Infantile Pain Episodes Associated with Novel Nav1.9 Mutations in Familial Episodic Pain Syndrome in Japanese Families

Painful peripheral neuropathy has been correlated with various voltage-gated sodium channel mutations in sensory neurons. Recently Nav1.9, a voltage-gated sodium channel subtype, has been established as a genetic influence for certain peripheral pain syndromes. In this study, we performed a genetic study in six unrelated multigenerational Japanese families with episodic pain syndrome. Affected participants (n = 23) were characterized by infantile recurrent pain episodes with spontaneous mitigation around adolescence. This unique phenotype was inherited in an autosomal-dominant mode. Linkage analysis was performed for two families with 12 affected and nine unaffected members, and a single locus was identified on 3p22 (LOD score 4.32). Exome analysis (n = 14) was performed for affected and unaffected members in these two families and an additional family. Two missense variants were identified: R222H and R222S in SCN11A. Next, we generated a knock-in mouse model harboring one of the mutations (R222S). Behavioral tests (Hargreaves test and cold plate test) using R222S and wild-type C57BL/6 (WT) mice, young (8–9 weeks old; n = 10–12 for each group) and mature (36–38 weeks old; n = 5–6 for each group), showed that R222S mice were significantly (p < 0.05) more hypersensitive to hot and cold stimuli than WT mice. Electrophysiological studies using dorsal root ganglion neurons from 8–9-week-old mice showed no significant difference in resting membrane potential, but input impedance and firing frequency of evoked action potentials were significantly increased in R222S mice compared with WT mice. However, there was no significant difference among Nav1.9 (WT, R222S, and R222H)-overexpressing ND7/23 cell lines. These results suggest that our novel mutation is a gain-of-function mutation that causes infantile familial episodic pain. The mouse model developed here will be useful for drug screening for familial episodic pain syndrome associated with SCN11A mutations.


Introduction
The recent identification of genetic variants causing pain-related disorders has created a new window into the etiology of hereditary pain. Voltage-gated sodium ion channels have been intensively studied, with a broad spectrum of new findings uncovered for peripheral neuropathy. Among the nine subtypes (Nav1.1-Nav1.9) encoded by the genes SCN1A-SCN5A and SCN8A-SCN11A, three subtypes (Nav1.7, Nav1.8, and Nav1.9) are strongly expressed in sensory neurons, and have been associated with human pain disorders. Mutations in SCN9A cause a spectrum of clinical phenotypes ranging from congenital insensitivity to pain [1], as well as chronic pain syndromes such as paroxysmal extreme pain disorder [2,3], inherited erythromelalgia [4,5], and small fiber neuropathy [6]. In contrast, Nav1.8 is associated with phenotypes ranging from painful symptoms [7] to Brugada syndrome [8].
Nav1.9, which is encoded by SCN11A, is a tetrodotoxin (TTX)-resistant (TTXr) channel primarily expressed in nociceptors. Nav1.9 channels exhibit hyperpolarized voltage-dependent activation close to the resting membrane potential (RMP), with slow inactivation leading to a persistent current [9]. Hence, Nav1.9 is thought to be an important regulator of membrane excitability. Recently Nav1.9 was identified as the genetic background for some peripheral pain syndromes. In a German family with an unusual syndrome involving loss of pain sensation, a de novo missense SCN11A mutation (p.L811P; we have abbreviated the prefix "p" in this paper) was found [10]. In another screening of SCN11A in 58 independent individuals with early onset severe sensory loss, a Swedish male was identified as a patient with exactly with the same heterozygous mutation (L811P), again occurring as a de novo event. This Nav1.9 variant was examined electrophysiologically in Scn11a +/L799P knock-in mice (L799P being the mouse ortholog), and exhibited excess sodium ion influx at rest with subsequent cell depolarization [10].
In contrast, two large multigenerational Chinese families with autosomal-dominant episodic chronic pain and two types of gain-of-function SCN11A mutations (R225C and A808G) were identified by genome-wide linkage analysis combined with whole-exome sequencing [11]. Overexpression of these mutants in mouse dorsal root ganglion (DRG) cells induced hyperexcitability [11]. electrophysiological mechanisms of the Nav1.9 variant in DRG neurons and Nav1.9-overexpressing ND7/23 cell lines, and found that the R222S mutation was associated with increased firing probability in DRG neurons.

Ethical statements
A part of the human study was approved by the Institutional Review Board and Ethics Committee of Kyoto University School of Medicine (approval no. G501; approval date 2 Aug. 2012), and Japan and Akita University Graduate School of Medicine (approval no. 960; 26 Sep. 2012), Japan. Written informed consent was obtained from all subjects and the parents of children and adolescents before participation.
Animal studies including animal care and all experimental procedures were in accordance with the Animal Welfare Guidelines of Kyoto University or the in-house guidelines of the Institutional Animal Care and Use Committee of Daiichi Sankyo Co., Ltd. Animal experiment protocols were reviewed and approved by the Animal Care, Use and Ethics Committee at Kyoto University (approval no. MedKyo15074; 27 Mar. 2015) and the Institutional Animal Care and Use Committee of Daiichi Sankyo Co., Ltd. (approval no. A1500635, 26 Mar. 2015 [8-9-weekold mice experiment]; A1502171, 16 Oct. 2015 [36-38-week-old mice experiment]).

Pedigrees and genomic DNA isolation
Six unrelated families were examined in this study (Fig 1A). Each family included a number of affected and unaffected members. Peripheral blood was collected from 23 affected members and 13 unaffected members from these families. Genomic DNA was extracted from whole blood samples using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany).

Behavioral experiments
Male knock-in (R222S mouse) and wild-type C57BL/6 mice (WT) were transferred from Kyoto University to Daiichi Sankyo Co., Ltd. Mice were housed in an air-conditioned room with controlled temperature (23 ± 2°C) and humidity (55 ± 10%) and a 12-h light/dark cycle (light from 7:00 to 19:00). They were fed and given water ad libitum. In the following behavioral experiments, the mice were used at 8-9 or 36-38 weeks old.
Irwin's test: multiple neurobehavioral signs (a total of 45 observation items including awareness, mood, motor activity, CNS excitation, posture, motor coordination, muscle tone, reflexes, and autonomic neurological signs and miscellaneous symptoms) were observed according to the method of Irwin [18].
Locomotor activity: locomotor activity was measured using SCANET MV-40 (MELQEST, Toyama, Japan), a device that uses infrared sensors to detect movement of an animal placed individually in a transparent plastic cage (20 cm × 20 cm × 20 cm). Once the animal was placed into the cage, detection of movement was started. Total horizontal movement was recorded for 30 min.
Von Frey test: sensitivity to a mechanical stimulus was assessed using von Frey filaments (Aesthesio 1 , DanMic Global, LLC, San Jose, CA, USA). Each animal was placed individually into a transparent plastic cage (8 cm × 7 cm × 8 cm) with a wire-mesh floor to allow insertion of the filament from below. The filament was placed against the plantar surface of the left hindpaw and acclimatized for not less than 30 min before testing. A 50% threshold, which indicates the filament force at which an animal reacts in 50% of presentations, was determined using a previously reported method [19]. Eight filaments with approximately equal logarithmic incremental forces (0.02, 0.04, 0.07, 0.16, 0.4, 0.6, 1.0, and 1.4 g) were used.
Hargreaves test: sensitivity to a thermal stimulus was assessed using the PAW THERMAL STIMULATOR (UCSD, San Diego, CA, USA). Each animal was placed individually in a transparent plastic chamber (8 cm × 8 cm × 18 cm) with a glass floor, and acclimatized for not less than 30 min before testing. A radiant light heat source located under the floor was focused onto the plantar surface of the left hindpaw, and paw withdrawal latencies recorded. Latencies were measured in triplicate for the left hindpaw of each animal, and the mean used for analysis. A 20 sec cut-off time was employed to avoid tissue damage.
Cold plate test: sensitivity to a cold stimulus was assessed using the MK-350HC Hot/Cold Plate Analgesia Meter (Muromachi Kikai Co., Ltd., Tokyo, Japan). Each animal was placed individually in a transparent plastic cylinder cage (8 cm φ × 20 cm) with a cold plate floor at 10°C. The total number of nociceptive behaviors (e.g., lifting and shaking of paws and jumping) was counted for 3 min.

Real-time quantitative PCR
cDNA was obtained from Nav1.9-overexpressing ND7/23 cells using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Real-time quantitative PCR for human SCN11A cDNA was performed using the primers described in S4 Table. Target cDNA expression levels were normalized to the corresponding Ppia expression levels.

Electrophysiology
Nav1.9-overexpressing ND7/23 cells. Nav1.9-overexpressing ND7/23 cells (WT and mutant [R222H, R222S]) and control cells (ND7/23 cells without Nav1.9 transfection) were inoculated onto 12 mm φ glass coverslips, and incubated at 28°C with 5% CO 2 for 16-24 h before patch clamp recording. Currents were recorded using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA), low-pass filtered at 5 kHz, and digitized using a Digidata 1440A (Axon Instruments, Union city, CA, USA). The pipette solution contained (in mM): 10 NaF, 110 CsF, 20 CaCl 2 , 2 EGTA, and 10 HEPES (pH 7.3 with KOH). The bath solution contained (in mM): 130 NaCl, 5 KCl, 1 MgCl 2 , 2 CaCl 2 , 10 glucose, and 10 HEPES (pH 7.4 with NaOH). To select cells with overexpressed sodium channels and block TTX-sensitive (TTXs) channels, blasticidin (50 μg/ml) and TTX (3 μM; Wako Pure Chemical Industries, Osaka, Japan) were added to the bath solution. Patch pipettes were fabricated from thin-walled borosilicate glass capillaries (King Precision Glass, Inc., Claremont, CA, USA), and had a resistance of~3.5 MO [22]. Series resistance was monitored during the recording session, and data were rejected if their value changed by > 20%. Leak currents were subtracted using electrical subtraction protocols (p/4). Data were acquired by pCLAMP 10 (Axon Instruments). Voltagedependent currents were acquired after achieving more than 5 min of whole-cell recording conditions. To record the current/voltage (I/V) relationship of TTXr, cells were held at −100 mV and step pulses applied from −120-30 mV for 100 ms in 10-mV increments. We did not include data for WT, R222S and R222H with the peak currents more than −10 pA, which corresponds with the peak amplitude of the control cells (See results). Current traces were filtered with a low-pass Bessel setting of 10 kHz. Activation curves were obtained above the step pulse and fitted to the Boltzmann function in the form where G max is the maximum conductance, V 1/2 is the mid-point of the activation curve, and k is the slope factor. All data were obtained at 28°C.
Isolated DRG neurons. Electrophysiological data from isolated DRG neurons were collected from small diameter (< 25 μm) cells from WT and knock-in mice. Data were obtained at 23-25°C within 8 h after isolation using an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany). Patch pipettes were fabricated from thin-walled borosilicate glass capillaries (GC150TF-10, Harvard Apparatus, MA, USA) and had a resistance of 4-6 MO [23,24]. Patch pipette tips were fire-polished before use. Electrode capacitance was compensated electrically, and series resistance was < 13 MO. Data were collected in cells were which required smaller current than -30pA to hold the membrane at -60 mV. A previous study found that mouse DRG neurons initially up-regulate I(NaN) after cell rupture [25], therefore to avoid dispersion of this effect on action potential (AP) firing, current-clamp recordings were obtained after achieving more than 5 min of whole-cell recording conditions. The pipette solution contained (in mM): 67 KCl, 65 K-gluconate, 1 MgCl 2 , 5 EGTA, 4 ATP-Mg, 1 GTP-Na 2 , and 10 HEPES (pH 7.3 with KOH). The bath solution for isolated DRG neurons contained (in mM): 130 NaCl, 5 KCl, 1 MgCl 2 , 2 CaCl 2 , 10 glucose, and 10 HEPES (pH 7.4 with NaOH). TTX (1 μM) was added to the bath solution to block TTXs sodium channels.
RMP was measured when the membrane potential was stabilized over 5 min after the whole-cell recording configuration was achieved. RMP was measured immediately before current injection in every trial. The following parameters of the first AP were measured: amplitude, 50% AP width, and maximum rate of rise/fall of AP. To evaluate input impedance, the voltage response amplitude was measured at a current injection of 10 pA. AP frequency was calculated from the AP number during step current injections (500 ms) from 10 to 210 pA in 25 pA increments. The current injection was limited to 210 pA because firing was not observed with currents larger than 210 pA owing to sodium channel inactivation. Cells that did not generate APs [26] or had only one AP in response to a 500-ms current stimulus in all step pulses were excluded from analysis. All data were acquired using Patchmaster (HEKA Elektronik). Electrophysiological data of isolated DRG neurons were analyzed using Igor Pro (WaveMetrics Inc., Portland, OR, USA).

Statistical analysis
Behavioral test and electrophysiological data were presented as mean ± standard error of the mean (S.E.M.). Statistical significance was determined by two-sided Student's t-tests. All statistical analyses were performed using SAS software (Version 9.4; SAS Institute, Cary, NC, USA). Statistical significance was indicated by p < 0.05.

Phenotypic characterization
The clinical features of the six pedigrees examined in this study (Fig 1A) are described below.
Family 1. The proband (V-1) is a 4-year-old girl born to nonconsanguineous parents after an uneventful delivery. Recurrent, unexpected crying and unpleasantness were recognized during her infancy. Around the age of 2 years, the mother noticed this was due to paroxysmal limb pain. The patient grew up normally and developed without other problems. The pain was partially relieved with oral acetaminophen, but was persistent, leading to referral to our hospital at the age of 4 years.
The characteristics of her limb pain were that it began abruptly and continued for 15-30 min, but occasionally for half or even a whole day. Typically the pain was induced by fatigue or weather changes, specifically atmospheric depression leading to cold weather, rain, and snow. The paroxysmal pain episodes occurred approximately 10 and 20 days a month during summer and winter, respectively. The patient is afebrile during an episode and has a cold sensation on the affected limb, being one or two elbows, wrists, knees, and ankles, without any other symptoms e.g., redness, rashes, swelling, and joint destruction. Physical examination did not reveal any abnormalities. Laboratory assessment did not show any suggestion of collagen diseases, autoimmune diseases, or other disorders. Her laboratory data was normal, as follows: peripheral blood count, coagulation system, erythrocyte sedimentation rate, C-reactive protein, ferritin, serum amyloid A protein, soluble interleukin receptor 2, IgD, complement levels, antinuclear and anti-cardiolipin antibodies, rheumatoid factor, anti-CCP antibody, serum lactate and pyruvate levels, amino acid analysis, and leukocyte alpha-galactosidase activity.
The father of the proband (IV-1), a 41-year-old man, showed similar symptoms during childhood. However, his paroxysmal pain episodes gradually decreased from around the age of 15 years, and now occur only once or twice a year. He sometimes has a throbbing headache, but has not yet been assessed in a hospital.
The other patients (IV-3, III-1, III-5, III-6, and II-5) all showed a similar clinical course. Their paroxysmal pain episodes decreased during their adolescence. This family is originally from Akita Prefecture.
Family 2. This family has previously been reported [13]. The proband (III-2), a 51-yearold woman, had paroxysmal and cyclic limb pain in her childhood. Although her paroxysmal pain episodes gradually decreased from around 15 years of age, intermittent episodes made her visit a physician when she was 36 years old. She noticed her son had the same symptoms (IV-3). The limb pain affected her quality of life; e.g., she could not attend elementary school. From our investigation, we identified four males and three females in her family who all showed clinical features similar to the proband's. Their limb pain is almost identical to that of Family 1. However, it lasts for a longer time in some individuals and repeats several times after an intermittent period of a few hours, resulting in a series of pain. Their limb pain characteristically precedes weather changes, in particular before low air pressure and bad weather; therefore, they can anticipate impending bad weather. Indeed, "weather forecaster" has become their nickname within the family. With regards age dependency, the paroxysmal pain episodes gradually decreased from around 15 years of age in all affected individuals. They have frequently used acetaminophen for pain relief. This family is originally from Chiba Prefecture.
Family 3. This family has previously been reported [14]. Briefly, eight males and five females are affected with similar limb pain as in the other families. The proband (IV-3), a 3-year-old boy, started recurrent episodes of unexplained crying at 6 months of age, which was recognized to be the result of limb pain. Besides the shared feature of limb pain, he occasionally had migraine episodes. Paroxysmal limb pain and migraine have not occurred simultaneously. This family is originally from Tottori Prefecture.
Families 4, 5, and 6. Four males and three females are recognized as affected with paroxysmal limb pain in Family 4. Ten males and 10 females are confirmed to have paroxysmal limb pain in Family 5. Headache (not yet diagnosed as migraine) is another complaint in an individual of this family (V-2). Finally, six males and nine females had paroxysmal limb pain during their infancy in Family 6. In these three families, age dependency was confirmed and the paroxysmal limb pain episodes decreased from around 15 years of age. These families are originally from Akita Prefecture.
In summary, the affected joints in frequency order are: knees, ankles, wrists, and elbows. Occasionally, the pain is localized to the forearms, brachium, palm, fingers, thigh, lower leg, and acrotarsium. The pain usually occurs on one side of the extremities, but occasionally on both sides. The pain typically lasts for 15-30 min and recurs several times a day. Moreover, it tends to be more frequent and severe at night than daytime. The patients had recurrent episodes of night-time crying and sleep disturbances in childhood. These episodes were usually present during the first few days of a series of painful days. The pain is often induced by fatigue and a prelude of bad weather. The patients feel cold in the pain region and get relief by warming the lesion. The paroxysmal pain episodes appear to start during infancy and gradually decrease from around 15 years of age. Finally, six affected individuals (family 2, IV-3; family 3, III-3, IV-3, IV-4; family 6, V-3, VI-2) complained of anorexia and diarrhea following cessation of a series of pain. These gastrointestinal manifestations are not observed in other mutation carriers or unaffected members, being not segregated by Nav1.9 mutation.

Clinical characterization
Almost all patients had a cold sensation in the affected limb. Therefore, we evaluated body surface temperature by thermography in four patients during their pain attacks.
In a previous report [14], the surface of the affected limb had a lower temperature than the contralateral side. In Family 3, it was higher in the affected forearm at the end of the pain attack. We also examined body surface temperature in the other families (1, 2, 4, and 6), but no significant difference was detected between the affected and contralateral limbs.
Nerve conduction velocity of the right median, right ulnar, right peroneal, and right tibial nerves was individually examined in an individual from Family 1 (IV-1), and all showed normal findings. None of the affected subjects had shown a past electrocardiographic abnormality.

Genetic analysis
To identify the causative genes, we used a combined approach of linkage analysis and exome sequencing. We first performed linkage analysis using Family 1 and 2, and identified a single locus on chromosome 3p22 with a LOD score of 4.32 (Fig 2), suggesting that the causative gene is located in this region.
We next performed exome analysis using three families (Fig 1A). Whole-exome sequencing generated 9.3-14.7 billion bases from nine affected and five unaffected individuals. After mapping to the human reference genome (UCSC Genome Browserhg19), we obtained 6.8-12.0 Gb of targeted exome sequence suitable for mapping, with a mean sequencing depth between 64-and104-fold. The mean percentage of exomes covered a read depth of 8-fold, which was > 94.7%. Filtering of variants was performed, with a focus on less common, deleterious variants in coding sequence regions. We removed synonymous and low coverage (read depth < 8) variants, and filtered against two reference databases (dbSNP135 and 1000 Genomes). Given that the families show an autosomal-dominant inheritance pattern, we selected variants that were heterozygote in affected members and not present in unaffected members. Consequently, 44, 10, and 91 variants were identified in Family 1, 2, and 3, respectively (S5 Table). Further searching within the families for variants in the same gene on chromosome 3p22 showed that all three families harbored a variant in the same codon of the SCN11A gene: R222H (Family 1 and 2) and R222S (Family 3) (Fig 3). No other variant was shared among the three families. We found no alterations in either SCN9A or SCN10A.
To determine segregation and confirm our exome analysis, direct sequencing of SCN11A was performed in all six families. We found complete segregation with the episodic pain phenotype of R222H in Family 1, 2, 4, 5, and 6, and R222S in Family 3 (Fig 1).
To investigate the MAF, we genotyped SCN11A R222H and R222S in 105 in-house Japanese controls by RFLP. Carriers of these two variants were not found in these controls.
Because five (Family 1, 2, 4, 5, and 6) out of six families harbor the same mutation i.e., R222H, we compared the haplotypes in these families using SNPs. Three different haplotypes were identified, indicating that R222H is derived either from a founder mutation or a private mutation (Fig 4).

Current-clamp characterization of DRG neurons in R222S and WT mice
To assess the effect of the R222S mutation on DRG neuronal excitability, we performed current-clump recordings in small DRG neurons (< 25 μm) from R222S and WT mice (Fig 6). We excluded the population of neurons that did not generate APs or had only one AP firing from our analysis, since > 80% of DRG neurons that generated APs had repetitive APs.
There was no significant difference in RMP of DRG neurons between WT and R222S mice (WT, −56.17 ± 9.64 mV, n = 3; R222S, −51.92 ± 5.74 mV, n = 4; p = 0.54). In contrast, the amplitude of the first APs was significantly different between WT and R222S mice when measured using a current injection of 210 pA (WT, 48.17 ± 26.76 mV, n = 5; R222S, 26.16 ± 12.51 mV,  Fig 6B and 6C), while input impedance in R222S mice were significantly higher than in WT mice (WT, 262.0 ± 118.57 MO, n = 6; R222S, 677.43 ± 209.44 MO, n = 5; p < 0.05, Fig 6D). These results suggest that DRG neurons of R222S mice can fire APs even with a low stimulus input. Next, we examined firing frequency in WT and R222S mice by current injection. Representative responses for DRG neurons from R222S and WT mice are shown (Fig 6E). In R222S mice, the firing frequency increased with current injection, while firing frequency remained at a low level in WT mice. Accordingly, firing frequency (calculated from the average number of APs in 500 ms) was significantly higher (using a 185 pA input current) in R222S mice than WT mice (Fig 6E). These results suggest that excitability is higher in R222S mice than WT mice.

Discussion
Here, we identified patients with typical childhood episodic pain symptoms in six unrelated families from throughout Japan, and further, identified two missense mutations in SCN11A (R222H and R222S) as the cause of the episodic pain. Given that these Nav1.9 variants are both gain-of-function mutations, the phenotypes elicited by the mutations are divergent. There are three subtypes of clinical manifestations i.e., loss of pain, painful small fiber neuropathy, and familial episodic pain (Table 1) in affected individuals with Nav1.9 variants (described in previous reports [10-12, 27, 28]). While L811P is associated with loss-of-pain phenotypes [10], other variants (e.g., G699R, L1158P, and I381T) are associated with small fiber neuropathy. These variants cause painful small fiber neuropathy of late onset i.e., fifth or sixth decades, and are complicated by skin discoloration and burning sensation, and autonomic dysfunction including a sensation of dry eyes, diarrhea, urinary problems, palpitations, and orthostatic dizziness [12,27]. The third phenotype is familial episodic pain. The Nav1.9 mutations, R225C, A808G, and V1184A have previously been reported in two Chinese families [11] and a family of mixed European ancestry [28]. These affected cases and our affected subjects share similar clinical characteristics (Table 1): early infantile onset of episodic pain and gradual amelioration with growth, primary localization of pain in the lower limbs, induction by cold exposure or weather changes, and alleviation of pain with anti-inflammatory medication. In some cases, it may be misdiagnosed with other diseases such as growing pains, because of gradual amelioration in adolescence and no abnormal laboratory findings. Clinical severity and symptoms are diverse, even in the same family, suggesting that the phenotype is modified by non-genetic factors. Further evaluation of the genotype-phenotype correlation is needed.
In our study, all patients had mutations at the arginine amino acid in position 222. Haplotype analysis revealed three different haplotypes carrying R222H, indicating that R222H is derived not only from a founder mutation, but also as an independent mutation. Another mutation, R222S, also occurred at the same position. Such a coincidental amino acid position with a childhood-restricted episode strongly suggests that gain-of-function SCN11A mutations and a painful childhood syndrome may have been overlooked in Japan because of self-limited mitigation of pain episodes by growth. We are tempted to speculate that our cases represent only the tip of the iceberg of such infantile patients.
Our behavioral experiments using knock-in mice successfully replicated the characteristic features of painful behavior i.e., elevated cold sensitivity in patients. Furthermore, our in vivo study confirmed that R222S is the gain-of-function mutation, with up-regulated excitability of small DRG neurons observed in R222S knock-in mice by electrophysiological experiments. Correspondingly, R222S mice demonstrate a chronic pain phenotype similar to that in human. This chronic pain phenotype is explained by increased DRG neuronal excitability in juvenile mice. It suggests that a mechanism of this pain is involved in higher electrical activity, thereby promoting AP firing in DRG neurons. Interestingly, the symptoms in these mice did not disappear after growth, which is different from the human case. A previous study has shown that Nav1.9 glycosylation state is paralleled by developmental change. Moreover, enzymatic desialidation of neonatal DRG cultures change TTXr properties in rat DRG cells [29]. However, in the present study, the mouse R222S knock-in model did not replicate growth-dependent action potentials between WT and R222S mice. Open and closed circles represent WT (n = 6) and R222S mice (n = 5), respectively. The range of 500-ms-step current injections was 10-210 pA. Data are presented as mean ± S.E.M. (*p < 0.05; two-sided Student's t test).
Recent studies in humans have reported that 12 amino acid substitutions of Nav1.9 are relevant to pain disorders, with seven of them associated with gain-of-function and painful/painless phenotypes [10-12, 27, 28] (Table 1). Particularly, in a Chinese family with autosomaldominant episodic pain, a point mutation of SCN11A (R225C) alters the firing property despite no significant difference in RMP and threshold [11]. This mutation is positioned at the S4 segment, which is the voltage censor domain. A previous study has shown that the S4 region in all four domains of rat brain IIA and II sodium channels are involved in both activation and inactivation gating [30,31]. Therefore, the R225C mutation was expected to change voltage-dependent gating of ion channels. However, G max and I max does not change significantly [11]. In mice, the R222S mutation is also in one of the multiple repeats of a motif consisting of a positively charged residue in the S4 segment, and did not change RMP by current-clamp recording. We attempted to confirm the gating properties of the Nav1.9 mutations, R222S and R222H. Activation curves from Nav1.9 mutation-overexpressing ND7/23 cells were not significantly different with voltage steps. Our results are in accordance with observations on the R225C mutation, which is located near this mutation [11]. However, inactivation gating was not confirmed in Nav1.9-overexpressing ND7/23 cells because these currents were so small. A lack of inactivation data is one limitation of our current study. Thus, understanding of the detailed mechanism underlying the gating property still requires further investigation. Nav1.7, 1.8, and 1.9 are preferentially expressed in peripheral neurons and are implicated in injury-induced neuronal hyperexcitability [11,32,33]. Recent studies have revealed an interaction between a scaffolding protein and TTXs/TTXr neuronal channels. The binding domains are weakly conserved and the core motif of the domain shared between potassium channels [34,35]. Particularly, Nav1.8 interacts constitutively with ankyrin G in skin nerve fibers. This constitutive binding may contribute to pathological aspects of the illnesses caused by Nav1.8 ectopic expression [36,37]. We demonstrated that the amplitude of the first AP was changed by R222S mutation. The ratio of TTXs/TTXr currents will affect excitability [38]. Additionally, Kv1 expression is reported in small DRG neurons [39]. Kv1 channels control the firing properties of neurons to generate a single spike at the onset of stimulus, and application of DTX, a selective potassium channel blocker, blocks Kv1 channels and induces repetitive firing in the auditory system [40,41]. We found that the R222S mutation altered input resistance but did not significantly affect the maximum rate of rise/fall of AP. These results are suggestive of a mechanism whereby the R222S mutation may change the interplay with axonal channel components in AP firing, including channel localization and distribution. Impaired Nav1.9 function is correlated with three types of neuronal phenotypes: painless, small fiber neuropathy, and familial episodic pain syndrome ( Table 1). Onset of the latter two phenotypes seems to be highly age dependent, but with the age of onset reversed: elderly for small fiber neuropathy and infancy for familial episodic pain syndrome. Genotype and phenotype association has never been explored for Nav1.9 age dependency. A previous study has shown that Dravet Syndrome, with progressive infantile onset epileptic encephalopathy caused by loss-of-function mutations in SCN1A, is correlated with natural loss of Nav1.3 channel expression in brain development, coupled with failure of increased functional Nav 1.1 channels [42]. Expression of TTXs and TTXr is regulated by developmental changes of neurotropic factors and/or a shift in expression of specific neurotrophic receptors [43]. Thus, further studies are warranted to elucidate the mechanisms of such an age dependency.

Conclusions
In the present study, we identified SCN11A R222H and R222S in six unrelated Japanese families with familial episodic pain syndrome. For R222S, we show the pathological role of this mutation in familial episodic pain syndrome using a mouse knock-in model combined with behavioral and electrophysiological examinations. The mouse model developed here will be useful for drug screening for familial episodic pain syndrome associated with SCN11A.