Temporal evolution of helix hydration in a light-gated ion channel correlates with ion conductance

Abstract

The discovery of channelrhodopsins introduced a new class of light-gated ion channels, which when genetically encoded in host cells resulted in the development of optogenetics. Channelrhodopsin-2 from Chlamydomonas reinhardtii, CrChR2, is the most widely used optogenetic tool in neuroscience. To explore the connection between the gating mechanism and the influx and efflux of water molecules in CrChR2, we have integrated light-induced time-resolved infrared spectroscopy and electrophysiology. Cross-correlation analysis revealed that ion conductance tallies with peptide backbone amide I vibrational changes at 1,665(-) and 1,648(+) cm(-1). These two bands report on the hydration of transmembrane α-helices as concluded from vibrational coupling experiments. Lifetime distribution analysis shows that water influx proceeded in two temporally separated steps with time constants of 10 μs (30%) and 200 μs (70%), the latter phase concurrent with the start of ion conductance. Water efflux and the cessation of the ion conductance are synchronized as well, with a time constant of 10 ms. The temporal correlation between ion conductance and hydration of helices holds for fast (E123T) and slow (D156E) variants of CrChR2, strengthening its functional significance.

Keywords: channel gating; channelrhodopsin; infrared spectroscopy; optogenetics; time-resolved spectroscopy.

Fig. 1.

Structural model of channelrhodopsin-2 in the inactive dark-state, based on the C1C2 chimera (21). (A) Intracellular view of the seven transmembrane helices. Helices A, B, C, and G frame the putative ion-conductive pathway, whose activation presumably involves an outward tilt of helix B (see arrow). Helices C, D, E, F, and G enclose the retinal and line the proton-pumping pathway (19). (B) Lateral view of the solvent-accessible surface of C1C2, clipped in half. The solvent-accessible surface is colored by atom type for the transmembrane region [oxygen (red), nitrogen (purple), sulfured (orange), carbon (gray)] and in gray for the intracellular and extracellular areas. Note the presence of two important internal cavities: one is occupied by the retinal (yellow) and the other forms an intruding pore from the extracellular side with E90 (helix B) and N258 (helix G) located at its tip (21). The aromatic side chain of Y70 (helix A) forms an intracellular constriction site (21). Molecular drawings were performed with Ballview (70).

Fig. 2.

Comparison of time-resolved currents and absorption changes of CrChR2 wild type following pulsed (10 ns) light excitation. (A) Photocurrents at various voltages (from −100 to +80 mV in 20-mV steps) for CrChR2 expressed in HEK cells measured by whole-cell electrophysiology at 23 °C. The photocurrents at 0 mV were subtracted to remove/attenuate electrical signals unrelated to passive cation flow (see Fig. S7 for uncorrected photocurrents). The t_max of the currents is indicated by green open circles. (B–E) Color surface with 40 equidistant contour lines of the transient absorption changes of detergent solubilized CrChR2 at 25 °C at various spectral ranges. The vertical black line, at 1 ms, indicates the t_max of the photocurrents. (F) Infrared difference spectrum at 1 ms, colored by the correlation coefficient between the photocurrent at −40 mV and the transient absorption changes. Hotspots of high correlation (red) are predominantly located in the amide I region.

Fig. S1.

Illustration of the correlation procedure used in the manuscript to compare time traces from absorbance changes and photocurrents. Absorbance changes at (A) 1,737 cm⁻¹ (D156), (B) 1,717 cm⁻¹ (E90), (C) 1,698 cm⁻¹ (D253 + other), and (D) 1,648 cm⁻¹ (hydrated helices) are compared with (E) the photocurrents at −40 mV. (F–I) These absorption changes are represented as a function of the photocurrents (black dots) and fitted by linear regression (red line). The correlation coefficient (R²), calculated with the MATLAB function regress, summarizes the similarity of the compared time traces. We accounted for differences in temperature between the photocurrents and the absorption changes (23 °C vs. 25 °C) by rescaling the time of the photocurrents by 0.83. To make the calculation of the correlation possible, the photocurrents and the absorption changes were interpolated to 200 common time values, uniformly spaced in logarithmic scale from 6 μs to 200 ms.

Fig. 3.

Hydration of helices monitored by amide A and I vibrations. (A) Sketch showing the effect of H bonding of water molecules to amide carbonyl groups on the amide I (C=O stretch) and amide A (N–H stretch) frequencies. (B) An amide carbonyl H-bonded to water shows through-bond coupling and through-space coupling (between the dipole moments of the two vibrations). The vibration of the C=O group is different when interacting with H₂¹⁶O and H₂¹⁸O. (C) Steady-state light-induced IR spectra at 2-cm⁻¹ resolution of CrChR2 hydrated with either H₂¹⁶O (blue curve) or H₂¹⁸O (red curve) between 3,800 and 2,700 cm⁻¹ (Left) and 1,825 and 1,500 cm⁻¹ (Right). The slight differences in band position between steady-state (C) and time-resolved (Fig. 2 D and F) difference spectra are due to the higher spectral resolution of the former experiments (Fig. S8). (Inset, Left) Zoom-in of the 3,400- to 3,200-cm⁻¹ region (amide A) after baseline correction and (Inset, Right) zoom-in of the 1,680- to 1,655-cm⁻¹ and 1,655- to 1,625-cm⁻¹ amide I regions. Note how the positive band at 1,650 cm⁻¹ downshifts to 1,647.5 cm⁻¹ in H₂¹⁸O.

Fig. S2.

Steady-state FTIR difference spectra of ChR2-WT at 2 cm⁻¹ resolution hydrated with (A) H₂¹⁶O and (B) H₂¹⁸O. Three replicates are shown in pale blue/red colors, with the average shown in blue/red. Note how band positions are preserved between replicates despite some differences in band intensity.

Fig. S3.

Steady-state light-induced FTIR difference spectrum of ChR2-WT hydrated with D₂O (>24 h incubation). Note how the intensity and position of the amide A bands are preserved in D₂O (compare with Fig. 3 in the MS), indicating that they arise from amide groups in secondary structures fully resistant to H/D exchange.

Fig. 4.

Lifetime distribution analysis of the photocurrents and the absorption changes in the amide I region. (A) Contour plot of the amplitude of the photocurrents to respect the time constant and the voltage. (B) Contour plot of the amplitude of the absorption changes with respect to the time constant and the wavenumber. For better band resolution, the time-resolved spectra were mathematically narrowed by Fourier deconvolution before performing the lifetime distribution analysis (see Fig. S4 for the lifetime distribution without band narrowing). (C) Comparison of the lifetime distribution of the photocurrents at −40 mV and of the absorption changes at 1,664 and 1,647 cm⁻¹.

Fig. S4.

Lifetime distribution of ChR2-WT in the 1,800- to 1,600-cm⁻¹ region. The lifetime distribution was estimated by the maximum entropy method, and bands resolved according to their wavenumber and time constant for rise (blue) and decay (red). Bands from dry (∼1,665 cm⁻¹) and hydrated helices (∼1,648 cm⁻¹) are labeled, as well as bands indicative of protonation changes of D156 (∼1,738 cm⁻¹). Contrary to Fig. 4, the MS spectra were not narrowed by Fourier self-deconvolution before estimating the lifetime distribution.

Fig. 5.

Comparison of time-resolved currents and absorption changes of fast (E123T; Left) and slow (D156E; Right) CrChR2 variants after 10 ns light excitation. (A and C) Photocurrents at various voltages: (A) CrChR2-E123T and (C) CrChR2-D156E. The maximum (t_max) of the currents is indicated by green open circles. (B and D) Color surface of the transient absorption changes in the amide I region: (B) CrChR2-E123T and (D) CrChR2-D156E. (E) Comparison of the lifetime distribution of the photocurrents and the amide I changes for CrChR2-E123T and CrChR2-E156E.

Fig. S5.

IR difference spectrum of ChR2-E123T 450 μs after photoexcitation, color-coded by the correlation coefficient with the photocurrents at −40 mV. Details of the correlation procedure are given in Fig. S1.

Fig. S6.

Lifetime distribution of (A and D) the photocurrents and (C and F) the absorption changes in the 1,700– to 1,600-cm⁻¹ region for (A–C) ChR2-E123T and (D–F) ChR2-D156E. Lifetime distributions were estimated by the maximum entropy method. In C and F, only bands assigned to dry and hydrated helices are labeled. (B and E) Extracted lifetime distribution at a representative voltage and wavenumber values.

Fig. 6.

Schematic representation of the gating steps in channelrhodopsin-2 following light excitation, including water influx and efflux steps, retinal isomerization, and proton transfers involved in proton pumping (and potentially in the gating mechanism). Water influx takes place in two cooperative steps (τ = 10 and 200 μs). Cation permeation starts with τ = 200 μs, when the hydration of the pore is completed, and ends with τ = 10 ms, when half of the pore dewets and collapses. The carboxylic groups above and below the retinal (only a small part of the retinal molecule is shown) correspond to D156 and D253, respectively. X–H it is the terminal proton release group, still unassigned.

Fig. S7.

(Upper) Uncorrected photocurrents for ChR2 wild-type, E123T, and D156E variants at various voltages. (Lower) Photocurrents corrected for the current recorded at 0 mV. The scaling factor used was optimized to remove the peak artifact below 20 μs, and ranged from 0.8 to 1.2 (for D156E, we used a fixed scaling factor of 1).

Fig. S8.

Steady-state FTIR difference spectra of ChR2-WT hydrated with H₂¹⁶O and H₂¹⁸O at (A) 4 and (B) 8 cm⁻¹ instrumental resolution (see Fig. S2 for spectra at 2 cm⁻¹ resolution). Note that as the spectral resolution is decreased, the positive band is downshifted (1,650.0, 1,648.6, and 1,648.2 cm⁻¹ at 2, 4, and 8 cm⁻¹ resolution), reducing the observed downshift in H₂¹⁸O (2.5, 0.8, and 0.5 cm⁻¹ at 2, 4, and 8 cm⁻¹ resolution).