《微生物学》课程教学资源（前沿进展）2型糖尿病与能穿透结肠黏液的细菌有关（英文原文）Structural modulation of gut microbiota during alleviation of type 2 diabetes with a Chinese herbal formula.pdf

OPEN ORIGINAL ARTICLE Structural modulation of gut microbiota during alleviation of type 2 diabetes with a Chinese herbal formula Jia Xu1,4, Fengmei Lian2,4, Linhua Zhao2 , Yufeng Zhao3 , Xinyan Chen2 , Xu Zhang1 , Yun Guo2 , Chenhong Zhang1 , Qiang Zhou2 , Zhengsheng Xue1 , Xiaoyan Pang1 , Liping Zhao1,3 and Xiaolin Tong2 1 State Key Laboratory of Microbial Metabolism, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, People’s Republic of China; 2 Guang’anmen Hospital, China Academy of Chinese Medical Sciences, Beijing, People’s Republic of China and 3 Ministry of Education Key Laboratory of Systems Biomedicine, Shanghai Center for Systems Biomedicine, Shanghai Jiao Tong University, Shanghai, People’s Republic of China The gut microbiota is hypothesized to have a critical role in metabolic diseases, including type 2 diabetes (T2D). A traditional Chinese herbal formula, Gegen Qinlian Decoction (GQD), can alleviate T2D. To find out whether GQD modulates the composition of the gut microbiota during T2D treatment, 187 T2D patients were randomly allocated to receive high (HD, n ¼ 44), moderate (MD, n ¼ 52), low dose GQD (LD, n ¼ 50) or the placebo (n ¼ 41) for 12 weeks in a double-blinded trial. Patients who received the HD or MD demonstrated significant reductions in adjusted mean changes from baseline of fasting blood glucose (FBG) and glycated hemoglobin (HbA1c) compared with the placebo and LD groups. Pyrosequencing of the V3 regions of 16S rRNA genes revealed a dosedependent deviation of gut microbiota in response to GQD treatment. This deviation occurred before significant improvement of T2D symptoms was observed. Redundancy analysis identified 47 GQDenriched species level phylotypes, 17 of which were negatively correlated with FBG and 9 with HbA1c. Real-time quantitative PCR confirmed that GQD significantly enriched Faecalibacterium prausnitzii, which was negatively correlated with FBG, HbA1c and 2-h postprandial blood glucose levels and positively correlated with homeostasis model assessment of b-cell function. Therefore, these data indicate that structural changes of gut microbiota are induced by Chinese herbal formula GQD. Specifically, GQD treatment may enrich the amounts of beneficial bacteria, such as Faecalibacterium spp. In conclusion, changes in the gut microbiota are associated with the antidiabetic effects of GQD. The ISME Journal (2015) 9, 552–562; doi:10.1038/ismej.2014.177; published online 3 October 2014 Introduction Type 2 diabetes (T2D), which is characterized by low-grade inflammation, insulin resistance (Shoelson, 2006) and b-cell failure (Butler et al., 2003), has become increasingly prevalent worldwide (Xu et al., 2013). The estimated proportion of diabetes among adults is 8.3% in 2010, among which T2D accounts for at least 90% (Alberti and Zimmet, 1998; Whiting et al., 2011). This proportion is projected to increase to 9.9% by 2030 (Whiting et al., 2011). Development of T2D results mostly from obesity, which has low-grade inflammation and insulin resistance (Hotamisligil, 2006). The gut microbiota may have a vital role in obesity development (Backhed et al., 2004; Collins et al., 2013; Le Chatelier et al., 2013; Zhao, 2013). For example, endotoxin produced by an opportunistic pathogen in the gut, such as Escherichia coli, induced obesity and insulin resistance when a purified form was subcutaneously infused into mice (Cani et al., 2007a). A greater abundance of opportunistic pathogens, such as Betaproteobacteria, was found in the gut of diabetic patients compared with healthy controls (Larsen et al., 2010). A more recent comparative metagenomic analysis of the fecal samples of 171 diabetic patients and 174 Correspondence: L Zhao, Ministry of Education Key Laboratory of Systems Biomedicine, Shanghai Center for Systems Biomedicine, Shanghai Jiao Tong University, Room 3-517, Biology Building, 800 Dongchuan Road, Shanghai 200240, People’s Republic of China or X Tong, Guang’anmen Hospital of China Academy of Chinese Medical Sciences, Room 432, Administration Building, 5 Beixiange Street, Xuanwu District, Beijing 100053, China. E-mail: lpzhao@sjtu.edu.cn or xiaolintong66@sina.com 4 These authors contributed equally to this work. Received 11 September 2013; revised 18 June 2014; accepted 15 August 2014; published online 3 October 2014 The ISME Journal (2015) 9, 552–562 & 2015 International Society for Microbial Ecology All rights reserved 1751-7362/15 www.nature.com/ismej

healthy controls showed that diseased samples had lesser abundance of butyrate-producing bacteria, such as Faecalibacterium prausnitzii, but greater abundance of opportunistic pathogens, including Clostridium bolteae and Desulfovibrio sp. (Qin et al., 2012). Another study found that the early onset of high-fat-diet-induced T2D was characterized by an increased bacterial translocation from the intestine towards tissues (Amar et al., 2011). An opportunistic pathogen, Enterobacter cloacae B29, isolated from the gut of a morbidly obese and diabetic patient, induced obesity and insulin resistance in germ-free mice (Fei and Zhao, 2013). Taken together, these studies indicate that a dysbiotic gut microbiota may causatively contribute to obesity and diabetes development, and thus may serve as a potential new target for disease control. To treat obesity, T2D and other metabolic diseases, several attempts have been made to target the gut microbiota (Cani et al., 2007b, 2009; Park et al., 2013). Berberine, the major pharmacologic component of a Chinese herb Coptis chinensis (HuangLian) originally used to treat bacterial diarrhea (Khin Maung et al., 1985; Rabbani et al., 1987; Tang et al., 2009), showed clinical efficacy in treating diabetes in a multicentered, randomized, double-blinded and placebo-controlled clinical trial (Zhang et al., 2008). The herb C. chinensis has been used in traditional Chinese medicine (TCM) for diarrhea control for nearly 2000 years. Our recent study showed that berberine prevented high-fatdiet-induced obesity and insulin resistance, enriched short-chain fatty acid-producing bacteria, reduced numbers of opportunistic pathogens and alleviated inflammation in Wistar rats (Zhang et al., 2012b). Drugs that treat bacterial diarrhea, such as berberine, might be useful for T2D treatment because both diseases share a dysbiotic gut microbiota. A standardized berberine-containing Chinese herbal formula, Gegen Qinlian Decoction (GQD), has been a treatment for diarrhea in Shang Han Lun since the East Han Dynasty. Its use was recorded by the prestigious physician Zhongjing Zhang (AD 150– 219). Subsequently, GQD has been reported to have potentially beneficial effects in the treatment of diabetes in animal trials, as well as in some clinical observations. For example, GQD significantly reduced fasting blood glucose (FBG) and glycated hemoglobin (HbA1c) in streptozotocin (STZ) and high-fat-diet-induced diabetic SD rats, and the serum of SD rats that received GQD enhanced glucose consumption in 3T3-L1 adipocytes (Zhang et al., 2013). T2D patients treated with a high dose of modified GQD two times daily for 3 months showed a reduction in HbA1c of 1.79% from the initial level of 9.2%. This decrease was significantly different from that of patients receiving a low-dose treatment (Tong et al., 2011). However, these studies are either animal trials or open, non-placebo-controlled clinical investigations with small sample sizes. Moreover, the mechanism underlying GQD’s impact on glycemic efficacy has barely been elucidated. A recent nuclear magnetic resonance-based plasma metabonomic study revealed that 5 weeks of GQD treatment conspicuously modulated gut microbial metabolism by degradation of choline into methylamines, together with a decrease in FBG and an expansion of islets in STZ and high-fat-diet-induced diabetic rats (Tian et al., 2013). This finding indicates that the gut microbiota might have a pivotal role in the effect GQD has on diabetic subjects. However, there is still a lack of direct evidence showing that GQD can modulate gut microbiota, particularly in humans. In this study, we conducted a randomized, double-blinded, placebo-controlled clinical trial to evaluate the efficacy and safety of GQD in the treatment of T2D. Furthermore, we examined the structural alterations of gut microbiota in response to GQD treatment intended to alleviate T2D. Materials and methods Study design The study was a 12-week, randomized, doubleblinded and placebo-controlled clinical trial that included a 2-week washout period. It was approved by the Ethics Committee of Guang’anmen hospital of China Academy of TCM. Participants were recruited by Guang’anmen hospital, Dong Zhimen Hospital affiliated to Beijing TCM University, China–Japan Friendship Hospital or Ji Shui Tan Hospital of Beijing from August 2010 to May 2011. All participants signed informed consent forms before beginning the study. The study was conducted in accordance with the principles of the Declaration of Helsinki. The inclusion and exclusion criteria of the patients’ enrollments can be found in the Supplementary Materials and methods. Using an initial screening, including FBG test and 75-g oral glucose tolerance test, 629 recently diagnosed T2D patients who had not received prior pharmacologic treatment for T2D were recruited into the study. After a 2-week washout period and the review of a series of examinations, 403 patients were excluded for not meeting the inclusion criteria and 2 patients were excluded for other reasons. The remaining 224 patients were randomly assigned to four groups of 56 patients. Each group received one of the following treatments: high (HD), moderate (MD) or low dose (LD) GQD, or the placebo for 12 weeks. Randomization was performed centrally and was concealed and stratified in blocks of eight by the PROC PLAN process using the SAS software (SAS Institute Inc., Cary, NC, USA). After the study was completed, a total of 187 patients were included for the final analysis by the verification of data examination committee (Supplementary Figure 1). Microbiota shift in alleviation of type 2 diabetes J Xu et al 553 The ISME Journal

Drug administration The TCM formula in our study was GQD, composed of four herbs, namely: Gegen (Radix Puerariae), Huangqin (Radix Scutellariae), Huanglian (Rhizoma Coptidis) and Gancao (Honey-fried Licorice Root) (Supplementary Figure 2a). Supplementary Table 1 lists the amount of each herb in one unit of GQD formula in each group. Herbs were all provided and quality controlled by Beijing Shuangqiaoyanjing Chinese herb manufacturer (Supplementary Materials and methods). The TCM intervention and placebo were given as decoction; these were prepared by Beijing Jiulong Pharmaceutical Factory according to a standard production process (Supplementary Materials and methods). Each unit of GQD formula or placebo yielded 300 ml of decoction. Each patient orally took 150 ml of the decoction two times daily for 12 weeks. All of the drugs and decoctions were quality controlled throughout the trial, and the placebo decoction was prepared by the same standardized process (Supplementary Materials and methods). Study evaluation and outcomes The following primary efficacy outcomes were used: changes in HbA1c, FBG and 2-h postprandial blood glucose (2h-PBG) levels. Secondary efficacy outcomes included changes in serum insulin, lipids levels and body mass index. Study assessments were performed at 0, 4, 8 and 12 weeks. Measurements of FBG, 2h-PBG, body mass index, waist circumference and hip circumference were taken at 0, 4, 8 and 12 weeks. Serum HbA1c, insulin, total cholesterol, triglycerides, high-density lipoprotein cholesterol and low-density lipoprotein cholesterol measurements were performed at weeks 0 and 12. Fecal samples were collected every 4 weeks until the end of trial for gut microbiota analysis. Clinical and biochemical measurements Biochemical measurements of glucose, serum lipids, HbA1c and insulin were performed in a central laboratory (Guang’anmen Hospital of China Academy of TCM, Beijing, China). Glucose, serum total cholesterol, triglycerides, high-density lipoprotein cholesterol and low-density lipoprotein cholesterol were measured by enzymatic methods (Olympus AU2700; Olympus Co. Ltd., Tokyo, Japan). HbA1c was measured by high-performance liquid chromatography using a variant hemoglobin HbA1c assay (ADAMS A1c HA-8160; Arkray Inc., Kyoto, Japan). Serum insulin was measured using a double-antibody RIA (ADVIA Centaur; Bayer Diagnostics, Leverkusen, Germany). ELISA Kits were used to measure plasma orosomucoid (Assaypro, St Charles, MO, USA), adiponectin (R&D System, Minneapolis, MN, USA), tumor necrosis factor-a (R&D System) and serum amyloid A protein (Invitrogen, Carlsbad, CA, USA) levels. Fecal DNA extraction and pyrosequencing Genomic DNA of each fecal sample was extracted by a InviMag Stool DNA Kit (Invitek, Berlin, Germany) combined with bead beating as previously published (Zhang et al., 2012b). The extracted genomic DNA was used as the template to amplify the V3 region of 16S rRNA genes. PCR reactions, pyrosequencing of the PCR amplicons and quality control of raw data were performed as described previously with minor modification (Zhang et al., 2009; Wang et al., 2011). Bioinformatics and multivariate statistics High-quality sequence alignments were performed using NAST. Sequence clustering by CD-hit and OUT delineation by DOTUR were performed as described previously (Zhang et al., 2012a, b). The representative sequences of operational taxonomy units (OTUs) with their relative abundance were used to calculate rarefaction analysis and Shannon diversity index by QIIME (Caporaso et al., 2010). In addition, the representative sequences were inserted into a pre-established phylogenetic tree of the full-length 16S rRNA gene sequences in ARB (Ludwig et al., 2004). Then, the phylogenetic tree and the relative abundance table of representative sequences of OTUs were used for UniFrac principal coordinate analysis (PCoA) (Lozupone and Knight, 2005). The statistical significance between different groups was assessed by multivariate analysis of variance in MATLAB 2010b (The MathWorks Inc., Natick, MA, USA). Redundancy analysis was performed using CANOCO for Windows 4.5 (Microcomputer Power, Ithaca, NY, USA) according to the manufacturer’s instructions (Braak and Smilauer, 2002). Statistical significance was assessed by MCPP with 499 random permutations under the full model. Ribosomal Database Project Classifier was used to assess the amounts of different genera by taxonomic assignment of all sequences. Real-time quantitative PCR of F. prausnitzii Real-time quantitative PCR (q-PCR) was used to determine the amounts of total bacteria and F. prausnitzii through detection of 16S rRNA genes. A set of universal primers was used to amplify a conserved 16S rDNA sequence in all bacteria as shown before (Wang et al., 2011). A set of specific primers was used to amplify a conserved 16S rDNA sequence in F. prausnitzii and the q-PCR reaction system and the program was described before (Balamurugan et al., 2008). A plasmid containing a F. prausnitzii full-length 16S rDNA from a previous study (Shen et al., 2006) was prepared using the EZNA Plasmid Mini Kit I (OMEGA, Doraville, GA, USA) and diluted from 1 103 to 1 109 (copies ml 1 ) to construct a standard curve for the detection of F. prausnitzii. We selected reactions with efficiencies ranging from 0.90 to 1.05 for Microbiota shift in alleviation of type 2 diabetes J Xu et al 554 The ISME Journal

further analysis. Standard and quantified samples were performed in triplicate. PCR reactions were performed using iQ SYBR Green SuperMix (BioRad, Richmond, CA, USA) on a MasterCycler ep Realplex 4s (Eppendorf, Hamburg, Germany). Spearman’s correlation coefficient (R) and P-value were used to compare the amounts of F. prausnitzii measured by q-PCR and pyrosequencing. This coefficient was also used to evaluate the relationship between F. prausnitzii and clinical parameters using MATLAB 2010b. Results The major components of GQD decoction There were four major categories of compounds in the GQD decoction. Flavones (baicalin, puerarin, wogonoside, daidzin, liquiritin, baicalein and wogonin), alkaloids (berberine, coptisine, palmatine and jatrorrhizine) and triterpenoid sapnins (glycyrrhizin) were detected in the decoction, among which baicalin, puerarin and berberine were the major components (Supplementary Table 2). The chemical structures of these 12 components are shown in the Supplementary Figure 2c. Carbohydrates (starch, sucrose, reducing sugar and soluble dietary fiber) were also detected. Insoluble dietary fiber was undetectable in GQD decoction (Supplementary Table 3). GQD significantly improved glycemic control in T2D patients In our 12-week, randomized, double-blinded, placebo-controlled clinical trial (Supplementary Figure 1), the data of 187 participants were analyzed as shown in Supplementary Table 4. The baseline variables were not significantly different among the four groups. After 12 weeks of treatment, GQD significantly improved glycemic control in T2D patients. The HD and MD groups, when compared with the placebo and LD groups, showed significant reductions in adjusted mean changes from baseline of FBG ( 1.46±0.23 and 1.09±0.21 vs 0.16±0.22 and 0.24±0.24 mmol l 1 ; Po0.001 for HD vs LD and placebo; Po0.01 for MD vs LD and placebo). Similarly, the HD and MD groups showed significantly reduced HbA1c ( 0.88±0.14 and 0.75±0.13 vs 0.35±0.13 and 0.36±0.15%; Po0.01 for HD vs LD; Po0.05 for HD vs placebo; Po0.05, MD vs LD and placebo) (Figures 1a and b). A decrease in the mean change of 2h-PBG from baseline was also observed in the treated groups, although not reaching significant level. (Supplementary Figure 3). In addition, ANCOVA analysis showed that HOMA-b was significantly improved by HD GQD treatment compared with the placebo and LD groups (Figure 1c). Plasma orosomucoid was significantly reduced by HD GQD treatment (P ¼ 0.023) (Supplementary Figure 4a) and the HD group showed a significant reduction (P ¼ 0.034) in mean change from baseline of plasma orosomucoid compared with the LD group (Supplementary Figure 4b). No significant difference was observed in plasma adiponectin, tumor necrosis factor-a or serum amyloid A among the four groups (Supplementary Figures 5a–c). Finally, no drug-related serious adverse events occurred in this study. Overall structural modulation of gut microbiota after GQD treatment First, we used a bar-coded pyrosequencing run to analyze the structural changes of gut microbiota in the four groups before and after GQD treatment. In total, 483 304 usable raw sequences (34 753 unique sequences) and 3222 OTUs were obtained from 235 samples with an average of 2057±396 per sample. Rarefaction and Shannon diversity curves revealed that, although no rarefaction curves plateaued with the current sequencing, most of the diversity had already been captured (Supplementary Figure 6). Weighted and unweighted UniFrac PCoA analysis revealed that gut microbiota structure of the treated groups showed a dose-dependent deviation Figure 1 GQD significantly improved glycemic control and HOMA-b in T2D patients. (a) Change in FBG, (b) change in HbA1c and (c) change in HOMA-b. Placebo (n ¼ 41), LD (n ¼ 50), MD (n ¼ 52) and HD (n ¼ 44). Data are presented as mean±S.E.M. *Po0.05, **Po0.01 and ***Po0.001 vs placebo using ANCOVA; þ Po0.05, þ þ Po0.01 and þþþ Po0.001 vs LD using ANCOVA. Microbiota shift in alleviation of type 2 diabetes J Xu et al 555 The ISME Journal

from the baseline structure, with the HD group reaching significant level in multivariate analysis of variance test (Figures 2a and b and Supplementary Figures 7a and b). To monitor the dynamic changes of gut microbiota during GQD treatment, we analyzed the fecal samples collected at weeks 0, 4, 8 and 12 in HD and placebo groups with a second pyrosequencing run. In total, we generated 680 774 usable raw sequences (37 498 unique sequences) and 4251 OTUs from 288 samples with an average of 2364±443 per sample (one sample was excluded in later analysis because only 81 reads were obtained). Rarefaction and Shannon diversity curves revealed that most of the diversity had already been captured (Supplementary Figure 8). UniFrac PCoA and PCA showed that after 4 weeks of treatment, the gut microbiota structure of the HD group had already significantly diverged from that of its baseline and of the placebo group (Figures 3a and b; Supplementary Figure 9). At that same 4-week analysis, 2h-PBG was significantly reduced, but FBG did not reach a significant level in treated groups (Supplementary Figure 10). As the treatment Figure 2 Dose-dependent alterations of the gut microbiota in T2D patients treated with different doses of GQD at weeks 0 and 12. (a) Weighted Unifrac PCoA of gut microbiota based on the OUT data from the first pyrosequencing run. (b) Clustering of gut microbiota based on mahalanobis distances calculated with multivariate analysis of variance (MANOVA). Each point represents the mean principal coordinate (PC) score of all patients in a group at one time point, and the error bar represents the s.e.m. The sample number (n) at week 0: placebo ¼ 30, LD ¼ 28, MD ¼ 32 and HD ¼ 28. The sample number (n) at week 12: placebo ¼ 30, LD ¼ 28, MD ¼ 32 and HD ¼ 28. ***Po0.0001. Figure 3 Trajectory of the gut microbiota in T2D patients treated with HD GQD and placebo at weeks 0, 4, 8 and 12. (a) Unweighted Unifrac PCoA of gut microbiota based on the OUT data from the second pyrosequencing run. (b) Clustering of gut microbiota based on mahalanobis distances calculated with multivariate analysis of variance (MANOVA). Each point represents the mean principal coordinate (PC) score of all patients in a group at one time point, and the error bar represents the s.e.m. Placebo: n ¼ 36; HD: n ¼ 36. ***Po0.0001. Microbiota shift in alleviation of type 2 diabetes J Xu et al 556 The ISME Journal