Genetically-Encoded Voltage Indicators


Jump to: navigation, search

Rationale for optical imaging of neuronal activity

The main advantage of optical imaging is, in principle, its minimal invasiveness in comparison to recording techniques based on needle electrodes because even the finest microelectrode will mechanically lesion or disintegrate brain tissue. In addition, optical methods can provide a superior spatial resolution since photons interact at the molecular level while electrodes are larger than molecules.

Historical Background

The first ideas on how to implement an optical measure of neuronal activity emerged in the late 1960s from the study of changes in light scattering, birefringence, and fluorescence associated with action potentials (Cohen et al., 1968; reviewed in Villringer, 1997). Larger optical signals were obtained in the 1970s by introducing voltage-sensitive organic molecules into neuronal membranes, opening the field of voltage-sensitive dye imaging which is still intensively explored today (Peterka et al., 2011). The use of exogenously applied dyes for reporting neuronal activities has been pioneered in particular by Cohen and coworkers who demonstrated changes in dye absorption during action potentials in giant axons from squid followed by the first multisite optical recordings of network activities by means of voltage-sensitive dyes (Grinvald et al., 1977 and Ross et al., 1974). While absorption dyes were initially the focus of development, fluorescent dyes became dominant afterwards as they provide signals with higher contrast due to lower background. They also allow for epi-illumination and, hence, application to nontransparent preparations like living rodents. Despite its enormous potential, classical voltage imaging has a small signal-to-noise ratio (SNR) and is also hindered by significant cell toxicity. It is largely for these reasons that voltage-sensitive dyes have lately been superseded by calcium indicators for the imaging of network activities.

Rationale for genetically-encoded voltage indicators (GEVIs)

Genetically encoded calcium indicators (GECIs) have already been proven to successfully report neural activity in zebrafish, worms, flies, and mice with intensity changes larger than those of voltage-sensitive dyes (Horikawa et al., 2010 and Tian et al., 2009). So, why are genetically encoded voltage probes so eagerly needed? One important difference between calcium and voltage imaging of network activities is that voltage signals report subthreshold activities while calcium imaging mainly captures action potential activities in neural network analysis. Although action potentials represent the main output for postsynaptic cells, the time structure of action potential events is the result of a neuronal computation that is reflected in the time course of subthreshold electrical activity in dendritic and somatic membranes. Moreover, subthreshold activity resolves information about the spiking activities of the ensemble of neurons that are synaptically interconnected while most of this information is lost in the action potential firing information of a single neuron. Indeed, voltage imaging at the population level usually favors the recording of subthreshold postsynaptic signals from dendrites; since, they constitute a significant fraction of labeled membranes. Another important difference between calcium and voltage imaging is that voltage signals are much faster than calcium responses. Because calcium enters the cell during action potential firing, calcium imaging can be used as an indirect measure of voltage transients. However, calcium imaging usually provides sufficient single-sweep sensitivity only for action potential spiking activities (Cossart et al., 2005 and Muri and Knöpfel, 1994) but not for subthreshold synaptic communication.

Evolutions of GEVIs

From FlaSh to VSFPs

Over the past 15 years, genetically encoded optical voltage reporter proteins have considerably evolved (for reviews, see Knöpfel et al., 2006, Mutoh et al., 2011 and Perron et al., 2009a). Initial design concepts exploited voltage-dependent conformational changes associated with voltage-gated ion channels (FlaSh: Siegel and Isacoff, 1997; SPARC: Ataka and Pieribone, 2002) or their isolated voltage sensor domain (VSFP1: Sakai et al., 2001), resulting in modulation of the reporter protein fluorescence intensity. Although these first generation voltage probes were shown to optically report changes in membrane potential, their application in mammalian systems is considerably limited by their poor membrane targeting in transfected cells (Baker et al., 2007). More recently, largely improved second generation of voltage-sensitive proteins (VSFP2s) were developed in the Knöpfel lab by replacing the actuator protein in our VSFP1 (Sakai et al., 2001) with the voltage-sensing domain of the non ion channel-forming protein Ciona intestinalis voltage sensor-containing phosphatase (Ci-VSP), leading to sensors with increased targeting to the plasma membrane and reliable responsiveness to membrane potential signaling in isolated cells, brain slices, and living mice (Akemann et al., 2010, Dimitrov et al., 2007, Lundby et al., 2008, Mutoh et al., 2009, Mutoh et al., 2011 and Tsutsui et al., 2008). The Knöpfel lab also generated a series of monochromatic fluorescent probes (termed VSFP3s) with spectral variants spanning the cyan to far-red region of the visible light spectrum ( Lundby et al., 2008 and Perron et al., 2009b). Whereas VSFP2s were based on a FRET reporting mechanism for protein conformational rearrangements, VSFP3s respond to voltage changes with a modulation in the fluorescence intensity of a single fluorescent protein through a FRET-independent mechanism that is only partially understood. VSFP3s offer the advantages of a broad coverage of the color spectrum with fast overall kinetics, but with signal amplitudes smaller than VSFP2s. Despite these limitations in signal size, VSFP3s were the first probes to report spontaneous electrical activity in neuronal cultures Perron et al., 2009b. Finally the Knöpfel lab also explored two other strategies: one that employs circularly permuted fluorescent proteins, which yielded proof-of-principle voltage-sensitive probes, albeit with small fluorescence changes (Gautam et al., 2009), and the other one in which the voltage sensor domain is sandwiched between two fluorescent proteins (VSFP variants named “Butterflies”) (Akemann et al., 2012). The rationale behind this second design was to combine the advantageous properties of VSFP2s (large dynamic range) with the fast kinetics of VSFP3s.


In the first VSFP (VSFP1), the actuator was the fourth transmembrane helix (the voltage sensing domain) of the potassium channel Kv2.1. In subsequent VSFPs, this actuator was replaced with the voltage-sensitive domain (VSD) of an ascidian voltage-sensitive protein (the self-contained non-ion channel protein Ciona intestinalis voltage-sensor-containing phosphatase, or Ci-VSP). This VSD is a homolog of the VSDs of voltage-gated potassium channels (Kv channels), but unlike them Ci-VSP's VSD can be inserted in the membrane as a monomer.

Schematic design of VSFP probes derived from a combination of a voltage sensor domain and fluorescent proteins. (a) VSFP2s are FRET-based voltage sensors comprising an acceptor (e.g., Citrine) and donor (e.g., Cerulean) fluorophore fused to the fourth transmembrane segment (S4) of the voltage-sensing domain of Ci-VSP. (b) VSFP3s involve a single FP. (c) cpVSFPs are based on circularly permuted fluorescent proteins (e.g., mKate). (d) In Butterfly VSFPs, the voltage-sensing domain is sandwiched between a fluorescent protein FRET pair. PM: plasma membrane. From Perron et al., 2012

Optimizing the intracellular targeting of GEVIs

It is important to stress that many fusion proteins involving FPs are mistargeted or form intracellular aggregates (Mutoh et al., 2011). Additional fluorescence originating from these aggregates adds photon noise and decreases the signal amplitude after normalization to baseline fluorescence values (ΔF/F0). Aggregate formation is particularly troublesome in sensors that are derived from fluorescent proteins isolated from the Anthozoa class as these proteins have a high tendency to form various kinds of deposits in mammalian brain cells even when expressed in the absence of a fusion partner (Hirrlinger et al., 2005). To some extent, this behavior may reflect their origin from tetrameric proteins but it might also indicate retention within intracellular compartments (e.g., endoplasmic reticulum, ER, or Golgi apparatus) or some yet unknown affinity for intracellular organelles. It is therefore not surprising that VSFP imaging of neuronal responses in mouse brain preparations has revealed signal amplitudes that are much smaller than those obtained in the more simple Xenopus model system (Tsutsui et al., 2008 and Villalba-Galea et al., 2009) or in selected regions devoid of background fluorescence (Mutoh et al., 2009). As the effectiveness of membrane potential indicators largely depends on the protein fraction that is correctly targeted to the plasma membrane, we have applied combinational molecular trafficking strategies (Gradinaru et al., 2010 and Zhao et al., 2008) to increase the sensitivity of our biosensors. Such modifications include the introduction of Golgi trafficking signals and additional ER export motifs to favor transport along the secretory pathway to the cell surface (unpublished observations).

Hybrid voltage sensors (hVoS)

While efforts were put into designing fully genetically-encoded voltage indicators, some labs kept on working with synthetic voltage-sensitive dyes but with the perspective of associating them with fluorescent proteins. The idea stemmed from some work performed in the Tsien lab, showing that intermolecular FRET between two fluorescent molecules could be used to signal changes in the transmembrane potential (Gonzalez and Tsien 1995). In this early study, a fluorescent tag such as a fluorescently labelled lectin (lectins are sugar-binding proteins so they can bind to glycoproteins) was attached to the outer leaflet of the cytoplasmic membrane, while a hydrophobic fluorescent anion was incorporated into the membrane. The first fluorophore remains attached to the membrane while the second translocates from one side to the other in response to changes in the transmembrane potential, allowing FRET to happen or not between the two molecules. FRET between the two fluorophores is disrupted when the membrane potential is depolarized because the anion is pulled to the intracellular surface of the cytoplasmic membrane. The same principle was applied later on in the lab of Bezanilla by replacing the membrane-bound fluorophore with a membrane-anchored fluorescent protein (a farnesylated enhanced GFP or eGFP-F) (Chanda et al., 2005). Here eGFP-F is bound to the cytoplasmic side of the membrane. In the role of the hydrophobic voltage-sensing molecule, the negatively charged dipicrylamine (DPA) proved very effective. Even though it is non fluorescent, DPA can absorb energy from eGFP-F via FRET (in other words can quench its fluorescence) when the membrane potential is hyperpolarized (and DPA is pulled toward the cytoplasmic side). Therefore, fluorescence from eGFP-F becomes a genetically-addressable optical readout for membrane potential changes. Together, eGFP-F and DPA formed the first "hybrid voltage sensor" (hVoS). This initial hVoS outperformed existing fully genetically-encoded voltage indicators, in particular through its fast (submillisecond) kinetics. Since then hVOS probes have been explored and optimized in the lab of Meyer Jackson to produce versions with a roughly 3-fold higher sensitivity and signal-to-noise ratio than the original hVoS (Wang et al., 2010). To improve hVoS signals, the Jackson lab systematically varied the optical properties, membrane targeting motifs and linkages of fluorescent proteins to optimize the ΔF/F and signal/noise ratio. The best probe (hVoS 2.0) was obtained with cerulean fluorescent protein tagged N-terminally with a GAP43 motif and C-terminally with a truncated h-ras farnesylation motif. Recently a Thy1-hVoS2.0 mouse line was generated and tested, giving encouraging results in acute slices (Wang et al., 2012). A major limit of hVoSs for in vivo applications remains the delivery of DPA to the side of interest.

Summary of available GEVIs

Monochromatic (single FP) GEVIs

Year Name Voltage-sensing domain Expression system for functional characterization Chromophore(s) peak emission wavelength(s) Sensitivity
(% ΔF/F per 100 mV)
@ xx mV[1]
Response time constant
Response time constant
1997 FlaSh (wtGFP) Shaker potassium channel Xenopus oocytes GFP/505 nm 5.1% 23 ms 105 ms [1]
2002 FlaSh Shaker potassium channel Xenopus oocytes GFPuv, Ecliptic GFP, eGFP, YFP [2]
2002 SPARC Voltage-gated sodium channel Xenopus oocytes GFP/505 nm 0.5% ~1 ms ~1 ms [3]
2008 VSFP3.1 ci-VSP PC12 cells cerulean/475 nm 2.2% @ -43 mV 1.8 ms 105 ms Lundby et al., 2008, Perron et al., 2009
2009 VSFP3.x ci-VSP Cultured neurons Citrine/530 nm
mOrange/562 nm
TagRFP/584 nm
mKate2/633 nm
2.5–5% @ -43 mV 2–100 ms ~100 ms Perron et al., 2009
2011 Split-Venus FlaSh variants Shaker potassium channel HEK293 or NIE115 neuroblastoma cells Split-Venus/528 nm 1.4% ~15 ms ~200 ms [4]
2011 Arch(D95N) Archeorhodopsin
(microbial opsin)
Neurons Retinal/687 nm 60% 41 ms 41 ms [5]
2012 ArcLight A242 Ci-VSP HEK293 cells and cultured neurons super ecliptic pHluorin A227D 35% ~10 ms - Jin et al., 2012
2012 ElectricPk Ci-VSP HEK293 cells and cultured neurons circularly permutated eGFP 1.5% 2 ms 2 ms Barnett et al., 2012
  1. The sensitivity is given as change of fluorescence relative to baseline (ΔF/F) measured in cultured cells. The sensitivity of most voltage indicators is voltage-dependent and, if data are available, is given at the membrane voltage of half maximal response. If available, data are from experiments where different indicators were compared side-by-side.
  2. 2.0 2.1 Measured with short lasting and large amplitude depolarizing voltage steps (< 10 ms); for kinetics described best with two-exponential functions the two time constants are given in order of weight.

FRET-based GEVIs

Year Name Voltage-sensing domain Expression system for functional characterization Chromophore(s) peak emission wavelength(s) Sensitivity
(% ΔR/R per 100 mV)
@ xx mV[1]
Response time constant
Response time constant
2001 VSFP1 Kv2.1 HEK cells CFP/477 nm
YFP/529 nm
1.8% < 1 ms < 1 ms Sakai et al., 2001
2007 VSFP2.1 ci-VSP PC12 cells and neurons Cerulean/475 nm
Citrine/529 nm
6.8% @ -70 mV 15 ms 75 ms Dimitrov et al., 2007
2008 VSFP2.3 ci-VSP PC12 cells, cultured neurons, mouse in vivo Cerulean/480 nm
Citrine/530 nm
22% @ -50 mV 30 ms
~2 ms
~80 ms Mutoh et al., 2009
2008 VSFP2-Mermaid ci-VSP Xenopus Oocyte, NT cells, Primary cultures of brain cells mUKG /490 nm
mKOκ/560 nm
21% @ -43 mV 30 ms
~2 ms
~80 ms Mutoh et al., 2009 & Tsutsui et al., 2008
2009 VSFP2.4 ci-VSP PC12 cells citrine/530 nm
mKate2/633 nm
20.5% @ -54 mV 30 ms
~2 ms
~80 ms Mutoh et al., 2009
2010 VSFP2.42 ci-VSP Neurons in culture, brain slices and in vivo citrine/529 nm
mKate2/633 nm
20.5% @ -50 mV 80 ms
2 ms
~80 ms Akemann et al., 2010
2012 Zahra and Zahra 2 Voltage sensitive phosphatases of Nematostella and Danio HEK293 cells cerulean/480 nm
citrine/530 nm
0.2% ~5 ms ~5 ms Baker et al., 2012
2012 VSFP2-Buterfly 1.2 ci-VSP Neurons in culture, brain slices and in vivo citrine/530 nm
mKate2/633 nm
22.2% @ -79 mV 10 ms
2 ms
~80 ms Akemann et al., 2012
  1. The sensitivity is given as change of fluorescence ratio (ΔR/R) measured in cultured cells. The sensitivity of most voltage indicators is voltage-dependent and, if data are available, is given at the membrane voltage of half maximal response. If available, data are from experiments where different indicators were compared side-by-side.
  2. 2.0 2.1 Measured with short lasting and large amplitude depolarizing voltage steps (< 10 ms); for kinetics described best with two-exponential functions the two time constants are given in order of weight.

List of available constructs

Name Description Map Available from
Lab Company


Error fetching PMID 11751337:
Error fetching PMID 9354320:
Error fetching PMID 12496128:
Error fetching PMID 21497167:
Error fetching PMID 22120467:
  1. Error fetching PMID 9354320: [Siegel1997]
  2. Error fetching PMID 12496128: [Guerrero2002]
  3. Error fetching PMID 11751337: [Ataka2002]
  4. Error fetching PMID 21497167: [Jin2011]
  5. Error fetching PMID 22120467: [Kralj2011]
All Medline abstracts: PubMed HubMed