All-optical reporting of inhibitory receptor driving force in the nervous system

Selfe, Joshua S.; Steyn, Teresa J. S.; Shorer, Eran F.; Burman, Richard J.; Düsterwald, Kira M.; Kraitzick, Ariel Z.; Abdelfattah, Ahmed S.; Schreiter, Eric R.; Newey, Sarah E.; Akerman, Colin J.; Raimondo, Joseph V.

doi:10.1038/s41467-024-53074-y

Download PDF

Article
Open access
Published: 16 October 2024

All-optical reporting of inhibitory receptor driving force in the nervous system

Nature Communications volume 15, Article number: 8913 (2024) Cite this article

8231 Accesses
2 Citations
19 Altmetric
Metrics details

Subjects

Abstract

Ionic driving forces provide the net electromotive force for ion movement across receptors, channels, and transporters, and are a fundamental property of all cells. In the nervous system, fast synaptic inhibition is mediated by chloride permeable GABA_A and glycine receptors, and single-cell intracellular recordings have been the only method for estimating driving forces across these receptors (DF_GABAA). Here we present a tool for quantifying inhibitory receptor driving force named ORCHID: all-Optical Reporting of CHloride Ion Driving force. We demonstrate ORCHID’s ability to provide accurate, high-throughput measurements of resting and dynamic DF_GABAA from genetically targeted cell types over multiple timescales. ORCHID confirms theoretical predictions about the biophysical mechanisms that establish DF_GABAA, reveals differences in DF_GABAA between neurons and astrocytes, and affords the first in vivo measurements of intact DF_GABAA. This work extends our understanding of inhibitory synaptic transmission and demonstrates the potential for all-optical methods to assess ionic driving forces.

GABAergic neuron-to-glioma synapses in diffuse midline gliomas

Article Open access 19 February 2025

Astrocytic chloride is brain state dependent and modulates inhibitory neurotransmission in mice

Article Open access 04 April 2023

Modulation of GluA2–γ5 synaptic complex desensitization, polyamine block and antiepileptic perampanel inhibition by auxiliary subunit cornichon-2

Article Open access 31 August 2023

Introduction

Ionic driving forces represent the net electromotive force available for generating ion fluxes across receptors, channels, and transporters. A driving force is the product of both an ion’s equilibrium potential and the potential difference across the membrane and constitutes the energy per unit charge that is available to mediate a host of fundamental cellular processes¹. Despite their fundamental importance, the only way to estimate ionic driving forces has been using intracellular electrophysiological recording methods. This has limited measurements to small numbers of cells and, more fundamentally, in all cases, these methods either directly or indirectly affect the ion gradients that underlie the driving forces². Meanwhile, methods for imaging intracellular ions cannot provide information on ionic driving force because they are blind to the cell’s membrane potential. Here we describe the first all-optical strategy for measuring ionic driving forces and demonstrate the utility of this approach for performing rapid, multi-cell measurements of undisturbed driving forces.

In the nervous system, fast synaptic inhibition is mediated by anion-permeable type A γ-aminobutyric acid receptors (GABA_ARs) and glycine receptors (GlyRs). These ligand-gated receptors are primarily permeable to chloride (Cl⁻) and, to a lesser extent, bicarbonate (HCO₃⁻)³. The transmembrane gradients for Cl⁻ and HCO₃⁻ determine the inhibitory receptor reversal potential (E_GABAA) and, together with the membrane potential (V_m), set the ionic driving force across these receptors (e.g., DF_GABAA). Therefore, DF_GABAA, by controlling anion flux across inhibitory receptors, is a key factor in determining inhibitory synaptic signalling in the nervous system. As an exemplar of an all-optical strategy for measuring ionic driving forces, we present ORCHID (all-Optical Reporting of CHloride Ion Driving force) as a tool for quantifying DF_GABAA, which is based on combining genetically encoded voltage indicators with light-gated ion channels.

Intracellular Cl⁻ concentration ([Cl⁻]_i), and hence E_GABAA, is a dynamic variable that is known to differ between subcellular compartments, cell types, and over a range of timescales and network states^4,5. Owing to the fact that E_GABAA is typically close to V_m, relatively small changes to E_GABAA or V_m can have a large effect on the polarity and magnitude of DF_GABAA and, thus, upon inhibitory signalling⁶. Alterations to neuronal DF_GABAA have been implicated in multiple neurological diseases, including epilepsy, chronic pain, schizophrenia, and autism^7,8,9,10. Similarly, DF_GABAA is thought to be modulated during brain development as part of the GABAergic system’s contribution to neural circuit formation^11,12. In regions of the mature nervous system, it is believed that DF_GABAA varies diurnally and is regulated as a function of wakefulness in neurons^13,14,15 and astrocytes¹⁶. Therefore, given the central importance of DF_GABAA for both brain function and dysfunction, techniques for estimating DF_GABAA are of major interest to the neuroscience field.

The most widely used electrophysiological strategy for estimating DF_GABAA is gramicidin-perforated patch-clamp recording². This technique provides electrical access to a target cell of interest without directly disturbing [Cl⁻]_i. When combined with a means of generating ion selective conductances, such as activation of GABA_ARs or light-sensitive channels with similar Cl⁻ permeability¹⁷, E_GABAA and DF_GABAA can be estimated. However, gramicidin recordings are technically challenging, have a low success rate and throughput, and perturb the intracellular levels of HCO₃⁻, directly modifying DF_GABAA, as well as K⁺ and Na⁺, which can indirectly change [Cl⁻]_i and DF_GABAA via the action of cation-Cl⁻ cotransporters (CCCs) such as KCC2 and NKCC1¹. Meanwhile, the best optical strategies for [Cl⁻]_i estimation include fluorescence lifetime imaging (FLIM) of the Cl⁻-sensitive dye MQAE^18,19,20 and genetically encoded Cl⁻ indicators²¹. Ratiometric genetically encoded Cl⁻ indicators allow for high-throughput measurements and can be targeted to different cell types, although there are challenges associated with ion selectivity, signal-to-noise ratio, and measurements in intact systems^22,23,24,25. More fundamentally, however, whilst currently available optical strategies can provide an estimate of [Cl⁻]_i, they do not incorporate information about the cell’s V_m and, therefore, cannot quantify the force available in the transmembrane gradient for Cl⁻ and HCO₃⁻, which is DF_GABAA.

In this work, we show that ORCHID provides an easy-to-use, genetically encoded optical estimation of a cell’s undisturbed DF_GABAA. ORCHID has been designed to incorporate the sensitive, chemi-genetically encoded voltage indicator (GEVI), Voltron2-ST²⁶, with an independent, spectrally separate means of generating anion-selective currents by using light activation of Guillardia theta anion channelrhodopsin 2 (GtACR2)²⁷. Voltron2-ST is therefore used to measure the magnitude and direction of GtACR2 anion-current-induced changes to V_m, thereby providing an estimate of DF_GABAA. We initially validate the ORCHID approach in mice by demonstrating that it replicates estimates of DF_GABAA that rely upon the activation of endogenous GABA_AR conductances. ORCHID is shown to provide reliable measurements of resting and dynamic DF_GABAA in different genetically targeted cell populations and across a timescale of seconds and hours. Furthermore, ORCHID confirms theoretical predictions regarding the biophysical mechanisms that establish DF_GABAA. Finally, we use ORCHID to reveal cell-type-specific differences in activity-dependent DF_GABAA changes during periods of network activity and to provide the first in vivo measurements of intact DF_GABAA in mouse cortical neurons. Together, our results demonstrate that ORCHID allows high-throughput, dynamic estimation of DF_GABAA in a cell-type-specific manner. This will assist in the study of inhibitory synaptic transmission and Cl⁻ homoeostasis in the nervous system and establish a precedent for using all-optical methods to assess ionic driving force more generally.

Results

ORCHID provides an optical estimate of the magnitude and direction of inhibitory receptor driving force

In order to optically measure DF_GABAA, we designed a strategy based on combining a genetically encoded optical reporter of V_m, with the simultaneous activation of a light-sensitive transmembrane anion channel (Fig. 1a). This provides a measurement of the difference between the V_m and E_GABAA (i.e., DF_GABAA), which can vary both in magnitude and polarity (Fig. 1b). The resulting ORCHID strategy utilises the recently developed soma-targeted GEVI Voltron2 (Voltron2-ST), and the light-activated anion channel GtACR2, separated via a P2A linker sequence for expression under the same promoter and translation of two separate proteins (Fig. 1c). The ORCHID construct has a double-floxed inverted-orientation (DIO) design for targeting Cre-expressing cells and is under the control of the EF1α promoter (Supplementary Fig. 1). Cre-expressing neurons in mouse hippocampal organotypic slices were successfully transduced with adeno-associated virus (AAV) vectors encoding ORCHID (“Methods”). Voltron2-ST works by irreversibly binding, via a HaloTag protein, one of a number of different fluorescent Janelia Fluor (JF) HaloTag ligands (dyes), each of which has different spectral properties²⁸. Prior to imaging cells expressing ORCHID, a dye incubation step was carried out with a dye that was selected for having spectral properties complementary to the experimental paradigm.

For ORCHID to work, it is critical to achieve spectral separation of the voltage readout using Voltron2-ST, from the manipulation of anion currents using GtACR2. Whole-cell patch-clamp recordings from pyramidal neurons expressing GtACR2 in hippocampal organotypic slices revealed a wide activation spectrum, with currents being generated by blue (475/28 nm), green (555/28 nm), and yellow (575/25 nm) excitation wavelengths, but not by orange (599/13 nm) or red (635/22 nm) wavelengths (Fig. 1d). Therefore, we utilised the HaloTag dye JF₆₀₈ together with continuous orange (599/13 nm) excitation light to ensure that Voltron2-ST imaging elicited no GtACR2 currents (Fig. 1e). This steady-state illumination produces negligible photocurrents by Voltron2-ST itself²⁶. Activation of GtACR2 with blue light (475/28 nm) does not significantly excite JF₆₀₈ fluorescence²⁹; nonetheless, blue light delivery was rapidly interleaved with camera acquisition to ensure that blue-light-generated autofluorescence would not cause artefacts in the imaging channel for Voltron2-ST labelled with JF₆₀₈ (Voltron2₆₀₈-ST) (“Methods”). The GtACR2 stimulation time was chosen to optimise the signal-to-noise ratio (SNR) in recordings (“Methods”). Imaging neurons expressing only Voltron2₆₀₈-ST confirmed that there was no change in fluorescence during blue light delivery (Fig. 1f). Voltron2-ST fluorescence readout (ΔF/F₀) was plotted as a negative for intuitive understanding of the underlying V_m shifts, as Voltron2-ST fluorescence increases with a decrease in V_m.

Whole-cell patch-clamp recordings from ORCHID-expressing neurons confirmed that Voltron2₆₀₈-ST reports linear fluorescence changes in response to voltage steps, thereby establishing that ΔF/F₀ can be used to estimate changes in voltage (ΔV). This was used to estimate DF_GABAA in subsequent analyses, including where cells were imaged alone (Fig. 1g–j; “Methods”). In ORCHID-expressing neurons, blue light stimulation elicited clear and reproducible V_m shifts, which showed reliable reproducibility within the same neuron (Fig. 1k, l and Supplementary Table 1).

To assess ORCHID’s ability to quantify DF_GABAA, we performed three variations of patch-clamp recordings with simultaneous voltage imaging in ORCHID-expressing neurons. Firstly, we used current-clamp recordings with a standard low-Cl⁻ internal solution to impose different DF_GABAA values in ORCHID-expressing neurons by injecting current via the patch pipette. Meanwhile, voltage imaging of ORCHID was performed under orange light, and anion currents were generated via blue-light activation of ORCHID (Fig. 1m, n). The relationship between ΔF/F₀ and ΔV (Fig. 1j) was used to convert the measured ΔF/F₀ values into estimations of anion-current-induced changes to V_m³⁰; when analysing recordings from multiple neurons, these optically determined DF_GABAA values were found to be equivalent to DF_GABAA measured from the simultaneous current-clamp recordings (Fig. 1n; R² = 0.9801; Runs test, one-tailed, deviation from linearity: P = 0.9360, n = 24 recordings from 17 cells). This demonstrated that the ability of ORCHID to estimate DF_GABAA is as accurate as using current-clamp recordings, with both methods accurately determining the direction of DF_GABAA and inclined to underestimate the full magnitude of DF_GABAA. Secondly, we used gramicidin-perforated patch-clamp recordings to quantify the similarity between DF_GABAA estimated optically using ORCHID and DF_GABAA measured through a perforated patch-clamp recording with undisturbed [Cl⁻]_i (Fig. 1n and Supplementary Fig. 2a). Measurements of DF_GABAA using these two techniques were consistent with one another (Fig. 1n and Supplementary Fig. 2b, Mann-Whitney test, two-tailed, P = 0.7969). Finally, ORCHID’s ability to detect a change in DF_GABAA was further confirmed by comparing measurements of DF_GABAA made before and after a neuron experienced an increase in [Cl⁻]_i imposed by patching with a high Cl⁻ internal solution (Fig. 1n and Supplementary Fig. 3).

As further validation, we assessed ORCHID’s ability to accurately determine DF_GABAA as compared to when anion currents were generated by activating endogenous GABA_ARs in different neuronal populations in hippocampal organotypic slices. Transgenic mice expressing Cre recombinase in different cell types were used to restrict reporter expression to either CaMKIIα+ pyramidal neurons or GAD2+ interneurons. Voltron2-ST labelled with JF₅₄₉ (Voltron2₅₄₉-ST) was used to assess DF_GABAA via endogenous GABA_ARs on the cell soma, which were activated by picolitre delivery of GABA (500 µM) via a glass micropipette and in the presence of a GABA_BR antagonist (CGP-55845; 5 µM). This was found to be an effective method of estimating DF_GABAA (Supplementary Figs. 4–6 and Supplementary Table 2) and revealed that both CaMKIIα+ pyramidal neurons and GAD2+ interneurons exhibit hyperpolarizing DF_GABAA under these conditions, with GAD2+ interneurons exhibiting slightly greater DF_GABAA (Fig. 2a–c). When these experiments were conducted using ORCHID, our all-optical strategy revealed similar hyperpolarizing DF_GABAA values for both neuronal populations, again with slightly greater DF_GABAA in GAD2+ interneurons compared to CaMKIIα+ pyramidal neurons (Fig. 2d–f). DF_GABAA estimates made using GtACR2 versus endogenous GABA_ARs were not statistically distinguishable (Supplementary Fig. 7; CaMKIIα+ pyramidal neurons: Mann-Whitney test, two-tailed, P = 0.0660; GAD2+ interneurons: Mann-Whitney test, two-tailed, P = 0.1819).

**Fig. 2: ORCHID reports endogenous inhibitory receptor driving force in different neuronal populations.**

ORCHID strategies confirm theoretical predictions regarding inhibitory receptor driving force

The accepted view is that DF_GABAA is predominantly established by the action of CCCs such as KCC2 and NKCC1, which are able to use secondary active transport to move Cl⁻ against its transmembrane concentration gradient by utilising cation gradients established by the primary active transporter, Na⁺/K⁺-ATPase¹. However, work using the genetically encoded Cl⁻ indicator, Clomeleon, challenged this idea by proposing that impermeant anions (and not CCCs) set [Cl⁻]_i, E_GABAA, and DF_GABAA³¹. We sought to examine these biophysical principles by first using a theoretical model to make predictions about the mechanisms that establish DF_GABAA. We extended a computational biophysical model of a single neuron based on the pump-leak mechanism that incorporated the permeant ions (Na⁺, K⁺, Cl⁻, HCO₃⁻) and their movement via leak channels, receptors, the Na⁺/K⁺-ATPase, and the major Cl⁻ extruder in mature neurons, KCC2. The model also incorporated impermeant anions, X, with average charge z³², pH control via the HCO₃⁻-buffering system, plus Na⁺/H⁺ exchange (NHE), dynamic volume, water permeability, and synaptic GABA_ARs that were able to elicit Cl⁻ and HCO₃⁻ conductances and resulting changes in V_m (Fig. 3a). As expected, simulating a block of KCC2 (i.e., a decrease in the conductance of KCC2 [g_KCC2]) in the model caused a positive shift in E_Cl, E_GABAA, and [Cl⁻]_i, with a small effect on V_m (Fig. 3b). This resulted in a substantial shift in DF_GABAA, causing a decrease in amplitude and an inversion in polarity of GABA_AR-mediated potentials (Fig. 3c). Meanwhile, simulating the addition of impermeant anions by reducing their average charge (z) generated a persistent negative shift in E_Cl, E_GABAA, and [Cl⁻]_i, consistent with previous observations³². However, this manipulation caused ionic redistribution such that V_m also shifted and in a manner proportional to the effects upon E_Cl and E_GABAA (Fig. 3d). Therefore, GABA_AR-mediated potentials were of a similar size following the addition of impermeant anions (Fig. 3e).

To empirically test these predictions, we used ORCHID strategies (i.e., activation of GABA_ARs or stimulation of GtACR2) to measure the effect of manipulations upon DF_GABAA in hippocampal organotypic CaMKIIα+ pyramidal neurons. First, we observed that the addition of the selective inhibitor of KCC2, VU0463271 (VU; 10 μM), caused a robust depolarising shift in estimated DF_GABAA (Fig. 3f–i). This was evident when DF_GABAA was measured either via activation of GABA_ARs (Fig. 3f, g) or all-optically (Fig. 3h, i), and the effect upon DF_GABAA reversed following washout of VU (Fig. 3f–i). Second, we assessed DF_GABAA before and after the addition of intracellular impermeant anions, which was achieved by single-cell electroporation of fluorescently tagged anionic dextrans³² (Alexa Fluor 488) (“Methods”) (Fig. 3j). Successful addition of impermeant anions was confirmed by directly observing the fluorescent dextrans within the neuron of interest, and DF_GABAA was estimated from the size and polarity of GABA_AR responses before and after the electroporation procedure (Fig. 3k, l). Consistent with the prediction from the modelling work, the addition of impermeant anions had no detectable effect upon the experimentally determined DF_GABAA (Fig. 3m). Therefore, ORCHID strategies are able to confirm theoretical predictions about the mechanisms that establish DF_GABAA.

ORCHID reveals dynamic ion driving forces over different timescales

In neurons, [Cl⁻]_i and E_GABAA are considered dynamic variables under the control of multiple activity-dependent processes and are thought to vary over a range of timescales⁴. Therefore, we sought to use ORCHID to investigate DF_GABAA dynamics and first examined periods of elevated network activity, during which intense activation of GABA_ARs is hypothesised to cause transient changes in DF_GABAA⁶. ORCHID was expressed in CaMKIIα+ pyramidal neurons or GAD2+ interneurons, and we used low-Mg²⁺ artificial cerebrospinal fluid (aCSF) to induce seizure-like events (SLEs) in mouse hippocampal organotypic brain slices. DF_GABAA was estimated using ORCHID in target cells, whilst a simultaneous whole-cell current-clamp recording from a nearby CA1/CA3 pyramidal neuron provided an independent readout of network activity (Fig. 4a–d). The SLEs were characterised by pronounced depolarising shifts in V_m and intense action potential firing, which could be observed both in the current-clamp recordings and in ORCHID’s voltage signals (Fig. 4a, c). CaMKIIα+ pyramidal neurons showed substantial activity-dependent changes in DF_GABAA, exhibiting shifts in polarity from a hyperpolarizing DF_GABAA at baseline to a depolarising DF_GABAA immediately post-SLE, which then returned to a hyperpolarizing DF_GABAA > 2 min after the SLE (Fig. 4b). Similarly, GAD2+ interneurons exhibited a hyperpolarizing DF_GABAA at baseline that flipped polarity to become a depolarising DF_GABAA immediately post-SLE, and subsequently returned to a hyperpolarizing DF_GABAA after > 2 min (Fig. 4d). These dynamics were confirmed by using endogenous GABA_AR activation to estimate DF_GABAA in CaMKIIα+ pyramidal neurons and GAD2+ interneurons (Supplementary Fig. 8). Furthermore, shifts in DF_GABAA have been postulated to occur immediately prior to seizure onset and to form part of the mechanism by which seizures initiate³³. In line with this, ORCHID detected more modest decreases in the magnitude of DF_GABAA between baseline and immediately prior to the SLE (< 15 s prior to SLE onset) in both CaMKIIα+ pyramidal neurons and GAD2+ interneurons (Fig. 4b, d), suggesting that DF_GABAA begins to break down before SLE onset (Supplementary Fig. 9). Interestingly, GtACR2 activation during ORCHID recordings from GAD2+ interneurons elicited further network bursts following an SLE (Fig. 4c). This occurred in recordings from 8/12 GAD2+ interneurons, and not when recording from CaMKIIα+ pyramidal neurons (0/15), highlighting the network effects of activity-induced intracellular Cl⁻ accumulation in interneurons (Supplementary Fig. 10).

**Fig. 4: ORCHID reveals dynamic inhibitory receptor driving forces over different timescales.**

Having demonstrated ORCHID’s ability to report DF_GABAA dynamics on a timescale of seconds, we next sought to assess ORCHID’s potential to measure longer-term changes in a neuron’s DF_GABAA, such as those that have been linked to activity-dependent changes in CCCs³⁴. To this end, ORCHID was used to track DF_GABAA in the same GAD2+ interneurons before and after a 180 min treatment with either 4-aminopyridine (4-AP; 50 µM) to increase neuronal activity or tetrodotoxin (TTX; 1 µM) to silence neuronal activity (Fig. 4e). When DF_GABAA measurements were made after the 180 min treatment and a subsequent 10 min incubation in TTX, ORCHID revealed that raising neuronal activity had caused a sustained and robust depolarising shift in neuronal DF_GABAA (Fig. 4f, g), which was not evident in neurons that had been silenced with TTX (Fig. 4h, i), or in neurons exposed to vehicle control for the equivalent 180 min period (Supplementary Fig. 11). Hence, ORCHID can provide high-throughput measurements of resting and dynamic DF_GABAA across timescales of seconds and hours.

Astrocytes sustain an outward anion driving force during enhanced network activity

Astrocytes are known to play important roles in regulating and maintaining neuronal ion gradients by modulating extracellular ion concentrations. However, due to the difficulty in performing electrophysiological recordings from this cell type, it has not previously been possible to measure astrocytic DF_GABAA across different types of network activity, although [Cl⁻]_i has been imaged in astrocytes before^16,35. Using the Cre-lox system, we restricted ORCHID expression to GFAP+ astrocytes in mouse hippocampal organotypic slices and optically estimated DF_GABAA. We observed that, in stark contrast to neurons, the astrocytic DF_GABAA under baseline conditions was strongly depolarising, consistent with previous reports³⁶ (Fig. 5a–c). A related idea that has been proposed is that astrocytes may serve to continuously supplement anions into the extracellular space during periods of enhanced network activity, and thereby help to maintain the effectiveness of GABA_AR-mediated inhibitory synaptic transmission^16,35. To test this explicitly, we used ORCHID to continuously monitor astrocytic DF_GABAA across periods of intense network activity, which was achieved by inducing SLEs through low-Mg²⁺ aCSF. Whilst there was a modest decrease in the magnitude of DF_GABAA during the SLE, ORCHID revealed that the astrocytes continued to exhibit a strongly depolarising DF_GABAA before and after the SLE, and are therefore capable of maintaining a driving force for anion efflux during substantial changes in network activity (Fig. 5d, e and Supplementary Fig. 12). The large increase in fluorescence at the initiation of the SLE is likely due to a large depolarisation in the astrocytic V_m. When compared to neurons, markedly different resting and dynamic DF_GABAA in astrocytes suggest that they differ in how they utilise major Cl⁻ regulatory mechanisms such as NKCC1 and KCC2, which act to transport Cl⁻ in and out of the cell, respectively. To assess this, we performed single-nucleus RNA sequencing (snRNAseq) on the same hippocampal brain slice preparations (“Methods”) and distinguished pyramidal neurons and astrocytes using unbiased clustering of the transcriptomic profiles (Fig. 5f). After normalising for total gene expression, we observed that KCC2 (Slc12a5) was more highly expressed in pyramidal neurons than astrocytes. Meanwhile, NKCC1 (Slc12a2), although comparatively low in both cell types, exhibited higher expression in astrocytes than pyramidal neurons (Fig. 5f). This confirms the established idea that KCC2 is highly expressed in mature neurons, and NKCC1 is expressed at a higher level than KCC2 in astrocytes^37,38,39. Thus, ORCHID reveals unique features of astrocytic DF_GABAA compared to neurons, that correlate with differences in their Cl⁻-transport mechanisms.

In vivo determination of resting and dynamic inhibitory receptor driving forces in neurons

Finally, ORCHID offers the opportunity to generate estimates of undisturbed DF_GABAA in the intact brain. We therefore assessed ORCHID’s ability to report resting and dynamic DF_GABAA in vivo. ORCHID was expressed in layer 1 (L1) GAD2+ interneurons in the mouse primary somatosensory cortex (S1), and anaesthetised in vivo imaging was combined with simultaneous local-field potential (LFP) recordings to provide an independent measure of local population activity (Fig. 6a; “Methods”). Under baseline conditions, we observed a range of DF_GABAA polarities and magnitudes across different neurons, including hyperpolarizing, purely shunting, and depolarising DF_GABAA (Fig. 6b, c). In a subset of mice, we performed continuous DF_GABAA measurements before and after infusing 4-AP (500 µM) into the cortex to elicit SLEs (“Methods”). The occurrence of SLEs was confirmed from the LFP recordings, and DF_GABAA could be tracked in the same individual neurons before and after the SLE (Fig. 6d). This revealed that L1 GAD2+ interneurons exhibit depolarising shifts in DF_GABAA as a result of SLEs (Fig. 6e). These data demonstrate that in vivo ORCHID affords easy and rapid reporting of resting and activity-dependent dynamic shifts in DF_GABAA in the intact brain, under control and disease-relevant conditions.

Discussion

All-optical approaches that combine voltage imaging and optogenetic stimulation represent a rapidly emerging strategy for studying neuronal dynamics in the intact nervous system^40,41. Here we present the first all-optical strategy for studying ionic driving forces. We describe ORCHID - a powerful all-optical tool for the estimation of DF_GABAA, which represents the electromotive force for ion flux across inhibitory receptors. Whilst ORCHID could be used to estimate DF_GABAA in any cell type or organ, we validate its use in the rodent nervous system. Taking advantage of ORCHID’s ability to provide high-throughput, genetically targeted estimates of DF_GABAA over extended periods of time, we confirm theoretical predictions regarding the mechanisms that establish DF_GABAA, utilise ORCHID to reveal cell-type-specific differences in activity-dependent DF_GABAA changes during network activity, and provide the first in vivo measurements of intact DF_GABAA in mouse cortical neurons.

ORCHID is easy to use with no spectral crosstalk between the orange-light excitation of Voltron2₆₀₈-ST, and blue-light activation of the integrated anion-permeable channel, GtACR2. This means that ORCHID is compatible with a wide variety of experimental paradigms. ORCHID’s accuracy in reporting DF_GABAA depends upon the size of the light-activated anion conductance (the GtACR2-induced change in the membrane conductance) relative to other conductances that are concurrently active in the cell of interest. For this reason, it is important that GtACR2 is known to generate large, consistent transmembrane conductances with minimal effects on neuronal function^27,42, and any future optimisations of light-activated anion channels are predicted to further improve ORCHID’s accuracy. GtACR2’s permeability sequence to different anions (NO₃⁻ > I⁻ > Br⁻ > Cl⁻ > F⁻) is the same as other anion channels, including GABA_ARs and GlyRs^27,43,44, which are also permeable to HCO₃⁻, and exhibit a Cl⁻-to-HCO₃⁻ permeability ratio of five to one^43,44,45,46. The equivalence of the anion permeabilities of GtACR2 and GABA_ARs is supported by the fact that their reversal potentials are the same^47,48, and also our evidence that ORCHID measurements of DF_GABAA are indistinguishable from those made by activating GABA_ARs. ORCHID, therefore, provides an effective readout of the driving force across the receptors that mediate fast synaptic inhibition in the brain. As a result, it is possible that changes in intracellular pH and HCO₃⁻, resulting in DF_HCO3- differences between cell types or network states, could also contribute to observed differences in DF_GABAA made using ORCHID. Nonetheless, as GtACR2 is primarily permeable to Cl⁻²⁷, ORCHID measurements can also be considered to be estimates of Cl⁻ driving force (DF_Cl), and future efforts to modulate the ion selectivity of light-activated Cl⁻ channels could further enhance the accuracy of all-optical reporters of DF_Cl.

ORCHID’s all-optical approach has several powerful advantages over currently available techniques for estimating DF_GABAA. Firstly, and most fundamentally, ORCHID does not perturb ionic gradients prior to measurement. Secondly, it is both genetically targetable and addressable by light, which affords excellent temporal and spatial control. Thirdly, ORCHID’s high-throughput nature allows for measurements that are orders of magnitude faster than traditional techniques. A typical gramicidin-perforated patch-clamp recording requires 30–60 min per cell, compared to 1–5 s for an ORCHID recording. Fourthly, ORCHID affords measurements from small cells and potentially from subcellular compartments, which are difficult to access using electrode-based techniques. Fifthly, a strength of ORCHID is that the stability of recordings over time allows for the possibility of measuring relative changes in DF_GABAA, even if no switch in the polarity of the DF_GABAA occurs.

In theory, a fluorescent reporter of [Cl⁻]_i could be used to report relative changes in a cell’s DF_GABAA, by measuring changes in the amplitude of the [Cl⁻]_i response to an evoked Cl⁻ conductance⁴⁹. However, this is unlikely to be achieved in practice because of the inherent limitations in the sensitivity of Cl⁻ imaging²⁵. Furthermore, as fluorescent reporters of [Cl⁻]_i are cytoplasmic, detecting [Cl⁻]_i changes would require much larger and longer conductances that alter [Cl⁻]_i in the cytoplasm, unlike ORCHID whose readout only requires Cl⁻ flux across the membrane to shift V_m. Finally, a fluorescent reporter of [Cl⁻]_i would not account for the HCO₃⁻ component of DF_GABAA. By integrating a transmembrane voltage-reporter element and an anion conductance, ORCHID is highly sensitive to transmembrane flux and able to simultaneously read out V_m, both of which are key to making accurate measurements of DF_GABAA. Nonetheless, ORCHID is limited to reporting DF_GABAA, and unlike other forms of Cl⁻ imaging, cannot estimate absolute [Cl⁻]_i or E_GABAA. Further, it does not provide a measure of absolute V_m, although ratiometric GEVIs are being developed and could be integrated with future iterations of ORCHID to potentially report DF_GABAA together with V_m and, hence, E_GABAA⁵⁰.

Our results demonstrate how ORCHID strategies can be harnessed to test theoretical predictions about the underlying mechanisms that establish ionic driving forces. Our computational modelling and empirical evidence support the view that DF_GABAA is primarily established by the action of CCCs such as KCC2¹. Therefore, whilst genetically encoded Cl⁻ indicators and other techniques that directly report [Cl⁻]_i are of considerable value, ORCHID highlights the danger of interpreting [Cl⁻]_i changes in the context of inhibitory synaptic transmission and DF_GABAA, without a concurrent readout of V_m. This is particularly the case for manipulations such as altering impermeant anions, which are predicted to shift E_Cl or E_GABAA and V_m in tandem^32,51. Given its ability to provide a sensitive readout of DF_GABAA, we note that ORCHID could serve as a high-throughput means for discovering new and improved enhancers, or inhibitors, of transmembrane Cl⁻ transport.

By avoiding intracellular recordings, ORCHID enabled us to monitor unperturbed resting and dynamic DF_GABAA. We confirm predictions from previous work by showing that intense network activity can induce a profound change in DF_GABAA polarity in neurons, with GABAergic synaptic input switching from inhibitory to excitatory, which is relevant for the management of status epilepticus^52,53. Whilst SLE-associated DF_GABAA estimates have been made using gramicidin-perforation in pyramidal neurons¹⁷, we provide the first data showing that inhibitory interneurons also undergo an inversion in DF_GABAA following seizure-like activity, which is predicted from observations that Cl⁻ increases in interneurons during SLEs⁵⁴. Recent work has suggested that the transition to seizures is associated with incremental increases in neuronal [Cl⁻]_i, which is predicted to reduce DF_GABAA prior to seizure onset³³. We provide direct evidence for this clinically relevant phenomenon and thereby demonstrate the utility of ORCHID in dynamically estimating DF_GABAA across network states.

By harnessing ORCHID’s ability to be genetically targeted, we recorded DF_GABAA from multiple cell types, including large populations of hippocampal CaMKIIα+ pyramidal neurons, GAD2+ interneurons, and GFAP+ astrocytes. This revealed that hippocampal astrocytes display a strong depolarising DF_GABAA at rest, which is opposite to the hyperpolarizing DF_GABAA exhibited by hippocampal pyramidal neurons in vitro, and is supported by the differential expression of CCCs between these two cell populations^17,36,48. We observed that in contrast to neurons, astrocytic V_m takes longer to respond to optogenetic activation, as has previously been described⁵⁵; this should be considered when using ORCHID to estimate DF_GABAA in this cell type. Previously, it has been suggested that astrocytes could serve as a source of extracellular anions that help to sustain the effectiveness of inhibitory GABAergic transmission⁵⁶, which has recently been supported by direct measurements of astrocytic [Cl⁻]_i during periods of network activity¹⁶. In support of this idea, our measurements of dynamic DF_GABAA reveal that astrocytes are able to maintain a force for the outward flux of anions even during intense network activity. As other cell types exhibit specific expression patterns of CCCs and other anion transporters⁵⁷, ORCHID could be used to determine how these translate into differences in the anion-driving forces in cells such as oligodendrocytes, ependymal cells, microglia, and endothelial cells. Finally, we also demonstrate ORCHID’s ability to provide measurements of undisturbed DF_GABAA in the intact brain. Our in vivo results captured a range of resting DF_GABAA values in cortical L1 GAD2+ interneurons, which is different to what was observed in vitro. We hypothesise that this heterogeneity may be due, in part, to differences in network state⁵⁸ or the animal’s sleep-wake history¹⁵. We revealed that DF_GABAA is a dynamic parameter in the intact brain, with periods of intense activity causing substantial shifts towards depolarising DF_GABAA. Indeed, through the use of fibre photometry, or the implementation of a GEVI that is suitable for use with 2-photon microscopy, ORCHID has the potential to generate measurements of resting and dynamic DF_GABAA throughout the mammalian brain⁴¹, including from deep brain structures.

In terms of future applications, a further important question is how DF_GABAA is regulated at a subcellular level. For example, it has been suggested that the properties of inhibitory synaptic input vary across a neuron and between different cellular compartments^{17,59,60,61,62,63}. However, this has been difficult to address given the issue of space-clamp and additional technical challenges associated with electrode-based recordings. One could imagine combining ORCHID strategies with protein-targeting motifs and the patterned delivery of light in order to record DF_GABAA from difficult-to-access subcellular compartments, such as distal apical dendrites, the axon initial segment, and astrocytic processes. In a similar manner, one could adopt targeted ORCHID strategies to estimate DF_GABAA/DF_Cl across the membranes of organelles, such as mitochondria, lysosomes, and vesicles. In doing so, ORCHID could be used to study a range of cellular phenomena, including cell division, growth, and migration, in which anion/Cl⁻ fluxes across different membranes contribute to the underlying intracellular and intercellular signalling processes.

In conclusion, we establish ORCHID as an all-optical strategy for estimating DF_GABAA in the nervous system. This underscores the potential for similar all-optical strategies for elucidating how ionic driving forces affect cellular signalling in a range of biological contexts.

Methods

Ethics statement

The use of the animals in this study was approved by the University of Cape Town Animal Ethics Committee (AEC Protocol 021/026 and AEC Protocol 022/038).

ORCHID construct subcloning

The ORCHID construct combined GtACR2 from the pAAV-EF1α-FRT-FLEX-GtACR2-EYFP plasmid, which was a gift from Mingshan Xue (Addgene plasmid #114369)⁴⁷, with Voltron2-ST from the pGP-pcDNA3.1 Puro-CAG-Voltron2-ST plasmid²⁶, which was kindly provided by Eric Schreiter and Ahmed Abdelfattah. The two genes were linked by a P2A linker sequence identical to that used by Holst et al.⁶⁴ to form an insert sequence, and include the Kozak sequence 5’ GCCACC(ATG) 3’. The vector used for the ORCHID construct originated from pAAV-hSyn-DIO {ChETA-mRuby2}on-W3SL, which was a gift from Adam Kepecs (Addgene plasmid #111389)⁶⁵, with the hSyn promoter replaced with the EF-1α promoter from pAAV-nEF Con/Foff hChR2(H134R)-EYFP, which was a gift from Karl Deisseroth (Addgene plasmid #55647)⁶⁶. The GtACR2-P2A-Voltron2-ST sequence of the ORCHID construct was synthesised in antisense (Thermo Fisher Scientific GeneArt Gene Synthesis) and inserted into the vector using the Bsp1407I/NheI restriction enzyme sites and the FastDigest versions of the two enzymes (Thermo Fisher Scientific). Sanger sequencing confirmed the sequence of the final construct. Production of the AAV vector (serotype 1) containing the completed pAAV-EF1α-DIO-GtACR2-P2A-Voltron2ST-W3SL construct was carried out by Dr R. Jude Samulski and the University of North Carolina Vector Core. The ORCHID plasmid (pAAV-EF1α-DIO-GtACR2-P2A-Voltron2ST-W3SL) is available from Addgene (plasmid #217676).

Animal husbandry

The animals used in this study were wild-type (C57BL/6 background) or GAD2-IRES-Cre mice (RRID: IMSR_JAX:010802), with the GAD2-IRES-Cre strain being characterised by expression of Cre recombinase in all GABAergic interneurons. Male and female mice were group housed on a 12 h/12 h light/dark cycle, with the lights on at 06:00. Food and water were provided ad libitum. Mice of both sexes were used, although sex was not considered during the analysis.

Mouse hippocampal organotypic brain slice cultures

Organotypic brain slice cultures were prepared using postnatal day (P) 7 mice and following the protocol originally described by Stoppini, Buchs and Muller⁶⁷ (for details, see Raimondo et al.⁶⁸). All reagents were purchased from Sigma-Aldrich unless otherwise stated. Briefly, the mice were killed by cervical dislocation, and the brains were removed and placed in 4 °C Earle’s Balanced Salt Solution (EBSS), with 6.1 g/l HEPES, 6.6 g/l glucose, and 5% 1 M NaOH. The hippocampi were removed and sectioned into 350 μm slices using a tissue chopper (McIlwain). The slices were placed on Millicell-CM membranes and cultured in a medium consisting of (v/v): 25% EBSS; 49% minimum essential medium, 25% heat-inactivated horse serum; 1% B27 (Invitrogen, Life Technologies), and 6.2 g/l D-glucose. The slices were incubated at 35–37 °C in a 5% CO₂ humidified incubator.

Viral transduction was performed at 0–1 days in vitro (DIV). For experiments in which endogenous GABA_ARs were activated, pGP-AAV-syn-flex-Voltron2-ST-WPRE-SV40 was utilised, while for ORCHID experiments, pAAV-EF1α-DIO-GtACR2-P2A-Voltron2ST-W3SL was utilised. AAV-CaMKIIα-Cre-GFP and AAV8-GFAP-GFP-Cre were used to target CaMKIIα+ pyramidal neurons and GFAP+ astrocytes, respectively. Additional targeting and validation were performed to ensure recordings were made from the correct cell types. Cells that were CaMKIIα+ were confirmed as being pyramidal neurons by their morphology (large apical dendrite extending into stratum radiatum) and the location of their cell bodies in the stratum pyramidale of CA1/CA3. Further, patch-clamp recordings provided electrophysiological confirmation in a subset of cells (Supplementary Fig. 13). GAD2+ cells were confirmed as being interneurons by their morphology (multiple large processes extending from the cell body) and the occurrence of their cell bodies in stratum radiatum. Further electrophysiological characterisation was used in a subset of cells (Supplementary Fig. 13). In addition, immunohistochemistry was used to confirm that DIO-ORCHID (i.e., DIO-GtACR2-P2A-Voltron2ST) and GAD1 were being co-expressed in the same cells in hippocampal organotypic slices from the GAD2-IRES-Cre mouse line (Supplementary Fig. 14; “Methods”). We confirmed the accurate targeting of astrocytes by GFAP-Cre through their morphological features (cells with small somata and multiple fine processes) and occurrence in the stratum radiatum. Additionally, patch-clamp recordings were performed from a subset of cells. In all cases (n = 14/14), we found a low resting V_m (−72.25 ± 1.84 mV), low membrane resistance (59.76 ± 12.53 MΩ), and an inability to fire action potentials, which is typical of hippocampal astrocytes^69,70 (Supplementary Fig. 15). We also confirmed that repeated stimulation of astrocytes expressing GtACR2 did not affect baseline electrophysiological properties (Supplementary Fig. 15). In addition, at the frequency (0.2 Hz) and duration (1 s) at which DF_GABAA was probed using ORCHID in GFAP+ astrocytes, we observed stable responses with no differences in the DF_GABAA values measured within the same network state (either pre-SLE or post-SLE; Supplementary Table 3).

Here, as in all relevant methods, micropipettes were prepared from borosilicate glass capillaries (Warner Instruments). The AAVs were loaded into a glass micropipette along with FastGreen (0.1% w/v) to aid visualisation and were injected into the slices using the Openspritzer system⁷¹.

Recordings were performed at 7–14 DIV, which is equivalent to P14-21 (as slices are prepared from P7 mice). This, as well as previous work, has shown that pyramidal neurons in the organotypic slices of the hippocampus have mature and stable Cl⁻ homoeostasis mechanisms at this stage, as evidenced by their hyperpolarizing DF_GABAA^72,73,74,75. Prior to imaging, slices were incubated for 30 min in aCSF bubbled with carbogen gas (95% O₂, 5% CO₂), containing either JF₅₄₉ or JF₆₀₈ HaloTag dye at 200 nM. The aCSF was composed of (in mM): NaCl (120), KCl (3), MgCl₂ (2), CaCl₂ (2), NaH₂PO₄ (1.2), NaHCO₃ (23), and D-glucose (11). During this incubation step, the dye binds irreversibly to the Voltron2-ST protein²⁸. After the incubation was complete, slices were transferred to a submerged recording chamber where they were continuously superfused with 32 °C carbogen-bubbled aCSF using a peristaltic pump (Watson-Marlow). Pharmacological agents included CGP-55845 (5 µM) and 2-chloroadenosine (4 µM) in the recording aCSF for all experiments in which GABA_ARs were activated, as well as picrotoxin (PTX; 100 µM), VU0463271 (VU; 10 μM), 4-aminopyridine (4-AP; 50 µM), and tetrodotoxin (TTX; 1 µM) for subsets of the in vitro experiments.

Surgical procedures for in vivo imaging

Intracranial bulk regional viral injections of pAAV-nEF-DIO-GtACR2-P2A-Voltron2-ST-W3SL were performed on P0-1 GAD2-IRES-Cre mouse pups^76,77. Adult breeders were removed from the home cage and placed in a holding cage for the duration of the procedure. A 30 G insulin syringe was loaded with 1 μl AAV combined with FastGreen (0.1% w/v) to aid visualisation and mounted in a stereotaxic frame. Pups were individually removed from their home cage and anaesthetised using 2 ml isoflurane pipetted onto a gauze swab in an enclosed space. Respiration rate and skin colour were monitored to assess the depth of anaesthesia. Once the depth of anaesthesia was sufficient, pups were rapidly transferred to a heating pad beneath the stereotaxic frame and the AAV was injected into S1 of the left cerebral hemisphere at a depth of ~ 1 mm, with a successful injection indicated by the green coloration of the hemisphere. The injection typically took ~ 30 s. The pup was then transferred back to its home cage and monitored until fully recovered from anaesthesia when the injection of the next pup was begun. We observed a 100% survival rate of injected pups using this method.

Imaging experiments began after P28. Retroorbital injections of JF₆₀₈ were performed 3–24 h prior to imaging^26,28. 100 nmol JF₆₀₈ was dissolved in 20 μl DMSO, 20 μl Pluronic F-127 (20% w/v in DMSO, Thermo Fisher Scientific), and 60 μl sterile phosphate-buffered saline (PBS). The mice were anaesthetised using an intraperitoneal (IP) injection of ketamine/xylazine mixture (ketamine 80 mg/kg and xylazine 10 mg/kg). The JF₆₀₈ solution was delivered using a 30 G insulin syringe as described by Yardeni et al.⁷⁸. The preparation for anaesthetised recordings was adapted from Burman et al.⁵⁸. Mice were anaesthetised using 3 l/s O₂ with 5% isoflurane for induction, and with the isoflurane reduced to 1–2% for maintenance. Once anaesthetised, the animal was secured in a stereotaxic frame (Narishige). An IP injection of buprenorphine (0.5 mg/kg) and a subcutaneous (SC) injection of 200 μl sterile saline were administered for analgesia and hydration, respectively. The mouse’s body temperature was maintained at ~ 37 °C using a heating pad and a temperature probe. The head was shaved, and eye-protecting ointment (Duratears) was applied to each eye. An incision in the scalp was made using surgical scissors, and the area was expanded by blunt dissection to expose the skull. Forceps were used to remove any membranous tissue from the surface of the skull. Tissue adhesive (Vetbond) was used to fix the edges of the incised scalp to the skull and to stabilise the cranial sutures. Several layers of dental cement (InterDent) were applied to create a recording chamber on top of the skull. A 0.5 mm craniotomy was drilled over S1 using a dental drill (Dental Lab). The craniotomy was submerged in cortex buffer containing (in mM): NaCl (125), KCl (5), HEPES (10), MgSO₄·7H₂O (2), CaCl₂·2H₂O (2), and D-glucose (10). The bone flap and dura were removed. The mouse was then transferred to the imaging setup. In a subset of animals, 4-AP was infused into the cortex buffer in the recording chamber using a pipette (500 µM working concentration). The recording session typically lasted 3 hours.

Electrophysiology

For patch-clamp recordings, micropipettes were filled with a low-Cl⁻ internal solution composed of (in mM): K-gluconate (120), KCl (10), Na₂ATP (4), NaGTP (0.3), Na2-phosphocreatine (10), and HEPES (10). For high-Cl⁻ recordings, pipettes were filled with a high-Cl⁻ internal solution ([Cl⁻] = 141 mM) composed of (in mM): KCl (135), NaCl (8.9), and HEPES (10). For all gramicidin-perforated recordings, gramicidin A from Bacillus brevis was added to the internal solution for a working concentration of 80 µg/ml. Adequate perforation of gramicidin-perforated recordings was assessed by monitoring access resistance and was defined as when access resistance was < 90 MΩ. Recordings were made using an Axopatch 200B amplifier (Molecular Devices), and data was acquired using WinWCP version 5.6.6 (University of Strathclyde). To calculate R_m in current-clamp mode, the change to V_m upon injection of 600 pA was used to calculate R_total, from which the pipette resistance was then subtracted. For experiments investigating the role of impermeant anions in setting DF_GABAA, anionic 10 000 MW dextran bound to Alexa-Flour 488 (Thermo Fisher Scientific) was electroporated into cells. This molecule is a hydrophilic polysaccharide, being both membrane impermeant and having a large negative average charge. A micropipette was filled with 5% dextran solution in PBS, and the micropipette was positioned near the soma of the targeted neuron. Voltage pulses (5–10 pulses, 20 ms duration, 0.5–1 V) were applied using a stimulus isolator, and successful electroporation was confirmed visually by observing the neuron filled with the fluorescent dye. For in vivo LFP recordings, micropipettes were filled with cortex buffer. The tip of the pipette was broken against tissue paper with the aim of achieving a tip diameter of ~ 5 µm. Recordings were made using a differential amplifier (A-M system, 1800), and data was acquired using LabChart 8 (ADInstruments).

Widefield fluorescence imaging

Neurons were visualised using a BX51WI upright microscope (Olympus) equipped with a 20x water-immersion objective (Olympus XLUMPlanFL) and a sCMOS camera (Andor Zyla 4.2). For imaging Voltron2₅₄₉-ST, a green LED with a 555/28 nm excitation filter (Lumencor Aura III) was used for excitation (20.6 mW/mm²) with a further multi-band filter set for collecting emission between 590 and 660 nm (Chroma, 69401-ET-380/55-470/30-557/35 Multi LED set). For Voltron2₆₀₈-ST or ORCHID, a high-powered orange LED (590 nm, SOLIS-590C, Thorlabs) was used for excitation (143 mW/mm²). Optogenetic activation of GtACR2 as a part of ORCHID was achieved using a blue LED (475/28 nm, Lumencor Aura III), with the intensity of this LED fixed at 3.6 mW/mm². A T570lpxr beam splitter was used for combining the blue and orange LEDs before a filter set with a 599/13 nm excitation filter, a 612 nm long pass beam splitter, and a 632/28 nm emission filter was used (Chroma, 49311-ET-Red#3 Narrow band FISH).

Illumination of the tissue was restricted to a ~ 140 µm diameter region using the built-in Olympus variable-diameter field stop aperture. Image acquisition was controlled using μManager version 2.0.0. For current-step recordings, exposure was 5 ms with an image acquisition frequency of 200 Hz. For all voltage-step recordings and Voltron2₅₄₉-ST recordings, exposure was 10 ms and image acquisition frequency was 100 Hz. For ORCHID recordings, exposure was 20 ms and image acquisition frequency was 25 Hz. The imaging region was 256 × 256 pixels, and with no overlap between successive camera exposures, this ensured a 20 ms gap between successive exposures. Transistor-transistor logic (TTL) signals from the camera were set to output throughout the exposure, and this signal was sent to an Arduino that ran a custom script to activate the blue LED immediately after each camera exposure ended and for 15 ms, allowing for a 5 ms gap between the blue LED being switched off and the subsequent camera exposure. We found this to be crucial for artefact-free imaging. Camera acquisition and blue-light activation were thus both at a frequency of 25 Hz, with 50% and 37.5% on time, respectively.

To calculate the duration of optogenetic stimulation during ORCHID recordings with a frame rate of 25 Hz, we aimed to detect a typical ΔF/F₀ of 1% (Eq. 1), with an SNR of ~ 10 per optogenetic stimulation. We assumed negligible dark noise and read noise and accounted for Poisson-distributed shot noise with a value of $\sqrt{{F}_{0}}$ (Eq. 2). The Andor Zyla 4.2 sCMOS camera that was used has a well depth (e⁻) of 30000, and a maximum quantum efficiency of 82%, giving a required number of e⁻ per frame of 36585.37. Thus, F₀ can be calculated as follows:

$$\frac{\Delta F}{{F}_{0}}=\frac{1}{100}$$

(1)

$${{{\rm{SNR}}}}=\frac{\Delta F}{\sqrt{{F}_{0}}}$$

(2)

$${F}_{0}=1\times {10}^{6}$$

(3)

Therefore, the number of frames required per optogenetic stimulation to achieve an SNR of 10 at a ΔF/F₀ of 1% is:

$$\frac{1\times {10}^{6}}{36585.37}=\, \sim 27$$

(4)

We used an optogenetic stimulation time of 1 s for longer recordings (which gives an expected SNR of ~ 9.6 for a ΔF/F₀ of 1% at 25 Hz frame rate), and for shorter recordings, we used a stimulation time of 2 s (which gives an expected SNR of ~ 13.5 for a ΔF/F₀ of 1% at 25 Hz frame rate). These times were additionally chosen to have a similar total time of optogenetic activation in the shorter recordings and the longer recordings (10 s vs 13 s).

When utilising Voltron2₅₄₉-ST with GABA_AR activation, 500 μM GABA in recording aCSF was loaded into a micropipette, and 100 ms puffs of synthetic air were delivered to the micropipette using the Openspritzer system. This duration was chosen to maximise GABA delivery while minimising tissue movement. Imaging traces were 7 s long with a 1 s baseline period prior to the puff. Care was taken to ensure no movement artefact from the puff. ORCHID recordings had a duration of 25 s and consisted of five 1 s or 2 s strobed optogenetic activations, which were averaged post-hoc for some experiments (see Data Analysis). These durations are broadly similar to the duration of GABA_AR stimulation during micropipette delivery of GABA. During low-Mg²⁺ wash-ins, in order to increase the chance of recording an SLE, the duration of recordings was increased to 65 s, and ORCHID was activated with 1 s periods of strobed light every 5 s. Here the averaging of multiple ORCHID activations was not possible due to the rapid DF_GABAA changes. Instead, we took the mean signal during a single optogenetic activation (1 s) during each relevant period, which was used alongside a baseline period immediately prior to the optogenetic activation. The relevant periods were: < 15 s before SLE onset for pre-SLE, < 15 s before SLE cessation for during SLE, and < 15 s after SLE cessation for post-SLE. For pre-SLE and post-SLE measurements, optogenetic activations where neither the activation nor the baseline period was contaminated by SLE-associated spiking/depolarisations were selected (which excluded any measurements during GtACR2-triggered network burst firing). For during-SLE measurements in astrocytes, a measurement close to the end of the SLE, where measurements were typically more consistent and thus representative, was selected. When testing the stability of DF_GABAA measurements made using ORCHID in astrocytes within the same network state (either pre-SLE or post-SLE), the same paradigm of ORCHID measurements made using 1 s optogenetic activations every 5 s was used, with a statistical comparison between measurements of DF_GABAA made 20 s apart. SLEs were defined as events with significant deviation from the resting potential in patch-clamp recordings (> 2 standard deviations) lasting for at least 5 s. Network burst events that occurred post-SLE upon GtACR2 stimulation were defined as constant depolarisations of > 10 mV above baseline that lasted for longer than 250 ms.

Single-nucleus RNA sequencing

Nuclei from mouse hippocampal organotypic slices were dissociated in lysis buffer (Nuclei EZ Prep, NUC101) on ice and processed via the 10X Genomics platform to prepare libraries for sequencing as outlined in the Chromium Next GEM 3’ Single Cell 3 Reagent Kits v3.1 user guide. Four samples, each comprising 36 hippocampal slices, were processed. Libraries were sequenced using an Illumina NovaSeq 6000 S2 flow cell. Analysis of the snRNAseq data was performed on facilities provided by the University of Cape Town’s ICTS High-Performance Computing team. Cell Ranger version 7.1.0 was used to map paired-end sequencing reads to the mouse reference transcriptome (refdata-gex-mm10-2020-A). The filtered feature-barcode matrix of each sample was converted into a Seurat object. Samples were processed according to the standard Seurat pipeline in R version 4.0.5 or 4.2.0⁷⁹.

Quality control (QC) involved removing nuclei with less than 500 unique molecular identifiers (UMIs), fewer than 250 genes expressed, log10GenesPerUMI less than 0.8, and/or a mitochondrial ratio greater than 0.2. In addition, all mitochondrial genes were excluded from the dataset. Three doublet identification tools were employed, including DoubletFinder⁸⁰, Scrublet⁸¹, and DoubletDecon⁸². For each sample, all doublets called by DoubletFinder, together with the intersection of doublets called by Scrublet and DoubletDecon, were filtered out. Following QC, normalisation and variance stabilisation were performed using the SCTransform() function. The mitochondrial ratio was regressed during this step. Integration of the different samples was performed via Seurat’s FindIntegrationAnchors() function using the top 2000 most variable features and 30 dimensions followed by the IntegrateData() function. A principle component analysis (PCA) was run on the integrated data followed by clustering using the FindNeighbors() and FindClusters() functions.

Several cluster annotation approaches were employed, including automated annotation, manual annotation, and Seurat’s label transfer method. For automated annotation, Seurat’s FindAllMarkers() function was used to identify differentially expressed genes (DEGs) between the clusters. A cluster resolution of 0.4 was set as the active identity and comprised 30 clusters to be annotated. An automated annotation tool known as SCSA⁸³ was used to obtain putative cluster annotations by comparing the list of DEGs per cluster obtained from FindAllMarkers() to reference datasets of known cell-type-specific markers, including SCSA’s built-in reference databases⁸⁴, as well as user-defined databases curated from mousebrain.org⁸⁵. Manual inspection methods involved visualising the expression of a smaller set of known cell-type-specific markers across the different clusters in bubble plots, as well as searching for the cell-type-specific markers in the list of cluster-specific DEGs from the FindAllMarkers() output. Using the automated and manual annotation methods, putative user-defined annotations for each cluster were generated. In addition, Seurat’s label transfer method was carried out using the Allen Mouse Brain atlas as the reference⁵⁷. Seurat’s FindTransferAnchors() function was used to find anchors between the reference and query datasets, with SCT specified as the normalisation method and the top 30 dimensions used. A relatively stringent filtering step was then performed to remove any nuclei that did not have an agreement between their broad user-defined and Allen Mouse Brain annotations. Following this filtering step, the user-defined coarse annotations were used for downstream analyses. The 4 samples were subsetted, and heatmaps were plotted to visualise the normalised average expression of several genes of interest across the pyramidal neuron and astrocyte clusters, respectively.

Immunohistochemistry

Mouse hippocampal organotypic brain slices (DIV 12) were incubated in 1 mM JF₆₀₈ dye in 2 ml carbogen-bubbled aCSF for 30 min and then transferred to a recording chamber perfused with aCSF for 1 h to allow the unbound dye to wash out. Slices were fixed in 4% paraformaldehyde in PBS for 15 min at 4 °C and permeabilized using wash buffer (0.3% Triton X-100 in PBS) for 30 min at 4 °C. Next, the slices were incubated in blocking solution (3% bovine serum albumin and 1% Triton X-100 in PBS) at room temperature (RT) for 4 h. This was followed by overnight incubation at 4 °C with rabbit anti-mouse GAD1 polyclonal antibody (Proteintech, catalogue number: 10408-1-AP) diluted in blocking solution (1:400). Slices were then washed three times and incubated with donkey anti-rabbit IgG conjugated to Alexa Fluor 488 (Abcam, catalogue number: ab150073; 1:500) in blocking solution for 4 h at RT. This was followed by three wash steps and nuclear staining with Hoechst (1:5000) in blocking solution for 15 min at RT. The slices were washed again and mounted.

Confocal imaging

Confocal images of live and fixed brain slices were obtained on an LSM 880 Airyscan confocal microscope (Carl Zeiss, ZEN SP 2 software) using a C-Apochromat 40 × objective. To visualise Hoechst a 405 nm laser was used, to visualise Alexa Fluor 488 a 488 nm laser was used, and to visualise ORCHID-expressing cells a 561 nm laser was used.

Data analysis

Data analysis was performed using custom scripts written in MATLAB R2022b (MathWorks). The image analysis pipeline extracted signals from a region of interest (ROI) targeted to the soma of the cell and calculated an integrated density of fluorescence (i.e., mean pixel brightness) in this region (F_s). Background subtraction was performed by selecting an ROI targeted to a representative area of the background, calculating an integrated density of fluorescence in this region, and subtracting this from F_s image-wise. For ORCHID recordings, this background fluorescence signal was smoothed with a smoothing window of 100 before being subtracted from F_s. Correction for photobleaching and low-frequency changes in signal was performed by fitting and subtracting a 9^th- or 10^th-order polynomial to and from the background-corrected F_s trace. The response to the GABA puff or optogenetic stimulation in the F_s trace was excluded from the fitting of this polynomial to avoid removing the signal from the trace. For recordings using Voltron2₅₄₉-ST with GABA_AR activation, this corrected F_s trace was smoothed with a smoothing window of 7, and ΔF/F₀ values were calculated using the maximum or minimum value of the smoothed signal post-puff and the mean of the smoothed baseline period prior to the puff. This same process was used for ORCHID recordings made during activity-dependent variation in DF_GABAA experiments, with the maximum or minimum value of the smoothed signal during an optogenetic stimulation being used to calculate ΔF/F₀. For all other in vitro ORCHID recordings, the 5 strobed optogenetic stimulation periods in the F_s trace were averaged before ΔF/F₀ was calculated using the mean during this averaged optogenetic stimulation, and the mean during the averaged baseline period prior to optogenetic stimulation onset. Due to the higher levels of spontaneous activity in vivo, for in vivo ORCHID recordings ΔF and F₀ were calculated using the mean fluorescence during a single optogenetic stimulus and an equivalent duration of recording on one side of the stimulus (these values termed ΔF_S and F_0S). To correct for neuropil contamination, the indices of these two periods (the period during the stimulus and the baseline period) were used to calculate an equivalent ΔF from the background fluorescence trace (ΔF_BG), which was subtracted from ΔF_S (giving ΔF_S-BG). ΔF/F₀ was then calculated using ΔF_S-BG and F_0S. The relationship between ΔV and ΔF/F₀ for Voltron2₆₀₈-ST and for Voltron2₅₄₉-ST, which were acquired and averaged from multiple simultaneously patched and imaged neurons, was linear (Fig. 1i, j and Supplementary Fig. 4c, d). Subsequently, a measured ΔF/F₀ in cells imaged but not patched could be translated to a ΔV, as previously performed by Cornejo et al.³⁰. This translated ΔV was then used to estimate DF_GABAA. This analysis pipeline used the WinWCP MATLAB Importer written by David⁸⁶.

Computational model

We simulated a neuron, using the pump-leak mechanism, as a single compartment within an extracellular environment consisting of fixed ion concentrations (i.e., an infinite bath model). The compartment was modelled as a cylinder with a diameter of 1 µm and a length of 25 µm. The ionic flow between the compartment and external environment was permitted via leak channels, Na⁺/K⁺-ATPase, and KCC2 transporters. For details of the simulation equations for ionic flux and volume change, see Düsterwald et al.³². In addition, GABA_ARs, HCO₃⁻, and pH regulation were added to the model in Düsterwald et al., which we describe below. All parameters are given in Table 1.

Table 1 Computational model parameters

Full size table

Fast GABAergic synaptic transmission mediated by GABA_ARs was modelled as an alpha function with a tau (τ) of 250 ms, and a max conductance g_{GABAA_max} of 10 nS, for t > t_onset:

$${g}_{{{{\rm{GABAA}}}}}={g}_{{{{\rm{GABAA}}}}\_{{\max }}}\cdot \frac{t-{t}_{{{{\rm{onset}}}}}}{\tau }\cdot e\left[\frac{-(t-{t}_{{{{\rm{onset}}}}}-\tau )}{\tau }\right]$$

(5)

Current through GABA_ARs $(I_{GABAA})$ was modelled as follows:

$${I}_{{{{\rm{GABAA}}}}}={I}_{{{{\rm{GAB}}}}{{{{\rm{A}}}}}_{{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}}}+{I}_{{{{\rm{GAB}}}}{{{{\rm{A}}}}}_{{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}}$$

(6)

$${I}_{{{{\rm{GAB}}}}{{{{\rm{A}}}}}_{{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}}}=\chi {\, \cdot \, g}_{{{{\rm{GABAA}}}}}\cdot \left({V}_{{{{\rm{m}}}}}-{E}_{{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}}\right)$$

(7)

$${I}_{{{{\rm{GABA}}}\_{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}=\left(1-\chi \right)\cdot \,{g}_{{{{\rm{GABAA}}}}}\cdot \left({V}_{{{{\rm{m}}}}}-{E}_{{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}\right)$$

(8)

$\chi$ represents the fraction of the total GABAergic current carried by Cl⁻ and is given by:

$$\chi=\frac{{E}_{{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}-{E}_{{{{\rm{GABAA}}}}}}{{E}_{{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}-{E}_{{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}}}$$

(9)

The reversal potentials E_Cl-, E_HCO3-, and E_GABAA were each updated throughout the simulation using the respective Nernst and Goldman-Hodgkin-Katz equations:

$${E}_{{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}}=\frac{R\cdot T}{F}{{\mathrm{ln}}} \, \, \left(\frac{{\left[{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}\right]}_{{{{\rm{i}}}}}}{{\left[{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}\right]}_{{{{\rm{o}}}}}}\right)$$

(10)

$${E}_{{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}}=\frac{R\cdot T}{F}{{\mathrm{ln}}} \, \, \left(\frac{{\left[{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}\right]}_{{{{\rm{i}}}}}}{{\left[{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}\right]}_{{{{\rm{o}}}}}}\right)$$

(11)

$${E}_{{{{\rm{GABAA}}}}}=\frac{R\cdot T}{F}{{\mathrm{ln}}} \, \, \left(\frac{\frac{4}{5}{\left[{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}\right]}_{{{{\rm{i}}}}}+\frac{1}{5}{\left[{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}\right]}_{{{{\rm{i}}}}}}{\frac{4}{5}{\left[{{{\rm{C}}}}{{{{\rm{l}}}}}^{-}\right]}_{{{{\rm{o}}}}}+\frac{1}{5}{\left[{{{\rm{HC}}}}{{{{\rm{O}}}}}_{3}^{-}\right]}_{{{{\rm{o}}}}}}\right)$$

(12)

Here, R is the ideal gas constant, F is Faraday’s constant, and T is temperature.

CO₂ was assumed to be equilibrium-distributed across the simulated plasma membrane. For a partial pressure of CO₂ (PCO₂) of 38 mmHg (5% CO₂), [CO₂] was calculated using Henry’s Law and a value for Henry’s Law constant (k_H) for CO₂ in the water of 0.031 (M/atm):

$$\left[{{{{\rm{CO}}}}}_{2}\right]={k}_{{{{\rm{H}}}}}\cdot {{{\rm{P}}}}{{{{\rm{CO}}}}}_{2}$$

(13)

Assuming that [CO₂] ≈ [H₂CO₃], [H₂CO₃] (inside and outside the compartment) was calculated as 1.55 mM.

The CO₂ hydration reaction outside the compartment was assumed to be under equilibrium:

$${{{{\rm{H}}}}}_{2}{{{\rm{O}}}}+{{{\rm{C}}}}{{{{\rm{O}}}}}_{2}\rightleftharpoons \,{{{{\rm{H}}}}}_{2}{{{{\rm{CO}}}}}_{3}\rightleftharpoons \,{{{{\rm{H}}}}}^{+}+{{{{\rm{HCO}}}}}_{3}^{-}$$

(14)

Using the Henderson Hasselbalch equation and a pK_a value of 6.1 as the dissociation constant for H₂CO_3, [HCO₃⁻]_o was calculated at a pH of 7.4:

$${\left[{{{{\rm{HCO}}}}}_{3}^{-}\right]}_{{{{\rm{o}}}}}=\left[{{{{\rm{H}}}}}_{2}{{{{\rm{CO}}}}}_{3}\right]\cdot {10}^{{pH}-{{pK}}_{a}}$$

(15)

This gives a [HCO₃⁻]_o of 31 mM.

Inside the compartment, pH was held constant at 7.2, and the production HCO₃⁻ by the forward reaction of the CO₂ hydration reaction (Eq. 14) was calculated using the forward rate equation and a forward rate constant K_f of 1 × 10³:

$${{{{\rm{Rate}}}}}_{{{{\rm{forward}}}}}={K}_{{{{\rm{f}}}}}\cdot \left[{{{{\rm{H}}}}}_{2}{{{{\rm{CO}}}}}_{3}\right]$$

(16)

The removal of HCO₃⁻ by the reverse reaction was calculated using the reverse rate equation and a reverse rate constant K_r of 2.539 × 10⁹:

$${{{{\rm{Rate}}}}}_{{{{\rm{reverse}}}}}={K}_{{{{\rm{r}}}}}\cdot \left[{{{{\rm{H}}}}}^{+}\right]\cdot \left[{{{{\rm{HCO}}}}}_{3}^{-}\right]$$

(17)

H⁺ produced or removed by the forward or reverse reactions was simulated as being immediately exchanged with Na⁺ via the Na⁺/H⁺ exchanger⁸⁷ in order to maintain intracellular pH at 7.2.

Statistics & reproducibility

Statistical measurements were performed using GraphPad Prism 7. Data were assessed for normality, and parametric or non-parametric tests were selected as required. Data is reported as mean ± SEM unless stated otherwise. No statistical method was used to predetermine the sample size. Sample sizes were chosen based on accepted standards in the field, aiming to minimise the number of animals sacrificed while still being able to demonstrate the robustness of any effects. No data were excluded from the analyses, with three predetermined exceptions. For ORCHID recordings, cells which were visually determined to have blebbed during recordings were excluded. For in vitro experimental data gathered using soma-directed puffs of GABA, recordings in which a movement artefact was present were excluded. Patch-clamp recordings in which the access resistance was greater than 30 MΩ were excluded. Mice were randomly selected for hippocampal slices or imaging experiments and were randomly assigned to experimental groups. The selection of cells for imaging was random, and within-cell comparisons were performed wherever possible. Investigators were blinded to experimental conditions where possible during data acquisition and analysis, although morphological differences between cells and technical differences between techniques meant that blinding was often not possible. Voltage imaging and electrophysiology experiments were performed independently for each individual cell in an experimental dataset.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Source data are provided with this paper as a Source Data file. Imaging raw data are available upon request, due to the size of the dataset. The expected timeframe for response to access requests is approximately two weeks, accounting for communication and upload time, and depending on the size of the data requested. Source data are provided in this paper.

Code availability

All data was analysed using custom scripts created in MATLAB, which are available at: https://github.com/joshs08/ORCHID (https://doi.org/10.5281/zenodo.13684771). Code used for the simulations is accessible at: https://github.com/Eran707/Single-Cell-Simulator (https://doi.org/10.5281/zenodo.13376723).

References

Kaila, K., Price, T. J., Payne, J. A., Puskarjov, M. & Voipio, J. Cation-chloride cotransporters in neuronal development, plasticity and disease. Nat. Rev. Neurosci. 15, 637–654 (2014).
Article CAS PubMed PubMed Central Google Scholar
Kyrozis, A. & Reichling, D. B. Perforated-patch recording with gramicidin avoids artifactual changes in intracellular chloride concentration. J. Neurosci. Methods 57, 27–35 (1995).
Article CAS PubMed Google Scholar
Farrant, M. & Kaila, K. The cellular, molecular and ionic basis of GABAA receptor signalling. Prog. Brain Res. 160, 59–87 (2007).
Article CAS PubMed Google Scholar
Wright, R., Raimondo, J. V. & Akerman, C. J. Spatial and temporal dynamics in the ionic driving force for GABA(A) receptors. Neural Plast. 2011, 728395 (2011).
Article CAS PubMed PubMed Central Google Scholar
Untiet, V. Astrocytic chloride regulates brain function in health and disease. Cell Calcium 118, 102855 (2024).
Article CAS PubMed Google Scholar
Raimondo, J. V., Richards, B. A. & Woodin, M. A. Neuronal chloride and excitability — the big impact of small changes. Curr. Opin. Neurobiol. 43, 35–42 (2017).
Article CAS PubMed Google Scholar
Huberfeld, G. et al. Perturbed chloride homeostasis and GABAergic signaling in human temporal lobe epilepsy. J. Neurosci. 27, 9866–9873 (2007).
Article CAS PubMed PubMed Central Google Scholar
Hyde, T. M. et al. Expression of GABA signaling molecules KCC2, NKCC1, and GAD1 in cortical development and schizophrenia. J. Neurosci. 31, 11088–11095 (2011).
Article CAS PubMed PubMed Central Google Scholar
Price, T. J., Cervero, F., Gold, M. S., Hammond, D. L. & Prescott, S. A. Chloride regulation in the pain pathway. Brain Res. Rev. 60, 149–170 (2009).
Article CAS PubMed Google Scholar
Tyzio, R. et al. Oxytocin-mediated GABA inhibition during delivery attenuates autism pathogenesis in rodent offspring. Science 343, 675–679 (2014).
Article ADS CAS PubMed Google Scholar
van Rheede, J. J., Richards, B. A. & Akerman, C. J. Sensory-evoked spiking behavior emerges via an experience-dependent plasticity mechanism. Neuron 87, 1050–1062 (2015).
Article PubMed Google Scholar
Ben-Ari, Y. Excitatory actions of GABA during development: the nature of the nurture. Nat. Rev. Neurosci. 3, 728–739 (2002).
Article CAS PubMed Google Scholar
Wagner, S., Castel, M., Gainer, H. & Yarom, Y. GABA in the mammalian suprachiasmatic nucleus and its role in diurnal rhythmicity. Nature 387, 598–603 (1997).
Article ADS CAS PubMed Google Scholar
Pracucci, E. et al. Daily rhythm in cortical chloride homeostasis underpins functional changes in visual cortex excitability. Nat. Commun. 14, 7108 (2023).
Alfonsa, H. et al. Intracellular chloride regulation mediates local sleep pressure in the cortex. Nat. Neurosci. 26, 64–78 (2023).
Article CAS PubMed Google Scholar
Untiet, V. et al. Astrocytic chloride is brain state dependent and modulates inhibitory neurotransmission in mice. Nat. Commun. 14, 1871 (2023).
Article CAS PubMed PubMed Central Google Scholar
Ellender, T. J., Raimondo, J. V., Irkle, A., Lamsa, K. P. & Akerman, C. J. Excitatory effects of parvalbumin-expressing interneurons maintain hippocampal epileptiform activity via synchronous afterdischarges. J. Neurosci. 34, 15208–15222 (2014).
Article PubMed PubMed Central Google Scholar
Verkman, A. S., Sellers, M. C., Chao, A. C., Leung, T. & Ketcham, R. Synthesis and characterization of improved chloride-sensitive fluorescent indicators for biological applications. Anal. Biochem. 178, 355–361 (1989).
Article CAS PubMed Google Scholar
Weilinger, N. L. et al. KCC2 drives chloride microdomain formation in dendritic blebbing. Cell Rep. 41, 111556 (2022).
Article CAS PubMed Google Scholar
Kaneko, H., Putzier, I., Frings, S., Kaupp, U. B. & Gensch, T. Chlorideaccumulation in mammalian olfactory sensory neurons. J. Neurosci. 24, 7931–7938 (2004).
Article CAS PubMed PubMed Central Google Scholar
Arosio, D. & Ratto, G. M. Twenty years of fluorescence imaging of intracellular chloride. Front. Cell. Neurosci. 8, 1–12 (2014).
Article Google Scholar
Arosio, D. et al. Simultaneous intracellular chloride and pH measurements using a GFP-based sensor. Nat. Methods 7, 516–518 (2010).
Article CAS PubMed Google Scholar
Kuner, T. & Augustine, G. J. A genetically encoded ratiometric indicator for chloride: capturing chloride transients in cultured hippocampal neurons. Neuron 27, 447–459 (2000).
Article CAS PubMed Google Scholar
Sulis Sato, S. et al. Simultaneous two-photon imaging of intracellular chloride concentration and pH in mouse pyramidal neurons in vivo. Proc. Natl. Acad. Sci. USA 114, E8770–E8779 (2017).
Raimondo, J. V. et al. A genetically-encoded chloride and pH sensor for dissociating ion dynamics in the nervous system. Front. Cell. Neurosci. 7, 202 (2013).
Article CAS PubMed PubMed Central Google Scholar
Abdelfattah, A. S. et al. Sensitivity optimization of a rhodopsin-based fluorescent voltage indicator. Neuron 111, 1547–1563 (2023).
Article CAS PubMed PubMed Central Google Scholar
Govorunova, E. G., Sineshchekov, O. A., Janz, R., Liu, X. & Spudich, J. L. Natural light-gated anion channels: A family of microbial rhodopsins for advanced optogenetics. Science 349, 647–650 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Abdelfattah, A. S. et al. Bright and photostable chemigenetic indicators for extended in vivo voltage imaging. Science 365, 699–704 (2019).
Article ADS CAS PubMed Google Scholar
Grimm, J. B. et al. A general method to optimize and functionalize red-shifted rhodamine dyes. Nat. Methods 17, 815–821 (2020).
Article CAS PubMed PubMed Central Google Scholar
Cornejo, V. H., Ofer, N. & Yuste, R. Voltage compartmentalization in dendritic spines in vivo. Science 375, 82–86 (2022).
Article ADS CAS PubMed Google Scholar
Glykys, J., Dzhala, V., Egawa, K. & Balena, T. Local impermeant anions establish the neuronal chloride concentration. Science 670, 676 (2014).
Google Scholar
Düsterwald, K. M. et al. Biophysical models reveal the relative importance of transporter proteins and impermeant anions in chloride homeostasis. ELife 7, 216150 (2018).
Article Google Scholar
Graham, R. T. et al. Optogenetic stimulation reveals a latent tipping point in cortical networks during ictogenesis. Brain 146, 2814–2827 (2023).
Article PubMed Google Scholar
Rivera, C. et al. Mechanism of activity-dependent downregulation of the neuron-specific K-Cl cotransporter KCC2. J. Neurosci. 24, 4683–4691 (2004).
Article CAS PubMed PubMed Central Google Scholar
Nguyen, T. D. et al. Astrocytic NKCC1 inhibits seizures by buffering Cl− and antagonizing neuronal NKCC1 at GABAergic synapses. Epilepsia 64, 3389–3403 (2023).
Article CAS Google Scholar
Ma, B.-F., Xie, M.-J. & Zhou, M. Bicarbonate efflux via GABA(A) receptors depolarizes membrane potential and inhibits two-pore domain potassium channels of astrocytes in rat hippocampal slices. Glia 60, 1761–1772 (2012).
Article PubMed Central Google Scholar
Virtanen, M. A., Uvarov, P., Hübner, C. A. & Kaila, K. NKCC1, an Elusive molecular target in brain development: Making sense of the existing data. Cells 9, 2607 (2020).
Article CAS PubMed Central Google Scholar
Payne, J. A. Functional characterization of the neuronal-specific K-Cl cotransporter: implications for [K+]o regulation. Am. J. Physiol. Cell Physiol. 273, C1516–C1525 (1997).
Article CAS Google Scholar
Engels, M. et al. Glial chloride homeostasis under transient ischemic stress. Front. Cell. Neurosci. 15, 735300 (2021).
Article CAS PubMed Central Google Scholar
Hochbaum, D. R. et al. All-optical electrophysiology in mammalian neurons using engineered microbial rhodopsins. Nat. Methods 11, 825–833 (2014).
Article CAS PubMed PubMed Central Google Scholar
Adam, Y. et al. Voltage imaging and optogenetics reveal behaviour-dependent changes in hippocampal dynamics. Nature 569, 413–417 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Wiegert, J. S., Mahn, M., Prigge, M., Printz, Y. & Yizhar, O. Silencing neurons: Tools, applications, and experimental constraints. Neuron 95, 504–529 (2017).
Article CAS PubMed PubMed Central Google Scholar
Bormann, J., Hamill, O. P. & Sakmann, B. Mechanism of anion permeation through channels gated by glycine and gamma-aminobutyric acid in mouse cultured spinal neurones. J. Physiol. 385, 243–286 (1987).
Article ADS CAS PubMed PubMed Central Google Scholar
Fatima-Shad, K. & Barry, P. H. Anion permeation in GABA- and glycine-gated channels of mammalian cultured hippocampal neurons. Proc. R. Soc. B Biol. Sci. 253, 69–75 (1993).
Article ADS CAS Google Scholar
Kaila, K., Pasternack, M., Saarikoski, J., Voipio, J. & Britain, G. Influence of GABA-gated bicarbonate conductance on potential, current and intracellular chloride in crayfish muscle fibres. J. Physiol. 416, 161–181 (1989).
Article CAS PubMed PubMed Central Google Scholar
Kaila, K., Voipio, J., Paalasmaa, P., Pasternack, M. & Deisz, R. The role of bicarbonate in GABAA receptor-mediated IPSPs of rat neocortical neurones. J. Physiol. 464, 273 (1993).
Article CAS PubMed PubMed Central Google Scholar
Messier, J. E., Chen, H., Cai, Z.-L. & Xue, M. Targeting light-gated chloride channels to neuronal somatodendritic domain reduces their excitatory effect in the axon. ELife 7, e38506 (2018).
Article PubMed PubMed Central Google Scholar
Burman, R. J. et al. Optogenetic determination of dynamic and cell-type-specific inhibitory reversal potentials. J. Neurosci. 44, e1392232024 (2024).
Grimley, J. S. et al. Visualization of synaptic inhibition with an optogenetic sensor developed by cell-free protein engineering automation. J. Neurosci.33, 16297–16309 (2013).
Article CAS PubMed PubMed Central Google Scholar
Rühl, P. et al. Monitoring of compound resting membrane potentials of cell cultures with ratiometric genetically encoded voltage indicators. Commun. Biol. 4, 1–11 (2021).
Article Google Scholar
Voipio, J. et al. Comment on ‘Local impermeant anions establish the neuronal chloride concentration. Science 345, 1130 (2014).
Article CAS PubMed Google Scholar
Burman, R. J. et al. Excitatory GABAergic signalling is associated with benzodiazepine resistance in status epilepticus. Brain 142, 3482–3501 (2019).
Article PubMed PubMed Central Google Scholar
Burman, R. J. et al. Why won’t it stop? The dynamics of benzodiazepine resistance in status epilepticus. Nat. Rev. Neurol. 18, 428–441 (2022).
Article CAS PubMed Google Scholar
Călin, A., Waseem, T., Raimondo, J. V., Newey, S. E. & Akerman, C. J. A genetically targeted ion sensor reveals distinct seizure-related chloride and pH dynamics in GABAergic interneuron populations. iScience 26, 106363 (2023).
Article ADS PubMed PubMed Central Google Scholar
Sasaki, T. et al. Application of an optogenetic byway for perturbing neuronal activity via glial photostimulation. Proc. Natl. Acad. Sci. USA 109, 20720–20725 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Egawa, K., Yamada, J., Furukawa, T., Yanagawa, Y. & Fukuda, A. Cl⁻ homeodynamics in gap junction-coupled astrocytic networks on activation of GABAergic synapses. J. Physiol. 591, 3901–3917 (2013).
Article CAS PubMed PubMed Central Google Scholar
Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241 (2021).
Article CAS PubMed PubMed Central Google Scholar
Burman, R. J., Brodersen, P. J. N., Raimondo, J. V., Sen, A. & Akerman, C. J. Active cortical networks promote shunting fast synaptic inhibition in vivo. Neuron, 111, 3531–3540 (2023).
Duebel, J. et al. Two-photon imaging reveals somatodendritic chloride gradient in retinal ON-type bipolar cells expressing the biosensor Clomeleon. Neuron 49, 81–94 (2006).
Article CAS PubMed Google Scholar
Szabadics, J. et al. Excitatory effect of GABAergic axo-axonic cells in cortical microcircuits. Science 311, 233 (2006).
Article ADS CAS PubMed Google Scholar
Glickfeld, L. L., Roberts, J. D., Somogyi, P. & Scanziani, M. Interneurons hyperpolarize pyramidal cells along their entire somatodendritic axis. Nat. Neurosci. 12, 21–23 (2009).
Article CAS PubMed Google Scholar
Földy, C., Lee, S.-H., Morgan, R. J. & Soltesz, I. Regulation of fast-spiking basket cell synapses by the chloride channel ClC-2. Nat. Neurosci. 13, 1047–1049 (2010).
Article PubMed PubMed Central Google Scholar
Currin, C. B., Trevelyan, A. J., Akerman, C. J. & Raimondo, J. V. Chloride dynamics alter the input-output properties of neurons. PLoS Comput. Biol. 16, 1–22 (2020).
Article Google Scholar
Holst, J., Vignali, K. M., Burton, A. R. & Vignali, D. A. A. Rapid analysis of T-cell selection in vivo using T cell-receptor retrogenic mice. Nat. Methods 3, 191–197 (2006).
Article CAS PubMed Google Scholar
Li, S.-J., Vaughan, A., Sturgill, J. F. & Kepecs, A. A viral receptor complementation strategy to overcome CAV-2 tropism for efficient retrograde targeting of neurons. Neuron 98, 905–917 (2018).
Article CAS PubMed Google Scholar
Fenno, L. E. et al. Targeting cells with single vectors using multiple-feature Boolean logic. Nat. Methods 11, 763–772 (2014).
Article CAS PubMed PubMed Central Google Scholar
Stoppini, L., Buchs, P.-A. & Muller, D. A simple method for organotypic cultures of nervous tissue. J Neurosci Meth. 37, 173–182 (1991).
Article CAS Google Scholar
Raimondo, J. V. et al. Tight coupling of astrocyte pH dynamics to epileptiform activity revealed by genetically encoded pH sensors. J. Neurosci. 36, 7002–7013 (2016).
Article CAS PubMed PubMed Central Google Scholar
O’Connor, E., Sontheimer, H. & Ransom, B. Rat hippocampal astrocytes exhibit electrogenic sodium-bicarbonate co-transport. J. Neurophysiol. 72, 2580–2589 (1994).
Article PubMed Google Scholar
McKhann, G. M., D’Ambrosio, R. & Janigro, D. Heterogeneity of astrocyte resting membrane potentials and intercellular coupling revealed by whole-cell and gramicidin-perforated patch recordings from cultured neocortical and hippocampal slice astrocytes. J. Neurosci. 17, 6850–6863 (1997).
Article CAS PubMed PubMed Central Google Scholar
Forman, C. J. et al. Openspritzer: an open hardware pressure ejection system for reliably delivering picolitre volumes. Sci. Rep. 7, 2188 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Streit, P., Thompson, S. M. & Gahwiler, B. H. Anatomical and physiological properties of GABAergic neurotransmission in organotypic slice cultures of rat hippocampus. Eur. J. Neurosci. 1, 603–615 (1989).
Article PubMed Google Scholar
Ilie, A., Raimondo, J. V. & Akerman, C. J. Adenosine release during seizures attenuates GABAA receptor-mediated depolarization. J Neurosci. 32, 5321–5332 (2012).
Article CAS PubMed PubMed Central Google Scholar
Raimondo, J. V., Kay, L., Ellender, T. J. & Akerman, C. J. Optogenetic silencing strategies differ in their effects on inhibitory synaptic transmission. Nat. Neurosci. 15, 1102–1104 (2012).
Article CAS PubMed PubMed Central Google Scholar
Wright, R. et al. Neuronal chloride regulation via KCC2 is modulated through a GABAB receptor protein complex. J. Neurosci. 37, 5447–5462 (2017).
Article CAS PubMed PubMed Central Google Scholar
Cheetham, C. E. J., Grier, B. D. & Belluscio, L. Bulk regional viral injection in neonatal mice enables structural and functional interrogation of defined neuronal populations throughout targeted brain areas. Front. Neural Circuits 9, 72 (2015).
Oomoto, I. et al. Protocol for cortical-wide field-of-view two-photon imaging with quick neonatal adeno-associated virus injection. STAR Protoc. 2, 101007 (2021).
Article CAS PubMed PubMed Central Google Scholar
Yardeni, T., Eckhaus, M., Morris, H. D., Huizing, M. & Hoogstraten-Miller, S. Retro-orbital injections in mice. Lab Anim. 40, 155 (2011).
Article Google Scholar
Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587 (2021).
Article CAS PubMed Central Google Scholar
McGinnis, C. S., Murrow, L. M. & Gartner, Z. J. Doubletfinder: Doublet detection in single-cell RNA sequencing data using artificial nearest neighbors. Cell Syst. 8, 329–337 (2019).
Article CAS PubMed PubMed Central Google Scholar
Wolock, S. L., Lopez, R. & Klein, A. M. Scrublet: Computational identification of cell doublets in single-cell transcriptomic Data. Cell Syst. 8, 281–291 (2019).
Article CAS PubMed PubMed Central Google Scholar
DePasquale, E. A. K. et al. DoubletDecon: Deconvoluting doublets from single-cell RNA-sequencing data. Cell Rep. 29, 1718–1727 (2019).
Article CAS PubMed PubMed Central Google Scholar
Cao, Y., Wang, X. & Peng, G. SCSA: A cell type annotation tool for single-cell RNA-seq data. Front. Genet. 11, 490 (2020).
Zhang, X. et al. CellMarker: a manually curated resource of cell markers in human and mouse. Nucleic Acids Res. 47, D721–D728 (2019).
Article ADS CAS PubMed Google Scholar
Zeisel, A. et al. Molecular architecture of the mouse nervous system. Cell 174, 999–1014 (2018).
Article CAS PubMed PubMed Central Google Scholar
WinWCP MATLAB Importer. https://www.mathworks.com/matlabcentral/fileexchange/49539-winwcp-matlab-importer (2024).
Chesler, M. Regulation and modulation of pH in the brain. Physiol. Rev. 83, 1183–1221 (2003).
Article CAS PubMed Google Scholar
Hernández, J. A. & Cristina, E. Modeling cell volume regulation in nonexcitable cells: the roles of the Na+ pump and of cotransport systems. Am. J. Physiol. Cell Physiol. 275, C1067–C1080 (1998).
Article Google Scholar
Lodish, H. et al. in Molecular Cell Biology (New York, 2009).

Download references

Acknowledgements

We would like to thank members of the Raimondo lab for advice and comments and Suzanne Duncan for assistance in naming ORCHID. We thank Sarah E. Plutkis, Katie L. Holland, Jonathan B. Grimm, and Luke D. Lavis (Janelia Research Campus) for Janelia Fluor HaloTag ligands, and Yoav Adam for providing advice on GEVI imaging. We thank Rodney Lucas for assistance in performing animal experiments, the Africa Microscopy Initiative Imaging Centre for their support and assistance with imaging, and Christina Steyn, Dorit Hockman, Muazzam Jacobs and Christopher Dulla for advice and support in performing snRNAseq analysis. The research leading to these results has received support from the Gabriel Foundation (J.V.R.), a Wellcome Trust Seed Award (214042/Z/18/Z) (J.V.R.), the FLAIR Fellowship Programme (FLR\R1\190829): a partnership between the African Academy of Sciences and the Royal Society funded by the UK Government’s Global Challenges Research Fund (JVR), a Wellcome Trust International Intermediate Fellowship (222968/Z/21/Z) (J.V.R.), ERC Grant Agreement 617670 (C.J.A.), and the project was supported by funding from BBSRC project BB/S007938/1 (C.J.A.).

Author information

Authors and Affiliations

Division of Cell Biology, Department of Human Biology, University of Cape Town, Cape Town, South Africa
Joshua S. Selfe, Teresa J. S. Steyn, Eran F. Shorer, Kira M. Düsterwald, Ariel Z. Kraitzick & Joseph V. Raimondo
Neuroscience Institute, University of Cape Town, Cape Town, South Africa
Joshua S. Selfe, Teresa J. S. Steyn, Eran F. Shorer, Kira M. Düsterwald, Ariel Z. Kraitzick & Joseph V. Raimondo
Department of Neurology, School of Medicine, Johns Hopkins Hospital, Baltimore, Maryland, United States of America
Eran F. Shorer
Department of Pharmacology, University of Oxford, Oxford, United Kingdom
Richard J. Burman, Sarah E. Newey & Colin J. Akerman
Gatsby Computational Neuroscience Unit, University College London, London, United Kingdom
Kira M. Düsterwald
Department of Neuroscience, Brown University, Providence, Rhode Island, United States of America
Ahmed S. Abdelfattah
Carney Institute for Brain Science, Brown University, Providence, Rhode Island, United States of America
Ahmed S. Abdelfattah
Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, Virginia, United States of America
Eric R. Schreiter
Wellcome Centre for Infectious Disease Research in Africa, Institute of Infectious Disease and Molecular Medicine, University of Cape Town, Cape Town, South Africa
Joseph V. Raimondo

Authors

Joshua S. Selfe
View author publications
Search author on:PubMed Google Scholar
Teresa J. S. Steyn
View author publications
Search author on:PubMed Google Scholar
Eran F. Shorer
View author publications
Search author on:PubMed Google Scholar
Richard J. Burman
View author publications
Search author on:PubMed Google Scholar
Kira M. Düsterwald
View author publications
Search author on:PubMed Google Scholar
Ariel Z. Kraitzick
View author publications
Search author on:PubMed Google Scholar
Ahmed S. Abdelfattah
View author publications
Search author on:PubMed Google Scholar
Eric R. Schreiter
View author publications
Search author on:PubMed Google Scholar
Sarah E. Newey
View author publications
Search author on:PubMed Google Scholar
Colin J. Akerman
View author publications
Search author on:PubMed Google Scholar
Joseph V. Raimondo
View author publications
Search author on:PubMed Google Scholar

Contributions

J.S.S., A.S.A., E.R.S., C.J.A., and J.V.R. conceptualised the study. A.S.A. and E.R.S. provided tools used in the study. J.S.S., T.J.S.S., R.J.B., A.Z.K., E.F.S., K.M.D., S.E.N., and J.V.R. carried out the investigations. J.S.S., C.J.A., and J.V.R. wrote the manuscript. J.V.R. supervised the study, and funding acquisition was by C.J.A. and J.V.R.

Corresponding author

Correspondence to Joseph V. Raimondo.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information (download PDF )

Reporting Summary (download PDF )

Peer Review File (download PDF )

Source data

Source Data (download XLSX )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Selfe, J.S., Steyn, T.J.S., Shorer, E.F. et al. All-optical reporting of inhibitory receptor driving force in the nervous system. Nat Commun 15, 8913 (2024). https://doi.org/10.1038/s41467-024-53074-y

Download citation

Received: 23 January 2024
Accepted: 30 September 2024
Published: 16 October 2024
Version of record: 16 October 2024
DOI: https://doi.org/10.1038/s41467-024-53074-y

This article is cited by

Functional KCC2 expression marks an evolutionarily conserved population of early-maturing interneurons in the perinatal cortex
- Rakenduvadhana Szrinivasan
- Kalevi Trontti
- Claudio Rivera
Nature Communications (2025)