Inulin-type fructan improves diabetic phenotype and gut microbiota profiles in rats

Study design and animal experiments

All procedures were undertaken with the approval of the Animal Care Committee of the Peking Union Medical Hospital Animal Ethics Committee (Project XHDW-2015-0051, 15 Feb 2015), and all efforts were made to minimize suffering. Five-week-old male Sprague-Dawley (SD) rats (158.3 ± 14.8 g) were obtained from the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences and Peking Union Medical College (Beijing, China, SCXK-2014-0013). All rats were maintained in cages at 24 ± 1 °C with lights on from 6:00 a.m. to 6:00 p.m. and were given free access to food and water.

Control rats were fed a standardized diet (kcal %: 10% fat, 20% protein, and 70% carbohydrate; 3.85 kcal/gm), and experimental rats were fed a high-fat diet (kcal %: 45% fat, 20% protein, and 35% carbohydrate; 4.73 kcal/gm; Research Diet, New Brunswick, NJ, USA) for four weeks and then intraperitoneally injected with streptozotocin (STZ, 30 mg/kg body weight) to induce diabetes (Chen et al., 2017). The diabetic animals were fed continuously on the high-fat diet throughout the remainder of the study. Fasting blood glucose (FBG) >11.1 mmol/L was determined to be the standard concentration for the type 2 diabetes model. Diabetic rats were randomly divided into two groups: an inulin-treated diabetic group (DM + inulin, Vilof™ Soluble Dietary Fiber; BAHEAL Medical Inc., Qingdao, China and Fengning Ping’an High-tech Industrial Co., Ltd., Hebei, China, orally administered 3 g Vilof™ Soluble Dietary Fiber powder/kg body weight/day, n = 6) and an untreated-diabetic group (DM, orally administered the same volume of normal saline, n = 6). Vilof™ Soluble Dietary Fiber powder contains 91% inulin-type fructan and 9% sucrose, fructose, and glucose. Control rats were randomly divided into an untreated control group (CON, n = 6, administered normal saline) and an inulin-treated control group (CON + inulin, n = 6, administered 3 g Vilof™ Soluble Dietary Fiber powder/kg body weight/day). After 12 weeks, fresh stool samples were obtained by stimulating the anus and were immediately frozen and stored at −80 °C for subsequent analysis. After 12 h of food deprivation, the rats were anesthetized (ketamine 100 mg/kg i.p.; Pharmacia and Upjohn Ltd., Crawley, UK), and blood samples were collected from the intraorbital retrobulbar plexus at 8 a.m. (Chen et al., 2017). The rats were thensacrificed. Epididymal fat and liver was quickly collected and kept at −80 °C.

Measurements of body weight and fasting blood glucose (FBG)

Body weight and FBG were monitored monthly with Bayer Contour TS glucometer (Bayer, Hamburg, Germany).

Oral glucose tolerance test (OGTT) and area under the curve

After fasting for 12 h, an OGTT was performed. Glucose (2 g/kg body weight) was orally administered to the rats. Glucose levels in the blood obtained from the tail were recorded before and 30, 60, and 120 min after the glucose load. The area under the curve (AUC) was calculated by the linear trapezoid method (Zhang et al., 2016).

Measurement of serum insulin, GLP-1, lipid profile, interleukin 6 (IL-6), and HOMA-IR

After 12 h of food deprivation, rat serum was obtained to analyze insulin (ELISA, Millipore, Billerica, MA, USA), GLP-1 (ELISA, Millipore, Billerica, MA, USA), total cholesterol (TC, enzyme end-point method; Roche Diagnostics, GmbH, Mannheim, Germany), triglyceride (TG, enzyme end-point method, Roche Diagnostics, GmbH, Mannheim, Germany), and IL-6 (ELISA, Millipore, Billerica, MA, USA) concentrations, according to the manufacturer’s instruction. The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated by the following formula: fasting blood glucose (FBG, mmol/L) × fasting serum insulin (µIU/mL)/ 22.5.

Measurement of liver TG content

Liver tissue (100 mg) was homogenized in 350 µL ethanolic KOH, then incubated overnight at 55 °C. The next day, 1,000 µL H2O: EtOH (1:1) was added to the tube. The samples were centrifuged for 5 min. The supernatant was used to measure the content of TG using a colorimetric kit (Biovision, Milpitas, CA, USA).

Quantitative PCR assay of Il6 mRNA expression in adipose tissue, G6pc and Pepck mRNA expression in liver

Total RNA from epididymal fat and liver was extracted by using RNA Isolation Kit (RNeasy kit, Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. RNA was reverse transcribed by Superscript II (Invitrogen, Carlsbad, CA, USA). Quantitative PCR was performed by using SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA). The primer was produced by Applied Biosystems (Foster City, CA, USA). Specific primers for rat Il6 mRNA were 5′-CTGGTCTTCTGGAGTTCCGT-3′ (forward) and 5′-TGGTCCTTAGCCACTCCTTCT-3′ (reverse), rat Pepck mRNA were 5′-CCAAGAGCAGAGAGACACCG-3′ (forward) and 5′-ATACATGGTGCGGCCTTTCA-3′ (reverse), rat G6pc mRNA were 5′-GCGTGCCATAGGACTCATCA-3′ (forward) and 5′-CACCAGCAAACAATTGCCCA-3′ (reverse). The products were run in the ABI Prism 7700 Sequence Detection system (Applied Biosystems, Foster City, CA, USA). The cycling conditions were set at 95 °C for 15 min, followed by 40 cycles of 95 °C for 15 s, 55 °C for 1 min and 72 °C for 1 min. Gadph was used for normalization. Relative quantification of mRNA was calculated using the comparative threshold cycle (Ct) method.

Fecal DNA extraction and amplification

Total DNA from fecal samples was isolated using a QIAmp DNA Stool Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. The 16S rRNA hypervariable regions V3–V4 were PCR amplified using barcoded, gene specific primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′) with Phusion High-Fidelity PCR Master Mix (New England Biolabs, Ipswich, MA, USA) followed by library preparation (TruSeq DNA PCR-Free Sample Preparation Kit; Illumina, San Diego, CA, USA).

Pyrophosphate sequencing and primary data analysis

Pyrosequencing was performed on an Illumina HiSeq 2500 platform (San Diego, CA, USA), and 250-bp paired-end reads were generated. Paired-end reads were connected using FLASH (John Hopkins University School of Medicine, Baltimore, MD, USA) (Magoc & Salzberg, 2011). Raw reads were filtered with a specific standard to gain high quality reads using QIIME version 1.7.0 (Caporaso et al., 2010). Sequences analysis was performed using UPARSE version 7.0.1001 (Edgar, 2013). The operational taxonomic units (OTUs) were generated from ≥97% similarity sequences. Representative sequences for each OTU were screened for further annotation.

Taxon composition and species diversity analysis

OTUs were annotated with taxonomic information based on the RDP classifier version 2.2 (Wang et al., 2007) algorithm using the Greengene database (DeSantis et al., 2006). OTU abundance data were normalized using a standard sequence number corresponding to the sample with least sequences. The relative proportion of each OTU was examined at the Phylum, Class, Order, Family, Genus and Species levels.

Alpha (within a community) and beta (between communities) diversity were analyzed with QIIME version 1.7.0. For alpha diversity, Chao1 and Shannon indices were calculated to identify community richness and diversity, respectively. For beta diversity, principal coordinates analysis (PCoA) plots were generated using both weighted (which considers the abundance of each species) and unweighted (which considers the presence or absence of each species) UniFrac (Lozupone & Knight, 2005). Unweighted pair-group method with arithmetic means (UPGMA) clustering was employed as the hierarchical clustering method to interpret the distance matrix using average linkage. In addition, linear discriminant analysis of the effect size (LEfSe) was performed to calculate OTU abundance and to determine the differences among groups (Segata et al., 2011). The threshold for the logarithmic linear discriminant analysis (LDA) score was >3.0.

Data analysis

The data are expressed as the mean ± SD. When the data were normal and variances were equal, differences among the groups were analyzed using one-way ANOVA followed by Tukey’s post hoc test. Otherwise, the Kruskal–Wallis test and the Mann–Whitney test were applied. Spearman’s correlation analysis was performed to identify the correlations. A P-value ≤0.05 was considered statistically significant (Prism version 5.0 for Windows; GraphPad Software, San Diego, CA, USA).

Body weight

Diabetic rats exhibited significantly lower body weight (P < 0.01, Fig. 1A), while inulin-treated diabetic rats had an increased body weight compared with the control diabetic rats (P < 0.01, Fig. 1A).

Fasting blood glucose level and glucose tolerance

Diabetic rats showed statistically elevated FBG levels compared to normal control rats (P < 0.01, Fig. 1B), and inulin treatment reduced FBG levels in diabetic rats (P < 0.01, Fig. 1B). According to the oral glucose tolerance test, blood glucose levels significantly increased before and after glucose load (P < 0.01, Fig. 1C) in diabetic rats. The elevated glucose area under the curve (AUC) on the OGTT (P < 0.01) in diabetic rats indicated severe glucose intolerance in diabetic rats (Fig. 1D). Inulin treatment significantly reduced blood glucose levels before and after glucose load (P < 0.05 or P < 0.01) in diabetic rats (Fig. 1C). Similarly, diabetic rats treated with inulin treatment exhibited a reduced glucose AUC by 25.8% (P < 0.01, Fig. 1D).

Fasting insulin and homeostasis model assessment of insulin resistance (HOMA-IR)

Dramatic increases in serum insulin levels and HOMA-IR in diabetic rats were noted (P < 0.01, Figs. 1E and 1F). Inulin treatment reduced serum insulin levels and HOMA-IR in diabetic rats (P < 0.01, Figs. 1E and 1F).

Serum biochemical parameters

Fasting serum GLP-1 levels decreased in the diabetic group (P < 0.01), while inulin treatment increased fasting serum GLP-1 levels (P < 0.01, Fig. 1G). Serum TC, TG, and IL-6 significantly increased in the DM group (P < 0.01, Figs. 1H–1J). Inulin treatment reduced serum TC, TG and IL-6 levels in diabetic rats (P < 0.05, Figs. 1H–1J).

Liver TG content

Liver TG content in diabetic rats significantly increased (P < 0.01, Fig. 1K). Inulin treatment reduced liver TG content slightly (P > 0.05, Fig. 1K).

Epididymal fat Il6 expression and hepatic Pepck and G6pc expression

The expression of Il6 in white adipose tissue and hepatic gluconeogenesis markers (Pepck and G6pc) expression increased significantly in diabetic rats (P < 0.01, Figs. 2A–2C). Inulin treatment reduced epididymal fat Il6 expression and hepatic Pepck and G6pc expression (P < 0.01, Figs. 2A–2C).

Characterization of gut microbiota

To describe the composition of the gut microbiota, a total of 1,402,994 pyrosequencing reads (69,281 unique sequences) were obtained from 24 stool samples. The 16s sequence data generated in this study were submitted to the NCBI Sequence Read Archive (SRA) database (accession number SRP095682). Among them, 1,108,097 reads (48,153 sequences) were generated after quality filtering, with an average of 2,386 ± 1,699 sequences being recovered per sample. The high-quality sequences were then delineated into 873 operational taxonomic units (OTUs) at the similarity cutoff of 97%.

In normal rats, inulin treatment did not affect OTU number, the Shannon diversity or the Chao1 index (Table 1). However, in diabetic rats, inulin treatment significantly reduced the Shannon diversity index of the gut microbiota (P < 0.05, Table 1). The richness of the gut microbiota was also significantly reduced by inulin treatment in diabetic rats, as shown by the Chao1 index (P < 0.01, Table 1). To compare the beta diversity of gut microbiota among groups, principal coordinate analysis (PCoA) was carried out. As shown in Fig. 3, the first two principal coordinates of PCoA (components 1 and 2) were separated into CON, DM, and DM + inulin groups. However, the CON and CON + inulin groups shared some overlapping regions.

Relative abundances of different bacteria in the gut microbiota after inulin treatment

As shown above, the CON and CON + inulin groups had similar alpha and beta diversities in gut microbiota. Next, we explored the similarities and differences of species distribution in the CON, DM, and DM + inulin groups. As shown in Fig. 4, we found that there were 602 species shared among the three groups, accounting for approximately three or four of the OTUs in each group. Of note, 88 species were found in the CON group, 31 OTUs in the DM group, and 24 OTUs in the DM + inulin group.

Consistent with other reports, the primary phyla were Bacteroidetes and Firmicutes in all samples. The DM group had an elevated relative abundance of Firmicutes (P < 0.01) and a reduced abundance of Bacteroidetes (P < 0.01). Inulin treatment reduced the relative abundance of Firmicutes (P < 0.01) and increased the abundance of Bacteroidetes (P < 0.01, Fig. 5A) in diabetic rats. The ratio of Firmicutes-Bacteroidetes has been suggested to be an indicator of gut microbial imbalance related to high-fat diet (Hildebrandt et al., 2009; Murphy et al., 2010; Turnbaugh et al., 2006). Our results showed that DM rats had a higher ratio of Firmicutes-Bacteroidetes (P < 0.01), whereas inulin treatment reuced this ratio (P < 0.01, Fig. 5B). Moreover, inulin treatment reduced the relative abundance of Tenericutes in diabetic rats (P < 0.05, Fig. 5A).

According to our LDA of the effective size, the abundances of many taxa were significantly different among the CON, DM, and DM + inulin groups (Figs. 6 and 7). At the family level, the Streptococcaceae and Acidaminococcaceae abundances were elevated in the DM + inulin group compared to the DM group, whereas the Christensenellaceae, Peptococcaceae, and Desulfovibrionaceae abundances were reduced in the DM + inulin group (P < 0.05, Table 2). At the genus level, the Phascolarctobacterium, Streptococcus, Lachnoclostridium, and Parasutterella abundances were all significantly elevated by inulin treatment, whereas the Ruminococcaceae_NK4A214_group, Alloprevotella, Ruminococcaceae_UCG_010, Ruminiclostridium_6, Christensenellaceae_R_7_group, Desulfovibrio, and Oscillibacter abundances were reduced by inulin treatment in diabetic rats (P < 0.05, Table 2). At the species level, Lactobacillus_animalis, Ruminococcus_gnavus, Phascolarctobacterium_faecium, Streptococcus_hyointestinalis, and Bacteroides_acidifaciens were more abundant in the DM + inulin group (P < 0.05, Table 2).

Correlations between metabolic biomarkers and bacterial abundance

At the genus level, Prevotellaceae_UCG-001 and Lachnospiraceae_UCG-006 abundances were negatively correlated with blood glucose AUC on the OGTT (P < 0.01, Figs. 8A and 8B). The Anaerovorax abundance showed a positive correlation with blood glucose AUC on the OGTT (P < 0.01, Fig. 8C).