Biomarkers for monitoring the equine large intestinal inflammatory response to stress-induced dysbiosis and probiotic supplementation

Abstract

Large intestine barrier disturbances can have serious consequences for the health of horses. The loss of mucosal integrity that leads to increased intestinal permeability may result from a local inflammatory immune response following alterations of the microbiota, known as dysbiosis. Therefore, our research aimed to identify noninvasive biomarkers for studying the intestinal permeability and the local inflammatory immune response in horses. Regarding the biomarkers used in other mammalian species, we measured the concentrations of lipopolysaccharides (LPS), reflected by 3-OH C14, C16, and C18 fatty acids, in blood, and fecal secretory immunoglobulin-A (SIgA). These biomarkers were evaluated in two trials including 9 and 12 healthy horses, which developed large intestinal dysbiosis experimentally induced by 5 d of antibiotic administration (trimethoprim sulfadiazine [TMS]) or 5 d of abrupt introduction of high starch levels (barley) into the diet. Horses were either control or supplemented with Lactobacillus acidophilus, Ligilactobacillus salivarius, and Bifidobacterium lactis. Correlations were performed between biomarkers and fecal bacterial diversity, composition, and function. No significant interaction between day and supplementation, or supplementation effect were observed for each biomarker. However, with the dietary stressor, a significant increase in blood concentrations of 3-OH C16 (P = 0.0125) and C14 (P = 0.0252) fatty acids was measured 2 d after the cessation of barley administration. Furthermore, with the antibiotic stressor, blood levels of 3-OH C16 progressively increased (P = 0.0114) from the first day to 2 d after the end of TMS administration. No significant day effect was observed for fecal SIgA concentrations for both stressors. These results indicate that both antibiotic- and diet-induced dysbiosis resulted in a local translocation of LPS 2 d after the cessation of the stressor treatments, suggesting an impairment of intestinal permeability, without detectable local inflammation. Blood LPS and fecal SIgA concentrations were significantly correlated with several bacterial variations in the large intestine, which are features of antibiotic- and diet-induced dysbiosis. These findings support the hypothesis that a relationship exists between dysbiosis and the loss of mucosal integrity in the large intestine of horses.

This study provides preliminary data on the use of blood and fecal biomarkers as non-invasive indicators of the loss of mucosal integrity or the occurrence of local inflammation in association with microbiota disturbances in the large intestine of horses. Such biomarkers could be beneficial in equine research and have future application in veterinary science practice.

Introduction

Intestinal barrier permeability is described as the facility by which intestinal epithelium allows molecules to pass through by nonmediated, passive diffusion (Travis and Menzies, 1992). The permeability of the intestinal epithelium is related to the mucosal barrier integrity, which is crucial for the general health and well-being of horses. Impairments in large intestine barrier function can have serious consequences, such as weight loss and poor performance in mild cases, and systemic inflammatory response syndrome, multiple organ dysfunction syndrome, or even death in the most severe cases (Stewart et al., 2017). In mammals, increased intestinal permeability is a well-accepted consequence of local mucosal inflammation (Ivanov et al., 2010; Lechuga and Ivanov, 2017), which can result from dysbiosis (Round and Mazmanian, 2009; Petersen and Round, 2014). Horses are very susceptible to microbial alterations in their large intestine driven by various environmental factors (Garber et al., 2020). However, in horses as in other mammals, it remains unclear whether intestinal dysbiosis is the cause or a consequence of mucosal inflammation (Buttó and Haller, 2016).

Assessment of large intestinal barrier integrity and the mucosal immune response requires a biopsy. Several recent studies used postmortem cecal and colonic tissue samples from healthy horses to evaluate the mucosal immune response (Lindenberg et al., 2019; Wambacq et al., 2020), but data from postmortem evaluations may differ from the examination of biopsies collected from live animals. Furthermore, in vivo examinations of rectal, rather than intestinal, mucosal samples have also been undertaken from horses with active inflammatory bowel diseases (Hjertner et al., 2013; Olofsson et al., 2015), but it is questionable whether investigations of rectal biopsies accurately represent events occurring in higher compartments of the intestine. Therefore, the use of easily accessible biomarkers found in the biological matrix, such as blood and feces, may allow suitable, noninvasive sampling in live animals. Based on the biomarkers used in other mammalian species, intestinal permeability and the local inflammatory immune response could be investigated by measuring the concentration of lipopolysaccharide-type endotoxins in the blood. The level of blood lipopolysaccharides (LPS) represents the proportion of LPS translocated across the intestinal mucosal barrier (André et al., 2021). In feces, inflammatory biomarkers, such as secretory immunoglobulin-A (SIgA), have been proposed as indicators of local intestinal barrier function (Bischoff et al., 2014), and the concentration of SIgA is related to the immune response to intestinal bacteria (Siddiqui et al., 2017).

Therefore, our research aimed to investigate whether these two biomarkers could indicate a compromised large intestinal permeability and a local inflammatory immune response, associated with dysbiosis in horses. We selected oral antibiotic administration and abrupt dietary change as the two microbiota stressors because of their well-documented capacity for inducing dysbiosis in the horse large intestine (Julliand and Grimm, 2017; Garber et al., 2020). We hypothesized that SIgA and LPS concentrations would increase in response to both antibiotic- and diet-induced dysbiosis in the large intestine, and that the variations would be correlated with fecal bacterial diversity, composition, and function. We also assessed the effects of dietary supplementation with three probiotic lactic acid bacteria strains, Lactobacillus acidophilus, Ligilactobacillus salivarius, and Bifidobacterium lactis. Our hypothesis was that supplementation with a combination of these strains would provide intestinal immunomodulatory effects, as reported in other mammals (Chaves et al., 2017; Quigley, 2017).

Materials and Methods

This research is part of two independent experimental trials that studied the effects of antibiotic oral administration (Trial 1) and dietary changes (Trial 2) on large intestinal microbiota. The trials were conducted at Lab To Field’s experimental facilities (Créancey, France). The experimental procedures and methods have been described in detail in Collinet et al., (2021a, 2021b). All experimental procedures were evaluated and approved by the institutional ethics committee (Comité d’Éthique de l’Expérimentation Animale Grand Campus Dijon). Both projects were authorized by the French Ministry of Agriculture (registration Nos.: APAFIS#12920-2016010516167886 and APAFIS #18307-2019010309072101).

Animals and experimental design

Nine (4 to 11 yr old, 523 ± 37 kg body weight [BW]) and 12 (4 to 10 yr old, 505 ± 35 kg BW) healthy adult Trotteurs Français geldings were involved in Trial 1 and 2, respectively. Seven individuals performed the two trials that were conducted at 210 d interval. Horse management was similar in both trials in terms of housing and exercising. In both trials, a high fiber diet (HF) composed of 86% dry matter (DM) of hay and 14% DM of pellet concentrates (Base 3, SERKO, Gemmelaincourt, France) was distributed in two equal meals. It supplied 2.15% BW in the form of DM and 0.58 g/kg BW of starch per meal. Prior to the start of each trial on day 0, horses were adapted to the experimental management. The two microbiota stressors were administered for 5 d from day 0, to induce a dysbiosis in the horses’ large intestine. In both trials, we tested Floréquilibre CVX (Wamine and PileJe Laboratories, Paris, France) containing L. acidophilus (LA201), L. salivarius (LA302), and B. lactis (LA304) as the probiotic supplementation. The supplement was given each morning before the meal from day 1 to the end of each experimental period. To ensure total ingestion of the supplement, it was incorporated into a mixture of 100 g alfalfa rehydrated with water. Horses from the control group were fed the same mixture without the supplement.

Trial 1

Horses were separated in three homogenous groups of three and enrolled in a 3 × 3 Latin square design composed of three experimental periods of 28 d (day 0 to day 28), separated by a wash-out period of 28 d. During each period, Trimethoprim sulfadiazine (TMS) was used as a stressor in all groups and was administered per os to horses (30 mg/kg BW, q. 24 h, day 0 to day 4). Horses received either Dose 1 (3 g; 4 × 10¹⁰ colony-forming units [CFU]) or Dose 5 (15 g; 2 × 10¹¹ CFU) of Floréquilibre CVX, or a placebo (control) each morning before their meal.

Trial 2

Horses were separated into two homogenous groups of six and enrolled in a longitudinal design of 56 d (day 0 to day 56). All horses were submitted to an abrupt change from the HF to the high starch diet (HS) as a stressor. Both diets were isoenergetic. The HS diet was composed of 60% DM of hay and 40% DM of rolled barley, and supplied 1.74% BW DM including 0.70% BW DM of barley. This provided 2.1 g/kg BW of starch intake per meal. It was distributed into 10 meals, (two per day, at 8:15 and 16:45), starting at day 0 with the afternoon meal, before return to the HF diet on day 5. One group of horses received the same Dose 1 as distributed in trial 1, while the second group did not (Control).

Sample collection

Both trials had a similar time course of sample collections: day 0, day 2, day 7, day 14, day 21, and day 28. A final sample was collected on day 56 in Trial 2 only. Fresh feces and blood samples were collected by rectal grab and venipuncture of the jugular vein, respectively. A fecal subsample was suspended in a 1× stock phosphate-buffered saline solution, centrifuged (1,500g for 20 min at 4 °C), and frozen (−20 °C) for SIgA analysis. Plasma was obtained after centrifugation (1,700g for 20 min at 4 °C) and stored frozen for further LPS analysis (–20 °C).

Blood lipopolysaccharides

Circulating lipopolysaccharide-type endotoxin concentrations were assessed by measuring plasma esterified 3-OH fatty acid (FA) concentrations by liquid chromatography–tandem mass spectrometry (André et al., 2021). In summary, esterified 3-OH FA concentrations were calculated as the difference between the amounts of total 3-OH FA released after strong acidic hydrolysis and the amounts of unesterified free 3-OH FA measured after direct extraction without the acidic hydrolysis step. 3-OH C14, C16, and C18 concentrations were analyzed independently, and the total 3-OH FA concentration was estimated by summing all the individual concentrations. Results are expressed in picomoles per milliliter.

Fecal secretory immunoglobulin-A

Fecal SIgA levels were quantified using enzyme-linked immunosorbent assays (Collinet et al., 2021a). Briefly, samples were centrifuged and analyzed using the manufacturer’s accessory kits (Horse IgA Quantification Set, Bethyl Laboratories Ltd., Montgomery, TX, USA) as adapted in our laboratory (Collinet et al., 2021a). Absorbance was read at 450 nm using a spectrophotometer (MRX Revelation Microplate Reader). Results are expressed in microgram per gram of fecal DM.

Microbiota analyses

Culture-independent and culture-dependent methods were used to study the fecal bacterial microbiota of Trial 1 and 2 horses. The materials and methods for these microbial analyses are thoroughly detailed in Collinet et al., (2021a, 2021b). Bacterial 16S ribosomal RNA gene extraction, amplification, and sequencing analysis was performed on fecal samples collected under sterile conditions to identify the bacterial richness and diversity (number of observed operational taxonomic units (OTU), and Shannon and InvSimpson indices), as well as composition, in terms of relative abundances. Enumeration of bacterial functional groups from fresh fecal samples was also realized using conventional anaerobic culture techniques to determine total anaerobic, cellulolytic, amylolytic, and lactic acid-utilizing bacteria concentrations. Results are expressed in CFU per gram of feces.

Statistical analyses

To achieve a normal distribution of the variables, a base 10 logarithm transformation was used for the esterified 3-OH FA, SIgA, and bacterial functional group concentrations. In each trial, statistical analysis was performed using the MIXED procedure of SAS software, version 9.3 (SAS Institute Inc., Cary, NC, USA), with a model including supplementation, day, and the interaction between supplementation and day as fixed effects. The model included “horse” as a random effect and “days” as repeated measures. For trial 1 (Latin square design), the period was added as a fixed effect, and was also used as an intercept for each horse to avoid individual variations along the duration of the trial. For trial 2, individual horse values were included as intercepts to avoid baseline differences because of the longitudinal design of the trial.

For each trial separately the SAS CORR procedure was computed to calculate the Pearson’s correlation coefficient between each biomarker (LPS and SIgA) and parameters of bacterial richness, diversity, composition (at a family level, with a relative abundance > 0.1%), and function (bacterial functional group concentrations), in the case of a significant day effect. The significance threshold was set at P ≤ 0.05.

Results

The mean 3-OH C14 concentrations (Trial 1: 19.43 ± 4.62 pmol/mL; Trial 2: 39.35 ± 5.16 pmol/mL) were lower than the mean 3-OH C16 (Trial 1: 95.28 ± 23.64 pmol/mL; Trial 2: 97.55 ± 13.11 pmol/mL) and 3-OH C18 (Trial 1: 57.86 ± 12.73 pmol/mL; Trial 2: 77.01 ± 18.28 pmol/mL) concentrations in both trials. Total 3-OH FA concentrations were 172.57 ± 33.53 pmol/mL and 213.91 ± 29.82 pmol/mL in Trial 1 and 2, respectively. The fecal concentrations of SIgA were 17.98 ± 17.82 μg/g fecal DM and 20.57 ± 35.02 μg/g fecal DM in Trial 1 and 2, respectively.

No significant day × supplementation interaction (P > 0.05) or supplementation effect (P > 0.21) was observed for each biomarker (Table 1). Thus, data reported in Table 1 reflects only the data for each day pooled across supplements. In trial 1, there was no significant period effect (P > 0.13) except for SIgA concentrations (P = 0.0010), which were lower in period 3 (3.78 ± 0.51 Log₁₀ μg/g DM) than in period 1 (4.19 ± 0.38 Log₁₀ μg/g DM) or period 2 (4.17 ± 0.42 Log₁₀ μg/g DM). Effects of the day were observed in each trial for blood LPS concentrations (Table 1). In Trial 1, the concentration of 3-OH C16 (P = 0.0114) increased from day 0 to day 7 and then decreased to its basal value at day 21. The concentration of total 3-OH FA (P = 0.0477) was significantly higher at day 2 and day 7 than at day 21 and day 28. In Trial 2, the concentrations of 3-OH C14 (P = 0.0252) and 3-OH C16 (P = 0.0125) increased from day 2 to day 7 and decreased at day 14. No significant effect of the day was observed for SIgA in the analysis of both trials (Table 1).

Table 1.

Effects of day, supplementation, and their interaction on LPS¹ and SIgA² biomarkers in Trial 1 (antibiotic stress) and Trial 2 (dietary stress)

Biomarker	Trial§	Mean value per day							Mean ± SD	Effect, P-value†
		Day 0*	Day 2	Day 7	Day 14	Day 21	Day 28	Day 56		D	S	D*S
3-OH C14 (Log10 pmol/mL)	1	1.26	1.29	1.29	1.27	1.27	1.28	-	1.28 ± 0.09	0.7544	0.3576	0.7978
	2	1.60ab	1.58bc	1.62a	1.56c	1.59abc	1.59abc	1.60ab	1.59 ± 0.06	0.0252	0.9597	0.0512
3-OH C16 (Log10 pmol/mL)	1	1.95bc	2.00ab	2.00a	1.97abc	1.93c	1.94c	-	1.97 ± 0.10	0.0114	0.6916	0.4120
	2	1.99ab	1.96b	2.00a	1.95b	1.98ab	2.01a	2.01a	1.99 ± 0.06	0.0125	0.6818	0.1239
3-OH C18 (Log10 pmol/mL)	1	1.78	1.77	1.75	1.73	1.74	1.75	-	1.75 ± 0.09	0.1265	0.6052	0.2975
	2	1.89	1.79	1.91	1.83	1.87	1.86	1.88	1.86 ± 0.20	0.8449	0.9447	0.3174
Total 3-OH FA (Log10 pmol/mL)	1	2.23ab	2.25a	2.25a	2.22ab	2.21b	2.22b	-	2.23 ± 0.08	0.0477	0.6213	0.4075
	2	2.34	2.32	2.35	2.29	2.32	2.33	2.34	2.33 ± 0.06	0.1817	0.5726	0.1833
SIgA (Log10μg/g DM3)	1	4.05	4.08	4.05	4.11	4.00	3.98	-	4.05 ± 0.48	0.7088	0.2058	0.1247
	2	4.18	4.06	4.11	3.70	4.05	3.94	3.79	3.98 ± 0.53	0.1833	0.7018	0.4352

LPS, lipopolysaccharides

SIgA, secretory immunoglobulin-A

DM, dry matter

Mean values with different superscripts in the same row differ at P < 0.05. Significant P-values are shown in bold.

Day 0: Basal value before the antibiotic administration or the abrupt starch incorporation.

D: Day, S: Supplementation, D*S: interaction between D and S.

Trial 1: trimethoprim sulfadiazine was used as a stressor and was administered per os to horses from Day 0 to Day 4. Trial 2: the horses’ diet was abruptly changed from a high-fiber to a high-starch diet, with high quantities of rolled barley introduced into 10 meals (two meals per day), starting at Day 0 with the afternoon meal, before return to the HF diet.

The results of the detailed microbiota analyses have been reported in Collinet et al., (2021a, 2021b). In the present study, only the correlations between each biomarker (LPS and SIgA) and microbiota diversity, composition (relative abundance at a family level), and function (bacterial functional groups concentrations) are presented. P-values and Pearson correlation coefficient (r) values are summarized in Table 2. In Trial 1, indices of fecal bacterial richness and diversity, such as the number of observed OTUs, and the Shannon and InvSimpson indices, were negatively correlated with 3-OH C18 concentrations in blood. All blood 3-OH FA concentrations were positively correlated with the relative abundance of the Prevotellaceae family in fecal samples. Blood 3-OH C16 concentrations were negatively correlated with the relative abundance of the Bacteroidales RF16 group, Defluviitaleaceae, and Spirochaetaceae. Blood 3-OH C18 and total 3-OH FA concentrations were negatively correlated with the relative abundance of the Bacteroidales RF16 group, Family XIII, Rikenellaceae, and Spirochaetaceae. In Trial 2, no significant correlations were found between blood 3-OH FA concentrations and fecal bacterial richness, diversity, or the relative abundance of families.

Table 2.

Pearson correlations between biomarkers and fecal bacterial richness and diversity indices, the relative abundance of several bacterial families, and the concentrations of bacterial functional groups in Trial 1 (antibiotic stress) and Trial 2 (dietary stress). Only the bacterial parameters significantly correlated with at least one biomarker are presented in the table

	Bacterial parameter		3-OH C14	3-OH C16	3-OH C18	Total 3-OH FA	SIgA1
Trial 1	Number of observed OTUs	r *	−0.024	0.007	−0.219	−0.095	−0.005
		P †	0.764	0.927	0.005	0.229	0.951
	Shannon index	r	−0.024	−0.028	−0.297	−0.149	0.041
		P	0.761	0.721	<0.001	0.058	0.607
	InvSimpson index	r	0.036	−0.034	−0.234	−0.116	0.006
		P	0.645	0.670	0.003	0.141	0.941
	Bacteroidales BS11 gut group	r	0.084	0.049	0.050	0.065	−0.181
		P	0.286	0.532	0.531	0.410	0.021
	Bacteroidales RF16 group	r	−0.135	−0.228	−0.171	−0.247	0.058
		P	0.086	0.004	0.029	0.001	0.461
	Christensenellaceae	r	0.127	0.097	0.013	0.086	0.199
		P	0.108	0.221	0.870	0.275	0.011
	Defluviitaleaceae	r	−0.068	−0.156	−0.020	−0.130	−0.076
		P	0.392	0.047	0.801	0.100	0.335
	Family XIII	r	−0.139	−0.110	−0.214	−0.189	0.210
		P	0.077	0.163	0.006	0.016	0.007
	Lachnospiraceae	r	−0.017	0.013	0.161	0.073	−0.134
		P	0.828	0.866	0.041	0.358	0.089
	p-251-o5	r	0.053	0.013	0.033	0.039	−0.210
		P	0.499	0.872	0.677	0.618	0.007
	Prevotellaceae	r	0.155	0.214	0.221	0.262	0.034
		P	0.050	0.006	0.005	0.001	0.669
	Rikenellaceae	r	−0.075	−0.140	−0.209	−0.197	0.032
		P	0.342	0.075	0.008	0.012	0.689
	Spirochaetaceae	r	−0.029	−0.263	−0.235	−0.283	0.050
		P	0.7163	0.001	0.003	< 0.001	0.524
	Total anaerobic bacteria	r	−0.191	−0.059	−0.055	−0.088	−0.119
		P	0.015	0.458	0.488	0.265	0.130
	Cellulolytic bacteria	r	−0.080	−0.132	−0.167	−0.176	0.055
		P	0.311	0.095	0.034	0.025	0.489
	Lactic acid-utilizing bacteria	r	−0.116	0.001	−0.191	−0.090	−0.133
		P	0.142	0.986	0.015	0.256	0.092
Trial 2	Bacteroidales RF16 group	r	−0.068	−0.106	−0.029	−0.089	−0.248
		P	0.544	0.340	0.797	0.424	0.024
	Amylolytic bacteria	r	0.125	0.248	−0.057	0.147	0.061
		P	0.255	0.023	0.605	0.181	0.583

SIgA, secretory immunoglobulin-A

Coefficient of correlation

P-value, significant P-values are shown in bold.

In Trial 1, cellulolytic bacterial concentrations in feces were negatively correlated with blood 3-OH C18 and total 3-OH FA concentrations, total anaerobic bacterial concentrations were negatively correlated with blood 3-OH C14 concentrations, and lactic acid-utilizing bacterial concentrations were negatively correlated with blood 3-OH C18 concentrations. In Trial 2, amylolytic bacterial concentrations were positively correlated with blood 3-OH C16 concentrations.

In Trial 1, fecal SIgA concentrations were positively correlated with the fecal relative abundance of Christensenellaceae and Family XIII, whereas they were negatively correlated with the relative abundance of the Bacteroidales BS11 gut group and p-251-o5. In Trial 2, fecal SIgA concentrations were negatively correlated with the fecal relative abundance of the Bacteroidales RF16 group.

Discussion

Whether intestinal dysbiosis is the cause or a consequence of inflammation in mammals remains unclear (Buttó and Haller, 2016). The aim of this study was thus to investigate if blood LPS and fecal SIgA concentrations could indicate a compromised mucosal permeability and local inflammation, respectively, in response to large intestinal dysbiosis in horses.

To experimentally induce dysbiosis, we chose two major microbiota stressors reported in the literature: TMS administration (Harlow et al., 2013; Costa et al., 2015) and an abrupt introduction of high levels of starch into the diet (Goodson et al., 1988; De Fombelle et al., 2001; Respondek et al., 2008; Warzecha et al., 2017). Using a holistic and comprehensive analysis of the large intestinal ecosystem, we confirmed that these stressors induced a loss of bacterial diversity, a bloom of potential pathobionts, and a loss of commensals (Collinet et al., 2021a, 2021b), features corresponding to the definition of dysbiosis by Petersen and Round (2014). Most of the microbiota changes were observed 2 d after the onset of the respective stressor.

In our study, both antibiotic- and diet-induced dysbiosis resulted in an increased blood LPS concentration 2 d after the completion of the stressor administration, suggesting an impaired intestinal integrity in both cases. Modifications of LPS-type endotoxin concentrations depended on the stressor. With the dietary stressor, a significant increase in the concentrations of 3-OH C16 and C14 was measured 2 d after stopping the HS diet, suggesting that the increase in permeability was concomitant to the shift in large intestinal microbiota. However, in a previous study, the blood concentration of 3-OH C14-type endotoxin was not modified in horses subjected to a progressive starch overload (Grimm et al., 2018), despite a similar level of starch administered to the horses. In this earlier study, starch was introduced gradually, whereas in the present study, it was an abrupt incorporation into the diet, suggesting that an abrupt introduction of starch into the diet may be more disruptive of the equine intestinal permeability than a progressive starch overload. In this regard, the abrupt introduction of high dietary starch levels in horses has been used experimentally to cause endotoxemia and laminitis that result from changes in the cecal microbiota (Garner et al., 1978; Sprouse et al., 1987).

With the antibiotic stressor, only the level of 3-OH C16 in the blood progressively increased from the first day to 2 d after the cessation of TMS administration. Irrespective of the stressor, circulating 3-OH C16 was the most affected LPS-type endotoxin in response to dysbiosis in the blood of the horses examined, compared with 3-OH C14 and C18. Consequently, 3-OH C16 appears to be the best LPS-type endotoxin to study further as a potential biomarker of intestinal permeability in horses.

In these trials, we found significant correlations between the microbiota changes and the two biomarkers we explored. Despite significant, all the correlation values were weak and must be interpreted with caution. Correlations between blood LPS concentrations and fecal microbial parameters were different between the stressors, suggesting distinct mechanisms underlying the loss of mucosal permeability. In antibiotic-induced dysbiosis, the levels of 3-OH C18-type endotoxin were negatively correlated with bacterial indices of richness and diversity. The concentrations of LPS-type endotoxins (C16, C18, and total FA) were also negatively correlated with the relative abundance of Rikenellaceae and Spirochaetaceae, two families containing beneficial fibrolytic members (Tokuda et al., 2018; Ren et al., 2020). By contrast, levels of all LPS-type endotoxins analyzed were positively correlated with the Prevotellaceae family-containing pathobionts (Larsen, 2017). As a consequence, LPS-type endotoxins were correlated with the three features of dysbiosis described by Petersen and Round (2014), and appear to be promising biomarkers of equine large intestine permeability associated with antibiotic-induced dysbiosis. Following the abrupt introduction of starch into the horse diet to provoke large intestine dysbiosis, we found that the level of 3-OH C16-type endotoxin was positively correlated with the concentration of amylolytic bacteria. The increase of such bacteria in the large intestine is the direct consequence of the high amount of starch that avoids pre-cecal digestion in horses (Julliand et al., 2006). These bacteria and their fermentation end products, such as lactic acid, lead to intestinal acidosis (Julliand and Grimm, 2017). Acidosis can damage the intestinal epithelium, alter its permeability (Shirazi-Beechey, 2008; Stewart et al., 2017), and consequently result in a higher level of 3-OH C16-type endotoxin translocation into the blood. Thus, 3-OH C16-type endotoxins may be a useful biomarker of equine large intestine permeability associated with HS dietary-induced dysbiosis.

Exploration of fecal concentrations of SIgA allowed us to investigate the possible intestinal inflammatory immune response in the experimental trials. However, irrespective of the stressor, SIgA levels in feces were not significantly modified. With the antibiotic stressor, concentrations of SIgA were correlated with the relative abundance of several bacterial families, which were significantly affected during antibiotic administration. The interpretation of these variations and correlations is quite difficult. In mammals, there is evidence that intestinal SIgA promotes host–commensal bacteria mutualism. This association protects the intestinal mucosa by mediating an immune exclusion, and thus prevents the development of intestinal inflammation (Kaetzel, 2014). In mammals, SIgA coating seems to be enriched for some members of the microbiota, which vary depending on the anatomical proximity of bacteria to the mucosal intestinal surface. However, the nature of this SIgA coating has yet to be further explored (Pabst et al., 2016). Based on the positive correlations observed in the trials, it is hypothesized that Christensenellaceae and Family XIII, two commensal and ubiquitous families that belong to the phylum Firmicutes, can enhance SIgA secretion and coating during antibiotic-induced dysbiosis. On the contrary, we found a negative correlation between Bacteroidetes families, such as the Bacteroidales BS11 gut group, p-251-o5, and the Bacteroidales RF16 group, and SIgA secretion in the intestinal lumen during antibiotic- and diet-induced dysbiosis, respectively. Modifications of the relative abundance of Firmicutes and Bacteroidetes were reported in horses suffering from diarrhea and colitis (Costa et al., 2012; Rodriguez et al., 2015; Weese et al., 2015), suggesting that SIgA could be a biomarker of dysbiosis-induced inflammation.

In spite of the immunomodulatory properties of L. acidophilus, L. salivarius, and B. lactis in mammals (Chaves et al., 2017; Quigley, 2017), feed supplementation with these strains had no significant effects on intestinal permeability and the local inflammatory immune response to dysbiosis in horses. Numerical differences of interest were observed, such as the fecal concentration of SIgA in Trial 1 that increased at day 2 (P = 0.1247) in the control group in response to antibiotic administration (in Log₁₀ μg/g DM, day 0: 3.97 ± 0.36; day 2: 4.26 ± 0.59; day 7: 3.99 ± 0.42; day 14: 4.16 ± 0.25; day 21: 3.98 ± 0.37; day 28: 4.13 ± 0.46), whereas it was not the case in the group of horses receiving Dose 1 of the supplement (in Log₁₀ μg/g DM, day 0: 4.07 ± 0.37; day 2: 3.77 ± 0.79; day 7: 4.11 ± 0.54; day 14: 4.07 ± 0.52; day 21: 3.95 ± 0.36; day 28: 3.69 ± 0.79). Consequently, it is hypothesized that including more horses in the trials might allow detecting a significant effect of Floréquilibre CVX supplementation on LPS and SIgA concentrations. As demonstrated in human and murine models, individuals are either permissive or resistant to the colonization of probiotics (Zmora et al., 2018). Monitoring fecal recovery of the three strains of Floréquilibre CVX at the end of the 28 d of administration would have made it possible to have better knowledge of their resistance to colonization in horses and their potential impacts on microbiome or immune functions. Moreover, in trial 1, we may hypothesize that the effect of the probiotics has been overpassed by the stress that the antibiotics have exhibited in each experimental period. The repeated antibiotic stress may have caused a shift of the microbiota to an alternative stable state, different from the initial state, as demonstrated in humans (Dethlefsen and Relman, 2011). This may contribute to the alteration of the immune response of horses, as well as the response to probiotics. This hypothesis can be supported by the lower concentrations of fecal SIgA concentration observed in period 3. Further studies should be conducted to assess whether supplementation with 4 × 10¹⁰ CFU of L. acidophilus, L. salivarius, and B. lactis might provide immunomodulatory effects and help maintain the integrity of the mucosal intestinal barrier.

In conclusion, LPS-type endotoxin and SIgA appear to be promising biomarkers for monitoring the equine intestinal barrier response to dysbiosis. The correlations between the biomarkers and parameters of the bacterial ecosystem studied for the first time here indicate that intestinal permeability and the inflammatory immune response were related to modifications of the richness, diversity, composition, and function of bacteria in the equine large intestine. The loss of mucosal integrity observed soon after the onset of dysbiosis in the large intestine suggested that a prior dysbiosis might lead to a subsequent modification of the local inflammatory immune response in horses.

Glossary

Abbreviations

BW

body weight

CFU

colony-forming units

DM

dry matter

FA

fatty acids

HF

high fiber diet

HS

high starch diet

LPS

lipopolysaccharides

OTU

operational taxonomic unit

SIgA

secretory immunoglobulin-A

TMS

trimethoprim sulfadiazine

Conflict of Interest Statement

The authors declare no real or perceived conflicts of interest.