Four-week-old male C57BL/6J mice (n = 40) were bought from the Shanghai SLAC Laboratory Animal Company Limited (Shanghai, China). During the experiment, the mice were kept in a well-maintained and hygienic environment (at a 12 h light/12 h dark cycle, with 55 ± 5% humidity and 23 ± 1°C) with free access to water and food. The mice were randomly divided into two groups after one week of acclimation (NC group: Normal chow diet, n = 20; HFD-induced group: high-fat diet, n = 20). The HFD-induced obese mice received a pellet diet (Rodent Diet D12492; Research Diets, United States), where fat provided 60% of calories, carbohydrates provided 20%, and protein provided 20%. Following an 8-week induction (p < 0.01 relative to the NC group), the HFD-induced obese mice (n = 20) were randomly divided into two groups: treatment with crocin-I (20 mg/kg/day; HFD-C20), or vehicle alone (HFD group). Both groups were continuously fed with an HFD for the next 10 weeks.

Mice in the HFD-C20 group were supplied with 20 mg/kg of crocin-I daily for 10 weeks using oral gavage. The NC group and HFD group received the same amount of RO water as the crocin-I group. Body weight was recorded once a week, and mean food consumption was measured three times a week for 10 consecutive weeks starting at the 8th week.

Serum was obtained from blood (4°C, 5000 rpm, 10 min) and stored at −80°C. The colon, cecal contents and fecal samples, and liver, kidney, and epididymal white adipose tissues (WAT) were stored at −80°C.

This animal study protocol was reviewed and approved by the Zhejiang University of Technology Ethics of Animal Experiments committee (Permission no. SYXK (Zhejiang) 2017-0001). The experimental design for the study is shown in Supplementary Figure S1.

Crocin- I was isolated as previously described (Xie et al., 2019). The HFD-C20 group was dosed with crocin-I (20 mg/kg/day bodyweight; human equivalent dose estimated to be ∼150 mg/day for a 70 kg-adult) via gavage. The dose was selected based on our previous study, which showed crocin-I’s hepatoprotective effects in C57BL/6J mice (Xie et al., 2019).

The liver tissues from each mouse were isolated and homogenized with nine volumes of cold PBS. These supernatants were collected using centrifugation (12,000 rpm, 10 min). The protein concentration was measured using a commercial BCA protein kit (Beyotime Institute of Biotechnology, Shanghai, China). TG and TC levels were detected using Wako Labassay kits (LabAssay^TM ALP, Wako, Japan).

The accumulation and droplet distribution of lipids in hepatic cells were measured using hematoxylin and eosin (H&E), and Oil Red O staining (n = 4 per group), as previously described in Xie et al. (2019). Briefly, we soaked fresh liver tissue slices in 4% paraformaldehyde fixation solution and placed them at room temperature for 24 h. The tissues were snap-frozen and cut into 5 μm-sections on glass slides. Next, the slides were incubated with Oil Red O and H&E, respectively. After a series of processes, the tissues (4 μm section) were stained with H&E based on a standard method, while the colon tissue was treated with H&E and Alcian blue-periodic acid Schiff (AB-PAS) solution, respectively. Pathological changes were observed under a light microscope (Olympus, Tokyo, Japan) and were analyzed with Image-Pro Plus 6.0 (Media Cybernetics Inc., United States).

Before IPGTT, the mice were fasted overnight for approximately 16 h. The fasting blood glucose was tested, which was recorded as t = 0 min. The mice were then intraperitoneally injected with 10% D-glucose (2 g/kg body weight). Blood glucose levels were tested 15, 30, 60, 90, and 120 min after glucose injection by placing a small drop of blood on a new test strip and recording the measurements. The area under the blood glucose curve was calculated.

Fecal pellets were collected after the crocin- I or HFD treatments. Fecal water was prepared by homogenizing collected pellets with 1 mM 2-ethylbutyric acid in 15% phosphate acid for 8 min using a vortex mixer. The supernatants were obtained after 12,000 rpm centrifugation for 10 min at 4°C, they were then syringe filtered with 0.22 μm filters, and transferred to 250 μL inserts (Agilent Technologies, Palo Alto, CA). As soon as the fecal supernatants reached a final concentration of 1 mM, they were stored in 2 ml GC vials. SCFAs were detected with gas chromatography using a Trace 1310 gas chromatograph (Thermo Fisher Scientific, Waltham, MA), coupled with a flame ionization detector (Thermo). A data assay was performed with gas chromatography–mass spectrometry (GC-MS) solution software (4.45).

Total RNA was extracted from liver, epididymal adipose, and colon tissues using TRIzol reagents (Invitrogen Co., Carlsbad, CA, United States). RT-qPCR was performed using methods that have previously been described in Xie et al. (2021). Expression levels of selected genes were normalized to β-actin gene levels. The specific primers are listed in Supplementary Table S1. The levels of relative expression were detected using 2^−ΔΔCT.

Microbial genomic DNA (gDNA) was extracted from thawed cecal contents using a QIAamp DNA stool Mini Kit (QIAGEN, Carlsbad, CA), and was then quantified via UV spectroscopy. The V3-V4 region in 16S rRNA gene of the bacteria was PCR-amplified from microbial genome DNA using specific primers (forward primer: 5′-ACT​CCT​ACG​GGA​GGC​AGC​AG-3′; reverse primer: 5′GGACTACHVGGGTW-TCTAAT-3′). QIIME (vision 1.6.0) was performed to detect alterations in the gut microbiota.

All of the data are expressed as means ± SEM. One-way Analysis of variance (ANOVA) in GraphPad Prism (version 7.0). Tuckey’s multiple comparison tests were used when multiple groups were being compared. A p value <0.05 was considered statistically significant.

Between the 3rd and 9th weeks, mice that were fed with an HFD showed increased weights compared to control mice. However, mice that then had 7 weeks of crocin-I treatment showed decreased weights compared to mice fed with HFD alone (Figure 1A; Supplementary Figure S2). At the end of the experiment, the HFD group had a 42.5% increase in body weight compared to the NC group, while crocin-I treatment decreased the body weight of the HFD group by 15.4% (Figure 1A). Additionally, no significant alterations in average food intake (g/d/mouse) were detected between the HFD and HFD-C20 groups (Figure 1B), which indicates that the influence of crocin-I on body weight was not associated with changes in food consumption. Furthermore, although liver weights increased significantly in the HFD group compared with the control group, this trend was reversed with the crocin-I treatment (Figure 1C). However, the ratio of liver to body weight (Figure 1D) and the ratio of the kidney to body weight (Figure 1E) decreased significantly in the HFD group relative to the NC group (Figure 1D) because increases in body weight were largely determined by increases in epididymal fat (Figure 1F). We also found that kidney weight and epididymal fat were not affected by crocin-I treatment (Figures 1E, F). H&E staining increased adipose cell volume in the WAT of HFD-induced obese mice, but crocin-I administration did not reverse this trend (Figure 1G). We then analyzed the mRNA expression of lipid-related genes in the WAT. The results show that mRNA expression of sterol regulatory element binding protein 1c (Srebp1c) and fatty acid synthase (Fasn) were increased in the HFD group compared with the NC group, whereas, crocin-I treatment did not reverse their changes in mRNA levels (Figures 1H,I). Additionally, a high-fat diet did not significantly influence mRNA expression of carnitine palmitoyltransferase 1 (Cpt1) in the HFD group, but crocin-I significantly increased mRNA expression of Cpt1 (Figure 1J). Taken together, these results suggest that crocin can possibly relieve obesity by regulating the phenotypes of the liver. Furthermore, IPGTT indicated that, in contrast with the NC group, the HFD group had increased blood glucose concentrations, and the HFD group had a higher area under the curve (Figures 1K,L). Although crocin-I treatment significantly reduced blood glucose concentrations in the HFD group, it did not restore them to the levels seen in the controls.

To examine the effect of crocin-I on liver physiology in mice in the HFD group, we analyzed lipid accumulation in the liver. Oil Red O staining showed increased liver lipid accumulation in the HFD group, but this effect was reversed after crocin-I treatment (Figure 2A). Liver TG and TC were significantly increased in the HFD group and were almost restored to the levels seen in controls after crocin-I treatment (Figures 2B,C). As depicted in Figure 2D, there were a large number of lipid droplets in hepatocytes in the HFD group, but the number decreased after crocin-I treatment. We next clarified that crocin-I affected liver metabolism and evaluated the expression of lipid-related genes using RT-qPCR. The mRNA expression of acetyl-CoA carboxylase (ACC), Fasn, and Srebp1c increased in the HFD group compared with controls (Figure 2E). Meanwhile, mRNA expression of PPARα was decreased, although not significantly (Figure 2E). Crocin-I reversed the changes of all these genes in the HFD group (Figure 2E). Our discoveries suggest that crocin-I rescued lipid metabolic disorders in HFD mice by counteracting the expression of genes controlling the fatty acid synthesis.

H&E staining showed no obvious alterations to colonic mucosa villi, or epithelial, crypts or goblet cells in either the HFD group or the crocin-I treated group (Figure 3A). However, we wanted to examine the effects of crocin-I on the HFD-induced phenotype (i.e., intestinal inflammation) at a molecular level. The levels of inflammatory cytokines (including IL-1β, IL-6, and TNF-α) were measured in colon tissues. Compared with the NC group, the HFD group showed increased mRNA expressions of these inflammatory cytokines (Figure 3B), while crocin-I reduced mRNA expressions of IL-1β, IL-6, and TNF-α in colon tissues (Figure 3B). Furthermore, we tested the effect of crocin-I on the intestinal barrier by detecting transcription levels of genes related to tight junction proteins. The results showed decreased mRNA expression of Occludin and Zo-1 in the HFD group when compared with the controls (Figure 3C), but these mRNA levels were restored by crocin-I treatment. However, no differences were seen in Claudin-1 mRNA expression amongst the three groups (Figure 3C). Western blot results showed that Occludin significantly decreased in the HFD group, and crocin-I treatment restored it to normal levels (Figure 3D). AB-PAS staining of the colon indicated that the secretion of mucin in HFD group was significantly decreased relative to the NC group (Figure 3E). We also explored the mRNA expression levels of genes that were correlated with mucin secretion in colon tissue. The Muc1 and Muc2 mRNA levels in the HFD group decreased compared with the NC group (Figure 3C), while treatment with crocin-I increased Muc1 and Muc2 mRNA expression (Figure 3C). Muc2 protein levels were also significantly reduced in the HFD group (Figure 3D) and were not restored by crocin-I treatment. These results indicate that HFD may damage the integrity of the intestinal barrier and that treatment with crocin-I can repair it.

We used high-throughput sequencing of 16S rRNA to evaluate the effects of crocin-I on the gut microbiota compositions of HFD mice. According to the Shannon and Simpson index, microbiota diversity was lower in the HFD group compared with controls (Figures 4A,B), but this diversity was restored with crocin-I treatment. In addition, unifrac-based PCoA demonstrated that microbiota composition exhibited distinct clustering in all three groups (Figure 4C). A total of 4012 OTUs in the NC group, 2751 OTUs in the HFD group, and 3337 OUTs in the HFD-C20 group were displayed in a Venn diagram, with 1590 OTUs shared (Figure 4D). At the phylum level, compared with the NC group, cecum Bacteroidetes, Cyanobacteria, and TM7 abundances were significantly decreased in the HFD group, while Proteobacteria and Firmicutes abundances were significantly increased (p < 0.05, Figure 4F). Additionally, the abundances of Defferribacteres and Actinobacteria were increased, and Verrucomicrobia and Tenericutes levels were reduced in HFD mice, but not to a significant degree (Figure 4F). Strikingly, treatment with crocin-I reversed the changes in Bacteroidetes, Firmicutes, and Proteobacteria in HFD group mice (Figure 4F; Heat Map: Figure 4G). The F/B ratio was increased in HFD mice but decreased after crocin-I treatment (Figure 4E).

Next, metabolic pathways related to differential metabolites were tested using the KEGG database. High-fat feeding changed many metabolic pathways, including ether lipid metabolism, butanoate metabolism, glycolysis/gluconeogenesis, valine, leucine and isoleucine degradation, and glycerophospholipid metabolism (Figure 4H). Crocin-I administration changed the metabolic pathways related to carotenoid biosynthesis, ether lipid metabolism, bacterial chemotaxis, and the synthesis of bacterial motility proteins (Figure 4I).

As shown in Figure 5, the total SCFAs level decreased in the HFD group when compared with the control group but this trend was rescued after crocin-I treatment (Figure 5A). Isobutyrate levels were reduced in the feces of the HFD group but treatment with crocin-I increased fecal isobutyrate levels (Figure 5B). The levels of acetate (Figure 5C), butyrate (Figure 5D), isovalerate (Figure 5E), caproate (Figure 5F), propionate (Figure 5G), and valerate (Figure 5H) had similar trends. Taken together, these results suggest that crocin-I could restore the reduction of fecal SCFAs in HFD-induced obese mice.