We used a mouse with deletion of exons 4, 5, and 6 of the SCN11A (sodium channel, voltage-gated, type XI, α) gene that encodes the voltage-gated sodium channel Nav1.9 to assess its contribution to pain. Nav1.9 is present in nociceptor sensory neurons that express TRPV1, bradykinin B2, and purinergic P2X3 receptors. In Nav1.9−/− mice, the non-inactivating persistent tetrodotoxin-resistant sodium TTXr-Per current is absent, whereas TTXr-Slow is unchanged. TTXs currents are unaffected by the mutation of Nav1.9. Pain hypersensitivity elicited by intraplantar administration of prostaglandin E2, bradykinin, interleukin-1β, capsaicin, and P2X3 and P2Y receptor agonists, but not NGF, is either reduced or absent in Nav1.9−/− mice, whereas basal thermal and mechanical pain sensitivity is unchanged. Thermal, but not mechanical, hypersensitivity produced by peripheral inflammation (intraplanatar complete Freund's adjuvant) is substantially diminished in the null allele mutant mice, whereas hypersensitivity in two neuropathic pain models is unchanged in the Nav1.9−/− mice. Nav1.9 is, we conclude, an effector of the hypersensitivity produced by multiple inflammatory mediators on nociceptor peripheral terminals and therefore plays a key role in mediating peripheral sensitization.
The peripheral terminals of nociceptors are the interface between noxious stimuli and the nervous system. Conversion of the stimuli into ion fluxes across nociceptor membranes is mediated by high-threshold thermal, mechanical, and chemical protein transducers (Julius and Basbaum, 2001). Transduction represents the first step in the generation of nociceptive pain, a sensation that warns of the presence of potentially tissue-damaging stimuli. The second step is initiation of action potentials in the nociceptors by voltage-gated sodium channels (VGSCs). Among the 10 VGSC isoforms, two, Nav1.8 (Akopian et al., 1996; Sangameswaran et al., 1997) and Nav1.9 (Dib-Hajj et al., 1998; Tate et al., 1998), are expressed almost exclusively by small nociceptor primary afferent neurons (Amaya et al., 2000) and are tetrodotoxin insensitive. Nav1.8 generates a tetrodotoxin-resistant (TTXr) current with a threshold of activation of −40 mV that inactivates slowly (TTXr-Slow) (Akopian et al., 1996; Sangameswaran et al., 1997; Renganathan et al., 2002b; John et al., 2004), whereas Nav1.9 is presumed from recordings in Nav1.8 null mice, to produce a persistent current with a more hyperpolarized voltage dependence and ultraslow recovery from inactivation (TTXr-Per) (Cummins et al., 1999; Baker et al., 2003).
In the presence of inflammation, the threshold to elicit pain falls and the response to noxious stimuli is exaggerated. The conversion of high-threshold nociceptive pain to a low-threshold inflammatory pain is contributed to by peripheral sensitization of the nociceptor peripheral terminal (Levine and Reichling, 1999). Inflammatory mediators, including amines, prostanoids, kinins, purines, protons, and nerve growth factor (NGF), sensitize the nociceptor peripheral terminal by producing a reduction in the threshold of transducer channels (McCleskey and Gold, 1999; Julius and Basbaum, 2001), an increase in terminal membrane excitability (Amir et al., 2006), and insertion of receptors into the terminal membrane (Zhang et al., 2005). The high threshold of the transducers are reduced by posttranslational processing in response to activation by the inflammatory mediators of multiple intracellular signaling pathways [PKA, PKC, extracellular signal-regulated kinase, phosphatidylinositol 3 (PI3) kinase, and phospholipase C] (Cesare et al., 1999; Aley et al., 2001; Bautista et al., 2006).
Prostaglandin E2 (PGE2), 5-HT, and bradykinin (BK) increase the Nav1.8-like TTXr current in DRG neurons, with an increase in amplitude and rates of activation and inactivation (Gold et al., 1996, 1998, 2002; England et al., 2001). Inflammatory pain sensitivity is, moreover, delayed in Nav1.8 null mutant mice (Akopian et al., 1999). PGE2 also increases TTXr-Per in DRG neurons, with steady-state activation shifted in the hyperpolarizing direction but no detectable change in channel activation or inactivation kinetics (Rush and Waxman, 2004). TTXr-Per is also increased by PKC activators (Baker, 2005). In an Nav1.9 knock-out (KO) with deletion of exons 2–4, there is loss of TTXr-Per, a reduction in inflammatory pain behavior (intraplantar carrageenan), and the response to peripheral PGE2 administration, but no disruption in basal or neuropathic pain (Priest et al., 2005). Nav1.9 may, like Nav1.8, play a pivotal role, therefore, in integrating the response of nociceptors to inflammatory mediators. We demonstrate here that Nav1.9 is a major effector of the peripheral pain-producing actions of multiple inflammatory mediators that act on G-protein-coupled receptors (GPCRs), TRP, and ligand-gated ion channels.
Materials and Methods
The SCN11A gene was isolated from a 129/Sv mouse genomic library and a “knock-out vector” generated containing an internal ribosome entry site (IRES)–lacZ reporter followed by a loxP-flanked neomycin resistance cassette to produce a targeting construct (Fig. 1A). Homologous recombination of the construct DNA in embryonic stem cells resulted in a disrupted SCN11A gene, and mice carrying the mutation of the Nav1.9 gene were generated according to standard protocols. The neomycin resistance cassette was not deleted. Germ-line chimeras were crossed with C57BL/6J females to generate heterozygotes and intercrossed, giving rise to overtly healthy and fertile mutant offspring in the expected Mendelian ratio. Successful targeting and transmission was confirmed by Southern blot analysis. An additional six backcrosses onto the C57BL/6J strain were performed before producing homozygote KO mice.
An automated behavioral analysis system (LABORAS) was used to measure the frequency and duration of locomotor activity, immobility, climbing, grooming, eating, and drinking behavior. Groups of eight male and eight female wild-type (WT) and homozygous Nav1.9−/− mice (9–12 weeks old) were housed in individual LABORAS cages with access to food and water ad libitum for a 24 h observation period.
For sensory testing, mice (between 8 and 15 weeks) were habituated and tests were performed blind to genotype. Plantar paw punctuate mechanical threshold and pinprick sensitivity were determined with von Frey hairs and a safety pin, heat sensitivity by withdrawal latency on a hotplate (50, 52, and 55°C), and reaction to cold on a cold plate and by acetone evaporation (Decosterd and Woolf, 2000). Intraplantar injections were made of BK (10 μl, 300 ng), capsaicin (10 μl, 2.5 μg), αβ-methyleneadenosine 5′-triphosphate (αβ-met-ATP) (10 μl 20 nmol), UTP (10 μl, 100 nmol), prostaglandin E2 (PGE2) (10 μl, 100 ng), interleukin-1β (IL-1β) (10 μl, 1 pg), and NGF (2.5S, 10 μl, 50 ng), and total licking time was recorded. Inflammation was produced by intraplantar injection of complete Freund's adjuvant (CFA) (20 μl; Sigma, St. Louis, MO), and paw edema was assessed by Evans blue extravasation (2%, 500 μl/mouse, i.p.) (Coderre et al., 1989). A spared nerve injury neuropathic pain model (Decosterd and Woolf, 2000) was used with section of the common peroneal and tibial nerves, leaving the sural nerve intact.
Mice were perfused with 4% paraformaldehyde, the tissue was cryoprotected in 20% sucrose, and DRGs (10 μm) were sectioned. Sections were blocked with Image-iT Fx signal enhancer (Invitrogen, Carlsbad, CA) and incubated with rabbit anti-Nav1.7 (1:1000; Alomone Labs, Jerusalem, Israel), anti- Nav1.8 (1:400), rabbit anti-peripherin (1:200; Chemicon, Temecula, CA), mouse anti-neurofilament 200 kDa isoform (NF200) (1:400; Sigma), and goat anti-β-galactosidase (1:300; Biogenesis, Poole, UK) in 0.1 m PBS, 0.025% Tween 20 and 0.5% BSA at 4°C for 2–3 d and were then incubated with rhodamine-conjugated anti-rabbit anti-mouse or anti-goat (1:100; Chemicon) antibodies. Double labeling was performed with rabbit anti-Nav1.9 (1:400) and either goat anti-transient receptor potential vanilloid receptor 1 (TRPV1) (1:400; Santa Cruz Biotechnology, Santa Cruz, CA) or guinea pig anti-P2X3 (1:1000; Chemicon). Double labeling for Nav1.9 immunohistochemistry and fluorescent in situ hybridization for the B2 bradykinin receptor was performed using a DIG-labeled RNA probe specific for B2 mRNA. The specificity of the Nav1.8 and Nav1.9 antibodies have been established previously (Amaya et al., 2000).
DRGs were placed into HBSS (Invitrogen) and digested with 5 mg/ml collagenase 1, mg/ml Dispase II (Roche, Indianapolis, IN), and 0.25% trypsin (Invitrogen) and triturated, the suspension was centrifuged through 10% BSA (Sigma), and the pellet was resuspended in Neurobasal (Invitrogen) containing B27 supplement (Invitrogen), penicillin and streptomycin (Sigma), 10 μm AraC, and 100 ng/ml 2.5S NGF (Promega, Madison, WI).
Whole-cell patch-clamp recordings were made using an Axopatch 200A amplifier and pClamp 8 (Molecular Devices, Palo Alto, CA). Small neurons from freshly (<48 h) dissociated DRG cultures (9–12 weeks old) were recorded at room temperature. Average cell size for WT was 24.6 ± 0.4 μm (n = 91) and for Nav1.9−/− was 27.5 ± 0.9 μm (n = 24). Fire-polished patch pipettes were made from 1.5 mm borosilicate glass (World Precision Instruments, Sarasota, FL) using a Sutter Instruments (Novato, CA) P-97 puller with 2–3 MΩ resistances. Signals were digitized at 10 kHz and filtered at 5 kHz. Capacity currents were cancelled, and series resistance was compensated by the “prediction” method (∼80%) and by the 10-μs-lag “feedback” method (70–80%). Linear leakage currents were digitally subtracted on-line with P/4 routines. Currents were recorded within 30 min of establishment of the whole-cell configuration to minimize time-dependent rundown. TTX at 300 nm was added to the bath solution to block TTX-sensitive (TTXs) currents. Bath solution contained the following (in mm): 130 NaCl, 4 KCl, 2 CaCl2, 1 MgCl2, 0.1 CdCl2, 10 HEPES, 5 glucose, and 10 tetraethylammonium (TEA)-Cl, pH 7.4 with NaOH (312 mOsm). Internal solution contained the following (in mm): 130 CsCl, 10 NaCl, 2 TEA-Cl, 10 EGTA, 0.5 CaCl2, 1 MgCl2, 5 MgATP, 0.5 NaGTP, 10 HEPES, and 2 glucose, pH 7.3 with CsOH (310 mOsm). After establishing whole-cell configuration, neurons were held at −90 mV to remove resting inactivation. A series of depolarizing voltage commands from −100 to 60 mV in steps of 10 mV for 120 ms with 1 s interval were used. To determine the voltage dependence of channel activation, the sodium conductance (GNa) was calculated. The normalized peak current for each voltage step was divided by the driving force (Vm − Vrev) and plotted against Vm. The voltage dependence of steady-state inactivation was measured by applying a double-pulse protocol consisting of one constant test pulse to 0 mV for 30 ms after 500 ms prepulses of amplitudes of −120 to 10 mV in steps of 10 mV. Normalized peak Na+ current during the test pulse was plotted against the prepulse voltage. The curves were fitted to the Boltzmann equation (y = A2 + (A1 − A2)/(1 + exp((x − x @ y50/dx), with x @ Y50 representing the potential at which half of the channels are activated or inactivated. Peak TTXs current density was measured by subtraction of TTXr current from total sodium current divided by cell capacitance. Peak TTXr-Slow current density in wild type was measured from cells without TTXr-Per currents.
Taqman reverse transcription-PCR.
Animal procedures were performed in accordance with the United Kingdom Home Office regulations as outlined in the Animals (Scientific Procedures) Act of 1986. L3, L4, and L5 DRGs were removed, snap frozen in liquid nitrogen, and stored at −80°C. Frozen DRGs were homogenized using a Mixer Mill MM 300 (Qiagen, Hilden, Germany), and total RNA was extracted using the RNeasy mini kit (Qiagen). A DNase step was included to eliminate DNA contamination in the RNA samples. The extracted RNA was quantified using the Agilent Technologies (Palo Alto, CA) Eukaryotic Total RNA assay and Nanodrop ND1000 spectrophotometer (LabTech, Andover, MA). First-strand cDNA was synthesized from 200 ng of total RNA from each sample. Reverse transcription (RT) was performed using Omniscript RT kit (Qiagen) in triplicate. An additional reaction in which the Omniscript reverse transcriptase enzyme was omitted and the value obtained for this sample was subtracted from the triplicate experimental values to correct for DNA contamination. The resulting cDNA products from the reverse transcription were divided into 20 aliquots using a Hydra 96 robot (Robbins Scientific, Sunnyvale, CA) for parallel Taqman PCR reactions using different primer and probe sets. Taqman RT-PCR was performed using an ABI 7700 sequence detector (Applied Biosystems, Foster City, CA) on the first-strand cDNA. The PCR reaction was performed at 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, 60°C for 1 min. A standard curve relating threshold cycle to template copy number was deduced using known dilutions of mouse genomic DNA (Clontech, Cambridge, UK). Specific PCR primers for genes/regions of interest were used (for sequence details, see supplemental Table 3, available at www.jneurosci.org supplemental material).
Lumbar DRGs were dissected and rapidly frozen [n = 3 animals per sample, 3 samples per tissue type (WT and −/−)]. The tissue samples were homogenized, and total RNA was obtained (TRIzol reagent; Invitrogen). Biotinylated cRNA was produced, hybridized to the Mouse Genome 430 2.0 array using standard methods, and array data were processed by GCOS 1.4 (Affymetrix, Santa Clara, CA). For sodium channels, probe set data are shown only if a present call occurred in at least one array and average probe signal intensity was >500 in at least one tissue type. If two or more qualifying probe sets were present per transcript on the array, the one with the highest average signal intensity across both comparisons was used. To be considered differentially regulated, a probe set had to be defined present, expression of >500, fold difference >1.5, and p value <0.05.
Statistical analysis was performed using one-way ANOVA, followed by Dunnett's or Student's t test, and data are represented as mean ± SEM. Analysis of the Taqman PCR data used ANOVA on log-transformed data.
Production of Nav1.9−/− mice
To investigate the functional role of the Nav1.9 α subunit, an Nav1.9 null allele mutant (Nav1.9−/−) was produced using a “knock-out vector” that disrupted the SCN11A gene with deletion of most of exon 4 and all of exons 5 and 6 (Fig. 1A). The Nav1.9−/− animals have no overt phenotype and appear indistinguishable from WT littermates. Locomotor activity, climbing, grooming, eating, and drinking behavior were measured, and differences between the Nav1.9−/− and WT animals were not considered significant (data not shown). The mutation of SCN11A does not result in a loss of DRG neurons, as indicated by presence of the nuclear lacZ reporter in adult DRG neurons in the Nav1.9−/− mice with a frequency similar to that of Nav1.9 mRNA in WT mice (Fig. 1B). More than 80% of the Lac-Z positive neurons colocalized with peripherin (a C-fiber marker) and ∼20% with NF200, a maker for neurons with myelinated axons (Fig. 1B), a similar proportion we find for Nav1.9 mRNA in WT mice. The relative proportion of peripherin- and NF200-labeled neurons in the Nav1.9−/− mice was not different from that in WT mice (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
Ion channel expression
Quantitative RT (QRT)-PCR analysis revealed that the levels in the DRG of the 5′ and 3′ regions of Nav1.9 mRNA were not different from WT but that exons 4–6 were absent in Nav1.9−/− mice. Nav1.3, Nav1.5, Nav1.7, and Nav1.8 mRNA levels were not altered in the Nav1.9−/− DRG relative to WT, as detected by both QRT-PCR and microarrays (Table 1). The relative numbers of Nav1.7- and Nav1.8-immunostained DRG neuronal profiles were also not altered (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). A significant decrease in mRNA for β1 and β4 sodium channel subunits was detected in DRGs from Nav1.9−/−mice, with no change in β2 and β3 (Table 1). Of 39,000 probe sets assayed by the microarrays, 111 had differential expression in the DRG between the WT and Nav1.9−/−, with 73 genes higher Nav1.9−/− than WT and 38 lower (supplemental Table 2, available at www.jneurosci.org as supplemental material).
Basal pain behavior
Basal mechanical and thermal pain behavior was unaffected in the Nav1.9−/− mice. Mechanical threshold (von Frey), paw-withdrawal latency to 50, 52, and 55°C hotplate simulation, and the number of flinches produced on exposure to a cold plate (0°C) for 5 min were identical in WT, Nav1.9−/+, and Nav1.9−/− mice (Fig. 2).
Fluoride-based internal solutions are used to stabilize recordings of TTXr-Per (Cummins et al., 1999). However, because fluoride causes a hyperpolarization shift in its voltage dependence (Coste et al., 2004), we preferred to use a chloride-based internal solution and a holding potential of −90 mV to remove resting inactivation. The activation threshold for Nav1.9 (approximately −60 mV) is more negative than that of Nav1.8 (∼30 mV) (Renganathan et al., 2002a; Coste et al., 2004), providing a window for observing Nav1.9 currents in WT DRG neurons. Na+ currents were recorded from 60 WT DRG neurons in the presence of 300 nm TTX using a series of depolarizing voltage commands from −100 to 60 mV for 120 ms. Two distinct subpopulations were identified among the small neurons recorded. One group (n = 27, 45%) exhibited both a TTXr-Per current between −60 and −30 mV that lasted the entire duration of the 120 ms voltage step and a TTXr-Slow current at more depolarizing voltage steps, whereas the other group of neurons (n = 33) exhibited only TTXr-Slow currents (Fig. 3A). In Nav1.9−/− DRG neurons, we observed only slow TTXr currents in all of the small DRG neurons recorded (n = 17) (Fig. 3A). We conclude that TTXr-Per currents are mediated by Nav1.9 and that the mice have a null mutation.
To study whether there are compensatory changes in sodium channel function in DRG neurons in the Nav1.9−/− mice, we subtracted TTXr currents (300 nm TTX) from total sodium currents to obtain TTXs currents. Peak current density of TTXs sodium currents in WT (n = 12) did not differ from those in Nav1.9−/− DRG neurons (n = 8, p = 0.7), and the TTXr-Slow current density in WT DRGs (n = 10) was comparable with that found in Nav1.9−/− neurons (n = 10). There is, therefore, no detectable compensatory change in sodium currents in the DRG of these Nav1.9−/− mice, in keeping with the unchanged channel mRNA levels. We conclude that the alterations in β1 and β4 in the knock-out mice are likely not functionally relevant.
Voltage-dependent activation and steady-state inactivation of sodium channels contributes to membrane excitability. We therefore studied the activation and inactivation properties of the TTXr current in both WT and Nav1.9−/− small DRG neurons. A series of depolarizing voltage commands from −100 to 60 mV were applied to activate sodium currents. Normalized peak conductance at each voltage step was plotted against voltage and fitted with the Boltzmann equation. We separately plotted the two subpopulations of small WT DRG neurons. The group with both TTXr-Per and TTXr-Slow current showed a significant (p < 0.05) hyperpolarization shift (V1/2 of −29.2 ± 2.7 mV) relative to that in the WT subpopulation with only TTXr-Slow currents (V1/2 of −8.1 ± 1.5 mV). Mean activation in Nav1.9−/− small DRG neurons had an intermediate value (V1/2 of −17.7 ± 1.3 mV). This population is a composite of those neurons that do not normally express Nav1.9 and those which lack Nav1.9 as a result of the knock-out (Fig. 3B). To study steady-state inactivation, a series of prepulses from −120 to 10 mV was followed by a testing pulse of 0 mV. The normalized peak current was plotted against the prepulse voltage and fitted with the Boltzmann equation. The voltage-dependent inactivation curves are comparable in the two WT and the KO neuronal populations (p > 0.05), with V1/2 of −17 ± 1 mV for WT DRGs with both TTXr-Per and TTXr-Slow (n = 11), V1/2 of −18.2 ± 1.2 mV for WT DRGs with only TTXr-slow currents (n = 15), and V1/2 of −21 ± 1.8 mV for Nav1.9−/− DRGs (n = 10) (Fig. 3B).
Double staining in DRG neurons for Nav1.9 immunohistochemically and for B2 bradykinin receptor mRNA by in situ hybridization or the TRPV1 or purinergic P2X3 receptor by immunohistochemistry revealed a high degree of colocalization in all cases (Fig. 4) (supplemental Table 1, available at www.jneurosci.org as supplemental material). We then tested whether the channel has a role in the generation of peripheral pain hypersensitivity by examining the effect of the null mutation of the channel on the direct and delayed pain responses to intraplantar injections of the inflammatory mediators bradykinin, PGE2, IL-1β, and NGF, the chemical irritant capsaicin, and P2X3 and P2Y receptor agonists.
Intraplantar bradykinin (300 ng) induced in WT mice an immediate paw licking/flinching response that was significantly briefer in Nav1.9−/− mice (Fig. 5A). Two hours after the bradykinin injection, WT mice had a reduction in the mechanical threshold for eliciting hindpaw withdrawal that was significantly less in Nav1.9−/− mice (Fig. 5B). Similarly, the reaction time on exposure to a 50°C hotplate decreased significantly after the bradykinin injection (2 h) in WT, but not in Nav1.9−/−, mice (Fig. 5C). WT mice exhibited a licking/flinching reaction immediately after intradermal injection of capsaicin (2.5 μg), a TRPV1 agonist, that, like bradykinin, was significantly briefer in Nav1.9−/− than WT mice (Fig. 5D). The capsaicin treatment produced at 25 min a mechanical hypersensitivity of the hindpaw in the WT mice that was significantly less in Nav1.9−/− mice (Fig. 5E). Both the selective P2X agonist αβ-met-ATP (Fig. 6A) and the P2Y agonist UTP (Fig. 6B) produced on intraplantar injection a rapid but relatively transient thermal hypersensitivity in WT mice that was reduced in Nav1.9−/− mice. We conclude that multiple inflammatory mediators and nociceptor receptor ligands acting on TRP channels, GPCRs, and ligand-gated ion channels produce on peripheral activation of nociceptors pain-like behavior and pain hypersensitivity that involves or requires Nav1.9.
PGE2 injection into the hindpaw produced a transient local mechanical allodynia that, like the other inflammatory mediators, was attenuated in the Nav1.9−/− mice relative to WT (Fig. 7A). The thermal hyperalgesia induced in WT mice by peripheral PGE2 was absent in the Nav1.9−/− mice (Fig. 7B). We also injected PGE2 into the subarachnoid space of the spinal canal to induce central sensitization (Harvey et al., 2004) (Fig. 7C,D). Spinal PGE2 (100 ng) induced an identical increase in thermal and mechanical sensitivity in Nav1.9−/−, Nav1.9+/−, and WT mice (Fig. 7C,D). We conclude that only the peripheral pain-producing actions of PGE2 require Nav1.9.
The inflammatory cytokine IL-1β (1 pg) administered into the hindpaw induced mechanical allodynia (Fig. 8A) and thermal hyperalgesia (Fig. 8B) that was significantly less in Nav1.9−/− than WT mice, but injection of NGF (50 ng), which is an inflammatory pain modulator (Lewin and Mendell, 1992; Woolf et al., 1994), into the hindpaw produced identical mechanical (Fig. 8C) and thermal (Fig. 8D) hypersensitivity in WT, Nav1.9+/−, and Nav1.9−/− mice. Most but not all proinflammatory agents require Nav1.9−/− to produce peripheral pain hypersensitivity.
Intraplantar injection of CFA in WT mice induced a localized inflammation of the hindpaw, the extent of which, measured by Evans blue extravasation, was similar in WT, Nav1.9+/−, and Nav1.9−/− mice. The peripheral inflammation produced in WT and Nav1.9+/− mice a significant reduction in the response latency to a hotplate at 50°C for 1–7 d (Fig. 9A). This inflammatory thermal hyperalgesia was absent, however, in the Nav1.9−/− mice (Fig. 9A). Three days after CFA injection, the latency of response to hotplate stimulation at 50, 52, and 55°C was significantly longer in the Nav1.9−/− mice than in WT mice (Fig. 9B). In contrast, WT, Nav1.9+/−, and Nav1.9−/− mice all developed similar degrees of mechanical hypersensitivity after intraplantar CFA administration (Fig. 10). Relative numbers of Nav1.9-positive profiles in WT L5 DRGs increased from 30% in naive to 50% after the CFA-induced inflammation (n = 5; p < 0.05) (Fig. 9C).
The decrease in mechanical threshold, the increase in pinprick response, and the cold allodynia (acetone) that occurs in a mouse spared nerve injury model (Bourquin et al. 2006) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material) and in a partial sciatic nerve injury (Seltzer) model (data not shown) were identical in WT, Nav1.9+/−, and Nav1.9−/− mice.
Disruption of exons 4–6 of the SCN11A gene results in a truncated transcript and complete absence of TTXr-Per in DRG neurons. All small (∼25 μm) diameter WT DRG neurons exhibit a TTXr-Slow current, and half have a TTXr-Per current as well. We never find any neuron in WT DRG cultures with TTXr-Per but no TTXr-Slow, consistent with our finding in rats that Nav1.9 always colocalizes with Nav1.8 in DRG neurons but that Nav1.8 is, in some rat DRG neurons, expressed without Nav1.9 (Amaya et al., 2000). The activation threshold of TTXr-Per was approximately −60 mV when recording with a chloride-based internal solution, which is relatively depolarized compared with fluoride-based solutions (Cummins et al., 1999). Fluoride, which inhibits phosphatases, produces a hyperpolarizing shift in TTXr-Per in myenteric sensory neurons (Rugiero et al., 2003). The presence in small DRG neurons of TTXr-Per caused a hyperpolarizing shift of activation of the total TTXr current but a similar inactivation. These results support a role for TTXr-Per in increasing membrane excitability and acting as a possible booster of subthreshold electrogenesis. We held the neurons at −90 mV to remove the resting inactivation of Nav1.9 and thereby maximally recruit the channels for analysis. This does not mean, however, that Nav1.9 is inactivated at resting membrane potentials. The half-maximal steady-state inactivation for Nav1.9 is approximately −45 mV (Cummins et al., 1999), and, at −60 or −70 mV, a reasonable fraction of Nav1.9 will still be available. A large number of channels with a low open probability attributable to ultraslow inactivation at the resting state will produce a small but consistent current, shifting the resting potential to a more depolarizing range and in this way will enhance the excitability of the cell (Cummins et al., 1999). QRT-PCR, microarray, and immunohistochemical studies failed to detect any compensatory alterations in other VGSCs in the Nav1.9 null mutant mice, and, although β1 and β4 accessory subunits decreased, these alterations did not affect the current density of TTXs and TTXr. Multiple small differences in many genes between WT littermates and the mutant mice detected by array analysis presumably reflect residual background differences, although the animals had been backcrossed for more than five generations. Although the neo cassette was not deleted, we found no indication in the microarray analysis of any change in expression of genes contiguous to SCN11A.
Involvement of Nav1.9 in peripheral nociceptor sensitization
Sensitizing agents act on nociceptor peripheral terminals via specific receptors expressed by the sensory neurons that are coupled with second-messenger systems to produce the changes that underlie peripheral sensitization. Nav1.9 appears from our data to be downstream of several quite distinct signal transduction pathways and to act as a point of convergence of multiple individual inflammatory mediators. Bradykinin evokes pain sensitivity in naive animals via the B2 bradykinin G-protein-coupled receptor (Dray and Perkins, 1993), which is coupled both to PKC-ε (Cesare and McNaughton, 1996) and TRP ankyrin repeat 1 (Bandell et al., 2004). UTP activates the P2Y purinergic GPCR, which is also coupled to PKC-ε. PKC-ε induces sensitization of nociceptors via changes in TRPV1 (Cesare et al., 1999) and contributes to inflammatory hyperalgesia (Hucho et al., 2005). PKC activators increase TTXr-Per current in rat and mouse DRG neurons (Baker, 2005) so that its sensitizing actions are likely to be attributable to changes in TRP and voltage-gated sodium channels.
PGE2 has a major role in the development of inflammatory pain and is produced by inflamed tissue as a result of induction of cyclooxygenase-2. There are four PGE2 G-protein-coupled receptor subtypes expressed by DRG neurons (EP1, EP3, and EP4). EP4 is coupled to Gs, some splice variants of EP3 to Gs and some to Gi, whereas EP1 is coupled to Gq/G11(Narumiya et al., 1999). PGE2, BK, and 5-HT all increase slow inactivating TTXr currents in DRG neurons with an increase in amplitude and rates of activation and inactivation (England et al., 1996; Gold et al., 1998). PKA potentiates Nav1.8 and depresses Nav1.7 currents, and PKC produces shifts in the steady-state activation of both channels in a depolarizing direction (Saab et al., 2003; Vijayaragavan et al., 2004). PGE2 increases TTXr-Per twofold in DRG neurons in a GPCR-mediated manner, with steady-state activation shifted 6–8 mV in the hyperpolarizing direction, and availability of the channels increased by 12 mV but with no detectable change in channel activation or inactivation kinetics (Rush and Waxman, 2004). This suggests that both PKC and PKA act on Nav1.9, as they do on Nav1.8. The diminished pain sensitivity we find after PGE2 treatment in Nav1.9 null mice is similar to that in another Nav1.9 knock-out mouse, with deletion of exons 2–4 (Priest et al., 2005). Because intrathecal application of PGE2 in the Nav1.9−/− mice produced similar pain hypersensitivity to that in wild-type mice, we conclude that the sodium channel has a minimal role in the spinal effects of PGE2, which appear to act mainly via the α3 subunit of glycine receptors (Harvey et al., 2004).
Capsaicin and αβ-met-ATP generate acute pain hypersensitivity through TRPV1 (Caterina et al., 1999) and P2X3 (Burnstock and Wood, 1996) ion channel/receptors, respectively. TRPV1 and P2X3 both increase intracellular calcium concentrations, which activates multiple calcium-sensitive kinases that can phosphorylate multiple ion channels and receptors. Although the expression of Nav1.9 and other receptors differs between mice and rats, Nav1.9 was highly coexpressed with TRPV1 and P2X3 in the mouse. The diminished behavioral response to capsaicin and αβ-met-ATP in the Nav1.9 null mouse indicates that the sodium channel may be downstream of calcium- as well as PKC- and PKA-dependent nociceptor sensitization.
IL-1β sensitizes primary sensory neurons, potentially by facilitating PGE2 and NGF synthesis (Safieh-Garabedian et al., 1995), in a manner we now show is Nav1.9 dependent. NGF has multiple actions on nociceptors via its TrkA receptor, including an action on TRPV1 receptors via (Chuang et al., 2001) or independent of (Zhang et al., 2005) PI3 kinase as well as slower onset changes that increase the levels of VGSCs or TRPV1 (Ji et al., 2002). We find no reduction in the acute hyperalgesia produced by NGF in Nav1.9 knock-out mice, indicating that its actions in generating peripheral sensitization do not appear in mice to require Nav1.9.
The Nav1.9−/− mice have markedly diminished thermal hyperalgesia for the week after peripheral inflammation induced by intraplantar CFA. Priest et al. (2005) also find a reduction in inflammatory (24 h after intraplantar carrageenan) thermal pain behavior in their Nav1.9 KO. This suggests that sensitization to heat after inflammation involves mediators that act through Nav1.9, such as BK, PGE2, or ATP, and that mechanical sensitization either relies on mediators that do not alter this channel (such as NGF) or is more a manifestation of central than peripheral sensitization.
We find that inflammation increases the number of Nav1.9-positive neurons in the DRG, consistent with our previous observation of an induction of Nav1.9 mRNA (Tate et al., 1998). Because neurons with both TTXr-Per and TTXr-Slow are likely to be more excitable than those with only TTXr-Slow, an increased number of Nav1.9-positive cells during inflammation might increase the numbers of neurons activated by peripheral stimuli. Deletion of Nav1.8 results in a delay in the development of inflammatory hyperalgesia (Akopian et al., 1999) but no difference in the magnitude of the hypersensitivity, although compensatory changes in Nav1.7 in the Nav1.8−/− mice have to be considered. Nociceptor-specific deletion of Nav1.7 induces insensitivity against noxious mechanical stimuli in naive mice and decreased inflammatory hyperalgesia (Nassar et al., 2004). Nav1.7 knock-out mice show no hyperalgesia in response to intraplantar NGF, indicating that Nav1.7 is downstream of different inflammatory mediators than Nav1.9.
Our results, like those of Priest et al. (2005), demonstrate that Nav1.9 does not contribute to neuropathic pain hypersensitivity. Nav1.9 expression is downregulated in injured neurons (Tate et al., 1998; Dib-Hajj et al., 1999) without a significant change in neighboring uninjured neurons (Decosterd et al., 2002). A Nav1.8 and Nav1.7 double knock-out mouse continues to show substantial hyperalgesia after nerve injury (Nassar et al., 2005). The data from Nav1.8 and Nav1.9 knock-outs imply that TTXr VGSCs may not contribute to the development of neuropathic pain, although a Nav1.8 knockdown produced by antisense oligodeoxynucleotide administration does reduce behavioral hyperalgesia after nerve injury (Lai et al., 2002).
We conclude that the Nav1.9 VGSC α subunit carries TTXr-Per in small DRG neurons and that it is a downstream effector of the increased pain sensitivity produced by diverse inflammatory mediators on nociceptor peripheral terminals. How the multiple different signal pathways converge on the channel, and the specific changes they produce that alter its properties, need now to be explored. Our data indicate, however, that sodium channel blockers that act on Nav1.9 may prove useful for the treatment of inflammatory pain.
This work was supported by GlaxoSmithKline and the National Institutes of Health Grants NS38253 and NS039518 (C.J.W.).
- Correspondence should be addressed to Clifford J Woolf, Neural Plasticity Research Group, Department of Anesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School, 149 13th Street, Charlestown, MA 02129.
- Akopian et al., 1996.↵
- Akopian et al., 1999.↵
- Aley et al., 2001.↵
- Amaya et al., 2000.↵
- Amir et al., 2006.↵
- Baker, 2005.↵
- Baker et al., 2003.↵
- Bandell et al., 2004.↵
- Bautista et al., 2006.↵
- Bourquin et al., 2006.
- Burnstock and Wood 1996.↵
- Caterina et al., 1999.↵
- Cesare and McNaughton, 1996.↵
- Cesare et al., 1999.↵
- Chuang et al., 2001.↵
- Coderre et al., 1989.↵
- Coste et al., 2004.↵
- Cummins et al. 1999.↵
- Decosterd and Woolf, 2000.↵
- Decosterd et al., 2002.↵
- Dib-Hajj et al., 1998.↵
- Dib-Hajj et al., 1999.↵
- Dray and Perkins, 1993.↵
- England et al., 1996.↵
- England et al., 2001.↵
- Gold et al., 1996.↵
- Gold et al., 1998.↵
- Gold et al., 2002.↵
- Harvey et al., 2004.↵
- Hucho et al., 2005.↵
- Ji et al., 2002.↵
- John et al., 2004.↵
- Julius and Basbaum, 2001.↵
- Lai et al., 2002.↵
- Levine and Reichling, 1999.↵
- Lewin and Mendell, 1992.↵
- McCleskey and Gold, 1999.↵
- Narumiya et al., 1999.↵
- Nassar et al., 2004.↵
- Nassar et al., 2005.↵
- Priest et al., 2005.↵
- Renganathan et al., 2002a.↵
- Renganathan et al., 2002b.↵
- Rugiero et al., 2003.↵
- Rush and Waxman, 2004.↵
- Saab et al., 2003.↵
- Safieh-Garabedian et al., 1995.↵
- Sangameswaran et al., 1997.↵
- Tate et al., 1998.↵
- Vijayaragavan et al., 2004.↵
- Woolf et al., 1994.↵
- Zhang et al., 1994.↵