Male TLR4^-/- (B6.B10ScN-Tlr4^lps-del/JthJ; 9 ± 1 weeks old) and sex- and age-matched wild-type (WT) C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME, United States) were housed in individually ventilated cages (IVC) at the animal facility of the Department of Pharmaceutical and Pharmacological Sciences, University of Padova. To normalize gut microbiota, mice colonies from both groups were housed in the same room and generally in the same cages, and maintained by the same personnel. All animals were specific pathogen-free and given standard chow diet and tap water ad libitum. All experimental protocols were approved by the Animal Care and Use Ethics Committee, University of Padova and Italian Ministry of Health (authorization number: 1142/2015-PR) and are reported in compliance with ARRIVE guidelines (McGrath and Lilley, 2015).

Ileal specimens, fixed in 10% buffered formalin, embedded in paraffin and cut into 4 μm-sections were stained with haematoxylin and eosin (H&E) (Dall’Olmo et al., 2014). Ten slides for each genotype were analyzed blindly.

Intestinal paracellular permeability was assessed as previously described (Aubé et al., 2006). Briefly, WT and TLR4^-/- mice were gavaged orally with absorbable fluorescein isothiocyanate (FITC)-dextran (4 kDa molecular weight; 200 μl, 600 mg/kg body weight). Preliminary experiments at various time points (0, 0.5, 1, 2, 4, and 6 h) showed that the appearance in blood of low molecular weight FITC-dextran peaked at 4 h following oral administration in WT and TLR4^-/- mice. After 4 h, FITC-dextran serum concentration was determined using a fluorimeter (PerkinElmer, Milan, Italy) at 490/530 nm.

WT and TLR4^-/- mice were gavaged orally with 70-kDa FITC-dextran (25 mg/ml in 0.9% saline solution). Preliminary experiments at various time points (0, 30, 45, 60, 75, and 90 min) showed that the distribution of high molecular weight FITC-dextran peaked at 30 min in the ileum following oral administration in WT and TLR4^-/- mice. After 30 min, the stomach and caecum were examined separately while small intestine and colon were divided into 10 and 3 comparable length-segments, respectively. Gastrointestinal transit was determined using the intestinal geometric center of FITC-dextran distribution throughout the intestine as described previously (Aubé et al., 2006; Brun et al., 2013). Gastric emptying was determined by percentage of fluorescent probe that emptied the stomach (Aubé et al., 2006).

Fecal pellet output was assessed daily as previously described. Mice were observed for 60 min. Fecal pellet numbers, fecal wet weights and dry weights were determined. The difference between wet and dry weights was used to calculate fecal water content (McQuade et al., 2016).

Contractility studies were performed as previously described (Giron et al., 2008; Brun et al., 2013; Zoppellaro et al., 2013). Briefly, 1-cm longitudinal segments from the distal ileum were mounted in 10-mL-organ baths equilibrated for 30 min in Krebs solution (37°C). Changes in muscle tension were recorded by isometric transducers connected to a PowerLab4/30 system (ADInstruments, Oxford, United Kingdom). Carbachol (0.001–100 μM) and ADP (0.001–1 mM) dose-response curves were obtained cumulatively. Non-receptor-mediated contractile responses to 60 mM KCl were assessed. Neuronal-mediated contractions were analyzed following electrical field stimulation (EFS; 0–40 Hz; 40 V) in basal conditions or in non-adrenergic non-cholinergic (NANC) conditions, obtained by adding 1 μM guanethidine and 1 μM atropine to the organ bath. 10 Hz-EFS-mediated NANC responses were evaluated in presence of the pan-NOS inhibitor L-NAME (100 μM), the iNOS inhibitor 1400W (10 μM), the P1-purinoceptor antagonist theophylline (100 μM), the P2-purinoceptor antagonist suramin (100 μM), the P2Y1 receptor (P2Y1R) antagonist MRS2500 (1 μM) or the P2X7R antagonist A804598 (0.1 μM).

Ten hertz-EFS-mediated tachykininergic responses were recorded in NANC conditions with 100 μM L-NAME. Contractile responses were expressed as gram tension/gram dry tissue weight of ileal segments. Ileal relaxation was calculated as the percentage reversal of the initial gram tension/dry tissue weight, setting 100% inhibition as the maximum relaxation (Zizzo et al., 2003).

Distal ileal 10 cm-segments were filled with fixative solution (4% paraformaldehyde in PBS) for 1 h at room temperature. Using a dissecting microscope, whole mount specimens of longitudinal muscle-myenteric plexus (LMMP) were prepared as previously described (De Mello et al., 2009; Brun et al., 2013). Briefly, LMMP preparations were gently stretched and pinned down on a wax support and subjected to acetylcholine esterase (AChE) or NADPH-diaphorase (NADPHd) biochemical staining (Anitha et al., 2006). Stained tissues mounted on glass slides were observed using a Leica DMI4000 B microscope (Leica Microsystems GmbH, Wetzlar, Germany). AChE⁺ myenteric fibers and NADPHd⁺ neuronal cells analysis was performed blindly by counting fibers or neurons in 10 randomly-chosen images per mouse (six animals/group), as previously described (De Mello et al., 2009).

Longitudinal muscle-myenteric plexus whole mount preparations were gently stretched and pinned down on a wax support and permeabilized in PBT (PBS with 0.3% Triton X-100) and blocked with 2% bovine serum albumin (BSA) for 1 h at room temperature, as previously reported (Brun et al., 2013). Distal ileum (0.5 cm-segments) were frozen in optimal cutting temperature mounting medium (OCT), sectioned (7 μm-thick) with a cryostat (Leica CM 1850 UV, Milan, Italy) and then mounted onto Superfrost Plus slides. From each ileal specimen, 100 sequential 7 μm-cross-sections were cut on a cryostat and 6–8 sections were subjected to immunohistochemistry as previously described (Giron et al., 2008; Zoppellaro et al., 2013). LMMP whole mount preparations or ileal cryosections were then incubated overnight at room temperature with the following antibodies: chicken polyclonal anti-mouse glial fibrillary acidic protein (GFAP; 1:100; Abcam, Milan, Italy), rabbit polyclonal anti-human GFAP (1:200; Merck Millipore Corporation, Milan, Italy), mouse biotin-conjugated anti-human HuC/D (1:50; Thermo Fisher Scientific, Milan, Italy), rabbit polyclonal anti-mouse neuronal nitric oxide synthase (nNOS, 1:100; Thermo Fisher Scientific, Milan, Italy), rabbit polyclonal anti-human inducible NOS (iNOS; 1:50, Santa Cruz Biotechnology, Milan, Italy), rabbit polyclonal anti-human vasoactive intestinal peptide (VIP, 1:100; GenWay Biotech, Milan, Italy), rabbit polyclonal anti-human P2X7 receptor-ATTO-488 (1:100; Alomone labs, Jerusalem, Israel), rabbit polyclonal anti-human P2Y1 receptor (1:100; Alomone labs, Jerusalem, Israel), rabbit monoclonal anti-human S100β (1:100; Merck Millipore Corporation, Milan, Italy) and guinea pig polyclonal anti-mouse substance P (1:100; Abcam, Milan, Italy). Then cryosections or LMMP whole mount preparations were washed and incubated for 2 h at room temperature with the following secondary antibodies: Alexa Fluor 488-conjugated goat anti-chicken IgG (1:1000; Thermo Fisher Scientific, Milan, Italy), Alexa Fluor 488-conjugated goat anti-guinea pig IgG (1:500; Thermo Fisher Scientific, Milan, Italy), Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:1000; Thermo Fisher Scientific, Milan, Italy), DyLight 649-conjugated goat anti-rabbit IgG (1:500, Jackson ImmunoResearch Laboratories, Milan, Italy) or Alexa Fluor 555-conjugated streptavidin (1:1000, Thermo Fisher Scientific, Milan, Italy). Nuclei were stained with TOTO-3 (1:500; Thermo Fisher Scientific, Milan, Italy) or with 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI) (1:1000; Thermo Fisher Scientific, Milan, Italy) added together with the secondary antibodies. After three washes, LMMP whole mount preparations or cryosections were mounted on glass slides using a Mowiol Mounting Medium (100-mM Tris-HCl (pH 8.5), 9% Mowiol 4–88, 25% glycerol and 0.1% DABCO). Negative controls were obtained by incubating sections with isotype-matched control antibodies at the same concentration as primary antibody and/or pre-incubating each antibody with the corresponding control peptide (final concentration as indicated by manufacturer’s instructions) (Brun et al., 2013).

Images were acquired using a Nikon D-Eclipse C1 confocal microscope (Nikon Instruments, Florence, Italy), equipped with an oil-immersion Nikon Plan-Apo 60×/1.4 objective or a low-magnification Nikon Plan Fluor 20×/0.5 objective. Five Z-series images (10 planes for ileum cryosections or 15 planes for LMMP whole mount preparations) of 1024 pixels × 1024 pixels were processed as maximum intensity projections. All microscope settings were kept constant for all images. Fluorescence intensity of S100β, substance P, P2X7 receptors, P2Y1 receptors and iNOS was assessed for each antigen in 20 images per mouse (N = 5 mice/group), as previously reported (Arqués et al., 2014). The obtained fluorescence values indicating the fluorescence intensity of each protein of interest (i.e., S100β, substance P, P2X7 receptors, P2Y1 receptors or iNOS) were normalized for the fluorescence intensity of its own TOTO-3 or DAPI and were reported as mean ± SEM, as previously described (Orso et al., 2005).

Fluorescence intensity of GFAP⁺ fibers was determined by applying the skeleton analysis method developed to quantify brain microglia morphology as previously described (Morrison et al., 2013). Briefly, for skeleton analysis, the maximum intensity projection of the GFAP⁺ channel was enhanced to image all enteric glial processes, followed by noise de-speckling to eliminate single-pixel background fluorescence. After converting the resulting images to binary, they were skeletonized using ImageJ software and then analyzed by the AnalyzeSkeleton plugin to determine the number of endpoints per frame and process length. These data were normalized to a total area of 13.14 mm², obtained from 20 images per mouse (N = 5 mice/group) in order to assess changes in EGCs. Analyses of images and related fluorescence intensity were performed using ImageJ software (Fiji, version 1.51n).

In ileal LMMP whole mount preparations, total neuron population analysis was performed by counting HuC/D⁺ cells in 10 randomly-chosen images per mouse (N = 8 mice/group). The total number of HuC/D⁺ neurons was recorded in each image and normalized to the total area of 4.05 mm². To evaluate the distribution of nitrergic and VIPergic neurons in ileal myenteric plexus, the number of nNOS⁺ and VIP⁺ enteric neurons was blindly counted in 10 randomly-chosen images per mouse (N = 8 mice/group) and normalized to the total area of 4.05 mm².

Unless otherwise specified, all chemicals were obtained from Sigma–Aldrich (Milan, Italy) and were of the highest commercially available analytical grade. OCT was purchased from Kaltek (Padua, Italy), PFA was from Electron Microscopy Sciences-Società Italiana Chimici (Rome, Italy), β-NADPH was obtained from Diagnostic Brokers Associated (Milan, Italy), and Triton-X-100 was from Applichem (Milan, Italy).

All data are expressed as mean ± SEM. Differences between the experimental groups were assessed using paired or unpaired Student’s t-test and one-way or two-way analysis of variance (ANOVA), followed by post hoc Bonferroni test. Data were analyzed using GraphPad Prism 3.03 (San Diego, CA, United States). The results were considered statistically significant at p < 0.05; “N” values indicate the number of animals.

Since TLR4 expression is required for normal growth (including villus height) of the small intestine (Riehl et al., 2015), we sought to determine whether the absence of TLR4 influences structural architecture by examining hematoxylin and eosin-stained sections. In agreement with previous findings (Riehl et al., 2015), TLR4^-/- ileal morphology was comparable to WT, except for villi height, which was significantly diminished (349 ± 9 μm in WT mice vs. 306 ± 15 μm in TLR4^-/- mice; N = 20 animals/group). No significant differences in serum levels of absorbable FITC-dextran were found between TLR4^-/- and WT mice (0.42 ± 0.1 μg/ml and 0.32 ± 0.1 μg/ml, respectively; N = 12 animals/group), to indicate no alterations in intestinal permeability. Considering that TLR4 is expressed in ENS (Rumio et al., 2006; Barajon et al., 2009), the impact of TLR4 absence on ENS integrity was evaluated by immunohistochemistry. In ileal cryosections, a 1.84-fold increase in the immunoreactivity of the glial marker S100β was found in TLR4^-/- myenteric plexus (Figures 1A,B). These increases in S100β immunoreactivity were associated to a 3.1-fold increase in process length of GFAP⁺ gliofilaments in TLR4^-/- ENS (Figures 1C–E).

Previous studies have shown a reduced gastrointestinal transit 4 h after non-absorbable FITC-dextran administration in C3H/HeJ mice with a spontaneous point mutation in Tlr4 gene (Tlr4^Lps-d) (Anitha et al., 2012). We hypothesized that the same functional impairment could be present in TLR4^-/- mice (B6.B10ScN-Tlr4^lps-del), which are homozygous for a null mutation of Tlr4 gene.

In WT mice, the non-absorbable FITC-dextran transited through the gastrointestinal tract over a 30-min period and localized at the terminal ileum (Figure 2A). Conversely, a significant reduction of geometric center and gastric emptying was observed in TLR4^-/- mice (Figures 2B,C). The number of fecal pellets/hour and stool water content were significantly lower in TLR4^-/- mice compared to WT mice (Figures 2D,E).

Based on the delayed gastrointestinal transit, we assessed spontaneous contractility that resulted comparable in frequency and amplitude in both genotypes (Figures 3A,B). To evaluate excitatory responses, cumulative concentration-response curves to the non-selective cholinergic receptor agonist, carbachol (CCh) were performed. Ileal segments from TLR4^-/- mice showed a significant downward shift of the concentration-response curve to CCh and a significant related decrease in the maximum response compared to WT (E_max = -25.6 ± 7.5%; Figure 3C). However, the response to high potassium-induced depolarization was similar in both genotypes (Figure 3D).

Since ENS structural abnormalities have been described in Tlr4^Lps-d mice (Anitha et al., 2012), we sought to test neuromuscular function by analyzing frequency-response curves to EFS. Altered neurotransmission in TLR4^-/- ileal segments was reflected by reduced EFS-elicited contractions (by 28.5 ± 9.7% at 10 Hz, Figure 4A). The EFS-induced contractions up to 10 Hz were of neuronal cholinergic origin as confirmed by their sensitivity to tetrodotoxin (TTX) and to the muscarinic receptor blocker atropine, as previously shown (Brun et al., 2013). However, no changes in the number of AChE⁺ fibers were found in TLR4^-/- whole-mount preparations (Figures 4B,C).

To evaluate the contribution of other excitatory neurotransmitters besides acetylcholine, we evaluated the post-stimulus excitatory responses in NANC conditions, which are determined by tachykininergic neurotransmission (Lecci, 2002). In WT mice, NANC responses evoked by EFS determined a transient relaxation of ileal preparations, followed by TTX-sensitive excitatory responses (Figures 5A,B) (Zizzo et al., 2003). These excitatory responses were significantly reduced in TLR4^-/- ileal segments (by 30 ± 8.3%; Figures 5A,B). Upon addition of L-NAME, a tachykinin-mediated excitatory response (Lecci, 2002) was found in TLR4^-/- preparations comparable to WT (Figure 5C). Accordingly, no differences were observed in the immunofluorescence distribution of substance P (SP) in WT and TLR4^-/- frozen sections (Figure 5D).

Considering that nitrergic neurotransmission (Zizzo et al., 2004; Lomax et al., 2010), the primary inhibitory pathway in the gut, is affected by dysbiosis (Kabouridis and Pachnis, 2015), we tested whether the reduced excitatory contraction could be the result of an increase of the inhibitory component. Consistent with this prediction, a reduction of nitrergic neurons stained with NADPH diaphorase (NADPHd) or anti-nNOS was observed in TLR4^-/- preparations (Figure 6). These changes were accompanied by a significant reduction of the total number of HuC/D⁺ neurons in TLR4^-/- myenteric plexus (Figures 6C–E). The reduction in nNOS⁺ neurons was associated with a proportional increase of VIP⁺ neurons in TLR4^-/- mice (Figure 6).

In NANC conditions EFS at 10 Hz caused a 1.48-fold increase in relaxation in TLR4^-/- mice (Figure 7A). Pretreatment with 1400W, a selective inhibitor of iNOS, significantly reduced the NANC-mediated relaxation in TLR4^-/- mice (by 25.2 ± 0.5%, Figure 7A), whereas a slight but not significant relaxation (13.8 ± 0.9%, Figure 7A) was recorded in WT. These findings support an involvement of iNOS in NO-mediated relaxation in the absence of TLR4. Furthermore, iNOS immunoreactivity increased by 5.7-fold in TLR4^-/- myenteric neurons and EGCs compared to WT (Figures 7B,C). Pretreatment with the pan-NOS inhibitor L-NAME almost completely blocked EFS-evoked NANC relaxation in WT mice. Conversely, in TLR4^-/- mice, this response was only partially abolished by L-NAME (Figure 7A) suggesting an influence of TLR4 in nitrergic-mediated relaxation and possibly in other inhibitory pathways (e.g., purinergic or VIPergic), known to sustain intestinal contractility (Zizzo et al., 2004). Accordingly, we evaluated the role of adenosine and adenosine-5′-triphosphate (ATP) in modulating relaxation in NANC conditions. Pretreatment with L-NAME and theophylline, a non-selective adenosine receptor antagonist, partially reduced the amplitudes of NANC-mediated relaxation in both genotypes whereas the addition of L-NAME and suramin, a non-selective ATP receptor antagonist, resulted in a significant reduction of the inhibitory response (by 75.3 ± 1.5%, Figure 7D) in TLR4^-/- mice, reaching a relaxation amplitude comparable to WT.

Considering that inhibitory neurotransmission in TLR4^-/- mice depends on both nitrergic and ATP-mediated relaxation, we evaluated the modulatory effect of ADP, the endogenous agonist of P2Y1 receptors (P2Y1Rs). TLR4^-/- mice showed a 1.43-fold increase in relaxation amplitude following addition of ADP in the organ bath with a significant shift to the left of the dose-response curve to ADP compared to WT mice (Figure 8A). Since also enteric P2X7 receptors (P2X7Rs) respond to ATP by mediating inhibitory neurotransmission and are involved in neuronal death during intestinal inflammation (Antonioli et al., 2014; Brown et al., 2015), we examined the influence of P2Y1Rs and P2X7Rs in NANC-mediated relaxation in the absence of TLR4. In ileal tissues from TLR4^-/- mice, P2Y1Rs blockade with MRS2500 in presence of L-NAME markedly reduced the amplitudes of NANC-mediated relaxation (by 45.1 ± 3%) whereas the addition of A804598 (a selective P2X7Rs antagonist) determined a reduction of the inhibitory response (by 24.2 ± 1.5%, Figure 8B), suggesting an involvement of both receptors in modulating relaxation in NANC conditions in the absence of TLR4.

Immunohistochemical analysis revealed a 1.79-fold increase of P2Y1Rs staining in myenteric ganglia, in both neurons and EGCs (Figure 9) and a 2.4-fold increase in P2X7Rs immunoreactivity in TLR4^-/- myenteric neurons, underlining the involvement of TLR4 in ensuring ENS homeostasis (Figure 10).