for repetitive stimulation at high frequency. This could be overcome by reducing the desensitization or facilitating the recovery rate.
Some of the chimera channelrhodopsins have several advantages over the wild-type ChR2 in applying to the light-induced membrane depolarization (14,31,41). For example, the chimera ChR(ABCDEfg) is sensitive to wavelengths of 480 -540 nm, has large effective conductance when expressed in HEK293 cells, and its photocurrent desensitization was almost negligible. Moreover, its ON and OFF kinetics were the same as those of ChR2. Therefore, this chimera, to which we give the name "channelrhodopsin/wide receiver," would become a good tool to stimulate neurons repetitively with low intensity flash light and also a potential tool as a visual prosthesis of photoreceptor degeneration (38,39). Another potentially useful chimera is ChR(ABcdefg) because it is very fast in both its ON and OFF kinetics. It is also characterized by small desensitization and large effective conductance. This chimera, to which we give the name "channelrhodopsin/fast receiver," would become a good tool to stimulate neurons repetitively at high frequency flash. The application of channelrhodopsins would be facilitated by this new lineup of chimeras.