Exercise Preconditioning Attenuates the Response to Experimental Colitis and Modifies Composition of Gut Microbiota in Wild-Type Mice

This study investigated the suppressive effect of exercise preconditioning against colitis induced by high-fat diet (HF) plus dextran sulfate sodium (DSS) in wild-type mice. Male mice (C57BL/6) aged 6 weeks were assigned to standard chow (SC, n = 10) or HF (n = 10) or HF followed by DSS (HF+DSS, n = 10) or exercise preconditioning (EX) followed by HF+DSS (EX+HF+DSS, n = 10) for a total of 15 weeks. After 12 weeks of dietary treatments and/or exercise preconditioning, mice in the DSS groups were subjected to administration of 2 cycles of 5-day DSS (2% w/v) with a 7-day interval between cycles. HF resulted in colitis symptoms and histological changes, infiltration of immunity cells, decreased gut barrier proteins, increased pro-inflammatory and chemotactic cytokines and decreased anti-inflammatory cytokine such as adiponectin, which deteriorated after administration of DSS. Exercise preconditioning alleviated HF+DSS-induced colitis and caused significant modifications in gut microbiota: decreased Bacteroides vulgatus (p = 0.050) and increased Akkermansia muciniphila (p = 0.050). The current findings suggest that exercise preconditioning attenuates the severity of HF+DSS-induced colitis in C57BL/6 mice.


Introduction
Inflammatory bowel disease (IBD) refers to immune disorders of the gastrointestinal (GI) tract, primarily Crohn's disease and ulcerative colitis. Although North America and Europe have high prevalence of IBD, its incidence is also rising at an unprecedented rate in developing countries due to rapidly Westernized lifestyles [1].
The leading hypothesis for IBD is that an interaction between environmental and immune factors triggers disruption of intestinal barrier integrity, resulting in an exaggerated immune response to the commensal bacterial resident in the gut lumen [2] and chronic activation and inflammation in the intestinal immune system [3]. Pathologically, chronic inflammation is a common feature of IBD and obesity in both animals [4] and humans [5]. In this perspective, dysbiotic gut microbiota often observed in IBD patients [6] suggest that aberrations in gut microbiota composition associated with Westernized diets may also contribute to the pathogenesis of IBD [6].
In the meanwhile, it is well established that regular physical activity improves immune function [7] and decreases risk of intestinal inflammatory diseases including IBD [8]. In addition, exercise training improves gut barrier function by up-regulating heat shock protein (HSP) in the gut [9] and alleviates inflammatory responses in animal models of experimental colitis [10]. Exercise training modulates gut microbiota-enhancing symbiosis such that the host and microbe benefit mutually [11]. Together, those findings suggest that physical activity and/or modifications in gut microbiota may lead to Figure 1A represents overall study design. One week after adaptation to environment, mice aged 6 weeks were randomly assigned to standard chow (SC, n = 10) or high-fat diet (HF, n = 10) or HF plus dextran sulfate sodium (DSS) (HF+DSS, n = 10) or exercise preconditioning (EX) with HF+DSS (EX+HF+DSS, n = 10).

Experimental Design
Life 2020, 10, x FOR PEER REVIEW 5 of 14

Exercise Preconditioning Alleviates Metabolic Complications Associated with HF+DSS-induced Colitis
Figure 1B-D represent changes in weights and GTT after 12 weeks of dietary treatments and exercise preconditioning. As expected, chronic exposure to high-fat diet resulted in significant weight gains and higher AUC during GTT (as shown in HF vs. SC), which were alleviated by exercise preconditioning (as shown in EX+HF vs. HF). Figures 1E-H represent hepatic damage markers associated with dietary treatments and exercise preconditioning. HF resulted in significant elevations of serum AST and ALT enzymes and higher hepatic expression of Timp1, Colla1, and Ly6d mRNAs (as shown in HF vs. SC), which were further elevated by administration of DSS (as shown in HF+DSS vs. HF). On the other hand, exercise preconditioning attenuated HF+DSS-induced elevations of liver damage markers (as shown in EX+HF+DSS vs. HF+DSS). during glucose tolerance test (n = 10 mice per group) (E) liver hematoxylin and eosin (H&E) staining (magnification, ×10; scale bar, 20 μm, n = 5 mice per group) (F) serum aspartate aminotransferase (AST) (G) serum alanine aminotransferase (ALT) (n = 6 mice per group) and (H) hepatic mRNA expression of Timp1, Col1a1, Ly6d, and Lgals (n = 6 mice per group). GTT: glucose tolerance test; SC: standard chow; HF: high fat diet; EX: exercise preconditioning; DSS: dextran sulfate sodium. Data are presented as mean ± S.D. Different letters correspond to significant differences between groups. Statistical significances were tested with one-way analysis of variance.  during glucose tolerance test (n = 10 mice per group) (E) liver hematoxylin and eosin (H&E) staining (magnification, ×10; scale bar, 20 µm, n = 5 mice per group) (F) serum aspartate aminotransferase (AST) (G) serum alanine aminotransferase (ALT) (n = 6 mice per group) and (H) hepatic mRNA expression of Timp1, Col1a1, Ly6d, and Lgals (n = 6 mice per group). GTT: glucose tolerance test; SC: standard chow; HF: high fat diet; EX: exercise preconditioning; DSS: dextran sulfate sodium. Data are presented as mean ± S.D. Different letters correspond to significant differences between groups. Statistical significances were tested with one-way analysis of variance. The SC and HF mice were fed with SC and HF, respectively, for a total of 15 weeks. The SC diet consisted of 10% fat, 70% carbohydrates, and 20% protein (kcal) and was provided in the form of regular pellets (Purina Mills, Seoul, Korea). The HF was prepared in small pellets of 60% fat (90% lard and 10% soybean oil), 20% carbohydrates, and 20% protein (kcal) (D12492 Research Diet, New Brunswick, NJ, USA). The HF used in this study is a well-established protocol to induce metabolic complications [10] along with pro-inflammatory and immune responses [11]. Body weights were measured weekly.

Exercise Preconditioning Alleviates the Colitis Symptoms and Histological Changes Associated with HF+DSS-induced Colitis
The EX+HF+DSS mice were trained on a motor-driven treadmill (Columbus Instruments, Inc., Columbus, OH, USA) with duration of 50 min per session and a frequency of 5 days per week for 12 weeks. Mice ran on the treadmill at zero inclination and a speed of 8 m/min for 5 min (warming up), 15 m/min for 40 min (main exercise at moderate intensity), and 8 m/min for 5 min (cool down) [12]. No electrical shock was used. In the meanwhile, the SC and HF mice were placed on the same treadmill without running for a matched period.

Glucose Tolerance Test
After 11 weeks of dietary treatments and/or exercise training, a glucose tolerance test (GTT) was performed with an intra-peritoneal injection (1.5 g/kg body weight) of glucose (Sigma-Aldrich, St. Louis, MI, USA) after 16-h fasting. Blood samples were collected from a cut at the tip of the tail before and 15, 30, 45, 60 and 120 min after glucose injection (a total of~18 µL). Serum blood glucose was measured with One Touch II glucose meter (Lifescan, Johnson & Johnson, New Brunswick, NJ, USA). The linear trapezoid method was used to calculate area under the curve (AUC) for GTT. One week of recovery time was allowed prior to administration of DSS. We confirmed that all the mice were fully recovered from GTT.

DSS-Induced Colitis Model
After 12 weeks of dietary treatment and/or exercise training, the DSS-induced colitis model was established using a modified model from the previously reported protocol [13]. In brief, mice were given 2 cycles of 2% DSS (w/v) (reagent-grade, molecular weight of 36,000~50,000, MP Biochemical, cat. no. 02160110) in autoclaved drinking water for 5 days per cycle with a 14-day rest period between cycles and water ad libitum. Fresh DSS water was prepared daily. Control mice had free access to tap water. Clinical symptoms of colitis (e.g., weight loss, positive fecal hemoccult, and diarrhea) were monitored daily during and following administration of DSS.

Blood and Tissue Sampling
At the end of the experiments, blood was sampled from mice without food overnight. Blood collections were obtained from the abdominal vena cava and the hepatic portal vein just before sacrifice of mice under anesthesia with a mixture of Zoletil (40 mg/kg) and Rompun (5 mg/kg). Blood samples were centrifuged at 3000 rpm and 4 • C for 10 min and stored at −80 • C. Liver, colon, and spleen were carefully excised and flash frozen in liquid nitrogen. Prior to freezing, colon and spleen were photographed, and colon length and spleen weight were measured.

Histopathology and Colitis Scoring
Tissues of colon and liver were fixed in 4% paraformaldehyde, embedded in paraffin, cut into 3 µm sections, and stained with hematoxylin and eosin (H&E). Morphological assessments of stained slides were performed with the Leica Qwin image analyzer system. Histological scoring was blindly performed by two researchers using the degree of surface epithelial loss, crypt destruction and inflammatory cell infiltration into the mucosa, scoring from 0 to 12 [14].

Flow Cytometry Analysis
For measurement of leukocytes in colon and blood, samples were stained with anti-CD45+, Ly6G, and Ly6C [16]. For mouse cell gating strategy, a live cell was stained with viability dye. CD45+ and CD11b+ were used as markers for total leukocytes and myeloid cell, respectively. Neutrophils were defined as CD45+CD11b+Ly6G+Ly6C+/low. CD45+CD11b+Ly6C+Ly6G-cells were defined as monocytes and excluded from neutrophil gating [16]. The cells were read by a FACSCanto II flow cytometer (BD Biosciences, San Jose, CA, USA) and analyzed using a FlowJo 7.6.5 analysis platform (Tree Star Inc., Ashland, OR, USA).

16 S rRNA Gene and Metagenome Sequencing
Fecal samples were collected daily during the 14th week. Fecal bacterial DNA was extracted using the PowerMax ® Soil DNA Isolation Kit (MO BIO, Carlsbad, CA, USA). After extraction, DNA quantity and quality were checked with a PicoGreen ® dsDNA quantitation kit (Thermo Fisher Scientific, Seoul, Korea) and Nanodrop TM (Thermo Fisher Scientific, Seoul, Korea), respectively. The DNA samples were then prepared according to Illumina 16S metagenomic sequencing library protocols for the amplification of the V3 and V4 region (519F-806R) with polymerase chain reaction (PCR). The bar-coded fusion primer sequences used for amplifications were 519F: 5 CCTACGGGNGGCWGCAG 3 , 806R: 5 GAC TACHVGGGTATCTAATCC 3 . The final purified product was then quantified by quantitative PCR (qPCR) (KAPA Library Quantification kits for Illumina Sequencing platforms). The integrity of gDNA was checked with a LabChip GX HT DNA High Sensitivity Kit. (PerkinElmer, Waltham, MA, USA). Paired-end (2 × 300 bp) sequencing was performed with the MiSeq™ platform (Illumina, San Diego, CA, USA).

Metagenome Sequences Analysis
After sequencing MiSeq raw data, a FASTQ file was created using real-time analysis and bcl2fastq (v2.20.0.422). The paired-end data for each sample were assembled into a single sequence using FLASH (v1.2.11). Low-quality, ambiguous, and chimera sequences were removed using a CD-HIT-EST-based operational taxonomy unit (OTU) analysis program. Clustering of sequences with at least 97% sequence similarity yielded an OTU at the species level. The representative sequence of each OTU was produced with BLASTN (v.2.4.0) in the reference database (NCBI 16S Microbial). A Shannon index and inversed Simpson index were used to confirm species diversity and uniformity of microbial communities in environmental samples. Alpha diversity information was confirmed via a rarefaction curve and Chao1 value. Beta diversity between samples was obtained based on unweighted UniFrac Distance. Principal coordinates analysis (PCoA) was used to visualize the relationship between samples.

Statistical Analysis
Data were presented as mean ± standard derivation (SD). A Shapiro-Wilk normality test was performed to check normal distribution of data. One-way ANOVA, followed by Fisher's Least Significant Difference (LSD) post-hoc test, if necessary, was used to compare any group differences with statistical significance of p = 0.05. A nonparametric t-test with 999 Monte-Carlo permutations was used to compare significant differences in alpha-diversity metrics (i.e., Chao1, Shannon, and Simpson). Nonparametric Kruskal-Wallis test using the Benjamini-Hochberg False Discovery Rate (FDR) correction was performed to test significant differences in relative abundance of taxa at p = 0.05. All analyses were carried out using SPSS-PC 25.0 (SPSS Inc., Chicago, IL, USA).

Exercise Preconditioning Alleviates Metabolic Complications Associated with HF+DSS-induced Colitis
Figure 1B-D represent changes in weights and GTT after 12 weeks of dietary treatments and exercise preconditioning. As expected, chronic exposure to high-fat diet resulted in significant weight gains and higher AUC during GTT (as shown in HF vs. SC), which were alleviated by exercise preconditioning (as shown in EX+HF vs. HF). Figure 1E-H represent hepatic damage markers associated with dietary treatments and exercise preconditioning. HF resulted in significant elevations of serum AST and ALT enzymes and higher hepatic expression of Timp1, Colla1, and Ly6d mRNAs (as shown in HF vs. SC), which were further elevated by administration of DSS (as shown in HF+DSS vs. HF). On the other hand, exercise preconditioning attenuated HF+DSS-induced elevations of liver damage markers (as shown in EX+HF+DSS vs. HF+DSS).         Figure 4A-C represents the total and percentages of immune cells in whole blood and colon. HF resulted in significantly higher numbers of circulating neutrophils and monocytes and increased levels of TLR4 gene and protein in the colon, with no significant changes in the immunity cells in the colon, as shown in HF vs. SC. Administration of DSS further increased levels of neutrophils and monocytes in the colon and blood in conjunction with further increased levels of TLR4 gene and protein in the colon, as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning alleviated HF+DSS-induced increases of the immunity cells in the colon and blood and HF+DSS-induced increases of TLR4 gene and protein in the colon, as shown in EX+HF+DSS vs. HF+DSS. Figure 4A-C represents the total and percentages of immune cells in whole blood and colon. HF resulted in significantly higher numbers of circulating neutrophils and monocytes and increased levels of TLR4 gene and protein in the colon, with no significant changes in the immunity cells in the colon, as shown in HF vs. SC. Administration of DSS further increased levels of neutrophils and monocytes in the colon and blood in conjunction with further increased levels of TLR4 gene and protein in the colon, as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning alleviated HF+DSS-induced increases of the immunity cells in the colon and blood and HF+DSSinduced increases of TLR4 gene and protein in the colon, as shown in EX+HF+DSS vs. HF+DSS. Figure 4D-I represent pro-and anti-inflammatory and chemotactic cytokines. Chronic exposure to HF resulted in higher serum levels of IL-6, GRO-α, and MCP-1 and lower serum levels of adiponectin, with no significant changes in IL-17a and TGF-β, as shown in HF vs. SC, implying activation of inflammatory and chemotactic responses to HF. Administration of DSS exacerbated HFinduced activation of inflammatory and chemotactic responses, with no significant change in adiponectin, as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning suppressed HF+DSS-induced up-regulation of inflammatory and chemotactic cytokines (i.e., IL-6, GRO-α, and MCP-1) and HF+DSS-induced down-regulation of adiponectin, as shown in EX+HF+DSS vs. HF+DSS.   Figure 4D-I represent pro-and anti-inflammatory and chemotactic cytokines. Chronic exposure to HF resulted in higher serum levels of IL-6, GRO-α, and MCP-1 and lower serum levels of adiponectin, with no significant changes in IL-17a and TGF-β, as shown in HF vs. SC, implying activation of inflammatory and chemotactic responses to HF. Administration of DSS exacerbated HF-induced activation of inflammatory and chemotactic responses, with no significant change in adiponectin, as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning suppressed HF+DSS-induced up-regulation of inflammatory and chemotactic cytokines (i.e., IL-6, GRO-α, and MCP-1) and HF+DSS-induced down-regulation of adiponectin, as shown in EX+HF+DSS vs. HF+DSS. Table 1 represents alpha diversity parameters of gut microbiota, including observed OTUs, Chao1, Shannon's and Simpson's diversity indices, and rarefaction curves. HF resulted in higher values of OTUs (p = 0.022) and Shannon (p = 0.038), with no significance differences in Chao 1 and Simpson indices, as shown in HF vs. SC. Administration of DSS resulted in lower values of OTUs (p < 0.001), Chao1 (p = 0.003), Shannon (p = 0.001) and Simpson (p = 0.003), as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning suppressed decreases in Shannon (p = 0.047) and Simpson (p = 0.028) associated with HF+DSS, with no such suppressive effects on OTUs and Chao1, as shown in EX+HF+DSS vs. HF+DSS. Additionally, unweighted PCoA was conducted to compare the individual groups of mice and showed a clear separation of HF (red color) or HF+DSS (orange color) or EX+HF+DSS (blue color) from SC (green color). PC1 explained 59.34% of inter-sample variance and revealed a sharp distinction among the groups of mice ( Figure 5A). Furthermore, the rarefaction curve showed significant differences in species richness among the groups of mice; SC vs. HF (p < 0.001), HF vs. HFD+DSS (p < 0.001), and HF+DSS vs. EX+HF+DSS (p = 0.032) ( Figure 5B).

Exercise Preconditioning Modifies Composition of Gut Microbiota
Simpson indices, as shown in HF vs. SC. Administration of DSS resulted in lower values of OTUs (p < 0.001), Chao1 (p = 0.003), Shannon (p = 0.001) and Simpson (p = 0.003), as shown in HF+DSS vs. HF. On the other hand, exercise preconditioning suppressed decreases in Shannon (p = 0.047) and Simpson (p = 0.028) associated with HF+DSS, with no such suppressive effects on OTUs and Chao1, as shown in EX+HF+DSS vs. HF+DSS. Additionally, unweighted PCoA was conducted to compare the individual groups of mice and showed a clear separation of HF (red color) or HF+DSS (orange color) or EX+HF+DSS (blue color) from SC (green color). PC1 explained 59.34% of inter-sample variance and revealed a sharp distinction among the groups of mice ( Figure 5A). Furthermore, the rarefaction curve showed significant differences in species richness among the groups of mice; SC vs. HF (p < 0.001), HF vs. HFD+DSS (p < 0.001), and HF+DSS vs. EX+HF+DSS (p = 0.032) ( Figure 5B).

Exercise Preconditioning-induced Modifications in Gut Microbiota and Their Potential Associations with HF+DSS-induced Colitis
As illustrated in Figure 5C, analysis of fecal microbiota showed that Bacteroidetes (53.25%), Deferribacteres (5.51%), Firmicutes (26.25%), and Verrucomicrobia (5.97%) were the top four abundant phyla, accounting for 90.98% of the reads. At the phylum level, HF mice had a significantly higher Firmicutes to Bacteroidetes (F/B) ratio (p < 0.001) compared to SC mice. On the other hand, HF+DSS and EX+HF+DSS mice had significantly lower F/B ratios (p < 0.001 and p < 0.001, respectively) compared to HF mice. HF mice had also a higher abundance of Deferribacteres (p = 0.050) and a lower abundance of Actinobacteria (p = 0.050) compared to SC mice. HF+DSS mice had a lower abundance of Proteobacteria (p = 0.050) compared to HF mice. EX+HF+DSS mice had a significantly higher abundance of Verrucomicrobia (p = 0.050) compared to HF+DSS mice (Table 2). As illustrated in Figure 5D, some noticeable modifications in gut microbiota were observed at the species level among the groups of mice. For example, HF+DSS mice had significantly higher Bacteroides vulgatus (p = 0.050) and Escherichia fergusonii (p = 0.050) compared to HF mice. On the other hand, EX+HF+DSS mice had significantly higher Akkermansia muciniphila (p = 0.050) and lower Bacteroides vulgatus (p = 0.050) compared to HF+DSS mice ( Figure 5E-G).

Discussion
In this animal study, we showed that chronic exposure to HF resulted in metabolic complications and increased susceptibility to colitis in wild-type mice. The clinical symptoms of colitis (i.e., weight loss, shortened colon, and enlarged spleen) were exacerbated by mild administration of DSS. On the other hand, exercise preconditioning alleviated the severity of the clinical symptoms of colitis associated with HF+DSS treatment, as evidenced by alleviations in the colitis symptoms, altered expression of gut barrier proteins, infiltration of immunity cells into the colon and blood, and inflammatory and chemotactic responses. In particular, exercise preconditioning-induced alleviations of colitis pathological markers appear to be associated with symbiotic modifications in gut microbiota, although underlying mechanism(s) remain to be elucidated.
With respect to HF and/or HF+DSS, the current findings of the study are in line with previous studies reporting western diet-related susceptibility to colitis [17,18]. Gulhane et al. [17] showed that chronic exposure to HF resulted in impaired intestinal development in mice. Kim et al. [18] also reported that BALB/c mice fed 4 weeks of western-style diet had increased inflammatory responses in conjunction with increased infiltration of macrophages into the colon, which were exacerbated by administration of DSS. With respect to exercise preconditioning, the current findings are consistent with previous studies reporting the therapeutic effects of exercise training against pathogenesis of experimentally-induced colitis models by suppressing pro-inflammatory responses and simultaneously stimulating anti-inflammatory responses [19] in conjunction with enhanced gut barrier function and symbiotic modifications in gut microbiota [11]. In addition, exercise training-induced enhancement of the immune system was observed in both animals [19] and humans [20] with exaggerated or chronic inflammation. In particular, exercise training-induced symbiotic modifications in gut microbiota have been found in animal models of inflammatory disease models, such as colon cancer and IBD [21], ulcerative colitis [22], HF-induced obesity [23], and sepsis [24].
Several explanations can be given for exercise preconditioning-induced alleviation of clinical symptoms of this experimental colitis. First, exercise preconditioning-induced increase of HSP70 protein may enhance its protective function of gut epithelial cells and thereby minimize loss of gut barrier proteins (i.e., ZO-1 and occluding) under the colitis associated with HF+DSS treatments. In support of this notion, an experimental induction of HSP70 protein resulted in protection against