Peripheral Ion Channel Genes Screening in Painful Small Fiber Neuropathy

Neuropathic pain is a characteristic feature of small fiber neuropathy (SFN), which in 18% of the cases is caused by genetic variants in voltage-gated sodium ion channels. In this study, we assessed the role of fifteen other ion channels in neuropathic pain. Patients with SFN (n = 414) were analyzed for ANO1, ANO3, HCN1, KCNA2, KCNA4, KCNK18, KCNN1, KCNQ3, KCNQ5, KCNS1, TRPA1, TRPM8, TRPV1, TRPV3 and TRPV4 variants by single-molecule molecular inversion probes–next-generation sequencing. These patients did not have genetic variants in SCN3A, SCN7A-SCN11A and SCN1B-SCN4B. In twenty patients (20/414, 4.8%), a potentially pathogenic heterozygous variant was identified in an ion-channel gene (ICG). Variants were present in seven genes, for two patients (0.5%) in ANO3, one (0.2%) in KCNK18, two (0.5%) in KCNQ3, seven (1.7%) in TRPA1, three (0.7%) in TRPM8, three (0.7%) in TRPV1 and two (0.5%) in TRPV3. Variants in the TRP genes were the most frequent (n = 15, 3.6%), partly in patients with high mean maximal pain scores VAS = 9.65 ± 0.7 (n = 4). Patients with ICG variants reported more severe pain compared to patients without such variants (VAS = 9.36 ± 0.72 vs. VAS = 7.47 ± 2.37). This cohort study identified ICG variants in neuropathic pain in SFN, complementing previous findings of ICG variants in diabetic neuropathy. These data show that ICG variants are central in neuropathic pain of different etiologies and provides promising gene candidates for future research.


Introduction
Neuropathic pain (NeuP) is defined as a pain condition usually caused by progressive nerve disease [1]. NeuP symptoms are often described as a shooting or burning pain accompanied by allodynia, hyperalgesia, sensory dysfunction and autonomic complaints [2]. Chronic pain is common in peripheral neuropathy, including diabetic neuropathy (DN) and small fiber neuropathy (SFN), where Aδ-fibres and C-fibres are affected [3,4]. Patients suffering from neuropathic pain report major negative impact on quality of life [5]. Unfortunately, the currently available treatment has a moderate effect and often does not bring the expected pain relief [2,6]. Several conditions such as diabetes mellitus, autoimmune disorders, viral infections, inflammatory disorders and chemotherapy have been linked to NeuP, but the pathophysiology is largely unresolved [2]. An increasing number of reports highlight a role for genetic factors involved in pain development [1,7,8]; still, in more than 80% of the cases, a possible genetic factor is unknown [9].
In the literature, other ion channels' genes (ICGs) have also been reported in pain modulation, mainly transient receptor potential (TRP) cation channels [16], potassium voltagegated (Kv) channels [17], hyperpolarization-activated and cyclic nucleotide-gated channels (HCN) [18] and Ca 2+ -activated Cl-channels, also known as anoctamins (ANO) [19]. TRP channels function as thermal, chemical and mechanical sensors [16]. Kv are a group of potassium channels involved in the modulation of sensory neuron excitability and pain processing [17]. HCN channels exhibit wide expression in peripheral nerves, and their impaired functioning has been linked to neuropathic pain [18]. The most studied member of the ANO family, ANO1, has been found to interact with TRPV1, leading to increased pain in sensory neurons [20]. Moreover, ANO3 modulates nociception in the dorsal root ganglion (DRG) via enhancement of the sodium-activated potassium channel Slack activity [21].
In a previous study, we investigated the role of variants in 15 ICG in different patient cohorts with painful and painless DN. GOF and LOF variants were present in both groups, suggesting that ICG variants contribute to NeuP [22]. Therefore, in this study, we extended the analysis of these ICG to a different patient cohort of neuropathic pain, SFN. We focused specifically on patients with SFN without a defined underling genetic cause, as analysis of the SCN3A, SCN7A-SCN11A and SCN1B-SCN4B genes was negative. We performed single-molecule molecular inversion probes-next-generation sequencing (smMIPs-NGS) to analyze exons and exon-intron junctions and classified the identified variants.

Genetic Variants Identified in Ion Channel Genes
Among 414 patients negative for SCN3A, SCN7A-SCN11A and SCN1B-SCN4B screening, 19 different potentially pathogenic variants were identified in ICGs. All detected variants were classified as VUS and localized in seven ICG genes: ANO3, KCNK18, KCNQ3, TRPA1, TRPM8, TRPV1 and TRPV3. In total, two missense variants were detected in ANO3, one frameshift variant leading to a shorter protein in KCNK18, two missense in KCNQ3, four missense and two nonsense variants in TRPA1, three missense in TRPM8, three mis-sense in TRPV1, one splice variant in the donor site of intron 9 and one missense in TRPV3 ( Table 1). All of them were heterozygous. The detected nonsense/frameshift variants were predicted either to cause a shortening of the protein length due to a premature stop codon or, most likely, cause nonsense-mediated decay (NMD) of the mRNA, while the splice site variant in the donor site of intron 9 (TRPV3 c.1242+1G>A) was predicted to alter the protein sequence via alternative splicing. Although the best predictable functional effect will be lack of the protein due to NMD or truncated, nonfunctional polypeptides (KCNK18 c.1107del, TRPA1 c.1177C>T and c.1954C>T), the substitution in conserved functional areas of the protein may affect proper functioning. Three variants (TRPV1 c.1348A>G and c.1735C>T and TRPV3 c.2006T>C) were found in transmembrane domains. One variant (ANO3 c.2656A>T) was detected in a linker between transmembrane domains, three variants were located in ankyrin repeats (ANK) (TRPA1 c.932C>A and c.980A>G and TRPV1 c.914T>G), three in the N-terminus (TRPM8 c.665A>G and c.1102C>T), four in the C-terminus (ANO3 c.3100G>C, KCNQ3 c.1885G>A and c.1706A>G, TRPA1 c.3136A>G and TRPM8 c.2945C>T) and one in the inositol-phosphate binding site of TRPA1 (c.3136A>G).
Each ICG variant was detected in one patient, with the exception of TRPA1 c.3136A>G, which was present in two siblings affected by SFN ( Table 1). Most of the reported ICG variants were novel (n = 16/19, 84%). The TRPA1 c.1954C>T variant has been seen before in a German painful-diabetic-peripheral-neuropathy patient [22]. The majority of the identified (n = 14/19, 73.7%) variants were located in TRP genes. No potential causing variant was identified in eight genes from our gene panel: ANO1, HCN1, KCNA2, KCNA4, KCNQ5, KCNN1, KCNS1 and TRPV4. None of the patients with an identified VGSC variant had an additional variant in one of the other ICGs ( Figure 1).
During the clinical assessment, multiple questionnaires, including the SFN Symptom Inventory Questionnaire (SFN-SIQ) [23] and the Visual Analogue Scale, were given to the patients (Table 3). All analyzed patients that completed the questionnaire reported severe pain; however, individuals carrying a potentially pathogenic ICG variant (n = 7) had higher maximal pain scores VAS = 9.36 ± 0.72 compared to those in the group without such an ICG/VGSC variant (n = 204, VAS = 7.47 ± 2.37 (Table 3)). Each individual with ICG reported maximal pain above the mean maximal pain calculated for patients without an ICG/VGSC variant. The patient with an ANO3 variant reported maximal pain (VAS = 9.7), the patient with KCNQ3 c.1885G>A variant reported VAS = 8.7, and the patient with KCNQ3 c.1706A>G reported VAS = 8.5. Interestingly, three out of four TRPA1-positive patients evaluated maximal pain as VAS = 10, which is the highest possible value in the scale. Only one patient carrying the TRPA1 variant, specifically c.2065A>G, reported VAS = 8.6. Therefore, the maximal pain scores in the group of patients having variants in the same gene were higher than those in the subgroup without an ICG/VGSC variant (Table 3).  ICG, ion channel gene; VGSC, voltage-gated sodium ion channel; SD, standard variation. Pain evaluated using the visual analog scale (VAS), where 0 represents "no pain" and 10 is "the worst pain that you can imagine". Patients with incomplete data not included in the table.
Several features, including SFN-specific symptoms like a changed sweating pattern, diarrhea, constipation, micturition problems, dry eyes, dry mouth, orthostatic dizziness, palpitations, hot flashes, sensitive leg skin, burning feet, sheet intolerance legs and restless legs were evaluated for the SFN patients (Tables 4 and 5). All patients with an ICG variant for which these data were available (n = 6) reported often or always burning feet and the restless legs symptom. Next to that, the most commonly mentioned complaint among patients with ICG variants was sheet intolerance on legs, which presented often or always (n = 5/6, 83.3%) and sometimes (n = 1/6, 16.7%) ( Table 6). However, burning feet, sheet intolerance and restless legs were also the most commonly mentioned features in patients without an ICG/VGSC variant. Burning feet presented always and often (n = 213, 79.1%), restless legs presented always and often (n = 156, 57.4%) and sheet intolerance presented always and often (n = 126, 46.3% (Table 7)). Statistical significance was not reached for any of investigated features.

Summary of ICG Screening in Patients with Small Fiber Neuropathy
In 414 patients with no (potentially) pathogenic variants in the SCN3A, SCN7A-SCN11A and SCN1B-SCN4B genes, twenty patients (4.8%) had one potentially pathogenic heterozygous variant in ICG. The detected variants were located in seven genes and were present in two patients (0.5%) in ANO3, one (0.2%) in KCNK18, two (0.5%) in KCNQ3, seven (1.7%) in TRPA1, 3 (0.7%) in TRPM8, three (0.7%) in TRPV1 and two (0.5%) in TRPV3. Most of the detected variants were novel; however, three variants have been reported before. The ANO3 c.3100G>C variant is present in the Human Gene Mutation Database (HGMD) and described as likely disease-causing with questionable pathogenicity in primary torsion dystonia [24]. Two KCNQ3 variants have been reported in ClinVar; c.1885G>A was reported two times, once as a VUS in benign familial neonatal seizures and as once as likely benign without linking to a specific phenotype, while the c.1706A>G variant was reported as a VUS, and the phenotype was not provided. Our patients did not have complaints typical for dystonia and familial neonatal seizures.
In our study population, each ICG variant was detected only once, with exception of the TRPA1 c.3136A>G variant present in a brother and sister with SFN, which was considered as one independent finding. None of the identified ICG variants have been reported before in neuropathic pain, except for the TRPA1 c.1954C>T variant, which we published recently, present in a 73-year-old male diagnosed with painful DN [22].
The frequency of potentially pathogenic ICG variants in our population (4.8%) was slightly lower than the frequency in patients with painful DN (5.4%) [22]. The most frequent variants in the SFN cohort were located in TRP genes, in total in fifteen patients (3.6%), which is consistent with data obtained for painful DN, where seven individuals (3.3%) had identified VUS TRP variants, three (1.4%) in TRPA1, three (1.4%) in TRPM8, one (0.5%) in TRPV4. Different heterozygous missense VUS were also present in three (1.4%) painful DN patients in ANO3 and two different KCNK18 variants leading to a premature stop codon in two (0.9%) individuals with painful DN [22]. Therefore, our data suggest that although the same genes may be involved, different variants underlay SFN and painful DN. No potentially pathogenic variants were detected in eight genes: ANO1, HCN1, KCNA2, KCNA4, KCNQ5, KCNN1, KCNS1 and TRPV4. Five of them, KCNA2, KCNA4, KCNQ5, KCNN1 and KCNS1, were also negative for screening in a diabetic neuropathy population [22]. Interestingly in an SFN cohort, variants have been detected in three genes: KCNQ3, TRPV1 and TPRV3, which did not appear in painful DN [22]. Therefore, although it seems that ICG variants can play a role in DN and SFN, it cannot be concluded due to the limited cohort sizes that specific ICG or ICG genes are responsible for either clinical manifestation. The list of genes, however, is a promising candidate for future research.

VUS Meaning in Context of Protein Function
All detected variants were classified as VUS, which reflects the limitations of bioinformatics analysis of novel variants in absence of functional data or segregation in the family [25,26]. In this study, we only relied on in silico prediction of pathogenicity, therefore, it is difficult to draw a definite conclusion about variant causality, which demands further validation. However, it is likely that a number of these VUS will change to (possibly) pathogenic, as they clearly have a strong impact on the protein. The nonsense/frameshift/splice site variants cause altered protein length or protein absence due to nonsense-mediated decay (NMD). Moreover, substitutions are present in conserved, functionally relevant regions of the protein, most likely affecting protein function, like channel opening and heat sensitivity (variants in the transmembrane domain) [27], channel gating and desensitization (variants in the linker between transmembrane domains) [28], protein-protein interactions (variation in ANK) [29,30], protein stability and folding (variants in N-and C-terminus) [31] and binding sites, such as the inositholphospathe binding site in the case of TRPA1 c.3136A>G, leading to channel inactivation [32].

Anoctamin 3
Anoctamin 3 is a member of the Ca 2+ -activated chloride channel family involved in pain processing and thermoregulation [33]. ANO3 indirectly inhibits pain signaling via enhancing sodium-activated potassium (SLACK) channels in DRG [21]. Since ANO3 knockout results in a decreased pain threshold in tested rats, GOF variants are expected in pain-related phenotypes [21]. Consistent with that, two heterozygous missense variants with predicted GOF were detected in individuals affected with painful DN [22]. In our study cohort of SFN patients, two heterozygous missense variants (ANO3 c.2656A>T and c.3100G>C) have been identified in two unrelated subjects. Both of them reported burning pain, allodynia and sheet intolerance. TTT was abnormal in both cases; moreover, a female patient identified cold temperature as a pain-provoking factor. Altogether, it may link the patients' clinical manifestation with disrupted thermoregulation.

Potassium Channels
Potassium channels are important in regulation of nociceptor excitability in sensory neurons [34]. A heterozygous deletion of one nt in KCNK18 c.1107del leading to a frameshift and a shortened C-terminus has been identified in an SFN patient. Although missense and frameshift mutations with a premature stop codon in KCNK18 are linked mainly to migraine, some studies indicate that downregulation of KCNK18 contributes to neuropathic pain [35,36]. Two patients from our cohort harbored two different heterozygous missense variants in KCNQ3 c.1885G>A and c.1706A>G. Interestingly, both of them had a similar phenotype with numb sensation and burning pain in the feet. In both cases, pain complaints were getting worse at rest. Additionally, both patients had disturbed warm and cold sensation in their feet. However, one patient reported that cold increased and warmth reduced pain symptoms, while the other patient did not indicate a temperature influence on pain. GOF mutations in KCNQ3 are recognized as contributors to pain resilience [37], therefore, a loss-of-function (LOF) effect of detected variants would be expected in our patients with SFN and neuropathic pain.

TRP Channels
Variants located in the TRP genes constitute the largest group of variants identified in our study cohort. TRP channels constitute a large, divergent gene family with a welldocumented role in mediating nociceptive behaviors in response to thermal, chemical and osmotic stimuli [38]. In total, six variants were detected in TRPA1 (two nonsense and four missense), three missense variants in TRPM8, three missense in TRPV1 and two variants in TRPV3 (one missense and one splice variant). TRPA1 channel activity has been functionally linked to pain hypersensitivity, cold hyperalgesia and mechanical nociception in inflammatory and neuropathic pain models [39][40][41]. The GOF mutation N855S in TRPA1 has been associated with familial episodic pain syndrome. The TRPA1 R919* variant is segregating in a family with the cram-fasciculation syndrome [42,43]. Interestingly, the R919* GOF effect was suggested based on clinical improvement with carbamazepine; however, it was not demonstrated whether the truncated TRPA1 protein was expressed at all [43]. In our study, one patient heterozygous for the missense VUS c.980A>G had complaints of numb sensation and painful cramps in the fingers, lower legs and feet, worsening after exercise. Several VUS leading to premature stop codons or nonsense-mediated RNA-decay variants have been reported in painful DN [22], and with the presence of two nonsense variants in our patient cohort (c.1177C>T and c.1954C>T in painful DN), a LOF mechanism would in our opinion be more likely. The clinical manifestation in the patients with TRPA1 VUS was diverse in our cohort; however, the most common complaints were numb sensation, tingling and burning pain. Four out of seven patients reported pain increase after exercise, and all the patients with performed TTT had abnormal TTT values.
Three different heterozygous VUS have been detected in patients with disturbed thermal sensation (abnormal TTT or QST) in TRPM8. Additionally, the patient with c.665A>G VUS reported increased pain due to warm temperature. As TRPM8 is a modulator of cold sensation and a potential target for neuropathic pain treatment [38], those variants remain interesting candidates for functional validation and determine a GOF and LOF effect.
TRPV1 and TRPV3 are heat-activated proteins involved in thermosensation and pain perception [38]. In addition to that, TRPV3 activation plays a role in enhanced itch sensation [44]. In our study, two patents had a TRPV3 variant (c.20006T>C combined with itch episodes and a splice-site variant c.1242+1G>A r.spl?), and three patients were heterozygous for a missense variant in TRPV1 (c.914T>G or c.1348A>G or c.1735C>T). Two of them reported increased pain complaints during rest, which might interfere with night rest. Interestingly, no pain change was reported in response of warm/cold temperature. The literature indicates that upregulation of TRPV1 increases amplification of pain signals [38]. Consistently with that, the GOF Q85R mutation of TRPV1 has been linked to neuropathic pain [45]. On the other hand, TRPV1 missense variants localized in transmembrane domains have been reported in diabetic neuropathy patients without pain [22], which highlights the need of functional variant validation before drawing a definite conclusion.

Patients with an ICG Variant Had More Severe Clinical Manifestations and Higher Pain Scores
Due to the small number of patients with an ICG variant, it is difficult to draw a definite conclusion whether these patients had more severe clinical manifestations in relation to basic clinical features or SFN-SIQ when compared to those of patients without an ICG/VGSC variant. It seems that patients with an ICG variant tended to have more frequent abnormal skin biopsies and a shorter duration of neuropathy; however, statistical significance was not reached for any of these features. There is evidence implicating enhanced Na + influx produced by gain-of-function Nav1.7 mutations as a trigger for calcium-importing reverse Na + /Ca 2+ exchange that can contribute to axonal degeneration [46]; further work will be needed to determine whether ionic imbalances due to any of the variants found in this study are injurious to axons. Nevertheless, patients with an ICG variant reported higher maximal pain VAS = 9.36 ± 0.72 than patients without an ICG/VGSC variant VAS = 7.47 ± 2.37 did. All of the patients with an ICG variant had severe mean maximal pain VAS ≥ 8.5, always higher than the mean calculated for patients without an ICG/VGSC variant. Interestingly, patients with TRPA1 reported very severe pain (mean max pain VAS = 9.65 ± 0.7), and three of them gave VAS = 10, which is the highest possible score. These data are consistent with a previous study showing that mean maximal pain was higher for painful DN patients with an ICG variant vs. painful DN without an ICG variant [22]. MUMC+ is a tertiary referral center for SFN in the Netherlands, and FINCB is a national referral center for several neurological conditions, including SFN. Medical history and clinical data were collected and recorded as described previously [12]. The diagnosis of SFN was confirmed based on clinical symptoms of SFN combined with reduced intraepidermal nerve fiber density (IEND) in skin biopsy without signs of large fiber involvement and/or abnormal temperature threshold testing (TTT) and/or nerve conduction study (NCS) [47]. The assessment of neuropathic pain was performed using the visual analog scale (VAS), where the scores are recorded by marking a value on a 10 cm line in which 0 represents "no pain" and 10 is "the worst pain that you can imagine", and the 11-point Numerical Rating Scale (PI-NRS) ranging from 0 to 10, where 0 means "no pain" and 10 refers to "the worst pain imaginable" [48,49]. Only patients with diagnosed SFN and neuropathic pain with VAS > 3 were included. In this study, we focused on patients negative for SCN3A, SCN7A-SCN11A and SCN1B-SCN4B variant screening. Most of the included patients (n = 348, 69%) were Dutch, while 156 (31%) patients were from Italy. The mean age of recruited SFN patients was 54.1 years (SD 14.2 years), with the mean age of complaints onset of 42.2 years (SD 13.6 years). More females than males were present in the study population. There were 59.5% (n = 300) of females and 40.5% (n = 204) of males. In both populations, females were over-represented, with 59.8% (n = 208) females in the Dutch cohort and 59% (n = 92) females in the Italian cohort, respectively. Twenty-three percent of the patients (n = 65) reported positive cases of neuropathy in the family, while negative family history was registered for 77% (n = 217). The data for familial cases of neuropathy was incomplete for 222 patients. More than one-third of the recruited patients had an abnormal skin biopsy (n = 120, 36.5%).

DNA Extraction and Storage
Genomic DNA was from blood samples using a NucleoSpin8 Blood Isolation kit (Macherey-Nagel, Düren, Germany) or a QIAamp DNA Blood Maxi Kit, Puregene ® Blood Core Kit (Qiagen, Hilden, Germany). DNA extraction was performed according to the manufacturers' instructions and stored in −20 • C.

smMIPs-NGS of Peripheral Ion Channels and Data Analysis
The smMIP-NGS of exonic regions with ±20 bps exon-flanking sequences of the 15 ion channel genes (Table 8) was performed according to a previously established protocol [50]. The detailed description of the technique, smMIPs characteristics, number of probes per targeted coding region per gene and smMIPs-NGS pipeline used for data analysis have been described before [22].

Variant Classification
Exonic and ±20 bp exon-flanking intron variants with >20x coverage and an alternative variant call of >30% were included for the analysis. The SNPs with high frequency, defined as dbSNP > 5%, ExAC > 5% and >1% in our Dutch in-house database of 12,244 exomes, were excluded from the analysis. All variants were examined individually, including BAM file visualization and variant interpretation performed in Alamut Visual software (Interactive Biosoftware, Rouen, France). Variants were classified according to the practice guidelines of the Association for Clinical Genetic Science (ACGS) [51].

Statistical Analysis
The analysis of continuous variables was performed using the independent Student's t-test. Categorical variables were analyzed using the chi-square test or Fisher's exact test in the case of small counts <5. Statistical significance of <0.05 was applied.

Conclusions
In a well-characterized cohort of patients with SFN, we identified twenty patients (4.8%) carrying a potentially pathogenic variant in ICG. Patients with an ICG variant might have more severe pain than patients without a (potentially) pathogenic ICG/VGSC variant. This complements previous results of the analysis of the same ICG in a large cohort of painful and painless DN. Moreover, it extends the list of potential pain-related genes, reporting variants in KCNQ3, TRPV1 and TPRV3 that were not present in painful DN. Apparently, ICGs are fundamental for both SFN and DN, although the numbers are too small to unambiguously identify gene-or variant-specific manifestations. Although all of the variants were classified as VUS and the consequences of the variation cannot be certainly predicted in silico, they are still promising risk factors for neuropathic pain as well as promising gene candidates for future research.  Institutional Review Board Statement: All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Local Medical Ethical Committee; Comitato Etico Regione Lombardia Sezione Fondazione IRCCS Instituto Neurologico "Carlo Besta" (n. 56, 07 November 2018) and METC azM/UM (METC 17-4-035.1/ivb, 11 May 2017). Patient information was anonymized prior to data analysis.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Conflicts of Interest: Professor
Stephen Waxman received SGW report grants from the European Union's Horizon 2020 research and innovation program Marie Sklodowska-Curie grant for PAIN-Net, Molecule-to-Man Pain Network (grant no. 721841) and European Union 7th Framework Programme (grant no. 602273) for the PROPANE study, was supported in part by the Rehabilitation Research and Development Service and Biomedical Laboratory Research Service, Department of Veterans Affairs, has served as a consultant to ChromoCell Corp. and Sangamo Therapeutics, and has served on the Scientific Advisory Board of OliPass Corporation, MedTronics and Navega Therapeutics, outside the submitted work.