Early versions of genetically-encoded voltage probes constructed with voltage-gated ion channels 
failed to traffic adequately to the plasma membranes of mammalian cells 
. CiVSP-based probes traffic to the cell membrane more completely and probes created with the CiVSP voltage sensing domain have been more successful in mammalian cells 
. The fluorescence changes, however, have been generally slow. The rapid response properties of the probes described here (2 ms) are in striking contrast to the dominant response kinetics (>12 ms) of all other currently published CiVSP-based probes 
. The only probes with similar characteristics are the sodium channel-based probe (SPARC; 2) and a probe based on the zebrafish voltage sensitive protein 
. The SPARC probe was able to capture rapid movements of the VSPs but is less useful due to low signal size and poor membrane targeting 
. The mechanism by which ElectricPk changes its fluorescence intensity is not clear, but it depends on the circular permutation of eGFP since the fusion of normal eGFP at the same site does not produce viable probes (data not shown).
There appears to be several voltage driven rearrangements in the CiVSP protein. One occurs quite quickly and can be measured as gating charge movement. Others occur more slowly 
, and it is these slow movements that likely drive the changes in fluorescence in many fluorescent voltage probes. The changes in ElectricPk, as well as in the majority of constructs described here, captures only the fast rearrangements in the CiVSP protein as changes in fluorescence. ElectricPk demonstrates that it is possible to capture a voltage dependent movement in CiVSP that previous sensor designs have missed, and to do this in neurons. It appears that the cpEGFP is capturing a fast, voltage-driven movement of CiVSP that both FRET pairs of FPs and single FPs have missed. In addition, the linear F/V response is quite different from all previously published CiVSP-based probes 
. This indicates that this probe is likely capturing a different protein rearrangement, one associated with gating charge movements of the S4, which may help to explain its speed.
Previously the Knöpfel laboratory used cpEGFP derived from GCaMP2 to create a few fusions analogous to our constructs 
. Their VSFP (B) cpEGFP is similar to our pLB1.1 or pLB2.1, and VSFP (C) cpEGFP as well as VSFP (D) cpEGFP are quite like pLB4.1 and pLB7.1, respectively. All of these constructs exhibited very low fluorescence in the initial screen and were not tested for voltage driven changes in fluorescence. Their results, and ours, demonstrate that small changes in the size of the hole in the fluorescent protein and linkage to CiVSP, have profound effects on a probe's response speed and size.
As hyperpolarizing steps cause a linear increase in fluorescence in ElectricPk, some of the constructs that showed no fluorescence during the initial fluorescence screen may have increased fluorescence upon changes in membrane potential. Therefore, we may have missed some probes which exhibited increases in fluorescence in response to voltage steps, as only visibly fluorescent probes were tested for voltage dependent changes in fluorescence using patch clamp. To date, most CiVSP-based voltage sensors have been too slow for following fast neuronal activity. Recently a probe has been developed that makes it possible to detect action potentials, but it too is to slow to clearly resolve the waveform 
. Tested in cultured neurons, ElectricPk detects action potentials with remarkable time resolution, comparable to what is seen with synthetic, small molecule voltage dyes. This is the first FP-based voltage sensor that could be, with additional adjustments of signal size, used for studies of the origin and propagation action potentials within single cells.
The weak fluorescence of a bacterial rhodopsin has recently been shown to be modulated by transmembrane potential 
. This approach to genetically-encoded voltage probe development has several potential strengths: large, linear fractional change in fluorescence with voltage; extremely red-shifted excitation and emission spectra and resistance to bleaching. However, the quantum efficiency of fluorescence in these rhodopsins is so low (QY of 0.001 vs 0.65 for GFP) that it is unlikely they will be of much use in optical recordings in vivo. While in vitro conditions are artificially optimized for detecting fluorescence, the low fluorescence of these probes require high intensity (200 mW optical power) laser illumination combined with an EMCCD camera for detection in vitro. These limitations make in vivo use highly improbable with lower levels of expression, more light scattering, and greater sensitivity to phototoxicity. In addition, attempts to produce non-conducting probes have 
, to date, significantly slowed the response kinetics of the probe (<1 ms for wild type vs 45 ms for the non-conducting Arch variant, ArchD95N).
ElectricPk demonstrates that the design principles that have evolved for the creation of better calcium probes can be applied to voltage sensors as well. The present study reconfirms 
that identifying a viable probe requires a systematic analysis of numerous different fusion location and FP types. While FRET-based sensors offer the theoretical advantage of ratiometric recordings, and should be exquisitely sensitive to small changes in the distance/orientation of the two fluorophores, recent history has shown that sensors designed around a single, circularly permuted FP can produce more robust signals. In this case, ElectricPk captures only the fast movement in the voltage sensing domain of CiVSP that has eluded other sensor designs. While remarkably fast, the signal generated by ElectricPk is relatively small. It is likely that continued rounds of evolution will improve the signal size, much as it has done for calcium sensors.