GTP up-regulated persistent Na⁺ current and enhanced nociceptor excitability require Na_V1.9

Abstract

Persistent tetrodotoxin-resistant (TTX-r) sodium currents up-regulated by intracellular GTP have been invoked as the site of action of peripheral inflammatory mediators that lower pain thresholds, and ascribed to the Na_V1.9 sodium channel. Here we describe the properties of a global knock-out of Na_V1.9 produced by replacing exons 4 and 5 in SCN11A with a neomycin resistance cassette, deleting the domain 1 voltage sensor and introducing a frameshift mutation. Recordings from small (< 25 μm apparent diameter) sensory neurones indicated that channel loss eliminates a TTX-r persistent current. Intracellular dialysis of GTP-γ-S did not cause an up-regulation of persistent Na⁺ current in Na_V1.9-null neurones and the concomitant negative shift in voltage-threshold seen in wild-type and heterozygous neurones. Heterologous hNa_V1.9 expression in Na_V1.9 knock-out sensory neurones confirms that the human clone can restore the persistent Na⁺ current. Taken together, these findings demonstrate that Na_V1.9 underlies the G-protein pathway-regulated TTX-r persistent Na⁺ current in small diameter sensory neurones that may drive spontaneous discharge in nociceptive nerve fibres during inflammation.

Reports in the literature ascribe three very different functional roles to SCN11A, Na_V1.9. Work in primary sensory neurones (Dib-Hajj et al. 1998; Cummins et al. 1999), reviewed by Dib-Hajj et al. (2002), and enteric neurones (Rugiero et al. 2003) indicates that Na_V1.9 is a tetrodotoxin-resistant (TTX-r) Na⁺ channel with a negative threshold of activation close to normal resting potentials, which gives rise to persistent currents. It is expressed in small diameter, potentially nociceptive sensory neurones in both dorsal root ganglia (DRG) and trigeminal ganglia, and also in the intrinsic sensory neurones of the gut (Rugiero et al. 2003; Fang et al. 2006; Padilla et al. 2007). Quantifying expression using RT-PCR and Northern analysis indicates a very low level of expression in the central nervous system (Dib-Hajj et al. 1998). Studies of the possible role of inflammatory mediators acting through G-protein coupled receptors are consistent with Na_V1.9 making a contribution to setting inflammatory pain thresholds (Baker et al. 2003; Baker, 2005), a concept supported by behavioural studies on other Na_V1.9 knock-out mice lines (Priest et al. 2005; Amaya et al. 2006). The other roles suggested for Na_V1.9 are generating a rapidly inactivating sodium current (Tate et al. 1998), and operating as a ligand-gated channel in the cortex (Blum et al. 2002). The human channel coexpressed with the TrKB receptor was reported to generate BDNF-gated currents in heterologous systems and to underlie excitatory currents evoked by crude preparations of BDNF in cortical neurones (Blum et al. 2002). Neither of these other suggested roles for Na_V1.9 has received further substantiation, and the first may have been the result of misidentifying an endogenous HEK293 cell current (Cummins et al. 1999).

Several lines of evidence now support the hypothesis that Na_V1.9 generates persistent Na⁺ current in nociceptive sensory neurones. Immunohistochemical studies detail expression of the channel in intact neurones shown to be nociceptive by the characterization of their receptive fields (Fang et al. 2006). Studies of the channel distribution localizes Na_V1.9 to unmyelinated nerve endings in the periphery, including the surface of the eye, lip skin and tooth pulp (Black & Waxman, 2002), and within the interplexus fibres of submucosal enteric neurones (Padilla et al. 2007). Furthermore Na_V1.9 mRNA is not normally found in large diameter neurones (Dib-Hajj et al. 1998) and the channel protein does not usually colocalize with neurofilament-200 (a marker for A-fibres) in DRG (Padilla et al. 2007). Na_V1.9 is reported to be expressed in enteric sensory neurones in the absence of Na_V1.8 mRNA, and these neurones have been demonstrated to generate only persistent TTX-r currents (Rugiero et al. 2003). The Na_V1.9 knock-out (KO) mouse described by Priest et al. (2005) shows an analgesic phenotype, primarily with respect to the second phase of pain behaviour during the formalin test, but also in a reduction of hyperalgesia brought about by subdermal application of PGE₂. Another Na_V1.9 KO also exhibits an inflammatory phenotype and has deficits in the response to intraplantar UTP, a P2Y agonist (Amaya et al. 2006). The persistent Na⁺ current that has been associated with Na_V1.9 is known to be functionally regulated by activation of G-protein pathways in primary sensory neurones and is expected to be modified by algogenic agents, including ATP (Baker et al. 2003; Baker, 2005). Although the mechanisms involved in the expression of phase II formalin pain behaviour are uncertain, evidence suggests that periods of spontaneous activity in small afferents must also be coincident (Puig & Sorkin, 1996), and possibly caused by functional up-regulation of the persistent Na⁺ current at nerve endings. Furthermore, electrophysiological characterization of dissociated DRG neurones from Na_V1.9 KO mice has indicated that the persistent Na⁺ current seems to be absent (Priest et al. 2005; Amaya et al. 2006). The reported properties of the KO are therefore consistent with the loss of a G-protein pathway regulated Na⁺ channel in the periphery, which is able to substantially modify the excitability and/or the firing properties of nociceptive afferents. However, whether the loss of Na_V1.9 also causes a loss of G-protein pathway induced changes in excitability has not yet been demonstrated. Therefore a crucial link between observations in voltage-clamped sensory neurones and Na_V1.9 KO mouse behaviour is yet to be described.

Whether or not the human channel can generate persistent Na⁺ current has remained uncertain, in part because of the difficulties involved in obtaining functional heterologous expression of hNa_V1.9, and also because of the claim that Na_V1.9 is a ligand-gated channel in the brain. To address these issues, we have generated a KO of Na_V1.9, and functionally characterized the TTX-r Na⁺ currents found in the small diameter DRG neurones from these animals. This allowed us to confirm that Na_V1.9 is the substrate for the persistent Na⁺ current in sensory neurones, and to test whether G-protein pathway-dependent changes in neurone excitability were also abolished. We have also used sensory neurones cultured from these KO mice as an expression system for the study of the hNa_V1.9 clone (Blum et al. 2002), which does not give rise to functional sodium channel activity when expressed in cell lines.

Methods

Na_V1.9 gene knock out

An RPCI-22 129S6/SvEvTac mouse BAC library was screened for Na_V1.9. DNA from Na_V1.9 BAC clones was prepared and subcloned into pBluescript (BS-SKII⁻).The 5′-arm, containing exons 2 and 3, of the SCN11A gene was a 4.8 kb EcoRI–NsiI fragment. The 3′ arm containing exons 6–10 was a 7.2 kb AccI–Smal fragment. The two arms were inserted around a neomycin cassette. Hence, exons 4 and 5 of the SCN11A gene were replaced by the neomycin resistance cassette, and this deletes the domain 1, S4 voltage sensor, of the Na_V1.9 channel. Cells were selected with G418 and correctly targeted single-copy integrations were identified using Southern blots. Chimeras were crossed with C57BL/6 and germ line transmission tested using Southern blotting of genomic DNA from tail biopsies. For Southern blots, DNA was digested with BamHI and probed with a 500 bp SacI-EcoRI fragment 5′-to 5′-arm.

RT-PCR

RNA isolation was performed using TRIzol Reagent (Invitrogen, Paisley, UK), according to the manufacturer/s protocol. The reverse transcription reaction was also performed according to an Invitrogen protocol, using random primers. The primers used were (exon 2: CCA-TCAGAAGCTTCATGATTCGCA) (exon 9: AAGACA-AAGTAGATCCCGGAGGTG) (exon 5: CCTTGTTTT-CTCGGTAACAAAGTC) and (exon 6: TGGAAAAAGCG-TTAGGGCCACAGT). RT-PCR bands were cut out from agarose gels and purified using a JetSorb kit and sequenced using an ABI BigDye kit.

Genotyping

DNA from mice ear biopsies was extracted using a Lysis Buffer (2% SDS, 0.2 μg ml⁻¹ proteinase K). Genotyping of the Na_V1.9 KO mice was performed by two PCR reactions using 3′ primers specific for the deleted exons 4 and 5 (5′-AACAGTCTTACGCTGTTCCGATG-3′) or the inserted neomycin gene (5′-CTCGTCGTGAC-CCATGGCGAT-3′). The same 5′ primer was used for both reactions (5′-ATGTGGCACTGGGCTTGAACTC-3′) giving a 450 bp band for the WT gene and a 600 bp band for the mutated gene on a 1% agarose gel.

DRG cultures

Adult mice were killed in accordance with Home Office guidelines by cervical dislocation. Primary sensory neurones were prepared by enzymatic dissociation of whole dorsal root ganglia, following the procedure described by Baker & Bostock (1997). The neurones were plated out onto poly l-lysine coated coverslips and maintained in culture for 1–2 days.

Intra-nuclear injection

Vectors containing hNa_V1.9 (the clone of hNa_V1.9 provided by Dr Robert Blum, University of Munich) and EGFP were injected into the nuclei of Na_V1.9 null neurones 1–2 days after preparation. This was achieved using a picospritzer to pressurize a fine injection needle carried on a micromanipulator, mounted on an inverted microscope. Injection solution contained 10 μg of hNa_V1.9 in pcDNA3.0, 3 μg EGFP in pBRS, and 2 μl of 10% dextran FITC or TRITC to enable visualization of the injection mixture. Neurones were subsequently incubated for 24 h before recording.

Electroporation

An MP-100 Microporator (Labtech) was used to electroporate primary sensory Na_V1.9 null neurones. We resuspended 5 × 10⁴ neurones in 10 μl of Microporation Buffer (Labtech) and 275 ng of pcDNA3.0 vector containing Na_V1.9 and 115 ng of an EGFP vector were added. The cells were pulsed using a 2 × 20 ms protocol with 1200 V and then plated on coverslips.

Electrophysiology

Whole-cell voltage-clamp and current-clamp recordings from small (< 25 μm apparent diameter) primary sensory neurones were made at 24–48 h following dissociation, using an Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA). Whole-cell voltage-clamp recordings were made from transfected neurones 24 h following intranuclear injection or electroporation of vectors, where the EGFP fluorescence intensity was clearly discriminated above background. A Dell PC generated the pulse protocols (pCLAMP 9, Axon Instruments) and recordings were made on-line, filtered at 1–5 kHz (4-pole Bessel), and usually sampled at 5 or 10 kHz, which was sufficient to allow the adequate resolution of kinetically slow Na⁺ currents and subthreshold potential changes for the determination of voltage threshold. In voltage-clamp, the holding potential chosen was −110 mV, where incrementing depolarizing clamp-steps were preceded by a pulse to −130 mV. Current recordings were usually the average of three responses to the repeated voltage-clamp protocol. Na_V1.8 could be recorded alone in some wild-type (WT) and heterozygote (Hetero) neurones either where persistent sodium current was not expressed, by using a more positive holding potential to minimize its contribution through inactivation, or where recordings were short-lived and the current did not have time to up-regulate. In current-clamp recordings, a holding current was applied to keep the membrane potential as close to −90 mV as possible, and the amplitude of this current was manually adjusted from time to time (Baker et al. 2003). This holding current was small (usually 200 pA or less, sometimes ∼20 pA), and maintaining the membrane potential near a predetermined value had two benefits. Firstly, it allowed persistent Na⁺ current to clearly influence threshold, when up-regulated, and secondly any complicating effects of different resting potentials on voltage-threshold were eliminated allowing easier comparisons between neurones. A depolarizing current increment was chosen to allow the recording of several subthreshold and suprathreshold responses within a single data file of eight sequential traces. Applied current steps were 160 ms in duration to potentially allow the recording of repetitive firing. Total transmembrane current was simultaneously recorded with the membrane potential, and neither of these data was averaged.

Estimates of neurone capacitance and series resistance were obtained in voltage-clamp using the capacity transient cancellation procedure provided by the amplifier. Series resistance compensation was set near 70% with a nominal feed-back lag of 12 μs. Compensation was used in voltage-clamp and maintained after subsequently switching to current-clamp mode. In order to compare the effects of the intracellular solution on the excitability of KO and WT-Hetero neurones and to measure changes in voltage threshold, it was necessary to make recordings over several minutes from the point at which recording began in whole-cell mode. Neurones were actually held for as long as possible and up to 31 min (11 ± 1 min and 10 ± 1 min, for KO (n = 32) and WT-Hetero (n = 30), respectively, mean ±s.e.m. However, persistent current up-regulation may not have reached a maximum in every case by the time a neurone was lost, and for this reason the experiments may theoretically underestimate the functional effects of Na_V1.9.

Electrodes were pulled from thin-walled glass capillaries (Harvard Apparatus, Edenbridge, Kent, UK). The electrodes had an initial resistance of between 1.5 and 2 MΩ once filled with intracellular solution. The solutions used for voltage-clamp experiments were as follows. Extracellular (mm): NaCl 43.3, TEA-Cl, 96.7, Hepes 10, CaCl₂ 2.1, MgCl₂ 2.12, 4-aminopyridine 0.5, KCl 7.5, CsCl 10, CdCl₂ 0.05, tetrodotoxin (TTX) 0.00025. Intracellular (mm): CsCl 145, EGTA (Na) 3, Hepes 10, TEA-Cl 10, CaCl₂ 1.21, ATP (Mg) 3, GTP-γ-S (Li) 0.5. Both extracellular and intracellular voltage-clamp solutions were buffered to 7.2–3 by the addition of CsOH. The solutions used for current-clamp experiments were the following. Extracellular (mm): NaCl 140, Hepes (hemi Na) 10, CaCl₂ 2.1, MgCl₂ 2.12, KCl 2.5. Intracellular (mm): KCl 143, EGTA (Na) 3, Hepes (Hemi Na) 10, CaCl₂ 1.21, MgCl₂ 1.21, ATP (Mg) 3, GTP-γ-S (Li) 0.5. Both extracellular and intracellular current-clamp solutions were buffered to 7.2–3 by the addition of NaOH. Reagents were obtained from Sigma-Aldrich (Poole, UK) with the exception of TTX, which was obtained from Alomone Laboratories (Botolph Claydon, Bucks, UK).

Where possible values are reported as mean ±s.e.m.

Results

We generated a conventional knock-out (KO) of Na_V1.9 (Fig. 1). Exons 4 and 5 of the SCN11A gene were replaced by a neomycin resistance cassette, eliminating the domain I, S4 voltage sensor. Analysis of genomic DNA from KO mice using Southern blots confirmed the insertion of the neomycin cassette (Fig. 1B). RT-PCR on total RNA extracted from KO and wild-type (WT) DRGs using primers flanking exons 4 and 5, and within exon 5 indicated that the KO allele produces a mRNA that lacks exons 4 and 5 (Fig. 1C). Sequencing the RT-PCR products confirmed that the splicing machinery skips the neomycin cassette and splices exon 3 to exon 6 (sequencing RT-PCR, Fig. 1D). This results in a frameshift mutation leading to the introduction of stop codons within exon 6. Na_V1.9 KO mice were healthy and fertile.

Figure 1. Construct and genotyping.

A, schematic diagram of the Na_V1.9 knock-out allele. The figure shows the size and location of the homologous arms. Also shown is the part of the Na_V1.9 gene replaced by a neomycin cassette. B, Southern blot confirmation of the targeting of the Na_V1.9 gene. An external 5′ probe identifies the wild-type band and the knock-out bands. C, analysis of Na_V1.9 mRNA from DRG of KO mice using RT-PCR. Using PCR primers flanking the deleted exons (filled arrow primer pairs) reveals a shorter band. Using a primer in exon 5 (open arrow primer pairs) produces a product in the wild-type DRG but not in the knock-out DRG. This indicates that the neomycin cassette is spliced out and that exons 3 and 6 are spliced together in the knock-out mouse. D, sequencing RT-PCR products reveals that the knock-out mRNA lacks exons 4 and 5 (in bold); the deleted part starts with the end of exon 3 and ends with the beginning of exon 6.

Electrophysiological characterization of Na_V1.9 KO neurones

Recording in the presence of tetrodotoxin (TTX), the GTP up-regulated persistent Na⁺ current, was not found in the small diameter Na_V1.9 KO neurones studied in voltage-clamp (n = 22). Persistent currents have been previously well described for WT rat and mouse small diameter sensory neurones (e.g. Cummins et al. 1999; Baker et al. 2003) and for enteric neurones in the guinea pig (Rugiero et al. 2003), and this finding is consistent with the reports of Priest et al. (2005) and Amaya et al. (2006), where neurones isolated from other Na_V1.9 KO mice have been studied in voltage-clamp. In the present experiments, the current–membrane potential (I–V) data for WT and Hetero neurones were not different (Fig. 2A) suggesting that the loss of one copy of SCN11A does not alter functional Na⁺ currents, a result similar to that found for the major TTX-r current with heterozygotes of Na_V1.8 (Stirling et al. 2005). The frequency with which the TTX-r persistent current was found in voltage-clamped WT-Hetero neurones (using our stated voltage-clamp solutions that block K⁺ channels) was 10 in 19, where the neurones generated an inward current of at least 50 pA at −40 mV. This finding is similar to our previous reports. However, larger persistent currents were less common.

Figure 2. Lack of GTP up-regulated, TTX-r persistent current in Na_V1.9 knock-out neurones.

A, separate mean current–voltage (I–V) relation for TTX-r currents for Hetero (•, mean –s.e.m.) and WT neurones (open circles, mean +s.e.m.), which are not different. B, mean current–voltage relation for TTX-r currents evoked in pooled Na_V1.9 Hetero (n = 7) and WT (n = 12) neurones (•±s.e.m.). Data from the Hetero neurone shown in Fig. 3 are also presented as an example I–V (^) with the maximum amplitude scaled to match the average maximum value. Apparent reversal of mean I–V relation more negative than E_Na is mainly due to the presence of residual K⁺ currents. C, mean current–voltage (I–V) relation for TTX-r currents evoked in Na_V1.9 KO neurones (•, n = 22), consistent with Na_V1.8 only contributing to the inward current. None of the neurones studied showed any persistent current that was up-regulated by GTP-γ-S. Data from a typical individual recording are also presented as an example I–V (^) with the maximum amplitude scaled to match the average maximum value. D, current amplitudes at negative potentials in WT-Hetero neurones (•) and in KO neurones (^). Current amplitudes, and derived current densities, show significant differences where currents at −30 mV in WT-Hetero are significantly larger than in KO (#P < 0.04, Student/s t test). The current density is significantly more negative at −40 and −50 mV in pooled WT-Hetero data than in KO (*P < 0.03 at −40, P < 0.02 at −50 mV, Student/s t test). Current density data at −50 mV is shown as a bar graph (inset), *P < 0.02.

Comparing I–V data from KO neurones with pooled data from WT and Hetero littermates (Fig. 2B and C) revealed that there was a significant loss of inward current at −30 mV in the KO, and a significant change in current density at −40 and −50 mV (Fig. 2D). Furthermore, there was also a significant change in current density in comparison with the WT alone (KO at −40 and −50 mV (n = 22) versus WT at −40 and −50 mV (n = 12), 1.30 ± 1.18 pA pF⁻¹ and 1.45 ± 0.93 pA pF⁻¹versus−4.79 ± 2.98 pA pF⁻¹ and −3.76 ± 2.53 pA pF⁻¹, P < 0.04 and < 0.03, respectively. Student/s t test). A reduction in the mean peak TTX-r inward current at 0 mV by 37% in the KO was not statistically significant (−0.76 ± 0.22 nA versus−1.21 ± 0.24 nA in KO versus WT-Hetero neurones, n = 22 and 19, respectively), but nevertheless consistent with the removal of one component of TTX-r current. Na_V1.8 was usually present in the KO (Fig. 2C and D), and had a current density not different from that found in many previous studies of wild-type mouse and rat sensory neurones (recording using the same Na⁺ gradient and solutions; Okuse et al. 2002; Nassar et al. 2004; Foulkes et al. 2006; −0.057 ± 0.009 nA pF⁻¹versus−0.075 ± 0.013 nA pF⁻¹, KO (n = 22) versus WT-Hetero (n = 12), respectively). Na_V1.8 peak currents were measured in isolation (see Methods) and included in the averages where the amplitudes were at least 100 pA. The loss of a component in the KO contributing to the I–V plot over a negative potential range was reflected in individual voltage-clamp experiments. For both WT and Hetero littermate neurones, there was clear evidence for a GTP up-regulated persistent current, but this was not the case for Na_V1.9 KO neurones (Fig. 3).

Figure 3. GTP up-regulated current present in heterozygote but not in knock-out.

Example recordings of TTX-r Na⁺ currents in a sensory neurone from an Na_V1.9 Hetero, compared with a KO. A, left-hand panels, Na⁺ currents recorded from Hetero neurone immediately on attaining the whole-cell configuration at the start of voltage-clamp recording (upper), and after 3.5 min (lower). The voltage step to −30 mV and current evoked at the same potential are indicated with a thick trace. The TTX-r current was seen to increase in amplitude, notably within the most negative potential range, because of the addition of a persistent current. Right-hand panels, Na⁺ currents recorded from an Na_V1.9 KO neurone over a similar time (upper, immediately on attaining the whole-cell configuration; lower, after 4 min). In this recording no current is activated more negative than −30 mV. The current was generated by Na_V1.8 operating alone, and also undergoes an increase in amplitude. B, in order to view the up-regulated current in isolation, an off-line digital subtraction was performed, where the current traces recorded at 0 min were subtracted from currents recorded at 3.5 min in Hetero (left) and at 4 min in KO (right). The up-regulated persistent Na⁺ current is clearly evoked over a potential range more negative than that associated with Na_V1.8, and only in the Hetero neurone.

In current-clamp experiments, the voltage threshold was measured repeatedly by applying incrementing depolarizing current steps after routinely holding the neurones at −90 mV (Fig. 4). This procedure allowed all the Na⁺ channel subtypes expressed in the neurones studied to contribute to subthreshold behaviour and also to action potential generation. When measured in this way a range of voltage thresholds are apparent, with firing threshold being determined usually by TTX-sensitive (TTX-s) Na⁺ currents (with an activation threshold near −40 to −45 mV), or more rarely TTX-r transient Na⁺ currents (Na_V1.8, with an activation threshold −25 to −20 mV). Following intracellular dialysis of GTP-γ-S, we found no change in threshold in the Na_V1.9 null neurones that could be attributed to the up-regulation of a persistent Na⁺ current (n = 32) (Fig. 5A and B). Also, no negative-threshold persistent inward current was observed in the same neurones when switching to voltage-clamp mode at the end of current-clamp recording (allowing the measurement of currents generated with a physiological Na⁺ gradient but in the presence of working K⁺ currents). This was not the case in neurones obtained from WT or Hetero littermates (n = 30; 18 neurones from WT and 12 from hetero, Fig. 5A and B), where our data indicate that there are two functional groups of neurones that either generate the current and undergo a substantial change in voltage threshold during recording (n = 5), or do not generate the current (n = 25). The threshold-change data gathered from WT and Hetero neurones are significantly skewed (skewness > 2 × standard error of skew: −2.15 and 0.43 for pooled WT-Hetero data, and −2.61 and 0.54 for WT alone; SPSS v14), with 5 of 30 neurones showing a threshold change greater than 2 s.d. from the mean of the KO distribution. Recordings were made from this subpopulation of neurones for 7.3 ± 2.0 min, no longer than for either the KO population, or the WT-Hetero population as a whole (see methods), or the WT-Hetero neurones that did not show a substantial threshold change (10.5 ± 0.5 min). Out of these five neurones, it was subsequently possible to get voltage-clamp data in only three. However, all three generated a persistent inward current, operating at potentials more negative than that necessary to recruit transient current, of −170.6 ± 53.3 pA (mean ±s.e.m. of largest current measured over a potential range of −75 to −50 mV, see Fig. 5D). In the remaining WT and Hetero recordings, voltage-clamp data were obtained in a total of 19 neurones, and no such inward currents were present. This finding suggests that persistent current up-regulation and a change in voltage threshold are inextricably linked. Taking the five neurones undergoing marked changes in voltage threshold as a separate group can be further justified by the analysis shown in Fig. 5B, where these data are compared with data from the rest of the WT-Hetero population and the KO, indicating a substantial and statistically significant difference in the means (P = 0.02 and < 0.02, Student/s t test, KO and residual WT-Hetero data, respectively). The change in voltage threshold seen in WT and Hetero neurones is accompanied by an increase in the latency of just-supra-threshold stimulus action potential generation, and an expected reduction in the applied depolarizing stimulus current (Fig. 5C). If persistent current up-regulation is substantial enough (Fig. 5D), neurones can fire action potentials after the cessation of an applied depolarizing current. These observations in Na_V1.9 KO neurones extend those previously published (Priest et al. 2005; Amaya et al. 2006) by implicating Na_V1.9 as the substrate of the functional plasticity controlling excitability, and for the voltage threshold for firing action potentials. In order to help rule out the possibility that the subpopulation of neurones from the pooled WT-Hetero group were found to undergo threshold changes simply by chance, in comparison with none in the KO group, we performed a Fisher exact test on the frequency data. With no responding neurones in the KO group, and five in the WT-Hetero group, P = 0.022. This may be taken as evidence that the WT-Hetero and KO groups are different.

Figure 4. Measurement of voltage-threshold in Na_V1.9 knock-out neurones.

A, example currents with quasi-physiological solutions recorded in voltage-clamp from a Na_V1.9 KO neurone. Transient Na⁺ current began to be recruited at −40 mV without there being any evidence for a persistent inward current operating at more negative potentials. A delayed-rectifier, voltage-dependent K⁺ current typically activated close to −50 mV. B, in the same neurone as in A, the total transmembrane current (both applied and endogenous active) and membrane potential responses are shown top and bottom panels, respectively. The subthreshold membrane potential changes are fitted with single exponentials. The voltage-threshold is defined as that value of potential where there is a clear deviation from a passive response subsequently leading to an action potential. This is shown as the dotted line (in this example, −40 mV). After the moment voltage threshold is reached, active inward current is also apparent in the membrane current records, and these two indicators may be simultaneous.

Figure 5. Action potential voltage thresholds in Na_V1.9 knock-out neurones and the effect of intracellular GTP-γ-S.

A, in current-clamp experiments, the intracellular dialysis of GTP-γ-S did not cause a substantial change in voltage threshold in Na_V1.9 KO neurones (•) in recordings lasting between 1.5 and 31 min. In both Hetero (▵) and WT (^) littermates (n = 12 and 18, respectively), there was a significant skew in the response to GTP-γ-S, with a fraction of the neurones showing a negative shift in threshold greater than 2 s.d. from the mean KO value (for recording times, see Methods). B, 5 neurones lying beyond −2 s.d. indicate that the WT-Hetero population can be described as two populations, those that underwent threshold change (n = 5) and that are therefore different from KO, and those that did not (n = 25). The mean threshold change in these 5 neurones is significantly different from both the KO and other WT-Hetero neurones (P = 0.02; P < 0.02 Student/s t test). C, simultaneous recordings of membrane potential and total transmembrane current near the start of whole-cell recording (left) and a few minutes later (right). The depolarizing current steps generate subthreshold and just supra-threshold responses in each case (thin and thick traces), the voltage threshold becoming more negative and action potential latency becoming longer with GTP-γ-S dialysis, even finally occurring after the offset of applied depolarization (indicated by the arrow). D, results of a voltage-clamp experiment on the same neurone, recorded after the data shown in C, and using the same solutions. A clear persistent inward current is seen operating at −50 mV (thick trace), which was not activated at −80 mV. Further depolarization activates a transient Na⁺ current and an inactivating K⁺ current (or A-current), *. The functional importance of the A-current in this particular neuron is reflected in the brief hyperpolarizing afterpotential following the action potential in C.

Expression of transfected hNa_V1.9 in Na_V1.9 KO neurones

Both intranuclear injection and electroporation of the human Na_V1.9 clone (Blum et al. 2002) gave rise to TTX-r persistent Na⁺ currents in transfected neurones (total n = 5; Fig. 6), providing confirmatory evidence that Na_V1.9 is indeed the substrate for the TTX-r persistent Na⁺ current. We were unable to obtain functional voltage-gated currents in either transfected HEK293 or COS-7 cells using the same vector. Possible explanations for this finding are that primary sensory neurones can provide essential auxiliary proteins or a necessary trafficking mechanism that allows functional channel expression.

Figure 6. Transfection of hNa_V1.9 results in the restoration of TTX-r persistent current in Na_V1.9 knock-out neurones.

Example currents recorded in voltage-clamp following intranuclear injection (A) and electroporation (B) of small diameter neurones cultured from Na_V1.9 KO mice (total n = 5). Currents elicited in the presence of TTX have the appropriate negative voltage dependence and slow kinetics previously described for persistent Na⁺ current. In B, a transient TTX-r current is recruited close to −30 mV, consistent with activation of Na_V1.8, presumed endogenous to the neurone.

Discussion

We have presented evidence that strongly supports the hypothesis that disruption of SCN11A eliminates the expression of a TTX-r persistent Na⁺ current in small-diameter sensory neurones, clearly indicating that Na_V1.9 is the substrate for the current. Furthermore, the persistent Na⁺ current can be restored when hNa_V1.9 is transfected into Na_V1.9 KO neurones. We envisage that the functional up-regulation of Na_V1.9 (Baker et al. 2003; Baker, 2005) within a nociceptor terminal arborization (e.g. Padilla et al. 2007), after exposure of the ending to ATP or other inflammatory mediators, may precipitate spontaneous action potentials and contribute to inflammatory pain. The substantial change in voltage threshold we found on intracellular dialysis of GTP-γ-S is caused by the functional up-regulation of Na_V1.9, providing an effectively new subset of Na⁺ channels in the neurone. These channels are capable of producing regenerative inward current over a more negative potential range than was previously possible with the repertoire of transient Na⁺ channels already operating.

Fewer than half of the neurones generating a TTX-r persistent Na⁺ current (as measured in voltage-clamp with K⁺ currents blocked) generate a substantial change in voltage threshold in current-clamp or provide a net inward current when studied using voltage-clamp in quasi-physiological solutions. The simplest explanation for this is that only the neurones with the largest persistent Na⁺ currents can change the voltage threshold, and generate net inward currents in voltage-clamp in the face of working K⁺ currents. Net inward current over the crucial negative potential range is essential to initiate a regenerative depolarization and cause the threshold change.

Effect of GTP-γ-S in small diameter neurones

GTP-γ-S is expected to non-selectively activate G-protein pathways that are known to act on ion channels other than Na_V1.9. Thus it seems surprising that a substantial voltage threshold change occurs only where the persistent current generated by Na_V1.9 is up-regulated. However, this finding is consistent with data obtained previously on both WT and Na_V1.8 knock-out neurones. In both sensory neurones and heterologous systems expressing Na_V1.8, the macroscopic current generated by the channel can be increased by the activation of protein kinase A (PKA) (England et al. 1996; Gold et al. 1998). PKA activation can cause not only an increase in amplitude but also an increase in the steepness of the current activation voltage dependence (England et al. 1996). The reason that such an effect does not change the voltage threshold in the present experiments is that there are usually TTX-s Na⁺ channels operating in the same neurones that have an activation threshold some 15–20 mV more negative than Na_V1.8. These TTX-s Na⁺ channels act as the ‘threshold channels’, and although the maximum amount of current they generate might be small (in comparison with the current generated by Na_V1.8 in the same neurone subject to a larger depolarization), they contribute regenerative inward current before Na_V1.8 is activated. Thus the voltage threshold for action potential induction is more negative than the activation potential for Na_V1.8. The kinase induced modifications of Na_V1.8 would still be expected to contribute to the pattern of firing generated in response to a ramp stimulus (Zhang et al. 2002) or to any large and prolonged depolarization, by making the neurone more prone to fire repetitively. The ability to fire repetitively requires the maintenance of excitability in the face of prolonged supra-threshold depolarization, a characteristic that has been associated with Na_V1.8 (Dib-Hajj et al. 2005; Rush et al. 2006), because the channel has a relatively positive activation range and can escape inactivation quickly following action potential repolarization.

The reason intracellular GTP-γ-S does not cause a negative shift in voltage-threshold mediated by altered TTX-s currents can also be explained. Published data clearly point to a reduction in TTX-s current amplitudes in response to protein kinase activation (e.g. Li et al. 1992), although the complement of TTX-s channels in our recordings could not be functionally quantified. Arguing on the basis of data obtained on example neuronal TTX-s Na⁺ channels, PKA activation reduces the amplitudes of Na_V1.2 and Na_V1.7 expressed in CHO cells and Xenopus oocytes, respectively (Li et al. 1992; Vijayaragavan et al. 2004), and PKC activation reduces peak currents for Na_V1.7 (Vijayaragavan et al. 2004). These considerations seem relevant as we found that in about half the KO neurones with an initial voltage threshold at or more negative than −45 mV there was a positive shift during the recording, consistent with a reduction of TTX-s current (possibly in parallel with a gradually more negative value of E_K occurring with a raised intracellular K⁺ ion concentration after minutes of dialysis; Baker et al. 2003). However, a positive shift in voltage threshold did not always occur, and this may be a reflection of the non-specific effects of GTP-γ-S, that would be expected to activate G_i pathways as well as G_s and G_q/11. In the population of neurones as a whole, there was very little change in voltage threshold in the absence of functional Na_V1.9. In terms of neuronal or axonal excitability in situations where protein kinases are inhibited or other mechanisms may come into play, such as altered axonal transport, even a small increase in TTX-s current density might have a subtle but potentially important effect on voltage threshold. Although channel behaviour would be constrained by an unchanged activation voltage dependence, a greater current density will recruit an action potential more readily in the face of already operating K⁺ channels. Therefore, up- or down-regulation of TTX-s Na⁺ channel density is expected to affect voltage threshold, but not as spectacularly as the recruitment of a new set of Na⁺ channels with an activation threshold some 15 or 20 mV more negative than those already operating.