Calcium Signaling in Intact Dorsal Root Ganglia: New Observations and the Effect of Injury

• ❖ Intracellular calcium concentration is an important second messenger in sensory neurons

• ❖ Traditional stimulation of sensory neurons in acute cultures by electrical fields has several drawbacks

• ❖ Recordings from neurons in intact ganglia reveal that each neuron type has a distinct firing frequency at which the calcium signal is maximized and that calcium is elevated even by single action potentials, particularly in slow conducting nocicepetive neurons

UNDERSTANDING the function of the peripheral sensory neuron is central to the science of anesthesia. Not only does activation of sensory neurons initiate the neurophysiologic events that trigger pain, but also other sensory neurons modulate reflexes that control muscle tone, ventilation, and thermoregulation during and after anesthesia. Also, temporary interruption of sensory neuron function is the core requirement of regional anesthesia. Furthermore, dysfunction of sensory neurons is a cause of neuropathic pain syndromes that are especially challenging to treat. These neurons serve their primary task of providing the central nervous system with environmental and internal information by selectively generating and transmitting action potentials (APs) in response to mechanical, thermal, and chemical stimuli. In addition to this electrophysiologic function of the neuronal membrane, important cytoplasmic and nuclear functions are regulated by the critical chemical messenger Ca²⁺, an ion that regulates such diverse events as neuronal differentiation, gene expression, neurotransmitter release, excitability, and apoptosis.1The increase and decrease in the cytoplasmic Ca²⁺transient that follows membrane depolarization are initiated by influx through the membrane and are further shaped by release from intracellular stores, sequestration into organelles, and extrusion from the cell. Better understanding of these complex events in sensory neurons is essential to devising safer local anesthetics and more effective chronic pain treatments.2,3 

Standard techniques have been used to explore cytoplasmic Ca²⁺signaling in sensory neurons using Ca²⁺-sensitive fluorophores and advanced cameras. Most often, sensory neuron somata are dissociated from the dorsal root ganglion (DRG) and thereafter activated by membrane depolarization using increased bath K⁺concentration or field stimulation. Although this model has yielded useful insights, there are significant limitations. The timing of activation by K⁺is limited to applications of at least a few seconds. This duration of exposure results in sustained membrane depolarization that may model events at transduction sites in the periphery, but is massively prolonged compared with trains of AP depolarizations that each last only a few milliseconds. Field stimulation, by which electrical current is passed between electrodes in a bath containing the neurons, has been devised to address this limitation.4However, there are theoretical concerns. Application of voltage greater than 1.23 V, the standard potential of water at 25°C, drives decompensation of water by the process of electrolysis that generates hydrogen at the anode. Further, in saline solutions, oxidation of chloride ions to chlorine gas may occur at the anode. Although dissociation and culturing of DRG neurons are also a part of the standard technique, removal of the axon from the sensory neuron soma eliminates as much as 99.8% of the volume of the cell.5In addition, the neurons are deprived of their natural environment by removal of the surrounding envelope of closely apposed satellite glial cells,6and digestion may affect membrane proteins, which in turn alters neuronal function.7 

Accordingly, in addition to examining neuronal performance under customary dissociation, bath, and stimulation conditions, the experiments described here were designed to adapt elements of various techniques to develop and test a method for recording Ca²⁺signals in sensory neurons with improved physiologic relevance while avoiding the limitations noted above. Our overall purpose was to provide improved tools for exploring sensory neuron function in anesthesia-related topics. Because nerve injury produces substantial derangement of sensory neurons Ca²⁺signals,8–11we examined this condition as an exercise in applying our improved technique.

All procedures were approved by the Animal Care Committee of the Medical College of Wisconsin, Milwaukee, Wisconsin. Male Sprague–Dawley rats (150 g; Taconic Farms, Inc., Hudson, NY) received anesthesia (2% isoflurane in oxygen), skin incision, and skin stapling in the control group, whereas the spinal nerve ligation neuropathic pain model was used in others for which the sixth lumbar (L6) transverse process was removed, and the L5 and L6 spinal nerves were ligated with 6–0 silk ligature and transected.12Sensory function was evaluated by determining the behavioral response to noxious punctate mechanical stimulation that was produced by applying a 22-gauge spinal anesthesia needle to the plantar surface of the hindpaw with pressure adequate to indent but not penetrate the skin.13The frequency of either a brief withdrawal or a sustained paw lifting, shaking, and grooming (hyperalgesia-type behavior) was determined during 10 touches on each of the three occasions between 10 and 17 days after surgery. For spinal nerve ligation rats, only those with more than 20% hyperalgesic response rate were studied in the experiments described below.

For dissociation, ganglia were harvested rapidly after decapitation. The fourth and fifth lumbar DRGs were incubated in 0.01% blendzyme-2 (containing collagenase and neutral protease; Roche Diagnostics, Indianapolis, IN) for 30 min followed by incubation in 0.25% trypsin (Sigma Aldrich, St. Louis, MO) and 0.125% DNase (Sigma) for 30 min, both dissolved in Dulbecco's modified Eagle's medium/F12 with glutaMAX (Invitrogen, Carlsbad, CA). After exposure to 0.03% trypsin inhibitor and centrifugation, the pellet was gently triturated in culture medium containing Neural Basal Media A with B27 supplement (Invitrogen), 0.5 mm glutamine, 10 ng/ml nerve growth factor 7S (Alomone Labs, Jerusalem, Israel), and 0.02 mg/ml gentamicin (Invitrogen). Neurons were plated onto poly-l-lysine-coated glass cover slips (Deutsches Spiegelglas; Carolina Biologic Supply, Burlington, NC), maintained at 37°C in humidified 95% air and 5% CO²for 2 h, and were studied no later than 6 h after harvest. Alternatively, intact ganglia with dorsal roots attached were harvested by laminectomy during isoflurane anesthesia and placed in artificial cerebrospinal fluid (128 mm NaCl, 3.5 mm KCl, 1.2 mm MgCl², 2.3 mm CaCl², 1.2 mm NaH²PO⁴, 24.0 mmc NaHCO³, 11.0 mm glucose) bubbled by 5% CO²and 95% O²to maintain a pH of 7.35.

Intracellular recordings were performed with microelectrodes fashioned from borosilicate glass with an Omega dot fiber (FHC, Bowdoin, ME) with resistances of 80–120 mΩ when filled with 2 m potassium acetate buffered to pH 7.6, mounted in a holder with a sideport to release luminal pressure. Dissociated neurons were superfused with Tyrode's solution (140 mm NaCl, 4 mm KCl, 2 mm CaCl², 10 mm glucose, 2 mm MgCl², 10 mm HEPES) at room temperature, whereas intact ganglia were superfused with artificial cerebrospinal fluid regulated to 35°C with an inline heat exchanger, and both solutions were delivered at 3 ml/min. (Noncomparable solutions were used for dissociated neurons and intact tissue to optimize preparation stability and to replicate customary use.) Lidocaine was administered through a microperfusion pipette stationed 100 μm from the tissue, with flow driven by pressure from a valve system (IIe, Toohey Company, Fairfield, NJ). The DRG was stabilized in the chamber with an anchor fashioned from gold wire holding parallel strands of 10–0 polypropylene monofilament at 1-mm intervals. Superficial neurons in the first or second layer were impaled while observed with an upright microscope (Eclipse E600FN; Nikon USA, Melville, NY) equipped with differential interference contrast optics and infrared illumination through a 40× water immersion objective (Nikon Fluor 40×/0.8W). Membrane potential was recorded using an active bridge amplifier (Axoclamp 2B; Axon Instruments, Molecular Devices, Sunnyvale, CA), filtered at 10 kHz, and digitized at 40 kHz (Digidata 1322A, Axon Instruments and Axograph × 1.1; Axograph Scientific, Sydney, Australia). Recording was initiated after resting membrane potential had stabilized (3 min) to less than −50 mV. Data were obtained only from neurons in which membrane potential remained stable, in which APs could clearly be discerned from the stimulus artifact, and in which an AP amplitude was greater than 45 mV. Leakage of neuronal contents after withdrawal of the electrode conceivably could alter function of adjacent neurons for which reason only nonadjacent neurons were examined.

Dissociated neurons were activated by one of the two techniques. Field stimulation was achieved using either a constant voltage stimulator built in-house that delivered 1-ms pulses at 0–30 V, monitored continuously, or a constant current stimulator (SIU-102, Warner Instruments, Hamden, CT). Coverslips carrying neurons were mounted in a 260-μl chamber (Warner RC-21BFRS), and stimuli were delivered either through the two built-in platinum electrodes (10 mm long, 6-mm separation) or through a custom-made pair of silver electrodes (5 mm long, 3-mm separation). Alternatively, impaled neurons were stimulated by delivery of depolarizing current through the stimulating electrode. Neurons in intact ganglia were activated by APs conducted from a stimulation site on the proximal end of attached dorsal roots in which a pair of platinum wire electrodes connected to an isolated current source (in-house design) delivered 0.06–1 ms pulses at 1.5× threshold for just generating an AP. Conduction velocity (CV) was measured by dividing the distance between stimulation and recording sites by the conduction latency.

The Ca²⁺-sensitive fluorophore Fura-2 was loaded into neurons by one of the two techniques. Dissociated neurons were incubated for 30 min in Fura-2-AM (5 μm; Invitrogen) in Tyrode's solution, followed by Tyrode's alone for 30 min. A technique previously reported for loading a fluorophore into other neuronal tissues14,15was adapted for neurons in intact ganglia. Neurons were loaded directly from the recording microelectrode filled with 10 mm Fura-2, a concentration recommended by the manufacturer (Invitrogen) and used by others for loading by microelectrode impalement,14,15dissolved in 2 m potassium acetate (pH 7.6 with HEPES).

Cells were imaged through a 20× or 40× objective, using a cooled 12-bit charge-coupled device camera (Coolsnap fx; Photometrics, Tucson, AZ). Excitation was achieved at 340 and 380 nm with a 150 W xenon lamp (Lambda DG-4; Sutter, Novato, CA), and images were collected at 510 nm using Metaflour software (Molecular Devices, Downingtown, PA). The ratio of fluorescent intensity at these wavelengths (R =I  ³⁴⁰/I  ³⁸⁰) was determined on a pixel-by-pixel basis for individual neurons after background subtraction. [Ca²⁺]^cwas estimated by the formula [Ca²⁺]^c=K  ^d·β· (R −R  ^min)/(R  ^max−R ), where β= (I  ^{380 max})/(I  ^{380 min}), K  ^d= 224 nm.16 R  ^min, R  ^max, and β were determined separately for dissociated neurons (0.38, 8.49, and 9.54, respectively; n = 33) and for intact ganglion recording (0.313, 11.28, and 28.64, respectively; n = 3). Ionomycin (5 μm for dissociated, 80 μm for intact DRG; Calbiochem, Gibbstown, NJ) in Ca²⁺-free bath produced a stable R^min, indicating equilibrium with bath solution, and exposure to Tyrode's solution with ionomycin produced R  ^max.8Although we report [Ca²⁺]^crather than fluorescence ratios, it should be noted that Ca²⁺concentrations more than 1 μm determined by Fura-2 become less reliable. No single fluorophore is fully reliable across the entire range of possible cytoplasmic Ca²⁺concentrations, but Fura-2 was chosen because of its optimal performance at [Ca²⁺]^cnear the neurons' resting level. Dimensions of cytoplasmic Ca²⁺signals are reported as amplitude of the activity-induced transient. Recovery of the transient in the form of decay time constant was determined for transients that resolved in a simple monoexponential fashion.

Dissociated neurons were chosen in the 30- to 35-μm diameter size range, whereas CV was used to categorize neurons in intact ganglia17because this tends to segregate neurons with specific receptive field properties.18,19Specifically, neurons with dorsal root CV less than 1.5 m/s were considered C-type, those with CV more than 15 m/s were considered Aβ-type, and neurons with CV more than 1.5 m/s but CV less than 10 m/s were considered Aδ-type. For neurons with CV between 10 and 15 m/s, long AP duration was used to classify the cells as Aδ-types.17Data are expressed as mean ± SEM. Student t  test (paired and unpaired), one-way ANOVA, and two-way ANOVA, with repeated measures if appropriate, were used (Statistica; StatSoft, Inc., Tulsa, OK) to detect differences in measured parameters between injury groups and stimulation protocols. Post hoc  testing was performed with paired comparisons by least-square differences with Bonferroni's correction for multiple comparisons. In many cases, data for more than one protocol were obtained from a single neuron. To describe the stability of transient dimensions with repetition, we subtracted each measurement from the average of the repeated measurements and averaged the absolute value of this difference. To describe variability of the decay time constant of a neuron (τ) across different stimulation frequencies (7, 20, 50, and 100 Hz), the τ of a neuron at frequency f (τ^f) was normalized to the average across frequencies of that neuron (τ^fnl=τ^f/([τ⁷+τ²⁰+τ⁵⁰+τ¹⁰⁰]/4)), then these were averaged for all neurons at a given frequency (τ^fAVG), which established the overall trend due to the effect of frequency. To describe the variation of a given neuron while accounting for the overall effect of frequency, we calculated the squared difference between the time constant and τ^fAVGof a neuron at each frequency, which was then summed across the four tested frequencies and divided by the degrees of freedom (i.e. , ∑^{f=7,20,50,100 Hz}(τ^fnl−τ^fAVG)²]/3), which is called as normalized variance hereafter. Significance was accepted at P < 0.05.

A total of 40 rats were used, 11 of which had spinal nerve ligation surgery, whereas 29 had skin incision only. Rats with spinal nerve ligation had a frequency of hyperalgesia-type responses to noxious mechanical stimulation of 39.8 ± 15.3%, whereas control animals had 0.0 ± 0.0 (P < 0.001).

Resistance between the field stimulation electrodes built into the chamber was 0.9–1.1 MΩ, and the resistance for the custom electrodes was 5.1 MΩ. Ca²⁺responses were maximal at voltages (recorded online) of 25–30 V. Close examination of the stimulating wires demonstrated accumulation of bubbles visible under 100× magnification even after two trains of 21 stimuli of 1-ms duration (fig. 1A). Indicator paper placed in the recording chamber to measure pH showed reciprocal color changes at the electrodes after the same stimulation, indicating pH values of less than 4 and more than 11 in the region of the electrodes (fig. 1B). We recorded the potential in the bath during field stimulation using either a low-resistance glass electrode (500 kΩ) attached to the electrophysiologic amplifier or a wire electrode connected to a digital oscilloscope. Both pulse generators and both stimulating electrode systems produced sustained bath potentials that fused into a continuous field potential with pulse durations of either 0.1 or 1 ms at 7 Hz (fig. 1C). Findings (data not shown) were similar with a biphasic stimulus pulse.

To identify the limits of field stimulation to generate trains of neuronal activity in neurons dissociated from adult rats, we tested the effect of stimulation frequency on Ca²⁺transient dimensions in dissociated control neurons (21 stimuli; 1-ms duration; 7, 20, 50, and 100 Hz in a random order; fig. 1D). Under these conditions, 11% (7/65) of neurons failed to respond at all, indicating highly variable response to field stimulation. In the remaining neurons, transient amplitude declined as stimulation frequency increased (P < 0.01, 40% decrease between 20 and 100 Hz; fig. 1E). If transient amplitude represents the summation of individual excitatory events, this finding could in part be due to a failure of field stimulation to activate neurons at higher stimulation frequencies. Although stimulation frequency had no effect on the average decay time constant, the decay time constant of individual neurons at different frequencies showed a large variability (fig. 1E). Specifically, the normalized variance of decay time constant of each neuron across the four frequencies was 13 ± 3%.

A possible explanation for the decline in [Ca²⁺]^ctransients with higher frequencies is that neuronal somata are not able to sustain firing rates beyond 20 Hz. To test this, we used direct intracellular current injection through a recording electrode to excite neurons that had been loaded by bath Fura-2-AM.20Somatic APs were discerned in all neurons (n = 7), and each of the 21 stimuli produced an AP in trains at 7, 20, 50, and 100 Hz (fig. 2A). In contrast to field stimulation, simultaneous [Ca²⁺]^crecording during direct stimulation showed no decrease in transient amplitude (P = 0.35, 7% difference between averages at 20 and 100 Hz, fig. 2B). The decay time constant was also not affected by frequency and had a normalized variance for each neuron across the four frequencies of 8 ± 3%. These findings indicate that adult rat sensory neuron somata are intrinsically able to generate APs at rates up to 100 Hz. If field stimulation were able to trigger APs at these same stimulation frequencies, comparable transient amplitudes should be generated. The significantly smaller transients (P < 0.01) and the 40% fall-off of transient amplitude with increasing frequency are strong evidences of a progressive failure of field stimulation to generate APs at frequencies at which the neurons are capable of firing.

Because dissociation alters the neuronal environment, traumatizes the neuron, and may digest cellular surface proteins, findings such as ability to fire at high frequencies could be dissociation-induced artifacts. Therefore, we developed a technique for imaging [Ca²⁺]^cin intact DRGs (fig. 2C). Dye loading with bath-applied Fura-2-AM proved unfeasible because individual neuronal profiles could not be identified during the fluorimetric recording (data not shown). Also, Fura-2-AM has been shown to preferentially enter satellite glial cells that surround the neuronal somata.21To overcome this, we loaded individual somata with Fura-2 via  the impaling microelectrode, which restricted the fluorescent responses to the single recorded neuron (fig. 2D) and has the added benefit of avoiding accumulation of fluorophore in intracellular organelles as may occur with bath-applied Fura-2-AM loading. The fluorophore entered the cell through passive diffusion alone in 76% of neurons and developed maximum fluorescence intensity within the 3 min required to achieve stable electrophysiologic recordings. In 19% of the neurons, fluorophore loading required a further 529 ± 443 s to reach peak fluorescence, during which fluorescence intensity (380 nm) increased by 58 ± 66%. In 5% of neurons, adequate Fura-2 loading required brief pulses of hyperpolarizing or oscillating current to the recording electrode, despite good electrical recording. Iontophoresis has been used by others for dye loading,15whereas the efficacy of electrical oscillation that we observed is unexplained. Loading by these techniques produced fluorescence levels comparable with bath loading, indicating comparable final cytoplasmic Fura-2 levels in the two loading methods. As in dissociated neurons, intracellular stimulation produced somatic APs that were clearly recorded in each of the seven control neurons (fig. 2E). Ca²⁺transient amplitude was highest at 20 Hz and then declined slightly with higher stimulation frequencies (P < 0.05, 7% difference between averages at 20 and 100 Hz), as did the decay time constant (fig. 2F). Generally, Ca²⁺transients in intact ganglia had higher amplitudes than in dissociated neurons (P = 0.006), comparable decay time constants (P = 0.24), and lower variability (normalized variance of decay time constant of each neuron across the four frequencies was 1 ± 0%, P < 0.05). These effects may be due to the influence of dissociation or alternatively to the differences in bath conditions (temperature and composition).

To examine cytoplasmic Ca²⁺signaling under the more physiologic conditions, we altered the above-described technique by inducing somatic APs through conducted axonal APs initiated by dorsal root stimulation, which was used in preference to spinal nerve stimulation so that neurons from the spinal nerve ligation model could be studied. Although the APs conveyed to the soma by this technique travel retrograde compared with normal sensory conduction, the only reported difference from peripheral nerve stimulation is a somewhat slower CV.22Simultaneous recordings of fluorometric and electrophysiologic events triggered by conducted APs were obtained from 117 neurons. All neurons that generated APs produced fluorometric Ca²⁺transients.

To confirm that transients generated using this simultaneous recording technique possess features expected of AP-induced neuronal activation, we demonstrated that transients were nearly eliminated in Ca²⁺-free bath solution while APs were still present in the simultaneous electrophysiologic traces (fig. 3A), indicating that Ca²⁺entry through voltage-gated Ca²⁺channels is necessary for initiation of the transient. Elimination of APs with lidocaine also abolishes the Ca²⁺transient and thus further demonstrates the dependence of Ca²⁺transients on APs in this model (fig. 3B). Also, failure of AP conduction into the soma, revealed as a reduced electrotonic residue passively conducted from the distant failure site,23was associated with failure to produce a Ca²⁺transient (fig. 3C). Baseline [Ca²⁺]^cwas 152 ± 10 nm (n = 57), with no differences between neuronal categories. Transients recorded after retraction of the recording electrode did not differ from immediately preceding recordings during impalement (P = 0.68 and 0.98, n = 12), confirming that the Ca²⁺signal is not altered by the presence of the recording electrode. Potential buffering of [Ca²⁺]^cby Fura-2 produced no effects on electrophysiologic parameters (table 1) compared with our previous findings.24 

Axonal stimulation and somatic recording of membrane potential allowed reliable generation of a single AP, accompanied by a Ca²⁺transient with a fast upstroke and rapid monoexponential recovery (fig. 4A). Four repeated single AP stimulations produced consistent transients in C-type neurons (average difference from the mean for amplitude 6.6 ± 0.6%, n = 39; for decay time constant 7.5 ± 1.0%, n = 29; fig. 4B) and for Aδ-type neurons (amplitude 5.8 ± 1.1%, decay time constant 10.4 ± 0.8%, n = 4). The size of transients generated in response to single APs tended to be much larger for C-type neurons than faster conducting neurons (figs. 4C and D) and could not be discerned in approximately 50% of Aβ-type neurons.

Activity of sensory neuron units naturally occurs in trains of many APs,25which are feasible to replicate by our technique. Using only data from trains in which an AP accompanied each stimulation (for this and all following data), stability of repeated transients generated by AP trains was tested by comparing four identical trains, which demonstrated stability in Aβ-type neurons (average difference from the mean for amplitude 5.5 ± 0.9%, n = 27; decay time constant 4.5 ± 1.2%, n = 7) and in C-type neurons (amplitude 5.2 ± 1.8%, n = 5; decay time constant 4.0 ± 1.7%, n = 5), regardless of frequency and number of APs (figs. 5A and B).

Transients produced by AP trains were composed of summated responses to individual APs that could be discerned at slow firing frequencies (fig. 5C). When the unitary transient triggered by a single AP occurs before the resolution of a prior one, the cumulative stacking of [Ca²⁺]^cincrements produces a higher amplitude [Ca²⁺]^csignal. This phenomenon occurred at lower stimulation frequencies in C-type neurons (usually starting at 0.3–1 Hz) than in fast conduction neurons (starting at 3–7 Hz). Further increases in frequency lessen the time available for each increment to dissipate before the next, so total transient amplitude for a burst of the same number of APs progressively increases (figs. 6A and B). We noted that an apparent maximum amplitude is reached at 50 Hz for Aβ neurons (fig. 6B). We hypothesized that once the interspike interval is shortened to the time it takes for the unitary single-AP transient to reach its peak, further increase in frequency would not produce greater stacking. Therefore, we characterized the performance of individual neurons at different frequencies. Transient amplitude continued increasing with greater AP frequency stimuli until AP conduction failed in six of the 11 C-type neurons. In the other C-type neurons and in A-type neurons (fig. 6C), a frequency was reached above which the transient amplitude did not further increase, giving each neuron a particular optimal frequency that differed on average by neuron type (medians 50 Hz for Aβ, 20 Hz for Aδ, 7 Hz for C-type, Kruskal–Wallis P < 0.01). Individual dissociated neurons stimulated by the impaling electrode (fig. 2B), which could not be characterized by CV, similarly showed transient amplitude maxima (20 or 50 Hz), which indicates that this tuning effect is inherent in the neuronal soma rather than deriving from a particular aspect of excitation by conducted APs or from the influences of the adjacent satellite glial cells.

Recovery of the transient after trains of 21 AP transients showed a monoexponential pattern in most neurons (C-type: 18/22, 82%; Aβ-type: 9/10, 90%) for which the decay time constant was independent of stimulation frequency (P = 0.14 for C-type neurons; P = 0.13 for Aβ-type neurons). On average, the size of transients generated in response to AP trains was larger in C-type neurons (fig. 6A) than for rapidly conducting neurons (fig. 6B).

Increasing the number of APs invading the soma should increase the total Ca²⁺load. Therefore, we subjected neurons to trains of increasing numbers of APs at fixed stimulation frequencies. C-type neurons generally could not generate more than 30 APs at stimulation frequencies that produced stacking of unitary transients, so no data are provided. For Aβ-type neurons, increasing the number of APs from 21–400 stimuli (50 Hz) showed a logarithmic increase in transient amplitude with AP number (data not shown; n = 8). This finding, together with the development of a sustained plateau during the resolving phase of the transient in all eight neurons subjected to 200 or 400 APs, suggests Ca²⁺sequestration into intracellular compartments at higher [Ca²⁺]^c.26 

Electrophysiologic changes induced by axotomy (table 1) duplicate the findings of our prior detailed investigation, including depolarization of the resting membrane potential, prolonged duration of APs, and diminished after hyperpolarization. Also, as previously observed, fluorometric recordings showed decreased resting [Ca²⁺]^clevels in axotomized neurons (127 ± 9 nm vs.  152 ± 10 nm, P = 0.03).

Injured C-type neurons showed substantial decrement in amplitude of transients evoked by single APs (fig. 4D) as well as by AP trains at different frequencies (fig. 6A), corroborating findings from small dissociated neurons.9Axotomized C-type neurons also displayed faster time constants for decay of transients evoked by single APs (fig. 4D) and in trains of APs (fig. 6A). Aδ-type neurons proved to be a heterogeneous group with highly variable transients that did not differ between control and injury groups, either with single AP stimulation (fig. 4D) or trains (data not shown). Similarly, Aβ-type neurons showed no significant differences in transients evoked by single APs (amplitude P = 0.31, decay constant P = 0.5; fig. 4D) or trains of APs (amplitude P = 0.49, decay constant P = 0.12; fig. 6B).

The validity of cellular research depends on using appropriate experimental conditions and probes. Various means of initiating activity in neurons are used, such as K⁺application and agonist receptor binding, but exposure of cultured neurons to electrical fields has valuable attributes, including ease of use, simultaneous exposure of multiple neurons, and avoiding breech of the plasmalemma, which provides relative stability of the preparation. Dissociation of the neurons also assures that agents applied in the bath reach the neuronal membrane at a precisely known concentration. However, our findings demonstrate important disadvantages to field stimulation as a means of neuronal activation, including altered pH in the bath and uncertainty about generation of neuronal activation. In addition, stimulation is unreliable, particularly at higher rates at which neurons are able to initiate APs during direct stimulation. In response, we have adapted a technique that allows recording of cytoplasmic Ca²⁺signals in intact neuronal tissue to the specific context of sensory neurons, providing useful particular features. Specifically, the addition of simultaneous electrophysiologic recording during the measurement of Ca²⁺signals allows precise regulation of activity rates and patterns as well as confirmation of successful stimulation, so that comparisons between experimental conditions may be standardized. Further, neuronal health is readily verified so that data are not obtained from depolarized or nonresponsive neurons. We also incorporate the use of intact tissue, which preserves the natural functional unit of neuronal somata with their surrounding satellite glial cells.6Intact tissue in addition allows natural excitation of sensory neurons by AP transmission via  the axons, thereby permitting the presentation of normal patterns of pulses. The full range of stimulation frequencies may be modeled to include natural maximal instantaneous firing frequencies for Aβ (up to 300 Hz), Aδ (up to 200 Hz), and C-type sensory neurons (up to 120 Hz).18,27–29 

It is important to note that field stimulation has been used successfully for the study of neurons dissociated from fetal or neonatal tissue and thereafter cultured for several weeks, which results in growth of neurites.30–32Both these features may make the neurons more readily responsive to field stimulation, especially when the field is directly applied to the neurites.30,31However, adult tissue is often needed for examination of disease models, such as effects of injury, and prolonged culture may result in loss of pathologic phenotypes. Acutely dissociated DRG neurons, which lack neurites, require voltages that range from 20 to 110 V4,9,26,33to generate activity.

We must also specify limitations to the simultaneous recording approach. Only one neuron may be studied at a time, reducing experimental throughput. Neuronal impalement is invasive and damages the plasmalemma, although most neurons are stable for at least 30 min. Dialysis of electrolytes from the electrode may produce membrane potential shifts. To limit this effect, we used high-resistance electrodes, self-filling glass tubing stock, a modern Brown–Flaming puller, and an electrode holder with a pressure relief channel, each of which limits entry of electrolytes into the neuron.34–36The use of acetate as the anion rather than Cl⁻minimized depolarization because of dialysis.37Another concern with the intracellular electrode technique is potential dialysis of small cellular metabolites, although our observation of stable membrane potential and sustained cell viability suggests an acceptable degree of this effect. The ability of bath-applied agents to penetrate into the DRG may be delayed relative to the action on a dissociated neuron, and concentrations of agents at the neuronal membrane cannot be known exactly. Finally, functional properties of the neuronal soma may only imperfectly reflect properties of membrane in other segments of the neuron.

A number of novel findings have emerged from our initial investigations with this improved recording method. The resting Ca²⁺level of 153 ± 11 nm is higher than the levels (70–136 nm) that we and others have previously observed in dissociated neurons.8,38–40This higher level may result from the more natural conditions of the neurons, including lack of dissociation and examination at body temperature rather than room temperature. The Ca²⁺transient evoked by a single AP has previously been recorded in sensory neurons,20,41but these previous demonstrations used dissociated neurons. Our measurement of [Ca²⁺]^cin nondissociated DRG neurons allows categorization of neurons by CV. These new observations show much larger increases in [Ca²⁺]^celicited by single conducted APs in intact DRGs and a markedly greater Ca²⁺signal in the probably nociceptive C-type neurons compared with the rapidly conducting A-type neurons. This difference may be attributed to the relatively prolonged AP duration of C-type neurons, which accentuates Ca²⁺influx.42The [Ca²⁺]^clevels achieved by single AP activation of C-type neurons reach the range that modulate important Ca²⁺-regulated processes, such as mitochondrial Ca²⁺uptake, kinase activation, and short-term enhancement of synaptic transmission.30,38,43 

We found that short trains (21 APs) generally produced Ca²⁺transients with a high amplitude and a monoexponential recovery, whereas a discrete plateau occurred only after 100 or more APs. These findings contrast with observations from K⁺-induced transients in which prolonged plateaus indicative of mitochondrial Ca²⁺release are expected.9Under more physiologic Ca²⁺loading conditions, such as we observed during natural AP excitation, shaping of the Ca²⁺signal by intracellular Ca²⁺sequestration into mitochondria and endoplasmic reticulum would be limited to slowly adapting sensory units exposed to sustained stimuli.

Our recordings show that AP-induced Ca²⁺transients are larger in C-type neurons than in Aβ-type neurons, which parallels our previous finding of greater Ca²⁺transients in large, capsaicin-insensitive neurons than in small, capsaicin- sensitive neurons. This may be due to geometrical considerations because slow-conducting neurons with small somata have a greater surface-to-volume ratio than rapidly conducting large neurons. This will result in greater accumulation of Ca²⁺per volume of cytoplasm if current density through the plasmalemma is similar in the somata of different size, as we and others have found10,44,45(although others have noted greater current density in large neurons46). In addition, the longer AP duration of C-type neurons produces a greater duration of voltage-gated Ca²⁺influx during each AP. The disproportionate Ca²⁺signal in small neurons may lead to relatively greater functional plasticity despite their typically slower pulse rates.

Although Ca²⁺transients from trains of APs in C-type neurons accumulate to higher levels than transients produced by single AP stimulation even at frequencies as low as 1 Hz (fig. 2C), this form of dynamic stacking did not occur in Aβ neurons until frequencies reached higher than 7 Hz. Furthermore, maximum transient amplitudes were achieved during trains of stimuli at much lower levels for C-type neurons than for Aβ neurons. From these new observations, it is apparent that each neuron has a dynamic response of the cytoplasmic Ca²⁺signal that is tuned to a distinct AP frequency. Furthermore, this tuning is matched to the typical pace of neuronal activity under natural conditions, specifically transmission of high-frequency activity in rapidly conducting A-fibers and trains of low-frequency activity in slow-conducting C-type neurons.18 

Finally, we observed distinct effects of a nerve injury pain model on Ca²⁺signaling. In addition to decreased resting [Ca²⁺]^clevels in axotomized neurons, transient amplitude is diminished and the transient decay is accelerated after injury, but only in the presumed nociceptive C-type neurons. Accelerated Ca²⁺clearance from injured neurons is also evident in the unitary transients that follow a single AP, which may in part explain the smaller cumulative transient amplitudes during repetitive stimulation, through diminished stacking of shorter duration unitary transients. However, our data also show that injury results in smaller transients after single APs, indicating a contribution of decreased Ca²⁺entry or release from stores. These observations indicate substantial disruption of processes regulating Ca²⁺after injury, possibly accounting for hyperalgesia through increased firing of sensory neurons24because of diminished Ca²⁺activation of K⁺currents.47,48A high degree of reliability is imparted to these findings by the use of our recording system by which we can drive neuronal activity in exact patterns and confirm the induced electrophysiologic activity, while preserving the natural structure of the DRG and avoiding injury-like artifacts of dissociation.7This recording approach may be particularly valuable for examination of experimental questions directly relevant to Ca²⁺transients in the DRG somata, such as the regulation of gene expression by the temporal dynamics of the Ca²⁺transients31and the increased depolarization-induced activity in DRG somata after injury.24,49This technique may also offer expanded research options for examining normal and diseased sensory neuron function related to clinical anesthesia, for instance, in studies considering local anesthetic toxicity, mechanisms of corticosteroid therapy for radiculopathy, and further exploration of nerve injury and other chronic pain models.