The TRPV1 channel in rodents is a major target for antinociceptive effect of the probiotic Lactobacillus reuteri DSM 17938

Abstract

Abstract

Certain bacteria exert visceral antinociceptive activity, but the mechanisms involved are not determined. Lactobacillus reuteri DSM 17938 was examined since it may be antinociceptive in children. Since transient receptor potential vanilloid 1 (TRPV1) channel activity may mediate nociceptive signals, we hypothesized that TRPV1 current is inhibited by DSM. We tested this by examining the effect of DSM on the firing frequency of spinal nerve fibres in murine jejunal mesenteric nerve bundles following serosal application of capsaicin. We also measured the effects of DSM on capsaicin-evoked increase in intracellular Ca²⁺ or ionic current in dorsal root ganglion (DRG) neurons. Furthermore, we tested the in vivo antinociceptive effects of oral DSM on gastric distension in rats. Live DSM reduced the response of capsaicin- and distension-evoked firing of spinal nerve action potentials (238 ± 27.5% vs. 129 ± 17%). DSM also reduced the capsaicin-evoked TRPV1 ionic current in DRG neuronal primary culture from 83 ± 11% to 41 ± 8% of the initial response to capsaicin only. Another lactobacillus (Lactobacillus rhamnosus JB-1) with known visceral anti-nociceptive activity did not have these effects. DSM also inhibited capsaicin-evoked Ca²⁺ increase in DRG neurons; an increase in Ca²⁺ fluorescence intensity ratio of 2.36 ± 0.31 evoked by capsaicin was reduced to 1.25 ± 0.04. DSM releasable products (conditioned medium) mimicked DSM inhibition of capsaicin-evoked excitability. The TRPV1 antagonist 6-iodonordihydrocapsaicin or the use of TRPV1 knock-out mice revealed that TRPV1 channels mediate about 80% of the inhibitory effect of DSM on mesenteric nerve response to high intensity gut distension. Finally, feeding with DSM inhibited perception in rats of painful gastric distension. Our results identify a specific target channel for a probiotic with potential therapeutic properties.

Key points

• Certain probiotic bacteria have been shown to reduce distension-dependent gut pain, but the mechanisms involved remain obscure.

• Live luminal Lactobacillus reuteri (DSM 17938) and its conditioned medium dose dependently reduced jejunal spinal nerve firing evoked by distension or capsaicin, and 80% of this response was blocked by a specific TRPV1 channel antagonist or in TRPV1 knockout mice.

• The specificity of DSM action on TRPV1 was further confirmed by its inhibition of capsaicin-induced intracellular calcium increases in dorsal root ganglion neurons. Another lactobacillus with ability to reduce gut pain did not modify this response.

• Prior feeding of rats with DSM inhibited the bradycardia induced by painful gastric distension.

• These results offer a system for the screening of new and improved candidate bacteria that may be useful as novel therapeutic adjuncts in gut pain.

Introduction

Abdominal pain is a hallmark symptom of both inflammatory bowel diseases (IBD) and most functional bowel disorders (Thompson, 1999; Long & Drossman, 2010). Gut inflammation may trigger many of these pathologies, and there is evidence that even irritable bowel syndrome (IBS) is characterized by low-grade proinflammatory cytokine responses (Dinan et al. 2006). IBS and IBD are associated with an altered gut microbiome (McKendrick, 1996; Manichanh et al. 2012) and there is evidence for a causal relationship (Manichanh et al. 2012). There is also increasing evidence which suggests that IBD may be a consequence of inappropriate inflammatory responses to intestinal microbes in genetically predisposed hosts (Abraham & Cho, 2009).

Ingestion of particular probiotics can alleviate IBS symptoms (O’Mahony et al. 2005; Whorwell et al. 2006; Sinn et al. 2008; Francavilla et al. 2010; Haller et al. 2010) and bacterial mixtures of a few probiotic strains may help to maintain remission or alleviate IBS (Kajander et al. 2005; Sinn et al. 2008) and even IBD symptoms (Haller et al. 2010).

That specific probiotic strains can reduce or inhibit visceral pain induced by gut distension in rodents has been shown by several investigators (Kamiya et al. 2006; Rousseaux et al. 2007; Collins et al. 2009; McKernan et al. 2010; Duncker et al. 2011) and the mechanisms whereby this occurs have begun to be explored (Rousseaux et al. 2007; Wang et al. 2010b; Mao et al. 2013; Perez-Burgos et al. 2013, 2014). A Lactobacillus rhamnosus (JB-1) has been shown to inhibit pain perception of colorectal distension (CRD) in association with altered signalling in dorsal root ganglion (DRG) fibres in rats (Kamiya et al. 2006), and it also reduces the pain perception induced by gastric distension (GD) (Duncker et al. 2011); however, the pathways through which these effects may be occurring remain uncertain.

Pseudoaffective or visceromotor responses to CRD and GD are approaches used in rodent models to test whether the ingestion of specific potentially beneficial bacteria may reduce visceral pain perception (Kamiya et al. 2006; Rousseaux et al. 2007; Collins et al. 2009; Duncker et al. 2011), and they are commonly used to test whether any molecule or receptor is involved in visceral pain (Hong et al. 2009; Zhi et al. 2013).

Spinal fibres respond to gut distension from the physiological to the noxious (high intensity) level and the resulting afferent impulses contribute to visceral pain signalling (Berthoud et al. 2004). One of the key receptors responsible for pain perception in the gut is a member of the vanilloid receptor family, the transient receptor potential vanilloid 1 (TRPV1) (Winston et al. 2007; Boesmans et al. 2011; Holzer, 2011). The major cellular expression of TRPV1 in the gastrointestinal tract is in spinal and vagal primary afferent neurons (Holzer, 2011). An increased expression of this receptor has been found on neurons in rectal biopsies from patients with IBS and severe pain (Akbar et al. 2008) and blocking the in vivo phosphorylation of this channel through the inhibition of its interaction with a scaffolding protein has been shown to reduce inflammation-induced hyperalgesia in mice (Btesh et al. 2013).

Lactobacillus reuteri DSM 17938 (DSM) is reported to be effective in treating infantile colic (Savino et al. 2010) and modulation of mouse intestinal motility (Wu et al. 2013). To help elucidate the anti-nociceptive action of DSM in general, we examine here whether the bacteria inhibit the action of capsaicin on TRPV1 channel activity on young adult mouse intestinal mesenteric nerve bundle fibres, and DRG neurons.

We show that DSM substantially inhibits activation of the TRPV1 receptor by insurmountable antagonism and this evidence offers an explanation of how this bacterium may be anti-nociceptive. We contrast the action of DSM with the anti-nociceptive bacterium L. rhamnosus (JB-1), which does not diminish responses to capsaicin.

Methods

Ethical approval

All procedures adhered to Canadian Council on Animal Care guidelines and were approved by the McMaster University Animal Research Ethics Board.

Extracellular recordings

Adult male Swiss Webster (SW) mice (20–30 g, ∼3–4 months old) were procured from Charles River Laboratories (Wilmington, MA,USA). The mice were killed by cervical dislocation. All ensuing procedures were ex vivo. Male TRPV1-deficient (B6.129X1-Trpv1tm1Jul/J;C57 BL/6 background) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) and these were 20–25 g, ∼3 months old.

Segments of distal jejunum (∼2.5 cm) with attached mesenteric tissue were removed from freshly killed animals and placed in a Sylgard-coated Petri dish filled with Krebs buffer (in mm): 118 NaCl, 4.8 KCl, 25 NaHCO₃, 1.0 NaH₂PO₄, 1.2 MgSO₄, 11.1 glucose, and 2.5 CaCl₂ bubbled with carbogen (95% O₂–5% CO₂). The oral and anal ends of each segment were cannulated with plastic tubing and emptied. The tissue was pinned to the Sylgard, and the mesenteric nerve bundle exposed. The Petri dish was placed onto the stage of an inverted microscope and the lumen gravity perfused at 0.5–1 ml min⁻¹ with oxygenated Krebs buffer or Krebs with additives (Perez-Burgos et al. 2013). The serosal compartment was separately perfused with prewarmed (34°C) Krebs buffer at 3–5 ml min⁻¹. The nerve bundle was gently sucked into a glass pipette attached to a patch-clamp electrode holder, and extracellular nerve recordings made using a Multi-Clamp 700B amplifier and Digidata 1440A signal converter (Molecular Devices Sunnyvale, CA, USA). Electrical signals were bandpass-filtered at 0.1–2 kHz, sampled at 20 kHz, and stored on a personal computer running pCLAMP 10 software (Molecular Devices). Repeated distensions of segments were made by raising intraluminal pressure above 2 hPa. A constant gravity pressure head of 48 hPa was applied to the Krebs buffer perfusing the lumen and pressure raised by closing the outflow tube for 1 min to a maximum of three consecutive distensions. Segments were allowed to rest for 9 min between distensions. Constitutive multi-unit electrical activity was recorded in the absence of positive intraluminal pressure. When bacteria or media were tested for effect on nerve firing, they were continuously perfused via the jejunal lumen for at least 30 min by switching the luminal inflow from one containing control Krebs buffer to one containing Krebs with bacteria or media (Perez-Burgos et al. 2013).

Vagotomy

Subdiaphragmatic vagotomy was carried out as previously described (van der Kleij et al. 2008; Perez-Burgos et al. 2013). Mice were allowed to recover for 10–14 days before harvesting the jejunum and mesenteric tissue for electrophysiological experiments. Sham vagotomies were performed in three animals. Postoperatively, the body weight and general health of the mice were measured daily (Bluthe et al. 1996; Hosoi et al. 2002). We found no evidence of significant differences in weight gain 1 week post-surgery in either vagotomized or sham-treated animals (data not shown) (O’Mahony et al. 2009; Perez-Burgos et al. 2013). Vagotomy was only deemed to have been effective when serosal application of cholecystokinin (25–33) (CCK) did not increase mesenteric nerve firing rate (Hillsley & Grundy, 1998; Perez-Burgos et al. 2013).

Treatment protocol for gastric distension

Seventeen Sprague–Dawley rats were assigned to two groups. Rats were gavaged each morning for 9 days with either 0.2 ml (1 × 10⁹ cfu ml⁻¹) live DSM in Krebs buffer or only Krebs as control (vehicle).

The methods for GD have been previously published (Tougas & Wang, 1999). Rats were fasted overnight, anaesthetized with a mixture of ketamine hydrochloride (75 mg (kg body weight)⁻¹) and xylazine (10 mg (kg body weight)⁻¹) intraperitoneally. After a mid-line laparotomy, a ball-shaped gastric balloon (2 cm internal diameter) affixed to a Teflon catheter (20 cm) was inserted into the stomach through a small incision in the proximal duodenum and connected to a barostat system (Distender, G&J Electronics, Toronto, Canada). The cardiac response was measured while inflating the balloon with air to pressures of 40 and 60 mmHg for 60 s. Ten minutes of rest was allowed for recovery after every distension. Only one set of distensions was applied to each rat to minimize possible compensatory mechanisms.

Continuous recordings of heart rate (HR) were performed through a surface electrocardiogram consisting of three needle electrodes applied to the left and right shoulders and the right hind legs. The signal was amplified and recorded on a personal computer using a data acquisition program (Experimenter’s Workbench, DataWave Technologies, Loveland, CO, USA). The HR was measured for 60 s before, during and after each distension for a total of 180 s. The data are presented as mean change from resting HR (100% = rest) using the mean HR recorded for a period of 10 s during distension (10, 20, 30, 40, 50 and 60 s). Groups were compared using the mean of HR changes (percent of resting HR) in all rats of the same group during the 60 s of each distension (40 and 60 mmHg).

Dorsal root ganglion (DRG) primary cultures

The spinal columns of mice were removed from the body and transferred to a beaker containing ice-cold Krebs buffer. DRGs were exposed and collected from thoracolumbar levels. Whole DRGs were washed twice with sterile Leibovitz’s L-15 medium (Gibco, Gaithersburg, MD, USA), and incubated for 40 min in collagenase type 1 at 1 mg ml⁻¹ (Sigma-Aldrich; Oakville, ON, Canada) and 0.5 ml trypsin (0.25%, Gibco) in 20 ml L-15 at 37°C. After further addition of 5 ml L-15 containing 10% fetal bovine serum (FBS; Gibco), the ganglia were centrifuged for 5 min at 125 g, then washed twice with growth medium (L-15, containing 10% FBS, 1% penicillin–streptomycin–glutamine, 1% Hepes and 1% sodium pyruvate). The DRGs were placed in 2 ml of growth medium and triturated 10 times. The ganglia were then centrifuged for 10 s at 31 g and the supernatant transferred to a sterile tube. They were resuspended in 2 ml of growth medium, triturated repeatedly until the volume of supernatant transferred was 10 ml, centrifuged for 5 min at 125 g, and the final pellet resuspended in 9 ml of growth medium. Neurons were plated onto three glass-bottomed Petri dishes coated with poly-d-lysine (MatTek, Ashland, MA, USA) and incubated for ∼24 h at 37°C with carbogen.

Ca²⁺ imaging

DRG neurons were placed within a polymethyl methacrylate (Perspex) recording dish and loaded with the Ca²⁺ indicator Fluo-4-AM (8 μm) diluted in Krebs buffer with 0.1% pluronic acid (in DMSO) at 37°C for 60 min. The dish was superfused with fresh Krebs buffer (∼34°C) for 15 min to allow dye wash out. Cells were observed on an inverted microscope (Nikon Eclipse TE 2000-S, Melville, NY, USA) and imaged using a Rolera-XR camera (Surrey, BC, Canada). Fluorescence intensity in individual neurons was recorded by Simple PCI 6 software (Compix Inc., Imaging systems, Sewickley, PA, USA). Drugs were delivered via a micropipette attached to a Picospritzer II (General Valve, Fairfield, NJ, USA). Images, recorded at 0.9 frame s⁻¹, were stored on a local hard drive and analysed off-line using ImageJ software (NIH, USA, http://imagej.nih.gov/ij). The increase in Fluo-4 Ca²⁺ when applying capsaicin was measured as the ratio (F/F₀) fluorescence intensity after capsaicin (F) divided by the intensity before capsaicin (F₀).

Whole cell patch clamp

DRG neurons were isolated as described earlier and added to an electrophysiology recording dish and allowed to adhere to the bottom. We tested the effect of DSM on neurons that were shown to respond to capsaicin in a before and after experimental design protocol. TRPV1 expressing DRG neurons belong to the class of small to medium sized neurons and are more prevalent in visceral than somatic sensory neurons (Holzer, 2011). We thus selected for recording only those neurons with a diameter <20 μm. DRG neuron whole cell patch clamp recordings were performed as described in Ma et al. (2009) with the exception that 4 mm K₂ATP was added to the standard intracellular electrode solution (in mm: KMeSO4 110, NaCl 9, CaCl₂ 0.09, MgCl₂ 1.0, Hepes 10, Na₃-GTP 0.2, K₄-BAPTA 0.2, titrated to pH 7.3 with 0.1 m KOH) to minimize tachyphylaxis of the capsaicin response (Docherty et al. 1996).

Patch pipettes of 2–4 MΩ resistance in saline were used to contact the DRG neurons, and cell-attached seals of ≥3 GΩ were achieved with mild negative pressure via a 1 ml syringe. The whole cell configuration was entered by applying further suction to the patch electrode and the neuron monitored for a stable resting membrane potential ≤−50 mV. Then, after switching to voltage-clamp mode and with voltage set to −60 mV, a brief (5 ms) pulse of 1 μm capsaicin was puffed onto the neuron. The capsaicin was delivered via a 40 μm tip diameter glass pipette connected to a General Valve Picospritzer II operated under computer control.

Drugs and bacteria

DSM was donated by BioGaia AB (Stockholm, Sweden) and JB-1 was taken from in-house stock (see Bravo et al. 2011). All procedures related to the bacteria were as reported previously (Kunze et al. 2009; Ma et al. 2009; Wang et al. 2010a). Bacteria were taken from frozen stocks, thawed and centrifuged at 500 g for 15 min, and the resulting pellet suspended in Krebs solution. This was again centrifuged and resuspended. Bacteria were killed by gamma irradiation (γ-Rad) with cobalt 60 at 8.05 G min⁻¹. DSM conditioned medium consisted of centrifuged medium (500 g for 15 min), and showed no growth after 72 h of culture (Kamiya et al. 2006). Prior to use, bacteria, conditioned medium or medium alone was diluted to working concentrations with Krebs solution.

CCK was obtained from AnaSpec (Fremont, CA, USA); nicardipine, capsaicin and 6-iodonordihydrocapsaicin from Sigma-Aldrich, and ω-conotoxin GVIA (ω-Cg-GVIA) and ω-conotoxin MVIIC (ω-Cg-MVIIC) from Alomone Labs (Jerusalem, Israel). The competitive TRPV1 antagonist 6-iodonordihydrocapsaicin and CCK were dissolved in dimethyl sulphoxide (DMSO); capsaicin was dissolved in ethanol to make stock solution aliquots. On the day of the experiment the aliquots were diluted in Krebs solution to working concentrations with final DMSO and ethanol concentrations of ≤0.001% and ≤0.01%, respectively.

Off-line data analysis

Multi- and single-unit spike recordings were used to determine changes in the mesenteric nerve fibre firing rates induced by exposing the gut to differing stimuli or pharmacological agents (Booth et al. 2001; Ibeakanma et al. 2011; Perez-Burgos et al. 2013, 2014). The timing of spikes in the multi-unit recording was determined using the peak detection module of Clampfit 10.2 (Molecular Devices), and average firing frequency was calculated from interspike intervals. Single units were extracted from the multi-unit signal by spike shape matching using the shape template detection tool of Clampfit (computerized waveform analysis). After running the template detection algorithm, single-unit spikes were always checked by visual inspection, and non-matching events were discarded (<0.2%) (Perez-Burgos et al. 2013).

Statistics

Data are presented as means ± SEM with N referring to the total number of the jejunal segments recorded from, and n referring to the number of single fibres activity extracted from multi-unit recordings. We extracted a maximum of six single units from each raw recording. The Wilcoxon test or Student’s unpaired t test was used for paired or unpaired data comparisons, respectively; one- and two-way ANOVA with Sidak’s or Holm–Sidak’s post hoc test were used to compare multiple groups as appropriate. Since substantial variations in background spiking rates may occur between preparations (Rong et al. 2004), comparisons were paired with before and after treatment recordings and made so that each nerve bundle served as its own control to assure significance of changes in each treatment or drug dose. The percentage increase in firing above baseline frequency vs. capsaicin concentration (in the presence or absence of bacteria) was plotted and this was fitted by a logistic dose–response equation [Y = Bottom + (Top − Bottom/1 + 10^logEC₅₀^−X)]. The parameters describing the logistic fits were compared using a sum-of-squares F test (Motulsky & Neubig, 2010). All statistical tests were performed using Prism software 5.0 (GraphPad software, San Diego, CA, USA).

Results

Effects of DSM on mesenteric nerve spontaneous firing frequency

Luminal DSM decreased spontaneous multi-unit discharge of the mesenteric nerve with onset latencies of 10–20 min. 1 × 10⁸ cfu ml⁻¹ intraluminal DSM reduced firing frequency by 19% from 21.6 ± 5.1 to 17.6 ± 6 Hz (N = 5, P = 0.06, Fig. 1A); DSM at 1 × 10⁹ cfu ml⁻¹ caused a decrease in spontaneous discharge by 22% from 36.3 ± 8.4 to 28.2 ± 7.2 Hz (N = 7, P = 0.02, Fig. 1B). DSM conditioned medium (1:5) also decreased the spontaneous discharge by 37% from 22.6 ± 4.2 to 14.2 ± 2.4 Hz (N = 7, P = 0.03, Fig. 1C). However, γ-irradiated killed DSM or unconditioned medium alone did not substantially alter afferent firing: 17.8 ± 5.3 vs. 18.2 ± 4.1 Hz (N = 6, P = 0.84), and 20.5 ± 1.6 vs. 21.5 ± 2.8 Hz (N = 6, P = 0.10), respectively (Fig. 1D and E).

Figure 1.

Adding DSM to jejunal lumen decreased mesenteric multi-unit spontaneous firing frequency

Effects of 1 × 10⁸ cfu ml⁻¹ DSM (A), 1 × 10⁹ cfu ml⁻¹ DSM (B), diluted DSM conditioned medium (1:5) (C), 1 × 10⁹ cfu ml⁻¹ γ-irradiated DSM (D), and diluted medium alone (1:5) (E). (Data plotted as means ± SEM, Wilcoxon test.)

Vagotomy and smooth muscle relaxation did not inhibit the reduction by DSM of mesenteric nerve spontaneous firing

We tested if the effect of DSM (1 × 10⁹ cfu ml⁻¹) on spontaneous mesenteric nerve discharge was cancelled by prior vagotomy. To control for possible direct actions of DSM on smooth muscle contraction, we added here, and in all subsequent experiments that examined mesenteric nerve firing, the L-type Ca²⁺ channel blocker nicardipine (3 μm). Thus, after vagotomy and in the presence of nicardipine, DSM reduced multi-unit firing frequency by 18% from 16.7 ± 1.9 to 13.8 ± 1.9 Hz (N = 18, P = 0.001, Fig. 2A and C).

Figure 2.

Intraluminal DSM reduced spinal afferent fibre firing rate

Vagal afferents were removed by prior vagotomy, and smooth muscle relaxed by addition of nicardipine. A, multi-unit firing rate decreased when 1 × 10⁹ cfu ml⁻¹ DSM was added to the lumen (Wilcoxon test). B, spinal single-unit firing was reduced by DSM (Wilcoxon test). C, upper panels, representative traces of spontaneous multi-unit discharge before and after adding DSM; lower panels, superimposed waveforms of a typical single-unit.

DSM also reduced single-unit firing frequency by 19% from 0.36 ± 0.05 to 0.29 ± 0.03 Hz (n = 30, P = 0.02, Fig. 2B and C); of these fibres, the majority (20/30) decreased their firing rate by 36% from 0.42 ± 0.06 to 0.27 ± 0.04 Hz (P < 0.0001), but the smaller remaining fraction of more slowly firing fibres increased their firing by 29% from 0.24 ± 0.04 to 0.31 ± 0.06 Hz (n = 10, P = 0.006).

DSM reduced the capsaicin-evoked facilitation of spinal nerve fibre firing by insurmountable partial antagonism of TRPV1 receptors

We tested whether DSM could moderate the capsaicin-induced excitation of mesenteric nerve spinal fibres with tissue taken from vagotomized mice. Capsaicin, applied to the serosal compartment, increased spontaneous multi- (Miranda-Morales et al. 2010) and single-unit firing rates, in a dose-dependent manner with onset latencies of ∼60 s. TRPV1 capsaicin sensitive receptors desensitize, so subsequent applications of the agonist may produce diminished effects. Therefore, we examined responses to a range of capsaicin doses (100 nm to 100 μm) in individual jejunal segments. We did this with, and without, 20 min of prior intraluminal application of DSM 1 × 10⁹ cfu ml⁻¹ (N = 25 for each curve, 5 segments per each concentration; ∼6 spinal single-units were analysed for each segment).

The percentage increase in firing frequency versus capsaicin concentration or capsaicin plus DSM concentration was plotted and fitted with a three-parameter logistic equation of the form Y = Bottom + (Top − Bottom)/(1 + 10^logEC₅₀^−X). EC₅₀ for capsaicin alone was 200 nm compared to 500 nm for capsaicin in the presence of DSM (P = 0.71). The maximal response (Top) obtained with capsaicin was 239 ± 27% vs. 129 ± 17% obtained with capsaicin plus DSM (P = 0.004, Fig. 3A and B). In agreement with previous reports (Caterina et al. 1997; Boesmans et al. 2011; Holzer, 2011), some spinal fibres were not excited by capsaicin.

Figure 3.

DSM antagonized the excitatory response of spinal fibres to capsaicin in the serosal superfusate

A, capsaicin dose–response curve from 113 spinal individual single units was plotted and fitted with a three-parameter logistic equation, EC₅₀ = 200 nm. Max = 238 ± 27%. For an additional 116 fibres a dose–response curve for capsaicin in the presence of 1 × 10⁹ cfu ml⁻¹ DSM was also plotted and fitted with the same logistic equation as for capsaicin only, and for which EC₅₀ = 500 nm and Max = 129 ± 17% (P = 0.7 and P = 0.004 for differences in EC₅₀ and Max respectively, extra sum-of-squares F test). B, summary scatter plots with means and SEMs showing how single-unit responses varied with increasing doses of capsaicin in the absence or presence of 1 × 10⁹ cfu ml⁻¹ DSM. Numbers above scatter plots give sample sizes.

We examined if capsaicin excited spinal fibres directly, or whether the excitation depended on intramural synaptic transmission from enteric neurons to intraganglionic spinal endings that have been reported to approach the neurons (Mazzia & Clerc, 1997). We added 1 μm capsaicin to the serosal compartment after intramural transmission was blocked by adding 500 nm each of the Ca²⁺ blockers ω-Cg-GVIA and ω-Cg-MVIIC. The synaptic blockers did not diminish the capsaicin-evoked excitation, which was 188 ± 43% (n = 17) in the absence of the conotoxins and 219 ± 72% (n = 12) in their presence (P = 0.95).

Effects of DSM conditioned medium reduce capsaicin actions on spinal fibres

We tested whether DSM products released into the culture medium might mediate the TRPV1-dependent reduction in mesenteric nerve firing rate. We applied 1 μm capsaicin to the serosal compartment in jejunal segments that had received a prior intraluminal application of DSM-conditioned medium or medium alone for 20 min. DSM-conditioned medium (1:5) but not medium alone inhibited the capsaicin-induced firing frequency increase on mesenteric single fibres from 188 ± 43% (control group, N = 17) to 74.9 ± 22% (N = 14). This effect was comparable to the percentage increase induced by capsaicin on spinal fibres with 1 × 10⁹ cfu ml⁻¹ DSM treatment (80.3 ± 22%, N = 17) (P = 0.02, one-way ANOVA; Fig. 4).

Figure 4.

DSM conditioned medium mimicked the effect of DSM bacteria on spinal afferent response to capsaicin

Summary plot showing the firing frequency increase of single-unit spinal fibres induced by 1 μm capsaicin in control conditions, with DSM conditioned medium (1:5), or with 1 × 10⁹ cfu ml⁻¹ DSM (P = 0.02, one-way ANOVA; post hoc P values, Holm–Sidak’s multiple comparisons test).

DSM, TRPV1 antagonism or TRPV1 knockout reduced distension-evoked firing

We next recorded multi- and single-unit firing frequencies of spinal fibres (presumed to be nociceptors; Grundy, 2004) in the absence of nicardipine and thus allowing muscle contraction. We tested whether DSM could reduce the excitatory response evoked by raising intraluminal pressure to a nociceptive intensity (48 hPa) (Grundy, 2004). The first response of these fibres to gut distension is larger than that obtained with a subsequent distension, but remains constant for up to three further distensions (Perez-Burgos et al. 2013). We thus used the second of three successive distensions for comparisons between treatments. Multi-unit firing was 112 ± 17 Hz during distension, but after adding DSM for 20 min, distension increased firing to only 86 ± 11 Hz (N = 5, P = 0.31). Spinal single-unit firing frequency was 3.0 ± 0.4 Hz during distension, but on applying intramural DSM firing frequency was 1.7 ± 0.2 Hz during distension (n = 28, P = 0.008, Fig. 5A). The pre-distension background firing frequency was 0.3 ± 0.05 vs. 0.25 ± 0.07 Hz (n = 28, P = 0.05, Fig. 5A) for control vs. added DSM.

Figure 5.

DSM or TRPV1 antagonist reduced distension-evoked excitatory response in spinal single-units

A, scatter plots showing that adding 1 × 10⁹ cfu ml⁻¹ DSM to the lumen reduced the increase in spinal single-unit firing rate evoked by raising intraluminal pressure to 48 hPa (the data plotted were recorded during distension). B, adding 10 μm of the TRPV1 antagonist 6-iodonordihydrocapsaicin (6-Cap) to the lumen mimicked the effect of adding DSM. C, paired differences between mean interspike intervals (ISIs) for responses during distension with 6-Cap-containing Krebs buffer bathing the serosal surface and DSM in the lumen vs. ISI for responses with 6-Cap applied serosally and only Krebs buffer in the lumen. D, paired differences between ISI for responses using jejunum, taken from TRPV1 knockout mice (KO), during distension with DSM in the lumen vs. ISI for responses with only Krebs buffer in the lumen. P values calculated using Wilcoxon matched-pairs signed rank tests.

The TRPV1 antagonist 6-iodonordihydrocapsaicin (6-Cap, 10 μm), when applied to the serosal compartment, mimicked the effect of DSM on the single-unit firing by decreasing the response to distension from 3.3 ± 0.7 to 2.2 ± 0.5 Hz (n = 13, P = 0.0002, Wilcoxon test, Fig. 5B).

To compare the degrees with which DSM vs. 6-Cap reduced single fibre responses to distension, we converted the firing frequencies given in Fig. 5A and B to interspike intervals (ISIs) because frequencies cannot readily be subtracted or divided, and errors were recalculated for ISI. Thus, the average ISI during distension was 255 ± 82 ms in the presence of DSM and 152 ± 122 ms with 6-Cap but no DSM. We directly tested whether DSM had effects on distension-induced firing beyond reduction of TRPV1 signalling in two further separate experiments. We measured the effect on 48 hPa jejunal distension before and after adding luminal DSM, where for all such paired experiments TRPV-1 function was eliminated throughout by continuously adding 10 μm 6-Cap to the Krebs buffer perfusing the serosa or by using jejunum taken from TRPV1 KO mice. Distension-evoked single unit ISIs (from 6 mice) were 446 ± 36 ms in the presence of only 6-Cap and 543 ± 33 ms when DSM was added (N = 31, P < 0.0001, Wilcoxon test, Fig. 5C). For TRPV1 KO jejunum (from 5 mice) single unit ISI during distension was 457 ± 31 ms and this was increased to 561 ± 30 ms when DSM was added (N = 27, P < 0.0001, Wilcoxon test, Fig. 5D). Thus both types of experiments indicate that modulation of TRPV1 function determined most (80%) of the inhibitory action of DSM on the mesenteric nerve fibre response to high intensity distension.

DSM blocked the capsaicin-evoked Ca²⁺ rise in DRG neuronal primary cultures

The specificity of the effect of DSM on spinal neurons was further investigated by testing the bacteria’s ability to inhibit the intracellular Ca²⁺ rise evoked by capsaicin application onto the soma. Capsaicin activates TRPV1 channels allowing an influx on Ca²⁺ into DRG somata, and we used Ca²⁺ imaging of DRG soma primary culture to measure the consequent increase in intracellular [Ca²⁺]. Puffing 1 μm capsaicin onto DRG neurons evoked a prolonged rise in intracellular Ca²⁺ within ∼30 s (Fig. 6A). Since the TRPV1 channel may desensitize with repeated ligand exposure, we only applied capsaicin once to each culture dish. We next added either DSM or JB-1 30 min before applying capsaicin. DSM at 1 × 10⁹ cfu ml⁻¹ decreased the fluorescence rise ratio from 2.36 ± 0.31 (control group, N = 14) to 1.25 ± 0.04 F/F₀ (N = 14). But 1 × 10⁸ cfu ml⁻¹ DSM changed F/F₀ to 2.67 ± 0.35 (N = 7) and 1 × 10^8.5 cfu ml⁻¹ DSM to 2.07 ± 0.27 (N = 12). Adding 1 × 10⁹ cfu ml⁻¹ JB-1 had little effect and changed the F/F₀ ratio to 2.48 ± 0.19 (N = 9), which is similar to the ratio obtained with capsaicin alone (P = 0.002, one-way ANOVA; Fig. 6B).

Figure 6.

DSM reduced capsaicin-evoked intracellular [Ca²⁺] rise in dorsal root ganglia neuron somata

A, 1 μm capsaicin evoked a rapid and then prolonged increase in [Ca²⁺]. The rapid [Ca²⁺] increase (reflecting influx) was dose-dependently inhibited by DSM, but unaltered by 1 × 10⁹ cfu ml⁻¹ JB-1. B, summary plot (means ± SEM) showing how the fluorescence ratio (F/F₀) varied with DSM concentration or JB-1. F denotes the maximal Ca²⁺ fluorescence intensity evoked by capsaicin, and F₀ is equal to the baseline Ca²⁺ fluorescence intensity recorded before applying capsaicin (P = 0.002, one-way ANOVA; post hoc P values, Sidak’s multiple comparisons test).

DSM reduced the capsaicin-evoked TRPV1 ionic current in DRG neuronal primary cultures

Each DRG soma received two successive puffs of capsaicin with an inter-puff interval of 20 min (Fig. 7). For five control neurons the average amplitude of the inward current evoked by the first capsaicin puff was 2.1 ± 0.2 nA and that evoked by the second puff was 1.7 ± 0.4 nA. In another set of experiments for six neurons the Krebs buffer superfusing the cells was switched immediately following the first capsaicin puff to one containing also 10⁸ cfu ml⁻¹ DSM. Then the second puff was applied after 20 min incubation in DSM. For these experiments the first puff evoked an inward current of 1.9 ± 0.2 nA and the second one evoked a current of 0.8 ± 0.2 nA. The normalized second response was 83 ± 11% for the control neurons, but this was reduced to 41 ± 8% in the presence of DSM (P < 0.001, unpaired t test, 2 tailed) (Fig. 7C and D). To show that capsaicin acted on TRPV 1 receptors we established that when 10 μm 6-iodonordihydrocapsaicin was added to the recording bath solution prior to puffing capsaicin (n = 4) or that puffing Krebs only solution (n = 4) failed to evoke any current response in the neurons being recorded.

Figure 7.

Effect of 20 min incubation with DSM on capsaicin-evoked currents in DRG neurons

A–D, representative whole cell current traces recorded from patch clamped DRG neurons; time of application of 5 ms capsaicin puffs indicated by filled circles above current traced. A, inward current evoked by applying capsaicin was partially reduced (B) when a second similar puff was applied 20 min after the first one. C, current response to capsaicin puff was substantially reduced (D) for a different DRG neuron when 10⁸ cfu ml⁻¹ DSM was applied for 20 min between puffs. E, summary statistics (means ± SEM) for 5 control and 6 DSM treated neurons. The percentage normalized second capsaicin response was substantially smaller for neurons incubated with DSM.

DSM inhibited heart rate slowing evoked by gastric distension in an in vivo rat model

The heart rate was unchanged by 40 mmHg distension and was not moderated by prior gavaging DSM for 9 days prior to testing (P = 0.1, Fig. 8A) (N = 8 and 9 with vehicle and DSM, respectively). Inflation to 60 mmHg decreased heart rate within 10 s which persisted for 30 s during distension (Fig. 8B). Gavaging with DSM decreased the bradycardia response to 60 mmHg (P = 0.03, unpaired t test, see Fig. 8A and B). Gastric compliance (volume/pressure) did not differ between animals treated with vehicle or DSM, for gastric distension pressures of 40 or 60 mmHg (data not shown).

Figure 8.

Nine-day feeding with DSM reduced gastric distension-evoked bradycardia

A, summary values for percentage decreases in resting heart rate evoked by 40 and 60 mmHg gastric distension (P values, unpaired t tests). B, summary plots showing how resting heart rate changed over time in response to 60 mmHg gastric distension. (P = 0.01, two-way ANOVA.)

Discussion

Since DSM decreased the firing frequency of nociceptive fibres, we wondered if the TRPV1 channel was involved since this is a major receptor involved in visceral nociception (Btesh et al. 2013). Verdu et al. showed that antibiotic therapy induced visceral pain hypersensitivity, which was inhibited by probiotic therapy (Verdu et al. 2006). Early life treatment of rats with vancomycin resulting in an altered gut microbiome, induced heightened visceral pain perception accompanied by a decrease in spinal cord TRPV1 expression (O’Mahony et al. 2014). Increased expression of this receptor by intestinal sensory fibres has been correlated with abdominal pain in patients suffering from IBS (Akbar et al. 2008). TRPV1 belongs to the transient receptor potential (TRP) ion channel family (Holzer, 2008) and is a polymodal thermo-, chemo- and mechanosensitive channel (Bielefeldt et al. 2006) that is non-selective for cations and permeable to Ca²⁺ (Cortright & Szallasi, 2004). It is expressed in unmyelinated visceral afferents and activated by capsaicin, noxious heat, acidosis (pH < 5.9), depolarization and endovanilloids (Caterina et al. 1997; Boesmans et al. 2011; Vay et al. 2012). Approximately 80% of spinal neurons express TRPV1 (Robinson et al. 2004; Brierley et al. 2005), but only a minority (32%) of vagal afferents do so (Brierley et al. 2005; Tan et al. 2010). A subgroup of spinal afferents (∼20%) reacts only to noxious levels of distension and fails to respond under normal circumstances (Cervero, 1994; Ozaki & Gebhart, 2001). These ‘silent nociceptors’ are activated under conditions of injury or inflammation (Grundy, 2004). They are mainly mechanosensitive and are likely to be involved in the afferent transmission of intestinal pain signals (Cervero, 1994). Both vagal and spinal fibres are involved in noxious stimulation of the stomach with mechanoceptive pain signals being carried mainly by spinal fibres (Lamb et al. 2003). Our single unit analysis of mesenteric nerve response clearly demonstrates that the TRPV1 channel mediates 80% of the afferent discharge response to high intensity mouse jejunum distension. Identification of the receptors involved in the residual response is beyond the scope of the present paper but we speculate that they might be any combination of non-TRPV1 mechanoreceptors such as TRPA1, ASIC (Page et al. 2005) and TRPV4 (Brierley et al. 2008).

The present work is the first to identify a likely sensory neuronal molecular target for the anti-nociceptive actions of a Lactobacillus (L. reuteri DSM 17938). A previous study reported that an anti-nociceptive L. acidophilus strain altered the expression of cannabinoid and opioid receptors in intestinal epithelial cells, but left open the crucial question as to whether sensory neurons were involved (Rousseaux et al. 2007).

The fact that DSM blocked the capsaicin-evoked rise in intracellular Ca²⁺ in isolated DRG neurons also supports our contention that DSM acts on extrinsic sensory neurons by inhibiting the activation of TRPV1 receptors. The substantial reduction of capsaicin-induced inward currents by DSM in cultured DRG neurons provides further supporting evidence that the actions of DSM on the extrinsic primary sensory terminals are on TRPV1 channels. Surprisingly the effects of the conditioned medium alone reproduced those of the live bacteria. Since dead (gamma-irradiated) DSM failed to show the inhibitory action of live DSM, it is likely that a secreted or shed product is responsible. The bacteria’s shed products consist in part of the microvesicles released by them, and these contain many of the surface and other bacterial components contained in the parent bacteria (MacDonald & Kuehn, 2012). We have recently shown that such microvesicles can reproduce most of the enteric neuronal and immune effects of a parent Lactobacillus delivered in the small intestinal lumen (Al-Nedawi et al. 2015). These microvesicles are less than 150 nm in diameter so that they can easily diffuse into tissues, and this offers one explanation of how luminal bacteria might influence the enteric nervous system.

The specificity of the activity for this nociceptive receptor offers an opportunity to identify the components responsible for the anti-nociceptive, and possibly other, beneficial effects of DSM. There is one study that found evidence that nitric oxide (NO) increment is involved in the anti-nociceptive effects of Lactobacillus farciminis (Ait-Belgnaoui et al. 2009). This may provide further avenues for future studies because NO has been implicated as an endogenous regulator of TRPV1 opening through S-nitrosylation of extracellular cysteines (Takahashi et al. 2012; Morales-Lazaro et al. 2013).

We have previously reported that JB-1 reduces the pain-related pseudoaffective responses and spinal nerve single fibre firing rates induced by gut distension (Kamiya et al. 2006; Duncker et al. 2011). Here we report that DSM, but not JB-1, reduced the capsaicin-evoked Ca²⁺ rise in DRG neurons. Thus, while inhibition of TRPV1 activation might be used to screen for potential anti-nociceptive agents, this method could not identify those that modulate the activity of other classes of pain-transducing molecular sensors.

Several clinical trials have shown diverse findings regarding DSM as an effective treatment to inhibit the symptoms of infantile colic (Savino et al. 2010; Szajewska et al. 2013; Sung et al. 2014), but there is as yet no data on its effect in young adults. Since there is evidence that dichotomizing axons from T4-L1 DRG neurons innervate both the stomach and the small intestine (Zhong et al. 2008) and because it is less invasive, we tested DSM’s effects on the pseudoaffective responses to GD in vivo to supplement our ex vivo electrophysiological data (Results). DSM moderated the gastric distension evoked bradycardia (Duncker et al. 2011). Since bradycardia is a reliable pseudoaffective sign of noxious gastric stimulation such moderation demonstrates an anti-nociceptive effect by the bacteria (Tougas & Wang, 1999). Therefore we believe that the functional evidence we have obtained for in vivo moderation of visceral pain strongly supports some of the clinical data, taken together with the electrophysiological and dynamic imaging data. The DSM-induced reduction of pain perception on gastric distension confirmed the effectiveness of this probiotic as an antinociceptive agent and speaks to its potential therapeutic uses.

The present results, in combination with previous work cited above, suggest a heterogeneity of mechanisms for the actions of anti-nociceptive lactobacilli, further emphasizing that the actions of non-pathogenic, potentially beneficial, anti-nociceptive bacteria may differ at the molecular level according to species and strains (Bercik et al. 2010; Wang et al. 2010a; Cryan & O’Mahony, 2011).

Taken together, our results strongly support the claim that DSM directly or indirectly can act as a visceral anti-nociceptive agent through the specific insurmountable antagonism of TRPV1 on extrinsic spinal primary sensory fibres and their corresponding DRG cell bodies. Further, our results offer a model system to screen candidate bacteria for potential visceral anti-nociceptive activity.

Glossary

CCK

cholecystokinin (25–33) sulphated

CFU

colony-forming unit

CRD

colorectal distension

DRG

dorsal root ganglion

DSM

Lactobacillus reuteri DSM 17938

FBS

fetal bovine serum

GD

gastric distension

HR

heart rate

IBD

inflammatory bowel diseases

IBS

irritable bowel syndrome

JB-1

Lactobacillus rhamnosus JB-1

SW

Swiss Webster

TRPV1

transient receptor potential vanilloid 1

ω-Cg-GVIA

ω-conotoxin GVIA

ω-Cg-MVIIC

ω-conotoxin MVIIC

Additional information

Competing interests

None.

Author contributions

A.P.B., J.B. and W.A.K. contributed to the study concept and design. A.P.B. performed most electrophysiological experiments, analysed and interpreted the data. L.W. performed gastric distension experiments. K.A.M.N. contributed to electrophysiological acquisition of data. Y.M. performed vagotomies and provided technical support. M.A. contributed to acquisition and analysis of imaging data. L.J.J. provided material support and advice for imaging analysis. A.M.S. and A.P.B. performed imaging experiments. A.P.B., J.B. and W.A.K. contributed to manuscript writing. All authors contributed to the critical revision and approved the final version of the manuscript. All contributors who qualify for authorship are listed.

Funding

This study was supported by a grant from the Guglietti Family Foundation, an unrestricted grant from BioGaia AB and a grant from the Natural Sciences and Research Engineering Council (NSERC) to W.A.K. (RGPIN-2014-05517). K.A.M.N. was supported by a Father Sean O’Sullivan Research Centre Fellowship.