Insulin resistant iNOS-/- mice display atypical gut microbiota with gram-positive bacteria dominance and altered serum metabolome
Chow fed iNOS-/- mice were glucose intolerant (Fig. S1A, B) and systemic insulin resistant (Fig. S1F, G), hyperglycemic and hyperinsulinemic as compared to WT (Fig. S1C-D), alongwith enhanced HOMA-IR (Fig. S1E) and decreased QUICKI (Fig. S1H). iNOS-/- mice also displayed enhanced gluconeogenesis as evident by PTT (Fig. S1K, L). Circulating total cholesterol, triglycerides (Fig. S1I, J), LDL and NEFA were significantly more in iNOS-/-mice while HDL levels were comparable to WT mice (Fig. S1M-O). The microbial alpha-diversity was reduced in insulin resistant iNOS-/- mice in the faecal samples as compared to WT (reduced observed, Shannon and Chao1 diversity indices). However, no significant difference was observed in the Simpson diversity index (Fig. 1A-D). Multidimensional scaling analysis through principal coordinate plot represents that the bacterial communities between the WT and iNOS-/- mice varied significantly (Fig. 1E). At the phylum level, relative abundance of Firmicutes was decreased with increased Verrucomicrobia. At the family level, Erysipelotrichaceae, Bifidobacteriaceae and Verrucomicrobiaceae were increased, while, Lactobacillaceae and Ruminococcaceae were decreased. Allobaculum, Bifidobacterium and Akkermansia were increased significantly, while, Lactobacillus genus was reduced in insulin resistant iNOS-/- mice as compared to WT (Fig. 1F) indicating differential gut microbiome with major changes in gram-positive bacteria. In PCA score scatter plot from serum metabolomics, distinctly separated clusters between the WT and iNOS-/- mice in ESI positive mode were seen (Fig. 1G) with significantly differential metabolites (p<0.05) being visualized through volcano plot (Fig. 1H). iNOS-/- mice displayed enhanced purine and pyridimidine metabolites, PE lipids, PE to PC ratio, 10-hydroxydecanoate, 3-nitrotyrosine, cysteamine, cysteate, carbohydrate metabolites, indole-3-ethanol, diosmetin and phosphonoacetate. PC, PA and PS lipids, laurate, lauroyl carnitine, anthranilate and cystathionine were decreased in iNOS-/- mice suggesting altered metabolic profile as compare to WT (Fig. 1I).
Vancomycin-induced modulation of gut microbiota rescue iNOS-/- mice from systemic IR and dyslipidemia
Subsequently, we used vancomycin to deplete the gram-positive and antibiotic cocktail to deplete both gram-positive and negative gut microbiota in iNOS-/- mice to assess the correlation, if any, between gut microbiome and altered metabolic homeostasis in iNOS-/- mice. Vancomycin (Fig. 2A-D) and Abx (Fig. S3A-D), as expected, further decreased the microbial diversity in insulin resistant iNOS-/- mice (decreased observed, Shannon, Simpson and Chao1 α-diversity indices) confirming the depletion of gut microbiota. Abx treated iNOS-/- mice showed the lowest alpha-diversity among all the groups. Bacterial communities between vancomycin and Abx treated WT (Fig. S2E, S5F) and iNOS-/- mice (Fig. 2E, S3E) vary distinctly and significantly from the untreated controls with markedly different clusters in PCA analysis. At the phylum level, Bacteriodetes and Actinobacteria were decreased significantly in vancomycin treated WT and iNOS-/- mice with increased Proteobacteria. Verrucomicrobia was increased significantly by vancomycin in iNOS-/- mice. Bacterial families- Bifidobacteriaceae, Ruminococcaceae, S24-7 were decreased significantly in vancomycin treated WT and iNOS-/- mice with increased Veillonellaceae, Verrucomicrobiaceae and Enterobacteriaceae. Erysipelotrichaceae was decreased with increased Lactobacillaceae, Staphylococcaceae and Porphyromonadaceae in vancomycin treated iNOS-/- mice. Lactobacillaceae and Lachnospiraceae were decreased in vancomycin treated WT mice. At the genus level, Bifidobacterium was decreased with increased Akkermansia and Veillonella by vancomycin treatment in WT and iNOS-/- mice. Vancomycin decreased Allobaculum and increased Lactobacillus, Staphylococcus and Parabacteroides in iNOS-/- mice; while decreased Lactobacillus and Oscillospira in WT (Fig. S2H, 2F). Abx decreased Firmicutes and Bacteroidetes with increased Proteobacteria in WT and iNOS-/- mice depleting majority of gut microbiota. Actinobacteria and Verrucomicrobia were decreased by Abx in iNOS-/- mice. Bacterial families- Ruminococcaceae, Lactobacillaceae, S24-7 were decreased with increased Enterobacteriaceae in Abx treated WT and iNOS-/- mice. Bifidobacteriaceae, Erysipelotrichaceae, Verrucomicrobiaceae and Bacteroidaceae were decreased in Abx treated iNOS-/- mice with increased Brucellaceae in WT. At the genus level, Lactobacillus, Bifidobacterium, Akkermansia were decreased by Abx treatment in WT and iNOS-/- mice with increased Klebsiella. Abx decreased Allobaculum and increased Serratia in iNOS-/- mice; while increased Pseudomonas, Elizabethkingia and Ochrobactrum in WT (Fig. S5H, S3F). Both, vancomycin (Fig. 2G-L) and Abx (Fig. S3G-L) reversed the systemic glucose intolerance, IR and enhanced gluconeogenesis in iNOS-/- mice as seen by decrease in AUC during GTT, ITT and PTT respectively. The increase in the circulating level of glucose and insulin in iNOS-/- mice was decreased by both the treatments (Fig. 2M, N; S3M, N). HOMA-IR was decreased by antibiotics treatment with increased QUICKI reversing the IR phenotype observed in iNOS-/- mice (Fig. 2O, P; S3O, P). The enhanced circulating TC, TG, LDL and NEFA in iNOS-/- mice were reversed by the antibiotics treatment, thus rescuing the dyslipidemia in iNOS-/- mice (Fig. 2Q-T; S3Q-T). The body weight (Fig. S6A, B) and food intake (Fig. S6D, E) however remained unaltered in obese iNOS-/- mice upon vancomycin and Abx treatment. Body composition and BMI also remained unaltered in vancomycin treated iNOS-/- mice with increased fat mass in Abx treated iNOS-/- mice (Fig. S6C, F). Liver, adipose tissue, and heart weight ratio was decreased in vancomycin and Abx treated iNOS-/- mice with increased weight and length ratio of small intestine and caecum (Fig. S6G, H).
Vancomycin-induced alterations in serum metabolites in iNOS-/- mice
The PCA score plots of vancomycin (Fig. 3A) and Abx (Fig. S4A) treated and untreated iNOS-/- mice and vancomycin (Fig. S2A) and Abx (Fig. S5A) treated and untreated WT mice in ESI (+) mode showed very distinct separations. Significantly differential metabolites (p<0.05) between untreated iNOS-/- and iNOS-/- treated with vancomycin (Fig. 3B) or Abx (Fig. S4B) were visualized through volcano plots. Nucleic acid metabolites- allantoin and pyrimidines were decreased by vancomycin (Fig. 3C) and Abx (Fig. S4H) in iNOS-/- mice with decreased methylthioadenosine by Abx. Vancomycin and Abx treatment did not decreased other purine metabolites in WT and iNOS-/- mice. Methylthioadenosine and allantoin were also decreased in WT by vancomycin (Fig. S2C) and Abx (Fig. S5D) along with decreased pyrimidines by Abx. 15-Methyl PGA2 and corticosterone was increased by both vancomycin (Fig. 3D) and Abx (Fig. S4F) in iNOS-/- mice. Serotonin was increased by vancomycin treatment in iNOS-/- mice with decreased thyrotropin-releasing hormone. Bile acid-cholate and cofactor-Pyridoxate was decreased by Abx (Fig. S5B, S4F) in WT and iNOS-/- mice but not by vancomycin (Fig. S2B, 3D). Glycolysis and Krebs cycle intermediates were decreased in vancomycin (Fig. 3E) and Abx (Fig. S4D) treated iNOS-/- mice while glyceraldehyde was increased. Vancomycin in WT (Fig. S2F) did not decreased the carbohydrate metabolites to a significant extent as compared to Abx treatment (Fig. S5C). Oxoproline, 2-methylhippuric acid, hippurate, indole-3-ethanol, ophthalmate, mevalonate and aspartame were decreased by vancomycin and Abx in WT (Fig. S2D, S5E) and iNOS-/- mice (Fig. 3F, S4C). 4-Acetamidobutanoic acid, indole-3-methyl acetate, sebacic acid, 4-chlorophenol and isopentenyl-adenine-7-glucoside were decreased by vancomycin and Abx in iNOS-/- mice while formate was increased by vancomycin suggesting the modulation of bacterial derived/dependent metabolites by antibiotics. Majority of amino acids were significantly decreased in vancomycin and Abx treated WT (Fig. S2G, S5G) and iNOS-/- mice (Fig. 3G, S4E) including aminoadipate except, 3-nitrotyrosine, cysteamine and cysteate. Aromatic amino acid metabolites- tryptophan was increased in iNOS-/- mice while 3-methoxytyrosine and N-acetyl phenylalanine was increased in WT by Abx (Fig. S5G, S4E). Spermidine was increased by vancomycin and Abx in WT mice (Fig. S2G, S5G). Many lipid species- PE, PC, PA, PS, PG, ceramides and 10-Hydroxydecanoate were decreased by vancomycin and Abx treatment in iNOS-/- (Fig. 3H, S4G) and WT mice (Fig. S2I, S5I); while palmitate was increased suggesting improved lipid metabolism along with improvement in PE to PC ratio. Laurate and lauroyl carnitine were decreased in iNOS-/- mice and were increased by vancomycin but not by Abx. These results suggest that vancomycin-induced gram positive bacteria depletion improved the metabolic perturbations in iNOS-/- mice.
Improvement in the disrupted lipid and glucose homeostasis in liver in iNOS-/- mice following treatment with vancomycin
Hepatic TC, TG and FFA levels were reduced in the vancomycin (Fig. 4A-C) and Abx (Fig. S7A-C) treated iNOS-/- mice. FFA levels were also decreased in vancomycin and Abx treated WT (Fig. 4C, S7C). The enhanced expression of lipid synthesis genes (PPARγ, SREBP-1c and ACC1) in the liver of iNOS-/- mice were decreased by vancomycin (Fig. 4F) and Abx (Fig. S7F) with unchanged LXRα, LXRβ, and HMGCR in liver. FAS and SREBP-2 expression was decreased significantly by vancomycin in iNOS-/- mice, but not by Abx. Expression of LXRα, LXRβ and HMGCR were increased in liver of vancomycin treated WT (Fig. 4F). PGC-1β expression was decreased in liver of vancomycin (Fig. 4G) and Abx (Fig. S7G) treated WT and iNOS-/- mice. PPARα and UCP2 expressions were decreased in liver of Abx treated WT and vancomycin treated iNOS-/- mice. Liver PGC-1α expression was increased in vancmycin treated WT. ACC2 expression remained unchanged in liver of vancomycin and Abx treated iNOS-/- mice. Expression of genes involved in the hepatic lipid uptake (CD36, LPL) were unaltered in iNOS-/- mice following vancomycin (Fig. 4H) and Abx (Fig. S7H) treatment with decreased SR-1B. CD36 and LPL expression was increased in vancomycin treated WT. LDLR expression was reduced in iNOS-/- mice and was unaltered by vancomycin and Abx. The expression of genes involved in hepatic lipid efflux- ABCG5 and ABCG8 was decreased in iNOS-/- mice and was increased significantly in Abx treated iNOS-/- mice (Fig. S7I) and were comparable to WT in vancomycin treated iNOS-/- mice (Fig. 4I). Hepatic PC expression was decreased by vancomycin with decreased G6PC expression by vancomycin (Fig. 4E) and Abx (Fig. S7E) in iNOS-/- mice along with unchanged FOXO1 and PEPCK. Hepatic glycogen levels also remained unaltered upon vancomycin treatment (Fig. 4D), while insulin stimulated glycogen levels were increased significantly by Abx (Fig. S7D).
Improvement in the disrupted lipid and glucose homeostasis in adipose tissue and intestine in iNOS-/- mice following treatment with vancomycin
The enhanced expression of lipid synthesis genes (PPARγ, SREBP-1c, FAS and ACC1) in adipose tissue of iNOS-/- mice were decreased by vancomycin, while, Abx treatment decreased SREBP-1c and FAS. LXRα expression was increased in eWAT of vancomycin (Fig. 5C) and Abx treated iNOS-/- mice (Fig. S8C). PGC-1α and UCP2 expressions in adipose tissue were increased in Abx treated iNOS-/- mice, while, PGC-1β, ACC2 and PPARα remains unchanged by vancomycin (Fig. 5D) or Abx (Fig. S8D). The enhanced expression of CD36 and LPL in adipose tissue of iNOS-/- mice was decreased by vancomycin (Fig. 5E) and Abx treatment (Fig. S8E). PEPCK and G6PC was decreased in adipose tissue by vancomycin (Fig. 5A) and Abx (Fig. S8A) in iNOS-/- mice with unchanged FOXO1 and PC. Decreased Akt2 gene expression was increased in white adipose tissue of vancomycin treated WT and iNOS-/- mice suggesting improved insulin signaling upon gram-positive bacteria depletion. The glucose transporters in adipose tissue remained unaltered upon vancomycin (Fig. 5B) or Abx treatment (Fig. S8B).
The LXRα, HMGCR and CYP7A1 expressions were decreased in small intestine of vancomycin treated iNOS-/- and WT mice (Fig. 5F). Abx decreased LXRα expression in iNOS-/- and WT mice (Fig. S8F). Expression of most of the lipid uptake genes in the small intestine (CD36, NPC1L1, FABP1, LDLR, ApoE, FFAR1, FFAR2) was not altered in the vancomycin (Fig. 5G) and Abx treated iNOS-/- mice (Fig. S8G), except SR-1B, which was reduced. Further, expression of CD36, NPC1L1, LDLR, FFAR1 and FFAR2 was reduced in the vancomycin treated WT mice (Fig. 5G). ABCG5 expression was decreased in small intestine in iNOS-/- mice and was increased by vancomycin (Fig. 5H) and Abx treatment (Fig. S8H) with unchanged ABCG8 and ABCA1.
The morphometric calculations (Fig. S9A, D) displayed increased intestinal villi and colonic crypt length in iNOS-/- mice (Fig. S9C, F). The barrier functionality of the gut remained unchanged in iNOS-/- mice as assesed by in vivo FITC-dextran gut permeability assay (Fig. S9G) and Alcian blue staining in intestinal (Fig. S9A, B) and colonic tissues (Fig. S9D, E). The gene expression of tight junction protein- Claudin-2 was increased with decreased ZO-1 and unaltered occludin in iNOS-/- mice. Mucins (Muc-2 and Muc-5AC) remained unaltered in iNOS-/- mice with reduced expression of antimicrobial peptide Reg3γ (Fig. S9H). The FITC-dextran gut permeability assay showed significant reduction in the intestinal permeability following vancomycin and Abx treatment in WT and iNOS-/- mice as compared to untreated controls (Fig. S9G). The mRNA expression for tight junction proteins, mucins and antimicrobial peptide in small intestine largely remained unchanged in antibiotics treated iNOS-/- mice. Claudin-2 mRNA expression was increased in Abx treated WT and iNOS-/-mice (Fig. S9H). Further, status of NO was checked and the decreased total nitrite levels in iNOS-/- mice were not altered in serum, adipose tissue and intestine after antibiotics treatment. In liver, the total nitrite levels were further decreased in iNOS-/- mice by antibiotics (Fig. S9I-L). nNOS expression remained unchanged in the liver and intestine, and was decreased in adipose tissue upon antibiotics treatment in iNOS-/- mice. Decreased eNOS expression was rescued in the adipose tissue and intestine by antibiotics and remained unaltered in liver (Fig. S9M-O).
Association of serum metabolites with metabolic profile of iNOS-/- mice upon treatment with antibiotics
The phenotypic, biochemical, functional, metabolic and molecular analysis suggests that iNOS-/- mice displayed IR and dyslipidemia along with altered gut microbiome as compared to WT mice. Gut microbiota depletion/ modulation by antibiotics showed marked improvement in metabolic parameters with precise alterations in the serum metabolites. Pearson’s correlation was used to identify the serum metabolites which strongly correlated with the metabolic biomarkers quantified in control (WT), insulin resistant (iNOS-/-) and antibiotics (vancomycin or Abx) treated WT and iNOS-/- mice. Lipid species- MGDGs, PEs, PAs, PSs, PGs, ceramides and 10-hydroxydecanoate positively correlated with bacterial diversity, dyslipidemia and IR; and negatively with caecum weight. Laurate and palmitate were negatively correlated liver lipids and bacterial diversity (Fig. 6A). Majority of amino acids positively correlated with biomarkers of IR, dyslipidemia and α-diversity and negatively correlated with HDL levels and caecum weight (Fig. 6B).
Indole-3-methyl acetate, 5-oxoproline, 2-methylhippuric acid, hippurate, indole-3-ethanol, ophthalmate, mevalonate, 6-carboxyhexanoate, sebacic acid, 4-chlorophenol, δ-undecalactone and aspartame positively correlated with IR, dyslipidemia and α-diversity and negatively correlated with caecum weight. S-carboxymethylcysteine and formate were negatively correlated with dyslipidemia and α-diversity (Fig. S10A). Nucleic acid metabolites- methylthioadenosine, xanthine, allantoin, thymine, cytosine and uracil positively correlated with biomarkers of IR and dyslipidemia and negatively correlated with caecum weight. Glycolysis and Krebs cycle intermediates were correlated positively and glyceraldehyde negatively with liver lipids and α-diversity. 4-Pyridoxate and melatonin were positively correlated with glucose intolerance and dyslipidemia. Bile acids and bile pigments were positively correlated with IR, dyslipidemia and negatively correlated with caecum weight (Fig. S10B). These results suggest that serum metabolites exhibited strong association with specific metabolic biomarkers and were modulated by microbiota depletion by antibiotics.
Association of gut microbiota with metabolic parameters in iNOS-/- mice following treatment with antibiotics
Further, association between antibiotics induced compositional changes in microbiota with the alterations in metabolic parameters, was investigated. Phylum- Bacteroidetes and Actinobacteria were positively associated with IR, dyslipidemia and α-diversity, while Proteobacteria correlated negatively. Firmicutes and Verrucomicrobia were negatively correlated with caecum weight and positively correlated with dyslipidemia (Fig. 7A). Bifidobacteriaceae (Actinobacteria), Erysipelotrichaceae (Firmicutes) and S24-7 (Bacteroidetes) were positively correlated with IR, dyslipidemia and α-diversity and negatively with caecum weight. Microbial families- Ruminococcaceae, Eubacteriaceae, Lachnospiraceae, Dehalobacteriaceae, Mogibacteriaceae, Lactobacillaceae (Firmicutes); Coriobacteriaceae (Actinobacteria); Verrucomicrobiaceae (Verrucomicrobia) were positively correlated with dyslipidemia and α-diversity and negatively correlated with caecum weight. Weeksellaceae, Sphingobacteriaceae (Bacteroidetes); Enterococcaceae, Bacillaceae, Tissierellaceae, Leuconostocaceae (Firmicutes); Microbacteriaceae, Micrococcaceae, Gordoniaceae, Nocardiaceae, Nocardiodaceae, Promicromonosporaceae, Beutenbergiaceae, Streptomycetaceae, Pseudonocardiaceae, Brevibacteriaceae, Geodermatophilaceae (Actinobacteria) and many families of Proteobacteria including Enterobacteriaceae were negatively correlated with serum NEFA and positively correlated caecum weight (Fig. S11).
Bifidobacterium (Actinobacteria); Allobaculum and Ruminocoocus (Firmicutes) were positively correlated with IR, dyslipidemia and α-diversity and negatively with caecum weight. Anaerofustis (Firmicutes) and Adlercreutzia (Actinobacteria) were positively correlated with serum NEFA and α-diversity and negatively correlated with caecum weight. Pediococcus, Bacillus, Enterococcus, Dialister (Firmicutes); Sphingobacterium, Elizabethkingia (Bacteroidetes); Gordonia, Aeromicrobium, Streptomyces, Rhodococcus, Mycetocola, Salana, Microbacterium, Luteimicrobium, Nesterkonia, Saccharopolyspora, Brevibacterium (Actinobacteria) and many genus of Proteobacteria phylum including Enterobacter were negatively correlated with NEFA and positively correlated with caecum weight. (Fig. 7B). These results suggest that Gram-positive bacteria depletion (Allobaculum, Bifidobacterium and Ruminococcus) following treatment with vancomycin is directly correlated with the improvement in metabolic biomarkers.
Gut microbiota association with serum metabolites in iNOS-/- mice upon antibiotics treatment
Pearson’s correlation coefficients were calculated between serum metabolites and microbial taxa, which positively correlated with blood glucose and glucose intolerance in WT and iNOS-/- mice without or with antibiotics (vancomycin or Abx) treatment to gain better insights into the host metabolism. Interestingly, most of these microbial taxa linked to impaired glucose metabolism are gram-positive bacteria. Lipid species- MGDG, PE, PC, PA, PS, PG, ceramide, 10-hydroxydecanoate, dodecanedioic acid and myristate were positively correlated and palmitate negatively correlated with taxa directly associated with glucose intolerance (Actinobacteria, Bacteroidetes, Erysipelotrichaceae, Bifidobacteriaceae, S24-7, Ruminococcaceae, Allobaculum, Bifidobacterium, Ruminococcus) (Fig. 8A). Most amino acids metabolites including anthranilate, histidine, carnosine, serine, N-formylmethionine, N-acetyl cysteine, N-acetyl norvaline, carnitine, aminoadipate, pipecolinic acid, methyl glutarate, glutamate, N-acetyl ornithine, betaine, creatine, creatinine, proline, leucyl proline and 3-amino-4-hydroxybutyric acid were positively correlated wih taxa associated with glucose intolerance (Fig. 8B).
Metabolites- Indole-3-methyl acetate, 5-Oxoproline, 2-methylhippuric acid, hippurate, indole-3-ethanol, ophthalmate, mevalonate, 6-carboxyhexanoate, sebacic acid, 4-chlorophenol, δ-undecalactone, aspartame and bis(2-ethylhexyl)phthalate were positively correlated and S-carboxymethylcysteine and formate were negatively correlated with taxa associated with glucose intolerance (Fig. 8C). Purines (allantoin, methylthioadenosine) and pyrimidines (uracil) correlated positively with taxa associated with glucose intolerance. Lactic acid, N-acetyl neuraminate, isocitrate and citramalate were positively correlated, while, glyceraldehyde was negatively correlated with taxa associated with glucose intolerance. 4-Pyridoxate, melatonin, bile acids and pigments- cholate, deoxycholate and bilirubin were positively correlated with taxa associated with glucose intolerance (Fig. 8D). These results suggest that metabolites involved in lipids, amino acids, bile acids, nucleic acids, carbohydrate, bacterial derived/dependent metabolism exhibited significant correlation with gram-positive bacterial taxa linked to IR and dyslipidemia in iNOS-/- mice.