A randomized and open-label, controlled clinical trial was designed and conducted according to Consolidated Standards of Reporting Trials guidelines (16). Participants were all recruited from the community healthcare center at West Renmin Road, Yanquan Road and Xiameiqiao Road in Chenzhou, Hunan, China from January 2019 to December 2020. All participants with T2DM signed written informed consentforms before enrollments. The trial protocol was registered at Chinese Clinical Trial Registry (http://www.chictr.org.cn/) and got a registration number of ChiCTR1900020482. Chinese newly onset, treatment-naive T2D patients from Han nationality with the age of 35- to 70-year old were enrolled in this trial. The inclusion criteria were as follows: (1) fasting plasma glucose (FBG) ≥7 mmol/L or/and 2-h oral glucose tolerance test ≥11.1 mmol/L; (2) HbA1c ≥6.5% (6.5%≦HbA1c≦12%) (17, 18). The exclusion criteria were listed in Table S1 ( Supplementary Material Content 1 ). The ethics committee of Affiliated Hospital of Xiangnan University approved the study protocol (Ethical approval department No. KY-20181226001).Patients who met all inclusion criteria were randomly assigned to either the C. paliurus leaves extracts group (CP group) or the Glipizide group (G group) in a 2: 1 ratio using a computer-generated random number allocation by a researcher not involved in the study.

According to Movahed et al., the HbA1c difference between intervention group and control group was expected to be 1.2% at the end of the intervention (19). Combined with a standard deviation of 1.09%, α=0.05 and power=0.80, we calculated that a minimum of 11 T2DM patients were required in each group. Assuming a shedding rate of 10%, each group needed12 T2DM patients at least according to the actual situation. For this clinical trial, a minimum of 36 T2DMpatients were required.

Before starting the clinical trial, all the enrolled participants underwent a 2-week run-in period and received a diet and exercise education about the daily management of T2DM. Once the 2-week run-in period completed, participants entered the intervention study immediately. In the CP group, the participants received CP powder (2g/time, 3 times/day orally before meals) (CP were provided by Chenzhou Mingrun biological products Co. LTD, Hunan, China) ( Figure S1 ). In the G group (the positive-control group), the participants received Glipizide Controlled-Release Tablets (5mg/time, 2 times/day orally before meals) (Pfizer Pharmaceuticals LLC, hereinafter referred to as Glipizide) that has no significant effect on gut microbiota of T2DM patients (20). During the intervention, all participants were required to monitor FBG, and 2-hour post-meal blood glucose (2hPBG) using Roche glucometer (Accu-Chek Performa) every day.

Cyclocarya paliurus leaves extracts (CP)were provided by Chenzhou Ming run biological products Co. LTD Hunan, China. The license number of this product is QS431006011417.The leaves of C. paliurus has long been used as a bitter Traditional Chinese Medicine which has also been historically used as an herbal tea in the folk. Nowadays, the products derived from the leaves of C. paliurus have become a very popular health product in China (11). CP were prepared according to the method described previously with some modifications (21). Briefly, the leaves of Cyclocarya paliurus were dried and pulverized before extraction. Cyclocarya paliurus leaves powders were soaked with distilled water at 90 °C for 1 h (1:30, g/mL), and the extraction solution was centrifuged at 4500 × g for 15 min. The above operation was repeated once. The supernatants (extraction solutions, two times) were combined and concentrated using a rotary evaporator under vacuum. The resulting residue was dissolved with deionized water, filtered through a 800 mesh standard sieve. The elution was performed using 80% ethanol and the effluent of ethanol solution was collected and concentrated, resulting in the CP.

The composition of the C. paliurus leaves extracts was determined by high performance liquid chromatography (HPLC, Agilent 1200, USA) equipped with Agilent 5 TC-C18 (4.6 mm ×250 mm, 5μm). The system parameters were set as follows: injection volume, 5μL; the column temperature, 30°C; and the mobile phase flow rate, 0.8mL/min. The mobile phase consisted of two solvents: (A) 0.2% aqueous acetic acid and (B) acetonitrile. The elution conditions were as follows: 0–5min, gradient 5% B; 5–10 min, linear gradient 5–10% B; 10–15 min, linear gradient 10–25% B; 15–25 min, linear gradient 25–40% B; 25–30min, linear gradient 40–90% B. Peaks were detected at 254 nm. The quantification of different component monomers was based on the peak areas and calculated as equivalents of the standard compounds; all contents were expressed as mg/g component dry weight. The temperature of the column oven was set at 40 °C, the flow rate was set at 0.6 mL/ min and the injection volume was 5 µL.

Participants were asked not to make any change to their daily physical activity and to maintain body weight. On Day 0, Day 42, and Day 84, the 7-day recall method was used to measure physical activity energy expenditure through an interview-administered survey instrument modified from the Cross-Cultural Activity Participation Study (22). Metabolic Equivalent Tasks (METs) were conducted to measure the physical activity energy expenditure of all the participants. On Day 0 and Day 84, the medical history and anthropometric evaluation (body weight (BW), waist circumference (WC), waist to hip ratio (WHR), body mass index (BMI), systolic blood pressure (SBP) and diastolic blood pressure (DBP)) were recorded and fasting blood samples and fecal samples were collected. Oral glucose tolerance test(OGTT), insulin release test and C-Peptide release test were also performed. Further details about the physical activity assessment can be found online ( Supplementary Methods, Supplementary Material Content 2 ). Adverse effects were also recorded by the research team.

The primary clinical outcome was a change in the level of HbA1c over 84 days in the two treatment groups. The secondary clinical outcomes included the changes in the levels of FBG, 2hPBG, the area under curve of OGTT blood glucose, fasting insulin (FINS), 2-hour post-meal insulin (2hPINS), the area under curve of insulin release test, fasting C- peptide (FCP), 2-hour post-meal C-Peptide (2hPCP), the area under curve of C-peptide release test, total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-c), high-density lipoprotein cholesterol (HDL-c), BW, WC, WHR, BMI, SBP, and DBP.

The stool samples were immediately transferred to the -80°Cfreezer for storage after collection on day 0 and day 84. Then the stool samples were used for bacterial 16sRNA gene sequencing, and SCFAs as well as BAs analysis. The detailed information was shown on Supplementary Material Content 2 .

The liver function (the total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), albumin (ALB), alanine aminotransferase (ALT), aspartate aminotransferase (AST)) and renal function (serum creatinine (Scr) and blood urea nitrogen (BUN))of all participants was assessed on day 0 and day 84.

The animal experimental protocol was approved by the animal ethics committee of Jinan University (20190821-02). A total 50 of C57BL/6J mice (weight, 25.00 ± 2g; age, 10 weeks) were purchased from Biotechnology Company Limited (Beijing, China). The mice were raised in the specific pathogen-free animal experiment center at Jinan University, with constant temperature (24 ± 2°C), constant humidity of 55 ± 10% and a 12-h light/dark cycle. All mice received adaptive feeding for 7 days, and then were randomly divided into a control group (NC, n=10) and an antibiotic treatment group (AT, n=40). Within the next 4 weeks, the mice in AT group were fed with antibiotic-containing sterile water and then divided into 4 groups. More detailed information was shown in Supplementary Material Content 2 .

Outcomes were calculated on a strict intention-to-treat (ITT) basis. The raw data of Day 84 – the raw data of Day 0 represented the changes in all variables. The Kolmogorov-Smirnov Test was used to detect whether the continuous data conform to normal distribution. If normally distributed, paired T-test was used for intra-group comparison and independent T-test was used for inter-group comparison, and the data were expressed as mean and standard deviations. If not normally distributed, non-parametric two-tailed Wilcoxon matched-pairs signed-ranks was used for intra-group comparison and non-parametric K-independent Wilcoxon signed-ranks were used for inter-group comparison, and the data were expressed as the median and inter-quartile ranges. Fold changes of diabetes-induced markers after treatments of CP & G were calculated as (the raw data of day 84/the raw data of day 0). Statistical Package for Social Science (SPSS) 28.0 was used to analyze all data and we defined P<0.05 as statistically significant. All data visualization were made by GraphPad Prism (version 9.0). We used the OmicStudio tools at https://www.omicstudio.cn.to make a clustering correlation heatmap.

Phytochemical compositions of the CP were clarified by using high performance liquid chromatography (HPLC). The 5 components in the CP were identified in the chromatogram with peaks at 17.19 min to3-cafeoylquinic acid, 20.19 min to isoquercitrin, 22.19 min to kaempferol-3-glucoside, 23.43 min to kaempferol 3-rhamnoside, 27.56 min to quercetin by the comparison to external standards ( Figure S2 ). Also, the contents of five compounds in the CP were measured by HPLC and their contents were 53.4, 42.5, 92.6, 4.8 and 2.1 mg/g in the CP, successively ( Table S2, Supplementary Material Content 1 ).

According to the inclusion and exclusion criteria, a total of 38 participants with T2DM were randomly divided into the G group (n=13) and CP group (n=25) in the study. During the intervention, 6 participants dropped out (1 participant in the G group and 5 participants in the CP group). Finally, a total of 32 participants completed the intervention ( Figure S3 ). At baseline, anthropometric parameters (BW, WC, WHR, BMI, SBP and DBP), blood glucose homeostasis (HbA1c, FBG, 2hPBG, FINS, 2hPINS, HOMA-IR, Fasting C-Peptide, 2hPCP, OGTT glucose AUC, Insulin release test AUC, C-Peptide release test AUC), blood lipid homeostasis (TG, TC, LDL-c and HDL-c) and other blood markers (TBIL, DBIL, IBIL, ALT, AST, ALB, Scr, BUN) were not statistically different between the two groups ( Table S3, Supplementary Material Content 1 ). In addition, there were no significant differences in physical activity levels in the intra-group and inter-group comparisons on Day 0 and on Day 84 ( Table S4, Supplementary Material Content 1 ).

Compared with Day 0, the level of HbA1c, the primary clinical outcome, was significantly decreased both in the G group and in the CP group on Day 84. At the end of intervention, there was no significant difference in the reduction of HbA1c between the two groups ( Figure 1A ; Table 1 ). We also calculated the proportion of participants who achieved adequate glycemic control (HbA1c<7%). On Day 0, there were no significant differences in the proportion of participants who achieved HbA1c<7% between the two groups, and both the G group and the CP group had a significant increase in the proportion of participants who achieved HbA1c<7% on Day 84 compared with Day 0. We also found that there was no significant difference in the proportion of participants who achieved HbA1c<7% on Day 84 between the two groups ( Figure 1B ; Table S5, Supplementary Material Content 1 ). These results demonstrate that the CP induce a significant reduction in HbA1c, which is similar to the effect of Glipizide on the reduction of HbA1c in T2DM patients.

From Day 0 to Day 84, the significant improvement in blood glucose homeostasis (FBG, 2hPBG, OGTT glucose AUC) ( Table 1 ; Figures 1C, D , 2A ), blood lipid homeostasis (TG, TC, LDL-c and HDL-c) ( Table 1 ; Figures 3A–D ) and anthropometric parameters (SBP and DBP) ( Table 1 ; Figures S4E, F )was observed, whereas the other blood glucose homeostasis (FINS, 2hPINS, HOMA-IR, fasting C-peptide, 2hPCP, insulin release test AUC and C-peptide release test AUC) ( Table 1 ; Figures 1E–I , 2B, C ) and anthropometric parameters (BW, WC, WHR and BMI) ( Table 1 ; Figures S4A–D ) were not significantly affected in the CP group. From Day 0 to Day 84, the significant amelioration in blood glucose homeostasis (FBG, 2hPBG, OGTT glucose AUC, FINS, 2hPINS, fasting C-peptide, 2hPCP, insulin release test AUC and C-peptide release test AUC) was found, while the other blood glucose homeostasis (HOMA-IR), blood lipid homeostasis (TG, TC, LDL-c and HDL-c) and anthropometric parameters (BW, WC, WHR and BMI) were not significantly affected in the G group. The fold changes in blood lipid homeostasis (TG and HDL-c) and anthropometric parameters (DBP) was significantly greater in the CP group compared with the G group. Moreover, there were no significant differences in the fold changes in the other blood glucose homeostasis (FBG, 2hPBG, FINS, 2hPINS, fasting C-peptide, 2hPCP, HOMA-IR, OGTT glucose AUC, insulin release test AUC and C-peptide release test AUC), blood lipid homeostasis (TC and LDL-c) and anthropometric parameters (BW, WC, WHR, BMI and SBP) over the 84-days between the two groups. Also, the liver and kidney function parameters (TBIL, DBIL, IBIL, ALT, AST, ALB, Scr and BUN)did not significantly change in both CP group and the G group over the 84-day period ( Table 1 ; Figures 1 - 3 , S4 , 5 ).These results suggest that CP can alleviate T2DM-associated clinical markers, blood lipid and blood pressure in addition to alleviating HbA1c in T2DM patients, having no significant impacts on the liver and kidney function.

To explore the potential role of the gut microbiota in CP-induced metabolic phenotypes improvement in T2DM patients, 16S rRNA gene sequencing was performed in the fecal samples from the patients on Day 0 and Day 84. At the phylum level, the overall microbial compositions were shown in Table S6 ( Supplementary Material Content 1 ). The dominant species (mean relative abundance >1%) were Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Verrucomicrobia and Fusobacteria ( Figure 4A ). From Day 0 to Day 84, only the abundance of Verrucomicrobia was significantly increased, whereas there were no significant differences in the abundances of Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Fusobacteria, Synergistetes, Cyanobacteria and Patescibacteria in CP group. In the G group, the abundances of the gut microbial at the phylum level were kept stable during the intervention ( Table S6, Supplementary Material Content 1 ; Figures 4B–F ).

At the genus level, the top 10 dominant genera was presented in Figure 5A . Compared with Day 0, the abundances of potentially beneficial bacteria (Faecalibacterium and Akkermansia) were significantly enriched, while the abundance of potentially pathogenic bacteria (Prevotella_9) was significantly reduced on Day 84 in the CP group ( Table S7, Supplementary Material Content 1 ; Figures 5B–D ). In addition, the abundances of other genera of bacteria showed no significant change in the CP group during the intervention. In the G group, the abundances of the gut microbiota at the genus level were not significantly affected from Day 0 to Day 84 ( Table S7, Supplementary Material Content 1 ; Figures 5E–G ).

Then we conducted a spearman correlation analysis to explore the potential relationships between bacterial abundance and metabolic phenotypes. We found that Faecalibacterium, which was enriched in the participants on Day 84 in the CP group, had a negative correlation with several metabolic phenotypes (HbA1c, FBG, 2hPBG, SBP and TC). Akkermansia, which was also enriched in the participants on Day 84 in the CP group, was negatively related to a few metabolic markers (HbA1c, FBG and TC). Prevotella_9 had a positive relationship with several metabolic markers (FBG, OGTT Glucose AUC, 2hPBG and TC) ( Figure 5H ). These findings indicate that CP-induced metabolic phenotype alleviation maybe associated with improved intestinal flora.

To reveal metabolites alteration, related to the gut microbiota, which is potentially involved in CP-induced T2DM improvement, we performed metabolic profiling of feces from the participants. From Day 0 to Day 84, a significant increase was observed in total SCFAs, propionic acid (PA) and butyric acid (BA), while the levels of acetic acid (AA), isobutyric acid (IBA), valeric acid (VA), isovaleric acid (IVA) and hexanoic acid (HA) had no significant change in the CP group. In the G group, the level of total SCFAs, AA, PA, BA, IBA, VA, IVA and HA were not affected over the 84-day period ( Table S8, Supplementary Material Content 1 ; Figures 6A–E ).

To explore the potential effect of CP on BA metabolism, a total of 20 species of BAs were detected. Compared to Day 0, the levels of unconjugated BAs (chenodeoxycholic acid (CDCA), 12-ketolithocholic acid (12-KLCA) and β-muricholic acid (β-MCA)) were significantly increased and the levels of conjugated BAs (taurodeoxycholic acid (TDCA),glycochenodeoxycholic acid (GCDCA) were significantly reduced, while no significant differences were found in other BAs in the CP group on Day 84. From Day 0 to Day 84, the levels of BAs showed no significant changes in the G group ( Table S9, Supplementary Material Content 1 ; Figures 7A–F ).

Correlation analyses were performed to determine the potential relationships between metabolites and gut microbiota. Coincidently, we observed that Faecalibacterium enriched in the CP group on Day 84 was positively related to SCFAs (total SCFAs, BA and PA) and unconjugated BAs (CDCA, 12-KLCA, β-MCA and LCA), but negatively correlated with conjugated BAs (GCA). Akkermansia, which was also enriched in the CP group on Day 84, showed a positive relationship with SCFAs (total SCFAs, BA, HA, IBA and PA), and unconjugated BAs (CDCA, 12-KLCA and β-MCA), but had a negative correlation with conjugated BAs (GUCDA, GCA, TDCA and TCDCA). Prevotella_9, which was depleted in the CP group on Day 84, was negatively correlated with SCFAs (total SCFAs, IBA and PA) and unconjugated BAs (CDCA and β-MCA), but positively associated with conjugated BAs (GCA, GLCA and TLCA). Megamonas was negatively correlated with conjugated BAs (GCA and GLCA), but positively associated with SCFAs (HA) and unconjugated BAs (β-MCA and CA) ( Figures 6F , 7G ).Taken together, the abovementioned results reveal that the changes in gut microbiome abundances induced by CP may be responsible for the alteration of metabolites, thus affecting metabolic phenotypes in patients with T2DM.

Experimental design for animal study for microbiota transplantation ( Figure 8A ). After the FMT, we observed that body weight was not also statistically significant both between intragroup and between intergroup ( Figure 8B ). Mice transplanted with the postintervention microbiota from the CP group showed better glucose metabolic parameters than those with the preintervention microbiota from the CP group, while mice transplanted with the postintervention microbiota from the G group showed no significant change in glucose metabolism parameters compared with the preintervention microbiota from the G group ( Figures 8C–E ).