|Home | About | Journals | Submit | Contact Us | Français|
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Changes in sodium channel activity and neuronal hyperexcitability contribute to neuropathic pain, a major clinical problem. There is strong evidence that the re-expression of the embryonic voltage-gated sodium channel subunit Nav1.3 underlies neuronal hyperexcitability and neuropathic pain.
Here we show that acute and inflammatory pain behaviour is unchanged in global Nav1.3 mutant mice. Surprisingly, neuropathic pain also developed normally in the Nav1.3 mutant mouse. To rule out any genetic compensation mechanisms that may have masked the phenotype, we investigated neuropathic pain in two conditional Nav1.3 mutant mouse lines. We used Nav1.8-Cre mice to delete Nav1.3 in nociceptors at E14 and NFH-Cre mice to delete Nav1.3 throughout the nervous system postnatally. Again normal levels of neuropathic pain developed after nerve injury in both lines. Furthermore, ectopic discharges from damaged nerves were unaffected by the absence of Nav1.3 in global knock-out mice. Our data demonstrate that Nav1.3 is neither necessary nor sufficient for the development of nerve-injury related pain.
Neuropathic pain is an important and unmet clinical problem. There is a lack of effective drugs for its treatment and in some cases the pain is resistant to the strongest analgesics . Hyperexcitability of damaged sensory neurons, as a result of injury to and/or demyelination of their peripheral axons, is thought to initiate the processes that lead to most neuropathic pain [2,3]. Hyper-excitable neurons can fire spontaneously, causing spontaneous pain, or become hypersensitive to otherwise innocuous mechanical and thermal stimuli giving rise to allodynia (noxious responses to innocuous stimuli) and hyperalgesia (increased perception of noxious stimuli) [2,3].
Injured sensory neurons undergo major changes in gene expression that have been catalogued in microarray studies [4-6]. Altered expression of voltage-gated sodium channels (VGSC), which underlie the electrical excitability of nerve and muscle, has been extensively studied . Ectopic activity in damaged neurons  and neuropathic pain behavior  have been shown to be sensitive to the VGSC blocker, Tetrodotoxin (TTX). Sensory neurons express multiple subtypes of TTX-sensitive and TTX-resistant VGSCs . The expression of Nav1.1, Nav1.2, Nav1.6, Nav1.7, Nav1.8 and Nav1.9 subunits is down-regulated, whereas only Nav1.3, a TTX-sensitive channel, is up-regulated following peripheral nerve injury [10-16]. Nav1.3 is expressed throughout the embryonic nervous system but is down-regulated in adults [11,17]. Coinciding with the re-expression of the Nav1.3 channel in injured neurons, the voltage-gated sodium currents recover from inactivation fourfold faster than that in uninjured neurons . This is thought to be a direct result of expression of Nav1.3 as it possesses similar inactivation kinetics when expressed in cell lines . Rapid recovery from inactivation could allow damaged nerves to fire at higher frequencies than otherwise . Therefore, it is hypothesized that mis-expression of VGSCs and in particular the re-expression of Nav1.3 is an important factor contributing to the hyperexcitability of injured neurons. Several pieces of evidence support this hypothesis. Firstly, the ectopic activity  and mechanical allodynia  associated with nerve injury has been shown to be sensitive to TTX. Secondly, Glial-derived neurotrophic factor (GDNF), which reverses neuropathic pain behaviour, reverses Nav1.3 up-regulation . Finally, intrathecal administration of antisense oligonucleotides directed against Nav1.3 mRNA reverses neuropathic pain behaviour and restores the inactivation kinetics of VGSC to that of uninjured neurons [20,21]. Taken together, these data suggest that Nav1.3 is a good target for analgesic drug development . However, a recent study using a different antisense oligonucleotide sequence directed against Nav1.3 failed to reverse neuropathic pain behaviour .
Here we describe the generation of 3 Nav1.3 mutant mouse lines to test the hypothesis that Nav1.3 has a causative role in nerve-injury-induced chronic pain . Our data shows, contrary to expectation, that Nav1.3 is neither responsible nor necessary for ectopic activity and neuropathic pain behaviour.
We used a genetic approach to investigate the contribution of Nav1.3 to setting pain thresholds, to the hyperexcitability of injured neurons and neuropathic pain behaviour in mice. We generated a floxed Nav1.3 mouse, (Figure 1a–b) and used a deletor mouse strain expressing Cre recombinase before E4  to generate a conventional global null mutant (Nav1.3 KO), (Figure (Figure1c).1c). To confirm deletion of Nav1.3 we used RT-PCR to amplify sequences between exons 2 and 10, figure figure1d.1d. Sequencing of the PCR product from the Nav1.3 KO brain confirmed the deletion of 221 bp representing the floxed exons 4 and 5, (Figure (Figure1e).1e). Sequencing of the PCR product from WT brain revealed that most of the Nav1.3 mRNA carries the adult exon five (about 80%), (Figure (Figure1f1f).
Nav1.3 KO mice were healthy, fertile, grew as well as their littermate controls (WT), (figure (figure2a),2a), and performed equally well on the rotarod, (figure (figure2b).2b). Peak sodium currents in cultured sensory neurons were unaffected by Nav1.3 gene deletion, (figure (figure2c2c).
We next examined acute pain thresholds in adult Nav1.3 KO animals. Thermal and mechanical thresholds are normal, (Figure 3a–d), indicating that Nav1.3 does not play a role in setting pain thresholds in the undamaged nervous system. Although it has been reported that Nav1.3 is up-regulated in an inflammatory pain model , we found that pain behaviour evoked by injection of formalin and CFA into the hindpaw was the same in both Nav1.3 KO and WT mice, (Figure 3e–g). This indicates that Nav1.3 does not play a role in inflammatory pain, in marked contrast to Nav1.7 KO mouse, which shows an almost complete loss of inflammatory pain .
Does Nav1.3 play a role in neuronal hyperexcitabtility following nerve injury? Voltage-gated sodium currents in injured neurons recover from inactivation fourfold faster than that in naive neurons possibly as a result of up-regulation of Nav1.3 [13,18]. Rapid recovery from inactivation could allow damaged nerves to fire at higher frequencies . The ectopic activity associated with nerve injury, which has been shown to be sensitive to TTX, is an important trigger for neuropathic pain behaviour [8,27]. We therefore measured ectopic discharges from teased fibres of the damaged (L5) and spared (L4) 24 hours after surgery in Nav1.3 KO and WT mice. L5 spinal nerve ligation induces spontaneous activity in L4 as well as L5 myelinated fibres, though the percentage of active neurons in mouse appears to be less than that in rat . The percentage of spontaneously active fibres from L5 DRG was three fold higher than that from L4 DRG, (Figure (Figure4b),4b), which is similar to earlier results in rats . Importantly, the percentage of spontaneously active fibres was not different between the Nav1.3 KO and WT, (Figure (Figure4a).4a). In addition, other characteristics of the spontaneous activity such as the number of bursts per minute, (Figure (Figure4b),4b), mean number of spikes per burst, (Figure (Figure4c),4c), and the mean burst length, (Figure (Figure4c),4c), were also not different. These findings suggest that Nav1.3 does not play an essential role in the ectopic hyperexcitability of damaged nerves.
GDNF reverses the up-regulation of Nav1.3 and neuropathic pain behaviour , and antisense oligonucleotides against Nav1.3 sequence reverse neuropathic pain behaviour [20,21]. Therefore, we measured mechanical pain thresholds in Nav1.3 KO mice following tight ligation of spinal nerve L5 (Chung model). Surprisingly, a robust mechanical allodynia developed in both Nav1.3 KO and WT mice from day 3 post-surgery and persisted throughout the experiment, (Figure (Figure5a).5a). Sodium currents in cultured sensory neurons were unaffected by Nav1.3 gene deletion, (Figure (Figure2c),2c), suggesting a lack of compensatory changes.
In order to rule out the possibility that genetic compensation had confounded the behavioural phenotype in the Nav1.3 KO, we generated two conditional knockout strains; we mated the floxed Nav1.3 mouse to the Nav1.8 Cre line  to delete Nav1.3 in nociceptors from E14, and with the NFH-Cre mouse line  to delete Nav1.3 in CNS and PNS from the 10th postnatal week. Mice in both conditional knockout strains developed a profound mechanical allodynia soon after surgery that was not different from that of littermate controls (floxed Nav1.3), (Figure (Figure5b5b and and5c),5c), confirming our results using the conventional global KO. We conclude that Nav1.3 does not play a critical role in either the induction or maintenance of neuropathic pain behaviour.
Our results can be explained as follows; the level of Nav1.3 expression in sensory neurons following injury is estimated to increase by about 1.4–2 fold [5,15] which is still very small considering the almost undetectable level of Nav1.3 in adult sensory neurons. Thus the increase in sodium currents due to the up-regulated Nav1.3 is unlikely to be significant in comparison to the total sodium current before injury. The reported fast kinetics of the Nav1.3 current  suggests that the qualitative aspect of the Nav1.3 up-regulation may be more important than the quantitative one. Our results using the global Nav1.3 KO rule out this in stimulus-evoked neuropathic pain. This is because Nav1.3 is deleted in all cell types regardless of the extent to which they up-regulate Nav1.3, the type of cells involved (small vs large sensory neurons and DRG vs central neurons) and the sub-cellular localization of Nav1.3 (somata vs terminals). Furthermore, although some antisense studies suggest a causal role for Nav1.3 in neuropathic pain [20,21], the specificity of the oligonucleotides used is not absolute (of 21 bases 15 are identical to Nav1.1, 17 to Nav1.2 and 14 to Nav1.6), and another antisense study using different specific oligonucleotides has failed to confirm these findings .
The role of VGSCs in the pathogenesis of neuropathic pain has recently come under increasing scrutiny. Changes in neuronal hyperexcitability have not been found to correlate with changes in sodium currents . Nevertheless, the effectiveness of non-subtype selective sodium channel blockers in treating neuropathic pain  has stimulated the search for a specific underlying sodium channel subunit against which, more potent and selective analgesics can be developed. It seems likely, however, that no single sodium channel is specifically associated with neuronal hyperexcitability in damaged nerves, but that aberrant expression of high densities of sodium channels at the damaged terminals of sensory neurons may result in hyperexcitability . In support of this we have recently shown that neither Nav1.7 nor 1.8 is necessary for the development of neuropathic pain . Nav1.9 null mutants have also been shown to develop neuropathic pain normally . By contrast, it is clear that Nav1.7 plays a major role in the development of inflammatory pain and subtype specific channel blockers directed against this channel may be desirable for the treatment of inflammatory pain.
The role in neuropathic pain pathogenesis of the other major VGSC subunits in DRG, Nav1.1, Nav1.2 and Nav1.6, remains to be investigated. Because of their relative abundance it is expected that deletion of these subunits may not have a specific effect on neuropathic pain. Almost certainly the usefulness of Nav1.1, Nav1.2 or Nav1.6 blockers in treating neuropathic pain will be greatly undermined by their broad expression in the CNS .
Neuropathic pain may be more amenable to local treatment with broad-spectrum sodium channel blockers, rather than those targeted at a specific sodium channel subunit such as Nav1.3.
An RCPI-22 129S6/SvEvTac mouse BAC library was screened using a probe representing exons 4–12. DNA from positive BAC clones was prepared and subcloned into pBluescript (BS-SKII-). Three subclones were obtained that covered Nav1.3 sequences from exon 2–8. The construct was prepared in three steps. Firstly, the 3' arm, Neomycin cassette and one LoxP site were cloned together as follows; two oligonucleotides were cloned in BS-SKII- to form new cloning sites. One LoxP site was cloned as an EcoRI-PstI fragment from PLneo vector . The 3' arm was cloned as ClaI-BamHI 7 Kb fragment. Neomycin cassette was cloned as BamHI-SalL fragment from Py6.0 plasmid . Secondly, the 5' arm, one LoxP and the floxed exons were cloned together as follows; 7 kb fragment ClaI-SacI was cloned in BS-SKII-. A LoxP site was inserted into the BsmBI site of the genomic fragment through blunt ligation. Finally, the two halves above were cloned into a vector containing the Thymidine-Kinase (TK) cassette as follows; the 3' side was prepared as Not-ClaI fragment, the 5' side was prepared as ClaI-SacII fragment and the TK containing vector was prepared as Not-SacII fragment. All were ligated together in the same tube. Positive clones were confirmed by restriction digests and sequencing. Blastocyst injection was performed by Monica Mendelson (Colombia University). Chimeras were crossed to C57BL/6 and germ line transmission tested using Southern blotting of tail DNA derived from F1 offspring. F1 heterozygotes were crossed to FLPe deletor animals  to excise the positive selection marker.
All tests were approved by the United Kingdom Home Office Animals (Scientific Procedures) Act 1986. They were performed in a Home Office designated room at 22 ± 2°C. Tests were carried out as previously descried . Formalin test: 20 μl of 5% formalin (37% formaldehyde (BDH) diluted with 0.9% saline (Sigma)) was injected subcutaneously into the plantar aspect of a hind-paw using a Hamilton syringe and 27 g needle and animals were placed in a 20 × 20 × 20 cm Plexiform chamber. The amount of time spent licking or biting the injected paw was then recorded in 5 minute periods over the next hour. Animals were habituated to the test environment for at least 1 hr before experiments were started. Activity during the first phase, from 0–10 mins, and the second phase, from 10–60 mins, was then calculated for each animal.
Neuropathic pain was induced according to the Chung Model (ligation of spinal nerve L5). Baseline mechanical thresholds for one paw were measured from mice using von Frey hairs (using the up-down method) respectively. Animals were anaesthetised using Halothane. A midline incision was made in the skin of the back at the L2-S2 levels and the left paraspinal muscles separated from the spinal processes, facet joints and transverse processes at the L4-S1 levels. The L5 transverse process was removed and the L5 spinal nerve tightly ligated using 8-0 silk thread.
DNA was precipitated from DRG and brain after overnight incubation in lysis-buffer (100 mM Tric-Cl, 5 mM EDTA, 0.2% SDS, 200 mM NaCl and 0.1 mg/ml Proteinase K). DNA pellet was resuspended in 200 μl TE and 1 μl was used for each 25 μl PCR reaction. Nav1.3 Primers used are (GAGAGAAAGACACTTAAATGCAGACATC), (GCTTTTTGTTCAAGTCTATCATATTCAAAG) and (AAG GAT GGC ATC ACC CAC AAG).
Deletion of exons 4 and 5 was confirmed by reverse transcription PCR (RT-PCR). RNA isolation was performed using TRIzol Reagent (Invitrogen, Paisley, UK), according to manufacturer's protocol. The reverse transcription reaction was also performed according to an Invitrogen protocol, using random primers. Primers used were (CACTAAACCCCGTTAGGAAAATTGCT) and (TCCAAGTGCTCCCTCTGTCTCCTC). The same primers were used for sequencing the RT-PCR bands.
Animals were terminally anaesthetised with 150 mg/kg sodium pentobarbitone (Rhône Mérieux), and then intracardially perfused with ice cold PBS (0.14 M NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.7 mMKH2PO4) followed by ice cold 4% paraformaldehyde/PBS. The tissues were removed and placed in O.C.T compound. 10 μm sections were prepared and mounted on Superfrost Plus slides (BDH). Sections were fixed in 4% paraformaldehyde for 5 mins and washed 3 × 10 mins in PBS + 0.1% Triton-X. They were blocked and permeabilised in 10% goat serum (GibcoBRL)/PBS + 0.1% Triton-X for at least 3 hrs at room temperature. Sections were then incubated in the following antisera in 10% goat serum/PBS + 0.3% Triton-X overnight at 4°C: rabbit anti-beta-galactosidase (5'-3') (1:500), mouse anti-peripherin (Chemicon) (1:1000) or mouse anti-neurofilament (Sigma N-52) (1:1000). They were then washed 3 × 10 mins in PBS + 0.1% Triton-X. The primary antisera were localised by immunofluorescence with Alexa Fluor 594 (Molecular Probes) (1:1000) and Alexa Fluor 488 (Molecular Probes) (1:1000). The slides were finally washed 3 × 30 mins at 4°C and mounted using Aqueous Mounting Solution (Sigma).
Cultures of dorsal root ganglion (DRG) neurons were prepared from adult Nav1.3 null and WT littermate mice. The animals were killed by cervical dislocation, and dorsal root ganglia (DRG) were removed and enzymatically dissociated (Dispase/collagenase; Sigma, Poole, Dorset, UK). The isolated and pooled DRG neurons were plated onto poly-L-lysine coated glass coverslips and maintained in culture for 1–2 days.
Conventional voltage-clamp recordings in the whole-cell patch-clamp configuration were made from neurones of varying sizes (corresponding to small and medium sized neurons with a maximum apparent diameter of 43 μm) using an Axopatch 200B amplifier (Axon Instruments, Union City, California, USA) driven from a PC generating pulse protocols (Pclamp 9, Axon Instruments). The neuronal capacitance was estimated through the procedure of capacity transient cancellation, soon after attaining the whole-cell configuration. Electrodes were made from thin-walled glass (Harvard Apparatus, Edenbridge, Kent, UK) and had an initial resistance of between 1.5 and 2 MΩ when filled with intracellular solution. Series-resistance compensation was set near 70% with a nominal feed-back lag of 12 ms. Recordings were filtered at 5 KHz (4-pole Bessel filter) and sampled at 10 KHz. Recordings were made at room temperature.
TTX-s and TTX-r Na+ currents were recorded in relative isolation by including pharmacological blockers of both K+ and Ca2+ currents in both extracellular and intracellular media. In order to distinguish TTX-s and TTX-r currents, recordings were made before and after the addition of 250 nM TTX to the extracellular solution, where the TTX-s component was subsequently derived by off-line analysis using digital subtraction. The extracellular solution contained (in mM): NaCl 43.3, Tetraethylammonium Chloride 96.7, HEPES 10, CaCl2 2.1, MgCl2 2.12, 4-Aminopyridine 0.5, CsCl 10, KCl 7.5, CdCl2 0.1. The intracellular solution contained (in mM): CsCl 130, CsF 13, EGTA (Na) 3, Tetraethylammonium Chloride 10, HEPES 10, CaCl2 1.21, ATP(Mg) 3 mM, GTP (Li) 500 μM. External and internal solutions were buffered to pH 7.2–7.3 with the addition of CsOH. Tetrodotoxin (TTX, 250 nM) was locally applied and removed by gravity fed superfusion. The reagents, with the exception of TTX, were obtained from Sigma-Aldrich (Poole, Dorset, UK). TTX was purchased from Alomone Labs (TCS Biologicals, Botolph Claydon, Bucks, UK).
In order to derive current density measurements, peak Na+ currents were recorded in response to a family of incrementing, depolarizing clamp-steps, filtered at 5 KHz (4-pole Bessel filter) and normally sampled at 20 KHz. The holding potential was -80 mV, and the clamp-steps were preceded by a 20 ms pre-pulse. These values of membrane potential, coupled with the presence of internal F-, were not optimal for recording TTX-r persistent Na+ current, and the TTX-r currents analysed always exhibited the voltage-dependent and kinetic characteristics of Nav1.8. Current density estimates were found by dividing the peak-current values by the value estimated for membrane capacitance. TTX-s current amplitudes were found by off-line digital subtraction.
At defined time points after L5 spinal nerve ligation mice were anaesthetised via a brief induction with isofluorane gas followed by an i.p. injection of pentobarbitone (100 mg/kg). A lumbar laminectomy was performed and the L4 and L5 dorsal roots were identified and cut close to the spinal cord. Both the ventral roots and the L4 spinal nerve were cut as far distally as possible and then the dorsal roots, ventral roots, DRGs and spinal nerves (in continuity) from L4 and L5 were carefully and quickly removed. Throughout the process of dissection the tissue was kept hydrated by regularly applying oxygenated (95% O2, 5% CO2) aCSF solution (in millimolar: NaCl 118; KCl 4.7; MgSO4 1.2; KH2PO4 1.2; NaHCO3 25; CaCl 2.5 and glucose 11) to the preparation. The tissue was then kept in oxygenated aCSF at room temperature for 30 minutes before the start of electrophysiological recording. The peripheral nerve preparations were mounted in a two chamber recording bath with the spinal nerve and DRG on one side continuously perfused with oxygenated aCSF at a rate of 10 ml/min. This solution was kept at 35 ± 1°C by running the perfusion tube through a jacket containing water heated and circulated by a thermostatically controlled water bath. The dorsal and ventral roots were sealed in the other chamber and covered in mineral oil which had been previously oxygenated with an air tube. Oil in this chamber was periodically replaced to maintain hydration and oxygenation of the dorsal root. A leak-proof partition barrier of perspex and high-vacuum grease ensured electrical isolation from the other compartment. For the purposes of electrical stimulation the end of the spinal nerve was taken up into a suction electrode with bipolar silver/silver chloride wires. Fine filaments were teased from the dorsal root using sharpened watchmakers forceps. These strands were placed over electrodes made of sliver wire coated with teflon that had been stripped away from the tip. A crushed strand from the ventral root was placed across another silver wire electrode to act as an indifferent recording. Signals from the electrodes were passed through a head-stage which was also connected to earth. From here the signal was amplified and filtered (low pass 500 Hz, high pass 5 KHz) using Neurolog equipment (Digitimer, UK). Residual 50 Hz noise was removed with a Humbug noise eliminator (Quest Scientific, Canada). The signal was viewed on a digital oscilloscope (Tektronix, USA) and also recorded for off-line analysis (Powerlab and Chart 5 software, ADI instruments, UK).
Spontaneously active units were recorded in both the L5 and L4 dorsal roots. In strands with multiple ectopically firing units separation was achieved on the basis of differing amplitude, action potential shape and interspike interval. The number of conducting fibres in each strand was determined by incremental electrical stimulation of the spinal nerve whilst observing the recruitment of units on a storage oscilloscope (Gould, UK). Strands that were studied typically contained a total of 8 to 12 conducting units and around 50 to 60 units were sampled from each dorsal root. The proportion of spontaneously active units in each preparation was calculated as a percentage and then a mean obtained for all data in a group with n representing a single L4 or L5 root taken from one animal. A large proportion of units observed firing spontaneously did so in a bursting manner and so further analysis of these patterns was undertaken using specialised software (Spike 2, Cambridge Electronic Design, UK).
Unpaired t-test was used to compare data unless otherwise indicated.
The author(s) declare that they have no competing interests.
We thank the Wellcome trust and the MRC for their support. MAN, SBM and JNW are members of the London Pain Consortium.