Journal of Neurophysiology

Electroporation Loading of Calcium-Sensitive Dyes Into the CNS

Agnès Bonnot, George Z. Mentis, Jesse Skoch, Michael J. O'Donovan


Calcium imaging of neural network function has been limited by the extent of tissue labeled or the time taken for labeling. We now describe the use of electroporation—an established technique for transfecting cells with genes—to load neurons with calcium-sensitive dyes in the isolated spinal cord of the neonatal mouse in vitro. The dyes were injected subdurally, intravascularly, or into the central canal. This technique results in rapid and extensive labeling of neurons and their processes at all depths of the spinal cord, over a rostrocaudal extent determined by the position and size of the electrodes. Our results suggest that vascular distribution of the dye is involved in all three types of injections. Electroporation disrupts local reflex and network function only transiently (∼1 h), after which time they recover. We describe applications of the method to image activity of neuronal populations and individual neurons during antidromic, reflex, and locomotor-like behaviors. We show that these different motor behaviors are characterized by distinct patterns of activation among the labeled populations of cells.


During the last decade, calcium imaging has become a powerful tool for analyzing the spatiotemporal patterns of activity generated by neural networks (Bonnot et al. 2002a; Denk et al. 1994; Miyawaki 2003). Despite progress in this field, the methods used for loading neurons with calcium-sensitive dyes have two significant limitations: the restricted amount of tissue that can be labeled and/or the time taken for loading.

Many applications would benefit from the ability to label large regions of the nervous system. For example, in the spinal cord, retrograde loading has been used to successfully label two to three segments for visualizing the spatiotemporal organization of motoneuron recruitment during locomotor-like activity (Bonnot et al. 2002b). However, the network producing this activity spans many more segments, including those from the sacrocaudal and thoracic region, and consists of both motoneuron and interneuron populations. Voltage-sensitive dyes have been used to monitor the population activity over several centimeters of cortex (for review see Wu et al. 1998), but the signals are small and subject to photobleaching, and the method does not provide cellular resolution in the vertebrate brain.

Membrane-permeant dyes can be loaded quickly (1–2 h) by direct application to exposed neurons (O'Donovan et al. 1994; Yuste and Katz 1991) or by injection into neural tissue (Stosiek et al. 2003). Tissue injection has been used to label the spinal cord of the larval zebrafish (Brustein et al. 2003), but the method relies on the ability of the dye to diffuse throughout the limited volume of the larval cord. In the mammalian cortex, unless several injections are made, labeling is restricted to a sphere ∼300 μm in diameter following a single bolus injection (Stosiek et al. 2003). Recently, membrane-impermeant dyes have been applied to tissues with a “gene gun” (Kettunen et al. 2002), allowing a rapid labeling of a few cells, but the targeted tissue remains superficial, and both the density and extent of the labeling are limited.

Another approach has been to apply membrane-impermeant (O'Donovan et al. 1993) or membrane-permeant dyes (Koshiya and Smith 1999) into an available tract of axons to allow retrograde transport to cells bodies. The most spatially extensive labeling (2–3 spinal cord segments) has been accomplished by retrogradely loading spinal motoneurons (Bonnot et al. 2002b) or spinal interneurons projecting into the white matter (A. Bonnot, unpublished data). However, the time required for transport to cell bodies in vitro (6–14 h) compromises the viability of the neural tissue. Furthermore, the target neurons must have a tract of axons available, and labeling is restricted to this specific population.

We now describe a new method—electroporation—that overcomes these limitations. In this technique, an electric field is imposed across a population of cells to transiently introduce membrane pores through which large molecules can enter. Electroporation has frequently been used to introduce genes, antibodies, dyes, and other molecules into cultured cells or cell suspensions (Neumann et al. 1982; Weaver 1993). In addition, it has been used in vivo to visualize cell migration and gene function in the developing nervous system (for reviews, see Ogura 2002; Trezise et al. 2003). While most studies have used electroporation for gene delivery, a few have used the technique to introduce fluorescent dyes into cultured neurons (Yang et al. 2004). However, with the exception of two studies in which calcium-sensitive dyes were electroporated into individual cultured neurons or neuronal precursors (Lundqvist et al. 1998; Teruel et al. 1999), no reports have described the use of electroporation to load these dyes into large populations of neurons either in vitro or in vivo. In particular, it was unclear if the technique would work for in vitro physiological studies in which neural function must be studied shortly after electroporation. In addition, the effectiveness of electroporation for loading neurons into a relatively intact tissue such as the spinal cord where the dyes do not have a direct access to exposed cells was unknown.

In this study, we show the feasibility of using electroporation to introduce membrane impermeant calcium indicators into neuronal populations of the isolated mouse spinal cord in vitro. We show that the technique enables rapid loading of calcium dyes into a large number of neurons over several millimeters, throughout the thickness of the tissue, and at levels sufficient to signal neuronal activity. Furthermore, we show for the first time that calcium dyes can be electroporated into neurons after intravascular injection. A preliminary report of this work has appeared as an abstract (Bonnot et al. 2003).


Surgical procedures

Experiments were performed on Swiss Webster neonatal mice (P0–P4; Taconic Laboratory). The mice were anesthetized with methoxyflurane, transected at T5–T7, and eviscerated. The remaining tissue was placed in a dissecting chamber and continuously perfused with a normal (in mM: 128 NaCl, 4 KCl, 1.5 CaCl2, 1 MgSO4, 0.5 NaH2PO4, 21 NaHCO3, and 30 d-glucose) or low Ca2+/high Mg2+ artificial cerebrospinal fluid (ACSF; in mM: 0.1 CaCl2 and 5 MgSO4), bubbled with 95% O2-5% CO2, at a flow rate of 35 ml/min. Initially the temperature of the ACSF was ∼17°C, and it was gradually allowed to rise to room temperature (23–25°C) during the dissection. A ventral laminectomy exposed the cord, which was removed from the vertebral column together with the attached roots and the cauda equina.

Dye application and electroporation

The isolated spinal cord was pinned down with the ventral side up, and a broken sharp glass electrode was used to pressure-apply 1–2 μl of calcium-sensitive dye solution under the dura matter in the caudal lumbar area (∼L5/L6) at the level of the ventral commissure (Fig. 1B). This location was chosen because the space between the dura mater and the ventral commissure allowed the injection electrode to be inserted through the dura mater without penetrating into the spinal cord, thereby allowing the dye to enter the subdural space. Proper introduction of the dye into the ventral subdural space was assessed postexperimentally on fixed sections by the absence of tissue lesions and/or labeled cells immediately surrounding an injection track. Some preparations were injected in the anterior spinal artery running along the midline of the ventral surface or in the central canal at the rostral lumbar level (∼L2/L3; Fig. 1A). In the latter cases, the dye spread up and down the spinal cord (within the spinal artery or the central canal) from the injection site, as opposed to subdural injections where the dye remained localized to the injection site (over a few segments). A cell-impermeant dye—either fluorescein dextran 3,000 MW (40 mM), calcium green-1 dextran 3,000 MW (30–40 mM), or calcium green-1 hexapotassium salt (1,147 MW, 55 mM; Molecular Probes, Eugene, OR)—was dissolved in the ACSF before application. Two gold-plated electrodes (BTX, 3 or 1 mm in length and 0.5 mm in diameter genetrodes or rectangular 3 × 5 mm gene paddles) were immediately submerged in the ACSF and positioned in parallel on either side of the lumbosacral cord with a 3-mm interelectrode distance (i.e., electrodes were <1 mm away from the cord edges; Fig. 1A). Therefore, while the electrodes were at the level of the injection site in the case of the subdural injection, they were remote from it in the case of the ventral artery or the central canal injections (site of injection in the rostral lumbar area). Because the dye spread over many segments following central canal or arterial injections, the electroporation electrodes could be placed away from the injection site. Under these conditions, this allowed us to unambiguously attribute the electroporation loading to dye originating from within the central canal or the vasculature.

FIG. 1.

Electroporation-induced loading of calcium dyes into spinal neurons. A: schematic showing the transverse plane of the spinal cord together with different sites for dye injection. B: schematic showing the injection site for subdural dye injections and location of electroporation electrodes. C: high power view of a cell located dorsally in L1 following electroporation of calcium green salt (T12–L3). D: labeling pattern on a transverse section at L6 following electroporation of the calcium green salt (L4–S2). E: transverse section at L5 following dye injection in the ventral subdural space but no electroporation. Note absence of cellular labeling and presence of fluorescent blood vessels, particularly the central branch of the anterior spinal artery (arrowhead). F: midline sagittal section showing the labeling following electroporation of the calcium green salt (L5–S2). Vertical dotted lines indicate the approximate border between spinal segments. In all figures, dorsal is up and ventral is down.

Square voltage pulses (5–40 V applied voltage, 5–36 V effective voltage, 50–100 ms duration, 4–10 pulses at 1 Hz) were applied between the electrodes within 5–10 min after dye application using the ECM 830 electroporation system (BTX). Typical parameters of a successful electroporation were 18–25 V effective, seven pulses of 50 ms or five pulses of 75 ms. We minimized the difference between the applied and effective voltages (∼1–5 V) by keeping the ACSF level in the chamber as low as possible during the electroporation. Voltages indicated in the text are effective voltages. The dura mater was usually peeled off from the lateral sides in the electroporated area to allow better visualization of the labeled cells. It was important not to remove the pia mater extensively, even though it improved visualization, because doing so compromised coccygeal-evoked locomotor like activity. After electroporation, the cord was left for ∼1 h to allow both wash-out of the excess of dye from the extracellular space and physiological recovery from electroporation. The cord was transferred to a recording chamber containing normal ACSF at 25°C and continuously oxygenated with 95% O2-5% CO2 at a flow rate between 8 and 10 ml/min.

We took several precautions to minimize the electroporation-induced depression of neural activity: 1) voltage, duration, and number of electroporation pulses were set as low as possible while still allowing enough penetration of the dye into the cells, 2) the temperature was generally kept below 20°C—a temperature at which motor networks remain mostly silent—during electroporation, and 3) we used a low Ca/high Mg solution in most (19/27) of our injected, electroporated, and recorded preparations to block electroporation-induced transmitter release.


Tight-fitting plastic suction electrodes were placed on individual ventral and dorsal roots to record either motoneuron electrical activity or to stimulate afferent or efferent axon bundles. They were placed within the electroporated area (L4–L6) as well as more rostrally (L1–L3) outside the electroporated area.

The recorded signals were filtered (DC to 3 kHz), digitized (NeuroData), and recorded on videotape. Episodes of data were acquired off-line using Clampex v.9 and analyzed with Clampfit (Axon Instruments) and Excel software. Spinal motoneurons were activated antidromically by square wave stimulation of the ventral roots (stimulus duration: 500 μs, trains 5–50 Hz, stimulus intensity: 50–80 μA, train duration: 2 s). To assess the recovery of function after electroporation, we recorded monosynaptic and polysynaptic reflex ventral root potentials evoked by a single electrical stimulus applied to a lumbar dorsal root (stimulus duration: 200 or 500 μs, stimulus intensity: 18–99 μA). Several electroporation protocols were sometimes performed on the same preparation. The monosynaptic response was measured from the peak-to-peak amplitude of the initial reflex spike, and the polysynaptic response was measured as the peak amplitude of the electronic potential within the first 50 ms after the stimulus. We also compared the potentials evoked and recorded within the electroporated area (L6) with those evoked and recorded outside the electroporated area (L1). Locomotor-like activity was evoked by a train of stimuli (4 Hz, stimulus duration: 200 or 500 μs, stimulus intensity: 20–50 μA, train duration: 10 s) applied to a coccygeal ganglion.

Intracellular recordings

Isolated spinal cords from P2–P3 neonatal mice were pinned down with the ventral side up, and the pia mater was removed before obtaining whole cell recordings from motoneurons under current clamp. Intracellular electrodes were pulled from borosilicate glass capillaries (1B100F-4, WPI) to a resistance ranging between 4–8 MΩ. The intracellular solution contained (in mM; all chemicals obtained from Sigma) 10 NaCl, 130 K-gluconate, 10 HEPES, 11 EGTA, 1 MgCl2, 0.1 CaCl2, and 1 Na2ATP; pH was adjusted with KOH to ∼7.2. The osmolarity was adjusted to that of the extracellular solution (∼299 mOsm). Intracellular recordings were performed using Multiclamp 700A (Axon Instruments). The recorded signals were filtered (DC to 10 kHz) and digitized using an analog to digital interface (Digidata 1320A, Axon Instruments) and fed to a PC computer for storage and off-line analysis using Clampex (Axon Instruments). Motoneurons were identified blindly by 1) their short latency antidromic action potential (mean latency: 2.1 ± 0.9 ms) and 2) all or none responses to a single threshold stimulus applied to the segmental ventral root (stimulus duration: 200-μs duration, stimulus intensity: 20–50 μA). The mean resting potential of the recorded neurons was −62.2 ± 7.2 mV, and the average amplitude of the action potential was 65.3 ± 19.2 mV. Values here and in the remainder of the text are given as mean ± SD.

Widefield epifluorescent optical imaging

The cord was placed in a perfusion chamber with a coverslip as its base. The chamber was mounted on an inverted epifluorescence microscope (Nikon Eclipse TE300) and continuously perfused with ACSF. The tissue was illuminated with a 75 W Xenon light source (excitation filter: 470–490 nm, emission filter: 520–560 nm) and viewed through the lateral white matter (see Fig. 6A). Suction electrodes were placed as indicated above, and activity-dependent changes in fluorescence were detected using an intensified CCD camera (Stanford Photonics) and stored on videotape (SONY SVO 9500MD). Changes in fluorescence were monitored on-line using a PC-based frame grabber (Matrox Image LC/Flashbus) and image processing software (Metamorph, Universal Imaging). Synchronization of the electrical and optical signals was achieved by gating the camera off and recording the gating pulse together with the electrical recordings. Stacks of images were collected at 30 frames/s during periods of antidromic, reflex, or locomotor-like activity. Difference images generated this way were linearly expanded to occupy the full 256 levels of the frame store. The fractional change in fluorescence (ΔF/F with respect to baseline fluorescence in the absence of evoked or spontaneous activity) was measured in a grid superimposed on the visualized field (∼100 × 100 μm). At 10× magnification, the grid divided the cord in six columns dorsoventrally, and the average activation of the ventral third, the intermediate part, and the dorsal third were compared during different stimulation protocols (ventral root train, dorsal root single, coccygeal train). One-way ANOVA statistical tests were used to establish the probability that variations of fluorescence induced by a given stimulation were significantly different in ventral, intermediary, and dorsal parts of the cord.

Two-photon optical imaging

To visualize neuronal calcium signals from cells embedded deep in the intact spinal cord, we used a confocal microscope (LSM 510, Carl Zeiss) equipped with a two-photon laser (Coherent). The confocal system consisted of an upright microscope (Axioscope 2 FS MOT, Carl Zeiss), allowing the emitted light to be directed to either a digital CCD camera (CoolSNAPHQ, Roper Scientific), or a photo-multiplier tube (PMT) in the scan head of the microscope (internal descanned detector). When using the CCD camera, excitation was produced by a 103 W mercury bulb through an excitation filter (480–500 nm).

In the confocal mode, the tissue was excited with a two-photon laser mode-locked at 780 nm. The emission filter in both cases passed wavelengths between 500 and 550 nm. In this way, we were able to compare the activity-dependent changes of fluorescence (ΔF/F) obtained by epifluorescence and two-photon imaging.

To synchronize the optical and the electrical signals recorded from the ventral roots, a TTL pulse from the Frame Synchronization (Frame Sync) signal produced at the E-Box of the confocal was acquired simultaneously with the electrical signals and digitized at 10 kHz (Digidata 1320, Axon Instruments). Each TTL pulse from the Frame Sync signal indicated the beginning of the scanning for each frame and was acquired on clampex v.9 along with ventral root signals. Frame acquisition was performed at 12-bit resolution with scan times from 37.7 to 294.9 ms. To limit photobleaching, a low magnification 10× water-immersion objective was used (Achroplan with 0.3 NA, Carl Zeiss). The strength of the two-photon laser varied between 6.5 and 15% depending on the depth of the neurons. Analysis of the electrical and optical signals was performed by off-line analysis using MATLAB.

Retrograde labeling

In some experiments, on termination of the recordings, the ventral roots belonging to the electroporated area were retrogradely loaded with Texas red 10,000 MW (Molecular Probes) using standard techniques (O'Donovan et al. 1993). For this purpose, several adjacent lumbar ventral roots were drawn into a suction electrode containing ∼6 μl solution of the dye (10 mM) dissolved in distilled H2O with 0.2% Triton X-100. The cord was left at 17°C for 7–13 h to allow labeling of motoneuron cell bodies and dendrites.

Confocal microscopy on fixed sections

At the end of the experiment, most spinal cords electroporated with a calcium-sensitive dye were fixed in 1-ethyl-3-(dimethylaminopropyl) carbodiimide hydrochloride (EDC; Sigma-Aldrich; 40 mg/ml) as described previously (Bonnot et al. 2002a). Cords injected with fluorescein dextran were fixed in 4% paraformaldehyde (PAF). The preparations we recorded from (n = 25) were usually fixed many hours after electroporation (5–20 h), causing the labeling to be less intense because of both photobleaching and leakage of the dye out of the cells (Takahashi et al. 1999). Therefore some preparations (25) were dedicated to neuroanatomical observation (no electrical or optical recordings) and fixed within 10 min to 2 h following application of the dye, allowing optimal preservation of the labeling. The cords were cut sagittally or transversely (70 μm) using a Vibratome. Sections were viewed on an Olympus FV300 confocal microscope (Argon laser 488 nm) using FluoView software to establish the labeling pattern in control or electroporated cords. In some initial experiments, the labeling on fixed sections was checked using a spinning disk confocal system (CARV, Atto Bioscience) coupled to the inverted microscope, and images were taken with a digital CCD camera (Roper Scientific). When motoneurons were labeled with Texas red, z-stacks were acquired at two wavelengths (argon laser, 488 nm; krypton laser, 568 nm) using the sequential mode to avoid bleed-through of the calcium green labeling into the red channel. In these double-labeling experiments, counts were made from stacks of images acquired at 40× (20–50 slices at 1 μm intervals). Images presented in this study are two-dimensional projection images from the z-stacks.


The experiments reported in this study were performed on 70 isolated spinal cords of the neonatal mouse (P0–P4). In the first section of this paper, we present data from 49 experiments in which the cords were fixed, sectioned (transverse or sagittally), and examined with confocal microscopy. In 41 of these experiments, electroporation was performed after dye injection in one of several locations: the subdural space at the level of the ventral commissure (lumbar levels L4–L6, n = 33), the anterior spinal artery on the ventral surface of the cord (L3, n = 4), or the central canal (L3, n = 4; Fig. 1A). The electroporation electrodes were typically positioned within the L4–S2 region on either side of the cord (Fig. 1B), except for one experiment in which electroporation was performed ventro-dorsally. We used three different dyes: the calcium-sensitive dyes calcium-green dextran (n = 15), the calcium green salt (n = 25), or fluorescein dextran (n = 8). Following electroporation, their labeling patterns in fixed sections were similar, although use of the salt appeared to produce the most homogenous labeling throughout the ventro-dorsal extent of the cord, whereas labeling obtained with the dextran tended to be clustered into particular areas of the cord. In seven control experiments, the dyes were applied as described above, but the cords were not electroporated. Under these conditions, no labeled cells were detected in fixed sections, and fluorescence could only be seen at the surface of the cord, around the central canal, and within some of the cord vasculature (Figs. 1E, 3, A and B, and 4A). In the absence of dye application (n = 1), tissue autofluorescence was extremely low and did not significantly contaminate the fluorescent labeling due to the dye.

Labeling pattern following subdural dye injection and electroporation

Few cells were labeled if the electroporation voltage was <15 V (n = 2). In contrast, reliable cellular labeling was observed when the electroporation voltage was between 18 and 27 V, with the most extensive labeling achieved at the highest voltages (n = 31). We also determined that the use of paddle electrodes (3 × 5 mm) that covered the ventro-dorsal extent of the spinal cord gave better results than thinner wire electrodes (0.5 mm diam). The labeling was usually strongest at the level of the dye application and got weaker rostrally and caudally. However, the rostrocaudal extent of the labeling always covered the entire length of the electrodes (1, 3, or 5 mm, 1–6 segments; Fig. 1F). Outside the electroporated area, only background fluorescence and fluorescent blood vessels could be seen, providing further evidence that the electric pulses are required to introduce the dye into the cells. Although there was some variation in the labeling pattern across experiments, typically we found that cells were labeled at all tissue depths from dorsal to ventral throughout the electroporated region (Fig. 1D). Labeling was found on both sides of the cord, although the side of the cord adjacent to the positive electrode was sometimes the most densely labeled (seen in 13 experiments). Cells around the central canal were always numerous and densely labeled and in most experiments (27/31), labeled cells were also abundant in the dorsal horn. The frequency of labeled cells was most variable in the ventral region of the cord. In some experiments (19/31), only a few ventral cells, recognizable as motoneurons, were labeled. In contrast, in the other experiments (12/31), a large proportion of the motoneurons were labeled, forming well-defined motoneuron clusters (similar to Fig. 4B). Images taken at high power revealed that labeled cells had a typical neuronal morphology and that soma and processes were both labeled (Fig. 1C). Both the nucleus and the cytoplasm of electroporated cells were labeled uniformly, although the cytoplasmic labeling was often punctate. Visualization at high power also revealed the presence of many less densely labeled cells (Fig. 2, A and C).

FIG. 2.

Confocal photomicrographs of electroporated and retrogradely labeled motoneurons. A: transverse section at L5 showing labeling of the ventral horn following electroporation (L3–S3) of the calcium green salt. B: same section as in A showing motoneurons retrogradely filled by application of Texas red dextran to the L5 ventral root. C: overlay of the green and red channels acquired sequentially. Arrowheads indicate some double-labeled motoneurons. Inset: whole transverse section, and dotted box corresponds to magnified area in C.

To show that some of the electroporated cells were motoneurons, we retrogradely labeled motoneurons with Texas red applied to the ventral roots within the electroporated area (n = 5). In these experiments, we first confirmed the presence of labeled motoneurons by optical imaging of antidromic signals and applied the Texas red to the appropriate roots. In all five experiments, after the tissue was fixed and sectioned, cells were found in which the electroporated calcium green and the retrogradely loaded Texas red were co-localized (Fig. 2, arrowheads), confirming that some of the electroporated cells were motoneurons. We found that ∼70% of the motoneurons were labeled following electroporation (one-half strongly labeled, one-half weakly labeled). Some of the electroporated cells in the motoneuron region consisted of smaller cells that were not double-labeled (see Fig. 2).

Labeling pattern following intravascular dye injection and electroporation

In control experiments in which the dye was injected subdurally but not subsequently electroporated, we noticed that blood vessels often contained fluorescent dye (Fig. 1E). This was not autofluorescence, because in the absence of dye, no vascular fluorescence was seen. This observation raised the possibility that the dyes might be distributed within the cord, in part, through the vasculature.

To test this idea, we injected the dyes into the anterior spinal artery at the level of the rostral lumbar segments (∼L3; n = 6). In four experiments, this was followed by electroporation (remote from the injection site, see methods), and in two, no electroporation was performed. In the non-electroporated preparations, blood vessels were extensively labeled both centrally (Fig. 3A) and laterally (Fig. 3B), but no fluorescent cells were detected. When this type of injection was followed by electroporation (24–33 V), in addition to the fluorescent vasculature, many cells were found labeled in the central part of the spinal cord where the vasculature is particularly dense (Fig. 3, C and E). Lateral cellular labeling was also apparent in three of four cords (Figs. 3D and 4, C and F).

FIG. 3.

Confocal photomicrographs of sagittal tissue sections following dye injection into the anterior spinal artery. A: midline section in the L3–L4 area following fluorescein dextran injection into the anterior spinal artery of a cord that was not electroporated. CC, central canal; ASA, anterior spinal artery. B: lateral section in the L3–L4 area (250 μm from lateral surface) from the same preparation as in A. C: labeling observed in an L5 midline section following fluorescein dextran injection into the anterior spinal artery followed by electroporation (L3–S2). D: lateral section in the L6 area (250 μm from lateral surface) from a cord injected with fluorescein dextran and electroporated (L5–S2). E: high power view of a midline sagittal section showing cells located around the CC from the same preparation as in D. In all images, dorsal is up and ventral is down.

FIG. 4.

Confocal photomicrographs of tissue sections following central canal or anterior spinal artery injection. A: transverse section at L5/L6 following injection of the calcium green salt into the central canal but no electroporation. Note the labeling of the central branch of the anterior spinal artery (arrowhead). B: transverse section at L5 following injection of the calcium green salt into the central canal and electroporation (L5–S1). C: transverse section at L6/S1 following injection of the calcium green salt into the anterior spinal artery and electroporation (L5–S2). D: high power view of the boxed area in B. Location and morphology of labeled cells suggests that they are mostly motoneurons. E: transverse section at L6 following injection of fluorescein dextran into the central canal and electroporation (L4–S2). Note the fluorescent blood vessel arising from anterior spinal artery (ASA). F: higher magnification view of boxed area in C. G: lateral sagittal section (200 μm from lateral surface) following injection of the calcium green salt into the central canal and electroporation (L4–S1). Vertical dotted lines indicate approximate borders between spinal segments. H: higher magnification view of boxed area in G. In all images, dorsal is up and ventral is down.

Labeling pattern following central canal dye injection and electroporation

We also observed that a subdural injection of the dye often resulted in dye entry into the central canal. This occurred even though the electrode was restricted to the subdural layer and did not penetrate the cord. This finding suggested another potential route for dye distribution. To investigate this possibility, we injected the dye directly into the central canal and electroporated the cord one to two segments caudal to the injection site (n = 4). In one non-electroporated cord, the dye remained restricted to the central canal, and no cellular labeling was seen (Fig. 4A). Some blood vessels, in particular the central branch of the anterior spinal artery, were often labeled (Fig. 4A, arrowhead) as seen in control cords injected subdurally (Fig. 1E). Electroporation (21–27 V) systematically induced a characteristic pattern of cellular labeling comprising two groups of laterally located cells: one in the ventral horn and another cluster of small cells in the lateral intermediate gray matter (Fig. 4B). Although more extensive, this labeling was similar to the one observed with ventral artery injection (cf. Figs. 4, B and C, D and F, and 3D and 4H). Particularly striking was the massive labeling of ventrally located cluster of cells, presumably motoneurons, occurring with this type of injection procedure (Fig. 4, D, G, and H). In these experiments, high power views of the central canal area revealed that some cells lining the central canal (clearly seen at the top of Fig. 3E) were making contact with blood vessels, often the central branch of the anterior spinal artery, through end feet projections (Fig. 4E).

Viability of electroporated cords

To establish if electroporation compromised the viability or function of the cord, we used ventral root electrical recordings to determine if reflex and motor behaviors were preserved after electroporation. In addition, we determined if the dye-loaded cells were capable of producing optical signals in response to antidromic stimulation, reflex, and motor behaviors.

Effects of electroporation on reflex function

We first tested the effects of electroporation on mono- and polysynaptic ventral root potentials evoked by dorsal root stimulation. Previous work has shown that a suprathreshold stimulus applied to a lumbar dorsal root evokes a monosynaptic reflex on the homonymous ventral root and a polysynaptic reflex on the homonymous and adjacent ventral roots. Therefore we stimulated and recorded from one dorsal/ventral root pair outside the electroporated area (L1) and one within the electroporated region (L6; electroporation electrodes spanning L5 to S2; Fig. 5D) and monitored the responses before and after electroporation. In this way, it was possible to establish if the effect of electroporation on spinal cord function was restricted to the tissue between the electrodes.

FIG. 5.

Depression and recovery of spinal cord function following electroporation. A: reflex responses (DC to 3 kHz) obtained in control and at different times after electroporation from ventral roots L1 and L6 (right side) in response to the stimulation of the homonymous dorsal roots. Arrowheads indicate the stimulus on the control traces. B: graphs showing percentage changes in the mono- and polysynaptic reflexes averaged from 5 experiments. Control value (C) is the amplitude before electroporation and is set at 100%. C: activity recorded in the L1 and L6 ventral roots in response to a train applied to a coccygeal ganglion before (control) and at different times after electroporation (3 mm genetrodes over L5–S1). Top panels show unfiltered activity (DC to 3 kHz), and bottom panels show discharge (AC 10–500 Hz) from a portion of each of the episodes (dotted box). D: schematic drawing showing location of segments L1 and L6 with respect to the genetrodes.

In 7/11 experiments, the effects of electroporation on reflex function were studied in the absence of dye application. Electroporations performed between 5 and 9V (3 experiments, 7 electroporations) did not abolish monosynaptic or polysynaptic components of the reflex responses within the electroporated area (data not shown). Electroporation performed at a voltage of 9–21 V (5 experiments, 7 electroporations) caused a temporary and local abolition of the monosynaptic and polysynaptic reflex followed by a partial or complete recovery (Fig. 5, A and B). The monosynaptic component of the dorsal root–evoked reflex recovered less completely than the polysynaptic component (Fig. 5B). In five experiments, the monosynaptic reflex evoked by stimulation of dorsal root L6 and recorded from ventral root L6 (within the electroporated area), recovered to 54.8 ± 19.2% of its control value (data measured 62–143 min after electroporation). In contrast, over the same period, the L6 polysynaptic dorsal root reflexes recovered to 117.3 ± 43.2% of their control value (5 experiments, Fig. 5B). In these same experiments, reflex function was also measured in L1, outside the electroporated area. We found no depression of either monosynaptic or polysynaptic reflexes evoked by stimulation of the L1 dorsal root and recorded in L1 ventral root. Instead, the reflex amplitudes were potentiated. The amplitudes of the monosynaptic and polysynaptic components of the reflex were increased to 172.8 ± 36.6 and 110.1 ± 22.3% of their respective control values.

Above 30 V (2 experiments, 2 electroporations), the reflexes evoked and recorded within the electroporated area were abolished and did not recover for the duration of the experiment (≤4 h after electroporation). However, the polysynaptic reflex evoked in L6 ventral root by the stimulation of the L1 dorsal root did partially recover, showing that the L6 area is still functional at these voltages and that the dorsal roots within the electrodes are likely to be damaged by the electroporation.

Effects of electroporation on locomotor-like activity

We also examined the effects of electroporation on rhythmic locomotor-like activity generated by a train of stimuli applied to a coccygeal ganglion (n = 11; 9 electroporation alone and 2 electroporation and sub-dural dye injection). This activity comprises rhythmic bursts that alternate between the left and right sides and between rostral (L1) and caudal lumbar (L6) ventral roots (Whelan et al. 2000).

When the electroporation voltage was ≤9 V (3 experiments, 7 electroporations), we found little effect on this type of rhythmic activity. In contrast, when the electroporation voltage was between 9 and 21 V (7 experiments, 9 electroporations), we found that locomotor-like activity was transiently but fully depressed after electroporation (Fig. 5C). Early after electroporation, not even reflex responses to individual stimuli could be seen. Typically, the ability to generate any kind of activity first recovered in L1, several segments away from the rostral limit of the electroporated area (L5). The sequence of recovery usually exhibited the following pattern: no response (1–10 min), reflex responses to individual stimuli of the train (5–15 min), sustained tonic depolarization (10–20 min), tonic bursting (15–20 min), and the first rhythmic discharges (20–30 min; Fig. 5C). A similar sequence of recovery occurred within the electroporated region (L6) but with a 10- to 30-min delay. After rhythmic discharges had appeared on both roots, the quality of the rhythm kept improving, and between 1 and 2 h after electroporation, the proper alternation between the rostral and caudal roots was observed (Fig. 5C, right). Finally, for voltages >30 V (2 experiments, 2 electroporations), only tonic bursting and rare nonalternating discharges could be recorded even 3–4 h after electroporation.

Optical signals from electroporated cords

It might be argued that the recovery of function after electroporation depended on cells that were not themselves electroporated or loaded with dye. This could occur if the electroporated cells were silenced by damage or if their function was compromised by an unusually high dye concentration. To test these ideas and to establish if loaded cells could produce optical transients, we monitored fluorescence changes during antidromic and reflex activity. In previous experiments in the chick, mouse, and rat spinal cord, we have suggested that the calcium transients evoked during antidromic stimulation or rhythmic behaviors are generated primarily by action potentials (Bonnot et al. 2002b; Lev-Tov and O'Donovan 1995; O'Donovan et al. 1994). As a result, the presence of optical signals synchronized with the electrical activity provides good evidence that dye-labeled cells are capable of generating action potentials and of participating in network activity.

In 25 electroporated preparations (9 calcium green dextran, 16 calcium green salt), optical signals were successfully recorded in response to stimulation of ventral or dorsal roots. Most recordings were made using conventional epifluorescence microscopy (Figs. 6, A–C, and 7, A–C). However, four electroporated preparations were also imaged using two-photon microscopy (Figs. 6, F–I, and 7, D–F) allowing cellular resolution as deep as 350 μm into the cord.

FIG. 6.

Calcium transients recorded from electroporated cells in response to antidromic stimulation of the ventral roots. A: epifluorescence images of the L4–L6 area following subdural application of the dye (same experiment as in Fig. 2) and viewed through the lateral white matter, as shown in the inset. B: difference image of the dotted area in A, obtained by subtracting the control fluorescence (image A) from an average generated during antidromic stimulation of the L5 ventral root at 20 Hz. C: fluorescent transients generated in the ventral area (black box in B) in response to ventral root stimulation at different frequencies. Single shock responses are arrowed. D: whole cell recording from a motoneuron in the L1 segment of the neonatal mouse spinal cord showing that a single action potential is evoked in response to a single stimulus applied to the ventral root. Note the prominent afterdepolarization (ADP), which is known to be accompanied by calcium entry through voltage-dependent calcium channels. E: comparison of averaged time course of the antidromic action potential recorded from 7 lumbar motoneurons in 6 preparations (red traces, ±SD) with the calcium transients recorded from lumbar motoneurons filled with either the calcium green salt by electroporation (averaged from 5 preparations; green trace, ±SD) or with calcium green dextran by retrograde loading (averaged from 4 preparations; blue trace, ±SD). Intracellular recordings were made in a different set of experiments from optical recordings, and cells were not filled with calcium dyes. F: a single optical scan 100 μm under the lateral surface of an electroporated cord obtained using 2-photon microscopy. Ventral area outlined by the red box is shown at higher magnification in G. G: average image generated over a 2-s period (frame scan: 180 ms) in the absence of activity. H: active difference image obtained by subtracting the control fluorescence (image G) from an average generated during 20 Hz stimulation of the L5 ventral root. I: calcium transients corresponding to the 5 regions of interest drawn in H (4 neurons and 1 background).

FIG. 7.

Calcium transients recorded from electroporated cells in response to a single stimulus to dorsal root afferents. A: fluorescence microgram of the L5–L6 area of the cord following subdural application of the calcium green salt and electroporation (L4–S4) and viewed through the lateral white matter. Vertical dotted line is the border between L5 and L6. B: difference image obtained by subtracting the control fluorescence (image A) from an average generated during the peak response to a single stimulus applied to the ipsilateral L6 dorsal root. C: optical (top trace measured from red boxed area shown in A) and electrical responses (bottom trace, L6 ventral root) following a single stimulus to the dorsal root. Arrowhead indicates stimulus. Note that the optical trace is derived from dorsally located interneurons, while the electrical trace corresponds to activity of ventrally located motoneurons. D: 16 frame average (7.8 s/frame) from dorsal horn cells located at a depth of 170 μm from the lateral surface. E: active difference image obtained by subtracting the control fluorescence from an average generated during a single stimulus applied to the L6 dorsal root. F: calcium transients corresponding to the 3 regions of interest drawn in E. Bottom trace is the simultaneously recorded ventral root L6 electrical response (DC to 3 kHz). Dotted line at top of panel indicates the period over which active frames were averaged to generate the difference image in E.

With epifluorescence microscopy, the labeling seen through the lateral white matter (after 1–2 h of wash-out following electroporation) consisted mostly of a diffuse fluorescence spanning the electroporated area (Fig. 6A). Dorsal and ventral columnar areas of stronger labeling could be delineated at low magnification, and blurred cellular profiles could sometimes be detected (especially with the dextran). As a consequence, when optical signals were recorded, they comprised diffuse population activity, and cellular signals were rarely seen. We detected optical signals following electroporation of both dextran and the salt, although signals were about one-half the size with the dextran compared with the salt. For this reason, the data described below are restricted to the electroporation experiments using the salt (16 experiments).

Antidromic stimulation of motoneurons

We first present the results of antidromic stimulation of motoneurons (Fig. 6). In 11/16 experiments, stimulation of any ventral root within the electroporated area resulted in optical signals from neurons spanning approximately one segment and located in the ventral third of the cord when viewed through the lateral white matter (Fig. 6, A and B). This region corresponds to the location of motoneurons. We examined the frequency dependence of the antidromic signal in six experiments (9 ventral roots). As shown in Fig. 6C, we found that the optical signal increased linearly with the stimulation frequency from 2.6 ± 0.7% at 5 Hz to 6.6 ± 2.0% at 50 Hz, and the optical response to a single shock could be resolved in each case (2.1 ± 0.9%; Fig. 6C). In the five remaining experiments, in which only a few motoneurons were observed in fixed sections, antidromic optical responses were weak (peak ΔF/F < 2.7%, 50 Hz train).

In a separate series of experiments, we established how many action potentials occurred in lumbar motoneurons following a single antidromic stimulus applied to the ventral root. Whole cell recordings obtained from lumbar motoneurons (15 cells, 10 preparations) showed that a single action potential was evoked by a single ventral root stimulus (Fig. 6, D and E). Ten of 15 motoneurons revealed a prominent afterdepolarization (ADP; Fig. 6D), which is known to be accompanied by calcium entry through voltage-dependent calcium channels (Kobayashi et al. 1997; Viana et al. 1993; Walton and Fulton 1986). This observation indicates that calcium transients detected in electroporated motoneurons in response to a single ventral root stimulus correspond to a single action potential. In Fig. 6E, we show the time course of the antidromic action potential together with the normalized and averaged calcium transients from electroporated and retrogradely loaded motoneurons. The peak of the calcium transients coincided with the end of the ADP, and the decay kinetics of the transients were similar for electroporated and retrogradely labeled motoneurons. The interpolated duration of the transient at 50% amplitude was 9.1 ± 1.5 frames (∼300 ± 50 ms; n = 5 preparations) for the electroporated motoneurons and 10 ± 5.4 frames (330 ± 178 ms; n = 4 preparations) for the retrogradely loaded motoneurons.

Because of the poor cellular resolution obtained with conventional epifluorescence microscopy, we performed additional experiments using two-photon confocal imaging. In three of four experiments, fluorescently labeled motoneurons were detected deep within the spinal cord tissue (60–170 μm; Fig. 6, F and G), and cellular signals could be resolved in individual motoneurons (Fig. 6, H and I). The absence of signal in the background extracellular region (region of interest 1) confirms the cellular nature of these signals. The fractional change in fluorescence measured during antidromic stimulation at 20 Hz was 40.6 ± 18.8% (11 cells in 3 experiments). In the same three experiments, 20 Hz antidromic optical signals were also measured using epifluorescence and averaged 5.86 ± 6.5%.

Dorsal root-induced optical signals

In 13/16 experiments, using the same lateral visualization, large optical signals were recorded in dorsal populations of neurons during reflex responses to a single shock or a train applied to the ipsilateral dorsal root located within the electroporated area (Fig. 7A and B). The peak fractional change in fluorescence measured in response to a single dorsal root stimulus averaged 9.7 ± 2.0% (SD; range, 7.2–12.7%, n = 11). The onset of the optical signal coincided with the onset of the simultaneously recorded ventral root potential. Dorsal root–activated optical signals were also observed in the ventral region, although they were smaller (<50%) than those in the dorsal region (P < 0.05, n = 7). In the remaining three experiments, the optical signals accompanying dorsal root stimulation were small (peak ΔF/F < 5%, single stimuli or 20 Hz train). Subsequent analysis of fixed sections obtained from two of these preparations revealed weak labeling in the dorsal area.

Confirmation of the cellular origin of these dorsal-root evoked signals was obtained with two-photon microscopy. As shown in Fig. 7D, cellular profiles were easily resolved, and signals were obtained from individual neurons. The fractional change in fluorescence during a train of dorsal root stimuli at 20 Hz averaged 41.8 ± 26.7% (16 cells in 3 experiments). In the same three experiments, the dorsal root evoked signals measured with epifluorescence averaged 10.5 ± 5.2%.

Optical imaging of locomotor-like rhythmicity

Locomotor-like activity is one of the most organized behaviors generated by the isolated mouse spinal cord. In the previous section, we established that this type of activity is initially depressed but recovers following electroporation. In the next set of experiments, we investigated if neurons participating in this behavior can be labeled by electroporation and can generate optical signals. This was particularly important because standard methods for labeling locomotor-related interneurons have significant limitations.

In 10/25 optically recorded preparations, trains of coccygeal stimulation successfully evoked alternating discharges on rostral and caudal ventral roots, characteristic of locomotor-like behavior (Whelan et al. 2000). Such trains gave rise to a tonic optical activation sustained for the whole duration of the train (10 s) with an average peak intensity of 7.4 ± 3.6% (range, 3.3–12.0%, n = 9). The activity was concentrated in the ventral and intermediate parts of the spinal cord (Fig. 8B), with significantly less activity in the dorsal portion of the cord (<50% of that in the ventral region, P < 0.05, n = 8). In three cases where the locomotor-like rhythm was robust and the labeling strong, we observed a 1–2% rhythmic variation of the fluorescence signal superimposed on a tonic increase of fluorescence (Fig. 8C). These rhythmic fluorescence variations were strictly coupled to the rhythmic discharge recorded from the ventral roots and were observed only in the most ventral column of cells, shown to be mostly motoneurons. Similar rhythmic signals have been recorded previously from retrogradely labeled motoneurons (Bonnot et al. 2002b). In contrast to the ventral recordings, signals from the intermediate region displayed a stable level of tonic activation for the whole duration of the coccygeal train.

FIG. 8.

Calcium transients of electroporated cells during an episode of locomotor-like activity evoked by a train of stimuli applied to coccygeal ganglia. A: epifluorescence images of the L4–L6 area following subdural application of the dye and electroporation (same experiment as in Fig. 2) and viewed through the lateral white matter. Field of view is the same as that shown in Fig. 6A. B: difference image obtained by subtracting the control fluorescence (image A) from an average generated during the peak response (black horizontal line in C) to a train of stimuli applied to a coccygeal ganglion. C: comparison of optical and electrical records during the episode. Top 2 traces show fluorescent transients generated from the ventral (orange) and intermediate (blue) rectangles shown in A. Bottom traces show the filtered (AC 100–500 Hz), digitized (1 KHz), rectified, and integrated (30 Hz) locomotor-like pattern recorded simultaneously on the ventral roots. Boxed region shows the alternation between the right L5 and L2 ventral root discharge and the synchrony between the ventral calcium transient and the right L5 discharge.


We have shown that electroporation can be used to rapidly load calcium-sensitive dyes into neurons over several segments of the spinal cord, that the electroporated cord can generate electrical activity, and that loaded cells can generate optical signals during antidromic, reflex, and locomotor-like activity. The extent of labeling throughout the three-dimensional volume of the cord and the diversity of cells labeled are unmatched by any of the currently available methods for introducing calcium dyes into neurons. In combination with confocal or two-photon microscopy, this new technique allows neuronal and network function to be visualized in relatively intact tissues in a manner not possible before. In the spinal cord, for example, it will now be feasible to analyze the dynamics of the interneuronal networks and neurons responsible for locomotion. To date, this has only been possible using intracellular techniques that are limited to recording from one or two cells simultaneously (Butt and Kiehn 2003). In addition, because of both the growing use of electroporation for the introduction of genes in vivo and the increased depth of optical resolution obtained with two-photon confocal imaging, we anticipate the extension of this approach to in vivo imaging of network function. We believe that in vivo studies will have an inherent advantage over in vitro studies because the electroporation can be performed well before the recordings, allowing adequate time for washout of extracellular dye.

Although we have focused on neuronal labeling in this paper, it is very likely that glial cells were also labeled by this method. Indeed, our observation of loaded central canal cells with end feet on blood vessels supports this idea. Unfortunately, nothing is known about glial calcium transients in the spinal cord at the ages we have studied. However, we note that the calcium transients we have recorded in response to antidromic stimulation and locomotor-like activity have the same time course as those recorded earlier from retrogradely filled neurons, a case in which glial cells were not labeled. Furthermore, using two-photon microscopy, we have been able to monitor the calcium transients of individual neurons. Nevertheless, in future experiments, it will be important to confirm glial labeling directly by immunocytochemistry and to establish if any of the optical signals we have recorded during antidromic, reflex, or locomotor-like activity contain contributions from labeled glial cells.

Mechanism of dye distribution throughout the cord

Our experiments have established that cellular loading is caused by electroporation and not by some other mechanism (e.g., endocytosis). Control cords that were exposed to the dye but were not electroporated or were electroporated at low voltages (<15 V) exhibited no labeled cells in fixed sections and did not produce optical signals during electrical activity. Second, when the dye was injected subdurally or spread widely throughout the cord following intravascular or central canal injections, the labeled cells were co-extensive with the electroporation electrodes, and no labeling was found outside this region.

Following subdural dye injection (at the level of the ventral commissure), we found that electroporation resulted in neuronal labeling throughout the three-dimensional extent of the cord. Although neuronal labeling was sometimes stronger on the side of the cord adjacent to the cathode, in many cases labeling was of similar extent and intensity on both sides. This is in contrast to electroporation following DNA injection into a closed cavity such as the developing neural tube of the chick embryo (Ogura 2002) or the brain ventricles of the Xenopus tadpole (Haas et al. 2002) and mouse (Saito and Nakatsuji 2001) or into the central canal of the mouse spinal cord (Saba et al. 2003). It seems unlikely that bilateral labeling is the result of lateral spread of the dye underneath the dura matter because lateral or dorsal subdural injections, followed by electroporation, did not result in significant labeling. Furthermore, when the dye was injected into the central canal, without the opportunity for lateral spread, labeling was observed on both sides of the cord.

To account for these findings, we hypothesized that the dye might enter the vasculature to be distributed through the cord. Several observations support this idea. First, following dye injections into either the subdural space or the central canal, we found fluorescently labeled blood vessels. Electroporation was not responsible for this labeling because blood vessels were also labeled when the dye was applied without subsequent electroporation. The most compelling evidence supporting intravascular distribution was the observation that intra-arterial injection followed by electroporation resulted in cellular labeling.

We also noticed that subdural injection of the dye frequently resulted in dye entry into the central canal, even though the electrode had not penetrated the cord. This observation raised the possibility that the dye entered the central canal through the vasculature. The tight communication between cells lining the central canal and blood vessels offers a potential route of communication between the central canal and the vasculature. Consistent with this idea, we found very similar labeling patterns if the dye was injected into the anterior spinal artery or directly into the central canal. In both instances, central and lateral clusters of cells were labeled. This labeling pattern might be due to a more intense vascularization of central and lateral areas or to thinner blood vessel walls that may render electroporation more efficient in these areas.

It is not clear, however, if access of the dye to the vasculature and the central canal is the sole mechanism for dye distribution within the cord. Cellular labeling was always more extensive following subdural application compared with intravascular or central canal injections. This may have occurred because more dye could be injected subdurally than into either the anterior spinal artery or the central canal. Alternatively, some of the dye may enter the cord by diffusion from the ventral surface, although it is difficult to see how this mechanism could produce labeling in the dorsal part of the cord several hundred microns from the ventral surface. It is also possible that some dye is driven electrophoretically into the extracellular space by the electroporation pulses.

Electroporation-induced depression of neuronal activity

To assess the effects of electroporation on neural function, we monitored mono- and polysynaptic dorsal root-evoked responses in motoneurons and the capacity of the cord to generate locomotor-like rhythmic behavior. These experiments were generally done without dye injections or the low calcium/high magnesium solutions we used during electroporation of the dyes. While it is likely that the low calcium/high magnesium solution offered some tissue protection during electroporation, we could not use it because it would have depressed neural function during the period we were monitoring its recovery.

Following electroporation, we observed recovery of polysynaptic reflex and rhythmic activity as long as the electroporation voltage was between 5 and 25 V, with a recovery time course strictly dependent on the applied voltage. While the polysynaptic dorsal root reflexes recovered to control values within 20–40 min (Fig. 5B), the monosynaptic dorsal root reflexes regained only 50–60% of their control level by 2 h after the electroporation. The reasons for this incomplete recovery are unknown, but we note that it did not compromise the ability of the cord to produce complex locomotor-like rhythmic behavior. One possibility is that the cut dorsal roots might have been damaged by their proximity to the electroporation electrodes. Typically, the electroporation electrodes were positioned <1 mm from the cord edges. As a result, the cut dorsal roots sometimes touched the electrodes during electroporation, possibly sustaining some damage. The fact that the polysynaptic dorsal-root-evoked responses did recover completely may reflect network homeostatic behavior that has been identified in the developing chick spinal cord (Chub and O'Donovan 1998).

The source of the electroporation-induced depression of function is unknown, but presumably it involves the exchange of the intracellular and extracellular solutions during the electroporation process (Weaver 1993). However, the fact that recovery could occur after the electroporation-induced depression indicates that cellular homeostatic mechanisms can restore cellular function under our experimental conditions.

Finally, we observed a potentiation of the monosynaptic reflex in L1, remote from the electroporated area that might be explained by a decreased presynaptic inhibition exerted in L1 by electroporated caudal primary afferents.

Optical signals from electroporated cells

One concern with the technique, which we have been able to disprove, is that the electroporated cells would be incapable of producing optical signals. This could occur if they were irreversibly damaged by the electroporation or alternatively because the dye might become compartmentalized within the cell. Although we did observe some evidence of compartmentalization in the sectioned material (e.g., Fig. 1C), electroporated cells were capable of generating optical signals during antidromic, reflex, and locomotor behaviors. In addition, we showed that these various sources of stimulation induce different and recognizable patterns of activation among the labeled population of cells.

These observations suggest that the dye concentration achieved during electroporation does not compromise network function. Although we did not estimate the dye concentration directly in electroporated neurons, it was our impression (based on the intensity of the fluorescence) that the concentration achieved with electroporation was similar or lower than that achieved with retrograde loading (Bonnot et al. 2002b). Consistent with this conclusion, we found that the decay kinetics of the calcium transients accompanying a single antidromic action potential were similar for electroporated and retrogradely loaded motoneurons (Fig. 6E). In a recent study employing retrograde loading of calcium dyes into spinal neurons, the final dye concentration was estimated to be ∼10 μM (Viana di Prisco and Alford 2004). Because retrograde loading has not been shown to alter network behavior (Bonnot et al. 2002b; Nakayama et al. 2002; O'Donovan et al. 1994; Ritter et al. 2001; Viana di Prisco and Alford 2004), it seems reasonable to conclude that the dye concentration achieved with electroporation is also unlikely to significantly perturb neuronal function.

When whole cords loaded by electroporation were visualized through the lateral or ventral white matter using epifluorescence microscopy, the fluorescent labeling was usually diffuse. Cellular profiles were more apparent with the dextran form of calcium green dextran than with the salt, but in either case, the cellular resolution was not as sharp as that seen with retrograde filling, and smaller optical signals were obtained (see Bonnot et al. 2002b). Epifluorescent optical signals obtained during antidromic stimulation of motoneurons following electroporation of the calcium green salt were about four to six times smaller than similar signals obtained from retrogradely filled preparations of the neonatal mouse or rat (Bonnot et al. 2002b; Lev-Tov and O'Donovan 1995). This could be due to several factors. First, in contrast to retrograde loading, with electroporation there is a significant amount of extracellular dye that degrades cellular contrast. Consistent with this idea, we noticed a reduction of background fluorescence and an improvement of cellular resolution, as the extracellular dye washed out during the experiment. Because extracellular dye is not reactive to intracellular calcium changes, it will reduce the activity-dependent changes of fluorescence that are normalized to the resting level of fluorescence. In addition, some of the dye was compartmentalized in the electroporated cells and was therefore not available to cytoplasmic calcium (Malgaroli et al. 1987). Second, as discussed above, the intracellular concentration of dye reached with electroporation may have been lower than with retrograde loading, making it difficult to detect cells through the white matter with conventional epifluorescence microscopy and probably contributed to the smaller size of the optical signals. Finally, retrograde filling labels a specific subgroup of neuronal somata and their associated processes, whereas electroporation tends to label many more cells and processes, again degrading our ability to resolve somatic profiles with epifluorescence. We do not, however, consider cellular resolution and signal size to be limiting factors with this technique because the optical signals exhibited a good signal to noise ratio and, using two-photon microscopy, much larger cellular signals could be resolved from neurons located 60–350 μm from the white matter surface.

Concluding remarks

Electroporation has allowed us to label and to visualize the activity of a greater diversity of spinal neurons, throughout a greater extent of the dorso-ventral and medio-lateral axis of the cord and over more segments than any currently available method. In the future, this will allow us to address questions that have hitherto been inaccessible. For example, the limitations of other loading methods have prevented us from visualizing the cellular and spatiotemporal organization of spinal interneuronal activity during reflex or locomotor-like behavior in the isolated spinal cord.

The coupling of two-photon microscopy with a spatially controlled delivery of calcium indicators to extended cell populations should allow calcium imaging to become even more informative in many different parts of the nervous systems. Ultimately, we expect to expand the use of this approach to optical imaging in vivo of the brain or spinal cord of adult rodents. Furthermore, in future experiments it may be possible to combine calcium imaging with gene delivery, a capability that only this method of loading can offer.


We thank D. Brewer for help with histology, E. Ralston and K. Zaal from the National Institute of Arthritis and Musculoskeletal and Skin Diseases imaging facility for technical support with the Olympus FV300, and J. Simmers for comments on the manuscript.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract