A method for imaging changes in membrane potential using genetically encoded voltage indicators is described.
Genetically encoded voltage indicators (GEVIs) have improved to the point where they are beginning to be useful for in vivo recordings. While the ultimate goal is to image neuronal activity in vivo, one must be able to image activity of a single cell to ensure successful in vivo preparations. This procedure will describe how to image membrane potential in a single cell to provide a foundation to eventually image in vivo. Here we describe methods for imaging GEVIs consisting of a voltage-sensing domain fused to either a single fluorescent protein (FP) or two fluorescent proteins capable of Förster resonance energy transfer (FRET) in vitro. Using an image splitter enables the projection of images created by two different wavelengths onto the same charge-coupled device (CCD) camera simultaneously. The image splitter positions a second filter cube in the light path. This second filter cube consists of a dichroic and two emission filters to separate the donor and acceptor fluorescent wavelengths depending on the FPs of the GEVI. This setup enables the simultaneous recording of both the acceptor and donor fluorescent partners while the membrane potential is manipulated via whole cell patch clamp configuration. When using a GEVI consisting of a single FP, the second filter cube can be removed allowing the mirrors in the image splitter to project a single image onto the CCD camera.
The major focus of this paper is to demonstrate the optical imaging of changes in membrane potentials in vitro using genetically encoded fluorescent proteins. Imaging changes in membrane potential offers the exciting possibility of studying the activity of neuronal circuits. When changes in membrane potential result in a fluorescence intensity change, each pixel of the camera becomes a surrogate electrode enabling nonintrusive measurements of neuronal activity. For over forty years, organic voltage-sensitive dyes have been useful for observing the changes in membrane potential 1-4. However, these dyes lack cellular specificity. In addition, some cell types are difficult to stain. Genetically encoded voltage indicators (GEVIs) overcome these limitations by having the cells to be studied specifically express the fluorescent voltage-sensitive probe.
There are three classes of GEVIs. The first class of GEVI uses the voltage-sensing domain from the voltage-sensing phosphatase with either a single fluorescent protein (FP) 5-9 or a Förster resonance energy transfer (FRET) pair 10-12. The second class of sensors uses microbial rhodopsin as a fluorescent indicator directly 13-15 or via electrochromic FRET 16,17. The third class utilizes two components, the genetic component being a membrane anchored FP and a second component being a membrane bound quenching dye 18-20. While the second and third classes are useful for in vitro and slice experiments19,20, only the first class of sensors are currently useful for in vivo analyses 6.
In this report we will demonstrate the imaging of membrane potential using the first class of GEVIs (Figure 1) in vitro. This first class of voltage sensors is the easiest to transition to in vivo imaging. Since GEVIs utilizing a voltage-sensing domain fused to a FP are about 50-fold brighter than the rhodopsin class of sensors, they can be imaged using arc lamp illumination rather than requiring an extremely powerful laser. Another consequence of the disparity in brightness is that the first class of GEVIs can easily exceed the auto-fluorescence of the brain. The rhodopsin-based probes cannot. The third class of sensor is just as bright as the first class but requires the addition of a chemical quencher which is difficult to administer in vivo.
We will, therefore, demonstrate the acquisition of a probe with a single FP (Bongwoori) 8 and a probe consisting of a FRET pair (Nabi 2) 12. The FRET constructs in this report are butterfly versions of VSFP-CR (voltage-sensitive fluorescent proteins – Clover-mRuby2) 11 consisting of a green fluorescent donor, Clover, and a red fluorescent acceptor, mRuby2, named Nabi 2.242 and Nabi 2.244 12. The introduction to these types of recordings should give researchers a better understanding of the type of information GEVIs can provide.
Figure 1. Two Types of Genetically Encoded Voltage Indicators (GEVIs) Imaged in This Report (A) A mono FP based GEVI having a trans-membrane voltage-sensing domain and a fluorescent protein. (B) A FRET based GEVI comprised of a trans-membrane voltage-sensing domain, a FRET donor and acceptor. Please click here to view a larger version of this figure.
Ethics statement: The animal experiment protocol was approved by the Institutional Animal Care and Use Committee at KIST animal protocol 2014-001.
1. Equipment Setup
Single FP based GEVI (Bongwoori) |
FRET pair based GEVI (Nabi 2.42 & Nabi 2.44) |
||
First filter cube placed in the microscope | excitation filter | 472 nm/30 | 475 nm/23 |
dichroic mirror | 495 nm | 495 nm | |
emission filter | 497 nm/long pass | – | |
Second filter cube placed in the beam splitter | dichroic mirror | – | 560 nm |
emission filter 1 | – | 520 nm/40 | |
emission filter 2 | – | 645 nm/75 |
Table 1. Two Different Filter Sets Used for a Single FP Based GEVI and a FRET Based GEVI Recordings
Figure 2. Equipment Setup for Voltage Imaging with GEVIs The workflow following the light path, (A) 75W Xenon arc lamp, (B) the excitation light from the arc lamp is filtered by the excitation filter and then reflected by the first dichroic mirror before it reaches to the specimen stage, an inset at the top right corner shows the whole cell configuration, (C) the slow speed CCD camera is used to aid both choice of a cell and patch clamp, (D) the image acquisition part; (1) the high speed CCD camera, (2) the image splitter for both FRET pair and mono FP GEVIs, (3) the demagnifier to fit the image onto the CCD chip in the high speed CCD camera, (4) the dual port camera adapter to switch the imaging pathway and (5) the slow speed CCD camera with high spatial resolution for identification of the cell to patch, (E&F) the image acquisition with a single-FP based GEVI (E) and a FRET-based GEVI (F). Please click here to view a larger version of this figure.
2. Expression of GEVIs
3. Voltage Imaging Protocol
4. Data Acquisition
5. Data Analysis
Transiently transfected cells can exhibit significant variation in fluorescence intensity and the degree of plasma membrane expression. Even on the same coverslip some cells will have varying levels of internal fluorescence. This is most likely due to the amount of transfection agent absorbed by the cell. Occasionally, too much expression causes the cell to experience the unfolded protein response resulting in apoptosis 27 (bright, rounded cells, with high internal fluorescence). The experimenter is therefore cautioned to test both bright cells and dim cells with as little internal fluorescence as possible since internal expression creates a non-responsive fluorescence that lowers the signal to noise ratio (SNR).
Another consideration is the maturation rate of chromophores. Figure 3 shows disparity in the fluorescence of the acceptor chromophore (mRuby2) which has a longer maturation time than the donor chromophore (Clover).
Figure 3. Confocal Images of HEK 293 Cells Transiently Transfected with a FRET Based GEVI, Nabi2.242. (A) Confocal images of an HEK cell showing membrane localized fluorescence from FRET donor and acceptor, (B) Confocal images showing a potential consequence of slow maturation of mRuby2. All four cells exhibit Clover fluorescence while only two exhibit mRuby2 fluorescence. Please click here to view a larger version of this figure.
The images were acquired by a confocal microscope. For the FRET donor, Clover, 488 nm laser was used for excitation and the emission was captured with 525/50 nm bandpass filter. The FRET acceptor, mRuby2, was illuminated with 561 nm laser for excitation and 595/50 nm bandpass filter was used for emission. The samples for confocal imaging were fixed with 4% paraformaldehyde/sucrose solution in phosphate buffered saline adjusted at pH 7.4 and then mounted with an anti-fade reagent.
Once the transfection conditions are optimized, the next source of variation comes from determining the region of interest to be analyzed. Using frame subtraction to identify the regions of the cell with the highest fluorescence change is often used to maximize the ΔF/F value. An alternative is to select all of the pixels receiving light from the cell. This increases the number of pixels resulting in the reduction of noise, but decreases the signal size since non-responsive internal fluorescence is included. Both methods are fine as long as the experimenter remains consistent.
Figure 4A shows the fluorescence change of an HEK cell expressing a single-FP based GEVI, Bongwoori. This is a typical fluorescence change in response to stepped voltage pulses. From this data one can plot voltage sensitivity of the GEVI and determine the on and off τ constants at different voltages by fitting to a single or double exponential decay function. 16 trials are typically averaged when imaging HEK cells to improve the SNR of small voltage steps and to detect any potential bleaching during the recording. Those probes that give a robust signal at 100 mV can be tested in dissociated hippocampal neurons (Figure 4C). The fluorescence trace from the hippocampal neuron is a single trial.
Figure 4. Voltage Imaging with a Single FP Based GEVI Expressed in HEK 293 Cells and Hippocampal Primary Neurons. (A) ΔF/F trace of a single FP based GEVI, Bongwoori, showing responses to stepped voltage pulses recorded at 1 kHz with a high speed CCD camera. (B) An HEK 293 cell imaged with the high speed CCD camera; (left) Resting Light Intensity (RLI) of a cell expressing Bongwoori & (right) the frame subtraction image indicating the pixels where fluorescence change was observed. (C) Optical recording of induced action potentials from mouse hippocampal primary neurons expressing Bongwoori. The action potentials were evoked under whole cell current clamp mode. The ΔF/F trace was selected from the pixels correlated to the soma. Please click here to view a larger version of this figure.
A representative fluorescence readout of the donor and acceptor chromophores is shown in Figure 5A. The polarity of the fluorescence change is opposite for the donor and acceptor enabling ratiometric analysis to reduce correlated noise sources, e.g., movement. For example, movement correlated noise will exhibit a change in fluorescence in the same direction for both donor and acceptor fluorescence. While the FRET based probe is capable of ratiometric imaging, it is often better to only analyze the brighter chromophore. This is because the dimmer chromophore can substantially increase the noise of the recording. Figure 5B shows this effect. The frame subtraction in Figure 5B from the brighter Clover clearly shows where the optical signal is in the cell. In contrast, the frame subtraction of mRuby2 fluorescence shows higher degrees of noise throughout the cell. Therefore, only the donor signal of Clover is shown in a hippocampal neuron in Figure 5C.
Figure 5. Voltage imaging with a FRET based GEVI expressed in HEK 293 cell and hippocampal primary neurons. (A) ΔF/F trace of a FRET based GEVI, Nabi2.242, showing responses at two wavelengths to stepped voltage pulses recorded at 1 kHz with a high speed CCD camera. (B) Top; the donor (Clover-RLI, Clover-frame subtraction) and acceptor (mRuby2-RLI, mRuby2-frame subtraction) images processed with two different thresholds for each FP, Bottom; the same threshold was used for both donor (Clover-RLI, Clover-frame subtraction) and acceptor (mRuby2-RLI, mRuby2-frame subtraction) images to illustrate the relative dimness of mRuby2. All the images were taken from the same HEK 293 cell expressing Nabi2.242. (C) The 520 nm wavelength trace of induced action potentials from a mouse hippocampal primary neuron expressing Nabi2.244 sensor. The ΔF/F trace was selected from the pixels correlated to the soma. Please click here to view a larger version of this figure.
Figure 6. Consequences of Varying Light Levels for ΔF and ΔF/F Values. (A) An image of an HEK 293 cell expressing a single FP based GEVI, Bongwoori, shown in Resting Light Intensity (RLI), (B) the fluorescence traces showing kernel averaged ΔF (Fx-F0) values from three different regions, region 1: membrane region with well localized fluorescence signal, region 2: a region with bright internal fluorescence, and region 3: region distant from optical signal, (C) the fluorescence traces showing kernel averaged ΔF/F values from the same regions in (B). Please click here to view a larger version of this figure.
Concepts | Equations | Remark |
Fractional fluorescence change (ΔF/F) | F1 = light intensity measured at a time point, F0 = light intensity measured at holding potential | |
Boltzmann function | y = ΔF/F, V = membrane potential in mV, V1/2 is the membrane potential in mV at half-maximal ΔF/F, A1 = the minimum value, A2 = the maximum value, dx = slope | |
Double exponential decay function | y = y0 + A1e -(t – t0)/ τ1 + A2e -(t – t0)/ τ2 | τ1, τ2 = time constants, A1,A2 = amplitudes, t, t0 = two time points, y0 = offset |
Table 2. Equations
The nervous system uses voltage in several different ways, inhibition causes a slight hyperpolarization, synaptic input causes a slight depolarization and an action potential results in a relatively large voltage change. The ability to measure changes in membrane potential by GEVIs offers the promising potential of analyzing several components of neuronal circuits simultaneously. In this report we demonstrate a fundamental method for imaging changes in the membrane potential using GEVIs.
A major key for imaging changes in voltage is the efficient expression of the GEVI in the plasma membrane. Intracellular expression creates a non-responsive fluorescence that reduces the SNR of the probe. Optimizing the transfection conditions vastly improves the consistency of the optical measurements. Indeed, when testing a novel GEVI it is advisable to also test a known probe to ensure that differences in optical activity are due entirely to the new probe and not the conditions of the cells.
Figure 6 shows the effect of internal fluorescence in decreasing the sensitivity of the voltage imaging, ΔF/F value. Figure 6B shows ΔF values an HEK cell expressing the single-FP based GEVI, Bongwoori. Trace 2 comes from region 2 (blue color) where relatively bright internal fluorescence is seen in Figure 6A. This trace has similar level of ΔF value as trace 1. However, in Figure 6C where the traces in Figure 6B were divided by RLI values for ΔF/F values, the trace 2 drops significantly due to the bright internal fluorescence which decreases ΔF/F values. Fluorescence trace 3 has a very small ΔF but also has a very small RLI value (F). The result is a misleading ΔF/F signal that has a substantial increase in the noise of the recording. This example was included to demonstrate the ΔF is also important. A large change in ΔF/F is not helpful if F is very low to begin with.
The use of an image splitter enables the concomitant measurement of two wavelengths onto a single CCD camera. This greatly assists the measurement of FRET dependent fluorescent changes for probe development and reduces the cost of the setup. However, due to chromatic aberration, these two images will be slightly out of focus necessitating the need for an apochromatic objective. The more light the better SNR, so choosing objectives with the highest numerical aperture also improves fluorescence imaging. In addition, one can use stronger light sources such as light emitting diodes (LED) or laser to increase SNR. LEDs do not require a mechanical shutter.
In this report we show the optical signals from a single FP GEVI and a FRET-based GEVI. The main advantage of a FRET-based GEVI is that ratiometric imaging enables the reduction of correlated noise due to respiration and blood flow in vivo. The opposite polarity of the fluorescent signals confirms a real change in membrane potential. However, since the fluorescence intensities of the two chromophores often vary significantly, the SNR will also vary. This confounds the ratiometric analysis and limits its usefulness 28. There can also be a FRET-independent signal29. It is therefore sometimes better to use only the brighter fluorescent chromophore when using FRET to analyze changes in membrane potential.
Voltage imaging is more challenging than calcium imaging due to the speed and variability of the changes in membrane potential compared to calcium imaging which measures the calcium flux. The membrane potential changes in very fast time scales compared to the calcium ion flux, and subthreshold events in neurons cannot be measured with calcium imaging 30. Furthermore, the voltage probe must be in the plasma membrane to be effective which reduces the amount of probe available to optically report changes in the membrane potential. Consequently, the number of photons that can actually arrive at the CCD chip is relatively few. This requires the need for a high speed and low read-noise camera with high quantum efficiency and a probe that elicits a high SNR upon changes in membrane potential. By imaging cells in vitro, researchers will better understand the potential benefits and pitfalls of GEVIs before attempting in vivo measurements.
The authors have nothing to disclose.
This work was supported by the World Class Institute (WCI) program of the National Research Foundation of Korea funded by Ministry of Education, Science, and Technology of Korea Grant WCI 2009-003 and Korea Institute of Science and Technology Institutional Program Project 2E24210. Sungmoo Lee was supported by Global Ph.D. Fellowship program (NRF-2013H1A2A1033344) of the National Research Foundation (NRF) under the Ministry of Education (MOE, Korea).
Inverted Microscope | Olympus | IX71 | |
60X objective lens (numerical aperture = 1.35) | Olympus | UPLSAPO 60XO | |
Excitation filter | Semrock | FF02-472/30 | For voltage imaging of super ecliptic pHluorin in Bongwoori |
Dichroic mirror | Semrock | FF495-Di03-25×36 | |
Emission filter | Semrock | FF01-497/LP | |
75W Xenon arc lamp | CAIRN | OptoSource Illuminator | LEDs and lasers are also effective light sources |
Slow speed CCD camera | Hitachi | KP-D20BU | |
Dual port camera adaptor | Olympus | U-DPCAD | |
High speed CCD camera | RedShirtImaging, LLC | NeuroCCD-SM | |
Image splitter | CAIRN | Optosplit 2 | |
Excitation filter | Semrock | FF01-475/23-25 | For voltage imaging of FRET pair based GEVI consisting of Clover and mRuby2) |
Dichroic mirror | Semrock | FF495-Di03-25×36 | |
Emission filter | Chroma | ET520/40 | |
Dichroic mirror | Semrock | FF560-FDi01-25X36 | |
Emission filter | Chroma | ET645/75 | |
Vibration isolation system | Kinetic systems | 250BM-IC, 5702E-3036-31 | |
Patching chamber | Warner instruments | RC-26G, 64-0235 | |
#0 Micro Coverglass (22x40mm) | Electron Microscopy Sciences | 72198-20 | |
Temperature controller | Warner instruments | TC-344B | |
#0 (0.08~0.13mm) – 10mm diameter glass coverslip | Ted Pella | 260366 | |
Lipofection agent | Life Technologies | 11668-027 | |
Calcium phosphate reagent | Clontech – Takara | 631312 | |
Patch clamp amplifier | HEKA | EPC 10 USB amplifier | |
Multi-channel data acquisition software | HEKA | Patchmaster | |
Image acquisition and analysis software | RedShirtImaging | Neuroplex | |
Spreadsheet application software | Microsoft | Microsoft Excel 2010 | |
Data analysis software | OriginLab | OriginPro 8.6.0 | |
Demagnifier | Qioptiq LINOS | Optem standard camera coupler 0.38x SC38 J clamp | |
Confocal microscope | Nikon | Nikon A1R confocal microscope | |
Anti-fade reagent | Life Technologies | P36930 |