New Molecular Scaffolds for Fluorescent Voltage Indicators

Abstract

The ability to non-invasively monitor membrane potential dynamics in excitable cells like neurons and cardiomyocytes promises to revolutionize our understanding of the physiology and pathology of the brain and heart. Here, we report the design, synthesis, and application of a new class of fluorescent voltage indicator that makes use of a fluorene-based molecular wire as a voltage sensing domain to provide fast and sensitive measurements of membrane potential in both mammalian neurons and human-derived cardiomyocytes. We show that the best of the new probes, fluorene VoltageFluor 2 (fVF 2) readily reports on action potentials in mammalian neurons, detects perturbations to cardiac action potential waveform in human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes, shows a substantial decrease in phototoxicity compared to existing molecular wire-based indicators, and can monitor cardiac action potentials for extended periods of time. Together, our results demonstrate the generalizability of a molecular wire approach to voltage sensing and highlights the utility of fVF 2 for interrogating membrane potential dynamics.

Graphical Abstract

Optical methods to measure biochemical and biophysical events in living cells provide a powerful approach to monitor cellular physiology in a non-invasive and high-throughput manner. The success of such light-based ventures depends critically on the ability to design and construct molecules that change their optical properties – for example, color, fluorescence intensity, or lifetime – in response to changes in the cellular environment.¹ Fluorescence microscopy is one of the most commonly used modalities because it is operationally simple, instrumentation is widely available, and an ever-growing library of small molecule fluorescent indicators exists to probe the dynamics of a host of cellular analytes, properties, and structures.^2–5 Of particular interest is the plasma membrane. Changes in the electrochemical potential across the lipid bilayer profoundly shapes cellular physiology. Excitable cells employ a consortium of ion channels both to maintain tight control over their membrane potential (V_m) and to initiate and propagate rapid changes in V_m. Rapid changes in V_m such as an action potential (AP), drive the unique physiology of excitable cells like neurons and cardiomyocytes. In neurons, APs evoke the release of neurotransmitter into the synaptic cleft, and, in heart tissue, waves of APs coordinate contraction and maintain regular rhythm. Disruption of the frequency, timing, and/or shape of APs are linked to serious human diseases ranging from epilepsy to long QT syndrome. Because of the importance of V_m to both health and disease, robust methods to optically monitor membrane potential remain a critical complement to more traditional approaches. The gold-standard for measuring V_m and APs in live cells is electrophysiology: direct determination of V_m through the physical interaction between the cell of interest and an electrode. Electrophysiology is highly invasive, low throughput, and difficult to interpret in samples like cardiac tissue or cardiomyocyte monolayers, where electrical coupling between cells confounds single-cell measurement. Optical recording of membrane potential using voltage-sensitive fluorescent indicators provides an attractive alternative to probe V_m and AP dynamics in multiple cells, in monolayers, or 3D tissue.⁶

Recently, we initiated a program to develop a new class of voltage-sensitive fluorescent indicators that utilize photoinduced electron transfer (PeT) as a rapid trigger to sense changes in V_m.^7–8 These small molecule voltage-sensitive fluorophores, or VoltageFluors (VF dyes), combine a xanthene-based dye as a fluorescent reporter and a conjugated molecular wire that localizes the indicator to the cell membrane and facilitates PeT from an electron-rich aniline donor to the fluorophore within the low dielectric environment of the lipid bilayer. Phenylene-vinylene (PV)-based conjugated molecular wires (Scheme 1, VF2.1.Cl),^9–10 are attractive because of their exceptionally low electron transfer attenuation values in donor-bridge-acceptor (DBA) systems (β = 0.04 Å⁻¹).^11–12 Within a PV wire framework, voltage sensitivity can be improved by altering the redox potentials of the fluorophore electron acceptor and the aniline electron donor.¹³ Additionally, fluorophores such as rhodamine⁹ and silicon-rhodamine¹⁰ can be substituted for fluorescein, after some adjustment to the identity of the aniline donor. To date, we have not explored alterations to the identity of the molecular wire component of PeT-based voltage indicators. Now, we present a new class of PeT-based voltage sensitive fluorescent indicators that use a 9,9-dimethyl-9H-fluorene monomer in place of the canonical 1,4-divinylbenzene moiety (Scheme 1). In other DBA scaffolds, 2,7-oligofluorene bridges effectively facilitate electron transfer across large distances and demonstrate low β values (0.09 Å⁻¹) comparable to values for PV wires (0.04 Å⁻¹).¹⁴ Because β values depend both on the identity of the wire and the donor/acceptor pair,^15–16 we wanted to explore fluorene molecular wires as a platform for optical voltage sensing, demonstrating the generalizability of a PeT-based approach to voltage sensing. We now report the design, synthesis, characterization, and application of a new series of fluorine-based VoltageFluors, or fVF dyes.

Scheme 1.

Synthesis of fluorene VoltageFluor dyes (fVF dyes).

The synthesis of fluorene-based voltage indicators starts with Suzuki-Mirayura cross-coupling of bromo-iodo-fluorene 1¹⁷ with either boronic ester 8¹⁸ or phenylboronic acid, providing monomeric fluorenes 2 and 3 as bright yellow solids

(Scheme 1). Attachment to a sulfonated dichlorofluorescein was achieved by Pd-catalyzed cross-coupling of a pinacol boronic ester with the terminal aryl bromide to provide 4 and 5. Suzuki-Mirayura cross-coupling yielded voltage indicator 6 (fVF 1) and indicator 7 (fVF 0), which lacks an aniline donor (Scheme 1). For the wires with electron rich donors (12, 13; Scheme 2), transformation of the nitro group to an aniline was performed with tin(II) chloride to yield wires 14 and 15. This was followed by reductive amination of formaldehyde with NaCNBH₃ to provide alkylated wires 16 and 17. These were prepared in a similar method to Scheme 1 to yield electron-rich voltage indicators 20 (fVF 2) and 21 (fVF 3).

Scheme 2.

Synthesis of electron-rich fluorene VoltageFluor dyes

New fVF dyes have a λ_max centered around 520 nm and a second major absorption band around 340 nm arising from the fluorene molecular wire (Fig. 1, Fig. S1, Table 1). Each fVF dye has a maximum emission around 535 nm, indicating little ground state interaction between the fluorene-based molecular wire and xanthene chromophore. fVF 1, 2 and 3 have lower fluorescence quantum yields (Φ_fl, 0.05 to 0.19) than control indicator fVF 0 (0.77, Table 1).

Figure 1.

Characterization of fluorene VoltageFluor 2 (fVF 2). a) Live cell fluorescence microscopy image of fVF 2 in HEK cells. Scale bar is 10 μm. b) Normalized absorption and emission spectra of fVF 2. Spectra were acquired in PBS (pH 7.2) +0.1% SDS. For emission scan, excitation was provided at 485 nm. c) Voltage sensitivity of fVF 2 in patch-clamped HEK cells. d) Plot of ΔF/F vs. membrane potential (in mV) for fVF 2. Red line is the line of best fit. Error bars are standard error of the mean for 8 independent determinations. e) Live-cell, wide-field fluorescence images of rat hippocampal neurons stained with 500 nM fVF 2. Scale bar is 20 μm. f) Representative ΔF/F plot of evoked neuronal activity of a single cell recorded optically with fVF 2.

Table 1.

Properties of fluorene VoltageFluor dyes (fVF dyes)

Entry[a]	R1	R2	Absorbance [nm][a]	Emission [nm][a]	QY[a]	% ΔF/F[b]	SNR [b]
fVF 0 (7)	-H	-H	519	535	0.77	−0.3 ± 0.03	1.9:1
fVF 1 (6)	-N(Me)2	-H	519	534	0.05	4.7 ± 0.5	22:1
fVF 2 (20)	-N(Me)2	-H	520	535	0.07	10.5 ± 0.8	39:1
fVF 3 (21)	-N(Me)2	-OMe	520	535	0.19	12.2 ± 1.6	13:1

^[a]Measured in PBS + 0.1% SDS (pH = 7.2).

^[b]per 100 mV.

Recorded in HEK 293T cells at 0.5 kHz optical sampling rate.

To measure the voltage sensitivity of these indicators, we used whole-cell voltage-clamp electrophysiology in tandem with epifluoresence microscopy. By applying voltage steps ranging from +100 mV to −100 mV in 20 mV increments to HEK293T cells stained with fluorene voltage indicators, we observe that indicators fVF 1–3 possess moderate sensitivity to changes in V_m. Similar to PV-based molecular wire voltage indicators, fVF 1–3 become brighter in response to depolarizing (more positive) membrane potentials. The two most sensitive compounds, fVF 2 and 3, have sensitivities of 11 and 13% ΔF/F per 100 mV, respectively; however, fVF 2 has an overall signal that was much brighter in cells, resulting in better SNR (Fig. 1, Fig. S2, Table 1). fVF 1 is also very bright in cells but has low SNR due to a low sensitivity (5% ΔF/F). Somewhat surprisingly, the electron-deficient compound fVF 0 (7, Scheme 1) exhibits a small amount of voltage sensitivity, −0.3%

ΔF/F per 100 mV, becoming less fluorescent in response to depolarizing potentials (Table 1, Fig. S2, Table 1), the opposite the opposite of every molecular wire indicator synthesized in our laboratory. We chose to characterize fVF 2 in subsequent experiments due to its brightness and superior SNR.

fVF 2 readily detects fast changes in membrane potential in mammalian neurons. In cultured rat hippocampal neurons, fVF 2 gives clear membrane staining (Fig. 1e) and faithfully records evoked action potentials with an average ΔF/F of 5.1% and SNR of 21:1 (N = 54 spikes, Fig. 1f, S4). fVF 2 clearly resolves spontaneous activity in cultured rat hippocampal neurons (Fig. S5). Despite the lower nominal voltage sensitivity of fVF 2 relative to a first generation VoltageFluor dye (VF2.1.Cl – 27% ΔF/F per 100 mV in HEK cells;⁷ 10% ΔF/F and SNR of 43:1 in evoked action potentials, n = 54 spikes, Fig. S4), the improved brightness of fVF 2 relative to VF2.1.Cl (1.2x brighter in neurons, Table S1) makes it useful for recording action potentials in neurons. When we evaluate fVF 2 against an electrochromic VSD with a similar voltage sensitivity, di-4-ANEPPS, we observe evoked spikes with a ΔF/F of −1% and SNR of 15:1. However, a 5-fold higher concentration was needed to make these recordings (Fig. S4, SI Table 1). In addition to reporting on neuronal activity, we detect no changes to neuronal membrane properties or action potential kinetics when comparing the electrophysiological parameters of neurons with or without the presence of fVF2 (Fig. S6, Table S2).

We sought to use fVF 2 for optical measurements of cardiac AP waveforms to provide a holistic assessment of drug cardiotoxicity in vitro, a major goal of the Comprehensive in vitro Proarrythmic Assay (CiPA) initiative.^19–21 We cultured hiPSC-CM monolayers and tested the ability of fVF 2 to report cardiac AP waveforms in spontaneously beating monolayers.^21–25 fVF 2 clearly stains the sarcolemma of hiPSC-CMs and faithfully reports ventricular-like AP waveforms, showing a large increase in fluorescence just before contraction of the monolayer (Fig. 2a-e). Using methods previously described,²⁶ we calculate action potential duration (APD) for each AP waveform in the fluorescence trace at 70, 50, and 10% of the maximum depolarization (APD30, APD50, APD90, respectively).^26–28 To correct for APD variation arising from difference in beat rate from spontaneously beating monolayers, we used Fridericia’s formula to provide a beat-rate corrected APD (cAPD).^28–29 From our optical measurements, we calculated cAPD90 values from 500–700 ms in spontaneously beating monolayers after 14 days in culture, consistent with previous reports in hiPSC-CMs. ^{27, 30}

Figure 2.

Fluorene VoltageFluor 2 (fVF 2) reliably reports on cardiac action potential (cAP) dynamics in human induced pluripotent stem cell derived cardiomyocytes (hiPSC-CMs). a) Brightfield. b) Membrane localization of fVF 2 (1 μM) in hiPSC-CM monolayers. c) Hoescht 33342 nuclear stain. d) Merge of membrane and nuclear stains. Scale bar is 20 μm. e) Representative image used to acquire functional AP data. Scale bar is 10 μm. f-i) Representative fluorescence traces acquired using 1 μM fVF 2. f) Baseline measurement of spontaneously contracting monolayer. Treatment with cisapride (300 nM) results in g) prolonged APs, h) shorter and more frequent APs, and i) early after depolarizations (EADs). j) Overlay of single APs from f-h to highlight observed waveform changes from cisapride treatment. The baseline trace in black shows a normal ventricular-like shape, the red trace depicts an extended AP after treatment (from panel g), and the blue trace depicts a shortened AP from a tachycardia-like phenotype (from panel h).

To demonstrate the utility of fVF 2 for parsing the pharmacological effects of drug treatment on cardiomyocytes, we treated hiPSC cardiomyocytes with cisapride. Cisapride, formally a useful gastroprokinetic agent, was withdrawn from the US market in 2000 due to its connection to torsades de pointes (TdP) induced by acquired long QT-syndrome caused by blockage of K_v11.1.³¹ Cisapride is also one of 12 training and calibration compounds used in the CiPA initiative.^{20, 32} Observation of cardiomyocyte monolayers treated acutely with cisapride results in several phenotypic alterations to cardiac AP, readily detectable by fVF 2. At a concentration of 300 nM, we saw three different manifestations of the effect of prolonged cAPD caused by I_Kr blockade:^33–34 an extended phase 3 (Fig. 2g), a tachycardia-like train of drastically shortened APs (Fig. 2h), and the appearance of early-after depolarizations (EADs) (Fig. 2i). K_v11.1/hERG channel blockade results in action potential prolongation (Fig. 2g) which has been connected to a higher risk of TdP and higher arrhythmogenic potential. The rapid, subthreshold spiking activity may be analogous to tachycardia,³⁵ which was accompanied by a loss in monolayer automaticity (Fig. 2h). The appearance of EADs (Fig. 2i) corresponds to the observed cAPD prolongation, and significantly increase the risk of arrhythmia and TdP.

Monitoring the effect of potentially cardiotoxic drug relies on the ability to make stable, long-term recordings from cardiomyocytes. When using voltage-sensitive fluorescent indicators, this often requires careful titration of illumination intensity and indicator concentration to minimize phototoxicity.^35–37 Therefore, we were pleased that fVF 2 displays lower phototoxicity in hiPSC-CMs relative to VF2.1.Cl, which itself requires an order of magnitude lower light power than di-4-ANEPPS, thereby avoiding phototoxic effects.³⁷ Others note phototoxic effects of di-4-ANEPPS can begin within seconds of illumination.³⁸ We stained hiPSC-CM monolayers with either fVF 2 or VF2.1.Cl and continuously illuminated for 10 minutes and optically recorded membrane potential dynamics every minute. Despite the initial photobleach of fVF 2 (Fig. 3a, Fig. S2i), the shape of recorded action potentials (Fig. 3c), SNR (Fig. 3e), and action potential duration (Fig. 3f,g) remain relatively constant (Fig. 3c and 3d, S7). In sharp contrast, however, VF2.1.Cl had a dramatic, detrimental impact on cardiomyocyte function. Although VF2.1.Cl initially has a high SNR compared to fVF2 (Fig. 3b,e), SNR drops quickly after the first minute of illumination (Fig. 3b,e), and both action potential shape (Fig. 3d) and duration (Fig. 3f,g) undergo substantial and significant changes, as early as 2 minutes into illumination (Fig. 3b,d,f,g). Even after 4 minutes, only sub-threshold activity was recorded with VF2.1.Cl and required an automated analysis script to detect these events (Fig. S7h). After 5 minutes, monolayers imaged with VF2.1.Cl cease to contract (Fig. 3d and S7). However, monolayers imaged with fVF2 continue to beat even after 10 minutes of continuous illumination (Fig. 3c and S7). Increasing the illumination intensity from 9 to 29 mW/mm² when making recordings with fVF 2 did not alter cAPD (Fig. S7l). Toxicity requires both dye and illumination, as regions of the coverslip not exposed to illumination light beat properly, in the case of fVF2 (Fig. S8). For VF2.1.Cl, we observe a small increase in cAPD values, which was not statistically significant (Fig. S8). Together, these results suggest that fVF 2 has lower phototoxicity than VF2.1.Cl and can be used to measure activity in cardiomyocyte monolayers for prolonged time periods. Previous reports note slight differences in the structure of transmembrane probes can explain differences in phototoxicity.³⁸ Experiments are underway to probe the precise molecular mechanisms underlying the faster photobleaching but reduced phototoxicity of fVF 2 compared to VF2.1.Cl.

Figure 3. fVF 2 displays low phototoxicity in cardiomyocyte monolayers.

Fluorescence intensity vs time for a) fVF 2 and b) VF2.1.Cl in monolayers of hiSPC-CMs. The raw fluorescence intensity from an entire field of view over an entire 10 s recording session is plotted vs. the total illumination time (in minutes). Individual action potential (AP) traces for c) fVF 2 and d) VF2.1.Cl are indicated by red stars in panels (a) and (b). Plot of mean e) signal-to-noise ratio (SNR ) f) cAPD30, and g) cAPD50 as a function of total illumination time for fVF 2 (black) and VF2.1.Cl (blue). For panels e-g, mean values are determined from n = 3 independent trials, and error bars are ± standard error of the mean. Statistical tests are two-tailed, unpaired t-test for each cAPD at the indicated time vs. t = 0. ** = p<0.005, *** = p<0.001, **** = p<0.0001.

To assess the ability of fVF 2 to measure changes to cardiac electrophysiology in responds to chronic drug treatment, we calculated IC₅₀ values for cisapride using in-well dose escalation with optical recording of cAPD. Measurements were taken after incubation with increasing cisapride concentrations in each well from 0.1 to 300 nM. Our optical measurements show an increase in cAPD90 (IC₅₀ = 10.6 nM, 14 days in culture) up to 300 nM of cisapride (Fig. 4e,f), which is in the range of IC₅₀ values previously measured for cisapride in other in vitro studies (SI Table 3).³² We also observe an increase in cAPD50 up to 100 nM, but a decrease at higher concentrations of cisapride. Similarly, measured cAPD30 seems to have little variation from DMSO vehicle control; however, a decrease is detected at higher concentrations of cisapride (Fig. 4e,f). Together, these results demonstrate the ability to fVF 2 to enable the rapid assembly of dose-response data in hiSPCs using an all-optical approach.

Figure 4.

Fluorene VoltageFluor 2 (fVF 2) reveals changes to cardiac action potentials (AP) upon treatment with cisapride. (a) Action potential duration (APD) values are calculated from the maximum derivative of the depolarization to the repolarization at 70, 50, and 10% of the maximum depolarization value, are corrected for beat-rate, and reported as cAPD30, cAPD50, and cAPD90 values, respectively. Treatment of monolayers cultured for 14 days with doses of cisapride from 0.1 to 300 nM cisapride (red) results in (b) little change in the cAPD30, (c) a moderate increase in the cAPD50, and (d) a clear increase in cAPD90. Black points indicate DMSO-treated control samples. Plots indicate mean values ± standard error of the mean for n = 4 independent experiments.

In summary, we present the design, synthesis and application of a new class of molecular wire-based fluorescent indicators. We show, for the first time, that fluorene-based molecular wires provide a platform for PeT-based voltage sensing. In general, these fluorene-based indicators have lower nominal voltage sensitivities (ΔF/F per 100 mV of 5 to 13% in HEK cells, compared to approximately 27% for VF2.1.Cl) to changes in membrane potential, but are brighter, than their phenylene-vinylene counterparts. fVF 2 exhibits adequate sensitivity and excellent brightness in cells for reporting AP waveforms in neurons and cardiomyocytes with high SNR. More importantly, fVF 2 displays substantially reduced phototoxicity in cardiomyocytes relative to VF2.1.Cl, allowing for prolonged, continuous measurement of cardiomyocyte activity. Fluorene-based molecular wires may provide an attractive, general solution to the phototoxicity often associated with voltage-sensitive fluorescent indicators.