Evaluation of bile salt hydrolase inhibitor efficacy for modulating host bile profile and physiology using a chicken model system

Seven day old chicks were randomly allocated into four groups (n = 10/group). Each group received none (50% propylene glycol control solution) or one of the BSH inhibitors via oral gavage (once a day) for 21 consecutive days. All of the chickens exhibited normal growth behavior and no mortality occurred during the 28 days of the experimental period. No weight loss was evident for any treatment administered relative to control animals (Table 1). In general, oral administration of each BSH inhibitor consistently enhanced overall body weight (BW) and actual BW gain at the different time-points and by 24 days of age (Table 1). However, despite these trends the differences in BW and BW gain were not statistically significant (P > 0.05).

Food intake was determined for each group throughout the duration of this study and the feed conversion ratio (FCR), an indicator for feeding efficiency, was recorded (Table 1). Oral administration of each BSH inhibitor consistently led to lower FCR than that in the control group during feeding periods examined: day 14–17, day 17–21, and day 21–24 despite a lack of significant difference (P > 0.05) (Table 1). However, FCR was significantly different during the initial period, for BSH inhibitor-treated groups (1.82, 1.85, and 1.75 for riboflavin, CAPE, and carnosic acid, respectively) than that of the control group (2.11).

When individual body weight and weight gain were examined over time following the different treatments, the animals could be stratified into two cohorts relative to the untreated group by principal component analysis (PCA analysis as shown in Fig. 1S). For each treatment non-responders (NRS) clustered with control, untreated animals while responders (RS) diverged and clustered together. On this basis, a threshold value of 1160 g/bird for NRS was calculated and any animal with a final weight above this value were classed as treatment RS. The actual division of animals on this basis for the different treatments are as follow: CAPE: RS n = 5, NRS n = 5; Riboflavin: RS n = 4, NRS n = 6; Carnosic acid: RS n = 4, NRS n = 6. Significant increases in body weight among RS was evident at 14 days for carnosic acid (P > 0.05) and at 17 days for both riboflavin and CAPE (P > 0.05) Fig. 1a–c. The RS body weight continued to increase significantly for the duration of the trial. The inhibitor potency to improve body mass among RS is as follows: riboflavin > CAPE > carnosic acid. When actual weight gain is measured (Fig. 1d–f), the potency varied in the order of riboflavin > carnosic acid > CAPE among RS. Accurate assessment of feed efficiency by RS and NS could not be performed since food intake was assessed collectively or per cage.

The collective effect of BSH (and its inhibitors) on bile acid signatures can feedback to alter BA synthesis, BA ileal uptake, host signaling locally and systemically, as well as microbial populations. In order to evaluate intestinal BSH activity and its inhibition, we examined both the local ileal and systemic (blood) BA composition in the presence and absence of dietary BSH inhibitors. Both the ileal contents and blood samples were collected from each individual chicken (n = 10) at end point, day 28, whether they were RS or NRS. Extracted samples were subjected to BA analysis following UPLC TMS based targeted analysis of 31 different BA moieties across the BA classes (Fig. 2a). For all samples, quality control spiking of each sample with internal standard deuterated cholic acid (D4 CA) prior to extraction allowed extraction efficiency assessment and BA level normalization. D4 CA recoveries were in an acceptable range indicating efficient and accurate sample extraction for ileal samples (Fig. 2b).

Individual bile acid moieties were quantified against their corresponding standard curve, the data for each moiety is recorded along with any statistical significance relative to the control animals (Tables 2 and 3, respectively). Interestingly, similar to rodents, control chicken ileal bile acid was dominated by tauro conjugated BA representing approximately 30% of total ileal BA while glyco conjugated BA was less well represented at just 0.3%. Primary BA was 100 fold higher than glyco conjugated BA and secondary BA represented 0.1% of the total BAs featured. In collective comparisons with the control groups, BA moieties secondary BAs and glyco conjugated BAs appeared significantly reduced for all three BSH inhibitors treatments and for CAPE and riboflavin treated groups respectively. With RS and NRS stratification these effects were maintained. In addition, TLCA emerged as significantly reduced in RS animals with BSH inhibitor riboflavin.

Heatplots showed enrichment for secondary BA in the majority of control animals relative to BSH inhibitor treated animals (Figs. 3 and 4). Furthermore, BSH inhibitor activity enriched for conjugated BA (tauro and glyco) for the majority of treated animals. Further analysis indicated that while there was a trend towards higher levels of total BA on treated with either riboflavin or carnosic acid they did not reach significance (P > 0.05) relative to untreated animals (Fig. 3b). These trends were due to increased levels of tauro and glyco conjugated BAs (Fig. 3f,g), consistent with this, significantly lower levels of secondary BAs were detected for the BSH inhibitor treated groups (Fig. 3d). In addition, all of the BSH inhibitor treatment groups tended towards reduction for taurine relative to untreated control animals again suggesting that BSH inhibitors are active in vivo. Subsequently, individual secondary BAs were examined, to confirm that oral administration of each BSH inhibitor led to significantly reduced levels of lithocholic acid (LCA) (P < 0.05; Fig. 4a). Riboflavin treatment showed the greatest reduction in LCA (Fig. 4a) and it correlated with a trend towards higher levels of taurolithocholic acid (T-LCA) (Table 2). The level of muricholic acid in the intestine was significantly lowered by all of the treatments (Fig. 4c). Significantly lower levels of glycolithocholic acid, glycodeoxycholic acid, and glycoursodeoxycholic acid (Fig. 4b,d,f), all of which are conjugated secondary bile acids, were evident in the riboflavin treated group with similar trends evident where carnosic acid was applied. CAPE treatment alone, significantly reduced levels of glycochenodeoxycholic acid (Fig. 4e). Taken together these data support effective gastrointestinal BSH inhibition in treated animals leading to less microbial deconjugation and dehydroxylation to secondary BAs.

BAs actively transported into enterohepatic circulation are detectible in the blood. Table 4 summarizes the levels of different BA moieties detected in the plasma of untreated and BSH inhibitor treated animals. Only GHCA was significantly reduced in circulation by BSH inhibitors, whether each group was stratified as RS or not (Table 5). Plasma BAs varied among individual animals for all of the treatment groups (Fig. 5a). However, BA categories: total BAs, total conjugated BAs and their subdivided glyco and tauro conjugates, hydrophobic and hydrophilic BAs as well as secondary BAs showed no significant differences or even trends towards differences in their overall content (Fig. 5a–g). Individual BA moieties were next examined (Fig. 6a) confirming that two bile acids LCA (Fig. 6b) and glyco hyocholic acid (G-HCA) (Fig. 6c) were significantly altered in response to CAPE and riboflavin BSH inhibitor. Riboflavin treatment elevated LCA levels significantly (P = 0.0220) however trends towards increased LCA were evident for the remaining BSH inhibitors (Fig. 6b). No significant differences were evident for any other BA moieties in circulation.

BA moieties can differentially alter gene expression through their interactions with bile acid receptors (BARS) including farnesoid X receptor (FXR), Takeda G protein couple receptor (TGR5), Vitamin D receptor (VDR) Pregnane X receptor (PXR) and Liver X receptor (LXR). The BA signatures generated could predict potential receptor interaction and signaling both locally in the ileum and in circulation to other tissues (Fig. 7). In the ileal, the combined levels of FXR antagonist (UDCA, TUDCA, GUDCA, T-MCA) was similar between the different groups. Interestingly, for all groups, the concentration FXR agonist (CA, CDCA, DCA, α-MCA and β- MCA) were within the same range but the level exceeded antagonist concentrations ca. 100 fold. In contrast, TGR5 agonists (LCA, DCA) were significantly reduced for all BSH inhibitor treatments compared to control animals. The levels of BA moieties ligands for VDR (LCA, CDCA, DCA, CA), PXR (LCA, DCA, CA) and LXR (CA, LCA, DCA, CDCA, HDCA) remained unchanged (Fig. 7). BAs in circulation are relatively low compared to ileal BA levels (Fig. 7b), here no significant differences were detected between the treatment groups on the basis of NR or GPCR agonism. Taken together, this work indicates that signals that may be elicited at the level of the gut do not translate systemically.

In order to assess gene transcription changes in the gut and in the liver one BSH inhibitor, carnosic acid (n = 4) was compared relative to untreated animals (n = 4), in order to assess gene transcription changes in the gut and in the liver by RNA-Seq. For BSH inhibitor treatment assessment they represented 1 RS and 3 NRS, based on body weight as described above. The high quality RNAs (reflected by RIN, Table) were extracted from tissue samples and applied for mRNA-Seq library construction and sequencing (Illumina Hi-Seq platform (150-cycle paired end). More than 685 million raw reads in pairs were obtained they ranged from 9.2 to 162.7 M raw reads per sample (Table). The mapping rate of RNA samples from liver samples and ileum samples ranged from 61.2% to 88.2%. S1 S1

In response to carnosic acid, 607 DEGs (495 up-regulated and 112 down-regulated) were evident from liver tissue analysis while 215 DEGs (116 up-regulated and 99 down-regulated) were detected in the ileum. As expected, many DEGs are consistent with the reported function of carnosic acid, as an antioxidant and anti-inflammatory agent, partly demonstrating the success of RNA-Seq analysis. For example, of more than 600 liver DEGs, approximately 80 DEGs were directly related to immunity and inflammatory response (data not shown). A notable panel of DEGs involved in bile biosynthesis, lipid metabolism and relevant physiology were also observed. The representative DEGs in response to carnosic acid treatment are described in Table 6.

In addition to those involved genes in immune response (CD3E, OASL, BLB1), antioxidant and anti-inflammatory roles (glutamine synthetase, glutathione S-transferase), highly upregulated genes in liver were associated with BA metabolism (AKR1D1, PHGDH) and secondary metabolites biosynthesis (P450 CYP2C23a, CYP3A80, CYP4B7), lipid metabolism (APOD, LIPI), energy metabolism (AK1), glucose metabolism (PRKAG3), and insulin regulation (IGFBP5) (Table 6). Interestingly, the highest fold increases in liver gene transcription were detected for oligoadenylate synthetase (>10 fold). Oligoadenylate synthase (OAS) proteins sense exogenous nucleic acid and initiate antiviral pathways and for major histocompatibility complex (>5 fold) initiating acquired immune system recognition. Growth factor binding protein (>3.8 fold increased) is a carrier of insulin–like growth factor 1 (IGF-1); IGF-1 can exert positive or negative effects, depending on its levels and distribution to include apoptosis induction, or alteratively, cancerogenic cell proliferation. Negative DEGs in the liver include avium defensins 2, 6 and 1, their expression was reduced 83, 39 and 22 fold respectively. Furthermore, lysosomal innate immunity molecule cathelicidin was reduced 24 fold by carnosic acid treatment.

Ileal DEGs were not as profoundly altered when compared to those of the liver and changes were in the region of 4–5 fold were recorded. Here, upregulation of antioxidant and anti-inflammatory coding genes and transcription factors were evident, including oligoadenylate synthetase (5.5 fold) eomesodermin (3.1 fold) and glutathione S transferase (2.44 fold). Interestingly cell division checkpoint G0/G1 switch was up regulated (4.17 fold). Regarding glucose metabolism, the fasting and energy reserve metabolism activator glucagon receptor was increased (2.8 fold).

Confirmation and validation of RNA-Seq analysis was performed by examining a number of DEGs associated with BA and lipid metabolism in the liver or ileum, by qRT-PCR analysis (Table 6). The qRT-PCR results are consistent with RNA-Seq analysis (Table 6), validating the expression of these genes and further demonstrating the integrity of RNA-Seq analysis.

In order to compare functional enrichment between the carnosic acid and non-treated groups a total of 607 DEGs (495 up-regulated and 112 down-regulated) from liver samples and 215 DEGs (116 up-regulated and 99 down-regulated) from the ileum were subsequently applied for GO term enrichment analysis (Figs. 8 and 9). The majority of alterations were associated with metabolic processes with 22.7% and 23.6% of up-regulated genes in liver and ileum, respectively and 14.6% and 15.8% down-regulated genes in liver and ileum, respectively. This was followed by cellular processes with 26.2% and 25.5% up-regulated genes in liver and ileum, respectively and 29.3% and 23.3% down-regulated genes in liver and ileum, respectively and then biological regulation with10.8% and 10.4% up-regulated genes in liver and ileum, respectively and 10.8% and 14.3% down-regulated genes in liver and ileum, respectively (Figs. 8 and 9).

The effects of all three BSH inhibitors relative to non treated controls were assessed for metabolic indicators including circulating lipid associated parameters (cholesterol, triglycerides, VLDL) and for glucose (Table 7). While BSH inhibitor treatment did not induce significant changes in the levels of circulating glucose (P = 0.08) glucose levels tended to be lower for all three BSH inhibitor treated groups. In contrast, total cholesterol levels appeared elevated in the treated groups (P = 0.33) a trend that was also consistent for VLDL (P = 0.18). Circulating triglyceride levels remained normal and equivalent among these different groups.

All of the methods described below were performed in accordance with the relevant guidelines and regulations. Chicken experiment was reviewed and approved by the University of Tennessee Institutional Animal Care and Use Committee (IACUC No. 2187). Forty newly hatched male broilers from Hubbard LLC (Pikeville, TN, USA) were raised in the Joseph E. Johnson Animal Research and Teaching Unit (JARTU, Knoxville, TN, USA). 7 days old chicks were randomly allocated into four groups (n = 10/group). They received one of caffeic acid phenethylester (Wuhan Yuancheng Gongchuang Tech, Wuhan, Hubei, China), riboflavin (Bulk Supplements, Henderson, NV, USA), carnosic acid (Purify, Chengdu, Sichuan, China) at 25 mg/kg, or simply control solution (50% propylene glycol), via oral gavage once a day for 21 consecutive days. The dosage was adjusted accordingly based on average chicken BW determined on Day 7, 10, 14, 17, 21, and 24. Water and feed were available ad libitum during the project. On Day 7, 10, 14, 17, 21, 24, and 28, BW and feed consumption were recorded. On Day 28, blood samples were collected from the brachial vein of each bird. Following blood sampling, all birds were euthanized by CO₂ asphyxiation and dissected for sampling. The ileal contents were collected from all birds, transferred to sterile cryogenic vials, immediately frozen on dry ice and stored at −80°C prior to shipment for metabolomic analysis. Approximately 5 cm of ileum from each chicken was collected and rinsed with saline (0.9% sodium chloride, w/v); the cleaned ileum and a sample of liver tissue from each bird were immediately frozen in liquid nitrogen and stored in cryogenic vials at −80°C prior to extraction for RNA-Seq analysis. The blood samples were centrifuged at 4,000 g for 15 min; subsequently, the supernatant (serum) was aliquoted in sterile cryogenic vials and stored in −80 °C freezer prior to further analysis.

The concentrations of glucose, triglycerides, total cholesterol, and very low-density lipoprotein (VLDL) in each serum sample were determined by using following commercial kits: Glucose Colorimetric Assay Kit and Triglyceride Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI, USA; for glucose and triglyceride), EnzyChrom Cholesterol Assay Kit (Bioassay System, Hayward, CA, USA; for total cholesterol), Chicken very low density lipoprotein, VLDL ELISA kit (Novatein Biosciences, Woburn, MA, USA; for VLDL).

Serum and ileal contents samples from individual chickens were packaged on dry ice and shipped for ‘blinded’ metabolomics measurements and statistical analysis. The extracted BAs from each sample were analyzed by Ultra Performance Liquid Chromatography-Tandem Mass Spectrometry (UPLC-TMS). The procedures for BA extraction, characterization and quantification were detailed in a previous publication⁸.

For each liver or ileum tissue sample (n = 4 for control and for treated groups for each tissue type) total RNA was extracted using Qiagen RNeasy mini kit (Qiagen Inc., Valencia, CA, USA) following the manufacturer’s instructions. The RNA integrity number (RIN), concentration and the 28 S/18 S ratio of RNA were then evaluated using Agilent RNA 600 Nano kit with Agilent 2100 Bioanalyzer (genomics Hub, University of Tennessee Institute of Agriculture, USA). The total RNA samples were stored at −80 °C prior to analysis. Extracted RNA was packaged on dry ice and shipped to the DNA Facility at Iowa State University (Ames, IA, USA) for mRNA-Seq library construction and RNA-Seq. Overall, sixteen sequencing libraries were constructed from RNA samples and subjected to sequencing using Illumina HiSeq 3000 system (150-cycle paired-end high-output sequencing).

The RNA-Seq differential gene expression analysis was performed in the Office of Advanced Research Computing (Rutgers University, NJ, USA). The raw reads were quality trimmed using Trimmomatic-0.33 with leading and trailing Q score 25, minimum length 25 bp, and minimum length 50 bp. The cleaned reads were mapped to Gallus_gallus genome - 5.0 version, using Tophat v.2.0.13. The reference genome sequence and annotation files were downloaded from ENSEMBLE, release.87 (Gallus_gallus.Gallus_gallus-5.0.dna.toplevel.fa and Gallus_gallus.Gallus_gallus-5.0.87.gtf). The aligned read counts were obtained using htseq-count as part of the package HTSeq-0.6.1.The Bioconductor package edgeR_3.1.2 with limma_3.22.7 was used to perform the differential gene expression analysis in R Studio v 1.0.143. The differentially expressed genes (DEGs) between the control group and carnosic acid group were identified at combined cut-offs with p value <0.05 and fold change>1.5. The DEGs were uploaded to DAVID Bioinformatics Resources (https://david.ncifcrf.gov/↗) online bioinformatics analysis system for functional annotation and canonical pathway analyses.

After gene annotation with DAVID-BR, Gene Ontology (GO) analysis (http://www.geneontology.org/page/go-enrichment-analysis↗) was applied to investigate the underlying functions (i.e., biological process, molecular function and cellular component). The P value denoted the significant of the GO term enrichment among differentially expressed genes (Adj P-value < 0.05).

Key PCR primers applied to this study are listed in Table 8. Extracted RNA (200 ng) was used for cDNA synthesis using Thermo Scientific RevertAid First Strand cDNA synthesis kit (Cat#: K1621) according to the manufacturer instructions. The qPCR assays were performed using iTaq universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA). Briefly, cDNA was diluted 10-fold, and 1 μL of each diluted sample was added to a 10-μL reaction mix containing 50 ng of forward and reverse primers of specific target gene. The β-actin house-keeping gene was used as an internal reference to normalize target gene transcript levels⁵⁴. Reactions were conducted in triplicate. The threshold cycle (C_T) values for genes of interest were normalized to an average C_T value of the house-keeping genes and the relative expression of replicate was calculated by applying the 2^-ΔΔCt method⁵⁵.

Chicken BW, BW gain, FCR, the levels of serum biomarkers (glucose, total cholesterol, triglycerides, VLDL) were compared among treatments using mixed model analysis of variance procedure. Analyses were conducted with SAS v9.4 (SAS Institute, Cary, NC, USA) and least squares means were compared at the significance level of P less than 0.05. BA analysis was performed using Waters MassLynx Software V4.2 SCN943 (Waters Corporation, USA). Prism 7.0 (GraphPad Software Inc., La Jolla, CA) was applied for graphic construction. For heatplot generation data was plotted using RStudio software version 1.1.453. Statistical significance for BA samples was calculated using using Kruskal wallis: $P < 0.05; $$P < 0.01; $$$P < 0.001 followed by Mann Whitney t-Test (2-tailed): *P < 0.05; **P < 0.01; ***P < 0.001.

RNA-Seq data generated for this study have been deposited in the National Center for Biotechnology and Information (NCBI) Sequence Read Archive (SRA) with accession number SRP149780.