The proteomic differences and expression of fatty acid‐binding protein 6 (FABP6) associated with gastrointestinal injury in horses with oral administration of a clinical dose of phenylbutazone

Abstract

Background

Phenylbutazone (PBZ) can potentially induce gastrointestinal ulceration, and early detection of PBZ‐induced gastroenteropathy will be useful for the diagnosis, treatment, and prevention of PBZ toxicity.

Objectives

To identify putative proteins associated with equine gastric ulcer syndrome after clinical dose (4.4 mg/kg) administration of PBZ by proteomic study.

Study Design

In vivo experiments.

Methods

Proteomic analysis using LC–MS/MS compared protein expression in serum and faeces of seven PBZ‐treated horses with seven placebo‐treated controls, and a novel putative biomarker was validated via enzyme‐linked immunosorbent assay.

Results

Differentially expressed proteins (DEPs) analysis on 5298 serum annotated proteins and 3538 faecal annotated proteins using the DESeq2 were performed between the control and treatment of EGUS groups. The results showed a list of 226 and 181 significant proteins in serum and faecal samples, respectively with a p adjust value <0.05. The proteomic serum and faeces samples were integrated into STITCH to illustrate PBZ interaction with bile acid homeostasis. FABP6 was significantly increased in PBZ‐treated horses. The serum FABP6 concentration in the treatment group on Day 8 (1.80 ± 0.37 ng/mL) was higher than on Day 0 (1.15 ± 0.33 ng/mL, p = 0.01, 95% CI [−1.07, −0.25]). On Day 8, the serum FABP6 concentration in the treatment group was also higher than the control group (1.20 ± 0.48 ng/mL; p = 0.02, 95% CI [−1.10, −0.11]).

Main Limitations

Validation of all expressed proteins is a main limitation.

Conclusions

Administration of PBZ at a clinical dose of 4.4 mg/kg twice daily for 7 days may cause gastric mucosal damage. PBZ treatment increased the expression of SLC10A1 and FABP6, suggesting that early gastric mucosal injury may be linked to the bile acid pathway. Bile acids could potentially exacerbate PBZ‐induced EGUS.

1. INTRODUCTION

The use of non‐steroidal anti‐inflammatory drugs (NSAIDs) to control pain, fever, and inflammation is widespread in horses. On the cell membrane, the enzyme cyclooxygenase (COX) catalyses arachidonic acid conversion to eicosanoids, including prostacyclins, prostaglandins, and thromboxane. ¹ Prostaglandins play a key role in vasodilation, anti‐nociception, and modulating temperature set‐point in the hypothalamus, while thromboxane mediates platelet adhesion. Inhibition of COX by NSAIDs causes fewer eicosanoids and limits their functions. The therapeutic effects of COX inhibitors have been documented in a variety of equine disorders, such as osteoarthritis, laminitis, colic, endotoxaemia. ² Although the efficiency of NSAIDs is very helpful in equine practices, ulceration because of these drugs in the digestive tract is a great concern for equine practitioners. Phenylbutazone (PBZ), one of the most common drugs used in horses, has been reported as the most toxic NSAID and is associated with equine gastric ulcer syndrome (EGUS) and right dorsal colitis. ³ , ⁴ , ⁵

Previous studies have shown high doses of PBZ can induce EGUS. ⁴ , ⁶ , ⁷ Lesions associated with high‐dose PBZ toxicity include diffuse severe hyperaemia, gastric mucosal oedema, and haemorrhage, in which most lesions appear in the glandular part of the stomach ⁷ and are typical of equine glandular gastric disease. Although gastroscopy has been a gold standard technique to diagnose EGUS for decades, it is expensive and requires equipment and expertise. Tesena et al. (2019) have shown candidate marker proteins appear in EGUS after being induced with a high dose of PBZ (5.25–10 mg/kg), but these doses are not regularly used in equine practices. Therefore, this study aimed to identify putative proteins associated with EGUS in serum samples, focusing on the key functional interactions involved in developing EGUS after a clinical dose (4.4 mg/kg) administration of PBZ. This technique might be useful for early detection and its progression.

2. MATERIALS AND METHODS

2.1. Animals

Fourteen adult light‐breed horses from a research herd were included in the study, 4 mares and 10 geldings. The median ages of horses were 14 years (range 4–20 years old). Median bodyweights were 426 kg (range 366–532 kg). The horses had a history of lameness but were free of systemic diseases, especially gastrointestinal tract diseases, and had not received any NSAID within 3 weeks before the study. The horses were allowed to pasture during the day and kept in the stable during the night, fed hay (2% of bodyweight daily), and allowed ad libitum water. Feed and water were withheld for 16 and 4 h, ⁸ respectively, before the study.

2.2. Experimental design

This randomised controlled trial was designed based on a priori power analysis using G*Power to determine the required sample size for detecting a substantial difference in gastric mucosal changes. The calculation, based on an independent t‐test (power = 0.95, α = 0.05), set Cohen's d at 2.2, representing a large standardised mean difference crucial for clinical relevance, as supported by prior research in similar settings. ⁹ The analysis indicated that seven horses per group were necessary. Fourteen horses were randomly assigned to two groups: the control group received 5 mL sweet syrup per os twice daily (6 AM, 4 PM) for 7 days, while the treatment group received 5 mL sweet syrup with 4.4 mg/kg PBZ per os on the same schedule. Simple randomisation was performed via coin flip, with allocation concealed in sealed envelopes. To minimise bias, treatment administration and outcome assessment were conducted under blinding.

2.3. Gastroscopy

After routine physical examination, all horses were sedated by 5–10 μg/kg detomidine (Detomo Vet, CEVA) ¹⁰ with or without 0.02–0.04 mg/kg acepromazine (Combistress, KELA N.V.) ¹¹ depending on horse's attitude, intravenously. A nasogastric tube was introduced into the stomach, followed by a 3‐m flexible gastroscope (KARL STORZ Endoscope, KARL STORZ SE & Co.). The entire stomach was investigated, including the non‐glandular part, glandular part, and margo plicatus, by 3 veterinarians (P.T., R.V., T.C.). The glandular and non‐glandular regions were assigned severity grades using previously reported 0–4 scales. ¹² , ¹³

2.4. Sample collection and preparation

Blood and faecal samples of each horse were collected before starting the study (D0) and after giving a placebo or PBZ (D8). Two millilitres of blood samples from each horse were collected in the plain tube from the jugular vein and then centrifuged at 10,000g for 10 min to obtain the serum. Ten grams of faeces were collected from the rectum of each horse. The serum and faecal samples were kept at −80°C until further processes.

The total protein concentration in serum and faeces was determined using the Lowry assay with bovine serum albumin as a standard. ¹⁴ In‐solution digestion was used on five micrograms of protein samples. Samples were completely dissolved in 10 mM ammonium bicarbonate (AMBIC), disulphide bonds were reduced using 5 mM dithiothreitol (DTT) in 10 mM AMBIC at 60°C for 1 h, and sulfhydryl groups were alkylated at room temperature for 45 min in the dark using 15 mM Iodoacetamide (IAA) in 10 mM AMBIC. For digestion, samples were incubated at 37°C overnight with 50 ng/L sequencing grade trypsin (1:20 ratio) (Promega). Before injecting the digested samples into the LC–MS/MS, participants must be dried and protonated with 0.1% formic acid.

2.5. Liquid chromatography–tandem mass spectrometry (LC/MS–MS)

The tryptic peptide samples were injected into an Ultimate3000 Nano/Capillary LC System (Thermo Scientific) coupled to an HCTUltra LC–MS system (Bruker Daltonics Ltd) equipped with a Nano‐captive spray ion source. Briefly, 5 μL of peptide digests were enriched on a μ‐Precolumn 300 μm I.D. × 5 mm C18 Pepmap 100, 5 μm, 100 A (Thermo Scientific), separated on a 75 μm I.D. × 15 cm and packed with Acclaim PepMap RSLC C18, 2 μm, 100 Å, nanoViper (Thermo Scientific). The C18 column was enclosed in a thermostatted column oven set to 60°C. Solvents A and B, containing 0.1% formic acid in water and 0.1% formic acid in 80% acetonitrile, respectively, were supplied in the analytical column. A gradient of 5%–55% solvent B was used to elute the peptides at a constant flow rate of 0.30 μL/min for 30 min. Electrospray ionisation was carried out at 1.6 kV using the Captive Spray. Nitrogen was used as a drying gas (flow rate of about 50 L/h). Collision‐induced dissociation (CID) product ion mass spectra were obtained using nitrogen gas as the collision gas. Mass spectra (MS) and MS/MS spectra were obtained in the positive‐ion mode at 2 Hz over the range (m/z) 150–2200. The collision energy was adjusted to 10 eV as a function of the m/z value. The LC–MS analysis of each sample was done in triplicate.

2.6. Quantification and identification of proteins

Maxquant 2.5.0.0 was used to quantify the proteins in an individual sample, while the Mascot search engine was used to correlate MS/MS spectra to the Uniprot Equus caballus database, ¹⁵ , ¹⁶ which was accessed on 10 July 2024. The Maxquant's standard settings were performed: a maximum of three missed cleavages, a mass tolerance of 0.6 Daltons for the main search, trypsin as a digesting enzyme, carbamidomethylation of cysteine as a fixed modification, the oxidation of methionine as variable modifications, and peptide charge state (1+, 2+, and 3+). The visualisation and statistical analyses were conducted using the Multi Experiment Viewer (MeV) in the TM4 suite software. ¹⁷ Proteins considered as identified proteins had at least one peptide with an individual mascot score corresponding to p < 0.05. The maximum peptide intensities level in log2 was transformed to Microsoft Excel and then established in the control and PBZ‐treated groups. As a result, raw MS/MS spectra data were examined for protein sequences and PELs of the control and treatment groups. The raw MS/MS spectra data are available in ProteomeXchange: PXD059214 and JPST003523.

2.7. Comparative differentially expressed proteins (DEPs) analysis toward functional annotation

The DEPs analysis was performed between control and PBZ‐treated groups using total PELs for each protein. The significant difference in proteins was examined using the likelihood‐ratio tests (LRT) in the DESeq2 model, and log2FC was added using the lfcShrink function with ‘apeglm’ adaptive shrinking. ¹⁸ For multiple testing correction, Benjamini–Hochberg correction was used for multiple testing to define differential proteins (FDR <0.05). This method, commonly used in bioinformatics and omics research, balances the need to identify meaningful biological associations without an overly conservative approach, which might increase false negatives. ¹⁹ The DEPs were functionally annotated by searching against the KEGG database ²⁰ and assigning KEGG Orthology (KO) IDs using GhostKOALA. ²¹ After the identification of KO IDs, they were categorised into main functions, subfunctions, and pathways. Additionally, pathway and functional enrichment analysis was performed using GSEA in the R package piano based on DEPs between control and PBZ‐treated groups. Pathways and functions with a distinct up‐directional p value <0.05 were considered significantly enriched.

2.8. Identification of key proteins associated with phenylbutazone treatment

Serum and faecal proteins in the control and PBZ‐treated groups were identified and compared. Then, the results obtained from proteomic serum and faeces samples were integrated into STITCH to explore potential protein interactions and regulatory mechanisms. This integrated data allows us to explore both the known and novel roles of proteins. After analysing, DEPs from serum and faeces samples under p adjust value <0.05 were then considered. To construct the protein–protein interaction (PPI) networks, the STITCH database (www.stitch.embl.de) ²² was used to predict the protein–protein–chemicals interaction involved in equine gastric disease. The framework of proteomic characterisation contributing to DEPs analysis and functional annotation is shown in Figure 1.

FIGURE 1.

The framework of proteomic characterisation contributing to DEPs analysis and functional annotation. Created in BioRender. Prapaiwan, N. (2025) g09c063.

2.9. Enzyme‐linked immunosorbent assay (ELISA)

Serum fatty acid‐binding protein 6 (FABP6) concentrations were quantified using a horse fatty acid binding protein 6 enzyme‐linked immunosorbent assay (ELISA) kit (MBS032862, MyBioSource), following the manufacturer's protocol. This equine‐specific ELISA measures serum FABP6 levels using a quantitative sandwich immunoassay. Briefly, standards and serum samples were added to coated microplate wells. Horseradish peroxidase‐conjugated secondary antibody was then added and incubated at 37°C for 60 min. After washing, chromogen solution and tetramethylbenzidine were added, followed by a 15‐min incubation at 37°C. Sulfuric acid was used to stop the reaction, and absorbance was measured at 450 nm. A standard curve was generated to calculate serum FABP6 concentrations. The assay exhibited a detection range of 0.625–20 ng/mL with intra‐ and inter‐assay coefficients of variation below 15%.

2.10. Data analysis

Gastric ulcer scores and FABP6 results were analysed using IBM SPSS version 18.0 for Windows (NY, USA). The normal distribution of residuals from the statistical models was evaluated using the Shapiro–Wilk test. Gastric ulcer scores were expressed as median and interquartile ranges and analysed by the Mann–Whitney U test to compare the control and PBZ‐treated groups at each time point. The Wilcoxon signed‐rank test was used to analyse the gastric ulcer scoring at each time point (Days 0 and 8) within each group. FABP6 was reported as mean ± standard deviation (SD). Comparisons between the control and PBZ‐treated groups were performed using the independent t‐test. The paired t‐test was used to analyse each group's time points (Days 0 and 8). Probabilities <0.05 were considered statistically significant.

3. RESULTS

3.1. Gastroscopy

In the control group, the median grade of the non‐glandular part of the gastric mucosa was 1 (range 0–1), and the glandular part of the gastric mucosa was 0 both before and after the study period. In the PBZ‐treated group, the median grade of the non‐glandular part of the gastric mucosa was 1 of 4 (range 0–1) before the treatment, while it was 2 (range 2–3) after the treatment. The glandular part of the gastric mucosa was 0 before the treatment, whereas it was 1 (range 0–2) after the treatment. The grading scores before the treatment were significantly increased (p = 0.01; Figure 2) in the treatment group's non‐glandular part of the gastric mucosa. Gastric scores for each horse are presented in File S1.

FIGURE 2.

Gastric ulcer scores were significantly increased in non‐glandular (A; p = 0.01) and raised but not significantly different in glandular (B; p = 0.06) parts of gastric mucosa after horses were given 4.4 mg/kg PBZ for 7 days. Error bars represent the interquartile ranges.

3.2. Analysis of DEPs

Our study delved into protein analysis using 28 samples from horses, resulting in the identification of 11,432 proteins in serum samples and 7700 in faecal samples. In all studied horses, proteomic profiles in both serum and faecal samples showed minor fluctuations over time, with overall stability between time points and treatment groups. In the serum, total spectral counts in the control group slightly decreased from Day 0 (D0: 29,545; 50,472) to Day 8 (D8: 28,965; 51,052), while in the PBZ‐treatment group, lower counts decreased (D0: 29,026 → D8: 27,056) and higher counts increased (D0: 50,991 → D8: 52,961). Similarly, in faecal samples, the control group showed a decrease in lower counts (D0: 11,630 → D8: 9337) and a slight increase in higher counts (D0: 42,270 → D8: 44,563). The PBZ‐treatment group exhibited a moderate decrease in lower counts (D0: 11,519 → D8: 10,578) and a slight increase in higher counts (D0: 42,381 → D8: 43,322). These results indicate consistent proteomic stability in both serum and faecal samples, with only minor variations between treatments and time points, as presented in Table 1.

TABLE 1.

Assigned the number of proteins and total spectral counts.

Protein annotation	Serum					Faeces
	Number of proteins	Total spectral counts (spectral counts per sample)				Number of proteins	Total spectral counts (spectral counts per sample)
		Control D0	Control D8	Treatment D0	Treatment D8		Control D0	Control D8	Treatment D0	Treatment D8
Annotated proteins	5298	29,545	28,965	29,026	27,056	3538	11,630	9337	11,519	10,578
Unannotated proteins	6134	50,472	51,052	50,991	52,961	4162	42,270	44,563	42,381	43,322
Total	11,432	80,017	80,017	80,017	80,017	7700	53,900	53,900	53,900	53,900

3.3. The annotated functions protein profiles

The proteomic profiles of serum and faecal samples in all studied horses were categorised into KEGG functional groups, revealing differences in the distribution of proteins across major categories. In serum, the largest categories were Environmental Information Processing (1265) and Metabolism (1131), followed by Genetic Information Processing (987), Organismal Systems (828), Cellular Processes (706), and Human Diseases (379). Similarly, in faeces, Environmental Information Processing (863) and Metabolism (790) were the most abundant categories, followed by Genetic Information Processing (630), Organismal Systems (549), Cellular Processes (433), and Human Diseases (273). These results highlight similar functional distributions between serum and faecal proteomes, with slightly higher representation across all serum categories than faeces. The sub‐functional KEGG categorisation of proteomes in serum and faeces of PBZ‐treatment horses revealed similar patterns, with serum generally showing higher protein representation across categories. Key pathways included Signal Transduction (serum: 940; faeces: 644), Lipid Metabolism (261; 197), Carbohydrate Metabolism (236; 152), and Translation (266; 197), reflecting conserved functional roles in both sample types with minor quantitative differences (Figure 3) The details of KEGG main‐functional and sub‐functional are shown in File S2.

FIGURE 3.

Functional categories datasets of all studied horses in serum and faeces. (A) Main‐functional KEGG categories. (B) Sub‐functional KEGG categories.

3.4. Analysis of DEPs and functional annotation

To identify DEPs, we conducted DEPs analysis on all 5298 serum annotated proteins and 3538 faecal annotated proteins using the DESeq2. This analysis identified a list of 226 significant proteins in serum samples and 181 in faecal samples with a p adjust value <0.05. After performing DEP analysis, we identified 47 putative proteins expressed in the serum proteins, while 20 were expressed in faecal proteins, as detailed in Files S3 and S4. All the significant proteins demonstrated a visual representation of the various groups categorised according to the statistically significant findings pertaining to the comparative protein expression levels (PELs), as shown in Figure 4.

FIGURE 4.

Heatmaps represent the comparative expression levels of serum protein (A) and faecal protein (B) expression levels of all studied horses. The heatmap was generated from the intensity of significant proteins database for each group using the CLUE program (https://clue.io/command, accessed on 15 December 2024). bC, tC, bT, and tT represent the control group before the study period, the control group after the study period, the treatment group before the study period, and the treatment group after the study period, respectively.

3.5. FABP6 analyses by ELISA

On Day 8, the serum FABP6 concentration in the treatment group (1.80 ± 0.37 ng/mL) was higher than in the control group (1.20 ± 0.48 ng/mL; p = 0.02, 95% CI [−1.10, −0.11]) as shown in Figure 5. Also, in the treatment group, the serum FABP6 concentration on Day 8 (1.80 ± 0.37 ng/mL) was higher than on Day 0 (1.15 ± 0.33 ng/mL, p = 0.01, 95% CI [−1.07, −0.25]). No difference in serum FABP6 concentration was observed between each time point in the control group (Day 0: 1.14 ± 0.45 ng/mL, Day 8: 1.20 ± 0.48 ng/mL; p = 0.64, 95% CI [−0.34, 0.23]).

FIGURE 5.

Comparison of the mean serum FABP6 concentration between groups on Day 8. ^A,BA significant difference between groups on Day 8 (p = 0.02, 95% CI [−1.10, −0.11]).

4. DISCUSSION

This study found differences in significant proteins in serum and faeces between the control and the clinical dose of PBZ‐treated groups. After the DEPs analysis, a variety of proteins, including beta‐1,4‐galactosyltransferase 1, beta‐glucuronidase, phosphatidylinositol‐4,5‐bisphosphate 3‐kinase, N‐acetylglucosamine‐6‐phosphate deacetylase, solute carrier family 10 member 1, 3‐hydroxyisobutyryl‐CoA hydrolase (HIB‐CoA hydrolase), acylphosphatase, aquaporin 8, and acetyl‐CoA carboxylase were presented. Following the DEPs analysis, the STITCH database program was analysed to show protein expression patterns and interactions (Figure 6). Interestingly, FABP6, which stimulates gastric acid and pepsinogen secretion ²³ was expressed in the STITCH network in this study. Consequently, the expression of FABP6 was compared between the control and PBZ‐treated groups by ELISA.

FIGURE 6.

The network of comprehensive protein expression patterns, highlighting notable proteins involved in bile acid homeostasis. PBZ engaged with bile acid in Equus caballus was a computational prediction by the Stitch program (www.stitch.embl.de, accessed on 7 March 2025) Version 5.0. ACYP2, acylphosphatase‐2; B4GALT1, beta‐1,4‐galactosyltransferases; CFTR, cystic fibrosis transmembrane conductance regulator; FABP6, fatty acid‐binding protein 6; GUSB, beta‐glucuronidase; HIBCH, 3‐hydroxyisobutyryl‐CoA hydrolase; PIK3CD, phosphatidylinositol‐4,5‐bisphosphate 3‐kinase; RRAS2, related RAS viral (r‐ras) oncogene homologue 2; SLC10A1, solute carrier family 10 member 1.

The STITCH database program showed the significant protein expression and interaction in this study in the way that PBZ was involved in bile acid homeostasis (Figure 6). The network included solute carrier family 10 member 1 (SLC10A1, NTCP), bile acids (BA) and fatty acid‐binding protein 6 (FABP6). Bile acid homeostasis is widely explained elsewhere. Briefly, bile acids are initially synthesised by the rate‐limiting enzyme cholesterol 7α‐hydroxylase (CYP7A1), which converts cholesterol to 7‐hydroxylation of sterol precursors, followed by modification of the ring structure and peroxisomal β‐oxidation of the side chain. Finally, bile acids conjugate with glycine or taurine. ²⁴ , ²⁵ The conjugated bile acids are secreted into the bile canaliculi via the bile salt export pump (BSEP) and released to the gastrointestinal lumen to facilitate lipid digestion and absorption. The conjugated bile acids are actively reabsorbed at the terminal ileum by apical bile salt transporters (ASBT, SLC10A2) and attach to the ileal bile acid binding protein (FABP6). This conjugated BA complex travels to the basolateral membrane and is effluxed into the portal vein. At the same time, unconjugated bile acids are passively reabsorbed in all parts of the small intestine and large intestine. Reabsorbed bile acids are circulated along with portal blood and re‐enter the liver via sodium/bile acid co‐transporter (NTCP, SLC10A1). In hepatocytes, free bile acids are reconverted into conjugated bile acids, which are secreted into the intestine with bile and reabsorbed and newly synthesised conjugated bile acids. This cycle is called the enterohepatic circulation of bile acid. Enterohepatic circulation of bile acids is a highly efficient physiological pathway; ~95% of bile acids are recycled in this pathway. ²⁴ , ²⁵

Bile acids (BAs) do not only play an essential role in aiding lipid digestion and absorption in the digestive tract, but also they act as signalling molecules that facilitate various physiological pathways, including glucose, lipid, and energy metabolism, as well as in cancer. BAs mediate immune responses, carcinogenesis, gestation, and gastrointestinal mucosal barrier function. ²⁶ The previous study has shown that bile acids induce an increase in the gene expression of cyclooxygenase‐2 (COX‐2) via the sequential transcriptional induction of small heterodimer partner (SHP) and caudal‐related homeobox gene (CDX1), which are associated with gastritis. ²⁷ BAs are cytotoxic. Due to their detergent property, BAs can alter mucosal membrane integrity. With this effect, bile reflux can cause gastric mucosal damage. ²⁵ , ²⁸ Together with BAs, NSAIDs worsen gastric mucosal injury by increasing duodenogastric bile reflux. ²⁹ NSAID integrates into phosphatidylcholine (PC) and alters the integrity of the lipid bilayer. This action enhances the isolation of bile acid molecules from PC lipids. The exitance of both NSAID and bile acids, which are effective detergents, enhances the toxicity of the complex and contributes to NSAID‐induced injury in the upper gastrointestinal tract. ³⁰ In the enterohepatic bile circulation pathway, SLC10A2 is a rate‐limiting mediator of bile acid uptake in the intestine, while SLC10A1 controls bile acid uptake in the liver. Increasing expression of SLC10A1 and FABP6 in PBZ‐treated horses in this study presumably PBZ might be related to bile acids secretion and responsible for raised gastroscopic grading scores in both non‐glandular and glandular parts of the gastric mucosa.

An ideal biomarker should be characterised by fast release from damaged cells and rapid appearance in the blood circulation. In normal healthy controls, this marker should be in high concentration in particular tissue, but low in plasma. ³¹ Fatty acid‐binding proteins (FABP) have shown these properties in many studies. ³¹ , ³² The concentration of FABP is relatively high in tissues but low in plasma. The level of FABP is increased in the blood circulation when the particular organ is injured. ³³ This phenomenon might be due to the fact that the small molecular size of FABP (about 15 kDa) makes it rapidly distribute from the damaged tissue into vascular space. ³² The previous study has shown that heart‐type fatty acid binding protein (H‐FABP) is useful in early detection of myocardial and cardiovascular diseases in humans. ³² Furthermore, the previous study illustrates that intestinal fatty acid binding protein (I‐FABP) and liver fatty acid binding protein (L‐FABP) have been increasing in the blood circulation following abdominal trauma. This finding suggests that I‐FABP1 and L‐FABP are potential candidates for early detection and rapid screening in abdominal injury patients. ³³ FABP has also been studied as a biomarker in equine practice. One study has compared the concentration of H‐FABP and creatine kinase (CK) activity in 13 Thoroughbreds before and after racing. The finding suggests that H‐FABP is not only an earlier marker, but also a more sensitive marker of muscle injury in horses. ³¹ Another study has shown that high concentrations of I‐FABP in abdominal fluid correlated with increased protein concentrations in peritoneal fluid and non‐survival, and plasma I‐FABP concentrations correlated with the necessity for abdominal surgery. The findings suggest that I‐FABP is useful in deciding on colic surgery and predicting survival. ³⁴ In this study, the level of gastroscopic grading scores in the non‐glandular part of gastric mucosa and the level of serum FABP6 in the horses treated with a clinical dose of PBZ in 7 days was significantly increased. Therefore, FABP6 might be associated with early damage of gastric mucosa by PBZ. FABP6, which is known as intestinal bile acid‐binding protein (I‐BABP), ileal lipid‐binding protein (ILBP), and gastrotropin (GT), belongs to the FABPs family, the intracellular lipid‐binding protein (iLBP). FABP6 plays an important role in the enterohepatic circulation of BA since it has greater affinity for BA and is manageable to attach to more than one BA ligand within the hydrophobic condition. ³⁵ Previous studies have indicated that GT is a potent enterooxytin and the first member of the FABP family, which has an extracellular function. ²³ , ³⁶ The study observes GT effects on both parietal and chief cells. The findings suggest that GT uses the same transduction mechanism as cholecystokinin octapeptide (CCK8) and carbachol and does not involve the activation of adenylate cyclase. ³⁶ However, one study has shown that GT does not influence acid secretion in rats with gastric fistulas or isolated canine gastric parietal cells. ³⁷ Also, beta‐1,4‐galactosyltransferase 1 (B4GALT1) was significantly expressed in this study. B4GALT1 belongs to the family of beta‐1,4‐galactosyltransferases (B4GALTs) and is responsible for the transfer of UDP‐galactose to terminal N‐acetylglucosamine. B4GALT1 is a key inflammatory mediator that plays a critical role in the initiation and maintenance of inflammatory reactions in diseases. ³⁸ In this study, high expression of B4GALTs in PBZ‐treated horses might reflect the ongoing inflammatory process. Beta‐glucuronidase (GUSB), a lysosomal enzyme, has been reported as a good indicator for inflammatory diseases such as chronic obstructive pulmonary disease (COPD). High levels of Beta‐glucuronidase are present when the numbers of neutrophils are elevated. ³⁹ In this study, high expression of GUSB in PBZ‐treated horses might also reflect the ongoing inflammatory process. Phosphatidylinositol‐4,5‐bisphosphate 3‐kinase (PIK3) has also been reported to regulate neutrophils (PMN) responding to antigen–antibody immune complexes (IC). The previous study shows adhesion of equine PMN to immobilised IC requires β‐2 integrins and this beta 2 integrin‐dependent function in equine PMN is dependent on PI3K activity. ⁴⁰ Therefore, high expression of PI3K in this present study might also support the ongoing inflammatory process. Cystic fibrosis transmembrane conductance regulator (CFTR), an electrogenic chloride channel, regulates anion secretion in the equine intestinal tract. The previous study has shown that CFTR is located at intestinal crypts. PBZ decreases CFTR function by lowering PGE₂ activity, causing mucosal injury in the colon. ⁴¹ However, there are no reports of CFTR located in the equine stomach; therefore, further study on CFTR involved in equine gastric disease might need to be investigated.

A validation of other proteins that are found in this study might be the first limitation of this study. Although B4GALT1, GUSB, and PIK3 are involved in inflammatory processes that are possibly caused by gastric mucosal changes in this study, these proteins seem to respond to general inflammation and might not be a good candidate for early detection of gastric mucosal injury. Therefore, validation in B4GALT1, GUSB, and PIK3 is less interesting. CFTR is also not a good candidate for early detection of gastric mucosal damage since this protein is mostly found in crypt cells, not blood or faeces. For these reasons, FABP6 was selected to further investigate in this current study. Although significantly increased expression of SLC10A1 and FABP6 after given PBZ, presumably involved in the bile acid pathway, has been shown in this study, the increasing level of bile acids that can cause gastric mucosal damage should be confirmed. Moreover, further studies of the FABP6 effect on gastric secretion in horses should be investigated.

5. CONCLUSIONS

The present study illustrated that the expression of SLC10A1 and FABP6 was significantly increased and related to increased gastric ulcer scores after the administration of a clinical dose of PBZ for 7 days to horses. The findings suggest that the bile acid pathway is involved in the early changes of gastric mucosal damage. These results highlight the potential role of bile acid transporters in the pathogenesis of gastric ulcers induced by PBZ, providing insight into the molecular mechanisms underlying gastric mucosal injury and offering possible targets for therapeutic interventions in horses. Further investigation into the modulation of these pathways could enhance strategies for preventing or mitigating gastric ulcers in equine clinical practice.

FUNDING INFORMATION

This research project has been funded by Mahidol University (Fundamental Fund: fiscal year 2024 by National Science Research and Innovation Fund (NSRF)).

CONFLICT OF INTEREST STATEMENT

The authors have declared no conflicting interests.

AUTHOR CONTRIBUTIONS

Ruethaiwan Vinijkumthorn: Conceptualization; investigation; funding acquisition; writing – original draft; methodology; validation; visualization; writing – review and editing; software; formal analysis; project administration; data curation; supervision; resources. Nawarus Prapaiwan: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; software; formal analysis; project administration; data curation; supervision; resources. Thanapon Chotikaprakal: Investigation; writing – review and editing; writing – original draft. Phirom Prompiram: Investigation. Narumon Phaonakrop: Investigation. Sittiruk Roytrakul: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; software; formal analysis; project administration; data curation; supervision; resources. Parichart Tesena: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; software; formal analysis; project administration; data curation; supervision; resources.

DATA INTEGRITY STATEMENT

Ruethaiwan Vinijkumthorn, Nawarus Prapaiwan, Thanapon Chotikaprakal, Phirom Prompiram, Narumon Phaonakrop, Sittiruk Roytraku, and Parichart Tesena had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

ETHICAL ANIMAL RESEARCH

The study protocols were approved by the Institutional Animal Care and Use Committee of the Mahidol University: permission No. MUVS‐2022‐11‐64.

INFORMED CONSENT

Not applicable.

ANTIMICROBIAL STEWARDSHIP POLICY

Not applicable.

Supporting information

Data S1. Spreadsheet includes the following:

File S1. Gastric ulcer grading and location of stomach (Nonglandular region, Glandular region) in the control and treatment groups.

File S2. List of main‐functional and sub‐functional KEGG categories of EGUS in serum and faeces.

File S3. List of putative serum proteins from differentially expressed proteins analysis under DESeq2.

File S4. List of putative faecal proteins from differentially expressed proteins analysis under DESeq2.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in ProteomeXchange: PXD059214 and JPST003523.