Painful Nerve Injury Decreases Resting Cytosolic Calcium Concentrations in Sensory Neurons of Rats

All procedures were approved by the Animal Care and Use Committee of the Medical College of Wisconsin, Milwaukee, Wisconsin.

Male adult Sprague-Dawley rats (n = 42; Charles River Laboratories Inc., Wilmington, MA) weighing 160–180 g were randomly assigned to an SNL group or control group. SNL (n = 24 rats) was performed similar to the originally reported technique.18During anesthesia with halothane (2–3%) in oxygen, the right lumbar paravertebral region was exposed. After subperiosteal removal of the sixth lumbar transverse process, both the right fifth and the sixth lumbar spinal nerves were tightly ligated with 6-0 silk suture and transected distal to the ligature (fig. 1). No muscle was removed, the intertransverse fascia were incised only at the site of the two ligations, and articular processes were not removed. The lumbar fascia was closed by 4-0 resorbable polyglactin suture, and the skin was closed with three staples. In control rats (n = 18), only lumbar skin incision and closure was performed. After surgery, the rats were returned to the colony, where they were kept in individual cages under normal housing conditions.

Identification of hyperalgesia was conducted as previously described.19At least 1 day after arrival at the animal care facility, rats were brought to the testing area for 4 h of familiarization with handling and the environment. Hind paws were stimulated in random order with a 22-gauge spinal needle applied with pressure adequate to indent but not penetrate the plantar skin 2 days before surgery and on the 10th, 12th, and 14th postoperative days. Control rats showed only a brief withdrawal. SNL animals that displayed a hyperalgesia-type response with sustained lifting, licking, chewing, or shaking of the paw were considered to express a phenotype of neuropathic pain, whereas others without hyperalgesic responses were considered to lack neuropathic pain.

The L4 and L5 DRG were removed from control rats as well as both hyperalgesic and nonhyperalgesic SNL rats (studied separately) after halothane anesthesia and decapitation. The operative field was perfused with cold, oxygenated, calcium and magnesium chloride free Hanks Balanced Salt Solution. Minced ganglia were enzymatically dissociated in a solution containing 0.018% liberase blendzyme 2 (Roche Diagnostics Corp., Indianapolis, IN), 0.05% trypsin (Sigma, St. Louis, MO), and 0.01% deoxyribonuclease 1 (150,000 U; Sigma, St. Louis, MO) in 4.5 ml Dulbecco's modified Eagle's medium F12 (Gibco, Carlsbad, CA) for 90 min in a shaker bath at 32°C. Cells were harvested by centrifugation and resuspended in a culture medium consisting of 0.5 mm glutamine, 0.02 mg/ml gentamicin, 100 ng/ml nerve growth factor 7S (Alomone Labs, Jerusalem, Israel), 2% (vol/vol) B-27 supplement (Life Technologies, Rockville, MD), and 98% (vol/vol) neurobasal medium A 1X (Life Technologies) for plating onto poly-l-lysine-coated 12-mm glass coverslips (Deutsche Spiegelglas; Carolina Biologic Supply, Burlington, NC), plating from two to four slips per ganglion. Cells were incubated for 2–3 h in humidified incubator at 37°C with 95% air and 5% CO²before dye loading and were studied within 5 h of dissociation.

Cells were loaded with the ratiometric calcium indicator Fura-2 AM (2.5 μm in 0.1% Pluronic F-127; Molecular Probes, Eugene, OR) for 45 min at room temperature and then washed three times with a Tyrode's solution consisting of 140 mm NaCl, 4 mm KCl, 2 mm CaCl², 2 mm MgCl², 10 mm glucose, and 10 mm HEPES. Cells were left in a dark environment for 30 min for dye deesterification. Coverslips were mounted in a 500-μl recording chamber superfused with room-temperature (22°C) Tyrode's solution at a gravity-driven flow rate of 2 ml/min and imaged at 400× magnification using an inverted microscope (Nikon Diaphot 200; Tokyo, Japan) and cooled charge-coupled device camera (Cool Snap fx-Photometrics; Tucson, AZ). Cell diameter was determined by calibrated video image. On bright-field examination, neurons were excluded from measurement if they showed evidence of lysis or crenulation of their surface, because these cells showed unstable recordings, and also if they had overlying glial satellite cells (fig. 2). Each neuron was specified as a region of interest in the digital image (MetaFluor; Universal Imaging Corporation, Downington, PA) for separate R measurement, and an additional background area was recorded in each field for on-line subtraction of background fluorescence. Only one field was studied per slip. Autofluorescence of unloaded cells had a signal strength of less than 5% of the fluorescence of loaded cells. Emitted Fura-2 fluorescence was recorded at 510 ± 20 nm wavelength during alternating 340- and 380-nm excitation (DG-4; Sutter, Novato, CA). The frame capture period was 200 ms at intervals of 10 s. The ratio (R) of fluorescence excited by 340 nm divided by fluorescence excited by 380 nm was determined on a pixel-by-pixel basis, and [Ca²⁺]^cwas calculated as

in which K^dis the dissociation constant for Fura-2 (specified as 224 nm),20R^minis the 340/380 fluorescence ratio in the presence of no calcium, R^maxis the ratio in the presence of saturating concentrations of calcium, and β is the ratio of fluorescence during 380-nm excitation at zero and saturating calcium concentrations.

Resting R (R^rest) was measured by averaging 10 stable sequential determinations. In this study, R^min, R^max, and β were determined by in situ  calibration for each cell, rather than applying averaged values as is typically done. Bath application of a calcium-selective ionophore (fig. 3), either ionomycin (5 μm) or 4-bromo A-23187 (5 μm; Molecular Probes), permitted equilibration of [Ca²⁺]^cwith bath calcium concentration during superfusion first with Tyrode's solution with 10 mm EGTA and no added calcium to determine R^minand then with normal Tyrode's external solution to determine R^max.21,22 

Drugs were delivered by bath change, except application of capsaicin, which was directed to the field through a microperfusion system 125 μm upstream from the imaged field. Using dye delivered through this system, onset time was measured as less than 200 ms, and there was no dilution of microperfusion solution despite bath flow.

Neurons were divided into two groups with diameters larger or smaller than 34 μm. This division is well established in similar studies of [Ca²⁺]^cin dissociated DRG neurons.23,24Although the size distribution of sensory neurons is a continuum (fig. 4),25grouping large and small neurons partially segregates the overlapping nociceptive and nonnociceptive neuron categories.26–28The chosen diameter for separating the large and small populations is derived from histologic studies.29,30 

In separate experiments, neurons were categorized as sensitive to capsaicin if R increased by 50% over R^rest. Results are presented as mean ± SEM, and n  refers to the number of cells tested unless otherwise stated. For [Ca²⁺]^cmeasurements, main effects were determined by two-way analysis of variance, and post hoc  assessment of within-group comparisons was performed conservatively using the Bonferroni test (Statistica 6.0; StatSoft, Tulsa, OK). Differences between two groups were considered to be significant when P  was less than 0.05. Significance of injury effects on frequency of capsaicin response was tested by nonparametric cross-tabulation. Because there is no established test for post hoc  comparisons within contingency tables, values for Fisher exact test are reported. Concentration-response analysis was performed using Prism 4.0 (GraphPad Software, Inc., San Diego, CA).

Of 40 rats subjected to SNL, 24 in which hyperalgesia developed were considered successful neuropathic models, and DRG were removed 19.1 ± 0.5 days after injury. In these rats, hyperalgesia-type responses occurred in 41.4 ± 4.8% of needle applications to the right foot. Post mortem examination confirmed accurate placement of ligatures and section for all SNL animals. Hyperalgesic behavior did not develop in control rats (n = 17; 0 ± 0% hyperalgesic response rate; P < 0.001 vs.  SNL), and DRG were removed 18.9 ± 1.6 days after skin incision surgery. Four rats in which hyperalgesia did not develop (1.7 ± 1.7% hyperalgesic response rate; P < 0.01 vs.  SNL) were also studied at 19.5 ± 0.9 days after injury.

Stable R^restrecordings were achieved after 5–10 min of imaging for neurons dissociated from 65 DRGs (24 L4 and 24 L5 after SNL; 17 L4 or L5 control), including control cells (n = 343), SNL L4 cells (n = 252), and SNL L5 cells (n = 225). To determine whether extreme values of R^min, R^max, or β should be used as exclusion criteria in addition to visual features, we plotted [Ca²⁺]^cacross the range of these parameters (data not shown). Because there were no outliers of [Ca²⁺]^cwith the highest or lowest values of R^min, R^max, or β, we did not exclude any cells on this basis.

We noted that R continued to decrease during application of ionophore/low bath calcium for as long as 25 min (fig. 3). The descending trace of R during this interval fit an exponential decay curve (r  ²= 0.98 ± 0.06), so that an extrapolated R^mincould be determined (Origin 7.0; Origin Lab Corp., Northampton, MA), which was lower than recorded R^minby an average of 0.10 ± 0.01 and resulted in an average increase of [Ca²⁺]^cby 26.4 ± 1.6 nm. We compared calibration using the ionophores 4-bromo A-23187 (n = 165) or ionomycin (n = 264). Because [Ca²⁺]^cdecreased more slowly during R^mindetermination and cells more frequently detached from the coverslip during R^maxdetermination using 4-bromo A-23187, further studies used only ionomycin as an ionophore during calibration. This choice is also supported by findings indicating that ionomycin has a higher selectivity for calcium and is more effective than 4-bromo A-23187 in binding and transporting calcium.21[Ca²⁺]^cvalues determined using 4-bromo A-23187 (n = 70) were higher in control animals than those (n = 122) determined using ionomycin, but the effects of injury on [Ca²⁺]^cwere comparable (n = 164; data not shown). For ionomycin, average values of extrapolated R^min, measured R^max, and β were 0.53 ± 0.01, 5.19 ± 0.10, and 5.81 ± 0.13, respectively. Comparison of [Ca²⁺]^ccalculated with either cell-by-cell calibration or mean values for R^min, R^max, and β (r  ²= 0.22; fig. 5) demonstrates the importance of determination of parameters for each cell. The absolute value of the difference between these two determinations is 58.7 ± 2.4 nm (overall for all cells), or 92 ± 6% of the fully calibrated [Ca²⁺]^c.

Two-way analysis of variance showed significant main effects for both injury and neuronal size on [Ca²⁺]^c, as well as a significant interaction of injury and size. Both large- and small-neuron groups showed significantly lower resting [Ca²⁺]^cin axotomized L5 neurons after SNL compared with control (fig. 6), but large neurons showed a proportionately greater decrease to 54% of control. Large neurons also showed a decrease of [Ca²⁺]^cin SNL L4 neurons to 29% of control.

To probe whether there is a relation between the decreased [Ca²⁺]^cand altered sensory behavior after injury, we examined a small number of animals without behavior change. Unlike neurons from hyperalgesic rats, large neurons removed from these four rats did not show a decrease in resting [Ca²⁺]^c(182 ± 25 nm, n = 13 for L4; 146 ± 35 nm, n = 5 for L5) compared with neurons from control rats (133 ± 11 nm, n = 25). This was also true for small neurons from such rats, which showed significantly higher resting [Ca²⁺]^cin injured neurons (160 ± 15 nm, n = 22, P < 0.001 for L4; 129 ± 10 nm, n = 31, P < 0.001 for L5) compared with control (78 ± 4 nm, n = 96).

Neurons with nociceptive and nonnociceptive modalities have broadly overlapping size distributions.15,16Because previous studies have established capsaicin sensitivity as a hallmark of nociceptors,17we categorized additional neurons according to their response to capsaicin, using a 50% increase of R to indicate sensitivity. A concentration-dependent response to capsaicin was evident between1 nm and 10 mm in uninjured neurons for both frequency of response at each dose and the response amplitude (fig. 7). A concentration of 100 nm resulted in an intermediate response rate comparable to the expected frequency of nociceptors among dissociated neurons and was used for further testing. For all control neurons, 63 of 130 (48%) responded to 100 nm capsaicin, with a smaller proportion in large neurons (5 of 20, 25%) than in small neurons (58 of 110, 53%; P < 0.05). SNL decreased the overall frequency of response to capsaicin in L5 neurons (7 of 57, 12%) but increased the incidence in L4 neurons (51 of 77, 66%) compared with both control and SNL L5 (fig. 8). This influence of injury on capsaicin sensitivity was evident among both large and small neurons (data not shown).

Two-way analysis of variance showed significant main effects for both injury and capsaicin sensitivity on resting [Ca²⁺]^c. Separate analysis of insensitive neurons (fig. 9) showed that resting [Ca²⁺]^cfor SNL L5 neurons was 60% lower than in control neurons and was also lower than in SNL L4 neurons. For capsaicin-sensitive neurons (fig. 9), the influence of injury did not reach significance (P = 0.077), due in part to the small number of SNL L5 neurons that were sensitive to capsaicin.

Although calcium is the principal signaling pathway controlling cellular processes of sensory neurons, there has been only minimal examination of effects of peripheral nerve injury on [Ca²⁺]^c. In this study, we demonstrate that axonal injury depresses resting calcium concentrations. This decrease is greater in directly axotomized neurons than in adjacent neurons exposed to degenerating axonal segments. The influence of injury is not uniform across neuronal types but is dominant in units presumed to have a nonnociceptive modality, as indicated by large somatic diameter and insensitivity to capsaicin.

We are not aware of other reports of resting [Ca²⁺]^cin painful neuropathic conditions, except for experimental diabetes mellitus. Although some studies of streptozotocin-induced diabetes have found increased resting [Ca²⁺]^c,23,24the relevance to traumatic or inflammatory neuropathy is uncertain. For example, diabetes is associated with increased voltage-activated calcium currents in the DRG neuronal membrane,31whereas we have determined that traumatic neuropathy decreases these currents.12,13Also, there are contrasting reports that show unchanged sensory neuronal resting [Ca²⁺]^cin streptozotocin-induced diabetic rats32–34and in a spontaneously diabetic rat strain.35 

Our intent in examining capsaicin response was to categorize cells according to sensory modality.17We also confirmed an EC⁵⁰for [Ca²⁺]^cresponse to capsaicin in the 55–72 nm range,36,37which is much less than the 350–728 nm necessary to produce a membrane current.38–41Sensory neurons express the capsaicin receptor TRPV1 on endoplasmic reticulum,42so release of calcium stored in endoplasmic reticulum43may explain the responses to low capsaicin concentrations we observed. Supporting our belief that the [Ca²⁺]^cresponse to capsaicin is a valid indicator of nociceptive modality, we noted a higher response rate in small cells, which tend to be nociceptors. Further, the response rate was substantially decreased in the axotomized neurons of the L5 DRG after SNL but increased in L4, consistent with immunohistochemical observations of TRPV1 expression.44,45A generally higher resting [Ca²⁺]^cin control cells tested with capsaicin compared with those categorized only by size may be due to an unexplained effect of capsaicin on ionophore calibration. Nonetheless, injury effects between groups persisted in cells treated this way.

Determination of resting [Ca²⁺]^cis highly sensitive to technical details. The use of a ratiometric indicator such as Fura-2 largely corrects for variations in dye loading, bleaching, or differences in fluorescent signal due to cell thickness. We have noted substantial variability between neurons in R^min, R^max, and β. This requires the determination of calibration parameters for each neuron if [Ca²⁺]^cis to be compared between cells, although averaged calibration parameters may be adequate if repeated measures for a given cell are compared (e.g. , response to an agonist).

Technical differences may in part explain the wide range of resting [Ca²⁺]^creported for sensory neurons that spans from 60 to 207 nm.24,36,46–48Further factors may influence measured [Ca²⁺]^c, however, including choice of K^d, age of the animal,47and variability in neuronal size and sensory modality of the sample population, as we have shown in the current study. Our finding of higher resting [Ca²⁺]^cin large-diameter sensory neurons than in small neurons contrasts with previous findings in mice32,49but is in accord with findings in rats.24 

Cytosolic calcium is regulated by the balanced actions of various membrane processes that acquire, store, and expel calcium. Although calcium entry through voltage-gated calcium channels does not influence [Ca²⁺]^cof the inactive neuron,50influx of calcium through voltage-independent channels regulated by calcium stores (capacitative calcium entry) is active at rest and sensitive to resting membrane potential.50,51Uptake into endoplasmic reticulum via  the sarcoplasmic or endoplasmic reticulum calcium adenosine triphosphatase pump and expulsion of calcium from the cell by the plasma membrane calcium adenosine triphosphatase pump both regulate resting [Ca²⁺]^c,52and mitochondria also sequester calcium in the inactive neuron.53In most sensory neurons, the plasmalemmal Na⁺–Ca²⁺exchanger is constitutively active such that removal of extracellular Na⁺increases resting [Ca²⁺]^c.54The decrease in resting [Ca²⁺]^cafter peripheral nerve injury may result from a disturbance in any of these regulatory processes, but these have not been examined.

Withdrawal of target-derived neurotrophins may be an upstream trigger for depression of [Ca²⁺]^c, because nerve growth factor,55brain-derived neurotrophic factor, and neurotrophin-356support resting [Ca²⁺]^c. However, nerve growth factor in the ligated L5 DRG recovers after 2 days,57and brain-derived neurotrophic factor expression is increased after nerve injury.58,59An alternate cause of depressed resting [Ca²⁺]^cmay be decreased neuronal firing, because activation of neurons increases resting [Ca²⁺]^c.60,61Decreased afferent traffic due to axotomy or disuse of the limb combined with injury-induced depression of voltage-activated calcium currents12,13may decrease the cytoplasmic calcium load. Although spontaneous activity develops in injured DRG neurons,8,62this is typically at very low rates.63,64The normal or higher levels of resting [Ca²⁺]^cin neurons from SNL rats that lacked hyperalgesia might reflect particularly high levels of spontaneous activity in these subjects. Differences we noted between putative nociceptive and low-threshold neurons may likewise be due to differences in neuronal activity between these modality categories.

A role for decreased resting [Ca²⁺]^cin generation of neuropathic pain is supported by our observation of a normal or increased resting [Ca²⁺]^cin a small number of rats lacking hyperalgesia after SNL. Although the functional consequence of dynamic [Ca²⁺]^ctransients is well established, the role of resting [Ca²⁺]^chas been less studied. In various neuronal tissues, decreased [Ca²⁺]^cprecipitates cell loss, including programed cell death by apoptosis.60,61,65DRG neuronal loss has been noted as a feature of neuropathy after SNL,66,67and induced cell activity may prevent cell loss after axotomy in the central nervous system, presumably by increasing resting [Ca²⁺]^c.68Therefore, the decrease in resting [Ca²⁺]^cwe have observed may directly lead to loss of sensory neurons after injury.

A second role of resting [Ca²⁺]^cis control of responsiveness to ligand stimulation. Increased resting [Ca²⁺]^cattenuates receptor-triggered calcium signaling in lymphocytes69and central nervous system microglia.70In DRG neurons, increased [Ca²⁺]^cinactivates responses to capsaicin.37,38,71In contrast, responsiveness to heat is potentiated when [Ca²⁺]^cis increased by capsaicin or a calcium ionophore,72but this is not found after [Ca²⁺]^cincrease by membrane depolarization.73Depressed resting [Ca²⁺]^c, as we have measured in sensory neurons after injury, may thus modulate the transduction of membrane receptor activation into intracellular calcium signals for a broad range of ligands and stimuli, including catecholamines, purines, pH, and inflammatory mediators such as bradykinin, complement, and cytokines. Altered expression of receptors may have a competing influence, however, as in the injury-induced changes in capsaicin responsiveness measured in this study. Low resting [Ca²⁺]^cmay itself produce complex genetic effects, increasing expression of inducible nitric oxide synthase in chondrocytes while depressing synthesis of cyclooxygenase II,74and increasing the expression of sodium channels in the surface membrane of adrenal chromaffin cells.75 

Finally, enzymatic signaling cascades sensitive to calcium, such as calmodulin, phosphatases, and protein kinases A and C, modulate neuronal activity. Through regulation of calcium-activated K⁺channels76and hyperpolarization-activated cation channels,77resting [Ca²⁺]^ccontrols membrane excitability and neuronal activity. Calcium/calmodulin-dependent protein kinase II has particularly diverse phosphorylation targets and may be an important pathway generating functional and genetic changes from shifts in resting [Ca²⁺]^c.78 

Our findings indicate that nerve injury associated with hyperalgesia depresses resting [Ca²⁺]^cin sensory neurons. This change is most evident in nonnociceptive axotomized neurons of L5 after SNL, which supports a central role of this neuronal group in the genesis of neuropathic pain.9However, these observations do not eliminate the possibly critical contributions of functional changes in calcium signaling of neurons that remain in continuity to their receptive fields, since we also identified decreased resting [Ca²⁺]^cin large L4 neurons after SNL.