In Vivo Two-Photon Voltage Imaging with Sulfonated Rhodamine Dyes.

Optical methods that rely on fluorescence for mapping changes in neuronal membrane potential in the brains of awake animals provide a powerful way to interrogate the activity of neurons that underlie neural computations ranging from sensation and perception to learning and memory. To achieve this goal, fluorescent indicators should be bright, highly sensitive to small changes in membrane potential, nontoxic, and excitable with infrared light. We report a new class of fluorescent, voltage-sensitive dyes: sulfonated rhodamine voltage reporters (sRhoVR), synthetic fluorophores with high voltage sensitivity, excellent two-photon performance, and compatibility in intact mouse brains. sRhoVR dyes are based on a tetramethyl rhodamine fluorophore coupled to a phenylenevinylene molecular wire/diethyl aniline voltage-sensitive domain. When applied to cells, sRhoVR dyes localize to the plasma membrane and respond to membrane depolarization with a fluorescence increase. The best of the new dyes, sRhoVR 1, displays a 44% ΔF/F increase in fluorescence per 100 mV change, emits at 570 nm, and possesses excellent two-photon absorption of approximately 200 GM at 840 nm. sRhoVR 1 can detect action potentials in cultured rat hippocampal neurons under both single- and two-photon illumination with sufficient speed and sensitivity to report on action potentials in single trials, without perturbing underlying physiology or membrane properties. The combination of speed, sensitivity, and brightness under two-photon illumination makes sRhoVR 1 a promising candidate for in vivo imaging in intact brains. We show sRhoVR powerfully complements electrode-based modes of neuronal activity recording in the mouse brain by recording neuronal transmembrane potentials from the neuropil of layer 2/3 of the mouse barrel cortex in concert with extracellularly recorded local field potentials (LFPs). sRhoVR imaging reveals robust depolarization in response to whisker stimulation; concurrent electrode recordings reveal negative deflections in the LFP recording, consistent with the canonical thalamocortical response. Importantly, sRhoVR 1 can be applied in mice with chronic optical windows, presaging its utility in dissecting and resolving voltage dynamics using two-photon functional imaging in awake, behaving animals.

Introduction

Emergent brain properties arise from the coordinated firing of neurons. The flux of ions into and out of these specialized cells give rise to changes in transmembrane potentials. Intracellular electrophysiological recordings provide the most accurate determination of membrane potential in single neurons, offering resolution of large action potential spikes or even synaptic currents, but are difficult to implement for more than one neuron for measurements of circuit activity in vivo.[1] A partial solution to the problem of recording from large numbers of neurons intracellularly is to record the ensemble extracellular potentials neuronal populations.[2] These extracellular electrical signals, reflecting contributions from multiple neurons, have the advantage that they can be readily obtained in vivo, and modern fabrication methods enable massively multiplexed recordings.[3,4]

The slow component of the extracellularly recorded signals, local field potentials, or local field potentials (LFPs) provides an integration of the synaptic events of large numbers of neurons, enabling high density recordings across a variety of species for use in understanding fundamental neurobiology and driving brain-machine interfaces. Despite the utility of LFP recordings for measuring ensemble neuronal activity in diverse brain contexts, they suffer from ambiguity in interpretation of the underlying membrane potential dynamics. For example, excitatory inputs and depolarization close to the brain surface produce the same LFP signature as inhibitory inputs and hyperpolarization in deep cortical layers.[5] Resolving this ambiguity requires prior knowledge of underlying neuronal architecture and physiology, complementary functional imaging approaches, and/or intricate computational models.[5] Further, LFP signals from the cerebral cortex of the brain lack a true depth resolution, because a signal recorded from the surface can arise due to synaptic inputs to the deep layers and vice versa.[5]

Fluorescence imaging of neuronal activity offers a promising strategy for complementing LFP recordings to resolve these critical ambiguities. LFP recordings have been combined with activity imaging using both small molecule[6,7] and genetically encoded[8,9] Ca2+ indicators as well as small molecule[10,11] and genetically encoded voltage indicators.[12] Voltage imaging with voltage-sensitive fluorescent indicators provides a direct readout of the local transmembrane potential and an important complement for interpreting the extracellular potential recordings of LFPs. However, most studies employ single-photon excitation coupled with widefield, epifluorescence imaging in superficial cortical brain areas. This type of imaging results in depth-averaged voltage signals that lack the depth resolution required to calibrate the interpretation of LFP.[10−14] Additionally, the use of powerful electrochromic voltage-sensitive dyes like the popular RH-1691 or RH-1692[15−17] compounds requires off-peak excitation and emission, resulting in approximately 99% photon loss compared to on-peak dye excitation and emission.[18] A recent study employed a genetically encoded voltage indicator targeted to defined cells, coupled with single-photon fiber photometry, to optically measure transmembrane potential in concert with electrophysiological field recordings.[19] By constraining indicator expression to specific cells, some depth resolution could be achieved, even in the absence of an imaging approach. We envisioned that two-photon imaging with untargeted voltage sensitive dyes would offer a more generalizable approach to directly record changes in transmembrane potential, without the need for transgenic animals or cellular resolution. In contrast to single-photon excitation, two-photon excitation offers optical sectioning due to a sharp fall of fluorescence intensity with distance from the focal plane. Therefore, to achieve the full potential of voltage imaging, especially in the context of existing technologies for recording extracellular potentials of large populations of neurons, requires indicators that are bright, excellent two-photon absorbers, highly voltage sensitive, and efficient with excitation and emission photons.

To address these challenges, we recently initiated a program to develop voltage-sensitive fluorescent indicators with sufficient speed, sensitivity, and brightness to monitor rapid changes in membrane potential. Our strategy relies on the use of photoinduced electron transfer, or PeT, as a voltage-sensitive trigger. Voltage-sensitive fluorophores, which we generically refer to as VoltageFluors or VF dyes, are designed to partition into the outer leaflet of the plasma membrane. Their fluorescence is diminished when cells are hyperpolarized. Correspondingly, when cells depolarize, VF fluorescence brightens. We hypothesize this is a result of the changing transmembrane electrochemical potential altering the rate of PeT and therefore modulating fluorescence.[20,21] Consistent with this hypothesis, VF dyes display rapid response kinetics, enabling them to detect action potentials in single-trial optical recordings, possess sensitivities of up to 63% ΔF/F per 100 mV,[22] can be tuned across a range of wavelength spanning the color palette from blue to far-red,[23−26] and can operate using two-photon illumination.[24]

A recent VF dye based on a rhodol chromophore, RhodolVoltageFluor-5, or RVF5, made use of a chlorinated, pyrrolidine-based rhodol. RVF5 possessed good photostability and moderate voltage sensitivity (28% ΔF/F per 100 mV), enabling detection of action potentials in cultured hippocampal neurons under conventional wide-field illumination and spiking events in mouse brain slices using two-photon illumination.[24] Building on this result, we wondered whether we could access VF dyes that made use of fluorophores with even higher two-photon absorption cross sections and even longer wavelength emission for use in in vivo applications (Scheme ). In this regard, rhodamine dyes, with symmetrical nitrogen substitution at the 3′ and 6′ positions of the xanthene chromophore, present themselves as an ideal choice because they have large two-photon absorption cross sections (σTPA), emission profiles bathochromically separated from typical fluoresceins and rhodols, and good photostability. We recently disclosed the synthesis of the Rhodamine Voltage Reporter (RhoVR) family of tetramethylrhodamine (TMR)-based voltage sensors, which incorporate an ortho-carboxamide group to prevent spirocyclization of the rhodamine fluorophore and ensure localization of RhoVRs to the outer leaflet of the plasma membrane.[25] The best of these dyes, RhoVR 1, shows improved sensitivity (47% ΔF/F per 100 mV) and red-shifted excitation and emission relative to RVF5,[24] but has lower solubility and requires several synthetic steps after generation of the key fluorophore-molecular wire scaffold.[25] We hypothesized that a sulfonated version of RhoVR would retain the essential characteristics of the carboxamide dye, but with improved solubility and fewer overall synthetic steps (Scheme ). We now disclose the design, synthesis, characterization, and application of sulfonated Rhodamine Voltage Reporters, or sRhoVRs (Scheme ). This study is enabled by a new synthetic route to ortho-sulfonated rhodamine dyes that provides regioisomerically pure sulfonated rhodamines in good yields and in just three steps from readily available starting materials. The best of the new indicators, sRhoVR 1, features good voltage sensitivity in HEK cells (44% ΔF/F per 100 mV), possesses a large σTPA (>200 GM at 840 nm), can detect action potentials in rat neurons in a single trial under widefield, confocal, and two-photon microscopy, and can be employed in vivo in both anesthetized and awake mice to report on the evolution of voltage changes during sensory stimulation.

Scheme 1

Molecular Redesign Enables in Vivo Voltage Imaging with Sulfonated Rhodamine Voltage Reporters (sRhoVRs)

Results

Design and Synthesis of sRhoVR Dyes

Preparation of sRhoVR dyes can be achieved through a Pd-catalyzed Heck coupling reaction between halogenated, sulfonated rhodamines (5 or 6) and substituted styrenes. We accessed sulfonated rhodamines dyes[27] from readily prepared 4-bromo-2-sulfobenzaldehyde[22] (3) and a novel 5-bromo-2-sulfobenzaldehyde (4). Both 3 and 4 could be generated in quantitative yield via reaction of bisulfite[28] onto commercially available fluorinated bromobenzaldehydes 1 and 2 (Scheme ). The improved yield for 3, relative to our previously reported yield (63%)[22] results from reducing the reaction temperature from 160 to 140 °C and increasing the reaction time from 16 to 48 h. Sulfonated benzaldehydes 3 and 4 were condensed with 3-(dimethylamino)phenol to afford the sulfoTMR dyes 5 (para-isomer) and 6 (meta-isomer) in 35% and 44% yield. Subsequent Pd-catalyzed Heck coupling with previously reported styrene molecular wires[22] gave sRhoVR dyes (7–10) in isolated crude yields ranging from 52 to 62%. Compared to carboxamide-substituted RhoVR, which requires six synthetic steps from commercially available starting materials,[25] sRhoVR can be accessed in just three steps, representing a 50% reduction in step count. Small amounts of the sRhoVR molecules were purified via preparative HPLC for spectroscopic characterization. The use of sulfonated benzaldehydes provides ready access to regioisomerically pure sulfonated rhodamines and may be a general strategy for creation of analogous, water-soluble xanthene dyes. The SulfoTMR dyes 5 and 6, as well as the final sRhoVR dyes (7–10), demonstrate absorption maxima centered between 548 and 553 nm (ε = 60 000 to 88 000 M–1 cm–1; Figure S1), similar to those of classic tetramethylrhodamine dyes. The sRhoVR compounds also possess a strong secondary absorption band near 400 nm due to the presence of the phenylenevinylene molecular wires, with the para-sRhoVR (7, 8; Figure S1) secondary band slightly red-shifted relative to that of the meta-sRhoVRs (9, 10; Figure S1). Fluorescence emission from all of the compounds was centered between 570 and 574 nm (Φ = 0.24–0.57; Table S1).

Scheme 2

Synthesis of Sulfonated Rhodamine Voltage Reporters (sRhoVRs)

Cellular Characterization and Performance of sRhoVRs

Live-cell imaging reveals that sRhoVR dyes localize to the plasma membrane of cells. Incubation of HEK cells with sRhoVR dyes (7–10) at a concentration of 200 nM results in clear membrane-localized fluorescence, as determined by confocal laser scanning fluorescence microscopy (Figure A, Figure S2). The apparent membrane staining was brighter for the meta-sRhoVR dyes (9 and 10) than for the para-sRhoVR dyes (7 and 8). The membrane staining of sRhoVR dyes indicates that the ortho-sulfonate is sufficient to prevent internalization of tetramethylrhodamine-based voltage indicators. We assessed the voltage sensitivity of each sRhoVR dye in HEK cells using patch-clamp electrophysiology. Hyper- and depolarizing steps from +100 to −100 mV in 20 mV increments from a baseline potential of −60 mV revealed a range of voltage sensitivities from 3 to 44% for the sRhoVR dyes, depending on the combination of fluorophore (para5 or meta6) and molecular wire (Table S1, Figure B,C, and Figure S3). The voltage sensitivities of sRhoVR compounds tracked well with previously reported RhoVR counterparts,[25] suggesting that the replacement of the carboxyl group associated with classical rhodamines with a sulfonate group minimally perturbed the electronic properties of the voltage reporters. Like we observed with the original RhoVR compounds, indicators[25] with meta-substitution patterns (9 and 10) were both more voltage-sensitive and demonstrated higher signal-to-noise ratios (SNR) than their para-substituted counterparts. We are currently undertaking studies to probe the molecular mechanisms underlying this difference. One simple explanation is that for the meta-sRhoVRs, the alignment of the sulfonate with the long axis of the molecular wire improves the alignment between the electron transfer vector and the transmembrane electric field, which should enhance sensitivity.[22] Given its high sensitivity—44% ΔF/F per 100 mV (compared to 25% for RVF5) and brightness in cells (Figure S2), we chose compound 10, which we call sRhoVR 1, for further characterization in subsequent experiments. Importantly, for use in imaging in tissue and in vivo, the water solubility of sRhoVR 1 (0.92 mM) was nearly 3-fold the solubility of RVF5 (0.29 mM) and 5-fold greater than RhoVR 1 (0.17 mM).

Figure 1

Characterization of SulfoRhoVR 1 (sRhoVR 1) in HEK Cells. (a) Confocal fluorescence image of sRhoVR 1 (10). Scale bar is 20 μm. (b) Plot of the percentage change in fluorescence vs membrane potential summarizing data from five separate cells, resulting in an average voltage sensitivity of 44% per 100 mV. Error bars are ± SEM. (c) Plot of fractional change in fluorescence vs time for 100 ms hyper- and depolarizing steps from +100 mV to −100 mV in 20 mV increments from a holding potential of −60 mV for single HEK293T cells under whole-cell voltage-clamp mode. (d) Two-photon cross section of sRhoVR 1 vs RhodolVoltageFluor 5 (RVF5), acquired in PBS, pH 7.4.

Neuronal Characterization of sRhoVR 1

Cultured rat hippocampal neurons incubated with sRhoVR 1 (10, 500 nM) and imaged using confocal laser scanning fluorescence microscopy display clear membrane staining (Figure a). Rat hippocampal neurons were subjected to whole-cell patch-clamp electrophysiology under current-clamp mode to record spontaneous and evoked action potentials. Dual optical and electrophysiological recording of spontaneous action potentials demonstrate that sRhoVR 1 exactly follows the electrophysiology recording (Figure b). We recorded evoked action potentials in hippocampal neurons under whole-cell current clamp mode, in the presence or absence of sRhoVR 1 (500 nM). The presence of sRhoVR 1 did not significantly alter the time to peak, half-width, rise tau, rise time, or decay time of the action potential, nor overall cell capacitance (n = 7 cells, 10 spikes each), confirming that sRhoVR 1 does not perturb underlying cellular physiology or membrane properties (Figure S4). Spontaneous activity recordings in cultured rat neurons revealed a SNR of 20:1 (n = 30 APs) with a ΔF/F of 5% per spike (Figure c). Cultured rat neurons were also subjected to external field stimulation and SNR for a single action potential was determined to be 11:1 (n = 50 APs) with a ΔF/F of 3% per spike (Figure d).

Figure 2

Imaging evoked and spontaneous activity in neurons with sRhoVR 1. (a) Rat hippocampal neurons were stained with 200 nM sVR 1. The scale bar is 20 μm. (b) Dual optical (red dots, 1 kHz sampling rate) and electrophysiological (black trace, 50 kHz sampling rate) recording of spontaneous action potentials in rat hippocampal neurons. (c) Imaging of spontaneous activity in rat hippocampal neurons using sRhoVR 1. (d) Optical recording of action potentials evoked in rat hippocampal neurons by external field stimulation (5 Hz).

Two Photon Performance of sRhoVR in Cultured Neurons

We evaluated the ability of sRhoVR to monitor membrane potential changes under two-photon illumination. Two-photon absorption relies on the essentially simultaneous absorption of two lower-energy photons to promote a chromophore to a singlet excited state. Due to the use of longer-wavelength light,[29,30] two-photon microscopy enables imaging in thick tissue samples such as brain slices and intact brains. We showed previously that rhodol-based voltage indicators display voltage sensitivity in both traditional single-photon and two-photon microscopy contexts.[24] Rhodamines have high two-photon absorption cross sections,[29] suggesting that sRhoVR 1, with its large voltage sensitivity (44% ΔF/F per 100 mV) and yellow-orange emission profile (570 nm), would be a promising two-photon voltage indicator. We measured the two-photon absorption cross section (σTPA) of SulfoTMR dyes 5 and 6 (in ethanol) as well as sRhoVR 1 and RVF5 (in PBS, pH 7.4) by normalizing to a rhodamine B standard.[24,31] A plot of two-photon absorption cross-section vs excitation wavelength reveals σTPA maxima of approximately 210 GM (830–840 nm), which is in good agreement with literature values for rhodamine B (a very similar fluorophore) in ethanol (204 GM at 830 nm)[31] (Figure S5). By comparison, RVF5 displays an almost 2-fold lower value, 125 GM at its maximum of 820 nm (Figure d), and the previously reported VF2.1.Cl has a much weaker 40 GM at its maximum of 780 nm.[24] Consistent with two-photon absorption, sRhoVR 1 emission under two-photon illumination demonstrates a quadratic dependence on illumination intensity (Figure S6).

With bright two-photon emission and high voltage sensitivity, we expected that sRhoVR 1 could be used for two-photon voltage imaging in vivo. To lay the groundwork for these studies, we assessed the ability of sRhoVR 1 to measure neuronal activity in cultured rat hippocampal neurons under two-photon illumination. Bath application of sRhoVR 1 (200 nM) resulted in well-defined membrane staining (Figure a,b). In single-trial, single-pixel (6.6 μm2) optical recordings (Figure c), hippocampal neurons isolated from rat showed robust spontaneous activity that could be detected without posthoc filtering, averaging, or photobleach correction (Figure d). To confirm that the activity we observed was due to action potentials, we treated active cultures with tetrodotoxin (TTX, 1 μM), to inhibit spontaneous action potential firing. In TTX-treated cultures, we did not observe spiking activity (Figure S7). Together these results establish that sRhoVR 1 can detect neuronal voltage changes under two-photon illumination.

Figure 3

Two-photon voltage imaging of sRhoVR 1 in rat hippocampal neurons. (a) sRhoVR 1 brightly stains the plasma membrane of rat hippocampal neurons under two-photon illumination. Scale bar is 50 μm for all images. (b) Zoomed version of boxed region in panel (a). (c) Still frame from a video recording (200 Hz, 425 × 52 μm, 64 × 8 pixels) of the neurons from panel b. (d) Fluorescence responses demonstrate robust spontaneous activity. Fluorescence traces are single-trial ΔF/F values from single pixels, are unfiltered, and uncorrected for bleaching.

In Vivo Imaging with sRhoVR 1

We next sought to deploy sRhoVR 1 for imaging voltage changes in the brains of live mice. sRhoVR is a promising candidate for in vivo brain imaging because it possesses a nearly 2-fold larger two-photon cross section than RVF5, shows a 50% improvement in voltage sensitivity over RVF5 (44% vs 28%), an emission peak red-shifted by 35 nm, and 3–5-fold greater solubility compared to RVF5 and RhoVR 1. In particular, the ability to use two-photon illumination for voltage imaging allows the interrogation of voltage dynamics from neuronal membranes localized to a two-photon focal plane of approximately 2 μm. This is in contrast to classically employed widefield voltage imaging techniques as well as extracellular electrophysiological recordings of local field potential (LFP) and multiunit activity (MUA). LFP reflects a current dipole that often extends hundreds of micrometers throughout a number of cortical layers,[5,32] while MUA reflects spiking of many neurons within ∼100 μm of the recording electrode tip.[2] Therefore, use of electrodes alone cannot resolve differences in layer-specific activity recorded by LFP. Similarly, pairing widefield epifluorescence microscopy with electrode-based recording cannot resolve layer-specific responses due to out-of-plane fluorescence collected during epifluorescence microscopy. We envisioned two-photon imaging with sRhoVR 1 would be a generalizable solution to this problem: bulk loading of sRhoVR 1 would widely label cells in vivo, and optical sectioning with two-photon microscopy would enable depth resolution for complementing LFP recordings.

We performed two-photon sRhoVR 1 imaging in layer 2/3 of the barrel cortex of anesthetized mice (Figure a). Pressure injection of sRhoVR 1 through a glass pipet at ∼200 μm below the surface (Figure b, asterisk, 100–200 μM in ACSF) resulted in diffuse fluorescence staining across an area approximately 500 μm in diameter (Figure b, red is sRhoVR 1 fluorescence, cyan is fluorescein isothio cyanate-conjugated dextran labeling of the vasculature). Typical staining patterns reveal comprehensive labeling of neuropil, with cell bodies appearing as dark silhouettes (Figure c,d, Figure S8a). In general, the staining was uniform, although occasionally we observed brighter cells, which are currently unidentified and may be microglia (Figure c, arrowheads). RVF5 did not perform as well as sRhoVR: injection of rhodol-based voltage indicator RVF5 did not give staining over as large of an area, which may be partially explained by the lower solubility of RVF5 relative to sRhoVR (0.29 mM vs 0.92 mM). Extracellular recordings of LFP and MUA were acquired simultaneously with two-photon imaging using a tungsten microelectrode inserted near the site of sRhoVR 1 injection (Figure b, white arrowhead). Optical recordings were acquired under two-photon excitation from a region of interest (ROI) shaped as a horizontal strip approximately 200 × 25 μm in size (Figure d, black rectangle), at a framerate of approximately 20 Hz. Weak electrical stimulation of the contralateral whisker pad in anesthetized mice resulted in clear, single-trial optical responses from sRhoVR 1(Figure e). The fluorescence increase ranged from 4 to 10% ΔF/F, with a signal-to-noise ratio (SNR) of 5.9 ± 3.2 in single trials and 11.6 after averaging 8 trials (Figure f). We observed very little photobleaching during the course of the experiments. A comparison of bleach rates for sRhoVR and RVF5 under nearly identically two-photon illumination conditions (31 mW for sRhoVR, 30 mW for RVF5) reveal no significant difference in bleach rates either in mouse cortext (Figure S8b) or in cultured rat hippocampal neurons (Figure S8c). The time course of the externally recorded field potential (LFP, Figure e,f) precisely matches the time course of the optically measured transmembrane potential (sRhoVR 1, Figure e,f). The positive deflection in sRhoVR 1 fluorescence indicates membrane depolarization. Combined with the negative deflections in the extracellularly recorded LFP, this confirms excitatory currents across neuronal membranes at a depth of 200 μm, resulting in depolarization. Because the sign of the LFP can vary with the position of the electrode, two-photon imaging of transmembrane potential with sRhoVR 1 provides an crucial complement for interpreting the extracellularly recorded LFP.

Figure 4

In vivo, two-photon voltage imaging in layer 2/3 of the barrel cortex of anesthetized mice using sRhoVR 1. (a) Schematic of in vivo imaging experimental setup. Cranial bone and dura mater were removed above the barrel cortex. The brain was covered with 0.7% agarose and on top, a coverslip was placed in a way that allowed access to the brain from one side of the preparation. sRhoVR 1 (100 to 200 μM in ACSF) was pressure-injected through a glass or quartz micropipette; local field potentials (LFP) and multiunit activity (MUA) were recorded in close vicinity to the injection site with a tungsten electrode (impedance: 5–7 MΩ). (b) A view from the top on the cortical surface after intracortical injection of sRhoVR 1 (red) and intravascular injection of fluorescein isothiocyanate-conjugated dextran (cyan); the sRhoVR 1-filled glass pipet (asterisk) and tungsten electrode (white arrowhead) are visible. Scale bar, 500 μm. (c, d) Typical two-photon images of tissue staining with sRhoVR 1 in cortical layer 2/3; the majority of cell bodies appeared dark indicating the lack of sRhoVR 1 in the intracellular compartment; arrowheads indicate a few exceptions. The rectangle in (d) depicts a typical region of interest (ROI) for data acquisition (approximately 200 × 25 μm). Scale bars, 50 μm. (e) Time-course of sRhoVR 1 fluorescence, relative to baseline (ΔF/F), and LFP traces acquired simultaneously. Dotted red lines indicate timing of contralateral whisker pad stimulation with a single 300 μs weak electrical pulse. (f) Single trials (gray) and trial average (black) for sRhoVR1 fluorescence, LFP, and MUA; eight trials from one ROI are overlaid. Dotted red lines indicate timing of contralateral whisker pad stimulation; asterisk in the MUA trace indicates artifact of electrical stimulation. For panels e and f sRhoVR 1 fluorescence was acquired in frame scan mode at a frequency of ∼20 Hz and normalized to mean fluorescence over the entire time course to obtain ΔF/F. All fluorescence time-courses are unfiltered, and uncorrected for bleaching.

To further establish the utility of sRhoVR 1 for bridging neuronal transmembrane potential dynamics to behavior in living organisms, we performed similar in vivo imaging experiments in awake mice (Figure ). Mice were implanted with chronic cranial windows that also allowed loading of sRhoVR 1 through a silicon injection port (Figure a). Mice were briefly anesthetized with isofluorane for intracortical injection of sRhoVR 1 (100 to 200 μM) using a quartz pipet positioned at ∼200 μm below the surface (Figure a). The staining pattern was similar to that in acute experiments (Figure b): sRhoVR 1 comprehensively labeled neuropil, with cell bodies appearing as dark shadows (representative images, Figure c,d)). Following recovery from isofluorane, we imaged responses to brief air puffs to the contralateral whisker pad. Figure e shows sRhoVR 1 signal time-courses from five different ROIs in layer 2/3 imaged consecutively at different depths (130–300 μm). In each case, the time-course was computed as an average from all pixels within the ROI (∼200 × 25 μm imaged at 20 Hz). Robust increases in fluorescence were observed in response to each air puff. Positive sRhoVR 1 responses indicate depolarizations in shallow (130 μm), intermediate (200 μm) and deeper cortical layers (300 μm, Figure e), providing a critical depth resolution constraint for the interpretation of LFP recordings in the cortex.[5] Following imaging sessions, mice recovered fully and could be again treated with sRhoVR 1 for voltage imaging. Together, these experiments establish the compatibility of sRhoRV-type dyes for in vivo imaging in mice and demonstrate the utility of sRhoVR 1 for recording optical transmembrane potential responses with depth resolution in the cortex.

Figure 5

In vivo, two-photon voltage imaging in the barrel cortex of awake mice using sRhoVR 1. (a) The chronic “cranial window,” made by fusing a stack of three 3 mm and one 5 mm glass coverslips, was used to close the exposure. As injection port, a 0.5 mm hole was drilled into the 5 mm coverslip and covered with silicone. During implantation surgery, the cranial bone was removed while dura mater was left intact. After 3–4 weeks following implantation, sRhoVR 1 (0.1–0.2 mM in ACSF) was pressure-injected with a quartz pipet inserted through the silicone injection port while the animal was under isoflurane anesthesia; imaging started after recovery from anesthesia (20–30 min after injection). (b) A view from the top on the cortical surface after intracortical injection of SulfoRhoVR1 (red); arrow indicates the injection path, the arrowhead points toward the injection port. Scale bar, 500 μm. (c, d) Typical two-photon images of tissue staining with SulfoRhoVR1 in cortical layer 1 (c) and 2/3 (d). Arrowheads indicate bright cell bodies. Scale bars, 50 μm. (e) Time-courses of sRhoVR 1 fluorescence relative to baseline (ΔF/F); each traces corresponds to a different ROIs in the same animal at the indicated cortical depth. Dotted red lines indicate timing of contralateral whisker pad stimulation with a single air puff. Black arrows point to motion artifacts. sRhoVR 1 fluorescence was acquired in frame scan mode at a frequency of ∼20 Hz and normalized to mean fluorescence over the entire time course. All fluorescence time-courses are unfiltered, and uncorrected for bleaching.

Discussion

In summary, we present the design, synthesis, and application of sulfonated Rhodamine Voltage Reporter 1 (sRhoVR 1). sRhoVR 1 represents the best of a new class of sulfonated rhodamine that uses PeT as a voltage-sensing trigger. With short synthetic route from commercially available starting materials (three steps), high voltage sensitivity (44% ΔF/F per 100 mV), an emission maximum centered at 570 nm, enhanced solubility (3–5-fold over RhoVR 1 or RVF5), improved two-photon cross section (∼2× over RVF5 and ∼5× over RhoVR 1), and fast response kinetics capable of tracking action potentials under both single-photon and two-photon conditions, sRhoVR is a promising candidate for use in vivo. We show that sRhoVR 1 can be deployed in intact mouse brains for two-photon imaging in vivo. By coupling optical recording of sRhoVR 1 responses in layer 2/3 of the barrel cortex of mouse, sRhoVR 1 provides a direct measure of transmembrane potential with essential depth resolution for interpreting extracellular potentials captured with the LFP recording. In the case of sensory input via whisker stimulation, depolarization measured via optical sRhoVR 1 signals provide direct evidence for excitatory synaptic transmission at multiple, optically resolved cortical depths. Depolarization of the transmembrane potential, directly observed with sRhoVR 1 via two photon microscopy, was accompanied by negative deflections in extracellular potential (LFP), consistent with a canonical thalamocortical response. Importantly, sRhoVR 1 is applicable for imaging in awake, head-fixed mice with implanted cranial imaging windows, opening the door for longitudinal in vivo two-photon membrane potential studies in behaving animals.