All-optical closed-loop voltage clamp for precise control of muscles and neurons in live animals

The study addresses how to achieve true, closed-loop control of excitable cell membrane voltage using an all-optical approach. Traditional methods such as patch-clamp electrophysiology provide high temporal precision but are invasive and low throughput. Calcium imaging is more compatible with intact organisms but has limited temporal resolution and cannot reliably report inhibition or subthreshold events. Previous optogenetic approaches combined voltage indicators and actuators but largely used static, open-loop stimulation, leading to adaptation and variable responses due to expression differences. The authors propose an optogenetic voltage-clamp (OVC) that integrates a genetically encoded voltage indicator (QuasAr2) for readout with bidirectional actuators (BiPOLES: Chrimson for depolarization and GtACR2 for hyperpolarization) in a closed-loop feedback system to clamp membrane potential in live cells. The purpose is to enable precise, non-invasive, high-throughput control and measurement of cellular electrophysiology in intact organisms and mammalian neurons.

Prior work includes: patch-clamp electrophysiology offering high temporal fidelity but being invasive; calcium imaging enabling in vivo application but limited by buffering, nonlinearity to voltage, and inability to detect inhibition. Rhodopsin-based GEVIs like QuasAr2 provide near-infrared emission and millisecond voltage tracking, allowing multiplexing with optogenetic actuators. The i-Optopatch approach combined CheRiff and QuasAr2 for unidirectional control without closed-loop. An optical dynamic clamp used archaerhodopsin with electrical feedback in cardiomyocytes. Light-induced electrophysiology (LiEp) enabled bidirectional modulation without feedback for drug screening. An optoclamp combined ChR2 and NpHR with extracellular arrays to clamp firing rates in ensembles using indirect voltage readout. These methods lacked true all-optical, single-cell, closed-loop voltage clamping with a GEVI and balanced bidirectional actuation.

The OVC integrates QuasAr2 for voltage readout and BiPOLES (a tandem Chrimson-GtACR2 actuator) for bidirectional control. QuasAr2 is excited with a 637 nm laser; voltage-dependent fluorescence from a selected ROI is captured by an sCMOS/EMCCD camera and processed in real time. A monochromator (400–600 nm) provides feedback light with 100 µs temporal and 0.1 nm spectral resolution. Custom software (Beanshell in µManager) computes bleaching-corrected ΔF/F0, compares it to a user-defined target (holding value), and adjusts monochromator wavelength via an integral controller within a tolerance range (typically ±1%). A PID+Kalman variant was tested without performance benefits. Initial calibration (∼20 s) estimates photobleaching parameters; ΔF/F0 is computed live. Sampling rates up to 100 Hz (10 ms exposure) yield ~1.8×10^6 photons per ROI per frame with shot-noise ~0.08% of signal. Step protocols (3- or 4-step) and an on-the-run mode allow dynamic selection of holding values. A pseudo I/V software applies consecutive clamp steps and records required wavelengths; simultaneous patch clamp provides calibration linking ΔF/F0 to voltage and wavelengths to currents via linear regression. Biological systems: C. elegans body-wall muscle (BWMs), cholinergic and GABAergic motor neurons, pharyngeal muscle, and the DVB neuron; rat organotypic hippocampal CA1 pyramidal neurons expressing QuasAr2 and soma-targeted BiPOLES (somBiPOLES). Transgenes were delivered via established C. elegans transgenesis; in slices, AAV9 CaMKII-somBiPOLES was injected at DIV3–5; single-cell electroporation of QuasAr2 at DIV14–16. Imaging was performed at up to 100 fps, 10 ms exposure, 4×4 binning; irradiances were specified for organisms. Concurrent patch clamp was used to validate voltage control, assess membrane resistance under illumination, determine calibration slopes, and measure currents induced by wavelength steps. Additional software implements optical current clamp (continuous/pulsed ramps) without feedback for evoking and recording APs.

• The OVC achieved fast closed-loop clamping of QuasAr2 fluorescence (voltage) between approximately −5% to +5% ΔF/F0 in C. elegans BWMs and motor neurons, and ±3% in rat hippocampal neurons. • Transition times: with BiPOLES in BWMs, median transition times were ~147.7 ± 25.5 ms for depolarizing steps and 88.0 ± 5.7 ms for hyperpolarizing steps, significantly faster than single actuators (ChR2 return 678.6 ± 148.9 ms; GtACR2 return 239.5 ± 29.9 ms). Nearly 50% of control events completed within 20–30 ms; control deviation was within ±1% at 84% of time points. • Calibration in BWMs showed a linear relation: ±3% ΔF/F0 modulated voltage by −7 to +8 mV (≈24 mV per 10% ΔF/F0). With 400 ms steps, ±5% ΔF/F0 covered ~22 mV (≈ −40 to −18 mV). BiPOLES mediated membrane currents spanning ~−190 pA, approximately linear for 420–580 nm. • All-optical pseudo I/V curves were obtained by mapping ΔF/F0 steps to required wavelengths and transforming to voltage/current via linear regressions: membrane potential vs ΔF/F0 (R2 ≈ 0.8); current vs wavelength (R2 ≈ 0.8). • Homeostatic excitability in unc-13(n2813) mutants: to achieve +5% ΔF/F0, the OVC required significantly bluer wavelengths (534.2 ± 2.8 nm) than wild type (550.4 ± 3.1 nm; p=0.0012), indicating higher muscle excitability. Current ramps produced larger voltage responses in mutants (p=0.04). Spontaneous APs had increased amplitude and duration compared to wild type (amplitude p=8.27E-4; duration p=0.0082). • Channelopathy analysis: egl-19(n2368) L-type VGCC gain-of-function mutants displayed larger, prolonged APs (amplitude p=3.04E-4; duration p=1.53E-6). Optical I/V curves diverged from wild type at positive clamp values (e.g., 2–5% ΔF/F0 steps significant), and difference curves (mutant − wild type) matched electrophysiological difference currents after capacitance normalization. • Mammalian neurons: In CA1 pyramidal neurons co-expressing somBiPOLES and QuasAr2, ±3% ΔF/F0 OVC steps produced voltage changes of ~−4 mV (hyperpolarization) and +3 mV (depolarization). QuasAr2-only cells showed no fluorescence modulation by monochromator without actuator, while electrical depolarization of 100 mV evoked ~21% ΔF/F0. somBiPOLES-only cells depolarized ~21.5 mV under 637 nm, partially counteracted by 530 nm. • Dynamic clamping: In pharyngeal muscle (~4 Hz APs), OVC with higher integral gain and tighter tolerance suppressed APs: amplitude reduced from 26.0 ± 1.6% to ~5.0 ± 0.7% ΔF/F0, duration reduced from 118.7 ± 7.3 ms to ~83.1 ± 8.8 ms; pump events were suppressed (p=2.03E-4). • In the DVB neuron, spontaneous APs (~7.7% ΔF/F0, ~500 ms FWHM) were significantly reduced in amplitude and duration under OVC. Patch clamp measured resting potential −49.0 ± 8.9 mV; AP threshold −23.4 mV, peak +26.3 mV, duration 359 ms (FWHM), after-hyperpolarization −38.5 mV, indicating ~50 mV AP amplitude corresponding to ~7.7% ΔF/F0.

The OVC achieves, for the first time, all-optical closed-loop voltage clamping at the single-cell level by integrating a near-infrared GEVI (QuasAr2) with bidirectional actuation (BiPOLES) and real-time feedback control. It combines the non-invasiveness and throughput advantages of imaging with capabilities typically associated with electrophysiology, enabling precise control over membrane potential in intact C. elegans tissues and measurable modulation in mammalian neurons. The system accurately holds voltage-equivalent fluorescence within tight tolerances, exhibits rapid transition times, and maintains stability over extended measurements. It detects physiological changes such as increased postsynaptic excitability in unc-13 mutants and altered conductance/activation in egl-19 gain-of-function channels, with optical I/V relations matching patch-clamp data. Dynamic clamping of ongoing rhythmic or sparse spiking (pharynx, DVB) demonstrates rapid feedback sufficient to suppress or shape native activity and behavior. Limitations arise in mammalian neurons where resting potential proximity to the chloride reversal potential and higher actuator expression lead to shunting, restricting the control range. The OVC’s speed is currently capped by 100 Hz software sampling; hardware acceleration and faster opsins could improve performance. Overall, the OVC advances optogenetics from perturbation toward true control, with potential for high-throughput screening and adaptive modulation of neural activity in behaving animals.

This work introduces an optogenetic voltage clamp that provides closed-loop, bidirectional, all-optical control of membrane voltage using QuasAr2 readout and BiPOLES actuation. It enables precise clamping in C. elegans muscles and neurons, calibration of fluorescence to voltage and wavelength to current, generation of all-optical I/V relationships, detection of homeostatic and channelopathic changes, and dynamic suppression of action potentials and behaviors. Translation to rat hippocampal neurons demonstrates feasibility with a reduced control window. The OVC paves the way for contact-less, high-throughput electrophysiology and adaptive control of neural circuits in vivo. Future work should minimize optical crosstalk with red-shifted sensors and blue-shifted actuators, optimize controller speed (e.g., FPGA-based implementations), explore faster opsins, and extend applications to other cell types and organisms, including potential therapeutic contexts such as seizure interruption or adaptive deep brain stimulation.

• In mammalian neurons, the controllable voltage range was limited to a few millivolts, likely due to resting potentials near the chloride reversal potential and relative expression levels causing GtACR2-mediated shunting during compensation. • Optical crosstalk: 637 nm GEVI excitation partially activates Chrimson; although compensable in C. elegans, it complicates mammalian applications. Improved spectral separation (more red-shifted sensors, blue-shifted actuators) is needed. • Speed constraints: Software sampling at up to 100 Hz limits transition times (~90–150 ms). Faster acquisition/control (e.g., FPGA, smaller ROIs) and higher-speed opsins are required to approach the ~20 ms system time constant. • Calibration dependence: Absolute voltages and currents require simultaneous electrophysiological calibration; fluorescence-based control reflects relative changes and may vary with baseline and GEVI membrane localization. • Not a full substitute for patch clamp: While non-invasive and high throughput, the OVC is constrained by ion reversal potentials and actuator kinetics; prolonged illumination can cause inactivation, though feedback compensates until saturation.