Painful Peripheral Nerve Injury Decreases Calcium Current in Axotomized Sensory Neurons

Two hundred forty-five cells were derived from 98 adult male Sprague-Dawley rats weighing 125–150 g provided by a single vendor (Charles River Laboratories Inc., Wilmington, MA), after approval of the Medical College of Wisconsin Institutional Animal Care and Use Committee (Milwaukee, Wisconsin). An additional 67 cells obtained from 21 adult male Sprague-Dawley rats (Taconic Farms, Inc., Hudson, NY) were studied in a separate group. Rats received either ligation and section of the right L5 and L6 spinal nerves (SNL) approximately 5 mm distal to the DRG4or control surgery with skin incision only. After fully recumbent recovery from halothane anesthesia, rats were returned to individual cages under climate- and light-controlled conditions for 10 days before behavioral testing as previously described.9Rats were weighed and placed on a 0.25-in wire grid in clear plastic enclosures and allowed to rest for 30 min. Five applications of a 22-gauge spinal needle were made to the plantar skin of each paw and repeated 3 min later, using a force adequate to indent the skin but not to puncture it. These mechanical stimuli produced either a normal brief flinch, or a hyperalgesic-type response characterized by sustained (> 2 s) paw lifting, shaking, and licking. Only SNL rats with an ipsilateral response rate of at least 20% averaged over 3 test days and normal contralateral responses were used for study. The validity of this approach to behavioral testing is supported by the observation that control rats never exhibited characteristic hyperalgesic responses.

After a postsurgical interval of 15.8 ± 1.4 days, the L4 and L5 ganglia were removed after decapitation during halothane anesthesia, placed into separate 35-mm Petri dishes containing Ca²⁺and Mg²⁺-free iced Hanks’ Balanced Salt Solution, and minced with iris scissors. Using separate sterile, silicone-coated (Sigmacote; Sigma, St. Louis, MO) Pasteur pipettes, each ganglion was placed into a 25-ml sterile tissue culture flask containing 0.0625% trypsin (Boehringer-Mannheim, Indianapolis, IN), 0.0125% DNase (Sigma), and 0.01% blendzyme 2 (Roche Molecular Biochemicals, Indianapolis, IN) in DMEM/F12 with glutaMAX (Invitrogen Corporation, Carlsbad, CA) for 1.5 h in a tissue shaking bath at 32°C and 70 rpm. Neurons were resuspended in adult neural basal media (1×; Invitrogen Corporation, Carlsbad, CA) containing 2% (vol:vol) B27 supplement (50×; Invitrogen Corporation), 0.5 mm glutamine, 0.02 mg/ml gentamicin, and 100 ng/ml nerve growth factor 7S (Alomone Labs, Ltd., Jerusalem, Israel). Dissociated L4 and L5 cells from each tube were plated onto separate poly-L-lysine (70–150 kd)–coated coverslips and placed in a 95:5 oxygen–carbon dioxide, water-jacketed incubator and allowed to plate for 2 h before study. All cells were studied within 8 h after plating.

Current clamp solutions duplicated natural cytosolic and extracellular conditions. A modified Tyrode solution was used externally in the bath during current clamp protocol, consisting of the following: 140 mm NaCl, 4 mm KCl, 2 mm CaCl, 2 mm MgCl, 10 mm D-glucose, and 10 mm HEPES at a pH of 7.4 with NaOH and an osmolarity of 300 mOsm. Internal pipette solution contained 120 mm KCl, 5 mm na-ATP, 0.4 mm na-GTP, 5 mm EGTA, 2.25 mm CaCl², 5 mm MgCl², and 20 mm HEPES at a pH of 7.4 with KOH and an osmolarity of 296 mOsm. Because internal pipette solutions cannot be changed after seal formation, I^Cawas isolated by external blockade of other voltage-sensitive currents. The following solution provided adequate separation from other currents: 2 mm BaCl², 1 mm 4-aminophyridine, 10 mm HEPES, 160 mm tetraethylammonium chloride, and 0.1 mm tetrodotoxin at a pH of 7.4 with tetraethylammonium hydroxide and an osmolarity of 300 mOsm. Tetrodotoxin blocked tetrodotoxin-sensitive I^Na, and 4-aminophyridine, tetraethylammonium chloride, and Ba²⁺blocked potassium currents. Because the change in external solutions results in a +4 mV shift in junction potential, no equivalent offset was added after switching from current clamp to voltage clamp. Cadmium (200 μm) was added at the end of some studies to verify the efficacy of I^Caisolation. In a separate protocol examining AP parameters and whole cell currents before and after I^Cablockade, Ca²⁺channel antagonists were applied directly to the cell through a pressure regulated microperfusion system (ALA-BP8 system; ALA Scientific Instruments, Westbury, NY). Complete ablation of high voltage–activated Ca⁺channels was achieved with a cocktail of antagonists that contained nisoldipine (200 nm), SNX-111 (200 nm), agatoxin IVA (200 nm), and SNX-482 (200 nm) to block L-, N-, P/Q-, and R-type calcium channel subtypes, respectively. Tetrodotoxin was obtained from Alomone Labs. SNX-111 was the kind gift of Dr. Scott Bowersox, Ph.D. (Department of Pharmacology, Neurex Corporation, Menlo Park, CA), agatoxin IVA was a gift of Dr. Nicholas Saccomano, Ph.D. (Assistant Director, Medical Chemistry Research at Pfizer, Groton, CT), and SNX-482 was purchased from Peptides International (Louisville, KY).

All cells were recorded in current clamp before changing to voltage clamp. Cells with resting membrane potentials more depolarized than −45 mV were excluded from study. Efficacy of potassium and sodium channel blockade was confirmed by observation of a reversal of negative membrane potential while still in current clamp and a lack of brief, inward currents during voltage clamp recording. Terminal application of Cd²⁺to block I^Cashowed no residual voltage-activated currents. Membrane capacitance was measured and access resistance compensated from 80 to 95%. Cells with greater than 10 MΩ access resistance were rejected. Most recordings were made on a List EP-7 amplifier (ALA Scientific Instruments) modified by the addition of a 150-kΩ resistor to the current injection circuit to increase maximum stimulus to 5 nA. An A/D converter (Axon 1200B; Axon Instruments, Union City, CA) attached to a personal computer running pClamp 8 (Axon Instruments) was used to record data. A second series of experiments on large and medium cells with I^Caof up to 72.5 nA were performed on an Axon 200B amplifier (Axon Instruments) with a different A/D converter (Axon 1320; Axon Instruments).

Initial observation of resting membrane potential determined cell viability before evoked potentials. To determine AP characteristics, a 1-s step protocol was used with current sufficient (between 10 pA and 2 nA) to evoke an AP (fig. 1). AP voltage threshold (AP^thresh) was defined as the level of the inflection at the beginning of the rapid upstroke,13whereas AP duration was measured at 50% of peak (AP⁵⁰). Inflection on the repolarization phase of the AP was observed, and the number of APs during 1 s current injection was determined to characterize the neuron as either adapting or repetitive firing. Continuous current injection and positive feedback from patch clamp head stage amplification prolongs AP duration.14Nevertheless, the relative effect of injury is still detectable. To avoid distortion from continuous current injection during measurement of the afterhyperpolarization (AHP), additional APs were evoked by a brief, 2-ms, 5-nA depolarizing pulse, and peak hyperpolarization (AHP^ampl) and duration at 50% amplitude (AHP⁵⁰) were recorded, as well as the time constant for resolution of the AHP (AHP^τ) by fitting the recovery phase to a single exponential (fig. 1).

All cells were exposed to both a sustained square wave protocol and a standardized AP waveform voltage command protocol. The square wave protocol consisted of 200-ms commands from a holding potential of −90 to +50 mV in 10-mV increments with 5-s intervals between steps. Measured inward current was normalized by membrane capacitance, which results in a current density corrected for cell size. Peak inward current was determined, and linear leak was subtracted post hoc  by fitting the first three steps before significant I^Caactivation to a linear function. Maximum conductance (G^max) was determined by fitting peak inward current to a Boltzmann function, as previously described.15Some large cells produced currents in excess of the bandwidth of the List EP-7 amplifier. Therefore, maximal I^Cafor large cells was calculated by linear fit of current-voltage peaks from the three traces before and at the reversal potential. G^maxdetermined by this procedure differed from that determined by Boltzmann fit by ±3% in 12 nonsaturating cells.

A standardized AP waveform command was used to reveal the effects of naturally rapid depolarization and repolarization on I^Caacross all cells. The basic features of the standardized wave were an inflection on the repolarization phase and a prolonged AHP to mimic standard features of a nociceptor (fig. 1). A modified P8 wave was used to subtract capacitance; namely, the AP waveform was inverted and divided by 8 to acquire membrane capacitance with minimal channel activation. Each P8 wave was subsequently multiplied by −8 and subtracted from the current in response to an AP waveform voltage command. Peak inward current and total charge transfer (current integrated over time) were determined by cursor measurements after current traces crossed 0 mV at the beginning and end of a wave (fig. 1). Charge transfer (coulombs) was normalized to cell area by converting cell capacitance to the specific capacitance of plasma membranes (0.01 pF/μm²) to compare Ca²⁺flux between cells of different sizes. Because cells reliably produced a triangular current waveform (fig. 1), peak and area for large cells with saturating currents were calculated by linear extrapolation of the nonsaturating portion of the trace. The reliability of this method was tested on artificially truncated traces of 12 nonsaturating cells, with results varying ±3% from actual measured values.

Cells were categorized according to cell size and surgical preparation. As noted elsewhere,16,17cell size correlates roughly to neuronal types, so we classified cells as small (C type, < 30 μm), medium (Aδ type, 30–40 μm), or large (Aα/Aβ type, > 40 μm). Neurons were further grouped according to surgical preparation (control or SNL) and DRG level (L4 or L5). Because there were no differences in responses from L4 and L5 ganglia from control rats, these cells were combined, resulting in a control group, an SNL L4 group, and an SNL L5 group. Recordings were evaluated post hoc  using Clampfit 8 (Axon Instruments), and data were summarized in Excel (Microsoft Co., Redmond, WA) and analyzed with Statistica 6 (StatSoft, Tulsa, OK) for analysis of variance and paired t  tests or GraphPad Prism 4 for Mac (GraphPad Software, San Diego, CA) for correlations. Values are expressed as mean ± SEM. For each dependent variable, the main effect of a group was tested with standard univariate analysis of variance, followed by planned Bonferroni post hoc  comparisons between groups. Normal distribution was not assumed for large cells in voltage clamp mode (confirmed by Kolmogorov-Smirnov test for normalcy) because of occasional outliers with very large currents. For this size group, a nonparametric Kruskal-Wallis analysis was used with the Dunn test for post hoc  comparisons. Significance was estimated at P ≤ 0.05 versus  control. Drug effects were evaluated with paired t  tests for dependent samples before and after drug administration. Pearson tests were calculated to determine the coefficient of correlation between G^maxin voltage clamp and AP or AHP parameters in current clamp. Two-tailed P  values and confidence levels were computed to determine whether covariance (r  ²) was different from zero.

Injury was associated with longer action potential duration (AP⁵⁰) in small and medium neurons from L5 ganglia compared with control cells, whereas medium and large cells from the same ganglion were characterized by AHPs with lower amplitudes and shorter durations after injury (table 1and fig. 2). The AP^threshwas significantly reduced only in small cells from the L5 DRG of SNL rats compared with control neurons.

Inflection of the descending limb of the AP identifies nociceptors.18Overall, inflected cells had a longer AP⁵⁰(6.4 ± 0.6 vs.  3.3 ± 0.5 ms, P < 0.05) and depolarized threshold (−15.4 ± 0.8 vs. −23.2 ± 1.2 mV, P < 0.05). Comparing neurons with and without inflection did not show any difference in the response of AP parameters to injury in any surgical group.

Cells from adjacent, noninjured L4 ganglia after SNL were statistically indistinguishable from control neurons in most parameters, except for isolated changes after injury in AHP amplitude in medium neurons and AHP duration in large neurons. Therefore, cells from adjacent, nonligated ganglia did not show the consistent pattern of AP or AHP parameter changes demonstrated in cells from the L5 ganglia of injured rats.

Voltage clamp studies were conducted in the same cells after fluid change. Efficacy of potassium and sodium channel blockade was confirmed by observation of elimination of the negative membrane potential while still in current clamp and lack of brief, inward currents during voltage clamp recording. In addition, terminal application of Cd²⁺to block I^Cashowed no residual voltage-activated currents.

Spinal nerve ligation reduced Ca²⁺conductance in L5 neurons across all size groups compared with neurons from control rats. Specifically, G^maxwas reduced across all three size groups from 36 to 37% in cells from L5 SNL ganglia compared with neurons from control rats. Reduced I^Cawas also reflected in responses to AP waveform voltage commands. Specifically, peak current amplitude and the integral of Ca²⁺flux over the course of the AP waveform (total charge transfer) fell between 37 and 60% in L5 ganglia from SNL rats compared with control. Less consistent decreases were found during voltage clamp examination of L4 neurons (table 1), which showed reduced G^maxonly in large neurons and altered AP waveform responses only in medium and large neurons.

Only L5 neurons showed decreased I^Caafter injury in the small neuron group, whereas there were changes in both L4 and L5 neurons in the medium and large groups. However, this part of the study was not specifically designed to compare L4 and L5 neurons. To directly assess the possibly distinct effects of injury on L4 and L5 neurons, we devised additional studies to optimally compare currents in these two groups. In these experiments, we examined currents in paired L4 and L5 medium and large neurons isolated from the same animal under identical conditions. This focused examination also eliminated inconsistencies that accrue due to data collection over a span of time (2 yr for the data set above), including genetic drift in rat breeding colonies and different batches of dissociation enzymes. Further, we used a different amplifier with expanded bandwidth to clamp whole cell recordings with up to 72.5 nA peak inward current. These experiments identified contrasting injury responses in medium and large L4 and L5 neurons. Specifically, I^Cain L4 neurons from medium cells was unchanged, whereas L5 neurons displayed reduced I^Cacompared with L4 and control neurons (fig. 3). I^Cain large L4 and L5 neurons did not differ from control.

Because reduced I^Camay have diverging effects on the AP depending on whether the direct inward Ca²⁺current or the induced outward I^K(Ca)dominates, we correlated G^maxagainst AP parameters measured during current clamp in the same cells to determine the role of I^Ca. AP⁵⁰correlated negatively with G^maxin small cells from L5 of injured rats (fig. 4A) but positively in large control and SNL L5 neurons (fig. 4B), indicating opposite contributions of I^Cato AP duration in these two neuronal groups. While G^maxdid correlate positively with AP⁵⁰in large L5 neurons after SNL, this correlation was also seen in large cells from control ganglia without difference in means as determined by analysis of variance (table 1). Small control neurons showed minimal dependence of AP repolarization on Ca²⁺-dependent processes (fig. 4A). G^maxcorrelated negatively with AHP^amplin large cells from L5 ganglia of injured rats (fig. 4C). G^maxdid not correlate with any parameters in L4 and medium-sized neurons.

To identify the biophysical consequences of reduced I^Ca, we duplicated the loss of I^Caseen after injury by application of Ca²⁺channel blockers and measured the effect of blockade on AP parameters and whole cell currents in the same cell (table 2). Medium or small neurons (n = 9) from nonoperated rats were studied in normal Tyrode solution before and after complete I²⁺ablation with a cocktail of high voltage–activated Ca²⁺channel antagonists containing nisoldipine (200 nm), SNX-111 (200 nm), agatoxin IVA (200 nm), and SNX-482 (200 nm) to block L-, N-, P/Q-, and R-type calcium channel subtypes, respectively. An AP was elicited with a brief (0.5-ms) pulse ranging from 1 to 3 nA. This same AP was used as a voltage command to elicit whole cell currents in the same cell. After capacity transients, an inward current representing calcium and sodium was followed by an outward current representing potassium (fig. 5). Calcium channel blockade significantly prolonged average AP⁵⁰and decreased AHP^amplduring current clamp recordings (fig. 5A), whereas underlying inward and outward currents were significantly reduced. It is noteworthy to observe the time course of ionic changes (fig. 5B). Reduced inward I^Cacorrelated with the delayed pace of AP depolarization, whereas outward current, presumably I^K(Ca), was reduced during the repolarization phase of the AP and the onset of AHP. This provides evidence of a critical role of I^Cain production of outward currents during AP repolarization in sensory neurons.

In this study, we have observed decreased I^Cain the somata of sensory neurons after peripheral nerve injury by SNL. In combination with our previous findings after more peripheral sciatic injury by chronic constriction,19this indicates that I^Cais a general feature of nerve injury. The chronic constriction injury model of neuropathic pain is limited by the inability to distinguish the contribution of axotomy versus  inflammatory mechanisms, because somata of axotomized neurons cannot be readily distinguished from those with intact axons. Incomplete injury is a prerequisite for provoked pain behavior, so we chose a model of neuropathic pain in which neurons injured by complete axonal transection might be compared with anatomically segregated neighboring neurons. We have found that reduced I^Cais predominantly a consequence of axotomy (L5 after SNL) and is most evident in small and medium neurons.

Various voltage-activated currents, especially N and T types, inactivate during sustained depolarization,20,21so dynamic aspects of voltage commands may substantially affect patterns of recorded I^Ca. Studies of ionic currents underlying neuronal membrane function typically rely on prolonged command pulses to detect voltage sensitivity and kinetic properties of channel gating. However, natural neuronal activity leads to opening of voltage-gated channels in response to only brief membrane depolarization events, under which conditions the specific activation, inactivation, and channel closure after repolarization (deactivation) of various currents critically controls Ca²⁺flux. We therefore used a voltage command in the form of an AP wave to determine whether an injury would produce a loss of I^Caas has been observed with prolonged step protocols. We find that injury reduces I^Cain response to an AP waveform to an even greater extent than I^Cameasured in response to 200-ms step pulses, which extends observations from traditional sustained square wave protocols and confirms that loss of I^Carepresents a true biophysical change associated with neuropathic pain.

We categorized neurons by size, whereas other studies have used AP characteristics, expression of membrane markers, or sensitivity to neuropeptides and algogenic agents. Clearly, sensory neurons show a high degree of heterogeneity, with some investigators describing seven subtypes in small cells alone.22However, nerve injury causes shifts in all these markers,23,24making their use problematic for categorizing neurons. A further limitation is that the exact electrophysiologic changes induced by injury in AP parameters are technique dependent. Specifically, intracellular recordings from nondissociated sensory neurons by us and others24,25contrast with the current patch recordings from dissociated neurons in showing prolonged AP duration after injury in large but not small neurons. To address these limitations, the current study examined Ca²⁺membrane currents together with AP characteristics in the same neuron, thus directly highlighting the role of I^Ca.

The functional implications of I^Caloss are not fully understood. Previous reports have shown that Ca²⁺contributes a large inward current during the repolarization phase of the AP, leading to a prolongation or inflection of the descending limb, whereas removing Ca²⁺shortens AP duration.26–28Accordingly, we have observed a direct relation between AP duration and I^Cain large neurons (fig. 3B).

Among small and medium neurons, however, diminished I^Caprolongs rather than shortens the AP (table 1and fig. 4). Blockade with specific Ca²⁺channel antagonists indicates that AP prolongation in these neurons is controlled by a shift in the balance of currents such that the loss of inward I^Cais overwhelmed by a much greater loss of outward I^K(Ca)(fig. 5). This relation by which decreased I^Caleads to prolonged AP duration is mirrored in the findings of others who have shown that a net outward current can result from Ca²⁺entry in neuronal cell types with a predominant expression of large conductance voltage- and Ca²⁺-gated K⁺channels.29,30Taken together, these results demonstrate the fine balance between ionic conductances and firing patterns which are altered in different ways by trauma. AP prolongation in small and medium neurons after injury may give rise to increased neurotransmitter release at synaptic junctions in the dorsal horn and thereby contribute to hyperalgesia.31 

We have also shown that I^Caloss is associated with AHP shortening in medium and large neurons after injury. Because currents through Ca²⁺-activated K⁺channels contribute substantially to the AHP,32our finding of shortened AHP provides further evidence of decreased outward K⁺current secondary to I^Caloss in these cell groups, although injury and inflammation could also have direct effects on these K⁺channels. Blocking Ca²⁺currents with specific toxins similarly caused a reduction of AHP amplitude and duration through diminished activation of Ca²⁺-dependent outward current. AHP loss primes the neuron for increased repetitive firing,33as we have demonstrated in excised ganglia after injury.24Therefore, the deficiency of I^Cawe have observed in sensory neurons after injury may substantially contribute to increased excitability and production of neuropathic pain. Further demonstration of this mechanism may be the antinociception provided by intrathecal Ca²⁺injection in mice that is eliminated by blockers of calcium-activated potassium currents (I^K(Ca))34and the burning pain in humans after delivery of a solution containing the Ca²⁺buffer EDTA to the DRG by epidural injection.35 

Calcium signaling performs different functions in different neuronal types. Accordingly, our finding of decreased I^Cain sensory neurons after nerve injury in rats with hyperalgesic behavior contrasts with relief of neuropathic pain in animal models by intrathecal administration of Ca²⁺channel blockers.36,37However, neuraxial blockade will affect both primary and secondary sensory neurons and interferes with neurotransmitter release in the spinal cord. Moreover, subarachnoid injection of blockers would not block sensory ganglia because of incomplete penetration into the DRG.38Finally, loss of internal Ca²⁺in the DRG has been shown to accelerate afferent conduction, whereas increased intracellular Ca²⁺reduced AP propogation.39 

We observed the greatest I^Caloss in axotomized L5 neurons after SNL. Although these transected sensory neurons lack a peripheral receptive field, they nonetheless may be an important source of conditioning activity that produces dorsal horn hypersensitivity (fig. 6).8Even in the absence of natural stimulation, axotomized neurons are particularly subject to chemical activation by catecholamines, proinflammatory cytokines, bradykinin and neurotrophins, and mechanical stimulation.40,41Cross-excitation from adjacent intact neurons,42such as those in the dorsal primary ramus of the segmental spinal nerve that remains intact after SNL, also excites the axotomized somata. Our findings indicating membrane hyperexcitability may amplify this induced L5 traffic, leading to hyperalgesic sensory events when naturally triggered L4 activity arrives in the hyperexcitable dorsal horn.

The authors thank Ksenija Modric-Jednacak, M.D. (Resident), Mike Bode, B.A. (Technician), and Mark Poroli, B.S. (Research Technologist), for technical assistance (all from the Department of Anesthesiology, Medical College of Wisconsin, Milwaukee, Wisconsin).