Nutrient composition influences the gut microbiota in chronic thoracic spinal cord-injured rats

Abstract

Chronic spinal cord injury (SCI) results in an increased predisposition to various metabolic problems that can be exacerbated by consuming a diet rich in calories and saturated fat. In addition, gastrointestinal symptoms have been reported after SCI, including intestinal dysbiosis of the gut microbiome. The effects of both diet and SCI on the gut microbiome of adult male Long Evans rats euthanized 16 wk after injury were investigated. The rats were either thoracic spinal contused or received sham procedures. After 12 wk of either a low-fat or high-fat diet, cecal contents were analyzed, revealing significant microbial changes to every taxonomic level below the kingdom level. Shannon α diversity analyses demonstrated a significant difference in diversity between the groups based on the surgical condition of the rats. SCI produced a unique signature of changes in commensal bacteria that were significantly different than Sham. Specific changes in commensal bacteria as a result of diet manipulation had high fidelity with reports in the literature, such as Clostridia, Thiohalorhabdales, and Pseudomonadales. In addition, novel changes in commensal bacteria were identified that are unique dietary influences on SCI. Linear regression analysis on body fat and lean mass showed that a consequence of chronic SCI produces uncoupled associations between some commensal bacteria and body composition. In conclusion, despite tightly controlling the protein content and varying the carbohydrate and fat contents, Sham and SCI rats respond uniquely to diet. These data provide potential direction for therapeutic modulation of the microbiome to improve health and wellness following SCI.

INTRODUCTION

The lifelong deficits are devastating for the ∼300,000 Americans who suffer from traumatic injury to the spinal cord (1). Sensory and motor impairments to the upper and lower limbs result in the loss of independent movement and associated metabolic expenditure. Hence, metabolic problems that are often exacerbated with diets rich in energy-dense nutrients, in parallel with reduced physical activity, are elevated in this population (2–4). In addition, nearly two-thirds of the SCI population is plagued with the various comorbidities of Metabolic Syndrome, which include obesity, type II diabetes, hyperlipidemia, and cardiovascular disease (3). These factors contribute to the already higher-than-average mortality rate for spinal cord injury (SCI) individuals, particularly in the first 2 yr after injury; overall, a reduced life expectancy persists for these individuals (5, 6). Furthermore, depending on the level of injury to the spinal cord, diverse neural damage to the visceral organs may cause metabolic dysfunction. Given the varied breadth of traumatic injuries possible to the spinal cord, these deficits are somewhat unique to each individual.

Persons with SCI may experience significant dysfunction of the digestive system. Damage to the visceral innervation, which may include damage to the enteric nervous system, the spinal cord’s communication with the enteric nervous system, or mismatch between the brainstem (vagal) and spinal cord control of the gut, results in reduced gastric emptying, along with altered peristalsis (7). Thus, SCI is associated with chronic problems with gastrointestinal (GI) symptoms, such as fecal incontinence and constipation (8, 9). Furthermore, intestinal dysbiosis, an imbalance of the gut’s microbial population leading to increased immuno-metabolic disease, has been recently described following SCI (10). Since intestinal dysbiosis is further exacerbated with diets high in fat and high-glycemic carbohydrates, these effects are only compounded in individuals with a preexisting SCI.

In recent decades, the exploration of the microbiota of the GI tract has been investigated to understand the diverse populations that coexist in the gut (11, 12) and modulate the populations to produce better metabolic health (13, 14). In addition, these studies have provided insight into bacterial complexity based on the type of diet consumed by the host. Though these studies can be more tightly controlled using rodent animal models, these changes are more difficult to extrapolate in the human population, which consumes a greater diversity of foods.

Few studies have addressed the knowledge gap regarding the investigation of the microbiome following traumatic SCI. Studies that employed a T10 weight-drop spinal contusion model in rats 8 wk after injury showed significant differences in the β diversity of the species, genus, and family levels (15). In this study, inflammatory markers, IL1β, IL12, and MIP2 correlate significantly with β diversity (15). In a mouse study using a controlled-contusion model in the mid-thoracic region, interesting associations showed an inverse correlation between specific taxa and injury status (16). Further, in a small human cohort study comparing cervical/thoracic SCI and lumbar SCI to able-bodied controls, the SCI groups demonstrated microbiota profiles associated with metabolic dysfunction (17). In total, these limited preclinical and human studies emphasize that greater characterization of the microbiome following traumatic injuries such as SCI may assist in attaining better metabolic health for this vulnerable population.

Our work here focuses uniquely on the enduring changes to the microbiome significantly after the acute rehabilitation phase that, for many patients, would be experienced in a hospital and specialized rehabilitation center consuming standardized diets. In the current study, we performed 16S rRNA gene sequencing of the cecal contents of T10 contused rats compared with controls 16 wk after injury. Directly following injury, animals were maintained on a chow diet. Then on postinjury week 4, animals were switched to either a high-butter-fat diet (HFD) or protein-matched, low-fat diet (LFD) for 12 wk before sampling. We report changes to the various taxa levels as a result of both diet and injury. Further, we performed analyses associating body fat and lean body mass to the various normalized counts of the taxa. Our studies show that SCI and diet impact the microbiota in unique ways that do not considerably overlap with able-bodied Sham controls.

MATERIALS AND METHODS

Animal Assurance

All procedures for animal use complied with the Guidelines for the Care and Use of Laboratory Animals by the National Research Council and were reviewed and approved by the University of Mississippi Medical Center (UMMC) Institutional Animal Care and Use Committee (IACUC #1469) and the US Army Animal Care and Use Review Office (ACURO). In conducting research using animals, the investigators adhered to the laws of the United States and regulations of the Department of Agriculture.

Animals

Cecal samples used for this study were obtained from rats whose phenotype was previously published (18, 19). Briefly, male Long Evans rats (250–300 g; Envigo, Indianapolis, IN) were initially housed and maintained in the UMMC vivarium on a 12:12-h light-dark cycle at 25°C and 50%–60% humidity. Rats had ad libitum access to water and standard chow (Cat. No. 8640, Envigo, 3.0 kcal/g; 17% fat, 54% carbohydrate, 29% protein). Rats were assigned to either Sham-laminectomy (Sham) or thoracic spinal cord injury (SCI) groups as previously described (18). Following surgery, rats were weight-stabilized for 28 days. After 4 wk of recovery, rats were assigned to one of two protein-matched diets: high-fat diet (HFD) (Cat. No. D03082706, Research Diets, New Brunswick, NJ, 4.54 kcal/g; 40% fat, 45% carbohydrate, 15% protein), or low-fat diet (LFD) (Cat. No. D03082705, Research Diets, New Brunswick, NJ, 3.81 kcal/g, 9% fat, 76% carbohydrate, and 15% protein) for 12 wk. GPower software was used to determine sample size. Power analyses show that an n = 8 for Shams and n = 8 for SCI was needed to achieve (1 − β) = 0.80 with a probability of 0.05 for metabolic endpoints. Final n sizes for the study were: Sham-LFD (n = 8), SCI-LFD (n = 10), Sham-HFD (n = 8), and SCI-HFD (n = 8).

Surgical Procedures

All surgical procedures were performed on animals as previously described (18). Using an Infinite Horizon Spinal Impactor Device (Precision Systems and Instrumentation, LLC, Fairfax Station, VA), moderate contusion injuries were delivered to the T10 spinal cord using 150 kdyn of force with a 1-s dwell. For the Sham surgery, a laminectomy was performed at T10 vertebrae, and then the overlying muscles were sutured and the skin securely closed using stainless steel wound clips.

Postoperative Care

Animals received one dose of buprenorphine SR (Sustained Release; 1.0 to 1.2 mg/kg SQ (ZooPharm, Laramie, WY) and 72 h later, single-dose buprenorphine for postsurgical pain management (0.025 mg/kg, twice daily for a period of 2 d, then as needed). Animals also receive 1) antibiotic, naxcel (5 mg/kg SQ, Zoetis, NJ) once daily for a period of 5 days, and 2) 3–5 mL of 0.9% saline, twice daily for a period of 3 days to ensure hydration. Bladder care was performed 2–3 times daily for ∼10 d and was discontinued for an animal when it exhibited an already voided bladder on two consecutive bladder care sessions.

Body Weight and Composition

Longitudinal body weights were measured as previously described (18). In addition, lean and fat mass were analyzed using Echo Magnetic Resonance Imaging (echoMRI; EchoMedical Systems, Houston, TX) (18).

rRNA Gene Sequencing

Cecal samples were provided to the UMMC Molecular and Genomics Core Facility to isolate DNA and 16S rRNA gene sequencing. Core operators were blinded to the groups and treatments. DNA was isolated using MoBio PowerMag Soil Kit (recommended for the isolation of DNA from stool samples) on the KingFisher Flex automated DNA/RNA isolation system (96 samples through-put). Subsequently, samples underwent an initial quality control step to determine DNA concentration (Nanodrop One and Qubit Fluorimeter) and integrity (Qiagen QIAxcel Advanced system). Samples that passed quality parameters (minimum concentration and quality DNA) were used to amplify the 16S V3 and V4 region bacterial ribosomal RNA using 16S Amplicon PCR Forward Primer = 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG and 16S Amplicon PCR Reverse Primer = 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC using a limited cycle PCR per Illumina protocol. Illumina sequencing adapters and dual-index barcodes were added to the amplicon target using Illumina NexteraXT indices (Illumina, San Diego, CA) that allow up to 384 samples to be indexed and run in a single library preparation (Illumina, San Diego, CA). For the current experiment, n = 33 samples were pooled into a single library for sequencing. The library was sequenced using the Illumina iSeq100 instrument. The average raw read depth per sample was 151,884, and after quality filtering, the read depth was 132,329. Thus, 87% of the reads were included.

Microbiome Data Processing and Analyses

The sequencing reads were automatically uploaded and evaluated for quality using the Illumina BaseSpace Cloud. The BaseSpace QIIME analysis pipeline (http://qiime.org/) was used for demultiplexing and quality filtering, OTU picking, taxonomic assignment, phylogenetic reconstruction, and diversity analyses. Sequencing reads were classified against the Greengenes database (http://greengenes.lbl.gov/cgi-bin/nph-index.cgi). The *.biom file generated from the QIIME output was analyzed using online MicrobiomeAnalyst (https://www.microbiomeanalyst.ca/) for statistical analysis and data visualization following default parameters.

Post Hoc Microbiome Analyses

Taxonomic level-specific data were downloaded into Microsoft Excel from MicrobiomeAnalyst. In Excel, microbiota counts for each taxonomic level were sorted into the host subjects’ respective groups (i.e., SCI-LFD, Sham-LFD, SCI-HFD, and Sham-HFD). At this time, the Core director unblinded the groups. Total number of microbiota per group was summed for each taxonomic level, and the individual bacteria counts were normalized to the percentage of the total number of microbiota counts for the individual subject. After normalizing the data, Student’s t tests were performed to compare Sham individuals to all SCI individuals (Fig. 3), Sham-LFD to SCI-LFD and Sham-HFD to SCI-HFD (Fig. 4), and Sham-LFD to Sham-HFD and SCI-LFD to SCI-HFD (Figs. 5 and 6) for each individual normalized bacteria count at all taxonomic levels. Finally, we performed simple linear regression analyses to determine associations between the normalized counts for each bacteria at each taxonomic level with body fat mass and body lean mass. These findings are reported in Tables 1 and 2.

Table 1.

Linear regression analysis of taxa for body fat for all four groups

Taxa	Name	F	R 2	P
Body fat	Sham-LFD
Order	Rhodobacterales*	8.67	0.59	0.03
Family	Rhodobacteraceae	8.38	0.58	0.03
Genus	Catonella	7.24	0.55	0.04
	Ehrlichia	6.75	0.53	0.04
	Holdemania	6.62	0.52	0.04
	Oribacterium*	6.40	0.52	0.04
Body fat	SCI-LFD
Order	Rickettsiales	6.47	0.45	0.03
Family	Anaplasmataceae	5.92	0.43	0.04
	Paenibacillaceae	11.45	0.59	0.01
Genus	Ehrlichia	5.35	0.40	0.05
	Roseburia	6.52	0.45	0.03
	Tenacibaculum	5.75	0.42	0.04
Body fat	Sham-HFD
Order	Desulfuromonadales*	13.00	0.68	0.01
	Entomoplasmatales	8.37	0.58	0.03
	Neisseriales	6.31	0.51	0.05
Family	Burkholderiaceae*	6.14	0.51	0.05
	Entomoplasmataceae*	8.37	0.58	0.03
	Erysipelotrichaceae*	6.01	0.50	0.05
	Lachnospiraceae*	7.20	0.55	0.04
	Neisseriaceae	6.30	0.51	0.05
Genus	Alkalibacterium	23.10	0.79	0.00
	Alkaliphilus*	9.75	0.62	0.02
	Anaerobacillus	22.18	0.79	0.00
	Bacillus	11.05	0.65	0.02
	Blautia*	10.49	0.64	0.02
	Delftia	9.15	0.60	0.02
	Desulfotomaculum	9.02	0.60	0.02
	Erysipelothrix	6.43	0.52	0.04
	Eubacterium*	11.99	0.67	0.01
	Fervidobacterium	7.51	0.56	0.03
	Kitasatospora	6.73	0.53	0.04
	Megasphaera	10.55	0.64	0.02
	Mesoplasma	11.60	0.66	0.01
	Mesorhizobium	8.51	0.59	0.03
	Moryella	7.92	0.57	0.03
	Oscillospira*	13.75	0.70	0.01
	Pedobacter	6.75	0.53	0.04
	Propionibacterium	10.23	0.63	0.02
	Rhizobium	9.96	0.62	0.02
	Sphingomonas	10.65	0.64	0.02
	Staphylococcus	7.67	0.56	0.03
	Sutterella*	9.33	0.61	0.02
	Thiothrix	6.03	0.50	0.05
Body fat	SCI-HFD
Order	Chromatiales	8.80	0.64	0.03
	Mycoplasmatales*	12.88	0.72	0.02
	Syntrophobacterales	7.95	0.61	0.04
Family	Burkholderiaceae*	12.21	0.71	0.02
	Chromatiaceae	11.26	0.69	0.02
	Dehalobacteriaceae	11.65	0.70	0.02
	Mycoplasmataceae*	13.32	0.73	0.01
Genus	Candidatus Blochmannia*	71.27	0.93	0.00
	Dehalobacterium	11.52	0.70	0.02
	Lautropia*	11.94	0.70	0.02
	Mycoplasma*	13.07	0.72	0.02
	Pectinatus	31.99	0.86	0.00

*Bacteria had normalized counts at or above 0.1. N = 8–10 male subjects per group. Statistical significance was determined by simple linear regression, with statistical significance as P < 0.05. Gray cells indicate a positive slope, whereas white cells indicate a negative slope. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

Table 2.

Linear regression analysis of taxa for body lean for all four groups

Taxa	Name	F	R 2	P
Body lean	Sham-LFD
Order	Bacillales	7.80	0.57	0.03
	Neisseriales	6.64	0.53	0.04
	Rhodobacterales	6.81	0.53	0.04
Family	Neisseriaceae	6.79	0.53	0.04
	Rhodobacteraceae	7.00	0.54	0.04
	Syntrophaceae	6.46	0.52	0.04
Genus	Catonella	42.88	0.88	0.00
	Erysipelothrix	13.25	0.69	0.01
	Micrococcus	12.24	0.67	0.01
	Oribacterium*	33.97	0.85	0.00
Body lean	SCI-LFD
Order	Lactobacillales*	6.98	0.47	0.03
Genus	Anaerovibrio	6.19	0.44	0.04
	Odoribacter	8.15	0.50	0.02
Body lean	Sham-HFD
Class	Deltaproteobacteria*	19.00	0.76	0.00
	Erysipelotrichi	11.78	0.66	0.01
	Fusobacteria	6.60	0.52	0.04
	Holophagae	6.31	0.51	0.05
Order	Bacillales	6.13	0.51	0.05
	Bdellovibrionales	11.02	0.65	0.02
	Desulfovibrionales*	18.18	0.75	0.01
	Erysipelotrichales*	11.79	0.66	0.01
	Gemellales	16.84	0.74	0.01
	Sphingomonadales	6.75	0.53	0.04
	Thiotrichales	9.17	0.60	0.02
Family	Aerococcaceae	11.04	0.65	0.02
	Bdellovibrionaceae	11.03	0.65	0.02
	Erysipelotrichaceae*	37.52	0.86	0.00
	Gemellaceae	16.87	0.74	0.01
	Heliobacteriaceae	9.84	0.62	0.02
	Lachnospiraceae*	20.50	0.77	0.00
	Micrococcaceae	11.40	0.66	0.01
	Mycobacteriaceae	9.62	0.62	0.02
	Oxalobacteraceae	9.37	0.61	0.02
	Sphingomonadaceae	6.77	0.53	0.04
	Thiotrichaceae	9.29	0.61	0.02
	Veillonellaceae*	6.28	0.51	0.05
Genus	Alkalibacterium	13.56	0.69	0.01
	Anaerobranca	6.11	0.50	0.05
	Bdellovibrio	10.99	0.65	0.02
	Blautia*	13.39	0.69	0.01
	Dorea*	6.63	0.52	0.04
	Erysipelothrix	9.97	0.62	0.02
	Eubacterium*	19.89	0.77	0.00
	Gemella	13.79	0.70	0.01
	Heliorestis	9.91	0.62	0.02
	Lautropia*	12.02	0.67	0.01
	Micrococcus	11.54	0.66	0.01
	Mycoplasma*	6.32	0.51	0.05
	Propionibacterium	6.85	0.53	0.04
	Sarcina	6.91	0.54	0.04
	Thiothrix	10.34	0.63	0.02
Body lean	SCI-HFD
Family	Comamonadaceae	14.96	0.75	0.01
	Desulfovibrionaceae*	7.96	0.61	0.04
	Synergistaceae	38.13	0.88	0.00
Genus	Candidatus Tammella	24.98	0.83	0.00
	Desulfosporosinus	12.88	0.72	0.02
	Desulfovibrio*	7.75	0.61	0.04
	Leuconostoc	6.63	0.57	0.05
	Sphingobium	9.87	0.66	0.03
	Thiomonas	28.07	0.85	0.00

*Bacteria had normalized counts at or above 0.1. n = 8–10 male subjects per group. Statistical significance was determined by simple linear regression, with statistical significance as P < 0.05. Gray cells indicate a positive slope, whereas white cells indicate a negative slope. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

Statistical Analysis

All statistical analyses were performed using GraphPad Prism version 9.1.2 (GraphPad Software, San Diego, CA). Two-way ANOVA was used to observe differences of diet and injury. The results are given as means ± SE. Results were considered statistically significant when P < 0.05. Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/) was used to produce comparisons and generate Venn diagrams (20). Associations were analyzed by performing simple linear regressions using GraphPad. Simple linear regressions were performed with normalized counts obtained by dividing the specific bacterial raw counts over the total raw count from each individual sample. Comparisons were made using body fat and body lean mass, from all four experimental groups, or only the two Sham groups, or only the two SCI groups. Inclusion criteria for the results included the association having a P value below 0.05, having an R² value above 0.2, and the bacteria having no normalized counts of 0. β Diversities were determined using permutational analysis of variance (PERMANOVA), which is a nonparametric multivariate statistical test that tests the null hypothesis that the centroids and dispersion of the groups of bacteria are equivalent for all groups.

RESULTS

SCI rats 16 wk postinjury had reduced body weight in comparison to Sham rats, P (injury) < 0.01 (Fig. 1A). Rats consuming HFD for 12 wk had a significantly increased body weight in comparison to LFD-fed rats, P (diet) < 0.05 (Fig. 1A). When examining average body fat mass, HFD-fed animals had increased fat mass compared with LFD-fed rats, P (diet) < 0.05 (Fig. 1B). However, when analyzing average lean mass, there was an effect of injury, where Sham animals had significantly higher lean mass than their SCI counterparts, P (injury) < 0.0001 (Fig. 1C).

Figure 1.

Body composition parameters. A: terminal body weight in grams. Post hoc tests showed a significant difference between Sham-HFD and SCI-LFD groups, where Sham-HFD was significantly higher. B: average body fat mass in grams. C: average body lean mass in grams. Post hoc tests revealed that there were significant differences between Sham-HFD and SCI-HFD, Sham-LFD and SCI-LFD, SCI-LFD and Sham-LFD, and Sham-LFD and SCI-HFD. n = 8–10 male subjects per group. Data are presented as means ± SE. Statistical significance was determined by two-way analysis of variance followed by Tukey’s post hoc test. Statistical significance was determined as *P < 0.05, **P < 0.01. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

Animals were euthanized 16 wk after injury, and contents were collected from the cecum. Bacterial DNA was extracted and subsequently processed for 16S rRNA gene sequencing. Analyses of α diversity showed similar read diversity between all groups (Fig. 2A). However, Shannon α diversity analyses demonstrated a significant difference in diversity between the groups based on the surgical condition of the rats, P (injury) < 0.01 (Fig. 2B). In addition, there was a significant difference between Sham-HFD and SCI-HFD following Tukey’s post hoc, P < 0.05 (Fig. 2B). Unlike α diversity, which only measures the abundance of all bacteria, Shannon diversity also considers each species type. So, the SCI animals had a greater richness in their Shannon α diversity than their Sham counterparts (Fig. 2B). When comparing the β diversities of all the groups by permutational multivariate analysis of variance (PERMANOVA), there were significant differences among the four cohorts, P < 0.01 (Fig. 2C). Although α diversity measures the richness within a single sample, β diversity examines the variability between the samples. When comparing Sham and SCI groups that were both fed LFD, no significant difference in the β diversities was observed (Fig. 2D). However, when comparing Sham and SCI groups that were fed HFD, a significant difference in β diversity was observed, P < 0.05 (Fig. 2E). The normalized percentage of the five most abundant phyla is presented in Fig. 2F When comparing LFD and HFD phyla among Sham animals, only Bacteroidetes was significantly different (P < 0.05; Fig. 2F), whereas in the SCI groups, three separate phyla, Bacteroidetes, Firmicutes, and Deferribacteres, were significantly different, P < 0.05 (Fig. 2F). In comparing the LFD-fed groups, only Verrucomicrobia was significantly different with SCI rats, demonstrating higher levels of this specific bacteria, P < 0.05 (Fig. 2F). In comparing phyla within the HFD-fed groups, there were no differences between Sham and SCI (Fig. 2F). The Firmicutes to Bacteroidetes ratio, which is often altered in obesity (21, 22), was reduced in HFD-fed animals, P (diet) < 0.01 (Fig. 2G). The Prevotella to Bacteroides ratio, which has been linked to glucose sensitivity in mice (23), is significantly elevated in SCI rats compared with Sham, P (injury) < 0.05 (Fig. 2H).

Figure 2.

α and β diversity. A: α diversity observed. B: Shannon plot of α diversity. Statistical significance was determined by two-way analysis of variance followed by Tukey’s post hoc test and Student’s t test. Post hoc tests revealed significant differences between the Sham-HFD and SCI-HFD groups; *P < 0.05. C: β diversity observed. Data are presented as means ± SE; P < 0.01. D: β diversity observed when comparing the two LFD groups. E: β diversity observed when comparing the two HFD groups; P < 0.05. F: major phyla identified and their abundances in each group; P < 0.05. G: Firmicutes to Bacteroidetes ratio. H: Prevotella to Bacteroides ratio. n= 8–10 male subjects per group. Data are presented as means ± SE. Statistical significance was determined by two-way analysis of variance followed by Tukey’s post hoc test. In C–E, diversities were determined using permutational analysis of variance (PERMANOVA). Statistical significance was determined as P < 0.05. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

Next, we examined significant differences in the taxonomic levels between Sham and SCI groups disregarding their postinjury diet. There were 2 classes (Fig. 3A), 9 orders (Fig. 3B), 16 families (Fig. 3C), and 17 genera (Fig. 3D) that were significantly different, P < 0.05. Taxa that were significantly higher in SCI animals are denoted with a black bar, whereas taxa that were significantly lower in SCI animals are denoted with a white bar (Fig. 3).

Figure 3.

Differences by injury alone. Significant differences found when comparing animals who received Sham surgery versus SCI. Graphs are illustrated using a logarithmic scale. A: class. B: order. C: family. D: genus. Black striped bars show classifications that are higher in SCI animals than in Sham. White striped bars show classifications that are lower in SCI animals than in Sham. n = 8–10 male subjects per group. Statistical significance was determined by unpaired Student’s t test with 2-tailed distribution. *P < 0.05, ** P < 0.01. SCI, spinal cord injury.

When we examined differences in the bacterial populations between Sham and SCI grouped by diet, we found significant differences at the order, family, and genus levels. These shifts are shown in Fig. 4. Significant differences between Sham-LFD and SCI-LFD are presented in Figs. 4, A, C, and E for the taxonomic classification of order, family, and genus, respectively. On the other hand, significant differences between Sham-HFD and SCI-HFD are presented in Fig. 4, B, D, and F for the taxonomic classification of order, family, and genus, respectively. Interestingly, diet uniquely influences SCI animals in that there is no overlap in the significant microbiota that appear under LFD conditions versus HFD conditions. In other words, no significant differences in Sham-LFD versus SCI-LFD overlapped with significant differences in Sham-HFD versus SCI-HFD.

Figure 4.

Comparison of Sham and SCI with respect to diet. Significant differences found in animals fed a LFD, dependent on injury illustrated on a logarithmic scale. A: order. C: family. E: genus. Significant differences found in animals fed a HFD dependent on injury illustrated on a logarithmic scale. B: order. D: family. F: genus. n = 8–10 male subjects per group. Statistical significance was determined by unpaired Student’s t test with 2-tailed distribution. *P < 0.05, ** P < 0.01. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

We further examined differences in the bacterial populations between LFD and HFD grouped by injury. We specifically compared Sham-LFD versus Sham-HFD and SCI-LFD versus SCI-HFD. We found significant differences at the phyla, class, order, family, and genus levels. These differences are demonstrated in Fig. 5. Only one phylum is significantly different in Sham-LFD versus Sham-HFD (Fig. 5A), whereas three are different in SCI-LFD and SCI-HFD (Fig. 5B). Bacteroidetes is common in both data sets, as indicated by the Venn diagram (Fig. 5C). Three classes are significantly different comparing Sham-LFD versus Sham-HFD (Fig. 5D), whereas three are different in SCI-LFD versus SCI-HFD (Fig. 5E). Clostridia and Bacteroidia overlap in both data sets (Fig. 5F). Seven orders each are significantly different, comparing Sham-LFD versus Sham-HFD (Fig. 5G) and SCI-LFD versus SCI-HFD (Fig. 5H). Clositridiales and Bacteroidales overlap in both data sets (Fig. 5I). With respect to the following taxa, Sham-LFD and Sham-HFD differ in ten different families (Fig. 5J), whereas 13 are different in SCI-LFD versus SCI-HFD (Fig. 5K). These overlap in two groups, Bacteroidaceae and Ruminococcaceae (Fig. 5L). Finally, differential counts in genera are shown in Figs. 6, A and B. Four overlapping genera were identified: Dorea, Phascolarctobacterium, Aggregatibacter, and Mannheimia (Fig. 6C).

Figure 5.

Comparison of LFD and HFD with respect to injury. Significant differences found in Sham animals, dependent on diet illustrated on a logarithmic scale. A: phylum. D: class. G: order. J: family. Significant differences found in SCI animals dependent on diet illustrated on a logarithmic scale. B: phylum. E: class. H: order. K: family. Respective Venn diagrams comparing significant differences found in Sham and SCI subjects. C: phylum. F: class. I: order. L: family. N = 8–10 male subjects per group. Statistical significance was determined by unpaired Student’s t test with 2-tailed distribution. *P < 0.05, **P < 0.01, ***P < 0.001. HFD, high-fat fiet; LFD, low-fat diet; SCI, spinal cord injury.

Figure 6.

Differences by injury and diet at the genus level. A: significant differences found in Sham animals dependent on diet illustrated on a logarithmic scale. B: significant differences found in SCI animals dependent on diet illustrated on a logarithmic scale. C: respective Venn diagram comparing significant differences found in Sham and SCI subjects. n = 8–10 male subjects per group. Statistical significance was determined by unpaired Student’s t test with 2-tailed distribution. *P < 0.05, **P < 0.01, ***P < 0.001. SCI, spinal cord injury.

We next were interested in understanding the relationship between normalized counts of each bacteria at the different taxonomic levels and body fat for each of the groups separately (Table 1). Therefore, we performed linear regression analyses of body fat and the following groups: Sham-LFD, Sham-HFD, SCI-LFD, and SCI-HFD. Genera Ehrlichia overlaps between Sham-LFD and SCI-LFD (P < 0.05), and the family Burkholderiaceae overlaps between Sham-HFD and SCI-HFD (P < 0.05).

Finally, we performed a similar analysis, this time comparing normalized counts of different bacteria at the different taxonomic levels and body lean mass for each of the groups separately (Table 2). Again, order Bacillales and genera Erysipelothrix and Micrococcus overlap between Sham-LFD and Sham-HFD (P < 0.05).

DISCUSSION

This study examined whether two unique diets varying by fat and carbohydrate content causes differential shifts to the microbiome and, further, whether the specific shifts in Sham or SCI animals vary with body composition. Significant differences were identified by both diet and injury. Overall, greater associations were identified between specific microbes and fat and lean mass in the Sham-HFD animals than in the other groups.

The microbiome sequencing data were generated using cecal contents from chronically injured animals having consumed their respective diets for over three months. α Diversity, a measure of the microbial variation within a single sample, showed some individual variability that did not correspond with injury status and diet type. However, the Shannon analysis demonstrates the increased richness in diversity in the SCI rats compared with Sham. The mismatch between the α diversity and the Shannon analysis suggests that SCI samples have a different composition than the microbial variations. These differences were not reported in the previous rodent studies using SCI intestinal samples (15, 16, 24). However, these previously published studies were performed using a typical rodent chow diet and at earlier time frames following injury [i.e., 21 days (24) and 35 days (16) in mice and 8 wk (15) in rats]. In recent work comparing patients with SCI with a broad range of injury levels to healthy controls, no differences in α diversity were reported. However, principal coordinate analysis (pCOA) performed at the genus level showed a significant separation in microbial profiles between patients with SCI and healthy controls (25). This human study focused on the acute phase of SCI recovery, obtaining samples predominantly within the first week at the rehabilitation center and no greater than 60 days after injury (25). Thus, our study focuses on a time frame after the injury that is chronic and has not previously been investigated or reported.

Many studies report a plethora of differences when comparing two unique diets (e.g., altering the fat and carbohydrate content) (26–31). At the phylum level, microbial samples obtained from subjects consuming a diet rich in fats compared with subjects having consumed diets with reduced fat content frequently show elevations in the ratio between Firmicutes and Bacteroidetes (32). In contrast, some studies suggest a reduction in the Firmicutes/Bacteroidetes ratio (21, 22), which is consistent with the findings reported in our current study. The difference we report regarding the Firmicutes/Bacteroidetes ratio may reflect the greater level of high-glycemic carbohydrate in the form of cornstarch in the LFD. In addition, the source, i.e., lard, and percentage of total fat (60%) differ in most rodent microbiome studies compared with the current study. At the genus level, the Prevotella/Bacteroides ratio was significantly elevated in animals with an SCI. This ratio is thought to be important in determining how successful a person will be at maintaining weight loss (33). It is also thought that lower Prevotella than Bacteroides are associated with glucose intolerance in mice (23).

When we performed analyses strictly comparing the two injury states (Sham and SCI) without considering diet (Fig. 3), we found significant differences below the phylum level. At the class level, Flavobacteriia and Erysipelotrichi were significantly lower in SCI rats compared with Sham. Flavobacteriia is lower in humans with type 2 diabetes (34). Individuals with SCI have a higher incidence of type 2 diabetes than able-bodied controls (34–38). In our dataset, Flavobacteriia were significantly reduced in the SCI groups. Higher Erysipelotrichi counts have been associated with obesity in one human study (39). In our study, Erysipelotrichi counts were higher in Sham animals compared with SCI; Sham animals were significantly heavier than SCI animals. There was no effect of Erysipelotrichi based on diet.

Certain bacteria that are elevated in SCI compared with Sham do have associations with metabolic syndrome phenotypes in the literature. For example, Pseudomonadales is increased in humans with obese compared with controls (40). Xanthomonadaceae is enriched in the salivary microbiome of humans with obese compared with controls (41). Ruminococcus is elevated in type 2 diabetes and produce proinflammatory states within the intestines (42). Many microbiome reports have been associative studies, and it remains unknown whether these specific bacteria are causal to the disease state or appear as a consequence of the underlying metabolic condition.

Previous studies examining the human microbiome during an injury state (i.e., trauma, burns, sepsis, and surgical injuries) show an increase in bacteria belonging to the Proteobacteria phylum (43), such as Pseudomonadales, Thiohalorhabdales, and Xanthomonadales. These were all increased in the SCI group compared with Sham in our data set, showing fidelity between these previously published studies and our data set. The great contrast between the already published studies using a rodent model of SCI (15, 16, 24) and the current work is that we are focusing on a time point representing chronic injury and thus demonstrate some enduring changes to the microbiome following SCI.

Given the profound impact that varying fat and carbohydrate content have on the microbiome, it is not surprising that the differences between Sham and SCI within each diet group did not overlap at any taxonomic level (Fig. 4). Said differently, the impact of the specific diet was unique when evaluating the injury groups. Thus, the changes in the specific microbiota in the SCI groups may result from interaction between the specific microbe and its influence on gastric emptying or peristalsis within the SCI animals, which are altered following injury to the cord in other studies (7, 44, 45).

Phylum Bacteroidetes, class Bacteroidia, order Bacteroidales, and family Bacteroidaceae were all significantly elevated in Sham-HFD and SCI-HFD when compared with their LFD counterparts. When Clostridia was interrogated first among Sham groups and then among SCI groups, reduced Clostridia was identified in HFD-fed animals compared with LFD-fed animals. These data align with other studies that showed Clostridia reductions with increasing BMI and obesity (46). At the genus level, Dorea levels are increased in both Sham-HFD and SCI-HFD animals. This is supported by the literature, which suggests that diets high in saturated fats result in elevations in Dorea (47), and Dorea is significantly elevated in patients with nonalcoholic fatty liver disease (NAFLD) (48).

These two diets influence body mass and composition. These changes have been reported previously (18, 19, 49). In summary, significant loss of lean body mass persists in SCI rats, and greater weight is gained on the palatable HFD over the less calorically dense LFD. Though body weights were, in fact, different as a result of diet and injury, we focused on dissecting the separate effects that body fat and lean mass had on the relative abundance of the various microbes within the taxonomic levels. The interest in the current study in lean body mass stems from the fact that injury to the cord results in atrophy of the musculature and reduced mass and density of the bone in SCI (50, 51). The relationship between bacteria and lean body mass composition has largely been overlooked, and body weight measurements serve as a proxy in many studies.

Sham animals exhibit the greatest number and strength of associations between body fat and groups of microbiota. In our data set, body fat is positively correlated with Rhodobacterales, Pseudomonadaceae, Rhodobacteraceae, and Lautropia, which has been previously shown in other studies using rodent models of obesity (52) or cohorts of humans with robust ranges of body size (53, 54). In addition, the relationship between Olivibacter and Roseburia (Roseburia faecis) and fat mass in SCI rats is negatively associated. However, in Sham rats, there is no association between these microbes and body fat. Limited reports exist on these particular genera. In other studies, Olivibacter is reduced following HFD feeding and is negatively correlated with weight gain and positively correlated with the presence of fecal propionate and acetate (55). These loosely indicate potential reasons for this relatively strong relationship in SCI rats, but more studies would be needed to understand this. Similarly, a negative association exists between Roseburia genus and body fat in SCI rats. Reductions in Roseburia appear to be linked to disruption of the intestinal barrier and dysbiosis (56). Since intestinal dysbiosis has received recent attention following injury to the cord (10, 16), the reduction in the abundance of Roseburia with increasing adiposity may indicate reduced gastrointestinal health (56).

Remarkably few studies in humans take into account lean body mass percentage. Similarly, few descriptive studies in rodents dive deeply into body composition and bacterial associations. Some work has identified that metabolically increasing gut bacterial short-chain fatty acid production may positively affect skeletal muscle mass and physical function in humans by ingesting high-fiber diets and increasing exercise (57, 58). Among the four groups, none of the significant linear associations overlapped. However, the number of positively associated microbes with lean body mass (white cells) poses the hypothesis that driving increases in these bacteria diet modulation may improve the quality and quantity of lean body mass. This requires rigorous testing.

Remarkably fewer associations were significantly correlated with any measures from SCI animals. It appears that the predictive relationship between measurements of body size and adiposity and the microbe is uncoupled when including measures from the SCI rats. Body size and composition in human SCI correlative to Metabolic Disease has recently been contested. For instance, BMI cut-offs for noninjured individuals (>25 overweight, >30 obese) do not well identify SCI persons at risk for obesity and metabolic disease because persons with SCI have significantly less lean body mass and 13% more fat per unit of BMI compared with noninjured controls (59). Instead, a BMI > 22 has been proposed to detect individuals with SCI at high risk for obesity and related diseases (60). The discordance between BMI and health risks after SCI is also supported by the findings that leptin, which in noninjured is positively correlated with BMI and body weight, in persons with SCI, is best explained by waist circumference and visceral fat area (61). In the future, it may be more plausible to use microbial biomarkers with greater concordance as evidence for disease risk in individuals with SCI than any anthropometric measurement.

In conclusion, the nutrients taken in by an organism would seem inherently important to influence the growth or decline of certain bacteria. Though similarities are observed in Sham and subjects with SCI consuming the same diet, the unique differences that filter to the top suggest that the neural changes to the innervation of the musculature and visceral organs may significantly change the physiology of the organ system, resulting in a unique interplay between the microbiota and disease state following SCI.

Hence, specific changes to physiology that are unique to SCI are not able to be controlled in this study. These include reduced gastrointestinal emptying, speed of digestive matter moving through the GI tract, the amount of locomotor activity performed by the animals, and the loss of overall muscle mass. These influences appear to affect the various bacterial populations, producing very different pressures to expand or contract certain microbial populations.

In the past, linear regression modeling comparing lean and obese states and various commensal bacteria populations produced associations between BMI and certain bacteria. Our data bring to the forefront whether BMI or body composition is as tightly linked to certain bacteria as has been previously reported. Overall, associations that seemed in the literature tightly controlled by body weight and/or composition are uncoupled in SCI animals. These lead us to question whether they indeed are markers of body size and, rather, greater indicators of other physiological processes.

Caveats and Future Directions

The studies we report here have some limitations that need to be addressed in future studies. The animals in this study were treated with naxcel to prevent bladder infection. Naxcel has been extensively utilized to treat domesticated farm animals and laboratory rodents, and its effects on the gut microbiome are well characterized. Studies investigating single and multiple exposures suggest changes to the microbiome, specifically with E. coli populations (62). However, in-depth analysis of the microbiome shows a return to baseline by 2 wk after treatment (62). Given that our studies evaluate the microbiome 16 wk postinjury, the impact of naxcel treatment is expected to be minimal. The current study design lacked a chow control group, having received both Sham and SCI surgery.

We believe some of the differences we report concerning others in the literature directly result from lacking this type of baseline information. Our study does not include female rats, whose microbiota may be greatly influenced by shifting reproductive hormones and may be lacking in their translatability. Our study only contains one-time point, which would only reveal chronic changes to the microbiome and does not acknowledge short-term changes caused by diet or injury state. Although other studies have used fecal matter to examine the gut microbiome, we used cecal contents. By using cecal material, we captured both the aerobic and anaerobic bacteria, whereas fecal material mainly includes anaerobicpopulations and, less abundantly, the aerobic populations. This contributes to some of the differences in overall populations reported. Furthermore, we did not collect the cecum to perform mRNA and protein analysis. Investigation of the cecum by mRNA or protein is not commonly performed. Finally, mechanistic studies, which specifically test biologically significant changes produced by perturbing individual microbe populations, are lacking in this study, as well as many others reporting microbiome differences. These types of mechanistic studies may include varying types of fiber supplements, which are known to alter microbiome compositions. Furthermore, probiotics, prebiotics, postbiotics, and synbiotics can be manipulated to improve the gut-brain axis positively; these positive benefits to the gut, increasing neurotransmitter and neuro-hormone concentrations in the body may benefit patients with injuries to the central nervous system. These studies must be performed methodically and optimally in human populations rather than rodent populations due to the apparent differences in diet and microbial composition, as duly noted in our study. The scientific community still needs to address the ever-important biologic significance to the changes in commensal bacteria populations.

DATA AVAILABILITY

Data will be be made available upon reasonable request.

GRANTS

B.E.G. has been supported by awards from the Office of the Assistant Secretary of Defense for Health Affairs supported by Award No. W81XWH-16-1-0349 and W81XWH-16-1-0387. The work performed through the UMMC Molecular and Genomics Facility is supported, in part, by funds from the National Institute of General Medical Sciences (NIGMS), including Mississippi INBRE (P20GM103476) and Obesity, Cardiorenal, and Metabolic Disease—COBRE (P20GM104357) and National Institute of General Medical Sciences of the National Institutes of Health under Award Number 5U54GM115428.

DISCLAIMERS

Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense or the National Institutes of Health.

DISCLOSURES

Michael Garrett is an editor of Physiological Genomics and was not involved and did not have access to information regarding the peer-review process or final disposition of this article. An alternate editor oversaw the peer-review and decision-making process for this article. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

B.E.G. conceived and designed research; A.M.S., B.A.W., K.K.H., M.R.G., and B.E.G. performed experiments; A.M.S., M.R.G., and B.E.G. analyzed data; A.M.S., M.R.G., and B.E.G. interpreted results of experiments; A.M.S. and B.E.G. prepared figures; A.M.S. and B.E.G. drafted manuscript; A.M.S., B.A.W., K.K.H., M.R.G., and B.E.G. edited and revised manuscript; A.M.S., B.A.W., K.K.H., M.R.G., and B.E.G. approved final version of manuscript.