The pTRKH3-slpGFP vector (Addgene, Watertown, Massachusetts, United States) was first modified to remove the GFP sequence at SalI/PstI restriction sites, insert the T7 transcriptional terminator at BamHI/EcoRV sites, and insert the linker sequence containing BsaI-BsaI at PstI/XmaI restriction sites. The cDNA of human NAPE-PLD was then inserted into the BsaI sites using In-Fusion method (Clontech, Mountain View, CA, United States). The resulting pTRKH3-slp-NAPE-PLD and parental plasmid (not expressing NAPE-PLD gene, used as negative control) constructs were transfected into the L. paracasei subsp. paracasei F19 strain (Arla Foods, Hoersholm, Denmark) by electroporation, and positive clones were obtained by erythromycin (5 μg/ml) selection. Both parental plasmid (pLP) and NAPE-PLD expressing bacteria (pNAPE-LP) were amplified anaerobically in Man, Rogosa, and Sharpe (MRS) broth (Conda, Torrejón de Ardoz Madrid, Spain) and isolated in MRS agar (Conda, Torrejón de Ardoz Madrid, Spain) both supplemented with erythromycin 5 μg/ml (Sigma-Aldrich, Milan, Italy) under anaerobic conditions for 72 h at 37°C. Bacteria viability was determined by manually counting colonies, and the colony forming units (CFU)/ml was obtained through a colony number correction for the dilution factor.

Six-week-old wild-type (WT) male C57BL/6J (Charles River, Laboratories, Italy) and PPARα knockout (KO) (Taconic, Germantown, New York, United States) mice were used for the experiments. All the procedures were approved by La Sapienza University's Ethics Committee in compliance with the IASP and European Community (EC L358/1 18/12/86) guidelines on the use and protection of animals in experimental research. As depicted in Scheme 1, C57BL/6J mice were randomly divided into four experimental groups (n = 10 per group): 1) vehicle (control group); 2) TcdA group; 3) TcdA + pLP group; and 4) TcdA + pNAPE-LP group, respectively. Both vehicle and TcdA groups received 200 μl MRS broth by intragastric gavage for one week, while the TcdA + pLP group was administered a daily volume of 200 μl of MRS broth suspension containing 10⁹ CFU of pLP strain with 0.0003 μg/ml of sodium palmitate (Sigma-Aldrich, Milan, Italy). In the same conditions, the TcdA + pNAPE-LP group was administered by intragastric gavage with a daily volume of 200 μl of MRS broth suspension containing 10⁹ CFU of pNAPE-LP and 0.0003 μg/ml of sodium palmitate. PPARα KO mice were randomly divided into three experimental groups (n = 6 per group): 1) vehicle (control group); 2) TcdA group; and 3) TcdA + pNAPE-LP group, respectively, and treated as above. One week following probiotic administration, mice received by intrarectal route, a single administration of phosphate buffered saline 1× (PBS 1×) or TcdA (50 μg/ml dissolved in PBS 1×), according to the method described by Hirota et al. (2012) to induce acute pseudomembranous colitis. Since animals’ death was not an acceptable experimental end point, based on the Institutional Animal Care and Use Committee (which ethically prevents the death of the animal as a terminal event of the procedure), we used a nonlethal dose of TcdA, and after 4 h from the intrarectal administration, all mice used for the experimental plan were deeply anesthetized before being euthanatized, as described by Hirota et al. (2012). Blood samples were collected by intracardiac puncture, and tissues were isolated and processed to further analyses (see below).

Extraction and analysis of PEA released in vitro and in vivo were performed according to Gachet et al. (Gachet et al., 2015), with slight modifications. Bacterial cultures were ultracentrifuged at 14,000 rpm for 10 min to separate the culture medium and bacterial pellet. Culture media were freeze-dried and then resuspended in a solution containing acetonitrile with 0.1% formic acid (extraction solution). Then, the samples were ultracentrifuged (14,000 rpm, 4°C, 5 min) and the supernatant injected for the mass spectrometry analysis. Both bacterial pellet and mice colonic tissues were firstly lysed using a lysis buffer and then evaporated under a nitrogen stream. Residues were suspended in extraction solution, ultra-centrifuged (14,000 rpm, 4°C, 5 min), and the supernatant injected for the mass spectrometry analysis. Analyses were run on a Jasco Extrema LC-4000 system (Jasco Inc., Easton, MD) coupled to an Advion Expression mass spectrometer (Advion Inc., Ithaca, NY) equipped with an electrospray (ESI) source. Mass spectra were recorded in the positive SIM mode. The capillary voltage was set at +180V, the spray voltage was at 3 kV, the source voltage offset was at +20 V, and the capillary temperature was set at 250°C. The chromatographic separation was performed on analytical column Kinetex C18 (150 × 4.6 mm, id.3 µm, 100 Å) and security guard column both supplied by Phenomenex (Torrance, CA, United States). The analyses were performed at a flow rate of 0.3 ml/min, with solvent A (water containing 2 mM ammonium acetate) and solvent B (methanol containing 2 mM ammonium acetate and 0.1% formic acid). Elution was performed according to the following linear gradient: 15% B for 0.5 min, 15–70% B from 0.5 to 2.5 min, 7–99% B from 2.5 to 4.0 min, and held at 99% B from 4.0 to 8.0 min. From 8 to 11.50 min , the column was equilibrated to 15% B and conditioned from 11.5 to 15.0 at 15% B. The injection volume was 10 µl, and the column temperature was fixed at 40°C. For quantitative analysis, standard curves of PEA (Sigma-Aldrich, Milan, Italy) were prepared over a concentration range of 0.0001–10 ppm with six different concentration levels and duplicate injections at each level. All data were collected and processed using JASCO ChromNAV (version 2.02.04) and Advion Data Express (4.0.13.8).

After sacrifice, mice distal colon specimens were fixed in 4% paraformaldehyde (PFA), sectioned into 15 μm slices, and stained with hematoxylin and eosin (H and E) (Sigma-Aldrich, Milan, Italy) for macroscopic and histopathological assessment. Colonic histological damage was evaluated through a complex score, according to the criteria proposed (Li et al., 2017), considering the following parameters: 1) distortion and loss of crypt architecture (0 = none; 1 = mild; 2 = moderate; 3 = severe); 2) infiltration of inflammatory cells (0 = normal; 1 = mild infiltration; 2 = moderate infiltration; 3 = dense infiltration); 3) muscle thickening (0 = normal; 1 = mild muscle thickening; 2 = moderate muscle thickening; 3 = marked muscle thickening); 4) goblet cells depletion (0 = absence; 1 = presence); and 5) crypt absence (0 = absence; 1 = presence). Slices were analyzed with a microscope Optika XDS-3L4 Ponteranica, BG, Italy), and images were captured at 4× magnification by a high-resolution digital camera (Nikon Digital Sight DS-U1). The cumulative histological damage score was expressed as average scores in each experimental group derived by the observations of two independent qualified observers (CC and GE).

Samples for immunohistochemical assessment of macrophages were isolated from both WT and PPARα KO mouse distal colon. Tissues were fixed in 4% PFA, embedded in paraffin, sectioned in 15 μm slices, and processed for immunohistochemistry. Slices were pretreated using heat-mediated antigen retrieval with a sodium citrate buffer, incubated with MAC387 (Abcam, Cambridge, United Kingdom) at room temperature (RT) (Thoree et al., 2008), and detected using the horseradish peroxidase (HRP)–conjugated compact polymer system. 3,30-diaminobenzidine (DAB) was used as the chromogen. Slices were then counterstained with hematoxylin, mounted with Eukitt (Sigma-Aldrich, Milan, Italy), and analyzed with a microscope (Optika XDS-3L4 Ponteranica, BG, Italy). Images were captured at 10× with a high-resolution digital camera and the data represent the median results of the two blinded assessors (CC and GE); in all cases, results of the assessments differed by no more than 5%. Results were quantified by ImageJ software (National Institutes of Health) and expressed as the number of macrophage marker antibody (MAC387) positive cells per area.

Myeloperoxidase (MPO), a marker of polymorphonuclear leukocyte accumulation, was determined as previously described (Mullane et al., 1985). After removal, colonic tissues from both WT and PPARα KO mice were rinsed with a cold saline solution, opened, and deprived of the mucosa using a glass slide. The resulting layer was then homogenized in a solution containing 0.5% hexadecyltrimethylammonium bromide (Sigma-Aldrich, Milan, Italy), dissolved in 10 mM potassium phosphate buffer, and centrifuged for 30 min at 20,000 × g at 37°C. An aliquot of the supernatant was mixed with a solution of tetramethylbenzidine (1.6 mM; Sigma-Aldrich, Milan, Italy) and 0.1 mM hydrogen peroxide (Sigma-Aldrich, Milan, Italy). The solution was then spectrophotometrically measured at 650 nm. MPO activity was determined as the amount of enzyme degrading 1 mmol/min of peroxide at 37°C and was expressed in milliunits (mu) per 100 mg of wet tissue weight.

Proteins were extracted from colonic tissue or bacteria pellets and processed by Western blot analysis. For protein extraction by the bacterial pellet, a specific CelLytic™ lysis buffer (Sigma-Aldrich, Milan, Italy) was used according to manufacturer’s instructions. Tissue samples were homogenized in ice-cold hypotonic lysis buffer [10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM phenylmethylsulphonylfluoride, 1.5 μg/ml soybean trypsin inhibitor, 7 mg/ml pepstatin A, 5 mg/ml leupeptin, 0.1 mM benzamidine, and 0.5 mM dithiothreitol (DTT)]. Both bacterial- and tissue-deriving protein extracts were mixed with a nonreducing gel loading buffer [50 mM Tris (hydroxymethyl)aminomethane (Tris), 10% sodium dodecyl sulfate (SDS), 10% glycerol, 2 mg/ml bromophenol] at a 1:1 ratio, and then boiled for 3 min followed by centrifugation at 10,000×g for 10 min. The protein concentration was determined using Bradford assay and equivalent amounts (50 μg) of each homogenate underwent electrophoresis through a polyacrilamide minigel. After the transfer the membranes were incubated with 10% nonfat dry milk in PBS overnight at 4°C and then exposed, depending on the experiments, with rabbit polyclonal anti-NAPE-PLD (Abcam, Cambridge, United Kingdom) (1:200 v/v), rabbit polyclonal anti–toll-like receptor-4 (TLR4) (Bioss Antibodies, Boston, United States) (1:1,000 v/v), mouse monoclonal anti-RhoA-GTPase (Santa Cruz Biotechnology, Santa Cruz, CA, United States) (1:100 v/v), rabbit polyclonal anti-p38 mitogen-activated protein kinase (p38 MAPK) (Bioss Antibodies, Boston, United States (1:1,000 v/v), rabbit monoclonal anti-phospho-p-38 (p-p38) MAPK (Santa Cruz Biotechnology, Santa Cruz, CA, United States) (1:1,000 v/v), rabbit polyclonal anti-NF-κB p65 (Sigma-Aldrich, Milan, Italy) (1:1,000 v/v), mouse monoclonal anti-NF-κB p50 (Santa Cruz Biotechnology, Santa Cruz, CA, United States) (1:1,000 v/v), mouse monoclonal anti–hypoxia-inducible factor-1-alpha (HIF-1α) (Novus biological, Abingdon, United Kingdom) (1:500 v/v), and rabbit polyclonal anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Cell Signaling Technology, Danvers, MA, United States) (1:1,000 v/v) according to standard experimental protocols. Membranes were then incubated with the specific secondary antibodies conjugated to HRP (Dako, Milan, Italy). Immune complexes were exposed to enhanced chemiluminescence detection reagents, and the blots were analyzed by scanning densitometry (Versadoc MP4000; Bio-Rad, Segrate, Italy). Results were expressed as optical density (OD; arbitrary units = mm²) and normalized against the expression of the housekeeping protein GAPDH. Immune complexes were revealed by enhanced chemiluminescence detection reagents (Amersham Biosciences, Milan, Italy) and exposed to the Kodak X-Omat film (Eastman Kodak Co., Rochester, NY, United States OK). Protein bands were then scanned and densitometrically analyzed with a GS-700 imaging densitometer. Results were expressed as OD (arbitrary units; mm²) and normalized on the expression of the housekeeping protein GAPDH for mice and proteins.

Before being sacrificed, mice were deeply anesthetized. Blood samples were taken by cardiac puncture and collected in 5% ethylenediaminetetraacetic acid (EDTA) vials, immediately prior to sacrifice. To determine nitric oxide (NO), interleukin-6 (IL-6), and vascular endothelial growth factor (VEGF) levels, plasma was then isolated from the blood, immediately frozen, and stored at −80°C until the assays.

Enzyme-linked immunosorbent assay (ELISA) for IL-6 and VEGF (all from Thermo Fisher Scientific Inc., Monza, Italy) was carried out on mice plasma according to the manufacturer’s protocol. Absorbance was measured on a microtiter plate reader. IL-6 and VEGF levels were determined using the standard curves method.

NO production was measured as nitrite (NO₂ ⁻) accumulation in murine plasma by a spectrophotometer assay based on the Griess reaction (Di Rosa et al., 1990). Briefly, Griess reagent (1% sulfanilamide and 0.1% naphthylethylenediamine in H₃PO₄) was added to an equal volume of plasma, and the absorbance was measured at 550 nm. NO₂ ⁻ concentration (nm) was thus determined using a standard curve of NaNO₂.

Segments of distal mouse colon were isolated and fixed in ice-cold 4% PFA and sectioned into 20 μm slices. Sections were thus blocked with bovine serum albumin and subsequently stained with mouse anti–zonula occludens-1 (ZO-1) antibody (Bioss Antibodies, Boston, United States) (1:100 v/v) or rabbit anti-occludin antibody (Novus biological, Abingdon, United Kingdom) (1:100 v/v). Slices were then washed with PBS 1× and incubated in the dark with fluorescein isothiocyanate–conjugated anti-rabbit (Abcam, Cambridge, United Kingdom). Nuclei were stained with 2-(4-amidinophenyl)-1H -indole-6-carboxamidine (DAPI) (Thermo Fisher Scientific, Massachusetts, United States). Sections were analyzed with a microscope (Optika XDS-3FL4 Ponteranica, BG, Italy), and images were captured by a high-resolution digital camera (Nikon Digital Sight DS-U1). The expression of zonula occludens (ZO-1) and occludin was measured as relative fluorescence units (RFU) fold change vs. vehicle groups.

Results are expressed as the mean ± standard error (SEM) of n sets of experiments in triplicate (see figure legends). Statistical analyses were performed using one-way analysis of variance, and multiple comparisons were performed using a Bonferroni post hoc test. *p < 0.05, **p < 0.01, and ***p < 0.001 were considered to indicate a statistically significant difference vs. control group, and °p < 0.05, ^oo p < 0.01, and ^ooo p < 0.001 were considered to indicate a statistically significant difference vs. TcdA group.

We first evaluated in vitro the ability of pNAPE-LP engineered strains to release PEA in the bacterial supernatant under the boost of the ultralow dose of exogenous palmitate and tested the optimal palmitate concentrations to use in our in vivo experiments. Our results demonstrated that NAPE-PLD protein expression increased in a time-dependent manner in pNAPE-LP bacteria, following culture medium supplementation with 0.000003–0.0003 μg/ml of palmitate, reaching an expression peak between 6 and 12 h and a plateau at 12 h (+89,000% vs. pLP) (Figures 1A,B), while NAPE-PLD was not detected in native pLP at the same time intervals. PEA concentrations significantly increased in the supernatant of pNAPE-LP, mirroring NAPE-PLD expression in a time and palmitate-dependent manner. The PEA level peaked at 12 h (+27900% vs. pLP) and reached a plateau concentration at the same time interval (Figures 1A,B). As anticipated, PEA levels were undetectable in pLP, even in the presence of the highest palmitate concentrations (0.0003 μg/l) (Figures 1A, B).

At sacrifice, NAPE-PLD protein expression and PEA release were also evaluated in mice colonic tissues from the different experimental groups. Our data show that TcdA challenge, per se, increased NAPE-PLD expression and PEA release (+336% and +400%, respectively, vs. vehicle), while the treatment with native pLP + palmitate 0.0003 μg/ml led to a further but not significant increase in NAPE-PLD expression and PEA release (+21 and +8.95%, respectively, vs. TcdA group).

Conversely, the treatment with pNAPE-LP + palmitate 0.0003 μg/ml resulted in a significant +85 and +72% relative increase of NAPE-PLD expression vs. TcdA and pLP + palmitate 0.0003 μg/ml treated mice, respectively. In line with NAPE-PLD expression, PEA concentrations in colonic specimens from pNAPE-LP + palmitate 0.0003 μg/ml treated mice were increased up to +1233% vs. vehicle and +150% vs. pLP + palmitate 0.0003 μg/ml, respectively, treated groups (Figures 2A,C,E).

The TcdA challenge induced a severe mucosal damage evaluated at histopathological analysis performed by H and E (Figures 3A,C) in WT mice (+350% vs. vehicle). Mucosal inflammation was featured by a markedly increased macrophage density in the colonic mucosa, as per immunohistochemical quantification of MAC387 positive cells (+450% vs. vehicle) and by increased neutrophils infiltration, indirectly confirmed by the increased MPO activity (+633% vs. vehicle) (Figures 3E,G,H). A negligible and not significant improvement in terms of histological score (−6%), relative MAC387 density (−5%), and MPO activity (−9%) was observed in mice treated with native pLP + palmitate 0.0003 μg/ml. On the contrary, pNAPE-LP + palmitate 0.0003 μg/ml administration significantly improved the histological damage score (−53%), with a consequent reduction of MAC387 + cell count (−70.4%) and a significant reduction in MPO levels (−82%) compared to the TcdA group in WT mice.

The expression of pro-inflammatory signaling molecules and their release were evaluated in colonic tissue homogenates and plasma samples, respectively. Our results demonstrated that the intrarectal TcdA challenge caused a marked increase in the protein levels of TLR-4 (881%), phospho-p38 MAPK (550%), HIF1α (+489%), and of the markers of NF-κB activation p50 (433%) and p65 (230%), vs. vehicle in C57BL/6 J mice. Immunoblot analysis also revealed a massive decrease of RhoA-GTPase protein expression in TcdA vs. vehicle group (−87%) (Figures 4A,C). In parallel, plasmatic levels of IL-6 (+740%), NO (+245%), and VEGF (458%) were significantly increased in TcdA as compared to vehicle (Figure 4E).

pLP + palmitate 0.0003 μg/ml coadministration failed to improve the above-described parameters, as we observed a not significant variation in TLR-4 (−10%), phospho-p38 MAPK (+6%), HIF-1α (+5%) p50 (−4.0%), p65 (-10%), and RhoA-GTPase (+9.5%) protein expression. Similarly, plasmatic pro-inflammatory mediators such as IL-6 (+5.6%), NO (−10%), and VEGF (−11%) were not significantly improved by native pLP + palmitate 0.0003 μg/ml vs. TcdA group.

Conversely, in the group of WT mice treated with pNAPE-LP + palmitate 0.0003 μg/ml, we observed a reduced expression of pro-inflammatory signaling molecules with a significant decrease of TLR-4 (−71%), phospho-p38 MAPK (−72%), HIF-1α (−53%), p50 (−74%), and p65 (−60%) expression as compared to the TcdA group. In line with this, pNAPE-LP + palmitate 0.0003 μg/ml also caused a significant recovery of RhoA-GTPase expression (+649%) and a significant inhibition of the systemic release of IL-6 (−86%), NO (−59%), and VEGF (−71%).

As a consequence of TcdA enterotoxicity, immunofluorescence analysis revealed a significant depletion of both ZO-1 and occludin protein expression, key factors regulating colonic mucosa integrity, as demonstrated by a severe loss of relative fluorescent units compared to vehicle.

Specifically, TcdA exposure caused a significant decrease of ZO-1 and occludin in C57BL/6 J mice treated with TcdA (-78% and -77%, respectively vs. vehicle) (Figure 5A). A one-week course of pLP and palmitate did not improve both ZO-1 (+18%) and occludin (-16%) expression vs. TcdA group (Figure 5A), while pNAPE-LP + palmitate 0.0003 μg/ml significantly improved the tight junction protein expression, with a relative increase in fluorescence intensity for both ZO-1 (+304%) and occludin (160%) in WT mice (Figures 5C).

In line with the findings in WT animals, the TcdA challenge caused a significant upregulation of colonic NAPE-PLD expression (+300%) and PEA release (+133%) also in PPARα KO mice. The treatment with pNAPE-LP + palmitate 0.0003 μg/ml caused a significant relative increase in NAPE-PLD tissue expression and PEA release in PPARα KO mice, similarly to what observed in WT animals (+168 and +207%, respectively, compared to TcdA-treated mice (Figures 2B,D,F)).

Nonetheless, despite the increase in PEA tissue levels, the treatment with pNAPE-LP + palmitate 0.0003 μg/ml failed to display all the aforementioned protective effects on the histological damage score and pro-inflammatory markers and tight junction proteins’ expression in this murine model. Indeed, under the same experimental conditions, pNAPE-LP + palmitate 0.0003 μg/ml administration failed to improve the histological damage score (−5%), macrophage density count (+1.5%), and MPO level quantification (+5%) (Figures 3B,F,D,H,J), in spite of the increased tissue production of PEA. Conversely to WT animals, no significant changes were detected neither in the expression of TLR-4 (+13%), RhoAGTPase (+3.7%), phospho-p38 MAPK (-3.7%), HIF-1α (+1.8%), p50 (-6.5%), and p65 (-1.5%) (Figures 4B,D) nor in the plasmatic levels of IL-6 (+3.4%), NO (-7%), and VEGF (+5%) in pNAPE-LP + palmitate 0.0003 μg/ml treated PPARα KO mice vs. the respective TcdA group (Figure 4F), further confirming the role of PPARα receptors in mediating PEA effects. Finally, TcdA exposure caused a significant decrease of ZO-1 and occludin in PPARα KO mice treated with TcdA (−78 and −77%, respectively vs. vehicle), but once again, the rescue of ZO-1 and occludin observed in WT animals treated with pNAPE-LP + palmitate 0.0003 μg/ml appeared to be mediated by PPARα receptors, since both ZO-1 (-0.89%) and occludin (+18%) signal intensity were unmodified by pNAPE-LP + palmitate 0.0003 μg/ml treatment in PPARα KO mice (Figures 5B,D). Taken together, these results suggest the crucial importance of PPARα receptors in mediating the effects of the engineered probiotic pNAPE-LP in our experimental conditions.