Microbial metabolite deoxycholic acid controls Clostridium perfringens-induced chicken necrotic enteritis through attenuating inflammatory cyclooxygenase signaling

Necrotic enteritis (NE) caused by Clostridium perfringens infection has reemerged as a prevalent poultry disease worldwide due to reduced usage of prophylactic antibiotics under consumer preferences and regulatory pressures. The lack of alternative antimicrobial strategies to control this disease is mainly due to limited insight into the relationship between NE pathogenesis, microbiome, and host responses. Here we showed that the microbial metabolic byproduct of secondary bile acid deoxycholic acid (DCA), at as low as 50 µM, inhibited 82.8% of C. perfringens growth in Tryptic Soy Broth (P < 0.05). Sequential Eimeria maxima and C. perfringens challenges significantly induced NE, severe intestinal inflammation, and body weight (BW) loss in broiler chickens. These negative effects were diminished (P < 0.05) by 1.5 g/kg DCA diet. At the cellular level, DCA alleviated NE-associated ileal epithelial death and significantly reduced lamina propria cell apoptosis. Interestingly, DCA reduced C. perfringens invasion into ileum (P < 0.05) without altering the bacterial ileal luminal colonization. Molecular analysis showed that DCA significantly reduced inflammatory mediators of Infγ, Litaf, Il1β, and Mmp9 mRNA accumulation in ileal tissue. Mechanism studies revealed that C. perfringens induced (P < 0.05) elevated expression of inflammatory mediators of Infγ, Litaf, and Ptgs2 (Cyclooxygenases-2 (COX-2) gene) in chicken splenocytes. Inhibiting the COX signaling by aspirin significantly attenuated INFγ-induced inflammatory response in the splenocytes. Consistent with the in vitro assay, chickens fed 0.12 g/kg aspirin diet protected the birds against NE-induced BW loss, ileal inflammation, and intestinal cell apoptosis. In conclusion, microbial metabolic product DCA prevents NE-induced BW loss and ileal inflammation through attenuating inflammatory response. These novel findings of microbiome protecting birds against NE provide new options on developing next generation antimicrobial alternatives against NE.

thioglycollate overnight for the NE challenge study, and was serially diluted and plated on Tryptic Soy Agar plus sodium thioglycolate for enumerating CFU. In previous experiments, birds infected with this aliquot of C. perfringens alone didn't show any signs of NE and had comparable body weight gain to noninfected birds (data not showed). In current experiments, birds were infected with 20,000 sporulated E. maxima oocytes/bird at 18 days of age and 10 9 CFU C. perfringens/bird at 23 and 24 days of age. Chicken body weight and feed intake were individually measured at 0, 18, 23, and 26 days of age. Bird health status was monitored daily after the pathogen infection. Four birds with average BW of the treatment group were sacrificed at 23 and 26 days of age. Because gross necrotic lesion was often observed in upper ileum in this NE model, the ileal tissue and digesta samples from all sacrificed birds were collected for RNA and DNA analysis. The ileal tissue of ~8 cm length was also Swiss-rolled for H&E staining and histopathology analysis. Images were acquired using a Nikon TS2 fluorescent microscope. Ileal inflammation was scored blindly using the H&E Swiss-roll slides (4 slides (birds)/treatment). Briefly, each Swiss-roll slide was divided into 4 areas and was scored. Total histopathological scores were then calculated by adding the four area scores. The following score scales used were based on previous ileitis 45, 46 and colitis scoring systems 47 : score 0: no inflammation, villi and crypt intact; score 1: small number infiltration cells in laminar propria of villi and crypts or villi minimally shortened; score 2: more extensive infiltration cells in laminar propria of villi and crypts, villi shortened >1/4 and edema, or crypt hyperplasia; score 3: pronounced infiltration cells in laminar propria of villi, crypts, submucosa, and muscularis, villi shortened >1/2 and edema, or crypt hyperplasia and regeneration; and score 4: necrosis, villus diffuse, ulcers, crypt abscesses, or transmural inflammation (may extend to serosa).
terminal deoxynucleotidyl transferase dUtp nick end labeling (tUneL) assay. Cell apoptosis in intestinal tissue was visualized using TUNEL assay similar to described before 48 . Briefly, ileal tissue slides were deparaffinized with xylene bath for 3 times and then rehydrated with 100%, 95%, and 70% ethanol. The tissue was then incubated with TUNEL solution (5 µM Fluorescein-12-dUTP (Enzo Life Sciences), 10 µM dATP, 1 mM pH 7.6 Tris-HCl, 0.1 mM EDTA, 1U TdT enzyme (Promega)) at 37 °C for 90 min. The slides were counter-stained with DAPI for nucleus visualization. The fluorescent green apoptosis cells were evaluated and imaged using a Nikon TS2 fluorescent microscopy. The green dots in representative 3 areas per slide were counted as apoptosis cells and blue dots of nuclei were counted as total cells using ImageJ 49 particle analysis and its plugin of Nikon ND2 reader. The results were showed as apoptosis cells per 1,000 total intestinal cells.
C. perfringens-induced inflammatory response using primary splenocytes. Splenocytes were isolated similarly to described previously 53 . Briefly, chicken spleen was resected, homogenized into splenocytes using frosted glass slides, and pooled together in RPMI 1640 medium supplemented with 2% fetal bovine serum, 2 mM L-glutamine, 50 µM 2-mercaptoethanol. After lysed the red blood cells, the collected cells were plated at 2 × 10 6 cells/well in 6-well plates. The cells were pre-treated with 1.2 mM aspirin for 45 min. Cells were then challenged with murine INFγ (1 μg/ml, Pepro Tech), chicken INFγ (1 μg/ml, IBI Scientific), or C. perfringens (multiplicity of infection 100). The cells were lysed in TRIzol (Invitrogen) for RNA isolation after 2 or 4 hours of cytokines or C. perfringens treatment, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ Dietary aspirin against ne. Birds fed diet supplemented with 0 and 0.12 g/kg aspirin were raised, infected, and sampled as abovementioned. Ileal histopathology images and scores were collected. To evaluate cell death, TUNEL assay was performed in the ileal slides of the birds and the apoptotic cells were quantified. The impact of aspirin on body weight gain was also measured.

Statistical analysis.
For in vitro assay of bile acids against C. perfringens growth, data were first analyzed by One-way ANOVA for significant difference and then Bonferroni's multiple comparison test using Prism 5.0 software. For other data, differences between treatments were analyzed pairwise using the nonparametric Mann-Whitney U test performed using Prism 5.0 software. The specific pairwise comparisons were showed in Results section and Figures

Results
DcA prevents C. perfringens in vitro growth. Based on previous findings and current state of knowledge, we reasoned that DCA would prevent C. perfringens growth. To test this hypothesis, in vitro inhibition experiments were conducted, in which C. perfringens was inoculated in TSB with sodium thioglycollate under anaerobic condition. The TSB was also supplemented with various concentrations of bile acids, including conjugated primary bile acid TCA, primary bile acid CA, and secondary bile acid DCA. The results showed that DCA inhibited C. perfringens growth at 0.01 (−33.8%) and 0.05 mM (−82.8%, clear broth), respectively, compared to control, while TCA (−16.4%) and CA (−8.2%) didn't prevent the bacterial growth (cloudy broth) even at 0.2 mM (Fig. 1A). We then examined whether other secondary bile acids were also bacteriostatic in TSB. Interestingly, C.

DCA prevents NE-induced intestinal inflammation in chicken ileum. To further address whether
DCA reduced coccidia E. maxima-and C. perfringens-induced NE in birds, broiler chicks were fed CA or DCA diet. Because coccidiosis and NE induce severe intestinal inflammation 7,8 , the impact of DCA on chicken NE was investigated. Upper ileum tissue was collected as Swiss-roll, processed with H&E staining, and performed histopathology analysis. Notably, E. maxima (Em) infection induced severe intestinal inflammation (ileitis) as seen by immune cell infiltration (yellow arrows) into lamina propria, crypt hyperplasia, and mild villus height shortening compared to uninfected birds ( Fig. 2A). NE birds displayed worse ileitis as seen by necrosis and fusion of villi and crypt (Green arrow), massive immune cell infiltration, and severe villus shortening. In contrast, DCA diet dramatically attenuated NE-induced ileitis and histopathology score (Fig. 2B), while CA diet also reduced NE-induced ileitis and histopathology score.

DcA attenuates ne-induced intestinal cell necrosis and apoptosis. Because inflammation
often induces cell death 54 , it was then sought to examine whether cell death was relevant in DCA attenuating NE-induced ileitis. Since it was difficult to find reliable chicken antibodies to detect apoptosis or necrosis in chicken histology slides, we first resorted to classical histological analysis under high magnification. Consistently, the epithelial nuclei (dark blue) in heathy control bird villi were distributed close to the basal membrane (at the right side of the yellow dash line, Fig. 3A lower left panel). In contrast, the nuclei in inflamed villi epithelial cells of Em and NE birds were scattered from basal to the apical membranes, indicating epithelial cell death in villi of www.nature.com/scientificreports www.nature.com/scientificreports/ those birds. As a contrast, the DCA diet prevented epithelial cell nucleus translocation to apical side, suggesting cell death reduction. To further characterize the villus cell death, TUNEL assay was used, which detects later stage of cell apoptosis. Consistent with histopathology results, coccidiosis and NE induced scattered (Em birds) or concentrated (NE birds) apoptosis cells (green dots) in villus lamina propria, while cellular apoptosis was attenuated in the DCA treatment birds (Fig. 3B). To quantify the apoptosis, ImageJ was used to count apoptosis cells (TUNEL, green) and total cells (DAPI, blue). Consistent with histopathology results, DCA reduced NE-induced cell apoptosis (Fig. 3C).  Given DCA inhibited C. perfringens growth in vitro, it was logic to reason that DCA might also reduce C. perfringens intestinal overgrowth in NE birds. To examine this possibility, C. perfringens colonization level was measured in the intestinal lumen using real-time PCR of C. perfringens 16S rDNA. Surprisingly, ileal luminal C. perfringens colonization in DCA birds was not significantly different from NE control birds (Fig. 4A), while bird histopathology was distinct between the two groups of birds. We then reasoned that the pathogen invasion into tissue was the main driving factors of NE pathogenesis, but not the pathogen luminal colonization level. To quantify the bacterial invasion, total DNA in ileal tissue was isolated and C. perfringens was measured using real time PCR. DCA attenuated more than 95% of C. perfringens invasion into ileal tissue (Fig. 4B). To have a better overview of the bacterial tissue invasion distribution, we used a fluorescence in situ hybridization (FISH) technique. We found that while C. perfringens was present deeply in the inflamed villus and crypt lamina propria of NE control birds, the bacterium was barely detectable in the ileal tissue of DCA birds (Fig. 4C). Because DCA reduced C. perfringens invasion and intestinal inflammation, the impact of DCA on various proinflammatory mediators was evaluated in ileal tissue using Real-Time PCR. C. perfringens induced inflammatory Infγ, Litaf (Tnfα), and Mmp9 mRNA accumulation in chicken ileal tissue, an effect attenuated by 51, 82, and 93%, respectively, in DCA fed chickens (Fig. 4D).
Body weight (BW) is one of important productivity parameters for meat chickens and reflects collective response during NE. The productivity results showed that DCA (solid black bar) but not CA (vertical-line bar) diet promoted bird daily BW gain during 0-18 days of age compared to birds fed control diets (open, tilted line, and dotted bars, Fig. 5). BW gain was reduced in birds infected with E. maxima (Em) (tilted line and dotted bars) during 18-23 days of age (coccidiosis phase) compared to noninfected birds (open bar). Subsequent C. perfringens infection reduced NE control birds (dotted bar) BW gain during 23-26 days of age (NE phase) compared to Em birds (tilted line bar). Consistent with histopathology analysis, DCA prevented productivity loss at coccidiosis and NE phases compared to the NE control birds. Interestingly, the primary bile acid CA diet attenuated body weight loss at NE phase but failed at coccidiosis phase compared to the NE control birds.

coX inhibitor aspirin alleviates C. perfringens-induced inflammatory response in splenocytes.
Inflammatory events shape intestinal diseases and targeting the inflammatory response attenuates disease progress such as in Inflammatory Bowel Disease 55 and campylobacteriosis 53 . To dissect how the host inflammatory www.nature.com/scientificreports www.nature.com/scientificreports/ response is involved in NE-induced ileitis, a primary chicken splenocyte cell culture system was then used similarly to previous report 47 . After isolation from chickens, the splenocytes were infected with C. perfringens (MOI 100) for 4 hours. The results showed that C. perfringens increased inflammatory mediators of Infγ, Litaf (Tnfα), Mmp9 and Ptgs2 (protein COX-2) mRNA accumulation by 1.54, 1.69, 1.72 and 8.65 folds, respectively, compared to uninfected splenocytes (Fig. 6A). Because COX-2 is an important mediator in the inflammatory response 32 , COX inhibitor aspirin was then used in C. perfringens-infected chicken splenocytes. Interestingly, aspirin failed to reduce C. perfringens-induced inflammatory gene expression (data not shown). We then reasoned that COX signaling acted on C. perfringens-induced inflammatory cytokines. Inflammatory cytokine of recombinant chicken INFγ (ch-INFγ) was then used to challenge splenocytes in the presence of aspirin. Aspirin reduced ch-INFγ-induced inflammatory gene expression of Infγ, Litaf (Tnfα), and Mmp9 by 44, 45, and 65%, respectively (Fig. 6B). Because no chicken TNFα was available, murine INFγ (mINFγ) and mTNFα were used. Consistently, aspirin also reduced murine INFγ (mINFγ)-induced inflammatory gene expression of Infγ, Litaf, and Mmp9 by 41, 27, and 45%, respectively (Fig. 7A). Similarly, aspirin reduced mTNFα-induced inflammatory gene expression of Infγ, Litaf, and Mmp9 by 49, 53, and 27%, respectively (Fig. 7B). These data indicate that C. perfringens induces inflammatory cytokines and COX-2 while blocking COX signaling by aspirin reduces inflammatory cytokine-induced responses, suggesting that aspirin poses protection potential against NE detrimental inflammatory response.  www.nature.com/scientificreports www.nature.com/scientificreports/ Aspirin attenuates ne-induced ileitis, intestinal cell apoptosis, and productivity loss. To functionally assess the protective effect of aspirin against NE-induced ileitis, broiler chickens fed with aspirin diet (ASP) were infected with E. maxima and C. perfringens as describe before. Ileal tissues were collected and histopathology analysis was performed to assess NE. Consistently, NE birds had severe ileal necrosis with immune cell infiltration and villus shortening (Fig. 8A). Notably, ASP attenuated NE-induced intestinal inflammation and histopathological score (Fig. 8A,B). At cellular level, ASP reduced NE-induced immune cell apoptosis in villus lamina propria (Fig. 8C,D). On growth performance, ASP birds grew slower compared to control diet birds during 0-18 days of age (Fig. 9). This is because aspirin inhibits all COX isoforms and COX-1 and -3 are important for intestinal homeostasis and growth. Notably, ASP attenuated NE-induced BW loss by 60% during NE phase of 23-26 days of age, while no difference between ASP and NE birds during coccidiosis phase of 18-23 days of age.  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Although NE has reemerged as a prevalent chicken disease worldwide in the antimicrobial free era 5 , the lack of comprehensive molecular mechanism insight into NE severely hinders the development of antimicrobial alternatives to control this disease 56 . Many virulence factors of Eimeria and C. perfringens are identified but few findings are effective to control NE in chicken production 9,57 , suggesting that important players/factors in NE pathogenesis were overlooked, such as microbiome and host response. We reported here that microbiota metabolic product DCA attenuated chicken NE by reducing chicken inflammatory COX signaling pathways. These new findings offer approaches for exploring novel antimicrobial alternatives to control C. perfringens-induced diseases.
It is a relatively new concept to manipulate microbiota and its metabolic products against infectious diseases. Fecal transplantation was used in chickens decades ago to prevent S. infantis infection 14 . Microbiome plays an important role in susceptibility to human C. difficile infection 58 . Anaerobe C. scindens-transformed secondary bile acids prevent C. difficile germination and growth 16 . LCA and DCA but not primary bile acid CA inhibit C. difficile vegetable growth and toxin production 16,59 . However, whether secondary bile acids prevent or treat C. difficile infection in human or animal models is still unknown. It has been recently found that orally gavaging DCA attenuated C. jejuni-induced intestinal inflammation in germ-free mice 20 . Based on the knowledge, we then reasoned that DCA might prevent E. maxima-and C. perfringens-induced chicken NE. Indeed, dietary DCA prevented NE and its associated productivity loss. The reduction of ileitis was coupled with reduced C. perfringens intestinal tissue invasion and intestinal inflammation and cell death. Intriguingly, DCA failed to reduce C. perfringens ileal luminal colonization, suggesting that the mechanism of DCA action is independent of intestinal luminal colonization exclusion and is possibly through modulating other factors such as inflammation.
At the cellular level, the intestinal tract of NE-inflicted birds displays severe small intestinal inflammation, showing massive immune cells infiltration into lamina propria, villus breakdown, and crypt hyperplasia 60,61 . Inflammation often induces cell death 54 . Normal intestinal epithelial cells have polarity 62 and their nuclei are located toward the basal membrane 63 , while stressed dying (apoptosis or necrosis) cells lose polarity 64 and their nuclei disperse from basal to apical membranes 65 . Intestinal inflammation is also critical to clear invaded microbes and to resolve inflammation, while overzealous inflammation causes more bacterial invasion and further collateral damage and inflammation 66 . Infectious bacteria often hijack the inflammatory pathways to gain survival and invasion advantage. For example, Salmonella Typhimurium induces extensive inflammation in mouse intestine and thrives on the inflammation 67 . Interestingly, S. Typhimurium infection causes immunosuppressive effect in neonatal chickens because of lymphocyte depletion 68 . Unlike S. Typhimurium infection in neonatal birds, coccidia infection induces strong immune response and intestinal inflammation in chickens 22,23 . Furthermore, NE birds suffered more intestinal inflammation compared to Em birds in current study. Consistent with the "over-inflammation" model, NE birds with severe intestinal inflammation showed extensive immune cell infiltration, increased inflammatory cytokine gene expression, and epithelial cell hyperplasia and death (dispersed nuclei and apoptosis). Conversely, DCA attenuated the intestinal inflammation and inflammatory cytokines and improved the growth performance and reduced the NE pathology. Consistent with the reasoning of anti-inflammation reducing NE, blocking inflammatory COX-2 signaling pathway by aspirin alleviated intestinal inflammation, villus apoptosis and NE-induced BW loss. These findings indicate that DCA attenuates NE through decreasing inflammatory signaling pathways.
Different strains of C. perfringens produce a variety of toxins including alpha (CPA), beta (CPB), epsilon (ETX), iota (ITX), enterotoxin (CPE), necrotic enteritis B-like toxin (NetB), and others 69 . Among them, researchers have reported that NetB but not CPA induced NE in chickens 70 . However, NE doesn't have strong association with NetB positive C. perfringens in US 71 . In our NE model, no difference of NetB gene expression in ileal digesta between healthy control and NE birds (data not shown) suggests limited role of the toxin. Generally, C. Figure 9. Aspirin attenuates NE-induced productivity loss. Cohorts of 13 broiler chicks were fed different diets and infected as in Fig. 6. Bird body weight gain was measured at 18 (13 birds/group), 23 (13 birds/group), and 26 (8 birds/group) days of age. Showed was daily periodic body weight gain. NE + ASP, NE birds fed aspirin diet. All graphs depict mean ± SEM. NS, not significant; *P < 0.01; Results are representative of 2 independent experiments.
Scientific RepoRtS | (2019) 9:14541 | https://doi.org/10.1038/s41598-019-51104-0 www.nature.com/scientificreports www.nature.com/scientificreports/ perfringens toxins induce cell death of apoptosis, necrosis, necroptosis, and pyroptosis in the presence of extracellular calcium influx 69 . Higher dietary calcium increases chicken mortality and reduces growth performance in coccidiosis-induced NE model 72 , suggesting the important role of toxin-induced and calcium-mediated cell death in NE. The reduction of intestinal cell death by DCA and aspirin suggests possible toxin involvement in our NE model. Future work on C. perfringens toxins, DCA, and inflammatory response will be conducted.
In contrast to the accumulating evidence of bile acids against infectious enteritis in recent studies 16,20 , the microbial-derived metabolites have long been implicated in chronic diseases 18 . Plasm bile acids is correlated with increased body mass index of obese patients 73 . Plasma chenodeoxycholic acid, CA and DCA concentrations are higher in type 2 diabetes patients compared to healthy subjects 74 . Bile acid promoting fatty acid digestion and absorption may play important role in the diseases. Furthermore, DCA and LCA in the colonic contents are increased in humans consuming a high fat diet 75 . The increase of the two bile acids in feces is associated with elevated incidence of colorectal cancer 76 . DCA damages genomic DNA by oxidation 77 and deficiency in base excision repair of oxidative DNA damage is linked to increased risk of intestinal tumors in mice 78 . Future work is needed to investigate how DCA induces the chronic diseases but attenuates the infectious enteritis.
Altogether, these data reveal that the microbial metabolic product secondary bile acid DCA attenuates NE, through reducing NE-induced host inflammatory response. These findings highlight the importance of elucidating the molecular relationship between infectious pathogen, microbiome, and host response. These discoveries could be applied to control NE and other intestinal diseases targeting microbiome and host inflammatory response.

Data Availability
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.