Discussion
We have demonstrated that metformin, acarbose, and sitagliptin monotherapy similarly lowered blood glucose levels, but differently affected the diversities, composition and functions of gut microbiota in ZDF rats.
In our study, alpha diversity analysis was used to detect the diversity of microbiota after the treatments. We found that acarbose reduced the microbial richness and diversity, but there were no changes after metformin and sitagliptin treatments. It has been previously shown acarbose strikingly increased the relative abundance of beneficial bacteria in patients with type 2 diabetes, along with the decreased diversity.18 19 Other antidiabetic medications, such as glucagon-like peptide-1 agonists, liraglutide decreased the alpha diversity in obese and diabetic rats.20 Interestingly, it has been reported recently that alpha diversity was reduced after 8 weeks of treatment with dapagliflozin, a sodium-glucose co-transporter-2 inhibitor, which mainly improved blood glucose levels by inhibiting reabsorption of glucose filtered from the renal glomerulus.21 22 The indices of alpha diversity normally serve as a summary measure tool and might not be accurate to detect the components of gut microbiota. Therefore, we did further analysis on core microbial composition to clarify the effects of interventions.
We observed that the relative abundance of phylum Firmicutes and the ratio of Firmicutes to Bacteroidetes were increased after using three hypoglycemic agents. The ratio of Firmicutes to Bacteroidetes was positively correlated with fecal SCFAs.23 Although Larsen et al24 found the opposite results, it has been shown that the ratio of Firmicutes to Bacteroidetes was decreased in diabetic rats.25 After 4-week treatments of metformin, acarbose and sitagliptin, the ratio was reversed in our study. The effect of sitagliptin was the most notable among the three medications. Another, an abnormal expansion of phylum Proteobacteria plays an important role in the inflammation. In our current study, metformin, acarbose and sitagliptin dramatically decreased the abundance of Proteobacteria. It was striking to find that Proteobacteria tended to decline after thiazolidinediones (TZD), pioglitazone treatment in the high-fat diet-fed rats.5 26 The interaction between TZD and gut microbiota needs to be clarified in the future study.
As a predominant genus in phylum Firmicutes, Lactobacillus exhibits prominent antidiabetic effects via stimulating incretin hormones secretion and reducing endotoxemia.27 28 This study first proposed that different hypoglycemic agents could selectively regulate the abundance of Lactobacillus sp. Metformin and sitagliptin rose the overall abundance of Lactobacillus genus including L. johnsonii and L. intestinalis spp. Acarbose dramatically increased L. intestinalis, but reduced the species of unclassified Lactobacillus and L. johnsonii. It was validated by previous studies that acarbose18 and metformin7 promoted the level of Lactobacillus genus. Wang et al showed that the genera Lactobacillus, Allobaculum and Turicibacter were also enriched following liraglutide and saxagliptin dosage.29 30 There was little research on the taxonomic composition of sitagliptin. Only one study in a diabetic rat model showed that sitagliptin did not change Lactobacillus genus.14 In fact, the genus level of Lactobacillus is unable to encompass the entire species due to the various fermentations of Lactobacillus, which include homofermentation or heterofermentation.31
In addition to Lactobacillus, Bifidobacterium has anti-inflammatory effects and the relative abundance is reduced in type 2 diabetes.32 We demonstrated that acarbose increased the relative abundance of Bifidobacterium which was not changed in the other three groups. High carbohydrates intake could result in higher abundance of Bifidobacterium.33 Thus, the elevated abundance of Bifidobacterium might be due to the more exposure of carbohydrates to the distal gut by acarbose.
In 2011, the MetaHIT team proposed the concept of ‘Enterotypes’, which divided the intestinal microbiota into three types: Bacteroides, Prevotella and Ruminococcus.34 Enterotypes are mainly determined by the dominant bacteria and relatively stable, independent of age, gender, race, body mass index, and nutritional status.35 Although three medications similarly lowered blood glucose levels, the bacterial cluster of acarbose group was notably different from the other three groups in terms of enterotypes and beta diversity. At genus level, Ruminococcus 2 was the main taxonomic driver after acarbose treatment whereas Lactobacillus was the dominant genus in the other three groups. Further analysis of enterotypes showed that there were differences between antidiabetic treatment and control at OTU level. L. johnsonii (OTU588) was the dominant taxon followed by metformin and sitagliptin treatment while the control group was dominated by unclassified Lactobacillus sp (OTU163).
Ruminococcus 2 is a genus of bacteria in the family Ruminococcaceae and class Clostridia. Ruminococcus was originally isolated from bovine rumen and then found in various mammalian hosts, including humans, rodents, and so on. Fermentable carbohydrates are required in growth of all Ruminococcus.36 Therefore, it is not surprising that acarbose increases the carbohydrates in the distal gut and may specifically promote the growth of Ruminococcus 2. Blaubia (Ruminococcus) obeum shares the same class Clostridia with Ruminococcus 2 and expresses α-glucosidases (Ro-αG1). Recently, a theory proposed that α-glycosidase inhibitors (acarbose, voglibose, miglitol) could affect the bacterial Ro-αG1 in human gut and exerted antidiabetic effects or created adverse gastrointestinal symptoms.37 Furthermore, Rumenococcus, SCFA-producing bacteria, mainly produced acetate and propionate38 to improve metabolic abnormalities and intestinal inflammation.39 Further functional predictions indicated that acarbose increased abundance of profiles in carbohydrate transport and metabolism [G], which mainly included transporter activity (COG1653) and binding-protein-dependent transport systems inner membrane component (COG1175 and COG0395). KEGG analysis confirmed that the enzyme function associated with carbohydrate metabolism was active in the intestinal flora after acarbose therapy, compared with metformin and sitagliptin.
There are some limitations in this study. First, the distribution of the flora significantly varies with the changes of pH and intestinal metabolites in different intestinal segments. There is comparatively less abundance of microbiota in the small intestine, and the dominant bacteria such as Ruminococcus and Clostridium mainly localize in the large intestine.40 We collected fecal samples to assess the composition of gut microbiota. Hence, it is hard to clarify the distribution characteristics of microbiota in the small intestine and proximal colon. Second, the animal model used in this study was a diabetic model, ZDF rat, which was mainly caused by genetic defects. It might be different in the distribution of the gut microbiota from that of the human body. Lastly, we used soluble starch which was extracted from corn during the IGSTT test. The type of starch, in particular, resistant starch, interacts differently with the microbiota, and the interaction between starch and drugs may also have a significant impact on it.41 In fact, we did collect the fresh fecal samples 1 week after the IGSTT. We believe that interaction between soluble starch and the different drugs would have little influence on the results.
In conclusion, our study found that metformin, acarbose and sitagliptin differently affected intestinal bacteria. Acarbose selectively increased the bacteria including genera Ruminococcus 2 and Bifidobacterium. Metformin and sitagliptin increased the relative abundance of Lactobacillus. Supplementation with specific probiotic may further improve the hypoglycemic effects of the antidiabetic drugs.