Herbal formula LLKL ameliorates hyperglycaemia, modulates the gut microbiota and regulates the gut‐liver axis in Zucker diabetic fatty rats

LLKL, a new traditional Chinese medicine formula containing Edgeworthia gardneri (Wall.) Meisn., Sibiraea angustata and Crocus sativus L. (saffron), was designed to ameliorate type 2 diabetes mellitus. Despite the therapeutic benefits of LLKL, its underlying mechanisms remain elusive. This study evaluated the LLKL anti‐diabetic efficacy and its effect on gut microbiota to elucidate its mechanism of action in Zucker diabetic fatty rats. We found that administration of different LLKL concentrations (4.68, 2.34 and 1.17 g/kg/d) improved several diabetic parameters after a 6‐week treatment. Moreover, LLKL modulated gut microbiota dysbiosis, increased the expression of occluding and maintained intestinal epithelial homeostasis, leading to a reduction in LPS, TNF‐α and IL‐6 levels. Hepatic transcriptomic analysis showed that the Toll‐like receptor signalling pathway was markedly enriched by LLKL treatment. RT‐qPCR results validated that LLKL treatment decreased the expressions of TLR4, MyD88 and CTSK. Furthermore, a gene set enrichment analysis indicated that LLKL enhanced the insulin signalling pathway and inhibited glycerolipid metabolism and fatty acid metabolism, which were verified by the liver biochemical analysis. These findings demonstrate that LLKL ameliorates hyperglycaemia, modulates the gut microbiota and regulates the gut‐liver axis, which might contribute to its anti‐diabetic effect.

characterized by hyperglycaemia, 5 hyperinsulinaemia 6 and impaired glucose and lipid metabolisms in the liver, 5 accompanied by chronic low-grade inflammation. 7 Accumulating evidence indicates that an altered gut microbiota composition and diversity contribute to the onset and progression of diabetes. 8 For example, a higher Firmicutes/Bacteroidetes ratio is always observed in both diabetic patients and animals, which further leads some substances, such as lipopolysaccharide (LPS), to impair the gut barrier function. [9][10][11] A lower expression of the gut barrier protein occludin was also observed in diabetic mice, indicating an increase in intestinal permeability. 12 LPS penetrates the intestinal epithelium into the blood circulation and liver via the portal circulation, leading to the activation of the mononuclear macrophage system and further regulates release of various cell "toxic factors", such as cytokines, inflammatory mediators, proteases and oxygen-free radicals. 13,14 The gut-liver axis communicates with organs of the digestive system through the biliary tract, portal vein and systemic crosstalk, which promotes the gut factors to regulate liver glucose and lipid metabolisms. 15 One major pathway by which the gut microbiota regulates the glycaemic control, inflammatory response and liver metabolism is the Toll-like signalling pathway. 16 In the liver, LPS binds to Toll-like receptors located on the membrane of hepatic cells and activates key signalling pathways, such as the myeloid differentiation primary response gene 88 (MyD88), 17 directly inducing the transcription of pro-inflammatory cytokines, including interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α), which then favours the insulin resistance. 18 Furthermore, a recent study also reported that cathepsin K (CTSK) could be up-regulated by an imbalance of microbiota and might function through Toll-like receptor 4 (TLR4) to up-regulate inflammatory factors. 19 The metabolic inflammation mediated insulin resistance through the inhibition of insulin signalling, which suppressed the hepatic glucose production or induced the production of ''second messengers'', such as fatty acids. It also stimulated hepatic lipogenesis, contributing to steatosis and elevated serum lipid levels. 20 Thus, the modulation of gut microbiota is a promising and feasible target for treating T2DM.
Traditional Chinese medicine (TCM) has been applied to the treatment of metabolic diseases such as diabetes and obesity for many years. 18  also called "lvluohua" in Tibet, China, has been widely used as a folk medicine to prevent and treat inflammation, cardiovascular disease and various metabolic diseases, including diabetes and hyperlipidaemia. 21 Sibiraea angustata, known as "liucha" and used as a common and civil traditional medicine in Tibet, China, contains many active components, such as terpenes, phenolic acid, saponins and polysaccharides, 22 with significant lipid-lowering and anti-obesity activity. 23,24 Moreover, saffron consumption combined with exercise reportedly improved diabetic parameters through redox-mediated mechanisms and stimulated the GLUT4/AMPK pathway to enhance glucose uptake. 25 In light of the above data, we hypothesized that the LLKL formula has more promising anti-diabetic benefits.
The present study evaluated the anti-diabetic therapeutic effectiveness and the underlying mechanisms of LLKL in Zucker diabetic fatty (ZDF) rats. We explored LLKL influence on insulin resistance and inflammatory status and assessed its capability to modulate the gut microbiota, hepatic glucose and lipid metabolisms, and the gut-liver axis activation via the Toll-like receptor signalling pathway.
Our findings present LLKL as a promising drug for the treatment of T2DM.

| Animals and experimental design
Male Zucker lean normoglycaemic rats (ZLN, +/fa, 13-14 weeks) and male Zucker diabetic fatty rats (ZDF, fa/fa, 13-14 week), which have already been administrated of purina#5008 diet for 4 weeks in order to induce T2DM status, were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd (licence number: SYXK (Jing) 2016 0011). Rats were housed at a temperature of 25 ± 2°C, relative humidity of 50 ± 5% and a 12/12-hour day-night cycle under specific pathogen-free conditions. All rats had free access to food and water. After 2 weeks of adaptive breeding, the fasting blood glucose (FBG) and bodyweight (BW) of each rat were measured. ZDF rats with FBG ≥ 7.8 mmol/L 26 were considered diabetic and used in the subsequent experiments. According to the FBG and BW values, diabetic rats were then randomly assigned to each of the following

| Measurement of FBG, OGTT and ITT levels
Blood glucose level was measured in the tail vein blood using a blood glucose meter (ARKRAY). FBG was tested after deprivation of food for 12 hours overnight. OGTT and ITT were performed after a 6-week treatment. Rats were fasted for 12 hours overnight and 4 hours before performing the OGTT and ITT, respectively. We

| Haematoxylin and eosin (H&E) and periodic acid-schiff (PAS) stainings
Fresh liver tissues and small intestine tissues were harvested from the rats, fixed in 4% paraformaldehyde (Solarbio) and embedded in paraffin. Prepared liver and small intestine slides were stained with haematoxylin and eosin (H&E) staining. Glycogen was measured using a PAS kit (Solarbio) following the manufacturer's instructions.
F I G U R E 1 Base peak chromatogram of LLKL in positive mode. Each peak number is consistent with

| Immunohistochemical (IHC) staining and analysis
For IHC staining, small intestine paraffin tissues were sectioned into 5-μm slices. Xylene was used for dewaxing and gradient ethanol was used for rehydration, respectively. The primary antibody of Occludin (ab216327, Abcam) was used. As described before, 27 IHC staining was performed and the intensity of staining and the proportion of positive cells were used to evaluate the immunostaining.

| Real-time qPCR
After the total RNA was extracted and its concentration was measured, cDNA was synthesized using the HiScript II Q RT SuperMix for qPCR (+gDNA wiper) (Vazyme). Real-time PCR was carried out by using the ChamQ SYBR qPCR Master Mix (Vazyme) and a LightCycler ® 480 II Real-time PCR Instrument (Roche). The relative mRNA levels were normalized to ACTB. The mRNA relative quantitation was calculated using the ∆∆Ct method. The primers used are listed in Table 1.

| Gut microbiota analysis
The fresh faecal samples were collected from the NC, MOD, LLKL_M and LLKL_H groups after the 6-week treatment and subsequently sent to Majorbio Biotech Co., Ltd. for 16S rDNA sequencing.
The primers 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V3-V4 variable regions of the 16S rDNA gene on a GeneAmp 9700 thermal cycler PCR system (Applied Biosystems). After the library was qualified, Illumina MiSeq PE300 platform was used for 16S rDNA sequencing. The data were analysed on the Majorbio Cloud Platform (www.major bio.com). The original 16S rDNA sequencing raw data were deposited into the NCBI database (accession number: PRJNA604387).

| Library preparation and illumina transcriptome sequencing
After extracting the total RNA from the liver tissues of the NC, MOD and LLKL_H groups, the RNA purity was determined using a NanoDrop 2000 spectrophotometer. The transcriptome strand library was prepared using the TruSeq™ stranded total RNA Kit from Illumina. After quantified by TBS380, the paired-end RNA-seq sequencing library was sequenced with the Illumina HiSeq xten (2×

bp read length) by Shanghai Majorbio Bio-Pharm Biotechnology
Co., Ltd. The raw data were deposited into the NCBI database (accession number: PRJNA601882).

| Correlational and functional annotation analysis
After quality control, R version 3.4.1 cor was used to analyse the correlation between each two samples. Principal components analysis (PCA) of gene expression was performed with the R3.4.1 psych version 1.7.8. We then identified the differentially expressed genes (DEGs) with FDR < 0.05 and |logFC| > 0.5 by using the R3.4.1 limma (version 3.32.5). 28 The DEGs heatmap was carried out using R3.4.1 pheatmap (version 1.0.8). 29 DEGs were subjected to the DAVID 6.8 online tool 30,31 for gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and the significant enrichment of DEGs was determined by a P-value < .05, which was a modified Fisher exact P-value.

| Gene set enrichment analysis (GSEA) and gene-pathway regulatory network analysis
GSEA is a computational method that determines whether a priori defined set of genes is statistically significant between two phenotypes. 32 We adopted one kind of reported methods to perform GSEA. 33,34 Briefly, GSEA software version 4.02 was used. We employed the liver transcriptome expression profile data of NC, MOD and LLKL_H groups and the "phenotype label" includes MyD88 and CTSK, in which expression was verified to be down-regulated by the LLKL_H treatment. GSEA generated an ordered list of all genes based on their association with MyD88 and CTSK expression, respectively. The Pearson correlation metric |r| > .25 was selected for the ranking genes. 1000 gene permutations were used to generate a null distribution for enrichment score, and then, each pathway will attain a normalization enrichment score (NES). The KEGG gene sets were used as the gene sets database. Gene sets enriched with |NES| > 1 were considered significant. The gene-pathway regulatory network was subsequently constructed by connecting the "phenotype label gene", "enriched pathway" and the related "co-expression genes" based on the results of GSEA analysis using the cytoscape software.

| Statistical analysis
Data were expressed as means ± SD with GraphPad Prism software (version 8.0). Two-factor repeated-measures ANOVA was performed for the FBG, BW, OGTT curve and ITT curve data analysis.
Other data were analysed by 1-way ANOVA. Statistical Package for Social Sciences (SPSS) software (version 23.0) was used for statistical analysis in this study. P < .05 was considered statistically significant.

| Identification of chemical composition of LLKL
Hclass-vion IMS QTof was used to analyse and identify the constituents of LLKL. As shown in Figure 2, we identified 18 phytochemicals, including 15 flavonoids and their glycosides and 3 organic acids. Table 2 shows the retention time, experimental mass and the identified compounds. used as a positive control and the diabetic group, respectively. As shown in Figure 2B and Figure S1 A, BW of the MOD group was increased compared with the NC group, but BW of rats between any of the treatment groups and MOD group were not significantly different. In addition, no significant differences in the food intake between the MOD group and TA B L E 2 The retention time, experimental mass and identified compounds any of the treatment groups at the end of week 6 were observed ( Figure 2C and Figure S1B). After LLKL_H, LLKL_M, LLKL_L, MET, Edgeworthia gardneri (Wall.) Meisn., Sibiraea angustata and Crocus sativus L. (saffron) administration, FBG levels were lower than those of the MOD group ( Figure 2D and Figure S1C). Moreover, LLKL_H were found to exert the highest hypoglycaemic effect and the LLKL_M had a more promising hypoglycaemic effect than the individual herbs.

| LLKL administration improves the glycaemic control and insulin resistance in ZDF Rats
The results of OGTT performed after the 6-week treatment showed that the blood glucose values peaked at 60 minutes after glucose administration and then decreased in all the groups ( Figure 2E and Figure S1D). Figure 2G and Figure

| Overall structural changes of the gut microbiota in response to LLKL treatment
To investigate the effects of LLKL on the gut microbiome, we performed sequencing of the V3-V4 region of 16S rDNA sequences present in the faeces collected at the end of week 6 from the NC, MOD, LLKL_H and LLKL_M treatment groups using Illumina MiSeq.
After removing low-quality sequences, a total of 903,965 highquality sequences and 379 OTUs were obtained from 24 samples at a 97% homology cut-off for subsequent analysis. The rarefaction curves indicated that although new rare phylotypes arose with additional sequencing, most of the diversity was already captured ( Figure S2A,B).
The α-diversity analysis of the intestinal contents showed that the Shannon value in the MOD group was significantly lower and Simpson value was markedly higher than those in the NC group, indicating that the gut microbiome diversity of the MOD group rats was lower than that in the NC group ( Figure 3A,B). As expected, compared with the MOD group, the Shannon and Simpson values increased and decreased, respectively, in both LLKL_H and LLKL_M groups ( Figure 3A,B), revealing that the α-diver-  Figure 3C).
To analyse the β-diversity of the gut microbiome, principal co-ordinates analysis (PCoA) and non-metric multi-dimensional scaling analysis (NMDS) based on the OTU profiles were conducted. As shown in Figure 3D,E, the gut microbiota structure of the LLKL_H and LLKL_M groups shared the same tendency and both were closer to that of the NC group compared with the MOD group, suggesting that the gut microbiota structure of LLKL_H and LLKL_M groups recovered to that of the NC group.
Next, we investigated the bacterial composition of each different group at taxonomic level. At the phylum level, as shown in Figure 3F,

| LLKL treatment alters the hepatic transcriptome and inhibits the Toll -like receptor signalling pathway
The liver is one of the most important organs for glucose and lipid metabolisms. Therefore, to further explore the hypoglycaemic mechanism of LLKL, a liver transcriptome analysis in the NC, MOD and LLKL_H groups was performed. We calculated Pearson's correlation coefficient to evaluate the association between samples and the result showed that the samples clustered closely with the treatments ( Figure 5A). Consistently, the PCA demonstrated a clear separation of the NC, MOD and LLKL_H groups ( Figure 5B), suggesting the high quality of our transcriptome data. Next, we identified 1495 (FDR < 0.05 and |logFC|> 0.5) and 740 DEGs, when comparing the MOD vs NC (Table S1) and the LLKL_H vs MOD groups (Table S2) Figure 5F,G). On the contrary, FOS was up-regulated in the MOD group as compared with the NC group, but there was no significant difference between the MOD group and the LLKL_H treatment group ( Figure 5H). As previous studies have demonstrated that TLR4 responses to LPS, we also measured the expression of TLR4 even though our transcriptome data showed that compared with the MOD group, TLR4 exhibited a relative reduction but not significantly in the LLKL_H group. As shown in Figure 5I, the expression of TLR4 was significantly increased in the MOD group  Table S3). Thus, these data evidently demonstrate that LLKL treatment promotes the insulin signalling pathway and inhibits glycerolipid metabolism and fatty acid metabolism.

| LLKL ameliorates the histological, glucose and lipid metabolic features in ZDF rat livers
To verify the histological, glucose and lipid metabolism characters in liver by biochemical experiments, the related parameters were analysed in livers of rats from the different treatment groups  Figure 7B). Moreover, compared with the NC group, the levels of serum FFA, liver TC and liver TG in the MOD group were notably elevated, and these effects were reversed by LLKL treatment ( Figure 7C-E). Therefore, these results reveal an effectively protective role of LLKL on hepatic morphology recovery and lipid metabolism in ZDF rats. Our data collectively led us to conclude that LLKL regulates gut microbiota and gut-liver axis which might contribute to the anti-diabetic effect of LLKL ( Figure 7F).

| D ISCUSS I ON
T2DM has grabbed increasing research attention worldwide to develop more effective therapeutic medicine, especially those from natural products. 35 In the present study, our results demonstrated that LLKL treatment improved insulin resistance in ZDF rats ( Figure 2) and the LLKL treatment showed a better effect than the individual herbs ( Figure S1). By performing a 16S rDNA sequencing analysis, we noted a disordered gut microbiota in diabetic rats, which was sufficiently improved after administration of LLKL, allowing the maintenance of homeostasis of the gut (Figure 3). 12.5 g/kg) were used to treat hyperglycaemia mice. 39 To investigate the effect and mechanism of Sibiraea angustata on lipid metabolism, F I G U R E 7 Effects of LLKL on hepatic glucose and lipid metabolism. A, Histopathological examination, Oil Red O and PAS staining of liver tissues. Representative images were presented. Scale bar represents 100 or 20 μm. B, Liver weight/BW. C, Serum levels of FFA. Levels of liver TC (D) and liver TG (E). F, Model depicts the improvement of diabetic features in ZDF rats after LLKL intake. This figure was created with BioRender.com. All data are presented as means ± SD (n = 6, *P < .05, **P < .01, ***P < .001 vs the MOD group) 1.05, 2.1 and 4.2 g/kg Sibiraea angustata were intragastrically administered to obese SD rats for 8 weeks. 40 Also, oral administration of Crocus sativus L. ethanolic extracts at doses up to 5 g/kg did not cause any mortalities or signs of toxicity in mice. 41 Taken together, our data indicate that LLKL effectively decreases the glycaemic levels and improves insulin resistance in ZDF rats.
Accumulating studies have revealed a correlation between diabetes and altered gut microbiota composition and diversity, suggesting a pivotal role of gut microbiota in the development of an anti-diabetes treatment strategy. 42 For example, probiotics have been identified as effective adjuvants to improve insulin resistance due to their gut microbiota regulation function. 43 55,56 In the present study, diabetic rats displayed abnormal morphological alterations, characterized by the loss of the normal villus structure in the small intestine epithelium, including disorganized, collapsed villi and decreased expression of occludin, which are consistent with previous studies. 57 As expected, LLKL treatment significantly reversed these changes. Accordingly, we also found that LLKL treatment alleviated the serum levels of LPS, IL-6 and TNF-α compared with those of the diabetic rats. These results indicate that the balance of the gut microbiota and the anti-inflammatory activity may be involved in the mechanism by which LLKL ameliorates diabetes in ZDF rats.
The Toll-like receptor signalling pathway plays an important role in the gut and liver crosstalk in metabolic disorders, such as NAFLD, diabetes and obesity. 58-60 TLR4 responds to ligands such as LPS, also referred to as endotoxin, and fatty acids, and initiates a response by forming a complex with myeloid differentiation factor 2 (MD-2), which leads to activation of both MyD88-dependent and non-MyD88-dependent signalling cascades. 61 In addition, MyD88 is considered to be a central hub of the inflammatory signalling cascades, 62  Moreover, according to our results, hepatic morphology, excessive hepatic lipid deposition and decreased glycogen were alleviated by LLKL supplementation. In addition, LLKL effectively inhibited serum FFA, liver TC and liver TG levels. Thus, LLKL may alleviate insulin resistance and regulate liver metabolism via the gut-liver crosstalk.
There are a few limitations to our experiments. Although we determined that LLKL inhibited MyD88 and CTSK expression to down-regulate the fatty acid metabolism and glycerolipid metabolism and up-regulated insulin signalling pathway through a series of genes, further efforts are necessary to study its precise molecular mechanisms. Other factors, including increased insulin sensitivity in muscle and adipocytes and protection of pancreatic β-cells, might also be involved in the underlying mechanism of anti-diabetes in LLKL. Further experiments are needed to comprehensively explore the underlying mechanisms of LLKL. In addition, to further verify the effect of LLKL on gut microbiota, we will perform faecal microbiota transplantation experiments in our future study. Additionally, we will further study the effect and metabolic processes of LLKL on the gut microbiota of T2DM patients. Overall, our data contribute to the understanding of the anti-diabetic effects and the underlying mechanisms of LLKL in ZDF rats.
In summary, our study indicates that LLKL elicited anti-diabetic properties in ZDF rats. In particular, its mechanism of action is mediated by modulating the gut microbiota dysbiosis and gut-liver axis activation via the Toll-like receptor signalling pathway. Taken together, these findings provide new evidence and insights into the anti-diabetic effects of LLKL.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.