Kinetic characteristics of chimeric channelrhodopsins implicate the molecular identity involved in desensitization.

Channelrhodopsin (ChR)-1 and ChR2 were the first-identified members of ChRs which are a growing subfamily of microbial-type rhodopsins. Light absorption drives the generation of a photocurrent in cell membranes expressing ChR2. However, the photocurrent amplitude attenuates and becomes steady-state during prolonged irradiation. This process, called desensitization or inactivation, has been attributed to the accumulation of intermediates less conductive to cations. Here we provided evidence that the dark-adapted (DA) photocurrent before desensitization is kinetically different from the light-adapted (LA) one after desensitization, that is, the deceleration of both basal-to-conductive and conductive-to-basal transitions. When the kinetics were compared between the DA and LA photocurrents for the ChR1/2 chimeras, the transmembrane helices, TM1 and TM2, were the determinants of both basal-to-conductive and conductive-to-basal transitions, whereas TM4 may contribute to the basal-to-conductive transitions and TM5 may contribute to the conductive-to-basal transitions, respectively. The fact that the desensitization-dependent decrease of the basal-to-conductive and conductive-to-basal transitions was facilitated by the TM1 exchange from ChR2 to ChR1 and reversed by the further TM2 exchange suggests that the conformation change for the channel gating is predominantly regulated by the interaction between TM1 and TM2. Although the exchange of TM1 from ChR2 to ChR1 showed no obvious influence on the spectral sensitivity, this exchange significantly induced the desensitization-dependent blue shift. Therefore, the TM1 and 2 are the main structures involved in two features of the desensitization, the stabilization of protein conformation and the charge distribution around the retinal-Schiff base (RSB+).

Channelrhodopsin (ChR)-1 and ChR2 are the first-identified members of ChRs which are a growing subfamily of microbial-type rhodopsins [1-4]. Each molecule consists of seven transmembrane helices (TM1–7) with a covalently bound retinal as a chromophore [5-7]. In the case of ChR2, the basal state with all-trans retinal (D480) is non-conductive to any ions. Light absorption is followed by the photoisomerization of the all-trans retinal to the 13-cis configuration and drives cyclic conformational changes of the molecule, called a photocycle, which consists of several intermediates such as P520, an intermediate conductive to cations. Consequently, very rapid (in the orders of ms) generation of a photocurrent is induced in cell membranes expressing ChR2. However, the photocurrent amplitude attenuates in the order of 10 ms and becomes a steady-state during prolonged irradiation [2,8,9]. This transition, which is termed desensitization or inactivation, has been attributed to the accumulation of intermediates less conductive to cations [6,7,10].

C1C2, one of the ChR1/2 chimeras, which consists of the N-terminal five domains of ChR1, each of which has a single TM, and the C-terminal counterpart domains of ChR2, is the first and only ChR the detailed structure of which has been crystallographically investigated [5]. However, the molecular dynamics involved in the desensitization has not been revealed since the molecule in a desensitized condition is hard to be crystalized. On the other hand, desensitized and non-desensitized photocycles can be differentiated by their photocurrent kinetics. In the present study, we compared the kinetics and spectral sensitivity between the desensitized and non-desensitized photocurrents of the ChR1/2 chimeras. The results suggest that the translocation of TM1 is involved in the kinetics and spectral change of ChR during desensitization.

Materials and Methods

Plasmids

The cDNAs encoding channelrhodopsin (ChR)-1 (ChR1, Met1–Glu345) and -2 (ChR2, Met1–Lys315) with 5′-EcoRI and 3′-BamHI restriction sites were prepared by conventional PCR and subcloned in-frame into pVenus-N1 [8]. Chimeric ChRs between ChR1 and ChR2 were prepared by overlap extension PCR and subcloned into pVenus-N1 [11,12], where the amino acid sequences of the ChRs were divided into seven domains so that each domain practically contained a single TM (Supplementary Fig. S1). These segments are referred to (from N-terminal to C-terminal) as “A,” “B,” “C,” “D,” “E,” “F,” and “G” for ChR1. The homologous counterparts of ChR2 are referred to as “a,” “b,” “c,” “d,” “e,” “f,” and “g.” The N-terminal domain of ChR2 was replaced in order with the corresponding counterpart of ChR1 and we prepared 6 chimeras; ChR-Abcdefg from domain “A” (Met1–Thr117) of ChR1 and domain “b”–”g” (Cys79–Lys315) of ChR2, ChR-ABcdefg from domain “A” and “B” (Met1–Leu164) of ChR1 and domain “c”–”g” (Leu126–Lys315) of ChR2, ChR-ABCdefg from domain “A”–”C” (Met1–Tyr184) of ChR1 and domain “d”–”g” (Ser146–Lys315) of ChR2, ChR-ABCDefg from domain “A”–”D” (Met1–Val212) of ChR1 and domain “e”–”g” (Lys174–Lys315) of ChR2, ChR-ABCDEfg from domain “A”–”E” (Met1–Val242) of ChR1 and domain “f “ and “g” (Pro203–Lys315) of ChR2, ChR-ABCDEFg from domain “A”–”F” (Met1–Phe269) of ChR1 and domain “g” (Ile231–Lys315) of ChR2. All constructs were verified by sequencing.

Cell culture

The electrophysiological assays of the ChRs were made using ND 7/23 cells, hybrid cell lines derived from neonatal rat dorsal root ganglia neurons fused with mouse neuroblastoma [13]. ND 7/23 cells were grown on poly-L-lysine (Sigma-Aldrich, St Louis, MO)-coated coverslip in Dulbecco’s modified Eagle’s medium (DMEM, Wako Pure Chemical Industries, Osaka, Japan) supplemented with 10% fetal bovine serum (Biological Industries, Kibbutz Beit-Haemek, Israel) under a 5% CO2 atmosphere at 37°C. The cells were maintained for no more than ten passages and grown to 80%–90% confluence in the culture dish. The expression plasmids were transiently transfected in ND 7/23 cells using Effectene Transfection Reagent (Qiagen, Tokyo, Japan) according to the manufacturer’s instructions. The medium was replaced with one supplemented with 2.5 μM all-trans retinal at 6 h after transfection. Electrophysiological recordings were then conducted 24–48 h after the transfection. Successfully transfected cells were identified by the presence of Venus fluorescence.

Electrophysiology

All experiments were carried out at room temperature (23±2°C). Photocurrents were recorded using an EPC-8 amplifier (HEKA Electronic, Lambrecht, Germany) under a whole-cell patch clamp configuration. The data were filtered at 0.7 kHz, sampled at 100 kHz (Digidata1440 A/D, Molecular Devices Co., Sunnyvale, CA) and stored in a computer (pClamp10.3, Molecular Devices). The pipette resistance was adjusted to be 2–5 MΩ (3.5±0.1, n=72) with a series resistance of 2.4–15 MΩ (7.6±0.3, n=72) and a cell capacitance of 17–56 pF (35±1, n=72). The series resistance was electrically compensated by 50%. As a result, the charging time constant was 130±6 μs (range, 33–302, n=72) after compensation.

The internal pipette solution for the whole-cell voltage clamp recordings from the ND 7/23 cells contained (in mM) 120 CsOH, 100 glutamate, 5 EGTA, 50 HEPES, 2.5 MgCl2, 2.5 MgATP, 0.1 Leupeptin and 0.01 Alexa 568, adjusted to pH 7.3 with CsOH. The standard extracellular Tyrode’s solution contained (in mM): 138 NaCl, 3 KCl, 2.5 CaCl2, 1.25 MgCl2, 10 HEPES, 4 NaOH, and 11 glucose (pH 7.4 adjusted with HCl).

Using software (pClamp10.3, Molecular Devices), the turning-on (ON) kinetics of each photocurrent was fitted by a single-exponential function for the transition phase between 10% and the peak of the maximal amplitude of photocurrent during irradiation. For the tuning-off (OFF) kinetics, the photocurrent transient after irradiation was fitted by a single-exponential function for the transition between 90 and 10% of the amplitude at the end of irradiation. Generally, no obvious deviation was observed between the raw data and the fitted curve (Supplementary Fig. S2). However, samples were not included in the statistics if their transients were deviated from the single exponential function.

Optics

To investigate the photocurrent kinetics, irradiation was carried out using a SpectraX light engine (Lumencor Inc., Beaverton, OR) controlled by computer software (pCLAMP 10.3, Molecular Devices) at wavelengths (nm, >90% of the maximum): 438±24 (blue), 475±28 (cyan) and 513±17 (teal). The power of the light was directly measured under a microscope by using a visible light-sensing thermopile (MIR-101Q, SSC Co., Ltd., Kuwana City, Japan). Every photocurrent was measured with a holding potential of −60 mV and at pH 7.4 outside. Two test pulses of light (20 ms and 100 ms) were applied before and after irradiation of 1-s strong light (blue: 438±24 nm, 5.1 mWmm−2 or cyan: 475±28 nm, 5.1 mWmm−2). Each irradiation protocol was applied every 60 s to enable full recovery from the desensitization.

Statistical analysis

All data in the text, figures and tables are expressed as mean±SEM and were evaluated with the Wilcoxon signed rank test for statistical significance for paired data and with the Mann-Whitney U-test for unpaired data unless otherwise noted. It was judged as statistically insignificant when P>0.05.

Results

Desensitization-dependent changes of ChR2 kinetics

When ChR2 was expressed in ND 7/23 cells, the light evoked a photocurrent with a transient peak and a steady-state plateau (Fig. 1A). To distinguish the photocurrent kinetics of the desensitized photocycle from the non-desensitized one, two test pulses of light (20 ms and 100 ms) were applied to measure the photocurrents immediately before and after irradiation of 1-s strong light (blue: 438±24 nm, 5.1 mWmm−2 or cyan: 475±28 nm, 5.1 mWmm−2) which completely desensitized the photocurrent. The test pulses of various colors of light were applied at various intensities: blue: 438±24 nm (0.58, 1.1, 1.5 and 1.9 mWmm−2), cyan: 475±28 nm (0.35, 0.72, 1.1 and 1.4 mWmm−2), teal: 513±17 nm (0.47, 0.81, 1.1 and 1.3 mWmm−2). The ON and OFF transients of the first test pulse of light (20 ms, Fig. 1B) should be largely dependent on the photocycle kinetics of the non-desensitized ChR because of the relatively slow process of desensitization (time constant >20 ms) [14]. However, the fraction of desensitized ChR should increase over time during the test pulse. In the present paper, the ON and OFF time constants of the first test pulse of light (20 ms) are respectively referred to as the dark-adapted ON time constant, τON(DA), and the dark-adapted OFF time constant, τOFF(DA), to distinguish them from the genuine kinetic parameters of non-desensitized ChR. On the other hand, the ON and OFF transient of the second test pulse of light (100 ms, Fig. 1C) was mostly dependent on the photocycle kinetics of the desensitized ChR because of the relatively slow recovery from desensitization (time constant >10 s) [2,8,9]. Even so, the fraction of non-desensitized ChR2 should increase over time during the gap of 250 ms before the second test. In the present paper, the ON and OFF time constants of the second test pulse of light (100 ms) are respectively referred to as the light-adapted ON time constant, τON(LA), and the light-adapted OFF time constant, τOFF(LA), to distinguish them from the genuine kinetic parameters of the desensitized ChR.

Figure 1

The differentiation of the dark-adapted (DA) and light-adapted (LA) kinetics of channelrhodopsin-2 (ChR2) photocurrent. (A) Sample trace of ChR2 photocurrent (bottom trace) evoked by two test pulses (upper blue trace, 20 and 100 ms) of light with blue (438±24 nm, 0.58 mWmm−2) before and after irradiation of 1-s cyan light (middle cyan trace, 475±28 nm, 5.1 mWmm−2), which completely desensitized the photocurrent. (B) The DA photocurrent with blue of variable irradiance (0.58, 1.1, 1.5 and 1.9 mWmm−2). (C) The LA photocurrent with blue of variable irradiance (0.58, 1.1, 1.5 and 1.9 mWmm−2).

As shown in Figure 1B and C, the photocurrents peaked earlier with the increase of irradiance (power of light). Actually, each ON rate constant of the DA and LA photocurrents, τON−1(DA) and τON−1(LA), experimentally followed a linear function of the irradiance (L) (Fig. 2A). That is,

Figure 2

Desensitization-dependent changes of the ON kinetics. (A)–(C) The relationship between the ON rate constant (τON−1) and the irradiance of the DA (blue symbols) and LA (red symbols) photocurrents with blue (A, n=10), cyan (B, n=10) and teal (C, n=10). (D) The comparison of the slopes to the irradiance (αDA and αLA) between DA and LA photocurrents with blue (n=10), cyan (n=10) and teal (n=10). (E) The comparison of the light-independent constant (βDA and βLA) between DA and LA photocurrents with blue (n=10), cyan (n=10) and teal (n=10). Wilcoxon signed rank test: *1 (P<0.05), *2 (P<0.005) and Mann-Whitney U-test: †1 (P<0.05), †2 (P<0.005), †4 (P<0.00005).

and

where αDA and αLA is the slope to the irradiance of the ON rate constant of the DA and LA photocurrents, respectively, and βDA and βLA is the light-independent component of the ON rate constant of the DA and LA photocurrents, respectively. The above relationship was different between the DA and LA photocurrents at any wavelength. In the response to blue light, αDA was significantly larger than αLA (P<0.005, n=10), whereas the difference between βDA and βLA was not significant (Fig. 2D and E). In response to cyan light, αDA was significantly larger than αLA (P<0.005, n=10), whereas the difference between βDA and βLA was not significant (Fig. 2D and E). In response to teal light, the difference between αDA and αLA was not significant whereas βDA was significantly larger than βLA (P<0.05, n=10, Fig. 2D and E).

On the other hand, the OFF rate constants of both DA and LA, τOFF−1(DA) and τOFF−1(LA), were almost independent on the irradiance at all wavelengths (Fig. 3A–C). Although they seemed to be somewhat negatively related to the teal irradiance, the differences were not statistically significant (n=10, Kruskal-Wallis test of ANOVA). This negative relationship may be attributed to the increasing contribution of desensitization during test pulse of 20 or 100 ms as the rate of desensitization is dependent on the irradiance. Thus the extrapolated values to 0 irradiance were used as the τOFF−1(DA) and τOFF−1(LA) estimates in the following experiments. As shown in Figure 3D, τOFF−1(LA) was significantly larger than τOFF−1(DA) at blue (P<0.005, n=10) and cyan (P<0.005, n=10).

Figure 3

Desensitization-dependent changes of the OFF kinetics. (A)–(C) The relationship between the OFF rate constant (τOFF−1) and the irradiance of the DA (blue symbols) and LA (red symbols) photocurrents with blue (A, n=10), cyan (B, n=10) and teal (C, n=10). (D) The comparison of τOFF−1 between DA and LA photocurrents with blue (n=10), cyan (n=10) and teal (n=10). Wilcoxon signed rank test: *1 (P<0.05), *2 (P<0.005).

Comparison of ON kinetics among ChR1/2 chimeras

The above kinetic differences between the DA and LA photocurrents should be attributed to the structural changes in the ChR2 protein moiety by the desensitization. To identify them, we divided the amino acid sequences of the ChR2 into seven domains “a”–”g” so that each one practically contained a single TM, and measured the kinetic parameters of the photocurrents for each ChR1/2 chimera in which some of the domains of ChR2 were replaced by their counterparts from ChR1, domains “A”–”G” (Supplementary Fig. S1) producing five chimeras, ChR-Abcdefg, ChR-ABcdefg, ChR-ABCdefg, ChR-ABCDefg and ChR-ABCDEfg [11]. The photocurrent could not be analyzed for ChR1 and ChR-ABCDEFg because of the low amplitude and signal/noise ratio. For each of the chimeras, the ON rate constants, τON−1(DA) and τON−1(LA), followed a linear function of L, whereas the OFF rate constants, τOFF−1(DA) and τOFF−1(LA), were not dependent on L. The slopes to the irradiance, αDA, αLA were varied among the chimeras. As shown in Figure 4A, a significant decrease in αDA was induced by exchanging the domain “a” of ChR2 with the counterpart of ChR1 (domain “A”) at all wavelengths. However, it was significantly increased by addition of the “b”-to-”B” exchange with blue and teal. Similarly the “d”-to-”D” exchange significantly decreased αDA at all wavelengths, whereas the “c”-to-”C” and “e”-to-”E” did not. The differences between αDA and αLA are expressed by the αLA/αDA ratio as shown in Figure 4B. For all chimeras, the αLA/αDA ratio was generally significantly smaller than 1.0 at all wavelengths. However, the change was insignificant for ChR2 with teal and for ChR-ABCDefg with blue and cyan. To test change in the spectral sensitivity by the desensitization, the ratio of αDA (or αLA) with blue over that with cyan (B/C ratio) and the ratio of αDA (or αLA) with teal over that with cyan (T/C ratio) were compared between the DA and LA photocurrents (Fig. 5). Although neither of the ratios of B/C and T/C was significantly changed for ChR2, a significant reduction of the T/C ratio was observed by the light adaptation for the other chimeras (ChR-Abcdefg, -ABcdefg, -ABCdefg, -ABCDefg and -ABCDEfg). The B/C ratio was also significantly reduced for ChR-ABcdefg.

Figure 4

Comparison of αDA and αLA among (from left to right) ChR2 and ChR1/2 chimeras. (A) αDA of ChR2 (blue, n=10; cyan, n=10; teal, n=10), ChR-Abcdefg (blue, n=12; cyan, n=11; teal, n=12), ChR-ABcdefg (blue, n=10; cyan, n=9; teal, n=11), ChR-ABCdefg (blue, n=9; cyan, n=10; teal, n=9), ChR-ABCDefg (blue, n=11; cyan, n=11; teal, n=10) and ChR-ABCDEfg (blue, n=11; cyan, n=11; teal, n=11). (B) The αLA/αDA ratio. Wilcoxon signed rank test between DA and LA photocurrents: *1 (P<0.05), *2 (P<0.005), *3 (P<0.0005). Mann-Whitney U-test between neighbors: †1 (P<0.05), †2 (P<0.005), †4 (P<0.00005) and between ChR2 and ChR1/2 chimeras: ‡2 (P<0.005), ‡4 (P<0.00005).

Figure 5

Comparison of the spectral sensitivity among (from left to right) ChR2 (n=10), ChR-Abcdefg (n=9), ChR-ABcdefg (n=9), ChR-ABCdefg (n=9), ChR-ABCDefg (n=9) and ChR-ABCDEfg (n=9). (A) The ratio of αDA (dark blue columns) or αLA (light blue columns) with blue compared to that with cyan (B/C ratio). (B) The ratio of αDA (dark green columns) or αLA (light green columns) with teal compared to that with cyan (T/C ratio). Wilcoxon signed rank test between DA and LA photocurrents: *1 (P<0.05), *2 (P<0.005). Mann-Whitney U-test between neighbors: †1 (P<0.05), †2 (P<0.005) and between ChR2 and ChR1/2 chimeras: ‡1 (P<0.05), ‡3 (P<0.0005), ‡4 (P<0.00005).

Next, the light-independent components, βDA and βLA, were compared among the chimeras. As shown in Figure 6A, the “a”-to-”A” exchange of ChR2 significantly decreased βDA, whereas the additional “b”-to-”B” exchange significantly increased it at all wavelengths. Similarly, the “c”-to-”C” and “e”-to-”E” exchanges significantly decreased it. The effect of the “d”-to-”D” exchange was significant only with blue. The difference between βDA and βLA are expressed by the βLA/βDA ratio, as shown in Figure 6B. For all the chimeras, the βLA/βDA ratio was generally significantly smaller than 1.0 at all wavelengths. However, the change was insignificant for ChR2 with blue and cyan and for ChR-ABCDefg and ChR-ABCDEfg with teal.

Figure 6

Comparison of βDA and βLA among (from left to right) ChR2 and ChR1/2 chimeras. (A) βDA of ChR2 (blue, n=10; cyan, n=10; teal, n=10), ChR-Abcdefg (blue, n=12; cyan, n=11; teal, n=12), ChR-ABcdefg (blue, n=10; cyan, n=9; teal, n=11), ChR-ABCdefg (blue, n=9; cyan, n=10; teal, n=9), ChR-ABCDefg (blue, n=11; cyan, n=11; teal, n=10) and ChR-ABCDEfg (blue, n=11; cyan, n=11; teal, n=11). (B) The βLA/βDA ratio. Wilcoxon signed rank test between DA and LA photocurrents: *1 (P<0.05), *2 (P<0.005). Mann-Whitney U-test between neighbors: †1 (P<0.05), †2 (P<0.005), †3 (P<0.0005), †4 (P<0.00005), †5 (P<0.000005) and between ChR2 and ChR1/2 chimeras: ‡1 (P<0.05), ‡4 (P<0.00005).

Comparison of OFF kinetics among ChR1/2 chimeras

The OFF rate constants (τOFF−1) varied among the chimeras but were dependent on neither the wavelength nor irradiance (Fig. 7A). Particularly, it was significantly decreased by the “a”-to-”A” and the “e”-to-”E” exchanges and significantly increased by the “b”-to-”B” at all wavelengths. The decreasing effects of “c”-to-”C” exchange and the increasing effect of “d”-to-”D” exchanges were also significant with blue and cyan. The differences between τOFF−1(DA) and τOFF−1(LA) are expressed by the τOFF−1(LA)/τOFF−1(DA) ratio as shown in Figure 7B. As noted previously (Fig. 3D), τOFF−1(LA) was significantly larger than τOFF−1(DA) with blue and cyan for ChR2. On the other hand, the τOFF−1(LA)/τOFF−1(DA) ratio was significantly less than 1.0 for ChR-Abcdefg, -ABcdefg and -ABCdefg at all wavelengths as well as for ChR-ABCDefg with cyan. The effects of desensitization were almost negligible for ChR-ABCDefg with blue and teal and for -ABCDEfg at all wavelengths.

Figure 7

Comparison of the OFF rate constants, τOFF−1(DA) and τOFF−1(LA) among (from left to right) ChR2 and ChR1/2 chimeras. (A) τOFF−1(DA) of ChR2 (blue, n=10; cyan, n=10; teal, n=10), ChR-Abcdefg (blue, n=12; cyan, n=11; teal, n=12), ChR-ABcdefg (blue, n=10; cyan, n=9; teal, n=11), ChR-ABCdefg (blue, n=9; cyan, n=10; teal, n=9), ChR-ABCDefg (blue, n=11; cyan, n=11; teal, n=10) and ChR-ABCDEfg (blue, n=11; cyan, n=11; teal, n=11). (B) The tOFF−1(LA)/tOFF−1(DA) ratio. Wilcoxon signed rank test between DA and LA photocurrents: *1 (P<0.05), *2 (P<0.005), *3 (P<0.0005). Mann-Whitney U-test between neighbors: †1 (P<0.05), †2 (P<0.005), †3 (P<0.0005), †4 (P<0.00005), †5 (P<0.000005) and between ChR2 and ChR1/2 chimeras: ‡1 (P<0.05), ‡4 (P<0.00005).

Although τOFF−1(DA) was significantly smaller than βDA with blue for ChR2 (P<0.05), ChR-Abcdefg (P<0.0005), ChR-ABcdefg (P<0.005), ChR-ABCdefg (P<0.005), ChR-ABCDefg (P<0.005) and ChR-ABCDEfg (P<0.005), the averaged values showed high correlation (r=0.99), as shown in Figure 8A. This relationship was similar with cyan (r=0.99). The averaged values were almost coincident with teal (r=0.99) and with insignificant differences except for ChR-ABCDefg, in which τOFF−1(DA) was significantly larger than βDA (P<0.005). A similar correlation was observed between τOFF−1(LA) and βLA with blue (r=1.00), cyan (r=1.00) and teal (r=0.98) (Fig. 8D–F). Although τOFF−1(LA) was again smaller than βLA for ChR-ABCDEfg (P<0.05) with blue and for ChR-ABcdefg (P<0.05) and -ABCdefg (P<0.05) with cyan, both values were almost coincident witht teal except for ChR2 (P<0.05).

Figure 8

Correlation between τOFF−1 and β. Each symbol is the mean±SEM for ChR2 (blue, n=10; cyan, n=10; teal, n=10), ChR-Abcdefg (blue, n=12; cyan, n=11; teal, n=12), ChR-ABcdefg (blue, n=10; cyan, n=9; teal, n=11), ChR-ABCdefg (blue, n=9; cyan, n=10; teal, n=9), ChR-ABCDefg (blue, n=11; cyan, n=11; teal, n=10) and ChR-ABCDEfg (blue, n=11; cyan, n=11; teal, n=11). (A)–(C) Relationship between τOFF−1(DA) and βDA with blue, cyan and teal, respectively. (D)–(F) Relationship between τOFF−1(LA) and βLA with blue, cyan and teal, respectively.

Discussion

Previously, multiple-photocycle models were proposed to account for the photocurrent kinetics and spectroscopic transitions of various ChRs [7,10,14,15-20]. That is, there should be at least two ground states (D1, D2) and two open states (O1, O2) [6,7,16]. When a flash of light is applied to a ChR-expressing cell under dark adaptation, some molecules enter the cation-conducting state (O1) to generate a photocurrent under whole-cell voltage clamp of the cell (Fig. 9). As the O1 is relatively unstable, the molecules become non-conductive with a certain probability and reactivated again through D1. However, some of them have different conformations of D2 and, with a certain probability, go into the independent desensitized photocycle with a different cation-conducting state (O2). With a smaller average conductance for O2 than O1, the photocurrent is progressively attenuated in amplitude. Finally, most of the molecules enter the desensitized photocycle with a steady-state plateau photocurrent as the recovery rate from D2 to D1 is relatively slow [2,8,9]. The present study provides the additional evidence of this multiple-photocycle model, that is, the D1-O1-D1 photocycle and the D2-O2-D2 photocycle are kinetically distinct. The τON−1(DA) of ChR2 was consistently larger than the τON−1(LA) for the same irradiance at a given wavelength. On the other hand, the τOFF−1(DA) of ChR2 was consistently smaller than the corresponding τOFF−1(LA).

Figure 9

Two-photocycle model of channelrhodopsin. The non-desensitized photocycle starts from the dark-adapted basal state (D1) to the open/cation-conducting state (O1) through intermediates (P 500 and P 390) with a relatively short dwelling time. The desensitized photocycle is similar, but starts from the light-adapted basal state (D2) to the open/cation-conducting state (O2) with a different conductance. The transition from D2 to D1 is relatively slow compared to the photocycle period.

In the present study, it was experimentally demonstrated that the turning-on rate (τON−1) of the DA/LA photocurrent is linearly related to a relatively wide range of irradiance (L), as previously noted [21-23]. This is consistent with the two-state model prediction that the transition from basal (D1/D2) to conductive states (O1/O2) is approximated by a single-photon reaction and the transition from conductive (O1/O2) to basal states (D1/D2) is a light-independent thermal reaction. That is,

and

where the constants ɛ1 and ɛ2 are respectively the molar absorption coefficient equivalents of D1 and D2, and are determinants of the spectral sensitivity of each state [24]. The constants ϕ1 and ϕ2 are respectively the quantum yield equivalents of D1 and D2. The ϕ1 is proportional to the probability of a molecule to change its conformation from D1 to O1, whereas the ϕ2 is proportional to the probability of a state transition from D2 to O2. The constants β1 and β2 are independent of L, but are dependent on the probability of the molecule to return from conductive (O1 and O2) to basal states (D1 and D2). In the present study, each steepness, ɛ1ϕ1 and ɛ2ϕ2, was experimentally approximated by αDA and αLA, and each constant, β1 and β2, was approximated either by βDA and βLA or by τOFF−1(DA) and τOFF−1(LA), which were directly measured from the photocurrents. These values would give us some insight into differences in the molecular dynamics between non-desensitized and desensitized photocycles and would also be key parameters for predicting the photocurrent kinetics of ChR as a function of irradiance (L) and time.

Deceleration of ChR2 kinetics by desensitization

In the case of ChR2, the αLA was smaller than αDA with blue and cyan (Figs. 2D and 4B) and without changes in the B/C and T/C ratios (Fig. 5A and 5B). Therefore, it could be predicted that ϕ2 is smaller than ϕ1 whereas ɛ1 and ɛ2 are similar. The insignificant difference in αLA and αDA with teal suggest that the difference between ϕ1 and ϕ2 may be cancelled by the small difference between ɛ1 and ɛ2. Similarly, βLA tended to be smaller than βDA, but the difference was significant only with teal (Figs. 2E and 6B). This could be attributed to the fact that βDA overestimated β1 because of the relatively fast rate of desensitization with blue and cyan. Therefore, it is suggested that both probabilities from the basal to conductive states and from the conductive to basal states are decreased by the desensitization. However, this is somewhat conflicts with the fact that τOFF−1(LA) of ChR2 was larger than τOFF−1(DA) with blue and cyan (Fig. 3D) as τOFF−1(LA) should be proportional to the probability of a channel to close.

Evaluation of the TM exchanges

We found that ChR2 and ChR1/2 chimeras differed even in the kinetics of the DA photocurrents as reported previously [11,12]. These differences could be attributed to structural changes of the opsins. Although the “a”-to-”A” exchange of ChR2 did not affect the spectral sensitivity (Fig. 5A and 5B), it decreased αDA (Fig. 4A), βDA (Fig. 6A) and τOFF−1(DA) (Fig. 7A) at all wavelengths, suggesting that the probabilities of D1-O1 and O1-D1 transition are dependent on the TM1. This domain should also be involved in the desensitization-dependent change of spectral sensitivity (Fig. 5B) as well as the desensitization-dependent reduction of the probabilities of state transitions during a photocycle. Particularly, the deceleration of the O2-D2 transition became manifest with the decrease of βLA/βDA (Fig. 6B) and τOFF−1(LA)/τOFF−1(DA) (Fig. 7B). The further “b”-to-”B” exchange increased αDA (Fig. 4A), βDA (Fig. 6A) and τOFF−1(DA) (Fig. 7A), suggesting that the probabilities of the D1-O1 and O1-D1 transition are also dependent on the TM2. Although the magnitude was decreased, the reduction of the state transition probability remained (Figs. 4B, 6B and 7B). The spectral sensitivity appeared to be red-shifted further by this exchange (Fig. 5B) with a significant reduction of the B/C ratio by desensitization (Fig. 5A). On the other hand, the effects of the “c”-to-”C” exchange were negligible for αDA (Fig. 4A) although its desensitization-dependent change remained (Fig. 4B). However, βDA and τOFF −1(DA) were significantly reduced (Figs. 6A and 7A), the T/C ratio was reduced in the DA photocurrent (Fig. 5B) and the B/C ratio became insensitive to desensitization again (Fig. 5A) even though a single amino acid, Ser181, in “c” was Ala in “C”. The effects of the “d”-to-”D” exchange were also considerably manifest particularly for αDA (Fig. 4A), with the enhanced T/C ratio at DA photocurrent (Fig. 5B) although TM4 is highly conserved and only 7 amino acids are different. In this chimera, ChR-ABCDefg, the effects of desensitization were smaller than in the others; the difference between αDA and αLA was negligible with blue and cyan (Fig. 4B), that between βDA and βLA was negligible with teal (Fig. 6B) and that between τOFF−1(DA) and τOFF−1(LA) was negligible at all wavelengths (Fig. 7B). In contrast, the “e”-to-”E” exchange reduced βDA and τOFF−1(DA) (Figs. 6A and 7A) with a significant red-shift of the DA photocurrent (Fig. 5A and 5B). Although its effect on αDA was insignificant at any wavelength (Fig. 4A), it induced a desensitization-dependent attenuation of the basal-to-conductive state transition (Fig. 4B).

As a general rule common to chimeras, high correlations were present between βDA and τOFF−1(DA) as well as between βLA and τOFF−1(LA) (Fig. 8). Therefore, the simple two-state model (equations (3)–(6)) may approximate the ON/OFF kinetics. However, τOFF−1(DA) was significantly smaller than βDA with blue and cyan. This is probably due to the increasing contribution of desensitization during the first test pulse of light (20 ms) to measure τOFF−1(DA) as the rate of desensitization with blue and cyan was larger than that with teal at the same irradiance. Indeed, τOFF−1(DA) was almost equal to βDA when the test light pulse was short (10 ms) (Supplementary Fig. S3). However, the distinctive trait of ChR-ABCDefg, that τOFF−1(DA) was significantly larger than βDA with teal, and the trait of ChR2, that τOFF−1(LA) was significantly larger than βLA with teal, has to be otherwise explained.

Molecular dynamics of desensitization

In summary, both the D1 to O1 and D2 to O2 state transitions are suggested to be regulated by the interaction between TM1 and TM2 and are decelerated by the heterogeneous combination of “A” and “b”. These transitions are also dependent on the TM4. Similarly, the O1 to D1 and O2 to D2 state transitions are suggested to be regulated by the interaction between TM1 and TM2, although not by TM4. On the other hand these transitions are both dependent on TM5. Probably, TM1 and 2 are involved in the general stabilization of the molecule, whereas the translocation of TM4 may contribute to the stabilization of D1 and D2, and TM5 may contribute the stabilization of O1 and O2. Indeed, Gln95, Thr98 and Ser102 in the TM1 of C1C2, the ChR-ABCDEfg equivalent, and the five Glu (Glu121, Glu122, Glu129, Glu136, Glu140) and Lys132 in TM2 (C1C2) have been suggested to form a hydrophilic channel with the amino acids in TM3 [5,21,22,25]. This notion is consistent with a recent molecular dynamics simulation showing that the movements of TM6, 7 and 2 are induced by the photoisomerization of retinal [26]. These conformational changes of the opsins are presumed to be almost similar for desensitized photocycles although the desensitization may cause the stabilization of both D2 and O2. It is possible that the more destabilized basal state in the non-desensitized photocycle, that is, the fact that ϕ1>ϕ2, is the consequence of the isomerization of all-trans, 15-anti-form of retinal to 13-cis, 15-syn-form in darkness [16] although no significant change in the spectral sensitivity of ChR2 was observed in the present study (Fig. 5).

The spectral sensitivity of a rhodopsin, either animal or microbial type, is influenced by a number of factors. Among them, the interaction between the retinal chromophore and the counterions around it are suggested to be critical [6,27]. That is, the negative charge distribution near the retinal-Schiff base (RSB+) stabilizes the basal state to blue-shift the spectrum. On the other hand, the negative charge distribution near the β-ionone ring red-shifts the spectrum. Although the position of TM1 is remote from the chromophore in C1C2 [5] and indeed its exchange from ChR2 to ChR1 showed no obvious influence on the spectral sensitivity, this exchange significantly induced the desensitization-dependent blue shift. A crystallographic study indicated that TM1 lies closer to the RSB+ than the β-ionone ring in C1C2 [5]. Therefore, the desensitization might move some negative amino acids in “A” such as Glu87, which is neutral Ala in ChR2, towards RSB+ by the translocation/de-protonation or other positive amino acids such as Lys88, which is neutral Gln in ChR2, away from RSB+. Alternatively, the interaction of TM1 of ChR1 and TM2 (ChR1 or 2) may change the charge distribution around RSB+ in desensitized ChR, since a small but significant reduction of the B/C ratio by desensitization was observed only for ChR2-ABcdefg. These two features of desensitization, stabilization of the protein conformation and a change in the charge distribution around RSB+, could be generated by the same structure, such as the H-bond formation between TM1 and other TMs. Indeed, Thr98 in TM1 affects the position of Glu129 in TM2, which should be involved in the pore constriction, through Ser102 (TM1) and Asn297 (TM7) [5]. Although the precise molecular dynamics underlying this transition should be investigated in the future through spectroscopy, X-ray crystallography, site-directed mutagenesis as well as the electrophysiology in combination, a previous Fourier transform infrared (FTIR) spectroscopic study suggested that the gating and desensitizing processes in ChR1/2 chimeras are different from those in ChR2 [28].

Conclusion

The desensitization of ChR was revealed to be accompanied by a deceleration of the state transition and by a spectral shift. TM1 and 2 are the main structures involved in the desensitization-dependent change of ChR.

Channelrhodopsins are desensitized during prolonged irradiation of light. Zamani et al. provided evidence that the desensitized photocycle is kinetically distinct and both rates from basal to conductive and conductive to basal states are decelerated during desensitization. Using channelrhodopsin-1/2 chimeras, they revealed that the transmembrane helices, TM1 and TM2, were the determinants of both basal-to-conductive and conductive-to-basal transitions, whereas TM4 may contribute to the basal-to-conductive and TM5 may contribute to the conductive-to-basal transitions, respectively. TM1 and 2 are the main structures involved in two features of the desensitization, the stabilization of protein conformation and the charge distribution around the retinal-Schiff base (RSB+).