Photoactivation of Channelrhodopsin*

Channelrhodopsins (ChRs) are light-gated ion channels that control photomovement of microalgae. In optogenetics, ChRs are widely applied for light-triggering action potentials in cells, tissues, and living animals, yet the spectral properties and photocycle of ChR remain obscure. In this study, we cloned a ChR from the colonial alga Volvox carteri, VChR. After electrophysiological characterization in Xenopus oocytes, VChR was expressed in COS-1 cells and purified. Time-resolved UV-visible spectroscopy revealed a pH-dependent equilibrium of two dark species, D470/D480. Laser flashes converted both with τ ≈ 200 μs into major photointermediates P510/P530, which reverted back to the dark states with τ ≈ 15-100 ms. Both intermediates were assigned to conducting states. Three early intermediates P500/P515 and P390 were detected on a ns to μs time scale. The spectroscopic and electrical data were unified in a photocycle model. The functional expression of VChR we report here paves the way toward a broader structure/function analysis of the recently identified class of light-gated ion channels.

conductance with a time constant of 200 s after the flash, thus closely resembling the photocurrent rise in living algae (2,6).
The observation that ChRs rapidly depolarize membranes within a few milliseconds in response to brief light flashes motivated neurophysiologists to use ChRs for controlling nerve cell firing. Originally, three groups in parallel expressed ChR in hippocampal neurons and triggered action potentials by short light pulses. The neurons followed the light stimulation protocols with action potential firing up to 30 Hz (10 -12). ChR was also used to control neuronal activity in living animals, such as chicken embryos (11), Caenorhabditis elegans (13,14), Drosophila (15), and mouse (16). In all cases, action potentials could be evoked by application of blue light to cells that were naturally light insensitive. The unique properties of ChR opened a new field of optogenetic technology for neuroscience and medical applications (17)(18)(19). However, despite the significant efforts that were made to make ChR available for such goals (20), the structure/function relationships of ChR, including the photoreactions as well as the ion channeling mechanism, remain unknown.
Here we describe the cloning and functional characterization of a ChR from the colonial alga Volvox carteri (VChR; see Fig.  1A). First, VChR was expressed in Xenopus oocytes and confirmed as a light-gated cation channel. Subsequently, it was expressed in COS-1 cells and functionally purified in detergent. This allowed us to obtain time-resolved UV-visible spectroscopic data and gain insight into the spectral properties and photocycle of this protein class. We identified two dark states and three photocycle intermediates with pH-dependent variations. Our studies on the transitions between the different states under various pH and light conditions are interpreted in a model that unifies the photocurrent kinetics and spectroscopic data.

EXPERIMENTAL PROCEDURES
Isolation and Cloning of the VChop DNA-Nucleotide sequences homologous to channelopsins Chop1 and Chop2 from Chlamydomonas reinhardtii were searched in a preliminary genome data base of V. carteri that is available within the Chlamydomonas genome project (genome.jgi-psf.org). Chop gene products are opsin apoproteins that form ChRs when reconstituted with the chromophore all-trans-retinal. Using the sequence information from the BLAST search, a 1797-bp fragment of the VChop cDNA was generated by reverse transcription-PCR from RNA isolated from a young V. carteri culture. See the supplemental data for details of cloning and sequence information on the resulting VChop cDNA fragment (GenBank TM accession number DQ094781).
Electrophysiology-Two-electrode voltage clamp measurements on Xenopus oocytes were performed 3-7 days after injection at 22°C under the conditions described before (21,22). For high resolution experiments a Turbo Tec-05X amplifier (NPI Electronic, Tamm, Germany) was used. The data obtained were normally the averages of three measurements. The resistances of microelectrodes were 0.5-1.5 M⍀. For recording of fast kinetics, 10-ns flashes (400 -600 nm) from a Rainbow OPO (OPOTEK, Carlsbad, CA) pumped by the third harmonic of a Brilliant b Nd-YAG-Laser (Quantel, Les Ulis Cedex, France) were applied via a 1-mm light guide. The current amplifier Tec-05X was compensated so that the voltage change was kept below 1 mV at a half-saturating laser flash. The data were recorded at high gain with a sampling rate of 250 kHz. The laser intensity varied from flash to flash within a range of 5%. Bath solution was 100 mM NaCl, 0.1 mM CaCl 2 , 0.1 mM MgCl 2 with 5 mM MOPS-NaOH, pH 7.5, or 5 mM citrate, pH 4.0. The conditions for light pulse experiments were 500 Ϯ 5 nm, 2 ϫ 10 22 photons m Ϫ2 s Ϫ1 .
Spectroscopy-For absorption spectroscopy and slow kinetics at 20°C, a Cary 50 Bio spectrophotometer (Varian Inc.) with a spectral resolution of 2 nm was used. The samples were illuminated for 10 s with a blue Luxeon LED (Philips Lumileds, San Jose, CA) with a wavelength of 456 nm, 90 milliwatt cm Ϫ2 , 2 ϫ 10 21 photons m Ϫ2 s Ϫ1 . Transient spectroscopy was performed on a LKS.60 flash photolysis system (Applied Photophysics Ltd., Leatherhead, UK) at 22°C. Excitation pulses (10 ns, 450 nm) were provided by a tunable Rainbow OPO/Nd-YAG-Laser laser system (see above). Laser energy was adjusted to 3.5 Ϯ 0.16 mJ/shot. The instrument used a Xe-Lamp (150 W) as a monitoring light source, which was pulsed during short time experiments. Monochromators before and after the sample were set to spectral resolutions of 18.6 and 9.3 nm, respectively. For detection, a 1P28 photomultipier (Hamamatsu Photonics) was used, and the signal was recorded with an Infinium Oscilloscope (Agilent Technologies). 32,000 data points were recorded in each measurement and compressed by LKS.60 software to files of 500 data points. 50 of these points were recorded before and 450 after laser excitation. To avoid artifacts and scatter, only data points 60 -500 were used for analysis. Data analysis was performed with Matlab 7.01 (The MathWorks, Natick, MA). Singular value decomposition of representative data sets was performed to identify significant components that were used for reconstruction of the three-dimensional spectra (see supplemental data). For measurements at pH 6, freshly prepared VChop was used. For experiments at pH 7.3 and 4.5, freshly prepared VChop was dialyzed and/or diluted with buffer containing 0.03% dodecyl maltoside, yielding 100 mM NaCl, 0.1 mM CaCl 2 , 0.1 mM MgCl 2 with 10 mM MOPS, pH 7.3, or 10 mM citrate, pH 4.5, and the pH was measured thereafter. pH titration was performed with 20 mM buffer (citrate, pH 4.5-6; MOPS, pH 7; or Tris, pH 8) in 100 mM NaCl, 0.1 mM CaCl 2 , 0.1 mM MgCl 2 , 0.03% dodecyl maltoside.

Functional Studies on VChR in Xenopus Oocytes-In the past,
ChRs from Chlamydomonas were difficult to express in standard expression systems like Escherichia coli, yeast, or COS cells (27). We therefore isolated a ChR-related cDNA from young V. carteri colonies (Fig. 1A). The cDNA encoded a large Volvox-ChR protein, VChR (599 amino acids), with seven transmembrane helices (TM) and a long C-terminal extension. We only studied the 7TM region of VChR (Fig. 1B) corresponding to amino acids 1-307 (see supplemental data) because earlier work on Chlamydomonas ChR1 and ChR2 (5,6) showed that the long C-terminal extension had little influence on the ion conducting properties. In VChR-expressing Xenopus oocytes, light pulses evoked an initial transient current that relaxed toward a stationary plateau (steady state; 1 ϭ 40 -85 ms depending on the extracellular H ϩ concentration (pH o ) and membrane voltage ( Fig. 2A). The current increased with progressing negative voltage and with decreasing pH o (Fig. 2 The current amplitude certainly also depends on the light intensity, but the stationary current saturates at lower light intensities than the transient component ( To further investigate the photocurrent kinetics, cells were stimulated with 10-ns laser flashes of 470-nm light (Fig. 3D). The photocurrent rose with a time constant ( 0 ) in the range of 200 s, and the current peaked at around 1 ms after the flash and relaxed to zero biexponentially with a dominant time constant 2 * (Ͼ90%) of about 16 ms at pH o 7.5 and Ϫ100 mV. Repetitive laser flashes did not significantly promote light adaptation, which we considered as ideal for action spectroscopy. The normalized action spectrum recorded at pH o 7.5 is rhodopsin-shaped with a half-bandwidth of 90 nm and a maximum at 470 nm (Fig. 3E). Interestingly, upon lowering the pH o to 4,

Photoactivation of Channelrhodopsin
the action spectrum was shifted to 484 nm, indicating two different dark states. We further determined that VChR is permeable for H ϩ Ͼ Ͼ Li ϩ Ͼ Na ϩ Ͼ K ϩ (Fig. 3F) and poorly permeable for Ca 2ϩ (Յ10% of the Na ϩ conductance) and is in this respect similar to its C. reinhardtii relative ChR2.
Spectroscopic Characterization of VChR-To get insight into the photocyle and the gating mechanism of VChR, spectroscopic measurements were performed with purified VChR. We used COS-1 cells to express the VChR 7TM region (amino acids 1-307; see supplemental data) plus nine additional amino acids corresponding to the C terminus of bovine rhodopsin (as epitope tag (23,28) for purification and detection by protein immunoblotting). The expressed VChR was reconstituted with all-trans-retinal and purified by immunoaffinity chromatography in dodecyl maltoside solution. Fig. 4 shows the purified protein as a double band of 31 and 34 kDa on a SDS-PAGE (left panel) and after protein immunoblotting (right panel). The 34-kDa band disappeared after treatment with glucosidase (PNGase F), suggesting that the protein was partially glycosylated.
The spectrum of dark-adapted VChR at pH 6 showed an absorption with vibrational fine structure and two maxima at 450 and 470 nm, indicating a rigid chromophore with protonated retinal-protein Schiff base linkage (Fig. 5A). The absorption is sensitive to hydroxylamine, which extracts the retinal by formation of retinaloxime (dotted line) and allows determination of the molar extinction coefficient ⑀ 470 Ϸ 45.500 M⅐cm Ϫ1 (based on ⑀ 360 Ϸ 54.000 M⅐cm Ϫ1 for retinal oxime) (29). Lys-252 in TM7 was identified as the site of retinal attachment, because the K252G mutant did not show any absorption beyond 400 nm (data not shown). Acidification of the medium altered the shape of the VChR spectrum asymmetrically in favor of the long wavelength peak (Fig. 5B). The dark form observed at pH 8 was named D 470 , whereas the acidic form was termed D 480 . When the sample was illuminated at pH 6 with a blue LED (456 nm) for 10 s, the absorption again shifted slightly to the red (Fig. 5, A, red versus black spectrum, and C) in a similar manner as observed after acidification. Again, the absorption slightly declined, but in this case the shift was transient with Ϸ 45 s at 20°C (Fig. 5, C and D).
Next, purified VChR was excited with 10-ns laser flashes (450 nm), and absorption changes were recorded from 300 to 600 nm in 10-nm intervals up to 150 ms after the flash. At pH 7.3, the flashes initiated a fast absorption increase between 480 and 580 nm, which reached 90% of its maximal amplitude already at the first recorded time point (50 ns after the flash). Series of absorption difference spectra at pH 7.3 for the time ranges of 0.3-15 s and 0.033-1.5 ms after the flash are shown in Fig. 6 (A  and B, respectively). The three-dimensional spectra were reconstructed using significant components identified by singular value decompositon (see "Experimental Procedures"). The fastest occurring absorption difference maximum was found at 520 nm (Fig. 6C, Early spectrum), corresponding to an absolute maximum of the photoproduct near 500 nm. Accordingly, this early photocycle intermediate was named P 500 . It converted with Ϸ 4 s into a species termed P 390 (Fig. 6A) seen as an absorption difference peak between 370 and 380 nm in the transient spectrum (Fig. 6C, Medium spectrum). P 390 decayed with Ϸ 200 s into a third photoproduct with an absorption difference maximum at 530 nm (Fig. 6, B and D At pH 4.5, flashes again initiated immediate absorption differences but with an early positive absorption difference peak at 535 nm (Fig. 6E), which is clearly red-shifted compared with the early photoproduct monitored at pH 7.3. The corresponding absorption maximum of the photoproduct is between 510 and 520 nm, and the species was termed P 515 accordingly. The rise time was again below 100 ns. At later times the absorption dif-  ference maximum shifted to Ϸ550 nm. This late species with a corresponding absorption maximum around 530 nm was named P 530 . (Fig. 6E). Unfortunately, the resolution of our detection system and the smaller absorption changes under acidic conditions did not allow observation of the transition between P 515 and P 530 . Because no photoproduct equivalent to P 390 was observed at pH 4.5, we conclude that only neutral conditions allow the formation or transient accumulation of a photoproduct with deprotonated retinal Schiff base during the VChR photocycle.

VChR Is Functionally Similar to Known ChRs from
Chlamydomonas-We found that VChR from V. carteri is a functional relative of the two ChRs from C. reinhardtii with closer relation to ChR2. VChR shows electrical properties that are not explained by an earlier discussed three-state model with one closed dark state (Dark), one open state (O), one desensitized (closed and inactive) state (Des), and a single photochemical reaction from Dark to O (6). If the Des 3 Dark reaction is slow, at high light intensities most protein would be trapped in Des, and the stationary current would decay to zero. If the Des 3 Dark reaction is fast, the model fails to explain the slow recovery of the transient current. In contrast, all observed currents are fully consistent with the four-state model of Fig. 7A originally developed for ChR1 from photocurrent kinetics (21). It comprises two dark states, D 1 and D 2 , and two conducting states, O 1 and O 2 . On the basis of the present study, the model is extended by the pH dependence of the D1^D2 equilibrium of dark-adapted VChR seen as a shift of the action spectrum (Fig. 3E). According to the four-state model, both D 1 and D 2 convert upon light excitation into separate conducting states, O 1 and O 2 , that are also in equilibrium (21). At neutral pH, the O 2 /O 1 ratio is assumed to be higher than the D 2 /D 1 ratio, resulting in an accumulation of the weakly conducting state O 2 in the light and explaining the decay of the transient current to the stationary level. Nikolic et al. (30) quantitatively described the channelrhodopsin photocurrents by an extended threestate model (Fig. 7B) that was already considered earlier qualitatively (6). The "Nikolic model" assumes that the recovery of the Dark state from the (closed and desensitized) Des state is accelerated by light. The model is attractive because it describes the photocurrents adequately and involves less rate constants compared with the four-state model. The Nikolic model, however, does not explain the pH-induced shift of the action spectrum. Based on the electrical measurements and spectroscopic (black circles) is seen as the difference maximum at 520 nm (arrow; Early spectrum), whereas the two later intermediates P 390 and P 510 are seen as difference maxima at 370 and 530 nm, respectively (blue triangles; Medium spectrum). D, absorption differences resulting from B. Although the medium difference spectrum (blue triangles) comprises P 390 and P 510 , the late difference spectrum shows a single photocycle intermediate P 510 only (red circles). E, absorption differences after a laser flash at pH 4.5. The early intermediate P 515 (black circles) is seen as the difference maximum at 535 nm (arrow), whereas the late intermediate P 530 is seen as difference maximum at 550 nm (red circles; arrow). The difference spectra result from two representative data sets: 0.25-59 s after a flash (Early spectrum) and 0.15-7.5 ms (Late spectrum). data generated for VChR, we are proposing here a photocycle (Fig. 7C) that is a hybrid of the four-state model (Fig. 7A) and the three-state Nikolic model (Fig. 7B). Because the acidic form and the preilluminated species (light-adapted VChR) are virtually identical, although there might be slight differences that are still unknown, we assume that D 480 (D 2 ) is an intermediate of the photocycle starting from D 470 (D 1 ). D 1 and D 2 are in a pHdependent equilibrium, and both states are photoconverted into a conducting state O. The model converges with the fourstate model by assuming a fast equilibrium between O 1 and O 2 (representing O) and a preferential thermal reaction from O to D 2 (D 480 ). A back reaction from O to D 1 (D 470 ) is not necessarily excluded but not explicitly needed for explaining the photocurrents. The transient nature of the current is explained in the model shown in Fig. 7C by a low quantum efficiency for the D 2 (D 480 ) to O transition rather than by a second conducting state. The three perspectives (Fig. 7, A-C) are kinetically quite equivalent and describe the photocurrents equally well (results of the calculations are available on demand). The assumption of a less efficient photoconversion of D 2 (D 480 ) is supported by the small light-induced absorbance changes observed at pH 4.5. Moreover, proteorhodopsin, a microbial rhodopsin from the noncultivated marine ␥-protobacterium, shows a red-shifted absorption under moderately acidic conditions, and the photoisomerization is less efficient compared with that of the alkaline form (31). It is known that protonation of the retinal Schiff base counterion in microbial rhodopsins (Glu-118 in VChR) lowers the quantum efficiency of photochemical 13-trans-to cisisomerization because of a rise of the C 13 -C 14 rotation barrier in the excited state (32). For VChR this results in a reduced stationary photocurrent as soon as the D 2 /D 1 equilibrium increases during illumination. The different amplitudes of the transient and stationary current even at saturating light are alternatively explained by a late photocycle intermediate because it occurs in all other microbial rhodopsins. Formation and decay of a corresponding state N would be adjusted in such a way that under all conditions, less than 60% of VChR is trapped in N.
Photocycle of VChR-The UV-visible absorption spectra presented above give a first insight into the absorption properties of the dark states and photointermediates of channelrhodopsin. The pH-dependent spectral equilibrium between D 470 and D 480 suggests protonation of an amino acid residue in the retinalbinding pocket. This residue is accessible from the extracellular bulk phase because the shift can be induced by extracellular acidification in oocyte experiments. At pH 7.3 the dominant photoproduct P 510 is assigned as the conducting state that is relevant under most physiological conditions. The equivalent under acidic conditions is P 530 .
The spectroscopic observations on VChR and most electrophysiological data are summarized by an extended model as shown in Fig. 7D that is compatible with that of Fig. 7C. It comprises two dark states in equilibrium (D 470^D480 ) and corresponding photoproducts. Starting from D 470 , the photoproduct P 510 (considered as the dominant conducting state) is populated sequentially via P 500 and P 390 . Because of the strongly blue-shifted absorption, P 390 is assumed to be the only photoproduct with deprotonated retinal Schiff base. Based on the assumption that P 510 is the conducting state at neutral pH, we conclude that the dark-adapted channel converts from the closed to the open state with Ϸ 200 s after light absorption (light gating). This value is compatible with the photocurrent rise, 0 , of the VChR currents in Xenopus oocytes (Fig. 3D) and in the living alga (2,4). At acidic pH, a species with deprotonated retinal Schiff base (P 390 ) is not observed, and light gating might be faster compared with alkaline pH conditions. The main intermediates P 510 and P 530 decay biphasically ( 2 Ϸ 20 ms and 3 Ϸ 100 ms), indicating that there is a spectrally similar intermediate N.
The D 480 3 D 470 transition functionally reflects the lightdark adaptation process D 2 3 D 1 as recorded by photocurrent measurements, even if the transient photocurrent recovers three times as fast as the slow recovery of D 470 after illumination with a light pulse (Fig. 5, C and D). The difference could derive from experimental conditions. The spectral data were obtained with VChR in solution, whereas photocurrents were recorded from VChR in oocyte membranes at different internal and external pH. The transition from D 2 3 D 1 (and thus D 480 3 D 470 ) is expected to be a two-step process as already proposed before (21) and possibly comprising an isomerization and deprotonation reaction.
The kinetic model presented in this study unifies electrophysiological data with spectral information and should help to interpret photocurrents recorded from intact algae under various physiological conditions. We expect that in algae a shift from D 1 to D 2 at low pH o and/or high light intensities reduces the H ϩ and cation influx and prevents major depolarization (3,4).
Conclusion-The functional purification and spectroscopic characterization of VChR we report here represents a major step toward a detailed structure/function analysis of ChR. It opens the possibility of tailoring ChRs for neuroscience and medical applications. In analogy to the development of green fluorescent proteins with specific absorption/fluorescence properties, desired spectral and kinetic properties of ChR will be obtained by protein engineering and subsequent spectroscopic analysis. This will enable the application of new ChRs with distinct electrical and spectral properties to selected cell types in tissues or living animals to remotely stimulate nerves and to potentially restore color vision in mouse or even human retina (16,20).