Visceral and somatic pain modalities reveal Na_V1.7‐independent visceral nociceptive pathways

Abstract

Key points

• Voltage‐gated sodium channels play a fundamental role in determining neuronal excitability.

• Specifically, voltage‐gated sodium channel subtype Na_V1.7 is required for sensing acute and inflammatory somatic pain in mice and humans but its significance in pain originating from the viscera is unknown.

• Using comparative behavioural models evoking somatic and visceral pain pathways, we identify the requirement for Na_V1.7 in regulating somatic (noxious heat pain threshold) but not in visceral pain signalling.

• These results enable us to better understand the mechanisms underlying the transduction of noxious stimuli from the viscera, suggest that the investigation of pain pathways should be undertaken in a modality‐specific manner and help to direct drug discovery efforts towards novel visceral analgesics.

Abstract

Voltage‐gated sodium channel Na_V1.7 is required for acute and inflammatory pain in mice and humans but its significance for visceral pain is unknown. Here we examine the role of Na_V1.7 in visceral pain processing and the development of referred hyperalgesia using a conditional nociceptor‐specific Na_V1.7 knockout mouse (Na_V1.7^Nav1.8) and selective small‐molecule Na_V1.7 antagonist PF‐5198007. Na_V1.7^Nav1.8 mice showed normal nociceptive behaviours in response to intracolonic application of either capsaicin or mustard oil, stimuli known to evoke sustained nociceptor activity and sensitization following tissue damage, respectively. Normal responses following induction of cystitis by cyclophosphamide were also observed in both Na_V1.7^Nav1.8 and littermate controls. Loss, or blockade, of Na_V1.7 did not affect afferent responses to noxious mechanical and chemical stimuli in nerve–gut preparations in mouse, or following antagonism of Na_V1.7 in resected human appendix stimulated by noxious distending pressures. However, expression analysis of voltage‐gated sodium channel α subunits revealed Na_V1.7 mRNA transcripts in nearly all retrogradely labelled colonic neurons, suggesting redundancy in function. By contrast, using comparative somatic behavioural models we identify that genetic deletion of Na_V1.7 (in Na_V1.8‐expressing neurons) regulates noxious heat pain threshold and that this can be recapitulated by the selective Na_V1.7 antagonist PF‐5198007. Our data demonstrate that Na_V1.7 (in Na_V1.8‐expressing neurons) contributes to defined pain pathways in a modality‐dependent manner, modulating somatic noxious heat pain, but is not required for visceral pain processing, and advocate that pharmacological block of Na_V1.7 alone in the viscera may be insufficient in targeting chronic visceral pain.

Key points

• Voltage‐gated sodium channels play a fundamental role in determining neuronal excitability.

• Specifically, voltage‐gated sodium channel subtype Na_V1.7 is required for sensing acute and inflammatory somatic pain in mice and humans but its significance in pain originating from the viscera is unknown.

• Using comparative behavioural models evoking somatic and visceral pain pathways, we identify the requirement for Na_V1.7 in regulating somatic (noxious heat pain threshold) but not in visceral pain signalling.

• These results enable us to better understand the mechanisms underlying the transduction of noxious stimuli from the viscera, suggest that the investigation of pain pathways should be undertaken in a modality‐specific manner and help to direct drug discovery efforts towards novel visceral analgesics.

Abbreviations

BSA

bovine serum albumin

CIP

congenital insensitivity to pain

CT

quantification cycles

DRG

dorsal root ganglia

FB

Fast Blue

GAPDH

glyceraldehyde‐3‐phosphate dehydrogenase

IC/BPS

interstitial cystitis/bladder pain syndrome

LS

lumbosacral

Na_V

voltage‐gated sodium channel

PEPD

paroxysmal extreme pain disorder

p.o.

per os

QST

quantitative standardized testing

TL

thoracolumbar

TRPV1

transient receptor potential cation channel V1

TTX‐R

tetrodotoxin‐resistant

TTX‐S

tetrodotoxin‐sensitive

Introduction

Chronic pain originating from internal organs affects significant proportions of the population, with analgesics restricted by dose‐limiting side‐effects. Persistent pain and visceral hypersensitivity manifest as reduced thresholds for mechanical distension of visceral organs and are strongly associated with inflammation. The targeting of peripheral sensory input, either by peripheral nerve block (Cherry et al. 1985; Brown, 1989; Eisenberg et al. 1995) or local anaesthetics (Verne et al. 2003, 2005) has proven effective in treating visceral pain. However, our understanding of key sensory afferent transduction mechanisms responsible for visceral nociception is limited. Here, we investigate voltage‐gated sodium channel Na_V1.7 in both visceral and somatic pain behaviours and show that peripheral pain pathways of the viscera are functionally distinct from classical nociceptors, providing evidence supporting functional diversity of nociception and confirmation that novel analgesic development must be applied in a mechanism‐specific manner.

Rare human genetic conditions link Na_V1.7 to pain perception, with loss‐of‐function mutations causing congenital insensitivity to pain (CIP) (Cox et al. 2006; Goldberg et al. 2007). Recapitulation of the human painless phenotype using knockout mice genetically engineered to globally lack Na_V1.7 results in complete loss of responses to acute, inflammatory and neuropathic pain (Gingras et al. 2014). Using tissue‐specific Na_V1.7 knockout mice (including nociceptor‐specific Na_V1.7^Nav1.8 mice (Nassar et al. 2004), pan‐sensory neuron Na_V1.7^Advill mice (Minett et al. 2012) and pan‐sensory and sympathetic neuron Na_V1.7^Wnt1 mice (Minett et al. 2012)) modality‐specific pain pathways associated with acute heat and mechanical detection, hyperalgesia and allodynia have been linked with differing Na_V repertoires.

Intriguingly, CIP patients feel no visceral pain, with reports of both painless childbirth and rupture of appendix (Melzack & Wall, 1988; Zimmermann et al. 1988; Wheeler et al. 2015), suggesting that Na_V1.7 may be required for visceral nociception. Rectal pain is a symptom of paroxysmal extreme pain disorder (PEPD), another condition associated with rare Na_V1.7 mutations (Fertleman et al. 2006), with defecation capable of triggering pain attacks implicating a link to anorectal distension. In patients with interstitial cystitis/bladder pain syndrome (IC/BPS), pain perception associates with Na_V1.7 mutations (Reeder et al. 2013). Like other chronic pain conditions, a hallmark of IC/BPS is ongoing pain in the absence of obvious pathophysiology (Dimitrakov & Guthrie, 2009). Therefore, Na_V1.7 could be involved in maintaining spontaneous pain, such as peripheral or central sensitization, in addition to evoked pain attributed to mechanical stimulation. Surprisingly, whilst broad‐spectrum sodium channel blockers are effective in treating chronic visceral pain, selective Na_V1.7 antagonists (ProTx‐II) and monoclonal blocking antibodies targeting Na_V1.7 have been unable to fully recapitulate loss of Na_V1.7 mutant phenotypes to other chronic pain models (Schmalhofer et al. 2008; Lee et al. 2014). Indeed, selective antagonism of Na_V1.7 with ProTx‐II also failed to block afferent responses to stretch of the colorectum (Feng et al. 2015), suggesting the contribution of Na_V1.7 to visceral pain processing is still unclear.

In light of recent findings that Na_V1.7 is essential for some (acute heat and mechanical pain, inflammatory hyperalgesia and neuropathic allodynia), but not all (acute cold pain, cancer‐induced bone pain and oxaliplatin‐evoked allodynia) pain modalities, we investigated visceral pain and referred hyperalgesia using a conditional nociceptor‐specific Na_V1.7 knockout mouse (Na_V1.7^Nav1.8) and selective Na_V1.7 antagonist PF‐5198007. Thus, using comparative behavioural models evoking somatic and visceral pain pathways we identify specific mechanisms regulating noxious heat pain threshold and show that Na_V1.7 in Na_V1.8‐expressing neurons is not required for visceral pain signalling.

Methods

Experiments were performed in adult mice weighing 20–35 g. Conditional nociceptor‐specific Na_V1.7 knockout mice (Na_V1.7^Nav1.8) and their littermate controls were generated as described previously (Nassar et al. 2004). Observers performing behavioural and ex vivo electrophysiological experiments were blind to the genotypes of the animals. Animals were acclimatized for at least 1 week before behavioural testing in temperature and light‐controlled (12 h light–dark cycle) rooms. All experiments were performed in accordance with the UK Animals (Scientific Procedures) Act of 1986 or with the EU Directive 2010/63/EU for animal experimentation, with approval of the University of Granada Research Ethics Committee (Granada, Spain). Human tissues were collected and utilized with approval of the East London and The City HA Local Research Ethics Committee (London, UK; NREC 10/H0703/71) in accordance with the Declaration of Helsinki and following full written informed consent.

Behavioural experiments

Experiments were performed on both male and female knockout and wild‐type littermate control mice. Visceral pain and referred hyperalgesia was assessed using previously described methods, with small modifications (Olivar & Laird, 1999; Laird et al. 2001; Gonzalez‐Cano et al. 2013). Briefly, mice were acclimatized for 40 min to test chambers (consisting of a transparent box on an elevated wire mesh floor) after which 50 μl of capsaicin (0.1 or 1%), mustard oil (0.01 or 0.1%) or vehicle was instilled intrarectally via a thin cannula inserted into the anus and the animal returned to the chamber. The number of spontaneous pain behaviours (including licking of abdomen, stretching of abdomen and abdominal retractions) were recorded for the subsequent 20 min. In a separate set of experiments, visceral pain behaviours caused by cyclophosphamide‐induced cystitis were examined following a previously described protocol (Olivar & Laird, 1999). Again after a 40 min habituation, animals were removed from the test chamber and cyclophosphamide (100 or 200 mg kg⁻¹) or vehicle injected intraperitoneally. The animals were returned to the chamber and pain behaviours recorded according to the following scale: 0 = normal, 1 = piloerection, 2 = strong piloerection, 3 = laboured breathing, 4 = licking of the abdomen and 5 = stretching and contraction of the abdomen. If more than one of these behaviours was noted during a single observation period, then only the type and not quantity of each different pain behaviour was scored (i.e. if two stretching and contractions (5 points) and one abdominal licking (4 points) was observed, then a score of 9 was assigned).

After the evaluation of spontaneous pain behaviours (primary behavioural endpoint), the presence of referred hyperalgesia was determined by measuring the withdrawal response to a punctate mechanical stimulation (von Frey hair filaments 0.02–2 g (0.19–19.6 mN), Touch‐Test Sensory Evaluators, North Coast Medical Inc., USA) of the abdomen using the up–down paradigm 20 min after algogen administration (Chaplan et al. 1994). Avoiding the perianal and external genitalia, the mid‐range 0.4 g von Frey hair filament was applied (three times for 2–3 s at 5 s intervals) to the lower and mid‐abdomen. If a positive response (consisting of immediate licking/scratching of the application site, sharp retraction of the abdomen or jumping) was observed, then probing was repeated in consecutive tests with a weaker von Frey filament. By contrast if there was no response to probing then a stronger von Frey filament was used. Once the withdrawal threshold (secondary behavioural endpoint) was ascertained, mice were humanely killed by concussion of the brain and cervical dislocation of the neck.

Electrophysiological recordings of visceral afferent activity

Nerves innervating murine and human gastrointestinal tissues were isolated and electrophysiological activity recorded using previously described methods (Peiris et al. 2011; Hockley et al. 2014). Mice were humanely killed by concussion of the brain and cervical dislocation of the neck. The distal colon with associated lumbar splanchnic nerves was removed and transferred to a recording chamber superfused (7 ml min⁻¹; 32–34°C) with carbogenated Krebs buffer (in mm: 124 NaCl, 4.8 KCl, 1.3 NaH₂PO₄, 2.5 CaCl₂, 1.2 MgSO₄.7H₂O, 11.1 glucose, and 25 NaHCO₃) supplemented with nifedipine (10 μm), atropine (10 μm) and indomethacin (3 μm). The same supplemented Krebs buffer was used to luminally perfuse (100 μl min⁻¹) the colon after cannulation.

To translate murine experimental recordings into human tissue, we recorded extrinsic nerve activity from resected human appendices. We have previously shown that the appendix represents a valid human ex vivo model of visceral afferent activity amenable to the testing of mechanical and chemical stimuli (Peiris et al. 2011). Specifically, the extrinsic nerves of the appendix are a branch of those innervating the right colon along the ileocolic artery and represent a readily available tissue in normal non‐inflamed (e.g. from colon cancer resections) states. Resected appendices were obtained from five patients undergoing elective surgery at Barts Health NHS Trust, London after full written consent was attained. Appendices were removed from patients undergoing right hemicolectomies as part of their normal surgical treatment for bowel cancer or slow transit constipation (see Table 1 for details) with the permission of the histopathologist and were returned to the morbid anatomy department after completion of the studies. Once removed, appendix specimens were immediately placed in cold Krebs buffer and handled in a comparable manner to mouse distal colon tissues. Removal of the tip and cannulation enabled intraluminal perfusion, in addition to superfusion with Krebs buffer (7 ml min⁻¹; 32–34°C) supplemented with 10 μm nifedipine and 10 μm atropine. Under a dissection microscope, mesenteric neurovascular bundles were blunt dissected and associated nerves identified and cleared of connective tissue. Using borosilicate glass suction electrodes, multi‐unit activity from whole lumbar splanchnic nerves (rostral to the inferior mesenteric ganglia) of mouse, or from mesenteric nerves of human bowel tissues, was recorded. Signals were amplified and bandpass filtered (gain 5 K; 100–1300 Hz; Neurolog, Digitimer Ltd, UK) and digitized at 20 kHz (micro1401; Cambridge Electronic Design, UK) before display on a PC using Spike 2 software. The signal was digitally filtered for 50 Hz noise (Humbug, Quest Scientific, Canada) and a threshold of twice the background noise (typically 100 μV) was used to determine action potential firing counts.

Table 1.

Patients details from which resected appendix specimens were used

Patient No.	Disease	Operation	Tissue	Age	Sex
1	Cancer	Right hemicolectomy	Appendix	83	F
2	Cancer	Right hemicolectomy	Appendix	42	F
3	Cancer	Right hemicolectomy	Appendix	72	F
4	Slow transit constipation	Subtotal colectomy	Appendix	69	M
5	Cancer	Right hemicolectomy	Appendix	70	M

Mean age of patients, 67 years; M:F ratio, 1:1.5. Appendix specimens from 5 patients were collected and used in electrophysiological nerve recordings.

Electrophysiological protocols

Following a stabilizing period of 30 min, noxious intraluminal distending pressures were applied by blocking the luminal perfusion out‐flow of the cannulated mouse distal colon or resected human appendix. The noxious pressures reached evoke pain behaviours in vivo and are above threshold for all known visceral afferent mechanoreceptors (Ness & Gebhart, 1988; Hughes et al. 2009). In murine experiments, a combined sequential protocol was used to initially assess multiple aspects of visceral afferent mechanosensitivity and chemosensitivity. Specifically, a set of 6 rapid phasic distensions (0–80 mmHg, 60 s at 9 min intervals) followed by slow ramp distension (0–145 mmHg, ∼5–6 min) were implemented prior to bath superfusion of separate 20 ml volumes of 1 μm bradykinin and 1 mm ATP at 40 min intervals. In separate experiments, the effect of pharmacological inhibition of Na_V1.7 on visceral afferent sensitivity to mechanical distension or noxious stimulation by capsaicin, mustard oil or bradykinin was tested. A set of 9 rapid phasic distensions (0–80 mmHg, 60 s at 9 min intervals) followed by a 30 min stabilization period and bath superfusion of 1 μm bradykinin in a 20 ml volume were performed. Prior to the 7th phasic distension, bath superfusion of the selective Na_V1.7 antagonist PF‐5198007 (100 nm; 500 ml; Alexandrou et al. 2016) or vehicle (0.1 % DMSO) was initiated and maintained for the duration of the remaining 3 distensions and bradykinin application. In some experiments, after a wash‐out period, repeat phasic distensions were performed during which 250 ml tetrodotoxin (TTX; 100 nm) was superfused. In separate experiments, a ramp distension (0–145 mmHg) was performed followed by bath superfusion of capsaicin (500 nm) and mustard oil (250 μm) at a 1 h interval. Five minutes prior to application of capsaicin, either 100 nm PF‐5198007 or vehicle (0.1% DMSO) was applied for the duration of the subsequent stimulations. Human appendix specimens were stimulated in a comparable manner by repeat ramp distension (0–60 mmHg, ∼30 s at 9 min intervals). Baseline responses were established for 3 distensions prior to the superfusion of PF‐5198007 (100 nm or 1 μm) for 50 min during subsequent distensions.

Retrograde labelling of gut‐specific sensory neurons and single‐cell qRT‐PCR

Distal colon‐specific sensory neurons were retrogradely labelled, picked and the expression of mRNA transcripts of interest determined by qRT‐PCR. A mid‐line 1.5 cm laparotomy was performed on male mice after induction of anaesthesia with 1.5% isoflurane. Multiple injections of Fast Blue (FB: 0.2 μl per site, 2% in saline, Polysciences Gmbh, Germany) were made using a fine pulled‐glass needle and microinfusion pump (0.4 μl min⁻¹) into the wall of the distal colon. Prior to suturing of the peritoneal muscle layer and securing the skin with Michel clips, the abdominal cavity was flushed with saline to remove any excess FB. Post‐operative care (monitoring body weight and soft diet) and analgesia (buprenorphine 0.05–0.1 mg kg⁻¹ daily) was provided for the duration of the protocol. Three to five days after surgery, mice were humanely killed by concussion of the brain and cervical dislocation of the neck, and thoracolumbar (TL: T10–L1) and lumbosacral (LS: L5–S2) dorsal root ganglia (DRG) were harvested and cultured separately for gene expression experiments. Dissected ganglia were incubated at 37°C (in 5% CO₂) in Lebovitz L‐15 Glutamax (GIBCO, UK) media containing 1 mg ml⁻¹ collagenase type 1A (Sigma) and 6 mg ml⁻¹ bovine serum albumin (BSA; Sigma, UK) for 15 min, followed by L‐15 media containing 1 mg ml⁻¹ trypsin (Sigma, UK) and 6mg ml⁻¹ BSA for 30 min. Ganglia were gently triturated and collected by brief centrifugation at 500 g. The supernatant (containing dissociated cells) was collected and the cycle of gentle trituration and centrifugation repeated. Cells from TL and LS DRG were plated separately onto poly‐d‐lysine‐coated coverslips (BD Biosciences, UK) and incubated in Lebovitz L‐15 Glutamax media containing 2 % penicillin–streptomycin, 24 mm NaHCO₃, 38 mm glucose and 10% fetal bovine serum. Fast Blue‐positive colonic sensory neurons were individually harvested from cultures of retrogradely labelled DRG (either TL: T10–L1 or LS: L5–S2) by pulled glass pipette. By breaking the pipette tip (containing the cell) into a tube containing preamplification mastermix (2.5 μl 0.2× primer–probe mix, 5 μl CellDirect 2× reaction buffer (Invitrogen), 0.1 μl SUPERase‐in (Ambion, TX, USA), 1.2 μl TE buffer (Applichem, Germany) and 0.2 μl Superscript III Reverse Transcriptase–Platinum Taq mix (Invitrogen)) and freezing immediately, mRNA transcripts were preserved. Only those individual Fast Blue‐positive neurons free from debris and other non‐neuronal cells (e.g. satellite glia) were collected. An image of each harvested neuron was also captured using a camera (DCC1545M, ThorLabs Inc, NJ, USA) attached to the inverted microscope enabling an estimation of cell size to be ascertained. In the absence of cells, samples of the bath solution were collected for no‐template control experiments. Using the following thermal cycling protocol, preamplification of cDNA was achieved: 50°C for 30 min, 95°C for 2 min, then 21 cycles of (95°C for 15 s, 60°C for 4 min). After dilution (1:5 TE buffer), Taqman qPCR assays were run for each gene of interest (Taqman Assay ID: Na_V1.1, Mm00450580_m1; Na_V1.2, Mm01270359_m1; Na_V1.3, Mm00658167_m1; Na_V1.4, Mm00500103_m1; Na_V1.5, Mm01342518_m1; Na_V1.6, Mm00488110_m1; Na_V1.7, Mm00450762_s1; Na_V1.8, Mm00501467_m1; Na_V1.9, Mm00449367_m1; GAPDH, Mm99999915_g1; Applied Biosystems) using the following cycling protocol: 50°C for 2 min, 95°C for 10 min, then 40 cycles of (95°C for 15 s, 60°C for 1 min). Glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) acted as an internal positive control, with all single‐cell RT‐PCR products expressing GAPDH and bath control samples negative for all Taqman reactions. Relative expression of Na_Vs was normalized to GAPDH quantification cycles (CT) using 2^−ΔCT formula. Quantitative assessment of gene expression was determined by quantification cycle values less than the threshold of 35 being considered as positive.

Ramping hotplate pain behaviours

Behavioural phenotyping experiments were performed using both male and female mice, and pharmacology experiments were carried out in male mice. Acute heat pain was assessed using a ramping hotplate comparable to that used in human standardized quantitative testing (QST) protocols (Rolke et al. 2006). Mice were acclimatized in a chamber for 6 min daily for the 3 days preceding dosing. After which, following a 30 s acclimatization, the chamber floor was slowly heated from 31°C at a rate of 3.4°C min⁻¹ and the temperature and time taken until observing a pain behaviour was recorded (behavioural endpoint; the occurrence of either licking or shaking of the hindpaw and/or rapid shifting of weight (stomping) from one foot to the other). After baseline measurements were made, mice were dosed via oral administration (p.o.) with either vehicle or PF‐5198007 at 1 or 3 mg ml⁻¹ with a dose volume of 10 ml kg⁻¹ and 1 h later, the ramping hotplate repeated. Mice were humanely killed by concussion of the brain and cervical dislocation of the neck immediately after final assessment of thermal pain threshold.

Skin–nerve preparation

Multi‐unit extracellular afferent recordings were made from the tibial nerve innervating the glaborous skin of the hindpaw as previously described (Milenkovic et al. 2008) but with some modifications. Briefly, mice were humanely killed by concussion of the brain and cervical dislocation of the neck, the hindlimbs were then shaved, removed and the tibial nerve and associated glaborous skin dissected free. The preparation was mounted glaborous skin downwards in a recording chamber superfused (10 ml min⁻¹; 36 ± 1°C) with carbogenated (95% O₂, 5% CO₂) Krebs buffer (in mm: 107 NaCl, 3.48 KCl, 26.2 NaHCO₃, 0.69 MgSO₄.7H₂O, 1.67 NaH₂PO₄, 9.64 sodium gluconate, 7.6 sucrose, 5.5 glucose, 1.53 CaCl). The epiperineurium was removed from the distal end of the tibial nerve and suction electrode recordings comparable to those of visceral afferent activity were made. Following a 60 min stabilization period, a heat stimulus (Krebs perfused onto the skin at a focal point equivalent to the heel portion of the paw) lasting 50 s was applied, this increased in temperature from 36°C to 52°C at a rate of 0.4°C s⁻¹ to mimic the noxious heat ramp used in vivo. In total, a series of 10 heat stimulations were performed at 15 min intervals. The first 4 heat stimulation formed the baseline reading, with bath superfusion of PF‐5198007 (30 nm) or vehicle (0.1% DMSO) initiated and maintained for the duration of the next 2 stimulations (30 min), PF‐5198007 (100 nm) or vehicle (0.1% DMSO) for the following 2 heat stimulations (30 min) and heat stimulations 9 and 10 carried out during the superfusion (15 min) of TTX (100 nm or 300 nm) and lidocaine (1 mm), respectively. In separate experiments, the effect of genotype and selective sodium channel antagonists were assessed in response to a cold stimulus (36 to 6°C at a rate of 0.4°C s⁻¹) delivered in the same manner as the heat stimulus, with comparable stimulation and protocols as above.

Data analysis

Pain behaviours and mechanical thresholds were compared across experimental groups with two‐way analysis of variance (ANOVA) followed by the Bonferroni post hoc test, using either SigmaPlot 12.0 (Systat Software Inc., CA, USA) or Prism 6 (GraphPad Inc., USA). Referred hyperalgesia, expressed as the mechanical threshold producing 50% of responses, was calculated using: 50% mechanical threshold (g) = [(10(X _f + κδ))/10], where X _f = value (in logarithmic units) of the final von Frey filament used; κ = tabular value for the pattern of positive/negative responses; and δ = mean difference (in log units) between stimuli (Dixon, 1980). Peak changes or total sum firing of electrophysiological nerve activity in multi‐unit experiments were determined by subtracting baseline firing (2 min before distension or drug application) from increases in nerve activity following distension or noxious chemical superfusion. Estimation of cell size from single‐cell images was achieved by averaging the height and width of each cell (ImageJ 1.49V analysis software, NIH, USA). Total sum firing of electrophysiological nerve activity in response to each hot or cold stimulation was obtained by subtracting any signal evoked by heat/cold stimulation in the presence of lidocaine (1 mm). Expression data were visualized using R and the ggplot2 graphics package (Wickham, 2009). Statistical significance was set at P < 0.05. Data are displayed as means ± SEM.

Drugs

Stock concentrations of capsaicin (1%; 10% ethanol, 10% tween, 80% saline), mustard oil (1%; 70% ethanol, 30% saline), cyclophosphamide (saline), bradykinin (10 mm; water), lidocaine (1 m; water) and ATP (300 mm; water) were purchased from Sigma‐Aldrich and prepared as described. Tetrodotoxin (15 μg ml⁻¹ stock) was purchased from Nanning Leaf Pharmaceuticals (Canada) and diluted in saline. PF‐5198007 was manufactured in‐house by Pfizer and solubilized in DMSO at a 10 mm stock. For in vitro experiments, PF‐5198007 was applied at a concentration of 100 nm (ensuring almost 100% inhibition of mouse Na_V1.7 (IC₅₀ 5.2 nm) and selectivity over Na_V1.1 and Na_V1.6 (IC₅₀ 149 nm and 174 nm, respectively; Alexandrou et al. 2016). For in vivo studies PF‐5198007, 1 mg ml⁻¹ or 3 mg ml⁻¹, was suspended in 0.5% methylcellulose + 0.1% Tween‐80 in distilled water. Doses of PF‐5198007 were selected to achieve a free plasma concentration of ∼100 nm (littermate: 1 mg kg⁻¹, 58 ± 10 nm, N = 5; 3 mg kg⁻¹, 842 ± 91 nm, N = 10; Na_V1.7^Nav1.8: 1 mg kg⁻¹, 68 ± 12 nm, N = 5; 3 mg kg⁻¹, 634 ± 69 nm, N = 9). Vehicle was dosed as a 10 ml kg⁻¹ solution of 0.5% methylcellulose + 0.1% Tween‐80 in distilled water. All other compounds were diluted in appropriate experimental buffer to working concentrations on the day of experimentation, unless otherwise stated.

Results

Visceral pain behaviours in response to colorectal sensitizing noxious stimuli were unaffected by deletion of Na_V1.7

We used a conditional Na_V1.7 knockout mouse strain, where floxed (SCN9A) Na_V1.7 mice were crossed with mice in which Cre expression is driven by the Na_V1.8 promotor (Na_V1.7^Nav1.8) resulting in tissue‐specific ablation of Na_V1.7 in sensory neurons expressing nociceptive markers (Nassar et al. 2004; Shields et al. 2012). Capsaicin acts at TRPV1 and will activate the vast majority of visceral afferent terminals (>85%; Christianson et al. 2006; Malin et al. 2009) leading to neurogenic inflammation and prolonged ongoing afferent activity due to sensitization (Laird et al. 2001, 2002). Intracolonic instillation of capsaicin in littermate control mice led to dose‐dependent increases in observed pain behaviours consisting of abdominal contractions and licking (Fig. 1 A). The deletion of Na_V1.7 from Na_V1.8‐positive neurons, however, did not attenuate pain behaviours at either dose of capsaicin tested (P = 0.72, N = 6–8, two‐way ANOVA). In separate experiments, the potent algogen mustard oil was instilled intracolonically leading to the activation and sensitization of afferents and induction of localized tissue damage as previously described (Laird et al. 2002). Substantial pain behaviours were observed in both Na_V1.7^Nav1.8 and littermate controls (Fig. 1 B), which were not significantly different in terms of the magnitude of their response (P = 0.79, N = 6–8, two‐way ANOVA). The time course of pain behaviours induced by capsaicin and mustard oil did not differ between littermate controls and Na_V1.7^Nav1.8 mice. These findings show that Na_V1.7 expressed in Na_V1.8‐positive neurons is not required for the development of visceral pain or for sustained spontaneous nociceptor activity as a result of sensitization.

Figure 1. Spontaneous visceral‐pain related behaviours in Na_V1.7^Nav1.8 and littermate mice following intracolonic administration of capsaicin (A and C) or mustard oil (B and D).

A and B, number of acute pain related behaviours (licking of abdomen, stretching, abdominal retractions) induced by capsaicin (A) or mustard oil (B) during a 20 min period. C and D, referred mechanical hyperalgesia (evaluated by stimulation of the abdomen with von Frey filaments) was measured 20 min after the administration of capsaicin (C) or mustard oil (D). Mean ± SEM of values obtained in 6–10 animals. ^* P < 0.05, ^** P < 0.01 vs. vehicle.

Referred hyperalgesia is a common characteristic of visceral pain, with the sensitization of somatic structures in the same metameric field to the affected organ driven in part by spinal convergence of somatic and visceral afferents inputs (Cervero, 1983; Mertz et al. 1995). Whilst primary inflammatory hyperalgesia has been shown to be dependent on Na_V1.7 in Na_V1.8‐expressing neurons, whether Na_V1.7 contributes to the development of secondary hyperalgesia remains unstudied. The development of mechanical sensitivity of the abdomen in response to intracolonic instillation of either capsaicin (0.1%) or mustard oil (0.01%) was independent of ablation of Na_V1.7 from Na_V1.8‐expressing neurons, with 50% withdrawal thresholds significantly reduced 20 min after treatment irrespective of genotype (capsaicin; P < 0.01, N = 6–8, two‐way ANOVA; mustard oil, P < 0.01, N = 6–8, two‐way ANOVA).

Pain responses to cyclophosphamide‐induced cystitis are unaffected by deletion of Na_V1.7

To model bladder pain/cystitis in Na_V1.7^Nav1.8 mice, cyclophosphamide was administered, leading to the progressive development of visceral pain behaviours for the duration of the 4 h observation window. Cyclophosphamide treatment produces mucosal erosion and haemorrhage of the bladder in addition to oedema (Fraiser et al. 1991). The development and time course of pain behaviours observed did not differ between littermate controls and Na_V1.7^Nav1.8 mice to either dose of cyclophosphamide tested (Fig. 2 A, P = 0.93, N = 6–8, two‐way ANOVA). Indeed both Na_V1.7^Nav1.8 mice and littermate controls also showed marked referred hyperalgesia when tested 4 h after cyclophosphamide treatment (Fig. 2 B). The referred hyperalgesia did not differ dependent on genotype, suggesting that persistent activation of nociceptors by a developing noxious chemical stimuli is not driven by a requirement for Na_V1.7 to be present.

Figure 2. Visceral pain related behaviours evoked by cyclophosphamide‐induced cystitis in Na_V1.7^Nav1.8 and littermate mice.

A, behavioural pain responses were recorded at 30 min intervals during the 240 min observation period after cyclophosphamide injection. B, referred mechanical hyperalgesia was evaluated by stimulation of the abdomen with von Frey filament 4 h after cyclophosphamide administration. Mean ± SEM of values obtained in 6–10 animals. ^* P < 0.05, ^** P < 0.01 vs. vehicle.

Visceral afferent mechanosensitivity is blocked by TTX but is unaffected by deletion of Na_V1.7 or blockade with a selective small‐molecule Na_V1.7 antagonist

In order to distinguish between the multiple roles that Na_V1.7 makes to nociceptive processing, we investigated the contribution of Na_V1.7 to mechanosensitivity and chemosensitivity at the peripheral terminals of sensory neurons innervating the gastrointestinal tract. To do this multi‐unit ex vivo extracellular electrophysiological recordings of lumbar splanchnic nerve activity were made from the distal colon of mice. Tissues were dissected free and cannulated to enable mechanical and chemical stimuli to be applied by luminal distension or bath superfusion. Phasic distension of the colon to noxious pressures (0–80 mmHg) was used to model mechanical stimulation of the bowel and evoke increased afferent firing for the duration (60 s) of the distension. Consistent with previous reports, adaptation in the response to repeat stimulation (at 9 min intervals) was observed during subsequent distensions, with the response stabilizing by the fourth to sixth distension (see Fig. 3 A and C) (Hockley et al. 2014). In Na_V1.7^Nav1.8 mice, there was no significant difference in either the initial peak distension response or in the degree of tachyphylaxis observed during repeat distensions compared to littermate controls (Fig. 3 C, P = 0.62, N = 13–14, two‐way repeated‐measures (RM) ANOVA). Previous studies have suggested that not only the magnitude, but also the dynamic quality, of the distension protocol used may be important for delineating gut motor events, specifically noxious stimuli (Sengupta & Gebhart, 1994; Booth et al. 2008). Given the proposed role of Na_V1.7 as a threshold channel contributing to the amplification of depolarizing stimuli in sensory neurons (Dib‐Hajj et al. 2013), we used a slow ramp distension protocol to supramaximal distension pressures (0–145 mmHg) in order to investigate the impact of loss of Na_V1.7 on responses across a range of innocuous and noxious distending pressures. In littermate controls, afferent firing increased proportionally to intraluminal pressure, with a peak firing rate of 37.5 ± 5.7 spikes s⁻¹ at 145 mmHg. Significantly less firing was observed in Na_V1.7^Nav1.8 mice to equivalent distending pressures (at 145 mmHg, 25.7 ± 4.2 spikes s⁻¹; P < 0.0001, N = 19, two‐way ANOVA). However, firing rates in Na_V1.7^Nav1.8 mice to ramp distension were unchanged within the physiologically‐relevant 0–80 mmHg range compared to controls (P > 0.05, Bonferroni's post hoc analysis). Within the supramaximal range (80–145 mmHg), there was a reduction in firing, suggesting Na_V1.7 may be involved in transducing non‐physiological extremes of pressure in the colon but not innocuous or even noxious mechanical stimuli.

Figure 3. Visceral afferent responses to noxious distension of the distal colon in Na_V1.7^Nav1.8 mice and following small‐molecule Na_V1.7 antagonism.

Example rate histogram of colonic splanchnic nerve activity and intraluminal pressure trace to repeat phasic distension (0–80 mmHg; 60 s; 9 min intervals) in Na_V1.7^Nav1.8 (B) and littermate (A) mice. C, peak change in firing rate during phasic distensions in both genotypes (P = 0.46, two‐way repeated‐measures ANOVA). D, average firing rates to ramp distension (0–145 mmHg) at 5 mmHg increments in littermate and Na_V1.7^Nav1.8 mice. E, effect of 100 nm PF‐5198007, vehicle (0.1% DMSO) or 100 nm TTX on total firing evoked during repeat 0–80 mmHg phasic distensions in littermate and Na_V1.7^Nav1.8 mice.

Given that Na_V1.7 is ablated only in Na_V1.8‐positive neurons, it is possible that visceral afferents that are both sensitive to noxious mechanical stimuli and are negative for Na_V1.8 may be contributory to the responses observed. In order to test this hypothesis, in a further set of experiments, repeat phasic distensions were continued and the effect of the selective small‐molecule Na_V1.7 antagonist PF‐5198007 (100 nm) was assessed on responses in both Na_V1.7^Nav1.8 and littermate control mice. Responses in littermate control mice to repeat phasic distensions were unchanged following pre‐incubation with, and in the presence of, 100 nm PF‐5198007 compared to vehicle (Fig. 3 E, P = 0.86, N = 7, two‐way RM ANOVA). Further, the afferent response following application of 100 nm PF‐5198007 in Na_V1.7^Nav1.8 mice also did not significantly differ from that observed in wild‐type animals (P = 0.87, N = 6–7, two‐way RM ANOVA). However, irrespective of genotype, application of 100 nm TTX to preparations did fully block afferent firing in response to noxious phasic distension (Fig. 3 E). Together this shows that mechanosensitivity in visceral afferents is dependent on TTX‐sensitive voltage‐gated sodium channels but not Na_V1.7.

Loss, or antagonism, of Na_V1.7 does not alter visceral afferent responses to acute inflammatory and algogenic mediators

To investigate the involvement of Na_V1.7 in modulating visceral afferent sensitivity to inflammatory and algogenic mediators used in our in vivo studies, capsaicin and mustard oil were applied to distal colon preparations and visceral afferent responses recorded from both littermate and Na_V1.7^Nav1.8 mice, and in the presence or absence of 100 nm PF‐5198007. In separate experiments, bradykinin and ATP, as inflammatory mediators typically present during injury or infection, and which may be evoked by mustard oil/cyclophosphamide treatment contributing to ongoing nociceptor sensitization, were also tested.

Responses to application of 500 nm capsaicin did not differ between Na_V1.7^Nav1.8 mice and littermate mice in vehicle control experiments (0.1% DMSO; Na_V1.7^Nav1.8 vs. littermate; P = 0.50, N = 6 both groups, Student's unpaired t test, Fig. 4 A). In addition, superfusion of 100 nm PF‐5198007 during, and 5 min prior to, capsaicin (500 nm) application did not significantly change the evoked afferent discharge in either genotype (Na_V1.7^Nav1.8: 100 nm PF‐5198007 vs. 0.1% DMSO, P = 0.82, N = 6, unpaired t test; littermate: 100 nm PF‐5198007 vs. 0.1% DMSO, P = 0.59, N = 6, unpaired t test, Fig. 4 A). Afferent firing evoked by mustard oil was also unchanged in both Na_V1.7^Nav1.8 mice and littermate controls (0.1% DMSO: Na_V1.7^Nav1.8 vs. littermate, P = 0.46, N = 6, unpaired t test, Fig. 4 B), irrespective of the presence of Na_V1.7 antagonist (Na_V1.7^Nav1.8: 100 nm PF‐5198007 vs. 0.1% DMSO, P = 0.44, N = 6, unpaired t test; littermate: 100 nm PF‐5198007 vs. 0.1% DMSO, P = 0.93, N = 6, unpaired t test, Fig. 4 B).

Figure 4. Effect of capsaicin and mustard oil on visceral afferent responses.

Change in peak firing rate to application of 500 nm capsaicin (A) and 250 μm mustard oil (B) in littermate and Na_V1.7^Nav1.8 mice, both in the absence and presence of 100 nm PF‐5198007.

Bath superfusion of 1 mm ATP in littermate mice resulted in significant afferent discharge, with a peak change in firing of 1.39 ± 0.50 spikes s⁻¹. In Na_V1.7^Nav1.8 mice, the response was comparable to littermate controls (2.33 ± 0.80 spikes s⁻¹, P = 0.32, N = 7–8, unpaired t test). Responses to application of 1 μm bradykinin were greater than that observed for ATP, but did not differ dependent on genotype (littermate, 9.11 ± 3.32 vs. Na_V1.7^Nav1.8, 8.56 ± 3.04 spikes s⁻¹, P = 0.90, N = 7–8, unpaired t test). Further, in distal colon preparations from littermate controls pre‐incubated with 100 nm PF‐5198007, peak firing response to 1 μm bradykinin was unchanged (vehicle (0.1% DMSO) 5.16 ± 2.00 vs. 100 nm PF‐5198007 4.31 ± 0.63 spikes s⁻¹, P = 0.70, N = 7, unpaired t test); this was also true of tissues from Na_V1.7^Nav1.8 mice pre‐incubated with the Na_V1.7 antagonist (100 nm PF‐5198007; P = 0.17, N = 6–7, unpaired t test). Collectively, these data suggest that Na_V1.7 within the peripheral terminal of colonic sensory neurons is not required in order to transduce both noxious mechanical and chemical algogenic stimuli, in agreement with behavioural experiments.

Localization of Na_V expression in colonic sensory neurons

We next investigated the expression of voltage‐gated sodium channel α subunits present in colonic sensory neurons. Specifically, using single‐cell qRT‐PCR we examined the expression of mRNA transcripts for Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.4, Na_V1.5, Na_V1.6, Na_V1.7, Na_V1.8 and Na_V1.9 in gut‐projecting sensory neurons. Both lumbar splanchnic and pelvic innervation have been shown to contribute to the transmission of noxious stimuli from the distal colon (Brierley et al. 2004). As such, the expression of these channels was determined in colonic sensory neurons in dorsal root ganglia (DRG) T10 to L1 levels (thoracolumbar: TL) that are known to possess the greatest number of sensory neurons projecting via the lumbar splanchnic nerve, and separately in DRG L5 to S2 levels (lumbosacral: LS); the afferents from which have been shown to project predominantly via the pelvic nerve. Of the 30 cells collected per mouse (N = 3), the average size of colonic sensory neurons harvested was 32.0 ± 0.2 μm for TL (N = 3) and 30.7 ± 1.0 μm for LS (N = 3). In the Na_V1.7^Nav1.8 mice used in the studies described here, Na_V1.7 was selectively ablated from all Na_V1.8‐positive sensory neurons. To confirm the proportion of colonic sensory neurons affected by this gene ablation, the expression of Na_V1.7 and Na_V1.8 was first examined. Na_V1.7 was present in 100% of thoracolumbar and 95.6 ± 2.22% of lumbosacral colonic sensory neurons. High expression of Na_V1.8 was also observed in both thoracolumbar (95.6 ± 2.22 %) and lumbosacral (91.1 ± 4.44 %) colonic sensory neuron populations. Importantly, significant co‐expression of both these sodium channels in individual colonic sensory neurons was found, with 95.4% of Na_V1.7‐positive neurons also expressing Na_V1.8, suggesting that the vast majority of colonic sensory neurons in Na_V1.7^Nav1.8 mice would be affected by the genetic deletion.

The expression of the remaining tetrodotoxin‐sensitive (TTX‐S: Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.4 and Na_V1.6) and TTX‐resistant voltage‐gated sodium channels (TTX‐R: Na_V1.5 and Na_V1.9) was also determined (Catterall et al. 2005). Of the TTX‐S sodium channels, Na_V1.6 was present in the greatest frequency (86.7%; Fig. 5 A) of thoracolumbar colonic sensory neurons after Na_V1.7. Significant proportions of thoracolumbar colonic sensory neurons also expressed either Na_V1.1 (44.4 ± 5.88 %), Na_V1.2 (68.9 ± 8.89 %) or Na_V1.3 (53.3 ± 10.2 %), although co‐expression was not always observed (see Fig. 5 C). As expected, both the skeletal myocyte voltage‐gated sodium channel Na_V1.4 and the cardiac myocyte Na_V1.5 channel were expressed by low proportions of thoracolumbar colonic sensory neurons (6.67 ± 6.67 % and 17.8 ± 5.88 %, respectively). In agreement with previous studies, mRNA transcripts for TTX‐R Na_V1.9 were observed in 84.4 ± 44.4 % of thoracolumbar neurons (Hockley et al. 2016). By comparison, the expression of Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.4, Na_V1.7 and Na_V1.8 did not significantly differ between populations of lumbosacral compared to thoracolumbar colonic sensory neurons (Fig. 5 A, all P > 0.05, TL vs. LS, unpaired t test). Interestingly, significant differences were observed between the frequency of expression of Na_V1.5 (TL vs. LS, P < 0.05, unpaired t test) and Na_V1.6 (TL vs. LS, P < 0.01, unpaired t test) in lumbosacral compared to thoracolumbar colonic sensory neurons. Indeed, transcripts for both Na_V1.5 and Na_V1.6 were observed in approximately half of lumbosacral colonic sensory neurons (48.9 ± 8.01 % and 51.1 ± 5.88 %, respectively). The expression of Na_V1.9 (which has been shown previously to contribute to afferent sensitivity of the lumbar splanchnic nerve; Hockley et al. 2014) in lumbosacral colonic sensory neurons were consistent with the frequency of expression observed in the thoracolumbar populations (P = 0.42, N = 3, unpaired t test). Taken together, these data not only support the expression of Na_V1.7 by a majority of colonic sensory neurons innervating the distal colon, but also highlight an as‐yet‐unexplored complexity in the molecular patterning of voltage‐gated sodium channels present in these neurons.

Figure 5. Expression of voltage‐gated sodium channel mRNA transcripts in mouse colonic sensory neurons by single‐cell qRT‐PCR.

A, proportions of thoracolumbar and lumbosacral colonic sensory neurons expressing transcripts for Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.4, Na_V1.5, Na_V1.6, Na_V1.7, Na_V1.8 and Na_V1.9. B, relative expression of Na_V transcripts in thoracolumbar and lumbosacral colonic sensory neurons. C, co‐expression analysis of voltage‐gated sodium channels in both thoracolumbar (left) and lumbosacral (right) colonic sensory neuronal populations. Each segment in the wheel‐diagrams is representative of a single cell, with a coloured segment signifying positive expression. [Color figure can be viewed at wileyonlinelibrary.com]

Deletion of Na_V1.7 impairs somatic noxious thermal thresholds, which can be recapitulated by Na_V1.7 antagonism

Given that no differences in acute visceral pain or referred hyperalgesia could be observed in mice lacking Na_V1.7 in Na_V1.8‐positive neurons or to block of Na_V1.7 by the selective inhibitor PF‐5198007, we next sought to investigate the role of Na_V1.7 in somatic acute pain behaviours. In order to investigate the contribution of Na_V1.7 in Na_V1.8‐positive sensory neurons to the modulation of thermal thresholds, we utilized a ramping hotplate behavioural paradigm. In littermate controls, this latency was 261 ± 5 s (N = 38) corresponding to a temperature rise of ∼13.6°C (baseline floor temperature 31°C ramping to 44.6 ± 0.2°C; Fig. 6 A). This increase in temperature required to evoke a behavioural response was equivalent to a previous study using a modified ramping Hargreaves’ test (Minett et al. 2014 a). Na_V1.7^Nav1.8 mice showed an attenuated response to ramping hotplate with an augmented latency (274 ± 5 s) and significantly increased thermal threshold (46.1 ± 0.3°C, N = 36, P < 0.0001, unpaired t test; Fig. 6 A) in agreement with previous observations (Minett et al. 2014 a). The attenuation of complex behaviours associated with the ramping hotplate test suggests involvement of Na_V1.7 in pain signalling to noxious thermal stimulation of the skin under certain conditions.

Figure 6. Somatic pain behaviours and tibial nerve activity to noxious thermal stimulation in Na_V1.7^Nav1.8 and littermate mice.

A, thermal pain thresholds in Na_V1.7^Nav1.8 mice are significantly increased following ramping hotplate behavioural testing. B, average thermal pain thresholds following the application of selective Na_V1.7 antagonist PF‐5198007 (1 or 3 mg kg⁻¹) or vehicle in Na_V1.7^Nav1.8 and littermate mice. C and D, example raw traces, rate histogram and temperature recordings of tibial nerve activity in littermate (C) and Na_V1.7^Nav1.8 mice (D). E, sum firing of tibial nerve activity during focal heat stimulation in skin–nerve preparations of Na_V1.7^Nav1.8 and littermate mice in the presence of TTX (100 nm) or vehicle (0.1% distilled H₂O). ^#### P < 0.0001, Na_V1.7^Nav1.8 baseline vs. littermate baseline. F, effect of PF‐5198007 on evoked tibial nerve firing by heat stimulation in Na_V1.7^Nav1.8 and littermate mice. G, sum firing of tibial nerve activity during focal cold stimulation in skin–nerve preparations of Na_V1.7^Nav1.8 and littermate mice in the presence of TTX (100 nm) or vehicle (0.1% distilled H₂O). H, effect of PF‐5198007 on evoked tibial nerve firing by cold stimulation in Na_V1.7^Nav1.8 and littermate mice. ^* P < 0.05, ^** P < 0.01, ^**** P < 0.0001.

Using the ramping hotplate, we went on to confirm the ability for the selective Na_V1.7 inhibitor PF‐5198007 to modulate thermal pain behaviours (see Fig. 6 B). In littermate mice, application of PF‐5198007 (1 mg kg⁻¹ p.o.) significantly increased the thermal threshold for observing pain behaviours with a concomitant increase in the latency to response when compared to vehicle controls (P < 0.01, N = 10, two‐way ANOVA with Bonferroni's post hoc test vs. vehicle; Fig. 6 B). In both vehicle and PF‐5198007 treatment groups, the thermal threshold of Na_V1.7^Nav1.8 mice remained significantly greater than littermate controls but did not differ between groups. Application of a higher dose of PF‐5198007 (3 mg kg⁻¹ p.o.) also led to an increase in thermal threshold during hotplate ramp, which was comparable to thresholds observed in Na_V1.7^Nav1.8 mice and significantly different from vehicle groups (P < 0.05, N = 10, two‐way ANOVA with Bonferroni's post hoc test vs. vehicle). These data suggest that whilst pain behaviours can be evoked in the absence, or antagonism, of Na_V1.7, the expression of Na_V1.7 in sensory neurons modulates heat pain thresholds for noxious thermal stimuli.

Na_V1.7 also contributes to cutaneous afferent firing in response to both noxious hot, but not cold, thermal stimuli

To investigate whether Na_V1.7 was necessary for sensory transduction at the peripheral terminal of somatic afferents, ex vivo multi‐unit electrophysiological recordings of the tibial nerve from skin–nerve preparations of Na_V1.7^Nav1.8 mice and littermate controls were made (Fig. 6 C and D). In support of hotplate experiments, a ramping thermal stimuli (focal water jet from 36°C to 52°C (at ∼0.4°C s⁻¹)) was applied to the corium side of the skin and the evoked nerve activity recorded. Total firing during the heat‐evoked stimuli was significantly attenuated in Na_V1.7^Nav1.8 mice compared to littermate controls (Fig. 6 E, P < 0.0001, N = 26–29, two‐way ANOVA with Bonferroni's post hoc test). Bath superfusion of 100 nm TTX led to significant inhibition of firing regardless of genotype compared to vehicle controls (Fig. 6 E, P < 0.05, N = 9–11 and P < 0.0001, N = 10–11, Na_V1.7^Nav1.8 and littermate controls, respectively, two‐way ANOVA with Bonferroni's post hoc test), suggesting that the transduction of noxious thermal stimuli at the peripheral terminal of sensory afferents is enhanced by the presence of Na_V1.7 in Na_V1.8‐positive neurons, but is dependent on other TTX‐S Na_Vs that might be present. Application of 100 nm PF‐5198007 in littermate controls was able to recapitulate the attenuated response observed in Na_V1.7^Nav1.8 mice (Fig. 6 F, P < 0.05, N = 9–10, two‐way ANOVA with Bonferroni's post hoc test vs. vehicle (0.1% DMSO)). In addition, PF‐5198007 in Na_V1.7^Nav1.8 mice further reduced afferent responses to heat ramp, suggesting that afferent firing at the peripheral terminal is dependent predominantly on expression of Na_V1.7 in Na_V1.8‐positive sensory neurons. However, this does not discount contributions of Na_V1.7 to other sensory populations spinally or supra‐spinally involved in the nociceptive processing of thermal stimuli.

In addition, we investigated cutaneous afferent firing in response to evoked cold stimuli by localized perfusion of a cooling perfusate over the receptive field from 36°C to ∼6°C (at ∼0.4°C s⁻¹). In previous studies, Na_V1.7 has been shown to be involved in acetone‐induced cooling, but not noxious cold sensation (Minett et al. 2012). Responses evoked by cold stimulation of the skin did not differ between Na_V1.7^Nav1.8 mice and littermate controls (Fig. 6 G, P > 0.05, N = 18, two‐way ANOVA with Bonferroni's post hoc test), but application of 100 nm TTX completely abolished cold‐evoked responses compared to vehicle (P < 0.01, N = 6 and P < 0.0001, N = 5–6, littermate and Na_V1.7^Nav1.8 mice, respectively, two‐way ANOVA with Bonferroni's post hoc test). Finally, incubation with the selective Na_V1.7 antagonist PF‐5198007 (100 nm) did not significantly attenuate cold evoked afferent firing (Fig. 6 H), supporting the posit that Na_V1.7 does not contribute to the transduction or amplification of cold‐evoked depolarizations at the peripheral terminal.

Mesenteric nerve responses to phasic distension in human appendix are unaffected by inhibition of Na_V1.7

Finally, in order to understand whether our findings in murine visceral afferents translate to human we used ex vivo extracellular recordings of surgically resected appendices to investigate Na_V1.7 function in response to mechanical stimuli. The human appendix has been used previously as a pre‐clinical model of visceral nociception (Peiris et al. 2011). The appendix was cannulated and stimulated by repeat noxious ramp distension (0–60 mmHg) and mesenteric nerve firing recorded. Ramp distension evoked a concomitant increase in human visceral afferent firing, with a peak change in firing of 10.1 ± 1.5 spikes s⁻¹ (N = 5), with reproducible responses observed to subsequent distensions. Application of PF‐5198007 did not significantly impair visceral afferent firing to ramp distension at either low or high distending pressures (Fig. 7 B, P = 0.26, N = 5, two‐way RM ANOVA). This confirms our mouse data highlighting that Na_V1.7 appears not to significantly impact visceral afferent sensitivity to acute mechanosensation. As such, Na_V1.7 imparts functionality on sensory neurons in a modality‐specific manner and therefore the analgesic assessment of Na_V1.7 antagonists should be determined in a mechanism‐dependent fashion.

Figure 7. Effect of selective small‐molecule antagonism of Na_V1.7 in resected human appendices following repeat noxious distension.

A, example rate histogram of appendix mesenteric nerve activity and intraluminal pressure trace following repeat ramp distension (0–60 mmHg; 10 min interval). Application of PF‐5198007 was initiated at the start of the black bar and maintained for 50 min during which distensions were continued. B, average firing rates to repeat ramp distension (0–60 mmHg; N = 5) of human appendix prior to, and after, addition of PF‐5198007; neither low‐threshold or high‐threshold afferent firing is affected by antagonism of Na_V1.7. C and D, both change in peak firing rate (C) and total afferent firing (D; area under curve) were unchanged by bath superfusion with PF‐5198007 (N = 5). [Color figure can be viewed at wileyonlinelibrary.com]

Discussion

Nociceptive processing in somatic and visceral pain has common underlying pathways, including convergence in neuroanatomy, overlap in psychological representation and commonality in cellular transductions. However, important differences exist in the manifestation, perception and psychology of these pain modalities. Traditionally, visceral afferents are characterized based on mechanical sensitivity and activation by chemical mediators (including bradykinin and ATP; Su & Gebhart, 1998; Brierley et al. 2004; Grundy, 2004), with functional assessment required to define nociceptive properties. Compared to somatic counterparts, visceral sensory neurons almost exclusively possess characteristics attributed to nociceptors (unmyelinated C‐fibres (Sengupta & Gebhart, 1994), peptidergic (Robinson et al. 2004) and high expression of Na_V1.8/TTX‐R sodium currents (Beyak et al. 2004)), yet collectively transduce innocuous unconscious and conscious sensations in addition to pain. As such, visceral sensory neurons do not fit well with classical views of nociceptors and established schema for nociceptive transduction pathways.

Here, we add to this by showing that visceral pain signalling in vivo in response to acute and sensitizing noxious stimuli is independent of Na_V1.7. We confirm by way of ex vivo electrophysiological recordings of mouse visceral afferent fibres that deletion of, or selective small‐molecule antagonism of Na_V1.7, does not attenuate responses to persistent noxious mechanical (including repeat phasic and sustained ramp distension) and chemical stimuli (including capsaicin, mustard oil, bradykinin and ATP). This lack of efficacy in Na_V1.7 antagonism in blocking visceral afferent activation extends to recordings from resected human appendix tissues when applying noxious distending pressures. Surprisingly, mouse visceral sensory neurons almost always express Na_V1.7, suggesting that, whilst present, Na_V1.7 appears not to contribute to the modulation of afferent excitability in response to depolarizing stimuli, or the propagation of action potentials. Furthermore, the lack of phenotype observed in Na_V1.7^Nav1.8 mice suggests Na_V1.7 is not necessary for transducing noxious visceral input centrally by Na_V1.8‐expressing neurons. By contrast somatically, deletion of Na_V1.7 does modulate acute heat pain thresholds, which can be replicated using selective Na_V1.7 antagonism. Strikingly, loss of Na_V1.7 from Na_V1.8‐expressing neurons, or small‐molecule antagonism, is able to attenuate afferent firing evoked by ramping heat stimuli applied to skin–nerve preparations. This implicates Na_V1.7 in modulating thermal transduction sensitivity in somatic afferents. This was not true of cold stimuli, where Na_V1.7 does not have a role in afferent responses. Our data demonstrate that whilst Na_V1.7 does modulate defined somatic pain pathways, it is not required for those visceral pain modalities investigated here, and advocate that selective pharmacological block of Na_V1.7 in the viscera may prove ineffective in targeting chronic visceral pain caused by spontaneous nociceptor activity, sensitizing inflammatory mediators or evoked mechanical distension: principal clinical drivers of visceral pain.

Voltage‐gated sodium channels are vital for the transmission of painful stimuli in primary afferents. Importantly, the relative significance of individual sodium channels is dependent on the pain modality considered, with Na_V1.7 essential in transducing somatic acute thermal and mechanical pain, in conjunction with inflammatory hyperalgesia and neuropathic allodynia (Minett et al. 2014 b). Similarly, Na_V1.8 is critical for extreme cold pain (Abrahamsen et al. 2008), with chemotherapy‐induced allodynia dependent on Na_V1.6 (Sittl et al. 2012; Deuis et al. 2013). Normal visceral nociceptor activity, by contrast, is dependent on both Na_V1.8 (Laird et al. 2002) and Na_V1.9 (Hockley et al. 2014). Surprisingly, the role of Na_V1.7 in visceral pain processing is poorly understood in spite of human genetic data linking Na_V1.7 to pain signalling.

Substantive evidence for the involvement of Na_V1.7 in visceral pain processing comes from human genetic studies. Patients with congenital insensitivity to pain linked to mutations in Na_V1.7 do not feel pain, including pain originating from internal structures (broken bones; Cox et al. 2006; Goldberg et al. 2007) and hollow organs (e.g. during appendicitis or child‐birth; Melzack & Wall, 1988; Zimmermann et al. 1988). Mutations in SCN9A gene encoding Na_V1.7 are also causal in paroxysmal extreme pain disorder (PEPD) where severe burning pain may occur in rectal, ocular and mandibular regions. Intriguingly, defecation and micturition can both trigger such rectal pain attacks (Fertleman et al. 2006; Meglic et al. 2014), implicating hypersensitivity of visceral mechanoreceptors in initiating pain attacks. Whilst Na_V1.7 is linked with multiple aspects of the pain pathway, this is the first report detailing the contribution of Na_V1.7 to visceral pain processing. Using single‐cell qRT‐PCR of gut‐specific sensory neurons we show that mRNA transcripts for Na_V1.7 are expressed by the vast majority of colonic sensory neurons, consistent with Na_V1.7 immunoreactivity in extrinsic afferent terminals of the distal colorectum (Feng et al. 2015). Co‐expression of Na_V1.7 in Na_V1.8‐positive neurons was substantial in gut‐projecting populations, suggesting that nearly all visceral sensory neurons would be affected by Na_V1.8‐specific knockout of Na_V1.7 (Nassar et al. 2004). However, it is possible that some Na_V1.7‐positive Na_V1.8‐negative colonic neurones remain, which may be sufficient to maintain pain behaviours.

Visceral afferent firing to mechanical and chemical activation were unaffected following loss of, or antagonism of, Na_V1.7, but could be blocked by TTX as shown previously (Campaniello et al. 2016). As such, TTX‐S Na_Vs other than Na_V1.7 are involved in transducing noxious visceral stimuli. Established roles for TTX‐R Na_V1.8 and Na_V1.9 correlate well with their extensive expression shown here; however, little is known about the expression of TTX‐S Na_Vs within a viscerally‐projecting population. Na_V1.6 is essential in pelvic afferent endings for spike initiation and repetitive firing (Feng et al. 2015), a concept that would fit with the extensive presence of Na_V1.6 mRNA transcripts observed here. Further, using toxin antagonists of Na_V1.7 (ProTx‐II) and Na_V1.6 (μ‐conotoxin GIIIa and μ‐conotoxin PIIIa), a requirement on Na_V1.6, but not Na_V1.7, was observed for the encoding of stretch‐sensitive pelvic afferents (Feng et al. 2015). Taken together, these observations present compelling evidence that Na_V1.7 is redundant in visceral afferent nociception in response to spontaneous or evoked noxious stimuli.

Clearly whilst not necessary for normal sensation in the gut, the high relative expression of Na_V1.7 suggests that aberrant Na_V1.7 function, such as that present in some monogenic pain disorders, could significantly impact visceral sensation. Intriguingly, the propensity for mutations in Na_V1.7 to evoke regional pain phenotypes in PEPD patients (i.e. rectal and not ‘true visceral’ pain) could be driven by differences we observe here in the expression of some sodium channels (Na_V1.5 (Renganathan et al. 2002) and Na_V1.6 (Cummins et al. 2005)) located in thoracolumbar, versus lumbosacral, visceral sensory neurons. Precedent for background neuronal phenotype contributing to the manifestation of functional effects already exists, with the same mutation in Na_V1.7 causing hypo‐ and hyper‐excitability when expressed in either sympathetic or sensory neurons (Rush et al. 2006). The extensive expression of Na_V1.7 suggests that mutations subverting its endogenous function may significantly alter phenotype even if not required for that pain modality normally. As such it is possible that non‐canonical roles of Na_V1.7 may help explain the contradiction of how CIP patients associated with loss of Na_V1.7 do not feel visceral pain. For example, recent evidence of Na_V1.7 deletion upregulating endogenous opioid expression suggests a complex transcriptional modulatory, as well as electrogenic, contribution by Na_V1.7, but this did not alter the expression of other Na_V subtypes present in DRG (Minett et al. 2015). Importantly, the use of a selective small‐molecule antagonist of Na_V1.7 enables us to discount developmental differences in gene deletion studies in the phenotypes observed here.

Comparison with somatic pain behaviours enables confirmation of a modality‐specific action for Na_V1.7 expression and confirms the ability of the antagonist PF‐5198007 in replicating gene deletion studies. Na_V1.7 is required for modulating heat pain thresholds after burn injury (Shields et al. 2012) and for acute noxious heat sensing in a population of Na_V1.8‐negative neurons (Minett et al. 2012). Surprisingly, we found, using an adapted ramping hotplate test, that loss of Na_V1.7 from Na_V1.8‐positive neurons could also alter acute heat pain thresholds and this could be recapitulated using PF‐5198007. In all cases, mice remained sensitive to noxious heat, suggesting that Na_V1.7 is not required in Na_V1.8‐expressing neurons but can modulate the thermal threshold sensitivity. Notably, we observed a desensitization of the heat pain threshold from ∼44°C by 2–3°C following antagonism of Na_V1.7, as such fixed temperature hotplate tests typically used to measure withdrawal latencies at 50°C or 55°C would be above threshold in either case, masking potential phenotypic differences. A similar non‐redundant role for Na_V1.7 in Na_V1.8‐expressing neurons was observed in response to an adapted Hargreaves’ test (Minett et al. 2014 a). This further highlights the involvement of multiple sub‐populations of neurons on stimulus‐intensity specific responses underpinning noxious thermal detection.

In summary, using a combination of gene deletion knockout mice and pharmacological tool molecules we demonstrate that Na_V1.7, although expressed extensively by gut‐projecting sensory neurons, contributes minimally to visceral pain pathways associated with algogenic sensitizing chemicals and evoked activation of visceral afferents by noxious stimuli. The patterning of sodium channel expression shown here reveals a previously unstudied molecular complexity to visceral sensory neurons. Combined with a detailed study of somatic thermal sensitivity, we show that assessment of candidate analgesic targets to pain mechanisms must be considered in a modality‐specific manner. As such, Na_V1.7 antagonism of peripheral visceral afferents may not represent a viable therapeutic rationale for the treatment of chronic visceral pain associated with evoked distension or inflammation of the viscera.

Additional information

Competing interests

The authors declare no competing financial interests or conflict of interest.

Author contributions

Study concept and design: J.R.F.H., R.G.C., S.M., M.A.T.G., W.J.W., D.C.B., C.M.C., G.M.; funding and supervision: J.M.B., W.J.W., D.C.B., C.K., C.M.C., G.M.; acquisition and analysis of data: J.R.F.H., R.G.C., S.M., M.A.T.G., C.M., A.T., A.L.W., V.C.G., F.R.N., T.P. J.N.W. provided reagents without which the studies would not have been possible. All authors contributed to the interpretation of data and writing the manuscript. C.M. is funded by the Dr Hadwen Trust and did not participate in experiments involving animals, or cells or tissues from animals or from human embryos. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by University of Granada‐GREIB (C.M.C.), an unrestricted educational grant from Neusentis (V.C.G.) and The Dr Hadwen Trust for Humane Research (C.M.). J.R.F.H., S.M., A.L.W., W.J.W. and G.M. are all employees of Pfizer Ltd.

Author's present address

W. J. Winchester: Takeda Cambridge Ltd, Science Park, Milton Road, Cambridge CB4 0PZ, UK.