Methods
Animals
All procedures were approved by the Animal Care and Use Committee (IACUC) of the Medical College of Wisconsin.
Induction of Experimental Neuropathic Pain
We randomized male Sprague-Dawley rats (120-140 g) by chance either to a spinal nerve ligation (SNL) or to a control group. All surgical procedures were performed under general anesthesia with isoflurane 1.5-2% in O
2. Ligation of the lumbar 5
th and 6
th spinal nerve was performed by the technique of Kim and Chung [
11], as we have described previously [
8‐
10].
Behavioral Testing, Tissue Harvesting and Cell Dissociation
Neuropathic pain following SNL was confirmed by behavioral testing for ipsilateral mechanical hyperalgesia at the 10
th, 12
th and the 14
th day after surgery [
8,
12]. After 30 min of rest we applied the tip of a 22 gauge spinal needle on the plantar surface of both hind paws in random order with pressure adequate to indent but not penetrate the plantar skin. Each needle application produced either a normal brief reflexive paw withdrawal, or a hyperalgesia-type response that included sustained (> 1 s) paw lifting, shaking, and grooming [
12]. Rats with probability of hyperalgesia ≥ 20% were considered as expressing neuropathic pain phenotype. These constituted 81% of the animals subjected to SNL. We determined that this threshold of 20% ensures optimal sensitivity and specificity for detecting neuropathic pain phenotype after SNL based on analysis of receiver operating plots from previous studies [
12].
Dorsal root ganglia were harvested between the 17
th and 28
th day after control or SNL operation, from rats euthanatized by decapitation under deep isoflurane anesthesia. For the electrophysiological experiments and confocal microscopy studies, we dissociated DRG neurons as described previously [
8].
RT-PCR for KATP Channel Subunits
Total mRNA was isolated from neurons dissociated from one L5 control DRG at each time. For cell isolation each ganglion was enzymatically dissociated in 0.5 ml DMEM/F12 (Dulbecco's modified Eagle's medium F12; Gibco, Invitrogen Corp., Carlsbad, CA) with 0.025% w:v liberase blendzyme 2 (Roche Diagnostics Corp., Indianapolis, IN) for 30 min in an incubator at 37°C. After centrifugation and removal of the supernatant, a second incubation at 37°C followed for another 30 min in 0.25 ml DMEM with 0.0625% trypsin from bovine pancreas (10,000-15,000 BAEE units/mg protein; Sigma, St. Louis, MO; Cat.No. T8802) and 0.0125% deoxyribonuclease 1 from bovine pancreas (≥ 2,000 Kunitz units/mg protein; Sigma, St. Louis, MO; Cat.No. D5025) in 0.25 ml DMEM.
After adding 0.25 ml trypsin inhibitor 0.1% w:v (Sigma), cells were centrifuged (600 rpm for 5 min). For RNA isolation cells were treated with TRIzol reagent (Invitrogen Life Technology) following the supplier's protocol. Isolated RNA was incubated with DNAse (Ambion, Austin, TX) to eliminate DNA contamination, and then 1.2 μg RNA was used for cDNA generation using the Retroscript Synthesis Kit (Ambion). The segment of interest was reverse-transcribed for 2 hours at 42°C.
PCR was performed with 1.5 μl cDNA in total 25 μl using specific primers for Kir6.1 [forward (fw): GGATAATCCCATCGAGAGCA, reverse (rv): CTCAGCCACTGACCTTGTCA), Kir6.2 (fw: TCCAACAGCCCGCTCTAC, rv: CAGCGTTTTGTCCCCATC), SUR1 (fw: TGAAGCAACTGCCTCCATC, rv: GACAAGCCGGAAAAGCTTC) or SUR2 (fw: ACCTGCTCCAGCACAAGAAT, rv: CCTGGTCATTGTGATGAAGAGA)] and Taq DNA polymerase (Roche). The 35 cycles included initial denaturation at 94°C, 4 min; 94°C, 10 sec; 56°C, 30 sec; 72°C, 2 min; and final extension 72°C, 10 min. In negative control experiments reverse transcriptase was omitted. PCR products (12 μl) were separated by electrophoresis on a 1.5% agarose gel stained with ethidium bromide and visualized with a Molecular Imager (Bio-Rad, Hercules, CA). The analysis was repeated with neurons dissociated from DRG of three different control rats.
Western blotting
Dorsal root ganglia, brain and pancreas from control rats were immediately frozen in liquid nitrogen after removal and stored in -80°C. After thawing in ice, they were homogenized in 100 μl lysis buffer (in mM: 20 MOPS, 2 EGTA, 5 EDTA, 30 NaF, 40 β-glycerophosphate, 10 Na-pyrophoshate, 0.5% NP-40). Samples were prepared for SDS-PAGE by 1:1 dilution with Laemmli buffer, and 30 μg of protein was loaded in each well. Staining with Ponceau S solution was used to verify protein transfer and protein loading to a PVDF membrane. Blots were incubated with 5% non-fatty dry milk (NFDM), 3% bovine serum albumin in 0.1% TBS-T solution for 1 h and then with antibodies (1:200) against SUR1, SUR2, Kir6.2 and Kir6.1, in 1% NFDM overnight at 4°C. Three 15 min washes in TBS-T followed before staining with each secondary antibody (goat anti rabbit HRP-IgG, Santa Cruz 1:10.000) for 2 h, at room temperature. The ECL-Plus chemiluminescence detection kit was used to detect the antigen-HRP conjugated antibody complex (GE Healthcare, Amersham, PA). Each determination was performed 3 times with tissue from different animals.
Immunohistochemistry
Dorsal root ganglia were cryoprotected in 4% PFA with 15% sucrose in 0.1 M PBS for 1 h, followed by 30% sucrose in 0.1 M PBS overnight. Subsequently DRG embedded in Tissue-Tek
® OCT, were sectioned (5 or 12 μm) with a Leica cryostat (Jung CM 1800; Vienna, Austria), plated onto Gatenby's solution subbed slides, and post-fixed in 4% PFA with 4% sucrose. After blocking with 4% normal goat serum for 1 h at room temperature, slides were incubated in polyclonal rabbit antibodies to Kir6.1, Kir6.2 (1:500, Alomone, Jerusalem, Israel), and SUR1 or SUR2 (1:50, Santa Cruz Biotechnology, Santa Cruz, CA). Some samples were also co-labeled with antibodies against neurofilament 200 (NF 200; 1:1000, Abcam, Cambridge, UK) and calcitonin-gene related peptide (CGRP; 1:50; Santa Cruz). The nodes of Ranvier were identified with contactin-associated protein 1 (Caspr, 1:500; clone K65/35, UC Davis/NINDS/NIMH NeuroMab Facility, University of California, Davis, CA 95616), which is localized in paranodal sites [
13,
14]. Samples were incubated at 4°C overnight. After series of washes, samples were stained with goat anti-rabbit Texas Red (1:500, Jackson Immunoresearch, USA), or goat anti-rabbit Alexa 546 (1:500, Molecular Probes, USA), or goat anti-mouse Alexa Fluor 488 (1:1000, Molecular Probes, USA) as secondary antibodies. DAPI (1:1000; Sigma, St Louis, MO, USA) was used for nuclear staining to distinguish neuronal somata from satellite glial cells. All antibodies were diluted in 1 × GDB containing 0.3% Triton. Slides were mounted with Shur/Mount medium and cover-slipped. For negative controls, only the secondary antibodies were applied. Rat pancreas, brain, and vascular tissues were used as positive controls.
To identify that the pore forming Kir6.2 subunits co-localize with the regulatory SUR1 subunits in DRG as they do in other tissues, we co-labeled some samples with antibody against Kir6.2, with FITC conjugated-glybenclamide (ER-Tracker Green BODIPY-Glybenclamide, Molecular Probes) [
15], and with DAPI. Specifically, frozen sections treated with anti-Kir6.2 were incubated with BODIPY-Glybenclamide (20 nM) in 0.1 M PBS for 60 min in room temperature, followed by two washes and DAPI staining. For negative controls, slides were incubated with 1 μM unlabeled glybenclamide for 30 min in room temperature prior to addition of 20 nM FITC conjugated-BODIPY-glybenclamide for 60 min.
Images were obtained using a digital camera and the Spot Diagnostics Instruments software (Version 4.5 for Mac OS X) through a Nikon Eclipse E600 (Nikon, Japan) microscope. Confocal microscopy employed a Nikon Eclipse TE 200-U (Nikon, Japan) equipped with EZ C1 laser scanning software. Images were captured using a 30 μm emission pinhole with a 40× objective.
The criteria used in analysis were consistent throughout the study of all images. The inspection and analysis of samples was performed by the same investigator, who was blinded to the groups of the animals the samples were obtained from. The prevalence and intensity of immunofluorescence in somata and nerve fibers were measured using the Metamorph (Version 7.1; Downingtown, PA). The intensity of fluorescence within cell targets was compared to background, as has been previously used in DRG immunostaining analysis [
16,
17]. Target was considered positive when intensity in area of interest was threefold greater than in background. This was confirmed by subjective evaluations of randomly chosen images. By this way, there was a consistent agreement between areas considered positive by direct visual inspection and areas exhibiting intensity/background ratio >3. Only cells with clearly visible nuclei were selected for further analysis. We measured the areas of the neuronal somata using the ImageJ software (Java, version 1.5.0-13; Wayne Rasband, Research Services Branch, NIH, Bethesda, MD).
Electron microscopy (EM)
For preparing samples for immuno-EM, freshly dissociated rat DRG tissues were high pressure frozen, freeze substituted in ethanol and 0.5% uranyl acetate, and embedded in Lowicryl K11 M resin. Ultrathin sections were prepared on formvar/carbon coated grids and incubated with anti-SUR1 or anti-Kir6.2 antibodies, washed, and then labeled with goat anti-rabbit antibodies conjugated to 10 nm colloidal gold. Sections were examined in a Hitachi H600 TEM 9 (Hitachi, Japan) and images recorded using an AMT 1K camera (Danvers, MA).
For preparing samples for measuring SLI by EM, a mixture of 2% glutaraldehyde and 4% PFA in 0.1 M sodium cacodylate buffer (pH 7.2) containing 0.5 mM CaCl
2 was applied onto exposed rat DRG in situ. After approximately 30 s, fixed pieces of the DRG and nerve were excised and placed into fresh fixative (as above) and fixed for a further 1 h on ice. Tissues were then washed in ice-cold buffer, 3 × 5 min changes, then post fixed with 1% osmium tetroxide and 1.5% potassium hexacyanoferrate for 2 h on ice. Following post fixation, the tissues were rinsed 3 × 5 min in distilled water, dehydrated through graded methanol and embedded in Embed 812 resin. Ultrathin (60 nm) sections were cut, stained with saturated uranyl acetate in 25% ethanol followed by Reynolds lead citrate and examined in a JEOL 2100 TEM (Japanese Electron Optics, Ltd). The areas of SLI were measured using the ImageJ software by an investigator blinded to the experimental group. For consistency reasons, we measured only incisures that traversed the whole width of the myelin sheaths (complete incisures) [
18]. Only longitudinal tissue sections were obtained for analysis. We examined sections from three separate sampling sites: from the L5 spinal nerve proximal to ligation, from within the DRG, and from the dorsal root proximal to DRG. The investigator who captured the images of SLI was blinded to the experimental group.
Electrophysiological recordings
We used electrophysiological recordings to further verify the specificity of our antibody against the SUR1 subunit, and to confirm that staining is localized in the plasmalemmal membranes of neuronal somata. We tested whether the anti-SUR1 antibody affected the blocking effect of glybenclamide on K
ATP single-channel currents recorded using the excised inside-out configuration of the patch-clamp technique. For this reason neurons from control (SS) rats were preincubated with anti-SUR1 antibodies (1:60) for 2 h and compared with neurons preincubated in control solution without antibody at 37°C for the same time. Single-channel K
ATP currents were recorded from large diameter neuronal somata at room temperature (20-25°C) as described previously [
8‐
10]. Glybenclamide, a selective K
ATP channel blocker [
19], was obtained from Sigma and diluted daily from 50 mM stocks in DMSO kept at 4°C. The bath (intracellular) solution contained (in mM): 140 KCl, 10 Hepes, 5 EGTA, and 2 MgCl
2. The pipette (extracellular) solution contained (in mM): 140 KCl, 10 Hepes, 10 d-glucose, and 0.5 EGTA. The pH of all solutions was adjusted to 7.4 with KOH. Membrane potential was clamped at -60 mV. A conventional 50% current amplitude threshold level criterion was used to detect open events. Channel open probability (Po) was determined from the ratios of the area under the peaks in the amplitude histograms fitted by a gaussian distribution. Channel activity was calculated as NPo, where N is the number of observed channels in the patch, from data samples of 30 s recordings duration in the steady state. Mean open time values were estimated from patches that contained predominantly one type of single channel opening. The NPo of the K
ATP channels was normalized to the baseline NPo value obtained before the drug at bathing solution without ATP, indicating relative channel activity (relative NPo).
Statistical analysis
Data were compared between control and SNL DRG, or between NF200 positive or negative neurons by Student's or Fisher's exact tests, as appropriate, using the SPSS 16.0 for Mac statistical software (SPSS Inc, Chicago IL). Concentration-response curves were plotted using the Prism 4 for Macintosh (GraphPad Software, Inc) and compared using ANOVA. Areas of SLI were also compared between groups and sites of sampling by ANOVA, using the univariate general linear function of the SPSS, followed by Bonferroni post hoc tests or Student's t test whenever appropriate.
Discussion
SUR1 or SUR2 subunits always co-assemble with either Kir6.1 or Kir6.2 subunits into functional K
ATP channel octamers [
2,
19]. Using immunostaining and Western blots, we showed that peripheral sensory neurons express SUR1, SUR2 and Kir6.2 protein, but not Kir6.1 protein. We identified this co-localization of Kir6.2 with SUR1 subunits by staining with antibody against Kir6.2, and with BODIPY-Glybenclamide, which specifically binds to SUR1 with high affinity at concentrations <40 nM [
15]. Our findings from these experiments highlight the presence of K
ATP channels in control and axotomized neurons. K
ATP channels in DRG neurons are of the Kir6.2/SUR1 (pancreatic-neuronal) and Kir6.2/SUR2 (cardiac) subtype.
Since suboptimal specificity may limit the reliability of immunostaining techniques [
31], we confirmed the specificity of our antibodies by Western blots. The size of the bands that we detected are consistent with those reported in other studies that have identified K
ATP channel subunits. Furthermore, we functionally confirmed the specificity of our anti-SUR1 antibody by electrophysiological recording, in which antibody application eliminated glybenclamide-sensitive current in excised membrane patches. Use of isoflurane anesthesia as well as the duration of hypoxia may have affected the recordings of potassium currents, as described previously [
32]. However, it is unlikely that these factors may have confounded the results because all rats were exposed to the same conditions of anesthesia and the duration of hypoxia. All these procedures were performed by the same investigator in the same fashion and by the same protocol for each animal. Antibodies directed against channel components such as the SUR1 regulatory subunit may alter active sites of the channel or its interaction with ligands, and thus can be used as specific inhibitors [
33‐
35].
Our RT-PCR technique detected mRNA encoding all K
ATP channel subunits. However, neither immunostaining nor Western blots detected any Kir6.1 at the protein level. Either immunohistochemistry and Western blots were not sensitive enough to detect Kir6.1 protein in DRG neurons, or Kir6.1 mRNA is not translated into protein in these neurons [
36,
37]. However, the same antibodies did not fail to detect Kir6.1 in brain tissue, which was used as control, indicating competence of our antibodies, and, therefore the lack of translation of Kir6.1 mRNA to protein in DRG neurons.
The biophysical and pharmacological properties of the K
ATP current recorded in DRG neurons indicate that the predominant membrane channel subtype is the Kir6.2/SUR1, since glybenclamide selectively blocks this subtype in the concentration range used in our experiments [
10,
19]. Immunostaining against SUR2 implies the presence of non-functional Kir6.2/SUR2 subunits.
We were able to identify the distribution of functional K
ATP channels and their alterations by axotomy by staining for SUR1 subunits. Previous reports have validated the use of anti-SUR1 only to study the cellular distribution of K
ATP channels [
38]. Identification of SUR1 subunits in membrane and SLI represents post-translational co-expression with pore-forming subunits, because SUR and Kir subunits always combine in functional K
ATP channels. In contrast, unbound SUR subunits are unstable [
39,
40].
In addition to neurons, K
ATP channels are present in glial satellite and Schwann cells, which are known to express K
+ currents [
41‐
43]. Since K
ATP channels are inward rectifiers, our findings are consistent with reports that Schwann cells express inward rectifying K
+ currents [
44]. Our findings that satellite and Schwann cells express K
ATP channels also help clarify a controversial issue about the presence of Kir6.2-containing K
ATP channels in glial cells [
45]. K
ATP channels in these cells, in cooperation with other Kir channels, may convey glial cell-mediated clearance of extracellular K
+, often termed "K
+ spatial buffering" or "K
+ siphoning" [
45,
46].
A consistent finding of our study is the presence of K
ATP channel subunits in nuclear envelopes of DRG neurons. These channels may regulate gene expression, as have identified in other tissues. Specifically in pancreatic β-cells, K
+ fluxes via nuclear K
ATP channels release Ca
2+ stored in the nuclear envelope leading to CREB phosphorylation and gene expression [
47]. It is therefore possible that K
ATP channels in the nuclei of DRG neurons play similar roles.
Kir6.2/SUR1 subunits have not been previously identified in peripheral nerve fibers, although inward rectifying K
+ channels have been detected in paranodal sites [
42,
44,
48]. The molecular identity of these inward rectifiers had not been identified previously, but we now show that both SUR1 and Kir6.2 co-localize with Caspr + paranodal areas in axons. This finding implies that K
ATP channels are the first inward rectifying channels identified in paranodal sites. These sites and the nodes of Ranvier have high metabolic demands [
49]. It is thus possible that K
ATP channels regulate excitability at paranodal sites, and may safeguard these regions of high metabolic activity from injury during energy depletion [
50].
Our novel finding that K
ATP channels are present in SLI, is supported by fluorescent microscopy, and more specifically by EM data, which together confirm the presence of SUR1 subunits in SLI. Schmidt-Lanterman incisures are accumulations of Schwann cell cytoplasm enclosed within the myelin lamellae all along the internodal length of the axon, and also adjacent to nodes of Ranvier [
18,
28]. These structures contain abundant gap junctions that mediate intra- and intercellular communications [
51]. K
ATP channels regulate opening of gap junctions in other cell types [
52], including astrocytes [
53], and they may do so in SLI as well. SLI are involved in metabolic functions of the myelin sheath and axon, including growth, maintenance and axonal support [
54]. Because of these functions, SLI have high metabolic activity and energy demands, which are elevated following nerve injury [
50]. It is thus possible that K
ATP channels in SLI mediate protection against energy depletion.
Part of the supportive functions of glial satellite and Schwann cells is the synthesis and transfer of ion channels to neighboring axons [
43,
55]. Indeed, an exchange of K
+ channels between Schwann cells and neuronal axons has been shown [
56]. Neurons need to supply 500-2000 nodes with channels that undergo rapid turnover. However, because axonal transport from the soma is not always sufficient, especially to the most distant parts of the fiber, Schwann cells may provide these channels to axons through SLI intercellular communications. Therefore, the SUR1 subunits that we have detected in the SLI may be parts of K
ATP channels in transit towards the adjacent axons, or may be functional constituents of the SLI themselves.
Neurons axotomized by SNL express K
ATP channel subunits at the same cellular and histologic locations as controls. However after injury, NF200+ somata are less likely to express SUR1 subunits less frequently than control somata. Like other types of K
+ channels [
57‐
59], K
ATP channels decrease after nerve injury, due to their downregulation or to their redistribution to axons.
A novel observation of our study is that the prevalence of SUR1 positive SLI, as well as the intensity of SUR1 immunofluorescence in the SLI, decrease after nerve injury. Our ultrastructural data from EM show that the geometrical area of the SLI after nerve injury is reduced, which may explain diminished SUR1 immunofluorescence. This further indicates disruption of myelin architecture proximal to the site of axotomy, consistent with other reports indicating altered SLI architecture in various peripheral neuropathies [
60], including Charcot-Marie-Tooth [
51].
Downregulation or decreased K
ATP channel activity may open gap junctions between SLI, between adjacent Schwann cells, and also between SLI and regenerating axons, as they do in other tissues [
52,
53]. These open gap junctions in Schwann cells transport ATP and other metabolic substrates to axons [
51]. Our observations of decreased SUR1 at these sites after injury suggests that withdrawal of K
ATP channel activity may support neuronal regeneration.
However, considering the alterations we observed in injured by SNL neurons, dissociated from rats with hyperalgesia, only comparison with neurons from rats subjected to the same treatment (SNL) but not exhibiting hyperalgesia would indicate whether these changes are pertinent in the pathogenesis of neuropathic pain. Thus, rats not exhibiting neuropathic pain phenotype after SNL would be appropriate to serve as an additional control group, the lack of which constitutes a limitation of our study.
We have previously reported the presence of K
ATP channels in DRG neuronal somata using electrophysiological recordings at the single-channel and whole-cell current level. We now identify the presence of the specific channel subunits that conduct K
ATP current not only in DRG somata but also in peripheral sensory axons, glial satellite and Schwann cells, including SLI. K
ATP channel opening is necessary to control excitability and neurotransmitter release in peripheral sensory neurons, as we have previously shown [
9]. Another limitation of this and our previous studies is a lack of description of the distribution and role of K
ATP channels at the presynaptic afferent terminals in the dorsal horns, although their presence at these sites is highly likely [
61]. K
ATP channels may also contribute to neuroprotective functions in DRG as in other neurons [
6].
Our previous work has emphasized the role of KATP currents in DRG neurons, which are the primary determinants of sensory function. However, lack of the functional confirmation of KATP currents in glial cells in contrast to DRG neurons is another limitation. Further exploration of the role of KATP channels in glial cell is an important topic, but the necessary studies should include extensive electrophysiological characterization and additional experiments that will be the subject of separate future investigations.
In glial satellite and Schwann cells, K
ATP channel opening may also support metabolic functions, including K
+ siphoning [
45]. However, in the peripheral nerves, recovery from injury may also benefit from loss of K
ATP channels, which facilitates restoration through gap junction opening and facilitated axonal regeneration.
Because of these diverging roles of KATP channels in different neuronal and non-neuronal sites, opposing effects of KATP channel opening in SLI and myelinated nerve fibers after painful nerve injury are likely. However, this is based on speculation. There is a possibility that the effects on one site may override these on another site, therefore further studies are necessary to elucidate the in vivo effects of peripheral application of KATP channel openers for treating neuropathic pain and neurodegenaration.
Authors' contributions
VZ, MYL, DW and BM carried out the immunohistochemical studies. VZ blindly evaluated the images and performed the measurements using Metamorph and ImageJ. TK carried out the electrophysiological experiments and data analysis. VZ, TK, and GG carried out animal surgery and behavior testing. VZ and MB carried out molecular biology experiments. VZ, MYL and CS conceived the study, and participated in its design and coordination. VZ, CS and QH wrote the manuscript. All authors have read and approved the final manuscript.