Phlorizin ameliorates obesity-associated endotoxemia and insulin resistance in high-fat diet-fed mice by targeting the gut microbiota and intestinal barrier integrity

C57BL/6 J mice fed a NCD or a HFD were coadministered with or without PHZ (80 mg/kg) for 12 weeks. The results showed that PHZ coadministration effectively suppressed HFD-induced weight gain, resulting in a significantly lower body weight from the 3^rd week (P < .001; Figure 1a). However, no significant difference was found in the food intake (Figure 1b) or the energy intake (Figure 1c) between the NCD and NCD+PHZ groups, or between the HFD and HFD+PHZ groups (P > .05 in all cases), indicating that the anti-obesity effect resulted from PHZ coadministration was not due to reduction in food consumption or energy extraction. This was further confirmed by the decreases in fat accumulation (Figure 1d) and changes in the weights of different organs (Figure 1e). Morphological observations indicated that PHZ significantly inhibited HFD-induced fat accumulation in the liver (Figure 1f), which might effectively reduce further development of complicating diseases, such as fatty liver. On the other hand, PHZ significantly inhibited HFD-induced fat accumulation (Figure 1g) and fat cell expansion (Figure 1h,i, P < .001).

As a component of the Gram-negative bacterial cell wall,³² LPS is regarded as a triggering factor that promotes the onset and progression of systemic inflammation and related MDs by multiple pathways.²⁰ Moreover, HFD-induced LPS elevation results in endotoxemia.¹⁷ Similarly, this study found a significantly elevated serum LPS level in HFD-fed mice (P < .01), and PHZ supplementation significantly inhibited the increase in serum LPS (P < .05; Figure 1j). Long-term systemic inflammation has also been proven to trigger glucose tolerance and insulin resistance,³³ which are strongly associated with the onset and progression of obesity. As one of the most important gut hormones, glucagon-like peptide 1 (GLP-1) stimulates glucose-dependent insulin secretion and consequently improves metabolic syndromes, such as obesity and T2DM.³⁴ In the current study, the serum GLP-1 level was significantly reduced by HFD (P < .05 compared with the NCD group; P < .01 compared with the NCD+PHZ group); however, such effect was significantly inhibited by PHZ supplementation (P < .05; Figure 1k). As a result, the elevated levels of fasting blood insulin (Figure 1l), glucose (Figure 1m), and HOMA-IR (Figure 1n) induced by HFD were significantly prevented (P < .001 in all cases). On the other hand, the HOMA-IS level was significantly suppressed by HFD, but such effect was also prevented by PHZ supplementation (P < .001; Figure 1o). Besides, the mice in the HFD+PHZ group showed significantly lower area under the curve (AUC) values in the oral glucose tolerance test (OGTT; Figure 1p) and insulin tolerance test (ITT; Figure 1q) compared with those in the HFD group (P < .001 and P < .01, respectively).

Previous works have demonstrated a critical role of PHZ in preventing and treating MDs by directly decreasing the blood glucose level via competitive inhibition of SGLT2;^35,36 however, the extremely low bioavailability of PHZ casted doubt on whether this was the only mechanism behind the therapeutic effect.¹⁰ Thus, the current work hypothesized that the gut microbiota was an alternative target responsible for the observed beneficial effects of PHZ intake. This work applied a high-throughput sequencing technology to systematically analyze changes in the fecal microbiota composition after PHZ supplementation. Principal components analysis (PCA) was firstly conducted to visualize differences in the fecal microbiota structure among the four groups (Figure 2a; PC1 and PC2 was 41.5% and 10.1%, respectively). Distinct group-based clustering patterns were observed in the score plot. Symbols representing the HFD mice were all distributed to the right quadrants while symbols representing the majority of mice of the three other groups were located to the left quadrants, suggesting drastic differences in the fecal microbiota structure between the HFD mice and those of the other three groups. Although a distinct clustering pattern was shown among the HFD+PHZ mice at the lower left quadrants, they were clustered closely with the NCD and NCD+PHZ mice (mainly distributed to the upper left quadrants on the score plot), indicating these three groups shared higher similarity in their gut microbiota structure though obvious differences still existed. These results together suggested that HFD significantly altered the gut microbiota structure; however, coadministering PHZ effectively inhibited or mitigated the drastic shift in gut microbiota. Symbols representing the NCD and NCD+PHZ mice were mainly distributed to the left upper quadrants, appearing as two closely located clusters, suggesting that a clear difference still existed between the gut microbiota structure of these two groups. Such results further confirmed that supplementing PHZ could specifically impact the gut microbiota. The results of PCA largely agreed with that of the hierarchical clustering analysis (Figure 2b), except that the sample 2–1 was not clustering with any of the groups on the PCA score plot. Interestingly, based on the PCA results, apparent differences existed in the gut microbiota structure among groups and HFD did alter the gut microbiota structure drastically, yet no significant difference was detected in the α-diversity (represented by Chao1 index) across groups (Figure 2c, P > .05), indicating that the different interventions did not result in significant changes in taxa richness.

To further investigate the specific changes in the bacterial communities, the relative abundance of the predominant phyla and genera was compared across four groups, especially taxa responding to PHZ supplementation (Figure 2d-p). At the phylum level, Firmicutes and Bacteroidetes were the leading phyla across all groups, occupying over 80% of the total sequences (Figure 2d). The relative abundance of Firmicutes significantly increased by feeding HFD (P < .001), and such change was evidently reversed by PHZ supplementation (P < .001; Figure 2e). In contrast, the phylum Bacteroidetes showed an opposite trend (P < .001; Figure 2f). Interestingly, the Firmicutes/Bacteroidetes ratio, an indicator associated with obesity reported in some studies,³⁷ was significantly higher in the HFD mice compared with mice of the other three groups (P < .001; Figure 2g). Coadministering PHZ also showed desirable effects in reversing some of the HFD-induced alterations in the gut microbiota composition, including Clostridia and Bacteroidia at the class level (P < .001 in both cases; Figure 2h,i), Desulfovibrionales and Bacteroidales at the order level (P < .01 and P < .001, respectively; Figure 2j,k), Desulfovibrion and Muribaculaceae at the family level (P < .01 and P < .001, respectively; Figure 2l,m), and Bilophila and GCA-900066575 at the genus level (P < .001 in both cases; Figure 2o,p). Moreover, PHZ supplementation significantly reduced the relative abundance of two obesity-associating genera,^38,39Mucispirilu and Bilophila, which were enriched by HFD (Figure 2n). Modulation of the above-mentioned taxa might together contribute to the mitigating effect of PHZ on the overall gut microbiota alterations. The overall gut microbiota modulation by PHZ supplementation could reduce the fecal LPS levels, which were significantly stimulated by HFD (P < .001; Figure 2q). Detailed results in the gut microbiota changes are available in the Supplementary Material Figure S1-S4.

The contents of fecal SCFAs were detected by gas chromatography-mass spectrometry (Figure 2r). The HFD group had significantly less total SCFAs compared with the NCD, NCD+PHZ, and HFD+PHZ groups (P < .001 in all cases), suggesting that HFD inhibited SCFAs production, which was mitigated by PHZ supplementation. Specifically, the fecal levels of acetic acid, propionic acid, and butyric acid were significantly higher in the HFD+PHZ group compared with the HFD group (Table 1; P < .05). However, PHZ supplementation did not change the fecal isobutyric acid, isovaleric acid, and valeric acid levels significantly (Table 1).

The gut epithelial integrity is considered as the first line of defense of the gastrointestinal tract.²⁶ The feeding of HFD could impair the intestinal epithelial barrier by increasing the gut LPS content;^25,40 and our results consistently showed that HFD increased the fecal LPS level (Figure 2q). Histological analysis confirmed that HFD induced intestinal barrier damage. The hematoxylin and eosin (H&E) and alcian blue periodic acid-Schiff (AB-PAS) stainings of the gut tissues of representative mice of each group (Figure 3a,b) showed that the HFD group exhibited obvious decreases in the goblet cell quantity (shown by the arrow), mucus thickness (shown by the black short line), intestinal wall thickness (shown by the blue short line), and villus height (Figure 3a,b). The supplementation of PHZ attenuated HFD-induced damage in epithelial integrity, accompanied by an increased blood GLP-2 level (P < .05, Figure 3c); GLP-2 is a peptide hormone secreted by L cells that helps repair and maintains the gut barrier integrity.⁴¹ Quantitative histological analysis revealed significant increases in the villus height (P < .01; Figure 3d), intestinal wall thickness (P < .001; Figure 3e), number of goblet cells (deep purple stained goblet cells in Figure 3b; P < .001, Figure 3f), and mucus layer thickness (P < .001; Figure 3g) in the HFD+PHZ mice compared with the HFD mice, suggesting an overall protective effect of PHZ against HFD-induced damage of the gut epithelial barrier.

Correlation analysis was then performed to identify association between the gut microbiota/metabolites and epithelial integrity-related parameters (Table 2; Figure 3). The results of the analysis revealed that most of the key bacterial species correlated positively with the levels of serum and fecal LPS while correlated negatively with SCFAs, including acetic acid, propionic acid, butyric acid, and total SCFAs (Table 2). The relative abundance of a number of these bacterial taxa, e.g., Eubacterium oxidoreducens group, Christensenellaceae R-7 group, Rikenellaceae RC9 gut group, Ruminococcaceae (Figure S3c), Lachnospiraceae, Desulfovibrio, Helicobacter, Bilophila (Figure S4c), and Mucispirillum (Figure S4f), increased significantly in the HFD mice. In the fecal samples of the HFD+PHZ mice, the relative abundance of these taxa often decreased, which was most likely resulted from PHZ supplementation. Some other species (e.g., Phascolarctobacterium) did not correlate with LPS but correlated negatively with SCFAs. Conversely, several key species (including NS3a marine group, Tyzzerella 3, and ASF356) were negatively correlated with LPS but were positively correlated with SCFAs; and their relative abundance was significantly higher in the HFD+PHZ mice compared with the HFD mice.

Besides, the results of correlation analysis also revealed positive correlation between SCFAs levels (especially acetic acid, propionic acid, butyric acid, and isobutyric acid) and the indicators of gut barrier integrity (including villus height, intestinal wall thickness, mucus thickness, and goblet cell quantity), as well as the blood GLP-2 level (Figure 3h). However, the indicators of gut barrier integrity was correlated negatively with the LPS level (Figure 3h).

To confirm that the attenuation of HFD-induced adverse effects was a result of the action the gut microbiota but not residues of PHZ (and its metabolites like PHT) remained in the feces of the donor mice, targeted chemical analysis was performed by using HPLC to quantify the spectrum of PHZ-related compounds in the collected feces before FMT. The two peaks, corresponding to the PHZ and PHT standards, appeared on the HPLC chromatograph of the standards but not detected in the fecal samples of the four groups of donor mice, suggesting that no PHZ or PHT was present in those samples (Figure 4a). The feces of the four groups of donor mice were transplanted daily to a new batch of acclimatized mice, which were then maintained on HFD for eight weeks, and the changes in their body weight were monitored (Figure 4b). A number of obesity- and gut epithelial barrier integrity-related parameters were recorded at eighth week, including the weights of epididymis fat and the liver (Figure 4c), H & E staining of epididymis fat (Figure 4d), adipocyte size of epididymis fat (Figure 4e), liver morphology (Figure 4f), fasting blood glucose and insulin (Figure 4g,h), serum levels of GLP-2, diamine oxidase (DAO), and D-lactate (Figure 4i-k), H&E and AB-PAS stainings of intestinal tissue sections (Figure 4l,m).

Transplantation of feces from the HFD+PHZ mice significantly attenuated HFD-induced weight gain and blood glucose increase in the recipient mice, accompanied by improvement in insulin resistance and reduction in serum LPS, suggesting that PHZ-modulated microbiota could attenuate HFD-induced MDs and that the gut microbiota was a target for the amelioration effect of PHZ (Figure 4 and S6). Our conclusion was also supported by a significantly higher serum GLP-2 level in the mice transplanted with feces obtained from the HFD+PHZ group than those receiving feces donated by the HFD group (P < .05; Figure 4i), suggesting that PHZ-modulated microbiota could stimulate GLP-2 secretion. Moreover, the HFD-induced damage of gut epithelial integrity was also attenuated after transplanting the PHZ-modulated microbiota into HFD fed mice, as evidenced by the microscopic morphology of the intestinal tissues (Figure 4l,m), as well as the significant increases in the intestinal wall thickness (Figure S6g), villus height (P < .01) (Figure S6h), thickeness of mucus layer (P < .001; Figure S6i), and number of goblet cells (P < .01; Figure 4m and S6j).

The gut barrier permeability and function were evaluated by the serum levels of DAO and D-lactate. The enzyme, DAO, is usually located at the upper villi of the small intestinal mucosa. D-lactate is an intestinal bacterial metabolite produced in the gut, which is not metabolizable by mammals due to the lack of a corresponding rapid enzyme system. Therefore, the serum levels of DAO and D-lactate could reflect the gut epithelial permeability and integrity.⁴² As shown in Figure 4j,k, the serum levels of DAO were 261.63 ± 10.60 pg/mL and 280.25 ± 7.96 pg/mL in mice transplanted with feces obtained from the NCD mice and the HFD mice, respectively, suggesting that transplanting the feces of the HFD mice could increase the serum level of DAO in the recipient mice. A similar increase was observed in serum D-lactate, which were 16.37 ± 2.68 μg/mL and 19.44 ± 0.68 μg/mL in mice receiving the feces from the NCD mice and the HFD mice, respectively. Such results suggested a higher gut barrier permeability in recipients transplanted with feces of the HFD mice compared with those receiving feces obtained from the NCD mice. However, the recipients transplanted with feces obtained from the HFD+PHZ mice had lower levels of serum DAO and lactate (235.08 ± 8.71 pg/mL and 15.18 ± 1.89 μg/mL, respectively) compared with mice receiving feces of the HFD mice, suggesting that transplanting PHZ-modulated microbiota could attenuate HFD-induced damage of the gut barrier.

The PHZ (≥98%) was purchased from Changsha Zhongren Biotechnology Co., Ltd. (Changsha, China), and it was dissolved in 0.9% of sterile saline before use. The PHZ standard was bought from Chengdu Herbpurify Co., Ltd. (No. Y-029-121123; Chengdu, Sichuan, China). Both acetonitrile and methanol (HPLC grade) were obtained from Tedia Co., Inc. (Fairfield, Ohio, USA). A Milli-Q water purification system (Millipore, Bedford, Massachusetts, USA) was used to produce deionized water (18.25 MΩ). All other chemicals were at least of analytical grade.

The Animal Ethics Committee of Sichuan University (No. KY160049↗) approved all the animal experiments. Male C57BL/6 J mice aged six weeks were bought from Chengdu Dossy Experimental Animals Co., Ltd. (Sichuan, China) and were acclimatized for one week under a controlled environment (temperature range of 20–22°C; relative humidity between 40%-60%; under a 12 h light/dark cycle by switching on lights at 8:00 AM, lights off at 8:00 PM). Food and water were available ad libitum during acclimatization.

After the acclimatization, the mice were assigned to four groups (n = 10 per group) in a feeding experiment: (1) NCD group (fed NCD, 10% of calories derived from fat; Supplementary Material Table S1); (2) NCD+PHZ group (fed NCD and 80 mg/kg of PHZ intragastrically); (3) HFD group (fed HFD, 60% of calories derived from fat; Supplementary Material Table S1); and (4) HFD+PHZ group (fed HFD and 80 mg/kg of PHZ intragastrically). All interventions lasted 12 weeks, and the body weight and food intake of mice were measured each week. The feces of the four groups of mice (as donor mice) were collected every day during the first four weeks of the feeding experiment for FMT and for further analysis. The fecal samples were collected and stored at −80°C until use.

At week 12, OGTT and ITT were performed using 2.0 g/kg glucose and 0.75 U/kg insulin, respectively, according to the methods described by Zhou et al.⁵⁸ After fasting for 12 h, all mice were sacrificed. Serum samples were collected by centrifuging the collected blood at 1700 rpm for 10 min, and the supernatants were stored at −80°C. Tissues and organs, such as intestine, liver, and epididymis fat, were also stored at −80°C. Formalin solution (10%, buffered) was used to fix parts of the intestine and epididymis fat collected from the mice. The experimental design is presented in Figure 6.

Before FMT, all recipient mice (a total of 24 mice, n = 6 per group) were acclimatized for one week as described above. After the acclimatization, the mice were fed a HFD throughout the FMT experiment. For FMT, the fecal samples daily collected from each group of donor mice from the fourth week of the feeding experiment were homogenized and suspended in 0.9% of sterile saline (200 mg/2 mL), followed by centrifugation at 800 rpm for 3 min. Each recipient mouse received 100 μL of respective bacterial suspension intragastrically, i.e., NCD→HFD group, NCD+PHZ→HFD group, HFD→HFD group, and HFD+PHZ→HFD group, correspondingly. The FMT was performed daily from the fourth week (Figure 6). The body weight and the food intake of the mice were recorded weekly. The fecal samples of each group of recipient mice were separately collected daily and stored at −80°C for further analysis. At eighth week of the FMT experiment, OGTT and ITT were performed according to the methods described by Zhou et al,⁵⁸ and samples of blood, tissues, and organs were also collected as described above after sacrificing the mice (Figure 6).

The fecal LPS, serum LPS, GLP-1, GLP-2, DAO, D-lactic acid, and insulin were quantified by commercially available enzyme-linked immunosorbent assay (ELISA) kits (Yinggong, Inc., Shanghai, China) according to the manufacturer’s instructions. Before the measurement of fecal LPS by ELISA, the LPS was extracted according to methods described by Han et al.⁵⁹

The epididymis fat and ileum section (3 to 5 cm above the cecum) of each mouse were collected for histopathological analysis. The H&E and AB-PAS stainings were used to stain the tissue sections according the methods described by Zhang et al.⁶⁰ A trinocular compound microscope with digital camera (BA410 Digital, Motic Group Co., Ltd.) was used to measure the size of adipocyte cell, small intestinal villus height, intestinal wall thickness, goblet cells quantity, and intestinal mucus thickness.

Prior to FMT, the fecal samples collected from the donor mice were extracted with methanol (m:v, 1:4) by ultrasonic extraction (250 W, 25°C for 20 min). After centrifugation (3000 rpm for 10 min) and filtration, the supernatants were collected. An established HPLC method was used to quantify the contents of PHZ and its aglycone, PHT, in the collected supernatants.⁷

The methods for analyzing the gut microbiota diversity and taxonomic profiles of the four groups of donor mice were described in detail in the Supplementary Methods.

Quantification of fecal SCFAs (including acetic acid, propionic acid, butyric acid, isbutyric acid, valeric acid, and isovaleric acid) was conducted by previously described methods,⁶¹ with slight modifications. Briefly, the fecal samples were suspended in and extracted with 25% methanol (m:v, 1:2.5) at 4°C for 20 min. The extracted samples were centrifuged (3400 rpm, 10 min) and filtered (membranes of 0.22 μm pore size) before analysis by using a PerkinElmer Gas Chromatography-Mass Spectrometer (Shelton, Connecticut 0648 USA) with a polar DB-FFAP capillary column (30 m × 0.25 mm i.d., film thickness of 0.25 μm). For the chromatographic separation, helium with a purity of 99.999% was utilized as carrier gas at a constant flow rate of 1 mL/min; hydrogen with a purity of 99.999% was utilized as auxiliary gas with the flame ionization detection (FID) temperature of 250°C, the injection port temperature of 250°C, and the shunt ratio of 10:1. For the analysis, 1 μL of sample was detected by the following temperature gradient: temperature was set to 100°C initially, then increased to 180°C at 5°C per min and maintained for 2 minu, and increased again to 250°C at 20°C per min and maintained for more than 5 min. The contents of SCFAs were quantified by using an external standard method with a standard mixture of SCFAs (Supplementary Material Table S2 and Figure S6).

The SPSS software (version 20.0, Chicago, Illinois, USA), GraphPad Prism software (version 5.01, San Diego, California, USA), and Origin Pro 9 software (Origin Lab Corporation, Wellesley Hills, Massachusetts, USA) were used to perform statistical analysis. One-way analysis of variance (ANOVA) and Tukey test were used to determine significant differences. All data were expressed as mean ± standard deviation. A P value < .05 was considered statistically significant.