Wnt1-Cre;R26R-GCaMP3 mice (aged 6–13 weeks) of either sex were killed by cervical dislocation. All procedures are approved by the Animal Ethics Committees of the University of Leuven (Belgium) and the University of Melbourne (Australia). Wnt1-Cre;R26R-GCaMP3 mice express the fluorescent Ca²⁺ indicator GCaMP3 in all enteric neurons and glia (Danielian et al., 1998; Zariwala et al., 2012). A segment of jejunum (approximately 5 cm long) was collected and immersed in 95% oxygen/5% carbon dioxide-bubbled Krebs solution (in mM: 120.9 NaCl, 5.9 KCl, 1.2 MgCl₂, 1.2 NaH₂PO₄, 14.4 NaHCO₃, 11.5 glucose and 2.5 CaCl₂). The tissue was opened along the mesenteric border and pinned flat with mucosa up in a silicone elastomer-coated dish. Some tissues were immediately fixed in 4% formaldehyde/PBS for 45 min at room temperature for immunohistochemical staining. In others that were used for imaging, the submucosal plexus with the mucosa intact was separated from the underlying muscle layers by microdissection, and the tissue was stabilized by stretching the tissue with the plexus facing up over an inox ring, before clamping it with a rubber O-ring (Vanden Berghe et al., 2002). Up to five ring preparations were obtained from each jejunal segment.

Two microscopy setups were used to image the ring preparations. In Leuven (BE), a 20× (NA 1.0) water dipping lens was used on an upright Zeiss Examiner microscope (Carl Zeiss, Oberkochen, Germany) equipped with a monochromator (Poly V) and cooled CCD camera (Imago QE), both from TILL Photonics (Gräfelfing, Germany). Images (640 × 512) on this system were acquired at 2 Hz. In Melbourne (AU), the ring preparations were imaged via an Olympus 20× (NA 0.5) water dipping lens on an upright Zeiss Axioscope microscope with a Zeiss AxioCam MRm camera and images (278 × 278) were acquired at 1 Hz. Preparations were constantly superfused (1 ml/min) with 95% oxygen/5% carbon dioxide-gassed Krebs at room temperature.

Ganglia of interest were stimulated chemically or electrically. Agonists were applied to the preparations by pressure ejection (2 s duration; 9 psi) via a micropipette (tip diameter ~20 μm) positioned approximately 50 μm from the selected ganglion. For time control experiments, each agonist was applied three times separated by 5 min to ensure reproducible responses. For electrical stimulation experiments, trains of 300 μs pulses (2 s, 20 Hz) were applied by a focal stimulating electrode (tungsten; 50 μm diameter) on an internodal strand leading to the ganglion imaged. For time control experiments, ganglia were stimulated twice separated by 5 min.

Antagonist experiments consisted of two consecutive applications of agonist or two consecutive trains of electrical stimulation, with the antagonist superfused prior to the 2nd stimulation. Thus the 1st response is considered the control response. Antagonists were superfused via a local perfusion pipette. Each preparation was only exposed to a single antagonist once and to only one type of agonist. At least three animals were examined for each set of experiments.

After calcium imaging experiments, selected preparations were fixed in 4% formaldehyde/PBS for 45 min at room temperature for post hoc immunohistochemistry.

Agonists used included VIP (human, mouse, rat; Bachem, Bubendorf, Switzerland), VPAC1 agonist, [K15, R16, L27] VIP(1-2)/GRF(8-27) and VPAC2 agonist, BAY 55-9837 (both synthesized by Mimotopes, Clayton, VIC, Australia), and the P2Y1 agonist 2-methyl-thio-ADP (2MeSADP; Sigma Aldrich, Castle Hill, NSW, Australia). Superfusion of 1 μM VIP depolarized myenteric neurons (Palmer et al., 1987). Hence, 100 μM VIP was used to fill the micropipette for local application by pressure ejection, as the solution may be diluted up to approximately 100-fold in superfused preparations (Liu et al., 1997). The VPAC1 agonist [K15, R16, L27]VIP(1-2)/GRF(8-27) and the VPAC2 agonist BAY 55-9837 were also used at 100 μM, as VIP and [K15, R16, L27]VIP(1-2)/GRF(8-27) have similar potencies on the VPAC1 receptor (Gourlet et al., 1997b), while BAY 55-9837 has shown to be effective on hypothalamic neurons when superfused at 1–10 μM (Hermes et al., 2009; Pantazopoulos et al., 2010).

VIP antagonists used include the VPAC1 antagonist PG97-269, and the VPAC2 antagonist, PG 99-465 (both from Mimotopes). PG97-269 inhibits VIP-evoked intestinal secretion at 1 μM (Fung et al., 2014), while PG 99-465 inhibits VIP-evoked responses in hippocampal neurons at 100 nM–10 μM (Pakhotin et al., 2005; Cunha-Reis et al., 2014). Accordingly, we superfused each VPAC antagonist at a concentration of 1 μM. Other antagonists include tetrodotoxin (TTX; 1 μM; Acros, Geel, Belgium); hexamethonium bromide (200 μM; Sigma; Bornem, Belgium), pyridoxal-phosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS; 30 μM; Sigma), and MRS2179 (10 μM; Tocris, Bristol, UK). The VPAC1- and VPAC2 antagonists were applied 5 min prior to agonist application as they have been shown to be most effective with short incubations (Pakhotin et al., 2005; Fang et al., 2006). TTX, hexamethonium, PPADS and MRS2179 were superfused for 10 min.

All drugs were prepared as stock solutions using water, except for TTX which was prepared in citrate buffer. TTX, 2MeSADP, and MRS2179 were stored at 4°C. Hexamethonium bromide, PPADS, VIP, [K15, R16, L27]VIP(1-2)/GRF(8-27), BAY 55-9837, PG97-269 and PG 99-465 were stored as aliquots at −20°C. Agonists and antagonists were diluted in Krebs to achieve the desired final concentration.

Only ganglia directly adjacent to the micropipette were considered for analysis. Neurons were identified by the size of their cell bodies (~20 μm diameter; Gabella and Trigg, 1984) and lack of fluorescence in their nuclei. Glia were identified by the size of their cell bodies (<5 μm diameter; Gabella and Trigg, 1984), and their morphology and location (either intraganglionic type I cells with irregular branched processes, or elongated type II cells within/at the edge of internodal strands; Boesmans et al., 2015). Ca²⁺-imaging analyses were performed with custom-written routines in Igor Pro (Wavemetrics, Lake Oswego, OR, USA). Regions of interest were drawn to calculate the average [Ca²⁺]_i signal intensity. Values were then normalized to the baseline fluorescence intensity (F_i/F_o). Responses were considered when the [Ca²⁺]_i signal increased above baseline by at least five times the intrinsic noise. [Ca²⁺]_i peaks were calculated for each response, with the peak amplitude taken as the maximum increase in [Ca²⁺]_i from baseline (ΔF_i/F_o). For time controls, the Δ_i/F_o of the 2nd agonist exposure is presented as a percentage of the 1st (% Δ_i/F_o). Similarly, the Δ_i/F_o evoked in the presence of antagonists is presented as a percentage of the 1st control agonist response. Responses that were observed with the 2nd agonist exposure, but not the 1st were not considered in this analysis. The total number of cells responding to each agonist exposure was also counted. The total number of cells responding to the 2nd agonist application was then similarly normalized to that of the 1st control response and presented as a percentage of the control.

Immediately-fixed jejunal segments from mice of a C57Bl6 background including Wnt1-Cre;R26R-GCaMP3 mice, and fixed submucosal ring preparations, were used for immunohistochemical staining. The mucosa and submucosa of immediately-fixed tissues were separated from the underlying muscle layers by microdissection. The mucosa of immediately-fixed preparations and fixed submucosal ring preparations was then removed and the preparations were incubated with a blocking buffer containing 4% donkey serum (Merck Millipore, Overijse, Belgium) and 0.5% triton X-100 (Sigma) in PBS overnight at 4°C. Preparations were subsequently incubated in primary antisera (Table 1) for 24–48 h at 4°C, washed in PBS (3 × 10 min) and then incubated in secondary antisera (Table 2) for 2 h at room temperature. Primary and secondary antisera were diluted in the blocking buffer. Preparations were washed in PBS (3 × 10 min) before mounting on slides with Citifluor (Citifluor Ltd., Leicester, UK) or Dakocytomation fluorescent mounting medium (Carpinteria, CA, USA).

Fluorescently labeled preparations were viewed under an epifluorescence microscope (BX 41 Olympus, Olympus, Aartselaar, Belgium) equipped with UMNUAUV, U-MWIBA3 and U-MWIY2 filtercubes for visualizing blue, green and red probes, respectively. Images were acquired with an XM10 (Olympus) camera using Cell^F software. Pictures were adjusted for contrast and brightness before overlay and quantification. Confocal images were recorded using a Zeiss LSM780 confocal microscope (Cell Imaging Core, KU Leuven, Belgium) or Zeiss Pascal confocal microscope (Biological Optical Microscopy Platform, The University of Melbourne, Parkville, VIC, Australia).

Data are presented as mean ± the standard error of the mean (SEM). One-way analysis of variance (ANOVA) followed by Dunnett’s post hoc test (where appropriate) were conducted to determine statistical significance, unless specified otherwise. P < 0.05 was considered significant. Analyses were performed using GraphPad Prism 5.0 (GraphPad softwares, San Diego, CA, USA).

Local VIP (100 μM) application by pressure ejection evoked transient increases in intracellular Ca²⁺ concentration ([Ca²⁺]_i) in neurons (Figure 1A); 44 ± 5% of neurons per ganglion responded with consistent [Ca²⁺]_i rises (ΔF_i/F_o: 0.28 ± 0.02, n = 103 neurons, 18 preparations; Figure 1B). Glial [Ca²⁺]_i transients were rarely observed (2/25 ganglia). Ganglia imaged in these experiments contained on average 9.3 ± 0.5 neurons. Post hoc immunohistochemistry was performed on some preparations to characterize the types of neurons which responded to VIP (100 μM; Figure 1C). Both ChAT⁺ neurons and ChAT⁻ neurons responded to VIP, but the characteristics of [Ca²⁺]_i transients observed in the two neuronal subtypes were distinctly different (Figure 1D). VIP-responsive ChAT⁺ neurons displayed [Ca²⁺]_i transients with a duration (at t50%; 50% of the peak amplitude) of 11.8 ± 0.8 s (n = 24 neurons), while ChAT⁻ neurons displayed slower responses with a significantly longer duration of 15.3 ± 1.0 s (n = 42 neurons; unpaired t-test, P = 0.016). The average latency of responses was 7.6 s ± 0.5 and ranged from 1.5 s to 18 s, but did not differ between ChAT⁺ and ChAT⁻ neurons.

The VIP-evoked response was nearly abolished by the Na⁺-channel blocker TTX (1 μM), which significantly reduced the amplitude of the response (to 15 ± 10% of the control; one-way ANOVA and Dunnett’s test, P < 0.0001; n = 40 neurons, 8 preparations; Figure 2A) and the number of responding neurons per ganglion (to 16 ± 7% of the control; one-way ANOVA and Dunnett’s test, P < 0.001; Figure 2B). This suggests that the majority of [Ca²⁺]_i transients observed were secondary to action potential generation. Furthermore, reminiscent of the effect of the VPAC2 antagonist, TTX uncovered VIP-evoked glial responses, with 1.9 ± 0.6 glial cells responding per ganglion (n = 15 glia; Figure 2C).

As VPAC1 receptors have been previously shown to be expressed on many cholinergic submucosal neurons (Fung et al., 2014), we examined the involvement of cholinergic transmission in the response to VIP using the nicotinic receptor antagonist hexamethonium (200 μM). Blocking this major form of fast excitatory neurotransmission within the ENS reduced the amplitude of VIP-evoked responses (to 32 ± 7% of the control; one-way ANOVA and Dunnett’s test, P < 0.001; n = 24 neurons, 6 preparations; Figure 2A). In contrast to TTX and the VPAC2 antagonist (see below), hexamethonium did not reveal VIP-evoked glial responses.

We next examined the relative contributions of VPAC1 and VPAC2 receptors to the VIP-induced [Ca²⁺]_i transients using the VPAC1 antagonist PG 97-269 (Banks et al., 2005) and the VPAC2 antagonist PG 99-465 (Moreno et al., 2000; Dickson et al., 2006). The VIP (100 μM)-evoked neuronal response amplitudes were significantly reduced in the presence of the VPAC1 antagonist PG 97-269 (1 μM; to 22 ± 5% of the control; one-way ANOVA and Dunnett’s test, P < 0.0001; n = 47 neurons, 9 preparations; Figure 2A), but the number of neurons responding did not change (Figure 2B). The VPAC2 antagonist PG 99-465 (1 μM) effectively reduced both the amplitude of the VIP-evoked response (to 22 ± 9% of the control; one-way ANOVA and Dunnett’s test, P < 0.0001; n = 30 neurons, 7 preparations; Figure 2A) and the number of responding neurons per ganglion (to 36 ± 18% of the control; one-way ANOVA and Dunnett’s test, P < 0.05; Figure 2B). These data indicate that the VIP-evoked [Ca²⁺]_i transients in neurons involve both VPAC1 and VPAC2 receptor activation. Interestingly, in 6/7 preparations, glial responses to VIP were uncovered in the presence of the VPAC2 antagonist, where a [Ca²⁺]_i rise was observed in 1.6 ± 0.4 glial cells per ganglion (n = 11 glial cells; Figure 2C).

As inhibiting VPAC2 receptors unveiled glial [Ca²⁺]_i transients in response to VIP, we investigated whether the VIP-induced glial response is mediated via VPAC1 receptors. Local spritz application of [K15, R16, L27]VIP(1-2)/GRF(8-27; VPAC1 agonist; 100 μM; Gourlet et al., 1997b) to submucosal ganglia predominantly evoked [Ca²⁺]_i responses in glial cells (Figure 3A). By contrast, BAY 55-9837 (VPAC2 agonist; 100 μM) did not evoke reproducible [Ca²⁺]_i responses in neurons or glia. The VPAC1 agonist evoked [Ca²⁺]_i transients in 3.5 ± 0.4 glia per ganglion (Figure 3A) with a peak amplitude (ΔF_i/F_o) of 0.42 ± 0.03 (n = 108 glial cells, 22 preparations; Figure 3B). With post hoc immunohistochemistry for glial fibrillary acidic protein (GFAP), we confirmed that the fibers activated in response to the VPAC1 agonist were glial (Figures 3C–E). The types of glial cells responding were classified based on their location and morphology (Boesmans et al., 2015). We found that predominantly type I enteric glial cells responded to VPAC1 agonist application: 2.1 ± 0.3 type I vs. 1.4 ± 0.2 type II cells displayed [Ca²⁺]_i transients per ganglion (paired t-test, P = 0.03; n = 31 ganglia).

The VPAC1 agonist also evoked some neuronal [Ca²⁺]_i responses (ΔF_i/F_o: 0.65 ± 0.1; n = 38 neurons, 22 preparations), where 13 ± 3% of neurons responded per ganglion. In this set of experiments, the mean number of neurons contained per ganglion imaged was 10.6 ± 0.6. Examining the time course of the VPAC1 agonist-evoked responses, some neuronal responses preceded glial activation and, interestingly, some neuronal responses were observed following the onset of a [Ca²⁺]_i rise in glia. Neurons that responded before glia displayed [Ca²⁺]_i transients with a latency of 4.3 ± 1.2 s (n = 10 neurons). Glial responses had a latency of 14 ± 1.0 s (n = 108 glia) and neurons that responded following glial activation had a latency of 19 ± 2 s (n = 28 neurons).

We used PG-97-269 (VPAC1 antagonist; 1 μM) to test the specificity of the VPAC1 agonist (100 μM), and it significantly reduced the number of responding glia per ganglion (to 58 ± 11% of the control; one-way ANOVA and Dunnett’s test, P < 0.05; Figure 4B), as well as the peak amplitude of the glial [Ca²⁺]_i transients (to 19 ± 5% of the control; one-way ANOVA and Dunnett’s test, P < 0.01; n = 50 glial cells, 13 preparations; Figure 4A). However, the VPAC1 agonist-induced glial response was unaffected by TTX (1 μM; n = 24 glial cells, 5 preparations; Figure 4).

Given that the VPAC1 agonist-evoked glial response was TTX-insensitive, we next assessed whether enteric glia express VPAC1 receptors. Localization of the VPAC1 receptor was examined using immunohistochemistry (Figure 5). VPAC1 receptor expression was specifically neural, with the antisera labeling a subset of submucosal neurons and neuronal, but not glial, processes. VPAC1 receptor-immunoreactivity was observed in 12 ± 2% Hu⁺ cells and these were predominantly ChAT⁺ (93 ± 4%; n = 284 neurons; 3 animals; Figures 5A–D). Some background nuclear labeling was observed in neurons, but this staining was not considered in the analysis. VPAC1 receptor staining also colocalized with peripherin, a label for neuronal fibers (Figures 5E–H; n = 3 animals). However, VPAC1 receptors did not colocalize with GFAP⁺ glial processes (n = 3 animals; Figures 5I–L). Since we found that VPAC1 receptors are expressed on some ChAT⁺ neurons, we also examined whether VPAC1 receptors may also be expressed on cholinergic varicosities by co-labeling for vesicular acetyltransferase (VAChT). However, there was no apparent overlap between VPAC1 and VAChT (n = 2 animals; Figures 5M–O).

As it has been reported that enteric glia receive innervation from sympathetic fibers in guinea pig myenteric plexus (Gulbransen et al., 2010), we addressed whether VPAC1 receptors may also be expressed on sympathetic adrenergic nerve fibers. However, we did not observe any colocalization between tyrosine hydroxylase (TH; labels adrenergic nerves) and VPAC1 receptor expression (n = 2 animals; Figures 6A–C). In addition to TH nerve fibers, many submucosal neurons of mouse ileum are surrounded by calcitonin gene-related peptide (CGRP)-immunoreactive varicosities (Mongardi Fantaguzzi et al., 2009). Some colocalization of VPAC1 receptors with CGRP-immunoreactive varicosities was observed (n = 2 animals; Figures 6D–F).

To confirm the presence of a VIP/VPAC neuron-to-glia signaling pathway, we stimulated fiber tracts electrically to release endogenously-produced VIP. In these experiments, trains of stimuli evoked neuronal [Ca²⁺]_i transients (ΔF_i/F_o: 0.80 ± 0.08; n = 86 neurons) that were unaffected by the VPAC2 antagonist PG 99-465 (1 μM; n = 115). However, this antagonist revealed delayed glial responses to electrical stimulation in 7/13 preparations, where 1.3 ± 0.2 glia responded per ganglion. These delayed glial responses had a latency of 20.5 ± 2.9 s (n = 10 glial cells) and were not observed in time controls.

Purinergic signaling is important for enteric neuron-glia communication (Gomes et al., 2009; Gulbransen and Sharkey, 2009), so we examined the effect of purinergic antagonists on the VPAC1 agonist-stimulated glial responses. The P2 receptor antagonist PPADS (30 μM), significantly reduced the peak amplitude of the VPAC1 agonist-induced glial [Ca²⁺]_i response to 7 ± 3% of the control (one-way ANOVA and Dunnett’s test, P < 0.01; n = 31 glial cells, 6 preparations; Figure 4A). We then tested a specific P2Y1 antagonist, as P2Y1 receptors are associated with enteric glial activation (Gomes et al., 2009; Gulbransen et al., 2012). The P2Y1-selective antagonist MRS2179 (10 μM) significantly inhibited the amplitude of the VPAC1 agonist-induced glial response (to 10 ± 5% of the control; one-way ANOVA and Dunnett’s test, P < 0.05; n = 20 glial cells, 5 preparations; Figure 4A).

To further investigate the role of P2Y1 receptors, we examined the response to spritz-applied 2MeSADP (100 μM; P2Y1-specific agonist). Consistent with observations in mouse myenteric plexus (Brown et al., 2016), stimulation with 2MeSADP primarily evoked [Ca²⁺]_i transients in submucosal glial cells. 2MeSADP evoked responses in 4.8 ± 0.4 glial cells per ganglion (ΔF_i/F_o: 0.16 ± 0.02; n = 43 glial cells, 6 preparations). We tested the specificity of 2MeSADP and found that the peak amplitude of the response was significantly reduced by MRS2179 (10 μM; to 21 ± 6% of the control; one-way ANOVA and Dunnett’s test, P < 0.05; n = 29 glial cells, 5 preparations; Figure 7A). TTX (1 μM) did not inhibit 2MeSADP-evoked glial responses, rather the number of glia responding was increased (to 122 ± 27% of control; one-way ANOVA and Dunnett’s test, P < 0.05; n = 6 preparations; Figure 7B).

We tested if agonists for the two receptors evoke [Ca²⁺]_i signals in the same glial cells to examine whether P2Y1 receptors act within the same pathway as that activated by VPAC1 receptors. We compared the [Ca²⁺]_i responses to 2MeSADP (100 μM) and VPAC1 agonist (100 μM) by sequentially applying these agonists to same ganglion, with each application separated by a 5 min washout. 2MeSADP evoked [Ca²⁺]_i responses in significantly more glia per ganglion than the VPAC1 agonist (4.6 ± 0.4 vs. 2.8 ± 0.5 respectively; paired t-test, P = 0.027; n = 10 ganglia, 7 preparations), and responses were unaffected by the order of agonist application. Notably, some glial cells responded to both agonists (Figures 7C,D), but 2MeSADP induced higher [Ca²⁺]_i peak amplitudes (ΔF_i/F_o: 0.23 ± 0.02 vs. 0.18 ± 0.02 respectively; paired t-test, P = 0.026; n = 24 glial cells). Moreover, the time course of the responses to the two agonists differed drastically (Figures 7E,F). The 2MeSADP-evoked glial response was rapid and short-lived with a latency of 2.8 ± 0.2 s, while the VPAC1 agonist glial response exhibited a significant delay with a latency of 16.3 ± 2.1 s (paired t-test, P < 0.0001; n = 24 glial cells).