Fluorescence Imaging of Cell Membrane Potential: From Relative Changes to Absolute Values

Membrane potential is a fundamental property of biological cells. Changes in membrane potential characterize a vast number of vital biological processes, such as the activity of neurons and cardiomyocytes, tumorogenesis, cell-cycle progression, etc. A common strategy to record membrane potential changes that occur in the process of interest is to utilize organic dyes or genetically-encoded voltage indicators with voltage-dependent fluorescence. Sensors are introduced into target cells, and alterations of fluorescence intensity are recorded with optical methods. Techniques that allow recording relative changes of membrane potential and do not take into account fluorescence alterations due to factors other than membrane voltage are already widely used in modern biological and biomedical studies. Such techniques have been reviewed previously in many works. However, in order to investigate a number of processes, especially long-term processes, the measured signal must be corrected to exclude the contribution from voltage-independent factors or even absolute values of cell membrane potential have to be evaluated. Techniques that enable such measurements are the subject of this review.

Keywords: membrane potential imaging, cell membrane potential, potentiometric sensors, genetically-encoded voltage indicators, voltage-sensitive dyes

Membrane potential is a fundamental physiological property of cells involved in the control of various biological processes [1]. Fast changes in membrane potential (up to hundreds of milliseconds) drive the activity of electrically excitable cells, such as neurons and cardiomyocytes. Slower changes of resting membrane potential characterize cell cycle progression [2,3], differentiation [4], insulin secretion [5], circadian firing cycles of SCN neurons [6], etc. Decrease of resting membrane potential induces cell proliferation and is related to cancer progression [7,8] as well as other disorders [9,10,11]. Therefore, the development of tools and techniques for membrane potential measurements is an important prerequisite for further progress in many areas of cell biology, biosensing, medicine, pharmacology, etc.

Electrophysiological methods provide accurate voltage measurements but possess several critical drawbacks. Injury of the membrane by microelectrodes during an experiment can alter membrane characteristics and become a non-negligible source of error [12,13]. The number of cells measured simultaneously is limited to a single-cell (patch-clamp) or hundreds of cells (multi-electrode arrays) with relatively poor spatial resolution [14]. Besides, electrophysiological methods require complicated preparation procedures, especially for in vivo measurements, and are challenging to use in long-term studies [15].

An alternative approach is membrane potential imaging, which utilizes methods to monitor voltage-dependent changes of the optical signal detected either from label-free cells [16,17,18], or from molecular, biomolecular or nanoparticle-based potentiometric sensors introduced into cells. Currently, the techniques based on different types of optical signals have been developed, including fluorescence, light scattering, transmitted light intensity, birefringence, second harmonic signals, etc. [18,19]. Voltage imaging does not have a limit on the number of analyzed cells, ranging from single-cell measurements to large cell populations. Spatial and temporal resolutions of voltage imaging depend on the applied sensor, and currently available toolkit provides sensors and methods to perform measurements with subcellular or single-cell spatial resolutions and investigate processes starting from millisecond timescale, which is typical for neuronal or cardiomyocyte action potentials, to days and weeks, which is an important timescale to study pathological processes.

The majority of membrane potential imaging studies record voltage-induced alterations of fluorescence intensity detected from the potentiometric sensors located in cells. This approach reports the relative changes of membrane potential and for a large number of biological processes was shown to provide robust results. For example, cell membrane potential imaging has been successfully applied in biological and biomedical research for the investigation of brain functioning [20,21,22,23,24], development of the vertebrate nervous system [25], cardiac electrical activity [26,27,28], cell development [29,30,31], etc. However, the detected fluorescence intensity alterations are associated not only with voltage changes, but also with a number of voltage-independent processes, such as cell motion or intracellular transport of macromolecules. When the value of voltage-induced fluorescence changes is much larger than the effect of non-target processes, the magnitude of error will be modest and can be neglected. Otherwise, specific techniques have to be applied to eliminate the contribution of voltage-independent factors from the detected fluorescence signal. Besides, to obtain quantitative results from membrane potential imaging experiments, the correspondence between the magnitude of detected optical signal and voltage values have to be derived using calibration techniques.

Techniques for membrane potential imaging, in which the contribution from voltage-independent factors into the detected optical signal is not taken into account, have been already covered in a number of reviews [32,33,34,35,36,37,38,39]. These studies reviewed classes of available sensors, principles of functioning, measurement methods, range of applicability, etc. Here, we rather focus on the techniques that provide the possibility to eliminate contributions in fluorescence signal that occur from factors other than voltage and methods that are applied to derive absolute cell membrane potential values. The two most common classes of sensors for membrane voltage imaging, fluorescent organic dyes and fluorescent genetically-encoded voltage indicators, will be considered.

The review is organized as follows. In Section 2, we will briefly describe two widely used approaches for membrane potential imaging: recording the intensity of fluorescence and measurement of excited state lifetime. In Section 3, we will describe the most commonly used types of molecular and biomolecular sensors. In Section 4, we will cover the techniques used to eliminate fluorescence changes caused by voltage-independent factors and perform reliable measurements of cell membrane potential changes in different processes. Finally, in Section 5 we will consider calibration techniques that allow deriving absolute values of cell membrane potential.

Two physical properties affected by voltage are most commonly used to record membrane potential: fluorescence intensity and excited state lifetime.

Fluorescence intensity can reflect the relative changes of membrane potential, but this property is also subjected to a number of voltage-independent factors (see Section 4). Measurements can be performed with standard fluorescence or confocal microscope equipped with a digital camera [40].

Excited state lifetime, the time between absorption of excitation photon by the fluorophore and the re-emission of fluorescence photon, can be expressed as follows:

Here, kfl is the rate constant for fluorescence, knr is the rate constant for non-radiative decay. For several sensors, linear voltage-dependence of τexc was demonstrated [41,42]. Measurements of excited state lifetime can be performed with equipment for time-resolved fluorescence microscopy [43].

Membrane voltage can affect both fluorescence intensity and excited state lifetime, and measurement of these properties is widely used for membrane potential imaging. To date, several classes of sensors with different functioning principles have been developed. Sensors with linear voltage-dependence of fluorescence properties are preferred for obtaining unambiguous correspondence between detected optical signal and membrane voltage.

The currently available toolkit for membrane potential imaging contains a large variety of molecular and biomolecular sensors with voltage-dependent fluorescence. These sensors can be composed of a single molecule or include several components. Below we follow one of the commonly-used classifications of molecular and biomolecular sensors.

Electrochromic dyes are characterized by a shift of absorption and fluorescence spectra in an external electric field (Stark effect). Photoexcitation of electrochromic dyes induces intramolecular charge transfer resulting in different electron distributions in the ground and excited states. Interaction with an external electric field alters the energies of the ground and excited states to a different extent, and the shift of absorption/emission bands is observed (Figure 1a). A number of methods utilizing electrochromic dyes have been developed for membrane potential imaging. These methods are based on monitoring the changes of absorption/emission intensity at the selected wavelength, monitoring the changes of total fluorescence excited at the selected wavelength, as well as methods based on two-photon absorption and second harmonic generation, as described in a number of comprehensive reviews [33,34,44,45,46,47,48].

The main advantage of electrochromic dyes is the ultrafast kinetics of fluorescence response to voltage change, which enables researchers to monitor processes with submillisecond temporal resolution. On the other hand, due to small magnitudes of voltage-induced spectral shifts, around 10 nm per 100 mV, electrochromic dyes do not demonstrate high voltage sensitivity. Usually, the magnitude of fractional change of fluorescence intensity (ΔF/F) per 100 mV lies in the 10–20% range, limiting the accuracy of measurements [33,34,49,50].

Fluorescence resonance energy transfer (FRET) is a process that can occur between two light-sensitive molecules, if the emission band of the first molecule (donor) and the absorption band of the second molecule (acceptor) overlap (Figure 1b). Emission of the donor fluorophore is absorbed by the acceptor via a non-radiative energy transfer, and the efficiency of this transition is proportional to the inverse sixth power of the distance between the donor and the acceptor. Two types of FRET-based sensors are most widely applied. In the sensors of the first type, the acceptor is not fluorescent. Therefore, in experiment only the donor fluorescence can be detected, which enhances when the distance between the donor and acceptor increases. In the sensors of the second type, both components are fluorescent and changes in donor fluorescence, acceptor fluorescence, or both signals can be monitored.

Usually, the donor is attached to the extracellular surface of the cell membrane, and its position does not change in the experiment. The negatively charged acceptor is placed inside the plasma membrane and moves between extracellular and intracellular membrane surfaces in response to voltage changes. Therefore, the change in the membrane potential leads to an increase or decrease in the distance between the donor and the acceptor. In such systems, FRET efficiency and, accordingly, the detected fluorescence intensity become potential-dependent (Figure 1b).

While early variants of FRET-based sensors had response times lasting several milliseconds [51,52], more recent examples provide sub-millisecond temporal resolution [53,54,55]. An advantage of FRET-based dyes is the high voltage sensitivity with fractional fluorescence change (ΔF/F) ranging from dozens to hundreds of percent per 100 mV [56,57]. A general obstacle related to sensors of this type is the necessity to accurately tune donor and acceptor concentrations. More information about the available FRET-based dyes and their applications in membrane potential imaging can be found in a number of reviews [33,57,58,59,60].

Molecular wire-based dyes are two-component systems consisting of a fluorophore and an electron-rich moiety (electron donor) connected with a molecular wire. After photoexcitation fluorescence emission and electron transfer from the donor become competing processes that convert fluorophore back to the ground state. The level of fluorescence quenching is determined by the rate of electron transfer. In an external electric field the rate of electron transfer depends on the magnitude of the field as well as the angle between the molecular wire and the electric field with the largest voltage-sensitivity observed when the wire is oriented parallel to the electric field [61].

This type of construct was implemented in VoltageFluors, a recently developed [pclass of dyes [62]. The fluorophore is attached to the outer surface of the cell membrane, an electron-rich moiety is located inside the lipid bilayer, and the connecting molecular wire is oriented perpendicular to the membrane surface, therefore, parallel to the membrane electric field (Figure 1c). In such a system, cell membrane potential efficiently alters the rate of electron transfer along the molecular wire and, therefore, the fluorescence intensity detected from the fluorophore.

The main advantage of sensors based on photoinduced electron transfer is the ultrafast response rates of fluorescence, which lie in the range of dozens of nanoseconds [63] and provide the ability to record the fastest processes in neurons and cardiomyocytes [64]. The most recent sensors also demonstrate high voltage sensitivity with relative change of fluorescence intensity exceeding 60% per 100 mV [65,66,67]. Recent works have extensively reviewed molecular wire-based dyes considering the sensors developed to date and their applications [33,35,63,68].

Membrane-permeable charged dyes placed in the vicinity of a cell with non-zero membrane potential will move between the cell interior and extracellular medium until the electrochemical equilibrium is established. At equilibrium the ratio of extracellular and intracellular dye concentrations is determined by cell membrane potential in accordance with the Nernst equation:

Here, Vmem is cell membrane potential, R is the universal gas constant, T is the absolute temperature, Z is the charge of the dye, F is the Faraday constant, [Dye]out and [Dye]in are the extracellular and intracellular concentrations of a charged dye, respectively. Therefore, if the relation between detected fluorescence intensity and dye concentration is established, cell membrane potential can be directly evaluated from the fluorescence measurements (Figure 1d). The fluorescence intensity of dye molecules located in the cell interior can be detected with the flow cytometry technique [69]. To evaluate extracellular dye concentration fluorescence from the intracellular region is measured after complete depolarization of cells (zero membrane potential), i.e., when extracellular and intracellular dye concentrations are equal. Such depolarization can be achieved by adding ionophores that form pores in the membrane, such as paraformaldehyde [70] or gramicidin [71]. For anionic dyes, usually oxonols [72], intracellular concentration is low in the normal state of a cell (negative membrane potential) and becomes higher upon cell depolarization, i.e., when cell membrane potential is increased. Cationic dyes, usually carbocyanines or rhodamine derivatives, demonstrate the reverse trend [29,73]. The main advantage of redistribution dyes is the high magnitudes of fractional fluorescence change in response to voltage. On the other hand, the response times are limited to the timescale of seconds and minutes required for dye molecules to penetrate the membrane. Therefore, cationic and anionic fluorescent dyes are used only to monitor potential changes in long-term processes, such as cell differentiation [29,69,74]. More information about available redistribution dyes and their applications for membrane potential imaging can be found in a number of works [68,73,75,76].

Sensors constructed as a fusion of proteins with voltage-independent fluorescence and non-fluorescent transmembrane proteins or protein domains that undergo structural reorganization upon voltage change were shown to possess voltage-dependent fluorescence (Figure 1e). Conformational changes in the voltage-sensitive domain induced by membrane depolarization or hyperpolarization are passed to the fluorescent protein via a peptide linker, leading to the change of the fluorophore local environment and the change of fluorescence intensity. Fluorescence voltage dependence properties were shown to be determined by the nature of the fluorescent protein and the voltage-sensitive domain, the length of the peptide linker, and the position for the fusion of two components [77,78,79,80].

Several sensors were constructed as a fusion of two fluorescent proteins with overlapping emission/absorption bands to the same voltage-sensitive domain. For such sensors structural reorganization of the domain alters the relative position of fluorescent proteins, therefore, the level of FRET between them (Figure 1e). The fluorescence intensities of the donor and acceptor become voltage-dependent and both signals can be used to monitor membrane potential changes [81,82]. The currently available sensors from this class demonstrate relative changes of fluorescence signal up to ∼50% per 100 mV and the response times in the range of several milliseconds [77,78,79,80,82]. Up to date, a large variety of voltage-sensitive domain-based genetically-encoded voltage indicators have been developed, and their spectral properties, voltage-dependence characteristics and applicability were the subject of many reviews [83,84,85,86,87,88,89].

Microbial rhodopsins are transmembrane proteins that demonstrate a large blue shift of absorption band upon decrease of cell membrane potential, which is associated with voltage-induced chromophore deprotonation. At positive voltage values absorption band of microbial rhodopsins largely overlap with the emission band of blue/green/yellow fluorescent proteins. Therefore, at positive voltages the emission of a fluorescent protein attached to non-fluorescent microbial rhodopsin will be effectively absorbed by the rhodopsin via FRET, resulting in the low intensity of the detected fluorescence signal. Voltage decrease is accompanied by the lowering of the red-shifted absorption band attributed to rhodopsins with protonated chromophore, resulting in the decrease of FRET efficiency and the enhancement of detected fluorescence (Figure 1f). A few examples were reported for this type of sensor with relative changes of fluorescence intensity up to 18% per 100 mV and the response times ranging from <1 ms to ∼7 ms [90,91,92].

Several proteins from the family of microbial rhodopsins, such as archaerhodopsin-3 and its mutants, possess the intrinsic voltage dependence of fluorescence intensity (Figure 1g) [93,94]. The mechanism of fluorescence voltage-dependence is still not clear, even though several studies on this problem have been reported [95,96]. Up to date, a large number of rhodopsin-based sensors have been developed using a directed evolution approach, which does not require detailed knowledge of underlying mechanisms [97,98,99]. The main advantages of sensors from this class are the fast response rates, up to submillisecond temporal resolution, high sensitivity of fluorescence signal with the fractional change of fluorescence intensity up to 90% per 100 mV, and the ability to vary absorption band maxima values in a wide range [97,98,99,100,101,102,103,104]. The main drawback limiting the application of sensors from this class is the dimness of the detected fluorescence. Fluorescence quantum yields of currently available rhodopsin-based sensors are in the range 0.8–1.2% [83,99,105]. Recent reviews have covered the currently available rhodopsin-based sensors focusing on their spectral properties and applicability for membrane potential imaging [83,84,87].

The fluorescence signal recorded in membrane potential imaging experiments is affected by a number of concomitant processes besides membrane potential alterations. When voltage-induced changes of fluorescence signal are much larger than the effect of side processes, the latter can be neglected. For such processes, robust results can be obtained by monitoring voltage-dependent changes of fluorescence intensity with a standard fluorescence or confocal microscope. However, in a large number of biological processes voltage-independent side factors altering fluorescence intensity become a significant source of error. Investigation of such processes with membrane potential imaging technique requires the application of methods that eliminate fluorescence changes caused by factors other than voltage.

The most frequent source of error in cell membrane potential imaging experiments is the change in the local concentration of sensors that occurs during the target process. Concentration changes can be a result of different factors, including redistribution of motile organic dyes, the motion of cells as in the case of contracting cardiomyocytes, intracellular transport of macromolecules, etc. This problem is particularly acute for long-term processes when the above-mentioned factor becomes non-negligible [28,41,106], and for in vivo studies due to sample movement relative to the camera caused by animal motion, breathing or other physiological processes [27,107]. Besides, the local concentration of fluorescent sensors decreases with time due to the bleaching of fluorophores. The rate of bleaching is not equal for all fluorophores and depends not only on the nature of the fluorophore, but also on the intensity of illumination and on the variation of the local environment, which often cannot be controlled in experiment [31,42].