Channelrhodopsin-2 is a leaky proton pump

Abstract

Since its discovery, the light-gated cation channel Channelrhodopsin-2 (ChR2) has proven to be a long-sought tool for the noninvasive, light-activated control of neural cells in culture and in living animals. Although ChR2 is widely used in neurobiological applications, little is known about its molecular mechanism. In this work, the unitary conductance of ChR2 was determined for different cations, for example 40 fS at 200 mM NaCl and -60 mV, using noise analysis. The kinetics of the ion channel obtained by noise analysis is in excellent agreement with the photocurrent kinetics obtained by voltage-clamp and time-resolved spectroscopy. The inward rectification of the channel could be explained by the single channel parameters. ChR2 represents an ion channel with a 7 transmembrane helix motif, even though the sequence homology of its essential amino acids to those of the light-driven H(+) pump bacteriorhodopsin (bR) is high. Here, we also show that when ChR2 is expressed in electrofused giant HEK293 cells or reconstituted on planar lipid membranes, it can indeed act as an outwardly driven H(+) pump, demonstrating that ChR2 is bifunctional, and in-line with other microbial rhodopsins, a H(+) pump but with a leak that shows ion channel properties.

Fig. 1.

Channelrhodopsin-2 currents from a whole-cell patch recording with standard patch solutions (see Methods; 11.5 °C). Currents were induced by a DPSS laser (473 nm). (A) Applied voltage steps reached from −80 mV to + 80 mV in 40 mV steps. (B) Current-voltage relationship of the currents shown in A.

Fig. 2.

Power spectra of whole-cell stationary recordings in standard solutions. Currents were recorded with and without illumination at light intensities that yielded half-maximal signal amplitude (−60 mV; 2 min; 11.5 °C; filtered at 2 kHz; see A and B, Insets). Power spectra of HEK293 cells with induction of ChR2 expression (A) and without (B, control). The red line shows the power spectra during illumination and the black one in the dark. (A and B, Insets) Macroscopic currents with illumination (red) and without illumination (black) are shown corresponding to the power spectra shown in A (ChR2 HEK293) and (B) (control HEK293). (C) Difference spectrum of the spectra shown in A fitted with a single Lorentzian (red line) between 2.5 Hz and 1 kHz. (D) Reduced difference spectrum of control cell spectra shown in B. Red line indicates the Lorentzian fit of C. Please note the logarithmic y axis [S(0) of the Lorentzian fit: 0.15 pA²/Hz; first data point in difference control spectrum: 0.04 pA²/Hz (0.84 Hz)].

Fig. 3.

Dependency of single channel parameters on guanidine⁺ concentration and temperature. (A) Change of the mean single channel conductance with increasing guanidine⁺ concentration. Conductances were calculated from the parameters of Lorentzian functions fitted to the power spectra (over a fixed interval of 2.5 Hz to 1 kHz; see Fig. 2). Approximation by a Michaelis-Menten function reveals a maximal single channel conductance of ChR2 of γ_max = 128.5 ± 8.6 fS (K_m = 82.3 ± 13.9 mM). Error bars indicate the standard deviation of the mean values. (B) Arrhenius plots of the rate constants k (2πf_c; ■; from the Lorentzian fitted to the respective power spectra) and k_off (1/τ_off; Δ; from macroscopic data), and γ (□) are depicted. Lines indicate activation energies of 75 kJ/mol for k, 64 kJ/mol for k_off, and 21 kJ/mol for γ. Whole-cell patches were clamped at −60 mV at 8 °C (n = 4), 11.5 °C (n = 22), 14 °C (n = 4), and 16 °C (n = 4) with standard patch solutions (see Methods).

Fig. 4.

Voltage and light dependency.(A) Voltage dependency of normalized mean (black; n = 25) and single channel (i; red) currents [normalized to the values measured at −40 mV; i: n = 22 (−60 mV); n = 4 (−40 mV); n = 4 (−20 mV); n = 5 (0 mV)]. (B) Light dependency of a single HEK293 cell expressing ChR2. The mean macroscopic current at different light intensities (473 nm; 0.05–5.3 mW mm⁻²; −60 mV applied) is plotted against the variance σ². Each variance is calculated from the parameters of the Lorentzian fit to the respective difference power spectra using σ² = [πf_cS(0)]/2. The data were approximated using σ² = Ii − (I²/n) yielding a single channel current of i = −7.5 ± 0.6 fA and n = 160,000 ChR2 channels. The error bars given for σ² are the combined error of the parameters S(0) and f_c resulting from the fit of the difference power spectra. (A and B) 11.5 °C and external 200 mM Guanidine⁺.

Fig. 5.

Pump current in relation to the photocyte. (A) Pump currents of ChR2: Proteoliposomes adhered to one side of the BLM in 20 mM Hepes/Tris, pH 7.4. The signals were recorded at different conductivities of the compound membrane directly after addition of the protonophore 1799 (0.4 nS, black trace) and after incubation for 1 h (2.8 nS, green trace) upon illumination (>435 nm). The insert shows the action spectrum for the stationary currents compared with the ground state spectrum of purified and solubilized ChR2 (red line). (B) Pump current measured on giant HEK293 cells after electrofusion (d ≈ 30 μm). The pump current was induced by a blue DPSS laser (473 nm, 40 mW mm⁻²). (C) The photocycle, the pump and the channel mode are related in a simplified photocycle under stationary conditions (9, 10). The ground state D470 reaches the early intermediate P500 after illumination. Deprotonation (no.1) to the extracellular (EC) side (P390) and reprotonation (no. 2) from the cytoplasmic (CP) side (P520) leads to the pumping of one proton per photocycle. During the lifetime of P520, the channel is open and allows permeation of protons and cations (M⁺) as indicated by the red arrow. Upon closing (no.3), ChR2 reaches the light-adapted state P480. Recovery to the ground state takes seconds in the dark and is accelerated by a second light reaction under continuous illumination. Therefore, the channel cycle can be lumped together to the 2-state model as indicated in the small circle inside. (D) Cartoon sketch of the mechanistic model for ChR2 in relation to the photocycle showing the retinylidene chromophore. Deprotonation of the Schiff-base (no. 1) takes place via the putative proton acceptor E123. Reprotonation occurs from an intramolecular donor toward (no. 2) that is going to be replenished from the cytoplasmic side (no. 3)