Dominant gut Prevotella copri in gastrectomised non-obese diabetic Goto-Kakizaki rats improves glucose homeostasis through enhanced FXR signalling.

Aims/hypothesis Drug and surgical-based therapies in type 2 diabetes are associated with altered gut microbiota architecture. Here we investigated the role of the gut microbiome in improved glucose homeostasis following bariatric surgery. Methods We carried out gut microbiome analyses in gastrectomised (by vertical sleeve gastrectomy [VSG]) rats of the Goto–Kakizaki (GK) non-obese model of spontaneously occurring type 2 diabetes, followed by physiological studies in the GK rat. Results VSG in the GK rat led to permanent improvement of glucose tolerance associated with minor changes in the gut microbiome, mostly characterised by significant enrichment of caecal Prevotella copri. Gut microbiota enrichment with P. copri in GK rats through permissive antibiotic treatment, inoculation of gut microbiota isolated from gastrectomised GK rats, and direct inoculation of P. copri, resulted in significant improvement of glucose tolerance, independent of changes in body weight. Plasma bile acids were increased in GK rats following inoculation with P. copri and P. copri-enriched microbiota from VSG-treated rats; the inoculated GK rats then showed increased liver glycogen and upregulated expression of Fxr (also known as Nr1h4), Srebf1c, Chrebp (also known as Mlxipl) and Il10 and downregulated expression of Cyp7a1. Conclusions Our data underline the impact of intestinal P. copri on improved glucose homeostasis through enhanced bile acid metabolism and farnesoid X receptor (FXR) signalling, which may represent a promising opportunity for novel type 2 diabetes therapeutics. Electronic supplementary material The online version of this article (10.1007/s00125-020-05122-7) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

randomly selected rats were inoculated with a single gavage of 5.10 8 CFU P. copri. The control group received an equivalent dosage of heat-killed P. copri.

Glucose tolerance tests and sample collection and analysis
Oral and intraperitoneal glucose tolerance tests (OGTT, IPGTT) were performed in conscious rats after an overnight fast. Due to changes in intra-abdominal organs secondary to gastrectomy, including extensive connective tissue, OGTT were preferred to IPGTT to assess glucose tolerance and glucose-induced insulin secretion in VSG treated rats (Experiment 1, Fig. 1). OGTT was also used to assess glucose tolerance in sham operated controls. For OGTT, a bolus of glucose (1g/kg body weight) was given by gavage in conscious rats. Blood samples were collected from the tail vein before glucose administration and 15, 30, 60, 90 and 120 min afterwards. For IPGTT carried out in GK rats inoculated with the caecal content of VSG or sham operated GK (Experiment 2, Fig. 1), treated with P. copri permissive antibiotics or no antibiotics (Experiment 3, Fig. 1) and inoculated with P. copri or heat inactivated P. copri (Experiment 4, Fig. 1), conscious rats were injected intraperitoneally with a solution of glucose (1g/kg body weight) after an overnight fast. Blood samples were collected from the tail vein before glucose injection and 5, 15, 30, 60, 90, 120, 180 and 240 minutes afterwards. Glycemia was determined using an Accu-Check® Performa (Roche Diagnostics, Meylan, France) and insulinemia was assayed by ELISA (Mercodia, Uppsala, Sweden). Glucose tolerance was assessed with the cumulative glycemia (increment of glucose values during the test) and the ΔG (cumulative glycemia above baseline).
Three days after the glucose tolerance tests, rats were fasted overnight and killed by lethal injection of sodium pentobarbital. Plasma samples, adipose tissue, liver and caecal content were quick-frozen in liquid nitrogen and stored at -80°C. Colorimetric assays were used to determine plasma and liver triglycerides (ab65336; Abcam, Paris, France) and liver glycogen (Sigma-Aldrich).

Metagenome sequencing
Bacterial DNA was independently prepared from caecal (10 from VSG-treated GK rats and 8 from sham operated GK controls) and colon (6 from VSG-treated GK rats and 3 from sham operated GK controls) samples (DNAStool mini kit, QIAGEN, Courtaboeuf, France). Wholegenome shotgun sequencing was performed on Illumina HiSeq 2000 instruments (Illumina, San Diego, CA) to generate 150nt sequence reads. For each sample, a single sequencing library was produced and bar-coded. Pools of four samples were run per lane. Every sample was sequenced on multiple instruments to minimize potential technical flow cell or instrument biases. Data was de-duplicated using standard in-house protocols. There were four separate sets of cluster building per sample in order to reduce errors of duplication and deduplication. To estimate the biodiversity of the samples, the WGS data was mined for sequence motifs corresponding to a 55nt segment of the bacterial 16S rRNA variable 1-3 region. A total of 8,767 unique rDNA motifs were found (517,697 occurrences), the vast majority of which were rare, both in terms of number of samples detected and prevalence in individual samples. To reduce data sparsity, we analyzed motifs that accounted for a minimum prevalence of 0.5% or better in any individual sample. For preliminary indications of species identifications, motifs were aligned to publicly available microbial databases to retrieve perfect and highly similar matches.

Quantitative PCR
Bacterial DNA from fresh caecum samples were extracted using DNA Stool mini kit (QIAGEN, Courtaboeuf, France). DNA yield and integrity were determined using a Nanodrop 2000c (Thermo Fisher Scientific, Illkirch, France) and agarose gel electrophoresis. P. copri enrichment was assessed by quantitative PCR using SYBR green assays (Life technologies, Saint Aubin, France) and comparison to universal 16S rRNA. Preparation of liver and adipose tissue RNA and quantitative RT-PCR analyses were performed as described [7] using the housekeeping gene HPRT. Relative quantification of mRNA levels of candidate genes were carried out using the Livak and Schmittgen method [8]. Reactions were run on taqman 7900 HT system for P. copri analysis and on a CFX96 Real-Time system (Bio-Rad) for gene expression. Oligonucleotide sequences are given in Supplementary Table 1.

Quantitative analysis of plasma bile acids
Plasma bile acids were quantified on 100μL plasma aliquots as described [9]. Plasma were spiked with isotopically labelled bile acid standards and treated with ice cold methanol in order to remove proteins. Bile acids were separated in an ACQUITY BEH C8 column and detected by a Xevo TQ-S mass spectrometer (Waters, Manchester, UK) operating in negative ionization mode and multiple reaction monitoring. The method allowed us to quantify up to 35 distinct bile acids.

Statistical analyses
R packages were used to assess differences in the frequency of 16SrDNA motifs between gastrectomised and sham operated control rats and to compute a p-value for each motif with a threshold of significance set to 0.05. P-values were corrected for multiple testing using the Benjamini-Hochberg method [10]. Blood glucose and insulin secretion data during the glucose tolerance tests were analysed with the Kruskal-Wallis test in order to account for conserved intra-individual distribution of blood glucose and plasma insulin. Non-parametric Mann-Whitney U tests were used to assess differences in physiological phenotypes, bile acid data and gene expression between experimental and control groups.
ESM Table 1. Oligonucleotides used for quantitative RT PCR.