The Metabolite β-Hydroxybutyrate of Lactobacillus Plantarum YZX21 Improves Type 2 Diabetes By Promoting Intestinal Secretion of GLP-1

Background: Probiotics and their metabolites regulate type 2 diabetes mellitus (T2DM) by promoting GLP-1 secretion, but the mechanism has not yet been fully claried. Results: In order to reveal the effect of Lactobacillus plantarum (cid:0) L. plantarum(cid:0)YZX21 on the regulation of T2DM, type 2 diabetic C57BL/6 mice induced by a high-fat diet and streptozotocin (STZ) were divided into different groups and were daily treated with L. plantarum YZX21 for 8 weeks. We identied that L.plantarum YZX21 reduced blood glucose, insulin levels and HOME-IR. Histopathology was found that L.plantarum YZX21 restored the mouse islet cells morphology and increased insulin secretion. The Elisa and immumohistochemical staining revealed concentration of glucagon-like peptide-1 (GLP-1) was increased in the mice colon. UPLC-MS/MS based widely-targeted metabolomics analysis was used to identify the differential intestinal metabolites in the mice colon. It was found that the metabolite β-Hydroxybutyrate(BHB) of L.plantarum YZX21 had a negative associated with T2DM. Subsequent, type 2 diabetic C57BL/6 mice was established and used to verify hpyerglycemic effest by daily treated with BHB for 8 weeks. It was found that BHB improved pathoglycemia, insulin resistance and increased the intestinal GLP-1 levels, especially reduced free fatty acid (FFA) levels. The expression of receptors G protein-coupled receptor (GPR) was veried by RT-PCR. The GPR109a receptor was signicantly stimulated which was up-regulated the expression of GLP-1, but not GPR41 and GPR43 in the mice colon. Conclusions: We found that the hypoglycemic effect of L.plantarum YZX21 was demonstrated by increasing the intestinal GLP-1 levels in the diabetic mice. Through the widely-targeted metabolomics, we identify the close correlation between serum concentrations of BHB and T2DM. In T2DM mice model, BHB reduced the FFA levels and increased the intestinal GLP-1 levels, which is associated with the downregulated expression of GPR109a. These results demonstrated that L.plantarum YZX21 alleviated T2DM by upregulating BHB/GPR109a/GLP-1 associated pathway.


Background
Type 2 diabetes mellitus (T2DM) was the third independent risk factor for premature mortality and disability in individuals [1]. Previous studies showed that hormone glucagon like peptide-1 (GLP-1) secreted by intestinal L cells stimulated insulin secretion and controlled blood glucose stability [2]. Therefore, increasing the production and secretion of endogenous GLP-1 had been widely concerned [3].
It is reported that probiotics can colonize in the intestine to regulate blood glucose by increasing the secretion of GLP-1 [4,5]. The probiotics regulated the expression of GLP-1 through metabolites such as bile acids, short chain fatty acids and tryptophan derivatives [6]. However, there was no direct correlation between these metabolites and FFA concentrations. It was report that prolonged FFA stimulation or longterm high-fat diet leaded to impaired secretion of GLP-1 [7]. Since more than half of insulin secretion after eating is induced by GLP-1, impaired GLP-1 secretion may be an important mechanism of lipotoxicity causing blood glucose disorders [8,9]. However, the mechanism which probiotics regulated FFA to increasing secretion of GLP-1 was not fully understood.
Metabolomics has been widely used in research related to disease diagnosis, as well as the exploration of drug action mechanisms and new biomarkers [10]. Widely-targeted metabolomics can re ect the metabolic pathways and be commonly used in research on the response mechanisms of different diseases and microorganisms [11]. With the help of Metabolomics, it was expected to reveal the mechanism of probiotics regulating GLP-1.
This paper used T2DM mice model to study the effects of Lactobacillus plantarum (L.plantarum)YZX21 on the glucose homeostasis as well as the secretion of intestinal GLP-1. And through the widely-targeted metabolomic analyses was to identify the target metabolites that stimulate GLP-1 secretion and reveal the regulatory mechanism of probiotics.

Results
The effect of L.plantarum YZX21 on body weight, food intake and blood glucose of mice From the fth week, diabetic mice showed typical symptoms of diabetes. The body weight and food intake of mice results are shown in Fig. 1A and 1B. Compared with control group, all diabetic mice showed signi cant weight loss from fth week to twelfth week (p < 0.05). After supplementing with L.plantarum YZX21, the weight of diabetic mice recovered signi cantly from 20.34g to 24.24g, which was not signi cantly different from the M group. The food intake of diabetic mice increased signi cantly from fth week to twelfth week. After the supplement of L.plantarum YZX21, the diet of mice was relieved from 13.28g to 7.03g (p < 0.05). The FBG and PBG levels are shown in Fig. 1C and 1D. Compared with control group, the FBG levels of diabetic mice were higher than 7.0 mmol/L and the PBG levels were higher than 11.1mmol/L. The FBG of mice in the L.plantarum YZX21 group decreased from 12.26 mmol/L to 7.13 mmol/L (p < 0.05). The PBG levels was signi cantly reduced, from 24.68 mmol/L to 12.95 mmol/L (p < 0.05). The OGTT results are shown in Fig. 1E. Compared with control group, the curve area was almost the 3 times larger than the control group. In the L.plantarum YZX21 supplement group, The area under the blood OGTT curve was signi cantly reduced, which was no different from the M group (Fig. 1F).
The effects of L.plantarum YZX21 on insulin secretion and pancreatic islets The results of insulin intraperitoneal injection (ITT) are shown in Fig. 2A. Compared with control group, the area of AUCglucose was 3 times than the control group. The insulin sensitivity of mice in the L.plantarum YZX21 group was signi cantly improved, and the area of AUCglucose decreased signi cantly (Fig. 2B), down 2.4 times (p < 0.05). Figure 2C and D show the fasting insulin levels and HOMA-IR. Compared with the control group, the fasting insulin levels of diabetic mice were signi cantly higher, and the HOMA-IR were also at high levels. After L.plantarum YZX21 intervention, serum insulin levels were signi cantly reduced from 38.42 mmol/L to 24.14 mmol/L and the HOMA-IR was decreased from 17.60 to 7.65, which was no different from the M group. The effects of L.plantarum YZX21 on islet Page 4/24 cells was further veri ed by histopathology. The histopathological staining results of the pancreas are shown in Fig. 2E. The pancreatic islet cells were round or oval, with clear cells, rich and uniform cytoplasm in control group. In diabetic mice, the area ratio of islets to pancreas decreased signi cantly, the cells showed irregular structures, and the volume of islet cells decreased. In the L.plantarum YZX21 group, some pancreatic islet cells shrank and stained very lightly, but the cell morphology was complete, and the morphology was close to normal pancreas. The results of immuno uorescence staining showed ( Fig. 2F) that the insulin of the control group occupied most of the pancreatic islet area, and insulin secretion was normal. While the insulin signal of diabetic mice was greatly reduced and the glucagon signal was signi cantly increased. In the L.plantarum YZX21 group, insulin secretion was signi cantly increased, glucagon secretion was decreased, and the release pressure of islet hormone was signi cantly relieved.
The colonize site of L.plantarum YZX21 and its effects on the intestinal GLP-1 secretion Through the uorescence staining experiment (Fig. 3A), it was found that L.plantarum YZX21 gavaged for 6 h showed high uorescence intensity in the colon, and at 12 h, the colon still maintained high uorescence intensity. This suggested that L.plantarum YZX21 was likely to mainly colonize the colon. As shown in Fig. 3B, the number of GLP-1 positive cells in the colonic mucosa of diabetic mice decreased. While after supplementation with L.plantarum YZX21, the colonic GLP-1 positive cells increased signi cantly from 17 to 31 (Fig. 3C). Consistent with this nding, in diabetic mice, the content of GLP-1 in the colon was signi cantly reduced (Fig. 3D). After supplementing with L.plantarum YZX21, the content of GLP-1 was signi cantly increased from 4.23 pmol/L to 11.07 pmol/L. Further the GLP-1 regulator gene GCG and PC3 mRNA levels were revealed in Fig. 3E and F. It was found that the expressions of GCG and PC3 were signi cantly increased in the colon after supplement of L.plantarum YZX21, where the expression was increased both about 3 times.

Metabonomics analysis of intestinal metabolites of L.plantarum YZX21
A signi cant difference between the metabolites of the diabetes group and the L.plantarum YZX21 group was observed in the OPLS-DA score chart (Fig. 4A), which suggested that there was a signi cant difference between the intestinal metabolism pro les of the two groups. Based on the variable importance in the projection value (VIP) greater than 1, the fold change value than 2 and p-value less than 0.05, it revealed that there were 87 metabolites that are signi cantly different between the diabetes group and the L.plantarum YZX21 intervention group (Fig. 4B). Based on the VIP score and the fold change value, we listed the top 10 metabolites in Table S3. FFA was the most changed metabolite, which was signi cantly reduced in YZX21 group. Through human metabolomic database (HMDB) and association analysis of metabolites with type 2 diabetes (Fig. 4C), it is found that BHB was closely related to FFA metabolic pathway and T2DM. It was found that free fatty acids (FFA) were highly increased in D group, which was nearly 4 times that of the YZX21 group (Fig. 5D). And BHB of L.plantum YZX21 group increased signi cantly, which was more than 3.5 times higher than that of D group (Fig. 5E).
Effects of BHB supplement on blood glucose and pancreatic islets in type 2 diabetic mice To further characterize the effects of BHB on T2DM and FFA, we measured effects of BHB supplement on diabetes-related indicators through the diabetic mice. The FBG levels of mice are shown in Fig. 5A. Compared with control group, the FBG levels of diabetic mice were higher than 7.0 mmol/L. The FBG of mice in the BHB group decreased from 9.72mmol/L to 7.52mmol/L (p < 0.05). The PBG levels were shown in Fig. 5B. The PBG levels of diabetic mice were higher than 11.1mmol/L. After supplementing BHB, the PBG decreased signi cantly, from 21.15mmol/L to 11.88mmol/L (p < 0.05). Fasting insulin levels and HOMA-IR are shown in Fig. 5C and D. Compared with the control group, the fasting insulin levels of diabetic mice were signi cantly higher, and the HOMA-IR levels were also at high levels. However BHB intervention signi cantly reduced the serum insulin levels from 37.67 mIU/L to 21.19 mIU/L and HOMA-IR from 16.57 to 7.32, which were no difference from the M group. The histopathology was shown in Fig. 5E. The islet cells in the control group were rich and uniform. In diabetic mice, the cells presented irregular cell structure. In the BHB group, except for occasional lymphocytes and hemorrhage, the morphology and structure of the islet cells were similar to normal pancreas. Immuno uorescence staining results (Fig. 5F) the insulin signal of diabetic mice was signi cantly decreased and glucagon signal was signi cantly increased. In the BHB treatment group, insulin secretion was signi cantly increased, glucagon secretion was decreased.

Effects of BHB on intestinal GLP-1 hormone
The number of GLP-1 positive cells in the colonic mucosa of diabetic mice was reduced. While the supplementation of BHB, the positive cells was restored to the level in control group ( Fig. 6A and B).

Effect of BHB on FFA and TG levels
The experimental results showed that the diabetic mice showed signi cant FFA metabolic disorder, and the serum FFA level was signi cantly increased, which was 2.73-fold that of the control group (Fig. 7A).
Serum TG level was also signi cantly increased, 2.41 times that of control mice (Fig. 7B). After BHB intervention, FFA disorder of diabetic mice were signi cantly improved, FFA level decreased to 187.56 µmol/L, signi cantly lower than the diabetic group. And TG level decreased to 1.56 mmol/L, also lower than the diabetic group.
Effect of BHB on GPR109a, GPR41 and GPR43 in colon In order to further explain the way BHB regulates FFA, uorescence immunoassay was used to detect the expression of GPR109a in colon. The results in Fig. 8A showed that the expression of GPR109a in the colon of the diabetic mice were signi cantly reduced. After BHB intervention, its levels were signi cantly restored, and the uorescence intensity was increased, and basically returned to a high level. GPR41 and GPR43 which highly expressed in colon tissue were also possibly receptors of BHB. Further testing the expression of other GPRs in the colon (Fig. 8B), it found that the diabetic group has a signi cant inhibiting effect on GPR41, GPR43 and GPR109a. After supplementation with BHB, it had a signi cant up-regulation effect on the expression of GPR109a, while the in uence of GPR41 and GPR43 was not obvious.

Discussion
Increased appetite and weight loss were the characteristics of T2DM [12]. After supplementation with L.plantarum YZX21, the weight of diabetic mice was restored, while food intake decreased. It was suggested that the glucose utilization was increased and glycogen and fat conversion was reduced. FBG and PBG were the criteria for diagnosing diabetes [13]. After supplementing L.plantarum YZX21, FBG and PBG were all relieved, and basically returned to normal levels. OGTT and its AUCglucose were the indicators of ability of pancreatic β-cells to regulate blood glucose [14]. The blood glucose of diabetic mice climbed rapidly and decreased slowly. The area of AUCglucose of diabetes was signi cantly higher. However supplementing L.plantarum YZX21, the peak blood glucose level decreased, and the blood glucose level dropped smoothly. This suggested that after supplementing with L.plantarum YZX21, the ability of pancreatic β-cells to regulate blood glucose was signi cantly enhanced.
Insulin is the only hypoglycemic hormone secreted by β cells. The main reason for the increase blood glucose in T2DM is insulin parasecretion [15]. The ITT experiment re ected the sensitivity of insulin in body [16]. After supplementing L.plantarum YZX21, the insulin sensitivity of the mice was obviously restored, and the area of AUCglucose was reduced. Fasting insulin level and HOME-IR are other key indicators of the body's sensitivity to insulin [17]. After supplementing L.plantarum YZX21, fasting insulin levels and HOME-IR were adjusted. These results suggested that L.plantarum YZX21 improved insulin sensitivity in T2DM mice. Histopathology also found the L.plantarum YZX21 supplement group reduced insulin resistance, protected the morphology of islet cells, restored insulin secretion, and inhibited insulin secretion. These results suggested that L.plantarum YZX21 reduced blood glucose levels in mice by improving insulin secretion. Improving insulin secretion and insulin sensitivity were the key of probiotics to play the role of hypoglycemia.
Through the uorescence colonization experiment, we found that L.plantarum YZX21 reached the colon and maintained with a high intensity. This suggested that L.plantarum YZX21 was likely to colonize in colon, which were a large number of L cells. GLP-1 is the intestinal hormone that protects the morphology of pancreatic islets, increase insulin secretion, and reduce foreign insulin resistance [18]. We also found that the number of GLP-1 positive cells and the GLP-1 content increased signi cantly in the colon. This suggested that L.plantarum YZX21 was likely to increase and regulate the content of GLP-1 by colonizing the colon.
Intestinal metabolites are the key to the connection between strains and intestinal cells [19]. The increase or decrease of speci c metabolites will affect the metabolism of the host [20,21]. Speci c metabolites are likely to be key components of hypoglycemic function. Through metabolomics techniques, we found that the FFA content in the intestine was signi cantly reduced in the L.plantarum YZX21 group. FFA has lipotoxic effects and is an important risk factor for diabetes [22]. As a kind of ketone body metabolite, BHB was closely related to metabolic diseases. In the L.plantarum YZX21 group, the BHB level increased signi cantly. The HMDB and correlation analysis suggested a close relationship between BHB and diabetes. However, the relationship between BHB, FFA and GLP-1 was not fully understood.
BHB was a common metabolite and the most important member of the ketone body family [23]. BHB was closely related to major diseases such as energy metabolism disorders in the organism and diabetes [24]. The relationship between BHB and GLP-1 had not been studied. In particular, whether probiotics could improve GLP-1 through BHB was still unclear. We constructed T2DM mice and gavage BHB to detect the mechanism and pathway of blood glucose regulation. After intragastric administration of BHB, the FBG and PBG of diabetic mice was decreased signi cantly and OGTT was improved. In the same way, insulin levels and HOME-IR in diabetic mice was also signi cantly improved. These results indicated that BHB intervention in type 2 diabetic mice reduced blood glucose by improving insulin resistance, which undoubtedly affects the process of T2DM. In addition, the results from pancreas histopathology and immuno uorescence analysis suggested that BHB contributes to increasing the insulin secretion, protecting islet cell morphology and inhibiting apoptosis. Hypoglycemic effects of BHB came down to increase the intestinal GLP-1 secretion. It was worth noting that BHB reduced TG and FFA levels, respectively. Metabolomics prompted that FFA was the greatest decrease in intestinal contents in diabetic mice. These results again suggested that BHB was likely to change intestinal GLP-1 content by in uencing FFA content.
Circulating FFA is believed to play a causal role in the pathogenesis of liver and peripheral insulin resistance [25,26] and is also a hallmark of T2DM [27]. Too much FFA aggravates insulin resistance, and is called lipotoxicity [28,29]. Studies found that high levels of FFA inhibit GLP-1 [30]. Inhibiting FFA levels can increase GLP-1 secretion, thereby improving T2DM [7,31]. Studies have shown that BHB signi cantly improved tissue damage, metabolic disorders and protein metabolism [32]. As an endogenous signaling molecule, BHB binds to an activate the intestinal hydroxyl carboxylic acid receptor (HCA2) [33,34], also known as GPR109a. BHB reduces the levels of FFAs and triglycerides in the circulation [35]. Our experiments found that after BHB intervention, serum FFA and TG levels were signi cantly reduced, while GLP-1 levels were signi cantly increased. The expression levels of its possible receptors GPR109a, GPR41 and GPR43 showed that GPR109a was mainly activated, but the effect on GPR41 and GPR43 was not obvious. Fluorescence histochemical experiments also found that BHB signi cantly increased the expression of GPR109a in the intestine. GPR109a in the intestine could inhibit the hydrolysis of TG into FFA and reduce its inhibitory effect on GLP-1 [36,37]. Thereby BHB restored the secretion of GLP-1 by reducing the inhibition of FFA through activating GPR109a receptor.

Conclusions
In summary, through the widely-targeted metabolomic analyses of intestinal contents, we found that L.plantarum YZX21 increased the content of BHB in the intestine. As the endogenous ligand of GPR109a, BHB inhibited fat degradation and reduce FFA levels, thereby increasing GLP-1 secretion and regulating diabetes. These results provide a new theoretical basis for the use of L.plantarum YZX21 to intervene in T2DM.

Experimental strains
The L.plantarum YZX21 was stored in the Functional Dairy and Probiotic Engineering Laboratory of Ocean University of China (preservation number CCTCC M 2020829). The strain was activated in de Man, Rogosa and Sharpe (MRS) broth at 37 ℃ for 48 h for twice, then inoculated in MRS broth at an inoculum size of 2% (v/v) and incubated at 37℃ for 24 h. Following that, the bacteria were centrifuged at 4 ℃ and 5000 ×g for 5 min. Sediments washed twice with sterile PBS buffer and resuspended in PBS to a nal concentration of 1×10 8 CFU/mL, stored at 4 ℃ before use.

Animal experiment
Experimental animals and model construction 6-week-old SPF C57BL/6J male mice (weighing 18-20g, n=80) were provided by Beijing Vital River Laboratory Animal Technology Co. Ltd. The mice were fed in a single cage with normal or high-fat feed. The speci c ingredients are shown in Table S1. The animal room had constant temperature and humidity (temperature 21-25°C, humidity 40%-60%, light from 8:00-20:00 for 12 h).
Before the experiment, the mice were adaptively fed in the animal room for 1 week. The control mice (n=8) was fed with normal diet and diabetic mice (n=24) was fed with high-fat diet for 4 weeks. After 4 weeks of dietary manipulation, the control mice were treated with 50 mmol/L citrate buffer (pH 4.5) by intraperitoneal injection. And diabetic mice were given intraperitoneal injection of Streptozotocin STZ at a dose of 50 mg/ (kg bw) [38]. After 7 days of STZ injection, the blood glucose of the mice was measured. The fasting blood glucose of the mice ≥7.0 mmol/L or the PBG level ≥11.1 mmol/L was considered as a successful model of diabetes [13]. β-Hydroxybutyrate supplementing experiment of mice by gavage Control group (C group): normal mice, gavage 0.25 ml 3% PBS solution; diabetic group (D group): diabetic mice, gavage 0.25 ml 3% PBS solution; Metformin group (M group): diabetic mice, gavage 0.25 ml of 10 mg/kg.bw metformin; β-Hydroxybutyrate group (BHB group): diabetic mice, gavage 40mg/kg.bw [39]. The mice were given intragastric administration once daily from weeks 1 to 12.

Blood glucose testing
The blood was collected from the tail using a rapid blood glucose meter (Sannuo, China). Fasting blood glucose (FBG) was measured in mice overnight fasting for 12 h. Postprandial 2 h blood glucose (PBG) was measured in mice after eating for 2 h.

Oral glucose tolerance test (OGTT)
The mice were fasted overnight for 12 h. The initial blood glucose was measured (0 min), and then a 2 g/kg glucose solution was given to the stomach. The blood glucose values were measured at 30, 60, 90 and 120 min.

Samples collection
At the end of the experiment, the mice were fasted for 12 h and injected 100 mg/kg bw of ketamine into the intraperitoneal cavity after anesthesia. Blood was taken from the eyeballs, and then dislocated and sacri ced. The blood was centrifuged for 10 min at 3000×g at 4°C, and then the supernatant liquid was drawn and frozen at -80°C. The collected pancreatic tissues were xed in 4% formaldehyde solution. The liver, intestinal tissue and intestinal contents were collect and freeze it at -80°C.

GLP-1 content detection
The colon contents were added the sample diluent and centrifuged for 10 min at 3000×g at 4°C. The supernatant liquid was drawn and the GLP-1 level was detected by ELISA kit (Nanjing Jiancheng, China), according to the instructions.
Determination of colonization of probiotic agents in intestinal tract of mice L.plantarum YZX21 was incubated with 1 mL FITC uorescent dye with a concentration of 1 mg/mL for 30 min at 37°C, washed with PBS until the liquid had no color, and diluted to 1×10 8 CFU/mL. The YZX21 group was gavaged the 0.5 mL of bacteria solution treated with dye. The mice were sacri ced at 1 h, 3 h, 6 h, and 12 h, and the entire digestive tract of the mice was taken and stored in liquid nitrogen. Intestinal tissue was photographed by IVIS Lumina XRMS Series III (PerkinElmer, Waltham, MA).

Widely-targeted metabolomics metabolic testing
The colon contents was thawed on ice. 50 mg of sample was taken and homogenized it with 500 ul of ice-cold methanol/water (70%, v/v). The sample was vortexed for 3 min, sonicated for 10 min in an ice water bath, and vortexed for 1 min. Then centrifuge it with 12,000×g at 4°C for 10 min.
The colon contents of the mice were analyzed using an LC-ESI-MS/MS system. The analytical conditions were as follows, UPLC: column, Waters ACQUITY UPLC HSS T3 C18 (1.8 µm, 2.1 mm*100 mm); column temperature, 40°C; ow rate, 0.4 mL/min; injection volume, 5μL [40]. LIT and triple quadrupole (QQQ) scans were acquired on a triple quadrupole-linear ion trap mass spectrometer (QTRAP), QTRAP® LC-MS/MS System, equipped with an ESI Turbo Ion-Spray interface, operating in positive and negative ion mode and controlled by Analyst 1.6.3 software (Sciex). Metabolome measurements were carried out through a facility service at Metware Biotechnology Co., Ltd.

Determination of FFA in the colon
The contents of the mouse colon were taken, and a commercial kit (Solarbio, China) was used to detect the levels of FFA according to the instructions.
Histological analysis HE staining: The mouse pancreas was xed in tissue xative for 24 h, and then stored in 70% ethanol solution. Part of the tissue was embedded in para n, and then stained with hematoxylin and eosin for histopathological analysis. The morphology was observed through a microscope (Olympus, Japan).
Immuno uorescence double labeling: The pancreas tissue was added insulin and glucagon primary antibody and incubated overnight at 4°C. DAPI was added to counter-stain the nucleus. The slides were mounted and sliced under a uorescent microscope (NIKON ECLIPSE TI-SR, Japan) to observe and collect images.
Immunohistochemistry: The primary antibody GLP-1 was added to incubate overnight, and the secondary antibody was added to incubate at room temperature. The hematoxylin stained nuclei were thick, dehydrated, transparent, and mounted. The sections were observed under an upright microscope (NIKON ECLIPSE TI-SR, Japan) and the number of positive cells was recorded.

Detection of related genes in diabetic mice
Take the colon tissue of mice added Trizol (Invertrogen, USA) lysate to extract total RNA from the tissues. CDNA was synthesized using ReverTra Ace®qPCR RT Master Mix Reverse Transcription Kit (Invertrogen, USA). The primer forward and reverse sequences are shown in Table S2 [41]. After adding SYBR Green (Invertrogen, USA), perform the reaction in a real-time PCR (Bio-Rad, USA). The relative levels of gene expression was detected by using the 2 -ΔΔ Ct method.

Statistical analysis
Data were presented as mean ± standard deviation. SPSS 21.0 and Prism 7.0 (GraphPad Software) were used for statistical analysis and gure drawing. One-way analysis of variance and Turkey's multiple comparison test were used to analyze the differences among the groups. p < 0.05 was considered statistically signi cant. Figure 1 Effects of L.plantarum YZX21 on body weight, food intake, and blood glucose in T2DM mice. (A) The body weights were measured for 12-week (n=8 mice/group). (B) The food intake were measured for 12week (n=8 mice/group). (C) Fasting glucose levels and (D) Postprandial blood glucose levels were determined in each group (n=8 mice/group). (E) The OGTT and (F) AUCglucose was calculated for each group (n=8 mice/group). Data are shown as mean ± SD. Different letters indicate signi cance between columns (p < 0.05). Representative images of pancreas with HE stain were shown (n=6 mice/group). (F) Pancreatic islets were identi ed by insulin immuno uorescence (green), and glucagon was determined by TUNEL (red); nuclei were visualized by DAPI staining and shown in blue (n=6 mice/group). ICCD: islet cell cavitating degeneration; IPI: irregular pancreas islet; VP: vascular proliferation; MAPI: Medium autolysis pancreas islet; SAPI: slight autolysis pancreas islet. Data are shown as mean ± SD. Different letters indicate signi cance between columns (p < 0.05).  Effects of L.plantarum YZX21 on intestinal metabolites in mice. OPLS scores plot (A) and corresponding S-plot (B) of the D and L.plantarum YZX2l groups (n=8 mice/group). (C) BHB was the highest negative metabolites association with T2DM (n=8 mice/group). (E) and (F) Comparison of relative peak areas of BHB and FFA in the different groups (n=8 mice/group). Data are shown as mean ± SD. Different letters indicate signi cance between columns (p < 0.05).

Figure 5
Effects of BHB on blood glucose and pancreatic islets in T2DM mice. (A)FBG levels and (B)PBG levels were determined in each group (n=8 mice/group). (C)insulin levels and (D)HOME-IR were determined in each group (n=8 mice/group). (E) Representative images of pancreas with HE stain were shown (n=6 mice/group). (F) Pancreatic islets were identi ed by insulin immuno uorescence (green), and glucagon was determined by TUNEL (red); nuclei were visualized by DAPI staining and shown in blue (n=6 mice/group). ICCD: islet cell cavitating degeneration; IPI: irregular pancreas islet; VP: vascular proliferation; MAPI: Medium autolysis pancreas islet; SAPI: slight autolysis pancreas islet. Data are shown as mean ± SE. Different letters indicate signi cance between columns (p < 0.05).