Abstract
Painful nerve injury disrupts levels of cytoplasmic and stored Ca2+ in sensory neurons. Since influx of Ca2+ may occur through store-operated Ca2+ entry (SOCE) as well as voltage- and ligand-activated pathways, we sought confirmation of SOCE in sensory neurons from adult rats and examined whether dysfunction of SOCE is a possible pathogenic mechanism. Dorsal root ganglion neurons displayed a fall in resting cytoplasmic Ca2+ concentration when bath Ca2+ was withdrawn, and a subsequent elevation of cytoplasmic Ca2+ concentration (40 ± 5 nm) when Ca2+ was reintroduced, which was amplified by store depletion with thapsigargin (1 μm), and was significantly reduced by blockers of SOCE, but was unaffected by antagonists of voltage-gated membrane Ca2+ channels. We identified the underlying inwardly rectifying Ca2+-dependent ICRAC (Ca2+ release activated current), as well as a large thapsigargin-sensitive inward current activated by withdrawal of bath divalent cations, representing SOCE. Molecular components of SOCE, specifically STIM1 and Orai1, were confirmed in sensory neurons at both the transcript and protein levels. Axonal injury by spinal nerve ligation (SNL) elevated SOCE and ICRAC. However, SOCE was comparable in injured and control neurons when stores were maximally depleted by thapsigargin, and STIM1 and Orai1 levels were not altered by SNL, showing that upregulation of SOCE after SNL is driven by store depletion. Blockade of SOCE increased neuronal excitability in control and injured neurons, whereas injured neurons showed particular dependence on SOCE for maintaining levels of cytoplasmic and stored Ca2+, which indicates a compensatory role for SOCE after injury.
Introduction
The concentration of cytoplasmic Ca2+ ([Ca2+]c) is the dominant regulator of numerous neuronal functions, including differentiation, excitation, synaptic transmission, and apoptosis (Ghosh and Greenberg, 1995; Paschen, 2001). Sensory neurons possess an array of plasmalemmal channels that admit Ca2+ in response to depolarization, binding of ligands, heat, cold, depressed pH, and mechanical distortion. Ca2+ signals initiated by these high-conductance channels are modulated by concurrent extrusion of Ca2+ from the neuron, as well as bidirectional exchange of Ca2+ between the neuronal cytoplasm and stores in endoplasmic reticulum (ER) and mitochondria. Inflammation and injury of sensory neurons disrupts this ensemble of interacting processes (Fuchs et al., 2007; Lu and Gold, 2008; Gemes et al., 2009; Rigaud et al., 2009).
There is expanding recognition in diverse cell types of Ca2+ entry through low-conductance plasmalemmal channels that are regulated by the level of Ca2+ stored in the ER, a process known as store-operated Ca2+ entry (SOCE). This pathway and its underlying Ca2+-release-activated current, ICRAC, are well defined in nonexcitable cells (Hofer et al., 1998; Braun et al., 2001; Mercer et al., 2006), for which SOCE is the dominant route of Ca2+ influx. Recent identification of stromal interaction molecule 1 (STIM1) as the ER Ca2+ sensor that regulates SOCE (Stathopulos et al., 2006), and Orai1 as a pore-forming subunit conducting ICRAC (Mercer et al., 2006), has allowed detailed characterization of SOCE in cells expressing these collaborating elements. Such studies have established cardinal features of SOCE, including amplification by depletion of Ca2+ stores, inward rectification of ICRAC, high inward conductance of Na+ through store-operated Ca2+ channels in the absence of divalent cations (DeHaven et al., 2007), and sensitivity of SOCE to certain semiselective blockers (Putney, 2001).
A small number of studies have examined SOCE in neurons. Interference with SOCE may depress [Ca2+]c and deplete sensory neuron intracellular Ca2+ stores (Usachev and Thayer, 1999), which also follows painful nerve injury (Rigaud et al., 2009). Additionally, SOCE may regulate kinase activity and synaptic transmission (Emptage et al., 2001; Cohen and Fields, 2006), providing strong motivation for expanded exploration of SOCE in sensory neurons under baseline and injured conditions. Accordingly, the present study pursues several goals. First, since previous findings that identified the basic effects of SOCE on resting [Ca2+]c and Ca2+ stores in sensory neurons examined neonatal sensory neurons after prolonged culture (Usachev and Thayer, 1999), we sought to extend those findings using acutely isolated adult sensory neurons, to limit effects of prolonged culture (Scott and Edwards, 1980) and to allow comparison in pain models using adult animals. Second, we wanted to determine the fundamental features of SOCE in sensory neurons by directly examining ICRAC and determining whether these neurons express the molecular components of SOCE described in other cell types. Third, since the functional role of SOCE is poorly defined in neurons, we examined the influence of SOCE on Ca2+ stores, resting [Ca2+]c, and excitability. Finally, to determine whether SOCE contributes to neuropathic pain, we also characterized SOCE in a model of peripheral nerve injury.
Materials and Methods
All methods and use of animals were approved by the Medical College of Wisconsin Institutional Animal Care and Use Committee.
Injury model.
Male Sprague Dawley (Taconic Farms) rats weighing 160–180 g were subjected to spinal nerve ligation (SNL) in a manner derived from the original technique (Kim and Chung, 1992). Rats were anesthetized with 2% isoflurane in oxygen and the right paravertebral region was exposed. After removal of the sixth lumbar (L6) transverse process, the L5 and L6 spinal nerves were ligated with 6-0 silk suture and transected distal to the ligature. The fascia was closed with 4-0 resorbable polyglactin suture and the skin closed with staples. Control animals received anesthesia, skin incision, and stapling only. After surgery, the rats were returned to their cages and kept under normal housing conditions with access to pellet food and water ad libitum.
Sensory testing.
Rats underwent sensory testing for a form of hyperalgesic behavior that we have previously documented to be associated with an aversive percept (Hogan et al., 2004; Wu et al., 2010). Briefly, on three different days between 10 and 17 d after surgery, right plantar skin was mechanically stimulated with a 22 gauge spinal needle with adequate pressure to indent but not penetrate the skin. Whereas control animals respond with only a brief reflexive withdrawal, rats after SNL may display a complex hyperalgesia response that incorporates sustained licking, chewing, grooming, and sustained elevation of the paw. The frequency of hyperalgesia responses was tabulated for each rat.
Neuron isolation and plating.
The right L5 ganglia were rapidly harvested after isoflurane anesthesia and decapitation and were incubated in 0.01% blendzyme 2 (Roche Diagnostics) for 30 min followed by incubation in 0.25% trypsin (Sigma-Aldrich) and 0.125% DNase (Sigma-Aldrich) for 30 min, both dissolved in DMEM/F12 with glutaMAX (Invitrogen). After exposure to 0.1% 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), and 0.02 mg/ml gentamicin (Invitrogen). Dissociated neurons were plated onto poly-l-lysine-coated glass coverslips (Deutsches Spiegelglas; Carolina Biological Supply) and maintained at 37°C in humidified 95% air and 5% CO2 for 2 h, and were studied no later than 6 h after harvest.
Solutions and agents.
Unless otherwise specified, the bath contained Tyrode's solution (in mm: 140 NaCl, 4 KCl, 2 CaCl2, 10 glucose, 2 MgCl2, 10 HEPES, with an osmolarity of 297–300 mOsm and pH 7.40). In some experiments, a Ca2+-free Tyrode's was used that contained the following (in mm): 140 NaCl, 4 KCl, 10 glucose, 2 MgCl2, 10 HEPES, and 0.2 EGTA.
Agents were obtained as follows: 2-aminoethyl diphenylborinate (2-APB), bovine albumin, caffeine, dimethylsulfoxide (DMSO), lanthanum chloride, 1-[2-(4-methoxyphenyl)-2-[3-(4-methoxyphenyl)propoxy]ethyl]imidazole (SKF-96365), thapsigargin (TG), N,N,N′,N′-tetrakis(2-pyridylmethyl)-ethylenedidiamine (TPEN), and 7-nitroindazole (7-NI) from Sigma-Aldrich, fura-2 AM from Invitrogen, and 1-(2-trifluoromethylphenyl)imidazole (TRIM) from Alexis Biochemicals. Stock solutions of 2-APB, SKF-96365, TG, 7-NI, TRIM, and fura-2 AM were dissolved in DMSO and subsequently diluted in the relevant bath solution such that final bath concentration of DMSO was 0.2% or less, which has no effect on [Ca2+]c (n = 20) (data not shown). The 0.5 ml recording chamber was constantly superfused by a gravity-driven bath flow at a rate of 3 ml/min. Agents were delivered by directed microperfusion controlled by a computerized valve system through a 500-μm-diameter hollow quartz fiber 300 μm upstream from the neurons. This flow completely displaced the bath solution, and constant flow was maintained by delivery of bath solution when specific agents were not being administered. Solution changes were achieved within 200 ms.
Measurement of cytoplasmic Ca2+ concentration.
Coverslips holding plated neurons were transferred to a room temperature 5 μm solution of fura-2 AM that contained 2% bovine albumin to aid dispersion of the fluorophore. After 30 min, they were washed three times with regular Tyrode's solution and left in a dark environment for deesterification for 30 min and then mounted onto the recording chamber. The fluorophore was excited alternately with 340 and 380 nm wavelength illumination (150 W xenon, Lambda DG-4; Sutter), and images were acquired at 510 nm using a cooled 12 bit digital camera (Coolsnap fx; Photometrics) and inverted microscope (Diaphot 200; Nikon Instruments) through a 20 or 40× Fluor oil-immersion objective. Recordings from each neuron were obtained as separate regions of interest by appropriate software (MetaFluor; Molecular Devices) at a rate of 3 Hz. After background subtraction, the fluorescence ratio R for individual neurons was determined as the intensity of emission during 340 nm excitation (I340) divided by I380, on a pixel-by-pixel basis. The calcium concentration was then estimated by the formula [Ca2+]c = Kd · β · (R − Rmin)/(Rmax − R), where β = (I380max)/(I380 min). Values of Rmin, Rmax, and β were determined by periodical in situ calibrations as described previously (Fuchs et al., 2005) and were 0.38, 8.49, and 9.54, respectively, and 224 nm was used as Kd (Grynkiewicz et al., 1985). Neurons were visually examined in the bright-field mode and those showing signs of lysis, crenulation, or superimposed glial cells were excluded. Similarly, only neurons with stable baseline R traces were further evaluated. Traces were analyzed using Axograph X 1.1 (Axograph Scientific). Neurons were characterized by diameter as large (>34 μm), which represent predominantly fast-conducting non-nociceptive Aβ neurons, or small (<34 μm), which represent a mix of Aβ neurons, slower conducting Aδ nociceptive neurons, and C-type nonmyelinated nociceptive neurons (data not shown). Unless otherwise stated, small neurons were examined. Fura-2 fluorescence during Sr2+ entry represents a mix of cytoplasmic Ca2+ and Sr2+, and transients were not calibrated, but rather are reported in R units.
Quantitative reverse transcriptase-PCR analysis.
Total RNA was isolated from the homogenized L5 dorsal root ganglia (DRGs) of control animals, and separately from the L4 and L5 DRGs of SNL rats harvested 21 d after surgery, following the manufacturer's (Invitrogen) instructions using Trizol reagent (from aqueous phase). After DNase treatment, cDNA was synthesized from equal amounts of RNA using SuperScript III first-strand synthesis kit (Invitrogen). Real-time PCR analysis was performed in duplicate for each run using iQ SYBR Green supermix (Bio-Rad) and specific primers to quantify the cDNA levels of STIM1 [forward primer (FP), GTGCGCTCGTCTTGCCCTGT; reverse primer (RP), TGCGGACGGCCTCAAAGCTG] and Orai1 (FP, CTGGCGCAAGCTCTACTTGA; RP, AGTAACCCTGGCGGGTAGTC). The expression level of housekeeping gene Tubb5 (FP, CATGGACGAGATGGAGTTCA; RP, GAAACAAAGGGCAGTTGGAA) was used for normalization. For each sample, two interrun determinations were averaged. Statistical evaluation was performed on normalized values. Figures show fold difference in expression of STIM1 and Orai1 in the DRGs from SNL animals, which was calculated by comparison with that of control DRGs.
Immunoblotting.
Total protein was isolated from homogenized L5 DRGs from control animals and separately from L4 and L5 DRGs of SNL rats harvested 21 d after surgery following manufacturer's (Invitrogen) instructions using Trizol reagent and sequential precipitation (from organic phase). Equal amounts of protein (20–50 μg; determined by Pierce bicinchoninic acid protein assay kit; Thermo Scientific) were separated on 4–15% SDS-PAGE gel (Bio-Rad) and transferred onto a polyvinylidene fluoride membrane. After blocking with 5% milk in TBST (Tris-buffered saline plus 0.1% Tween 20), blots were sequentially probed with anti-β-Tubulin I mouse monoclonal antibody (1:20,000; Sigma-Aldrich; catalog #T7816), anti-STIM1 rabbit polyclonal antibody (1:500; ProSci; catalog #4119), and anti-Orai1 rabbit polyclonal antibody (1:1000; ProSci, catalog #4281). Because of nonspecific bands using this antibody, Orai1 protein expression was examined with a second anti-Orai1 rabbit polyclonal antibody (1:1000; Abcam), which also showed nonspecific binding. The ProSci antibody was used for data shown here. Western blot Restore stripping buffer (Thermo Scientific) was used to strip antibodies from the membrane. Horseradish peroxidase-conjugated goat anti-rabbit and goat anti-mouse antibodies (1:2000) were used as secondary antibodies (Pierce). Enhanced chemiluminescence (GE Healthcare) was used for the detection of the protein bands. The bands obtained were quantified using NIH ImageJ program, and β-Tubulin I was used to normalize the protein loading.
Immunohistochemistry.
Twenty-one days after surgery, control and injured rats were perfused with saline followed by 4% paraformaldehyde. The control L5 DRGs and L4 and L5 DRGs from SNL rats were harvested and postfixed in 4% paraformaldehyde overnight, followed by incubation in 30% sucrose for 8 h. Tissues were frozen in TissueTek optimal cutting temperature compound (Ted Pella). Sections (15 μm) were permeabilized with PBS plus 0.1% Triton X-100 (PBST) for 20 min, blocked with 8% normal goat serum for 2 h, and then incubated overnight with anti-STIM1 rabbit polyclonal antibody (1:1500; ProSci). After three washes with PBST, sections were incubated with Alexa Fluor 568 goat anti-rabbit antibody (1:500; Invitrogen) for 1 h. The sections were washed thrice with PBST and examined by confocal microscopy. The expression level of STIM1 protein was represented by the average image intensity of two sites in each cell in images that were captured using standardized camera parameters, and cell area was determined by outlining the neuronal profile.
To determine colocalization of STIM1 with the neuron-specific nuclear protein (NeuN), sections were blocked with 8% NGS, incubated overnight with anti-STIM1 rabbit polyclonal antibody (1:1500) and anti-NeuN mouse monoclonal antibody (1:500; Millipore) followed by incubation with Alexa Fluor 568 anti-rabbit IgG (1:500; Invitrogen) for STIM1 antibody and Alexa Fluor 488 goat anti-mouse IgG conjugated with (1:1000; Invitrogen) to bind NeuN antibody for 1 h. To determine colocalization of STIM1 with glutamine synthetase, sections were blocked with 8% NGS, incubated overnight with anti-STIM1 rabbit polyclonal antibody (1:1500) followed by incubation with Alexa Fluor 568 goat anti-rabbit IgG (1:500) (Invitrogen). After three washes with PBST, the sections were incubated with anti-glutamine synthetase rabbit polyclonal antibody (1:500) (Santa Cruz Biotechnology) for 2 h. After washes, sections were incubated with Alexa Fluor 488 goat anti-rabbit IgG (1:1000; Invitrogen) for 1 h. Sections were washed three times with PBST and examined by confocal microscopy.
Intracellular electrophysiological recording.
Intracellular recordings were performed with microelectrodes fashioned from borosilicate glass (1 mm outer diameter, 0.5 mm inner diameter; with Omega fiber; FHC) using a P-97 programmable micropipette puller (Sutter). Pipettes were filled with 2 m potassium acetate, which was buffered with 10 mm HEPES, with a resulting resistance of 70–100 MΩ. For recording from dissociated neurons, coverslips carrying the neurons were mounted onto a 500 μl chamber and constantly superfused with Tyrode's solution at 3 ml/min. Neurons were selected in bright-field mode on an upright microscope using a 40× water-immersion objective and impaled under direct vision with the aid of an oscillating current to the recording electrode. Membrane potential was recorded using an active bridge amplifier (Axoclamp 2B; Molecular Devices). Voltage recordings were filtered at 10 kHz and then digitized at 40 kHz (Digidata 1322A; Molecular Devices; and Axograph X 1.1) for data acquisition and analysis. Alternatively, for recording from neuronal somata in intact DRGs, ganglia were perfused with a bath solution (in mm: 128 NaCl, 3.5 KCl, 1.2 MgCl2, 2.3 CaCl2, 1.2 NaH2PO4, 24.0 NaHCO3, 11.0 glucose) bubbled by 5% CO2 and 95% O2 to maintain a pH of 7. Neurons were impaled using differential interference contrast imaging with infrared illumination. Voltage error was minimized using a switching amplifier (Axoclamp 2B) operating in discontinuous current-clamp mode with a switching rate of 2 kHz, while monitoring for complete settling of electrode potential between sampling. Voltage recordings were filtered at 1 kHz. Recordings were not started until resting membrane potential had stabilized and resting membrane potential (RMP) was less than −45 mV (typically within 2 min). Somatic action potentials (APs) were generated by direct membrane depolarization with current injection through the recording electrode. The rheobase current was determined as the minimal depolarization adequate to produce an AP, and the resting voltage during this depolarization was considered the voltage threshold. AP duration was measured at 50% resolution, whereas afterhyperpolarization (AHP) duration was measured at 80% resolution toward RMP. Excitability was assayed two ways. Examination of the firing pattern during depolarizing current injection (100 ms, 0.2 nA increments) through the recording electrode allowed categorization of neurons as either repetitively firing versus those that fire only a single AP despite depolarization (accommodation). Data were included only from neurons that generated an initial AP at 10 mV or less, and firing behavior was evaluated during additional depolarization up to 30 mV. A second, additional analysis of repetitively firing neurons examined the slope relating the number of APs evoked at different transmembrane potentials during depolarization. This frequency gain (number of APs per millivolt) was determined as a liner fit only for neurons that showed at least three different levels of firing (number of APs) at potentials <30 mV.
Patch-clamp electrophysiological recording.
Voltage and currents were recorded in small- to medium-size neurons (28.8 ± 0.4 μm; n = 36), using the whole-cell configuration of the patch-clamp technique at room temperature. Patch pipettes, ranging from 2 to 5 MΩ resistance, were formed from borosilicate glass (Garner Glass) and fire polished. Currents were recorded with an Axopatch 200B amplifier (Molecular Devices), filtered at 2 kHz through a 4-pole Bessel filter, and digitized at 10 kHz with a Digidata 1320 A/D interface and pClamp 9 software (Molecular Devices) for storage on a personal computer. After achieving gigaohm seal and breakthrough, membrane capacitance was determined and access resistance was compensated (60–85%). Access resistance was typically between 5 and 10 MΩ after breakthrough.
A modified Tyrode's solution was used for external bath solution, consisting of the following (in mm): 140 NaCl, 4 KCl, 2 CaCl2, 2 MgCl2, 10 d-glucose, 10 HEPES at pH of 7.4 adjusted with NaOH and an osmolarity of 300 mOsm adjusted with sucrose. The internal pipette solution contained the following (in mm): 120 KCl, 5 Na-ATP, 0.4 Na-GTP, 10 EGTA, 2.25 CaCl2, 5 MgCl2, 20 HEPES at a pH of 7.2 with KOH and osmolarity of 296–300 mOsm. This produced a calculated [Ca2+]c of 70 nm (Maxchelator program; http://maxchelator.stanford.edu). A Na+-free/Ca2+-free external solution was prepared by removing the NaCl and CaCl2 from the modified Tyrode's solution and adding 85 N-methyl-d-glucamine (NMDG), 0.1 EGTA, 50 tetraethylammonium (TEA), 5 4-aminopyridine.
In Ca2+ readdition experiments, Na+-free/Ca2+-free solution was applied after the whole-cell configuration was achieved, which was then changed to an identical solution to which Ca2+ had been added to 10 mm final concentration (Hoth and Penner, 1992). Before and after solution changes, currents were recorded during hyperpolarization steps (−100 mV for 100 ms from a holding potential of −65 mV) presented every 5 s during Ca2+ readdition. In a second protocol, currents were recorded during voltage ramps (−100 to +20 mV over 100 ms) that immediately followed a conditioning depolarization (0 mV for 500 ms) that inactivated voltage-gated Ca2+ channels (VGCCs). Neurons were otherwise held at a potential of −65 mV, and currents were normalized based on cell capacitance. Five traces were averaged from the baseline Na+-free/Ca2+-free condition compared with an average of five traces obtained after steady-state responses were achieved with readded Ca2+.
Monovalent permeation experiments used a divalent-free (DVF) bath solution that was prepared by removing the CaCl2 and MgCl2 from the modified Tyrode's solution and adding 0.1 mm EGTA and 50 mm TEA (DeHaven et al., 2007). This solution was applied after the whole-cell configuration was achieved, in alternation with a solution that differed only in having 10 mm CaCl2. Current was recorded while neurons were continuously hyperpolarized (−100 mV). Baseline current was determined from the initial recording (0–5 s) and subtracted from subsequent recording for each neuron. Responses to application of DVF solution and DVF containing La3+ or Ca2+ were determined from averaging the trace across 10 s once a steady-state response was achieved.
Except as described above for ramp protocols, depolarization was avoided before current recordings, to avoid Ca2+ influx that could alter the state of Ca2+ stores. External solutions were administered by bath change (3.3 ml/min; bath volume, 0.7 ml). The rate of bath change, represented by the time constant of the exponential fitted to the time course of dye washout monitored photometrically, was τ = 21 ± 2 s (n = 7). To deplete Ca2+ stores and maximally activate ICRAC, TG (1 μm) was applied for 7 min in the external bath after establishing whole-cell patch configuration.
Statistical analysis.
Statistical analyses were performed with Statistica (StatSoft). The t test or one-way ANOVA was used to detect the influence of injury group on measured parameters, except for immunoblot and quantitative reverse transcriptase-PCR (rtPCR) for which the nonparametric Kruskal–Wallis test was used. Parameters determined from intracellular recordings were analyzed by two-way ANOVA to determine the effect of injury, the effect of TRIM, and their interaction. Where main effects were observed in ANOVA, either Bonferroni's post hoc test or Tukey's test (when all possible comparisons were considered) was used to compare relevant means, and a value of p < 0.05 was considered significant. Results are reported as average ± SEM.
Results
A total of 139 rats was used for the study, of which 75 were control animals and 64 were subjected to SNL. The average rate of hyperalgesia-type behavioral responses was 0.6 ± 0.3% in control animals and 41.3 ± 2.8% in SNL animals (p < 0.001). All SNL animals used in this study had >20% hyperalgesia responses to pin stimulation.
SOCE is present in sensory neurons
After a 7 min interval in Ca2+-free bath, which is long enough to establish a stable [Ca2+]c (50 ± 2 nm; n = 57), readdition of 2 mm bath Ca2+ to sensory neurons initiated Ca2+ influx and resulted in a rise in [Ca2+]c (Fig. 1A) to levels that exceeded the original baseline in all neurons (baseline, 73 ± 4 nm; readdition, 100 ± 6 nm; n = 37; p < 0.001). This suggests a regulatory process driven by a signal that originates in a site other than the cytoplasm. Extension of the duration of Ca2+ deprivation to 30 min before Ca2+ readdition resulted in transients with amplitudes that did not differ from those after 7 min in Ca2+-free bath (Fig. 1A,B). Sensory neurons are a heterogeneous population and include nociceptors that characteristically respond to capsaicin with Ca2+ influx through transient receptor potential vanilloid 1 (TRPV1) receptors (Caterina et al., 1997). The Ca2+ readdition transient amplitude in neurons that were subsequently shown to be sensitive to 10 nm capsaicin did not differ from those that were unresponsive to capsaicin (Fig. 1B). Large soma diameter typically characterizes fast-conducting neurons that respond to low-threshold mechanical stimuli (Waddell and Lawson, 1990). There was no difference in Ca2+ readdition transient amplitude between large-diameter (39 ± 1 μm) and small-diameter (27 ± 1 μm) neurons (Fig. 1B).
Manipulation of SOCE by withdrawal of Ca2+ from the bathing solution and the readdition of Ca2+. A, Ca2+ withdrawal reduces resting [Ca2+]c, which is mostly complete within ∼3 min. Withdrawal for 30 min (gray trace, 32-μm-diameter neuron) depresses [Ca2+]c only slightly more than a 7 min withdrawal in a different neuron (dark trace, 30 μm). Readdition of Ca2+ to the bath causes elevation of [Ca2+]c to a level higher than the original baseline (dashed line). B, Amplitudes of the [Ca2+]c transients induced by readdition of Ca2+ to the bath in separate groups of neurons. Unless otherwise indicated, neurons are small (<34 μm diameter) and the Ca2+ withdrawal is for 7 min. There was no difference between bath Ca2+ withdrawal for 7 min versus 30 min. Neurons sensitive to capsaicin (10 nm; Cap+) showed the same transient amplitude as insensitive neurons (Cap−). Amplitude in large neurons (42 ± 2 μm diameter) did not differ from small neurons. Chelation of endoplasmic reticulum Ca2+ stores with TPEN (100 μm; 3 min) or depletion of Ca2+ stores with TG (1 μm; 7 min) each increased the Ca2+ readdition transient amplitude. The numbers in the bars indicate number of neurons; the brackets indicate p < 0.05. Shown are mean ± SEM. C, Trace (typical of n = 9) showing a greater increase in [Ca2+]c with Ca2+ readdition to the bath solution after exposure to TG (1 μm). Additionally, the maximal rate of [Ca2+]c during Ca2+ readdition, indicated by the tangential dotted lines, is greater after TG (3.2 nm/s) than before (1.1 nm/s).
A cardinal feature of SOCE is its sensitivity to the level of intracellular Ca2+ stores. We therefore tested whether the Ca2+ readdition transient is amplified by complete depletion of ER Ca2+ stores achieved through exposure of neurons to the sarco-endoplasmic Ca2+-ATPase (SERCA) inhibitor TG. By leaving the constitutive leak of Ca2+ from the ER unopposed, TG itself causes a [Ca2+]c elevation that resolves within 5–7 min in Ca2+-free bath (see Fig. 3A). Application of bath Ca2+ to neurons that had been exposed to TG (1 μm; 7 min) produced Ca2+ transients with amplitudes that were greater than those in other neurons without TG (Fig. 1B) (p < 0.01 vs control neurons). Although the amplitude of the Ca2+ readdition transient is a commonly accepted measure of SOCE (Usachev and Thayer, 1999; Mercer et al., 2006), it is possible that competing processes that extrude Ca2+ from the neuron influence the level of the steady state that underlies the [Ca2+]c peak. Accordingly, we also examined the initial slope determined from the differentiated [Ca2+]c trace, which may provide a more direct and valid measure of the rate of Ca2+ influx during bath readdition (Glitsch et al., 2002). By this measure, the SOCE was also greater with TG (3.85 ± 0.39 nm/s; n = 93) than without TG (0.86 ± 0.14 nm/s; n = 35; p < 0.001).
To confirm the influence of TG on SOCE, separate experiments were performed in which Ca2+ readdition transients were compared in the same neuron under baseline conditions and again after TG application (Fig. 1C). TG increased the transient amplitude 2.7 ± 0.4-fold compared with the transient before TG (p = 0.0001; n = 9). This was significantly greater amplification (p < 0.01) than the slight increase in the second transient (1.2 ± 0.1-fold; p = NS; n = 24) when Ca2+ readdition was repeated in the absence of TG. Measuring slopes confirmed that TG increased the Ca2+ influx rate 3.3 ± 1.1-fold compared with baseline (Fig. 1C) (p < 0.05; n = 9), whereas the amplification during repeat Ca2+ readdition without TG (1.4 ± 0.1-fold; p < 0.01; n = 23) was significantly less (p < 0.001).
Chelation of ER Ca2+ stores provides an alternative method for testing the influence of Ca2+ stores on Ca2+ influx. TPEN passes freely through membranes and only binds Ca2+ at micromolar concentration, thereby having a selective effect on stored Ca2+. In investigations of other cell types, TPEN depresses free Ca2+ concentration in ER stores within 10 s of application, without an effect on cytoplasmic Ca2+ levels (Hofer et al., 1998). Ca2+ readdition to sensory neurons incubated in TPEN (100 μm; 3 min) produced transient amplitudes that were on average 1.7-fold greater (p < 0.05) than those in other neurons incubated in Ca2+-free bath (7 min) without the chelator (Fig. 1B). Compared with TPEN, TG may have a relatively greater effect on [Ca2+]c rise after bath Ca2+ readdition because of its blockade of Ca2+ sequestration from the cytoplasm via SERCA. Together, the findings from depletion and chelation of Ca2+ stores reveal a Ca2+ entry process that is regulated by the state of ER Ca2+ stores.
Since sensory neurons possess a variety of VGCCs, we examined the possibility that activation of these channels by membrane depolarization during Ca2+ readdition might underlie a component of the observed Ca2+ influx. Intracellular electrode recordings showed membrane potential was unchanged during withdrawal of Ca2+ from the bath and during subsequent Ca2+ readdition (Fig. 2A,B), which indicates that neither action potentials nor incremental membrane depolarization contribute to generating the Ca2+ readdition transient. It is possible nonetheless that Ca2+ may enter neurons through VGCCs that are conducting at resting membrane potential, particularly low-voltage-activated T-type Ca2+ channels (Lee et al., 1999). However, the presence of mibefradil (200 nm), a T-type VGCC blocker, did not alter the transient amplitude during Ca2+ readdition to TG-treated neurons (1 μm; 7 min) compared with a preceding Ca2+ readdition in the absence of mibefradil (96 ± 14 nm baseline; 92 ± 10 nm after blockade; p = 0.63; n = 14). Combined application of selective VGCC blockers (Fig. 3A) (mibefradil, nitrendipine, 10 μm, for L-type current; SNX-111, 200 nm, for N-type current; ω-conotoxin Aga-IVA, 200 nm, for P/Q-type current; SNX-482, 100 nm, for R-type current) also did not alter Ca2+ readdition transient amplitudes (94 ± 10 nm baseline; 135 ± 10 nm after blockade; p = 0.70; n = 32), which suggests that no Ca2+ enters through this pathway and confirms Usachev and Thayer's previous findings in cultured embryonic neurons.
Membrane potential of a sensory neuron after dissociation from a dorsal root ganglion, recorded by intracellular electrode during inactivation of SOCE by bath Ca2+ withdrawal and SOCE activation by Ca2+ readdition. A, Sample trace from a simultaneous recording of membrane potential and [Ca2+]c from a control neuron (32 μm diameter). B, There is no difference in potential immediately after impalement (time point 1), during decreased [Ca2+]c with Ca2+ withdrawal (time point 2), or on readdition of Ca2+ to the bath solution (time point 3). Shown are mean ± SEM.
Effect of blockers on SOCE measured as the [Ca2+]c rise on bath Ca2+ readdition. A, Blockade of VGCCs with combined administration of mibefradil (200 nm) for T-type current, nitrendipine (10 μm) for L-type current, SNX-111 (200 nm) for N-type current, ω-conotoxin Aga-IVA (200 nm) for P/Q-type current, and SNX-482 (100 nm) for R-type current does not influence the [Ca2+]c increase on bath Ca2+ readdition in a control neuron (28 μm diameter) treated with TG (1 μm). Typical of n = 32. B, Specific blockers of SOCE reduce the amplitude of [Ca2+]c rise on bath Ca2+ readdition. La3+ (10 μm), 2-APB (100 μm), and TRIM (400 μm) significantly (*p < 0.05) reduced transient amplitude. SKF-96365 (10–50 μm) had no effect. Shown are mean ± SEM.
Sensitivity to blockers of SOCE, such as low concentrations of La3+, has been used in previous studies to identify SOCE (Szikra et al., 2009). During Ca2+ readdition in TG-treated sensory neurons (Fig. 3B), we found a substantial suppression of transient amplitude by La3+ (10 μm), which eliminated 79 ± 7% of the Ca2+ readdition transient compared with other neurons without blocker. 2-APB (100 μm), another commonly used SOCE inhibitor (Bootman et al., 2002), moderately decreased the Ca2+ readdition transient amplitude in sensory neurons by 39 ± 6%. TRIM (400 μm), which has been used to block SOCE in neurons (Tobin et al., 2006), suppressed the readdition transient by 61 ± 5%. The commonly used SOCE blocker SKF-96365 (10–50 μm) did not have any significant effect on sensory neurons. ML-9, a novel SOCE blocker (Smyth et al., 2008), itself elevated [Ca2+]c by 83 ± 24 nm (n = 15) in neurons bathed in Ca2+-free solution, indicating release of Ca2+ from stores, and was not investigated further.
ICRAC underlies SOCE in sensory neurons
Direct measurement of the Ca2+ release-activated current ICRAC has been achieved in expression systems and in several native cell types. The exceptionally small conductance of store-operated Ca2+ channels eliminates the option of single-channel recording, so we used the whole-cell patch-clamp technique to identify ICRAC.
We initially attempted to record a Ca2+-dependent inward current at the approximate natural resting potential of sensory neurons (−65 mV). Ca2+ readdition produced small increases in inward current (Fig. 4A), but this approach produced inconsistent results, as has been reported previously (Liu et al., 2003). We therefore used voltage conditions that amplified the observable influence of extracellular Ca2+. During hyperpolarization steps to −100 mV (Fig. 4B) (Hoth and Penner, 1992), ICa in bath containing 10 mm Ca2+ (−1.4 ± 0.1 pA/pF) was greater than the current recorded in the same neurons in bath without Ca2+ (−1.0 ± 0.1 pA/pF; n = 5; p < 0.001), confirming a Ca2+-dependent inward current in the absence of depolarization. During ramp depolarization (Parekh, 1998; DeHaven et al., 2007), comparison of currents before and after readdition of bath Ca2+ to TG-treated neurons (Fig. 4C) revealed a greater current in the presence of bath Ca2+ (−3.1 ± 0.8 pA/pF; measured at −80 mV) than during Ca2+-free conditions in the same neurons (−0.8 ± 0.1 pA/pF; n = 5; p < 0.05), again demonstrating a Ca2+-dependent inward current. Inward rectification, which is a characteristic feature of ICRAC (Hoth and Penner, 1992), was observed in the Ca2+-dependent component of the current (Fig. 4C, subtracted current), supporting the identification of this current as ICRAC.
Recording of the Ca2+ release-activated current (ICRAC). A, Withdrawal of bath Ca2+ lowers inward current, and readdition of bath Ca2+ (5 mm) elevates current during voltage clamp (−65 mV) of a control neuron (30 μm diameter; top panel) and an injured neuron (30 μm; bottom panel) by SOCE. The scale bars apply to both traces. B, Current traces generated during step hyperpolarizations in a control neuron (30 μm). The baseline trace (BL) was during 0 bath Ca2+. Inward current induced by hyperpolarization increased during 10 mm bath Ca2+ and returned to baseline levels with washout of Ca2+ from the bath (WO). The difference current (baseline trace subtracted from the 10 mm Ca2+ trace) represents ICRAC. (Typical of n = 5.) C, Presentation of a ramp depolarizing voltage command (12 mV/ms) generates current (plotted here against voltage) in a control neuron (28 μm) that is less with 0 bath Ca2+ than during 10 mm bath Ca2+. The difference trace represents ICRAC. Outward current is carried by K+ channels that were sensitive to (data not shown), but incompletely blocked by, bath TEA (50 mm). The ICRAC trace shows inward rectification, with an inflection at approximately −34 mV, above which Ca2+-sensitive K+ current contributes an outward component. (Typical of n = 5.) D, After incubation in TG (1 μm for 7 min), exposing a control neuron (32 μm) to bath solution lacking divalent cations (DVF) produced a large inward current through the store-operated Ca2+ channels, which is blocked by La3+ (10 μm). E, Data from experiments such as shown in D. In control neurons (top panel), DVF produces minimal current under baseline conditions in the absence of store depletion (filled circles). In separate control neurons incubated in TG (open circles), DVF produces large inward currents that are blocked by La3+. Injured neurons (SNL L5; bottom panel) show greater DVF-induced currents without store depletion by TG (filled squares), which are not significantly altered by TG (open squares). Shown are mean ± SEM. *p < 0.05; **p < 0.01.
Store-operated Ca2+ channels become nonselective for Ca2+ ions in the absence of divalent cations, whereupon a high rate of Na+ influx provides a more readily measurable manifestation of ICRAC (Hoth and Penner, 1993; DeHaven et al., 2007). Only small currents were induced on exposure of sensory neurons to DVF bath solution under baseline conditions. However, sensory neurons incubated in TG (1 μm for 7 min) showed robust ICRAC on exposure to DVF conditions, unlike neurons without store depletion (p < 0.01 for DVF with vs without TG) (Fig. 4D,E), which identifies this as a current conducted by store-operated channels. The initiation the DVF-induced inward current showed a delay (53 ± 7 s; n = 6) (Fig. 4D), despite previous store depletion by incubation in TG. We attribute this delay to the time required to achieve the low level of bath Ca2+ that is necessary to allow nonselective conductance, for which reported Kd range from 1.7 to 4.5 μm (Lepple-Wienhues and Cahalan, 1996; Kerschbaum and Cahalan, 1998; Rychkov et al., 2001). Specifically, an exponential model incorporating the bath volume (0.7 ml) and inflow rate (3.3 ml/min) predicts 110 and 98 s to reach these levels of bath Ca2+, although imperfect mixing may cause faster washout in the central area used for recording. Sensory neurons did not show the fast depotentiation of DVF-induced ICRAC as has been noted in other cells and expression systems (Zweifach and Lewis, 1996; DeHaven et al., 2007; Smyth et al., 2008). Return of Ca2+ to the bath resulted in immediate termination of Na+ permeation (Fig. 4D). Additional recognition of the observed current as ICRAC was obtained through the application of La3+ (10 μm), an established blocker of ICRAC (Hoth and Penner, 1993), which eliminated 62% of the current initiated by DVF solution in TG-treated neurons (p < 0.05 vs neurons without TG) (Fig. 4D,E).
Identification of molecular components of SOCE
STIM1 protein has previously been confirmed in neuronal tissues (Gasperini et al., 2009; Klejman et al., 2009), but not in adult sensory neurons, whereas Orai1 has been colocalized with STIM1 in brain neurons (Klejman et al., 2009). In DRG neurons, we have found expression of both components of SOCE in sensory neurons at the protein level by immunoblotting (Fig. 5A,B) and at the transcript level by quantitative rtPCR (Fig. 5C) (n = 3 control animals for both determinations), indicating that STIM1 and Orai1 are available in sensory neurons as potential constituents of SOCE. Immunohistochemistry of DRG tissue from control animals (n = 3) revealed the presence of STIM1 in neuronal somata as a homogeneous distribution in the cytoplasm (Fig. 6A), consistent with its location in the ER. There was no preferential expression of STIM1 in DRG subpopulations of different neuronal size (Fig. 6B). Double staining for STIM1 with NeuN, a neuron-specific marker (Fig. 6A), revealed expression of STIM1 in all neurons. Using glutamine synthetase as a marker for satellite glial cells (Weick et al., 2003) demonstrated that STIM1 is also present in the cytoplasm of satellite glial cells. Since available antibodies to Orai1 protein reacted with nonspecific bands in immunoblotting, immunohistochemical selectivity could not be assured and anatomic identification was not performed.
Identification of molecular components of store-operated Ca2+ channels. A, Western blotting detected Orai1 and STIM1 proteins in lysates from control (C) dorsal root ganglia (DRG), from fourth lumbar DRGs after fifth lumbar spinal nerve ligation (SNL L4), and from axotomized fifth lumbar DRGs after SNL (SNL L5). A dominant STIM1 band at the expected molecular weight is accompanied by a closely coherent secondary band that others have shown to be sensitive to STIM1 knockdown by RNAi (Ong et al., 2007). (Typical of n = 3.) B, Quantification of Western blots as fold difference compared with control showed no effect of injury on Orai1 and STIM1 protein levels. Shown are mean ± SEM. C, Quantification of transcript levels by quantitative rtPCR also showed no effects of injury on Orai1 and STIM1 expression. For B and C, the numbers in bars indicate n.
Immunohistochemical identification of STIM1 distribution. A, Staining for STIM1 (left panels) was uniformly distributed within cytoplasmic areas of cells identified as neuronal profiles through costaining with NeuN (middle panels). Only minimal nonspecific staining was present in fiber tracts. The scale bar applies to each panel. B, Morphometric analysis showed comparable intensity (arbitrary units, recorded using standardized image acquisition) of staining in neurons of all sizes (ANOVA, p = 0.40). Shown are mean ± SEM. C, STIM1 (left panel) was also found in cellular components identified as satellite glial cells by their expression of glutamine synthase (GS) (middle panel). Identically placed arrows in the left and middle panels indicate cellular areas clearly identifiable as satellite glial cell cytoplasm that expresses STIM1. The scale bar applies to each panel.
Functional role of SOCE in sensory neurons
The influence of SOCE on neuronal function is poorly defined (Putney, 2003). SOCE has been noted to modulate resting [Ca2+]c in rat sympathetic neurons (Wanaverbecq et al., 2003). We found that most (57 of 71; 80%) sensory neurons responded to Ca2+withdrawal with a fall in [Ca2+]c (Fig. 1A) by 17 ± 3 nm (p < 0.001 withdrawal vs baseline) after 7 min exposure to Ca2+-free bath (Fig. 7A), which indicates a dependence of resting [Ca2+]c on ongoing SOCE. In a subset of neurons, we extended the interval of Ca2+ withdrawal to 30 min, but did not observe any significant difference compared with the 7 min interval (14 ± 2 nm; p < 0.001, withdrawal vs baseline; p = 0.42 for 7 vs 30 min).
Functional roles of SOCE. A, Decreasing bath Ca2+ from 2 to 0 mm reduced [Ca2+]c in control (C) neurons, as well as in fifth lumbar (L5) neurons axotomized by SNL and those in the adjacent L4 dorsal root ganglion (*p < 0.05). SNL L5 neurons had lower resting [Ca2+]c levels before and after bath Ca2+ withdrawal compared with control and SNL L4 neurons (brackets). B, The size of intracellular Ca2+ stores that are releasable by caffeine (20 mm), measured as the area under the curve (AUC) of the caffeine-induced [Ca2+]c transient, was diminished by withdrawal of bath Ca2+ for 7 and 30 min compared with baseline conditions (0 min of Ca2+ withdrawal; ANOVA main effect for time in 0 bath Ca2+, p < 0.001; paired comparisons, *p < 0.05, **p < 0.01). Stores are reduced in SNL L5 neurons before bath Ca2+ withdrawal (ANOVA main effect for injury group, p < 0.001; bracket indicates p < 0.01 for paired comparison). The numbers in the bars indicate number of neurons. Shown are mean ± SEM. C, Absence of bath Ca2+ prevents refilling of intracellular Ca2+ stores. Caffeine (20 mm) releases stored Ca2+ in a control neuron (29 μm diameter; left panel), which is replenished during 10 min in 2 mm bath Ca2+, as shown by a second caffeine-induced transient of almost equal size. In another neuron (32 μm; right panel), absence of bath Ca2+ after Ca2+ store depletion results in a much reduced release of Ca2+ by the second caffeine application. The scale bar applies to both traces.
Neurons, particularly slowly conducting C-type neurons with small-diameter somata, may remain quiescent for a sustained period of time (Schmidt et al., 1995). Since extrusion of cytoplasmic Ca2+ by the plasma membrane Ca2+-ATPase continues in resting neurons (Wanaverbecq et al., 2003), SOCE may be an important source of Ca2+ influx for maintaining intracellular stores during inactivity. To test this, resting sensory neurons were incubated in Ca2+-free bath solution for various time intervals and then exposed to 20 mm caffeine, which leads to a quantifiable emptying of Ca2+ stored in the ER (Rigaud et al., 2009). Releasable Ca2+ significantly decreased after 7 min in Ca2+-free bath and even more after 30 min (Fig. 7B), revealing a constitutive role of SOCE in the maintenance of intraneuronal Ca2+ store levels.
Ca2+ stores in sensory neurons may be depleted through the activation of metabotropic receptors, such as those for bradykinin, ATP, and glutamate (Thayer et al., 1988; Crawford et al., 2000; Kruglikov et al., 2004), or by activation of high-affinity TRPV1 channels on the ER (Liu et al., 2003). In the absence of neuronal depolarization, the SOCE influx pathway may serve a critical role in generating plasmalemmal Ca2+ influx for replenishing Ca2+ stores. We found that Ca2+ stores, measured by release with caffeine (20 mm), recover almost completely during 10 min with 2 mm bath Ca2+ (second transient compared with first; amplitude, 89 ± 5%; area, 72 ± 3%; n = 26) (Fig. 7C), but recovery fails in the absence of bath Ca2+ (amplitude, 5 ± 2%; area, 3 ± 1%; n = 14), indicating that SOCE is needed for replenishing intraneuronal Ca2+ stores after a release event.
Effect of painful nerve injury on SOCE
In previous studies, we found that nerve injury depresses resting [Ca2+]c and diminishes Ca2+ stores (Fuchs et al., 2005; Gemes et al., 2009; Rigaud et al., 2009). Since our present findings show that SOCE functions to maintain intracellular levels of cytoplasmic and releasable Ca2+, we examined the possibility that SOCE is deficient after peripheral nerve injury by SNL, a standard model of neuropathic pain. This proved not to be the case. The amplitude of the SOCE transient on Ca2+ readdition was increased in axotomized L5 neurons after SNL compared with neurons from control animals, after both 7 and 30 min of Ca2+ withdrawal, whereas there was no effect on adjacent L4 neurons (Fig. 8A). Analysis of the transient slope similarly showed an amplification of SOCE after injury (Fig. 8A). In contrast, when SOCE was maximized by store depletion with TG, transient amplitude and slope in injured SNL L5 neurons were no different from control neurons (Fig. 8B), implying that injured neurons have increased regulatory drive for SOCE but not increased maximal efficacy of SOCE.
Effect of injury by SNL on the axotomized fifth lumbar (L5) and L4 neurons compared with control (C) neurons. A, Both the maximum slope of the [Ca2+]c trace and the amplitude of the transient induced by Ca2+ readdition are increased in L5 neurons after SNL (ANOVA main effect for injury group, p < 0.01; bracket indicates p < 0.05 for paired comparison). Increasing the time of Ca2+ withdrawal from 7 to 30 min has no effect on the amplitude of readdition transients. B, Maximizing store-operated Ca2+ entry by store depletion with thapsigargin incubation (1 μm; 7 min) increased average transient slope and amplitude compared with transients without thapsigargin (shown in A) and eliminated the influence of injury because of a diminished effect on SNL L5 neurons. The numbers in the bars indicate number of neurons. Shown are mean ± SEM.
To determine whether STIM1 and Orai1 expression is affected by injury, levels of protein and transcript of both genes were measured after SNL. The levels of STIM1 and Orai1 protein determined by immunoblotting were not significantly different between the DRGs of control (n = 3) and injured (n = 3) animals (STIM1, p = 0.12; Orai1, p = 0.73) (Fig. 5B). Similarly, quantitative rtPCR analysis (Fig. 5C) indicated that transcript levels of STIM1 (p = 0.67) and Orai1 (p = 0.49) in L5 DRGs of control animals (n = 3) and L4 and L5 DRGs of the injured animals (n = 3) were comparable. The anatomical distribution of STIM1 in L4 and L5 DRG sections from SNL animals (n = 3) double stained for STIM1 and NeuN were not different from control findings described above (Fig. 6A). These analyses together indicate that STIM1 and Orai1 expression is not altered by peripheral nerve injury.
Direct measurement of ICRAC induced by DVF conditions in injured SNL L5 neurons again showed delays in onset of the current (without TG, 105 ± 21 s; n = 8; with TG, 77 ± 18 s; n = 6). In the absence of store depletion, ICRAC was greater in SNL L5 neurons (−4.15 ± 0.64 pA/pF; n = 8) than in control neurons (−0.41 ± 0.14 pA/pF; n = 6; p < 0.01) (Fig. 4E), indicating that ICRAC is increased in sensory neurons after nerve injury. TG failed to have a significant effect on measured ICRAC in injured neurons (average, −7.04 ± 2.67 pA/pF; n = 6) (Fig. 4E), suggesting that injury itself depletes stores and activates ICRAC. After store depletion by TG, there is little difference in DVF-induced ICRAC between injured neurons (−7.04 ± 2.67 pA/pF) and control neurons (−6.44 ± 1.22 pA/pF) (Fig. 4E). These findings, together with lack of effect of injury on Ca2+ readdition transients after TG and comparable expression of molecular subunits after injury, suggest that injury activates SOCE through store depletion but does not alter the intrinsic capacity of SOCE.
Injured neurons demonstrate an elevated dependence on SOCE. Although resting [Ca2+]c is depressed in injured neurons (54 ± 4 nm; n = 30) compared with control neurons (66 ± 3 nm; n = 57; p < 0.05) and adjacent SNL L4 neurons (70 ± 3 nm; n = 24; p < 0.05), termination of SOCE by bath Ca2+ withdrawal imposes additional depression of [Ca2+]c (Fig. 7A), provoking much lower [Ca2+]c after injury in SNL L5 neurons (34 ± 2 nm) than in control conditions (50 ± 2 nm; p < 0.001) and SNL L4 neurons (44 ± 3 nm; p < 0.05). The level of stored Ca2+ in resting sensory neurons is also diminished by injury (Fig. 7B) (Rigaud et al., 2009). Elimination of SOCE has a particularly severe effect on Ca2+ stores in injured neurons, such that SNL L5 neurons retain only 10% of their original stored Ca2+ after 30 min of SOCE termination through bath Ca2+ removal, compared with retention of 33% by control neurons and 23% by SNL L4 neurons. The combined effect of injury and SOCE termination reduces Ca2+ stores to 4% of that in baseline control neurons (Fig. 7B). Together, these observations indicate that SOCE plays an amplified role in maintaining Ca2+ homeostasis after axonal trauma in sensory neurons.
SOCE regulation of neuronal excitability
We have previously observed that blockade of Ca2+-induced Ca2+ release (CICR) from intracellular stores increases sensory neuron excitability associated with a decreased AHP duration (Gemes et al., 2009). We therefore reasoned that store depletion from loss of SOCE might have a comparable effect. We chose not to eliminate SOCE by bath Ca2+ withdrawal since this can directly increase membrane excitability, independent of the loss of SOCE, by disrupting membrane charge and by decreasing depolarization-induced ICa through voltage-gated channels, which in turn decreases Ca2+-activated K+ currents and AHPs that follow APs (Lirk et al., 2008). We instead examined the effects of SOCE blockade while recording transmembrane potentials from small- to medium-sized neurons, using an intracellular electrode technique in excised but intact DRGs that avoids the excitatory influence of neuronal dissociation (Zheng et al., 2007). Since La3+ is a nonselective blocker of Ca2+ channels including VGCCs, we therefore used TRIM (200 μm) as the best available blocker despite its incomplete efficacy (Fig. 3B). We paired recordings on each day such that an L5 DRG from a control animal was incubated in TRIM and its other L5 DRG was incubated in vehicle (0.2% DMSO) for 30 min, randomizing the sequence. For injured neurons, two L5 DRGs ipsilateral to an SNL injury were harvested from two different animals, and used in a similar paired fashion. There was no effect of TRIM on RMP or AP width (Table 1), comparable with previous findings in neurons of the supraoptic nucleus (Tobin et al., 2006). In vehicle-treated neurons, injury produced longer AP duration and decreased rheobase, as has been noted previously (Sapunar et al., 2005). TRIM increased the rheobase current necessary for initiating an AP in both control and injured neurons, and additionally decreased input resistance and lowered voltage threshold for AP generation in injured neurons. The late phase of the AHP is produced by the slow SK isoform of the Ca2+-activated K+ channel, which is particularly dependent on Ca2+ released from stores. We therefore examined AHP duration, which was decreased by TRIM, particularly in injured neurons. The duration of the AHP regulates repetitive firing behavior of sensory neurons (Sapunar et al., 2005). We evaluated this by injection of suprathreshold depolarizing currents, which produced either a repetitive firing pattern (Fig. 9A) or a completely accommodating pattern in which only a single spike was generated despite depolarization beyond threshold (Fig. 9B). TRIM increased the incidence of repetitively firing neurons in both control and injured neurons. In these repetitively firing neurons, we characterized neuronal excitability further by plotting the number of evoked APs at each membrane potential, for which the slope of the fitted line represents the gain of the relationship. TRIM increased the gain in both control neurons (DMSO, 0.13 ± 0.02 APs/mV, n = 6; TRIM, 0.28 ± 0.06, n = 12) and injured neurons (DMSO, 0.13 ± 0.02 APs/mV, n = 6; TRIM, 0.20 ± 0.03, n = 18; ANOVA main effect of TRIM, p < 0.01), showing that blockade of SOCE elevates neuronal burst firing generally. Although there is a higher resting level of SOCE in injured neurons, we did not demonstrate a difference in the effect of TRIM on excitability of control and injured neurons, possibly because TRIM leaves 39% of SOCE intact (Fig. 3).
Electrophysiological behavior of control and injured fifth lumbar dorsal root ganglion neurons during either blockade of store-operated Ca2+ entry with TRIM (200 μm) or application of vehicle (DMSO, 0.2%)
Effect of SOCE blockade on neuronal excitability, showing action potential generation patterns recorded by intracellular electrode in two neurons from the fifth lumbar dorsal root ganglion of rats subjected to spinal nerve ligation, during comparable depolarization (1.2, 2.4, and 3.6 nA). Noise is attributable to the use of a switching amplifier (2 kHz), used to minimize voltage error from current injection through the high-resistance electrode. A, Recording from a neuron (30 μm diameter) incubated with the SOCE blocker TRIM (200 μm for 45 min) shows repetitive firing during depolarization past rheobase. B, A different neuron (30 μm) bathed in vehicle (DMSO, 0.2%) demonstrates accommodation, with only a single action potential despite depolarization past rheobase.
Apart from blocking SOCE, TRIM may also inhibit neuronal nitric oxide synthase (NOS) (Handy et al., 1995; Gibson et al., 2001), and NO may affect voltage-gated Ca2+ and Na+ currents in DRG neurons (Kim et al., 2000; Renganathan et al., 2000). We therefore examined the effect of incubating control neurons with 7-NI (200 μm; n = 22), a selective neuronal NOS inhibitor (Moore et al., 1993), which produced no elevation of repetitive firing compared with neurons of matched ganglia incubated in vehicle (0.1% DMSO; n = 19). From these findings, we infer that the increased repetitive firing of DRG neurons produced by TRIM is the result of loss of SOCE.
Influence of store depletion on depolarization-induced Ca2+ influx
Recent reports indicate that STIM1, when activated by store depletion, suppresses Ca2+ influx through L-type VGCCs (Park et al., 2010; Wang et al., 2010). To determine whether this regulatory pathway functions in sensory neurons, we incubated neurons from uninjured animals for 30 min in normal Ca2+ bath containing TRIM (200 μm) or vehicle (0.2% DMSO). Thereafter, the bath was changed to a solution in which Ca2+ was replaced by Sr2+ to block SOCE, and dantrolene (10 μm) was added to block CICR from stores. After 1 min in this solution, neuronal depolarization (rapid bath application of K+, 50 mm, 3 s) triggered Sr2+ entry through VGCCs, which produced transients in TRIM-treated neurons that had diminished amplitudes (TRIM, 0.81 ± 0.12 R units, n = 18; DMSO, 1.08 ± 0.08 R units, n = 29; p < 0.05) and initial slopes (TRIM, 0.54 ± 0.07 R units/s, n = 18; DMSO, 0.76 ± 0.07 R units/s, n = 29; p < 0.05) compared with DMSO control. These diminished transients are unlikely to be attributable to decreased CICR from TRIM-depleted stores since dantrolene will have blocked CICR in both DMSO and TRIM groups. Also, it is known that TRIM at a 10-fold higher concentration does not directly block VGCCs (Tobin et al., 2006). So, in addition to regulating SOCE, it is likely that STIM1 activation in sensory neurons inhibits VGCCs, which may in turn contribute to elevated neuronal excitability (Lirk et al., 2008).
Discussion
Our data confirm and extend previous observations that infer the presence of SOCE in central and peripheral neurons. In sensory neurons from neonatal rats, Usachev and Thayer (1999) have shown that replenishment of intracellular stores requires bath Ca2+. They and others (Liu et al., 2003; Lu et al., 2006) have observed SOCE activity through the rise in [Ca2+]c on return of Ca2+ to the bath solution, which is modulated by store level (Usachev and Thayer, 1999). We have confirmed that SOCE, represented by the readdition transient, is a general feature of acutely dissociated adult neurons, including putative nociceptors with small diameters and capsaicin sensitivity, as well as large, capsaicin-insensitive non-nociceptive neurons. Both the depletion of stores by SERCA blockade and the chelation of stores by TPEN result in amplification of the transient, suggesting the regulation of SOCE by store level. Calcium influx on return of bath Ca2+ cannot be attributed to currents through VGCCs or to changes in membrane potential. Although the pharmacological tools for manipulating SOCE are poorly developed, blockers that have proved successful in other reports, including La3+, TRIM, and 2-APB, showed efficacy in reducing the readdition transient in sensory neurons. These observations provide strong inferential support for the existence of SOCE in sensory neurons.
We have also been able to obtain direct evidence of SOCE in neurons through the recording of ICRAC. Although this current is very small at RMP, we identified an inward current component during step hyperpolarizations or ramp depolarizations that depended on the presence of external Ca2+ and exhibited inward rectification, which are characteristics of ICRAC. The development of large inward Na+ flux on removal of divalent cations from the bath solution has been shown to be a typical feature of SOCE in nonexcitable cells and expression systems (Hoth and Penner, 1993; DeHaven et al., 2007). We observed a similar phenomenon in sensory neurons. The modulation of the DVF current by store level and bath-applied La3+, an established blocker of SOCE, confirms this current as a representation of ICRAC.
We additionally sought direct evidence of the presence of a SOCE mechanism in sensory neurons through identification of the molecular components underlying the process. There is now substantial agreement on the collaborative roles of STIM1 as the sensor of stored Ca2+, and of Orai1 as the pore-forming protein seated in the plasmalemma. Expression of these two proteins alone is adequate to generate SOCE, and their colocalization as puncta occurs on store depletion (Putney, 2007a,b). The involvement of canonical transient receptor potential channels (Ong et al., 2007) or the TRPV1 channel (Liu et al., 2003) in store-regulated Ca2+ influx has been proposed, but their participation in SOCE under the control of STIM1 is unlikely (DeHaven et al., 2009). Previous studies of neuronal tissues have identified both STIM1 and Orai1 in the brain, especially the cerebellum (Klejman et al., 2009), and have located STIM1 in the fetal peripheral nervous system, including the DRG (Dziadek and Johnstone, 2007; Gasperini et al., 2009). Our new data demonstrate the expression of both STIM1 and Orai1 at the transcript and protein levels in adult sensory neurons, thereby establishing that the molecular hardware for SOCE activity is present in these cells. Our anatomic observations show that STIM1 is present in all neurons, with no difference in intensity between groups of differing neuronal diameter. This is consistent with our findings that SOCE is present during bath Ca2+ readdition in all neurons, without difference in magnitude between subgroups. We identified expression of STIM1 also in satellite glial cells. There is a growing recognition that the close apposition between the sensory neuronal soma and its surrounding satellite glial cells constitutes a functional unit (Hanani, 2005), which may also apply to Ca2+ signaling through a shared extracellular Ca2+ pool.
SOCE serves a clear purpose in nonexcitable cells such as epithelial and blood cells, by providing the dominant Ca2+ entry pathway for replenishing stores and for sustained elevations of [Ca2+]c. Identifying functional roles of SOCE in excitable cells that are equipped with high-conductance Ca2+ entry pathways is at an early stage. Support of the resting [Ca2+]c by SOCE has been surmised from the observation of depressed [Ca2+]c after removal of bath Ca2+ in neurons of both the central and peripheral systems (Lipscombe et al., 1989; Nohmi et al., 1992; Wanaverbecq et al., 2003; Szikra et al., 2009), which we have confirmed in adult sensory neurons. Depression of resting [Ca2+]c in neurons may increase sensitivity of the TRPV1 channel (Cholewinski et al., 1993), decrease sensitivity to thermal stimuli (Guenther et al., 1999), and trigger apoptosis (Tsukamoto and Kaneko, 1993; Galli et al., 1995; Bian et al., 1997; Wei et al., 1998). We also identified a dependence on SOCE for maintenance of releasable intracellular Ca2+ stores and their replenishment after release, which confirms previous findings in embryonic sensory neurons (Usachev and Thayer, 1999; Cohen and Fields, 2006). In the absence of SOCE, a quiescent neuron would potentially suffer depletion of stores, which may trigger ER stress that involves accumulation of unfolded protein, global suppression of protein synthesis, and activation of a variety of transcription factors, resulting in neuronal dysfunction and apoptosis (Paschen, 2001). Investigations of long-term potentiation in the hippocampus (Emptage et al., 2001; Baba et al., 2003) have revealed that presynaptic SOCE contributes to synaptic plasticity, so spinal cord dorsal horn plasticity may be similarly dependent on sensory neuron SOCE.
The Ca2+ that enters a sensory neuron during AP-induced depolarization provides membrane stabilization through the activation of Ca2+-sensitive K+ channels (Hogan et al., 2008; Lirk et al., 2008) The resulting afterhyperpolarization and diminished input resistance that follows each AP limits the impulse generation rate, or may fully eliminate repetitive firing. Release of stored Ca2+ (CICR) contributes to this regulation of spike frequency (Gemes et al., 2009). Additionally, data in the present report show that SOCE functions to suppress neuronal excitability. This raises the possibility that a loss of SOCE contributes to the excessive excitability noted in DRG neurons proximal to an injury (Devor and Seltzer, 1999; Sapunar et al., 2005). Our data have not supported this hypothesis. Rather, we have found that axotomized sensory neurons display amplified SOCE function under baseline conditions. However, pharmacological store depletion reveals an unchanged maximal efficacy of SOCE, and there is no evidence of altered levels of transcript or protein for STIM1 or Orai1 after injury. Combined with our previous observations of decreased releasable Ca2+ stores and decreased concentration of Ca2+ in the ER lumen after injury (Rigaud et al., 2009), our findings suggest the persistence in axotomized sensory neurons of normally functioning SOCE that is driven into a high activity state by elevated STIM1 triggered by depletion of stores.
Although a normal functioning feedback control response to store depletion satisfactorily explains elevated SOCE activity after neuronal injury, alternative mechanisms accounting for SOCE upregulation might be considered. Recent work has identified regulation of SOCE by signaling pathways involving phosphoinositides (Korzeniowski et al., 2009), tyrosine kinase, which potentiates SOCE (McElroy et al., 2009), and protein kinase C, which inhibits SOCE through phosphorylating Orai1 (Kawasaki et al., 2010), but there is no direct evidence that shifts in these factors contribute to stimulating SOCE function after injury. Ca2+/calmodulin-dependent protein kinase II (CaMKII) activates SOCE (Machaca, 2003), but we have found decreased, rather than increased, CaMKII activity in sensory neurons after injury (Kawano et al., 2009; Kojundzic et al., 2010), so the increase of SOCE that we see in injured neurons occurs despite a loss of CaMKII activity.
A supportive role of SOCE in neuronal homeostasis is indicated by the pathogenic consequences of its loss. Although there has been only limited exploration in neurons, the importance of SOCE in neurological disease is highlighted by identification of diminished SOCE in neurons from mice with presenilin-1 mutations related to familial Alzheimer's disease (Yoo et al., 2000), and the direct inhibition of SOCE in hippocampal neurons by mutant presenilin-1 (Herms et al., 2003). In our present study, however, we have identified an apparent compensatory role, since SOCE is increased after injury. Although resting [Ca2+]c and stores are both reduced by injury, the withdrawal of SOCE has a proportionately greater depressive effect on these factors after injury compared with the healthy state. Our data also confirm findings by others (Park et al., 2010; Wang et al., 2010) that store depletion inhibits VGCC function, which elevates excitability in sensory neurons (Lirk et al., 2008). Therefore, supporting Ca2+ store levels by the concurrent activation of SOCE is particularly important so CICR can provide a means to augment the otherwise reduced activity-induced cytoplasmic Ca2+ signal. Our findings expose a particular dependence of injured neurons on SOCE for Ca2+ homeostasis and functional regulation, and highlight the utility of SOCE as a restorative mechanism.
Footnotes
-
This work was supported by National Institutes of Health Grants NS-42150 (Q.H.H.) and DA-K01 02475 (H.-E.W.).
-
We thank Dr. Ranjan K. Dash (Department of Physiology, Biotechnology, and Bioengineering Center, Medical College of Wisconsin) for expert assistance.
- Correspondence should be addressed to Dr. Quinn H. Hogan, Department of Anesthesiology, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226. qhogan{at}mcw.edu
