Bioorthogonal, fluorogenic targeting of voltage-sensitive fluorophores for visualizing membrane potential dynamics in cellular organelles

Abstract

Electrical potential differences across lipid bilayers play foundational roles in cellular physiology. Plasma membrane voltage is the most widely studied; however, the bilayers of organelles like mitochondria, lysosomes, nuclei, and endoplasmic reticulum (ER) also provide opportunities for ionic compartmentalization and the generation of transmembrane potentials. Unlike plasma membranes, organellar bilayers, cloistered within the cell, remain recalcitrant to traditional approaches like patch-clamp electrophysiology. To address the challenge of monitoring changes in organelle membrane potential, we describe the design, synthesis, and application of LUnAR RhoVR (Ligation Unquenched for Activation and Redistribution Rhodamine based Voltage Reporter) for optically monitoring membrane potential changes in the endoplasmic reticulum (ER) of living cells. We pair a tetrazine-quenched RhoVR for voltage sensing with a transcyclooctene (TCO)-conjugated ceramide (Cer-TCO) for targeting to the ER. Bright fluorescence is observed only at the coincidence of LUnAR RhoVR and TCO in the ER, minimizing non-specific, off-target fluorescence. We show that the product of LUnAR RhoVR and Cer-TCO is voltage sensitive and that LUnAR RhoVR can be targeted to intact ER in living cells. Using LUnAR RhoVR, we use two-color, ER-localized, fast voltage imaging coupled with cytosolic Ca²⁺ imaging to validate the electroneutrality of Ca²⁺ release from internal stores. Finally, we use LUnAR RhoVR to directly visualize functional coupling between plasma-ER membrane in patch clamped cell lines, providing the first direct evidence of the sign of the ER potential response to plasma membrane potential changes. We envision that LUnAR RhoVR, along with other existing organelle-targeting TCO probes, could be applied widely for exploring organelle physiology.

Graphical Abstract

Introduction

Electrical potentials across biological membranes, including those that form membrane-bound organelles, regulate and initiate a host of physiological processes.¹ Although electrical potentials across external plasma membranes have been extensively studied, organelle membrane potentials remain relatively under-explored, by comparison. The membranes of organelles, ensconced within the confines of the cellular membrane, are difficult to access with electrodes, making determination of organelle membrane potential difficult.^2–4 While optical methods for monitoring changes in or measuring values of plasma membrane potentials are a powerful complement to electrode-based methods, delivering and targeting voltage-sensitive dyes to specific, internal organelle membranes for visualizing organelle membrane potential is an outstanding challenge.

Many strategies exist to target voltage sensitive fluorophores to mitochondria: most rely on lipophilic, cationic groups which accumulate in the mitochondrial matrix in a potential-dependent fashion. Approaches to other organelles are limited and most often involve bulk measurements of isolated organelle extracts.^5–7 Recent efforts to optically image membrane potential of internal organelles in intact cells have used DNA nanodevices⁸ or genetic targeting with small molecule quenchers to target lysosomal membranes.⁹ Other approaches use completely genetically encoded voltage-sensitive fluorescent proteins which express in both external and internal membranes to track membrane potential, but labeling can be ambiguous because of variable expression levels.¹⁰ A need exists for robust approaches to monitoring membrane potentials of cellular organelles.

We wondered whether we might adapt VoltageFluor dyes for delivery to organelle membranes and take advantage of the photoinduced electron transfer (PeT) voltage-sensing mechanism to monitor membrane potential in organelles. Most existing VoltageFluors are designed to be cell impermeant for measuring cellular membrane potential and thus do not effectively stain organelle membranes.¹¹ However, we recently found that modifications to a rhodamine-based voltage reporter (RhoVR) enabled passage through cell membranes and localization to mitochondria.¹²

Fluorescent dyes can be targeted to subcellular locations either through charge,^13–15 covalently attached groups specific to a subcellular location,¹⁶ genetic targeting of protein constructs,^17–21 or bioorthogonal chemistry in which one reaction partner is attached to a targeting group and another is attached to the fluorescent probe to allow targeting in living cells.²² A VoltageFluor targeted using bioorthogonal chemistry would be targetable to multiple subcellular locations using different targeted reaction partners. We selected a transcyclooctene (TCO)/tetrazine click reaction to use for subcellular targeting of VoltageFluors due to its rapid reaction kinetics^23–25 and the ability of tetrazines to quench the fluorescence of xanthene dyes.^26–28 Fluorescence becomes unquenched after the tetrazine group reacts with a strained alkene to form a covalent adduct, tethering unquenched dye to the subcellular location targeted by the alkene. We hypothesized that this could be general strategy for targeting and measuring membrane potential dynamics in internal organelles. A recent report described coupling of plasma membrane and internal membrane potential changes, and we thought a chemical targeting strategy along with a fast-response indicator like RhoVR might provide insight into this phenomenon.¹⁰

Results and Discussion

To realize this strategy, we made a Ligation Unquenched for Activation and Redistribution Rhodamine Voltage Reporter (LUnAR RhoVR). In this approach, a TCO is targeted to the membrane of interest using either a chemical targeting group or genetically targeted protein, after which the voltage reporter reacts with the targeted TCO to covalently bind it to the membrane of interest. Off-target background fluorescence is minimized because of quenching by the unreacted tetrazine group, (Scheme 1). After synthesizing and screening four different rhodamine-tetrazine linker chemistries for fluorescence turn-on (Figure S1), we settled on an amide linkage between the RhoVR and the tetrazine due to its superior turn-on and synthetic compatibility with published VoltageFluor syntheses (Scheme 1). Although direct conjugation of tetrazines to the pendant ring of the xanthene dyes give large fluorgenic responses (~100× fold)²⁷ via a through bond energy transfer mechanism (TBET),²⁹ amide-linked tetrazines afforded synthetic tractability and resonable unquenching responses ranging from 3× to 20× increase in fluorescence.^{20, 30}

Scheme 1. Structures of LUnAR RhoVR and Cer-TCO for targeted delivery of RhoVRs to organelles.

A small molecule rhodamine voltage sensor functionalized with a tetrazine quencher (LUnAR RhoVR 1) reacts with a targeted transcyclooctene (TCO) to form a covalent adduct with unquenched fluorescence. When this reaction is performed in live cells with a targeted transcyclooctene, for example ceramide-TCO, the small molecule voltage sensor is targeted to the membrane of interest with quenched fluorescence from any off-target localization.

The synthesis of tetrazine-containing LUnAR RhoVR proceeds from a reduction of a brominated carboxytetramethyl rhodamine (1)¹¹ to the corresponding alcohol (2) followed by a Heck reaction with a molecular wire. The alcohol is then converted to a primary amine and coupled using car-bodiimide chemistry to 3-methyl-6-carboxyphenyl tetrazine(24)³¹ to give LUnAR RhoVR 1 (Scheme 2). A similar synthesis with a molecular wire lacking aniline and methoxy groups gives LUnAR RhoVR 0, a voltage insensitive compound which serves as a useful control for cellular studies.^32–34

Scheme 2.

Synthesis of LUnAR RhoVRs

LUnAR RhoVR 1 and 0 possess absorbance and emission spectra characteristic of tetramethyl rhodamine, with absorbance centered at 557 nm and an emission maximum of 578 nm (Figure 1, Figure S2). The fluorescence of LUnAR RhoVR 1 increases 7-fold upon reaction with TCO-ol (25); LUnAR RhoVR 0 increases 10-fold (Figure 1, Table 1). Although the fluorescence increases, the absorbance shows very little change before and after reaction with TCO (Figure 1), indicating that the increase in brightness results primarily from a change in the quantum yield of fluorescence. Similar turn-on ratios are achieved with TCO compounds bearing HaloTag ligands and ceramide (Table 1). The second-order rate constant for the reaction of LUnAR RhoVR 1 with TCO-PEG₂₅-Halo is 150 (±10) M⁻¹ s⁻¹, which similar to previously reported values for tetrazines bearing a methyl substituent, which range from 40²⁰ to 800 M⁻¹ s⁻¹.³¹

Figure 1.

Optical properties of LUnAR RhoVR before and after reaction with transcycloocteneol (TCO-ol). a) Plot of absorbance vs wavelength for LUnAR RhoVR1 before (dark blue) and after (magenta) reaction with transcycloocteneol. b) Plot of fluorescence emission vs. wavelength for LUnAR RhoVR 1 before (dark blue) and after (magenta) reaction with transcycloocteneol from the samples shown in a. Spectra were acquired in ethanol with 0.01% HCl. LUnAR RhoVR 1 (200 μM) was reacted with transcyclooctenol (10 mM) in 5:1 water:dioxane and then diluted to 500 nM for spectral measurements.

Table 1.

Reactions of LUnAR with Transcyclooctenes

RhoVRa	TCO	Φflb
LUnAR 1	---	0.042
LUnAR 1	TCO-ol	0.26 (6.2×)
LUnAR 1	TCO-PEG25-Halo	0.28 (6.7×)
LUnAR 1	Cer-TCO	0.20 (4.8×)
LUnAR 0	---	0.032
LUnAR 0	TCO-ol	0.32 (10.0×)
LUnAR 0	Cer-TCO	0.17 (5.3×)

^aThe indicated LUnAR RhoVR (200 μM) was reacted with the indicated transcyclooctene (10 mM) in 5:1 water:dioxane and then diluted to 500 nM in ethanol with 0.01% HCl for spectral measurements.

^bQuantum yield of fluorescence. Values in parenthesis indicate fold increase over unreacted RhoVR.

We first tested a HaloTag targeting system for targeting LUnAR RhoVRs to both the exterior and interior leaflets of the plasma membrane. TCO tethered to a HaloTag chloroalkane ligand via a flexible polyethyleneglycol (PEG) linker would first be localized to HaloTag. Then, treatment with LUnAR RhoVR would result in labeling of HaloTag proteins (Scheme 1). LUnAR RhoVR1 labels the plasma membrane when HaloTag is localized to the outer leaflet of the plasma membrane (Figure S3a–c); however, when HaloTag is localized to the endomembrane via prenylation³⁵ (Figure S4), we observe very low levels of cellular fluorescence, likely because the long PEG linkers used to ensure voltage sensitivity of RhoVRs in HaloTag targeted systems³⁴ prevent substantial internalization of RhoVR (Figure S3d–f). Labeling of cells expressing the same endomembrane HaloTag show robust fluorescence when labeled with the cell-permeable JF646-HaloTag ligand, supporting this hypothesis (Figure S3f–h).

We verified the voltage-sensitivity of the product of LUnAR RhoVR 1 and TCO by reacting TCO-PEG₂₅-Halo (25) with LUnAR RhoVR 1 in vitro and applying this to cells expressing outer leaflet HaloTag (Figure S5).³⁴ Patch clamp electrophysiology on these cells reveal that LUnAR RhoVR 1 / TCO conjugates respond to membrane potential change, with a sensitivity of approximately 21% (± 0.72%, S.E.M. for n = 3 cells, Figure S5).

After establishing the voltage sensitivity of LUnAR RhoVR1-TCO conjugates at the outer leaflet of the plasma membrane, we turned to the organelle specific lipid-TCO developed by the Schepartz lab for their High Density Environmentally-sensitive probes (HIDE probes)³⁶ to target LUnAR RhoVRs to organelles. We loaded HeLa cells with a ceramide-TCO (Cer-TCO) followed by loading with LUnAR RhoVR 1 or LUnAR RhoVR 0 at 37 °C. We previously showed that, at temperatures below 20 °C Cer-TCO enables selective labeling of the Golgi with quenched tetrazine fluorophores.³⁷ However, when loaded at 37 °C, Cer-TCO enables labeling in the ER and Gogli.³⁶

Cell treated with Cer-TCO followed by LUnAR RhoVR 1 show strong fluorescence associated with internal structures (Figure 2a, Figure S6, S8). Cells that pre-treated with Cer-TCO prior to LUnAR RhoVR treatment are significantly brigher than cells without Cer-TCO (up to 4.4× brighter, mean of 2.3× brighter; p = 4.1×10⁻²³, Figure S7). The bright LUnAR signal colocalizes with the ER (Pearson’s coefficient of 0.82 ± 0.03 with ER Tracker Green, Figure 2d) and shows negilible plasma membrane staining (Pearson’s coefficient of 0.33 ± 0.07 with membrane-associated BeRST,³⁸ Figure 2d, Figure S6). Similar colocalization analysis reveals much lower colocalization with mitochondria (when using MitoTracker Deep Red as a marker for mitochondria,^{13, 39} 0.75 ± 0.02). Mitochondria (measured by MitoTracker Deep Red) and ER (ER Tracker Green) show substantial colocalization themselves (0.68 ± 0.04)(Figure S6, S9). Colocalization values are even lower for Golgi (using GalNAcT2 – BFP as Golgi marker, 0.47 ± 0.05).⁴⁰ Together, these data show localization primarily to the ER. The voltage insensitive control compound, LUnAR RhoVR 0, shows similar cellular localization (Figure S6, S9).

Figure 2.

LUnAR RhoVR 1 + Cer-TCO localize to the endoplasmic reticulum of HeLa cells. Confocal slices of HeLa cells pre-incubated with Cer-TCO followed by labeling with a) LUnAR RhoVR 1 (magenta) and c) the plasma membrane marker BeRST (cyan). b) An overlay of the two signals shows no significant overlap. Scale bar is 20 μm. Confocal slices of HeLa cells pre-incubated with Cer-TCO followed by labeling with d) LUnAR RhoVR 1 (magenta) and f) ER Tracker Green (green). e) An overlay of the two signals shows overlap in white. Scale bar is 20 μm. d) Plot of Pearson’s coefficients for colocalization analysis of LUnAR RhoVR 1 and ER green or BeRST. 90° control indicates analysis in which one of the images has been rotated 90° during the colocalization analysis. * indicates p < 0.05; “ns” indicates p > 0.05 (student’s t-test, two tailed unequal variance).

Having established that LUnAR RhoVR 1 can label the ER via reacting with pre-localized Cer-TCO (Figure 2) and that the reaction product of LUnAR RhoVR 1 + TCO is voltage sensitive (Figure S5), we next sought to determine whether LUnAR RhoVR 1 is voltage sensitive in the ER membrane. Because of the difficulty associated with patch clamp electrophysiological calibration of ER membrane potential in living cells, we modified a previously reported literature method for inducing voltage changes in intracellular membranes.⁹ In our method (Figure 3a, Figure S10), cells loaded with LUnAR RhoVR1 targeted to the ER via Cer-TCO are first treated briefly with digitonin to selectively permeabilize the cell membrane, but not organelle membranes. Digitonin permeabilizes cholesterol-rich membranes.⁴¹ The plasma membrane contains up to 50% cholesterol content;^42–43 while ER membranes contain only about 5% cholesterol content.^44–45 Thus, the digitonin treatment selectively permeabilizes the plasma membrane while leaving the ER membrane intact.

Figure 3.

LUnAR RhoVR1 senses voltage changes across the ER membrane. a) Schematic representation of permeabilization of plasma membranes with digitonin, resulting in loss of calcein fluorescence (green) but retention of LUnAR RhoVR1 staining in ER, treatment with ionophores valinomycin (beige) and nigericin (yellow), and modulation of [K⁺] to change ER membrane potential. b) When cytosolic and luminal [K⁺] is equal, there is no difference in membrane potential, and ΔV_ER ~ 0. c) When the cytosolic [K⁺] is switched to 10 mM, valinomycin facilitates the diffusion of K⁺ ions out of the ER lumen, resulting in a net negative luminal potential (ΔV_ER is negative), decreased PeT, and increased RhoVR fluorescence. d) When cytosolic [K⁺] is returned to 130 nM, K⁺ again diffuses down its concentration gradient, resulting in a net positive luminal potential (ΔV_ER is positive), increased PeT, and decreased RhoVR fluorescence. e-h) Plots of relative RhoVR fluorescence intensity vs. time for Cer-TCO loaded, LUnAR RhoVR 1 stained HeLa cells with e) digitonin-permeabilized plasma membranes or f) intact plasma membranes. Grey bars indicate high [K⁺] (130 mM) in both cytosolic/external and luminal spaces, blue bars indicate low cytosolic/extracellular [K⁺] (10 mM) with high luminal [K⁺], and red bars indicate high cytosolic/external [K⁺] with low luminal [K⁺]. g) Plot of relative RhoVR fluorescence intensity for cells as in (e), but with a constant high [K⁺] (130 mM) perfusion. h) Plot of relative RhoVR fluorescence intensity for cells as in (e), but with voltage-insensitive LUnAR RhoVR 0 in place of LUnAR RhoVR 1. Data represent mean values (solid lines) ± error (S.E.M.) for n = 9, 5, 4, or 3 for conditions e, f, g, and h, respectively. ΔV designations at the bottom of the plots indicate the relative potential across the ER membrane, relative to the lumen, where ΔV_ER = V (lumen) – V (cytosol).

We established plasma permeability by co-incubation with calcein-AM: cells with intact cell membranes retained calcein-AM fluorescence, while cells with permeabilized membranes showed little to no calcein fluorescence (Figure 3a).⁴⁶ After treatment with digitonin, cells are incubated in a buffer mimicking the intracellular environment (high [K⁺] of 130 mM) and containing the potassium ionophores valinomycin and nigericin which allows K⁺ ions to equilibrate along their electrochemical gradient (Figure 3b).

Subsequently, a solution of low [K⁺] (10 mM) is perfused onto the cells (Figure 3c) while fluorescence images are taken over time (Figure 3e–h, S11). A short voltage transient is induced due to the different K⁺ concentrations on either side of the ER membrane and high K⁺ permeability induced by valinomycin and nigericin, resulting in a net negative potential on the luminal side of the ER (Figure 3c). Accompanying this hyperpolarization is a transient increase in LUnAR RhoVR 1 fluorescence of approximately 5% (Figure 3e). When a high [K⁺] solution is perfused back on the cells, the potential across the ER membrane is positive on the lumenal side (Figure 3d), and a small decrease in fluorescence occurs (Figure 3e). This response is likely smaller than the response to the initial hyperpolarization due to incomplete equilibration of the ER lumen with the low [K⁺] solution and thus a smaller magnitude of the reverse voltage transient.

The increase in fluorescence associated with ER lumen hyperpolarization is only observed in cells with permeabilized plasma membranes; non-permeabilized cells in the same dish, identified by their high intracellular calcein fluorescence, do not respond to differing [K⁺] perfusion (Figure 3f, Figure S12). In separate control experiments, perfusion with identical high [K⁺] solutions results in no fluorescence change (Figure 3g). Finally, if the high/low [K⁺] perfusion is repeated with voltage insensitive LUnAR RhoVR 0, we observe no change in fluorescence (Figure 3h). As an additional control confirm that LUnAR RhoVR 1 is not sensing mitochondrial membrane potential changes, we perfused FCCP, which transports protons across membranes down their concentration gradient,⁴⁷ rather than [K⁺]. FCCP perfusion results in minimal fluorescence change (Figure S13). This supports localization of LUnAR RhoVR 1 to the ER, rather than mitochondria, and corroborates colocalization data (Figure 2, Figure S6, 7–8), since we previously established that low doses of FCCP (500 nM) cause a change in the mitochondrial membrane potential.¹² It further corroborates that the fluorescence changes we observe upon switching [K⁺] result from membrane potential changes in the ER, rather than off-target pH sensing, since the pH of the ER is similar to the cytosol,¹ while the pH gradient between the cytosol and other organelles like mitochondria and lysosome is much steeper.^{1, 48} Finally, these data support localization of the majority of Cer-TCO/LUnAR RhoVR to the luminal leaflet of the ER membrane. An even distribution of Cer-TCO to cytosolic and luminal leaflets would give no response, but even a small bias for the luminal leaflet would give optical responses to ER membrane potential in the direction we observe. Taken together, these results demonstrate that LUnAR RhoVR 1 is able to sense voltage transients in the ER.

We next use LUnAR RhoVR1 to examine two physiological processes in which ER membrane potential changes may play a role: Ca²⁺ release from ER stores and electrical coupling between the ER and plasma membranes. We first examine ER voltage changes during Ca²⁺ release. The ER and sarcoplasmic reticulum (SR) are major cellular stores of calcium ions.⁴⁹ During cellular signaling processes, Ca²⁺ channels in the ER/SR open and Ca²⁺ is released from these internal stores,⁴⁹ moving down its electrochemical gradient and into the cytosol.⁵⁰

The movement of Ca²⁺ out of the ER down its electrochemical gradient results in negative potentials in the ER lumen relative to the cytosol, unless cations enter the ER lumen to balance the charge. Potassium may flow into the ER lumen from the cytosol to compensate for the outward flow of Ca²⁺. Previous reports find that SR membrane potential does not change during Ca²⁺ release.^51–52 However, the fluorescent reporters used in these experiments have limited time resolution due to their slower kinetics.^{6–7, 53} VoltageFluors, because of their PeT mechanism, sense rapid changes in membrane potential,^{33, 54} and thus we saw an opportunity to validate the conclusions of prior work by using LUnAR RhoVR to examine evidence for rapid changes in ER membrane potential in a non-excitable cell line.

Histamine treatment induces Ca²⁺ oscillations in HeLa cells which necessitates release of Ca²⁺ from internal stores.⁵⁵ By simultaneously measuring ER membrane potential in HeLa cells with LUnAR RhoVR 1/Cer-TCO at 200 Hz (Figure 4a) and Ca²⁺ oscillations via Oregon Green BAPTA (OGB, Figure 4b), we were able to look for rapid changes in ER voltage that could be linked to Ca²⁺ release. Intact cells were treated with Cer-TCO and LUnAR RhoVR 1 to label ER and OGB (Figure 4ab) and stimulated with histamine to evoke Ca²⁺ release. Fluorescence traces were then aligned so that the steepest increase in OGB fluorescence is synced to t = 0 in all traces (Figure S14). We observe no change from the baseline fluorescence of LUnAR RhoVR during histamine-evoked Ca²⁺ release when examining either individual cells (Figure 4c) or average responses across multiple trials (Figure 4d).

Figure 4.

ER membrane potential does not change during Ca²⁺ release in HeLa cells. a) Fluorescence image of HeLa cells treated with Cer-TCO and a) LUnAR RhoVR 1 to stain the ER and b) OGB-AM to visualize Ca²⁺ transients. Scale bar is 20 μm. Plots of LUnAR RhoVR 1 fluorescence (purple) and OGB fluorescence (green) vs time after histamine (100 μM) stimulation from a single HeLa cell 100 μM histamine stimulation (100 μM). Data are for either c) a single cell or d) the average across 3 different coverslips. e) Adding a fluorescence change of 0%, 1%, 2%, and 5% into the raw trace of a single cell (purple trace) and the threshold for 10 standard deviations above the local average fluorescence (dashed orange trace) used for detection. f) Plot of the fraction of individual cells in a single coverslip that have an event detected above the threshold shown in c on the raw fluorescence data from histamine trials as well as raw data where a single frame has an added fluorescence change of the amount indicated on the x-axis. *** indicates p<0.001 (student’s t-test, 2-tailed, unequal variance)

To establish a limit of detection, we developed an automated analysis routine (Figure S14). We examined each cell region of interest (ROI) to search for any cell on a coverslip that has a fluorescence change 10 standard deviations above the noise during the RhoVR recording. We then created artificial positives to test the limit of detection, increasing the fluorescence of an ROI at a single timepoint by 1 to 5% (Figure 4e). Then we asked what fraction of cells had detectable single-frame fluorescence changes. The unmodified data had no cells with a detectable fluorescence change, but the automated analysis could readily detect changes of 2.5% or greater (p < 0.001, Student’s t-test, 2-tailed unequal variance, Figure 4f). Taken together, these results extend the current literature hypothesis that voltage changes during Ca²⁺ release from the ER are not electrogenic and are likely compensated for by the movement of other ions into the ER. The more rapid time sensitivity of LUnAR RhoVR1 compared to existing probes rules out rapid voltage transients down to 5 ms in duration and <2.5% ΔF/F. The lower time bound is due to the limits of the image collection rate rather than the limits of probe response kinetics.

To further explore the ability of LUnAR RhoVR1 to measure biologically relevant voltage changes, we turned to patch-clamp electrophysiology to examine the electrical relationship between plasma membrane potential and ER membrane potential. A previous report using an internally targeted genetically-encoded voltage indicator (GEVI) gave evidence for electrical coupling and synchrony between the plasma membrane and internal membranes in HEK cells and hippocampal neurons.¹⁰ Although optical signals from the plasma membrane and internal membranes moved in the opposite direction, the orientation of the GEVI in intracellular membranes is unknown; therefore, the direction of fluorescence change cannot be used to determine the direction of the voltage change. We set out to disambiguate the direction of the optical response and extend these results into another cell line through simultaneous electrical measurements of plasma membrane potential and optical measurements of ER membrane potential with LUnAR RhoVR 1. We induced plasma membrane depolarization (more positive inside) and hyperpolarization (more negative inside) in LUnAR RhoVR1 / Cer-TCO stained HeLa cells using whole-cell voltage clamp (Figure 5a).

Figure 5.

Voltage imaging with LUnAR RhoVRs in the ER under whole-cell patch-clamp conditions. a) Widefield fluorescence image of HeLa cell treated with Cer-TCO, followed by LUnAR RhoVR 1. b) Plot of command voltage vs. time for patch-clamped HeLa cells with LUnAR RhoVR 1 labeling. c) Plot of LUnAR RhoVR 1 fluorescence vs. time for patch-clamped HeLa cells. d) Zoomed-in overlay of voltage command (blue) and LUnAR RhoVR 1 response (magenta) from panels b/c. e) Plot of command voltage vs. time for patch-clamped HeLa cells with LUnAR RhoVR 0 labeling. f) Plot of LUnAR RhoVR 0 fluorescence vs. time for patch-clamped HeLa cells. g) Two-color imaging of plasma membrane potential with BeRST 1 and ER membrane potential with LUnAR RhoVR 1. Black traces (upper) are the voltage commands, bright purple is BeRST 1, and magenta is LUnAR RhoVR 1. h) Plot of percent change in fluorescence from fluorescence at −60 mV as a function of plasma membrane potential from n=5 cells voltage clamped as in a. Error bars represent ± S.E.M. Dashed red lines are the line of best fit for values from −100 mV to 0 mV and from +20 mV to +100 mV.

Depolarizing steps in the plasma membrane (more positive potentials, Figure 5b) are accompanied by decreases in LUnAR RhoVR1 fluorescence (Figure 5c), which corresponds to a depolarization of the ER lumen (more positive on the inside). This result is consistent with [K⁺] perfusion experiments (Figure 3) and, along with the [K⁺] perfusion calibration, provides the first unambiguous confirmation of the correspondence of the sign of membrane potential change between ER and PM membrane potentials. Depolarization at the PM results in depolarization of the ER.

Notably, the fluorescence response of LUnAR RhoVR1 exactly follows the time course of the plasma membrane voltage command (Figure 5d, zoom-in of a single voltage step), suggesting a direct, or tight coupling of ER-PM potentials. When cells are stained with voltage insensitive LUnAR RhoVR0, the fluorescence does not change with membrane potential (Figure 5e/f), confirming that the observed fluorescence change with LUnAR RhoVR 1 is a response to ER membrane potential.

A two-color experiment with the plasma membrane potential reporter BeRST and LUnAR RhoVR 1 in Cer-TCO treated HeLa cells corroborates the electrophysiological recording. Depolarizing command voltages result in BeRST fluorescence increases (plasma membrane depolarization) and LUnAR RhoVR 1 fluorescence decreases (ER membrane depolarization) and are aligned precisely with the electrode-delivered plasma membrane potential command (Figure 5g).

Finally, a plot of the change in LUnAR RhoVR 1 fluorescence in the ER (ΔF/F) vs. the plasma membrane potential command reveals a plot with a substantial deviation from linearity. This is in contrast to the voltage sensitivity of pre-formed LUnAR RhoVR 1 on the plasma membrane of patch-clamped HEK cells, which shows a linear relationship between ΔF/F and plasma membrane command voltage (Figure S5e).

The ΔF/F response of LUnAR RhoVR 1 in the ER appears to have two linear components. At negative plasma membrane potentials (≤ 0 mV), we find a linear response of −1.4% per 100 mV (± 0.39, S.E.M., n = 5, R² = 0.997, Figure 5g). However, at positive potentials (> 0 mV), the ΔF/F voltage dependence becomes shallower: −0.4% per 100 mV (± 0.12, S.E.M., n = 5, R² = 0.97, Figure 5g). This greater than 3-fold change in the response at negative vs positive potentials is statistically significant (p = 0.037, paired t-test, n = 5, Figure S15). Because the fluorescence response of pre-formed LUnAR RhoVR on the plasma membrane is linear with respect to voltage (Figure S5e), this non-linear behavior may be due to underlying voltage changes in the ER. An intriguing possibility is that the ER rectifies, or opposes, plasma membrane potential commands at values more positive than 0 mV, and investigations are underway to explore the cellular and molecular mechanisms underlying this observation.

Conclusion

In this work, we report a new small-molecule sensor capable of targeting subcellular membranes. We show that LUnAR RhoVR is capable of targeting the ER of living cells and reporting on membrane potential fluctuations in this organelle in intact cells. This is the first report of a small molecule voltage dye used to measure ER potential dyanmics in intact cell; previous methods performed measurement in isolated membranes or vesicles.^5–7 Although the electrical properties measured in cell-free/isolated preparations of ER suggest that there is no potential difference across the ER,⁵⁶ recent reports show that synchrony exists between the plasma membrane potential changes and optical signals from an internal-membrane expressed GEVI.¹⁰

Using dual plasma membrane patch clamp and ER voltage imaging (Figure 5), we corroborate these previous observations and extend them into a new cell type. By combining patch clamp experiments with ionic calibration (Figure 3) and chemical targeting we directly establish that the directionality of the ER membrane potential change corresponds to the voltage change in the plasma membrane – meaning a depolarization at the plasma membrane (positive voltage outside) leads to a depolarization of the ER membrane (positive voltage in lumen). Further, our results reveal that ER membrane potential signals linearly track plasma membrane at negative potentials, but appear rectified at positive potentials. Further studies are geared towards exploring the nature of this functional coupling between the plasma membrane and ER membrane. We envision that LUnAR RhoVR will be an important addition to the suite of methods available for monitoring ER and other organelle physiology. Partnering LUnAR RhoVR with other lipid-based HIDE probes will allow access to the physiology of additional organelles.