Metf and Cana were purchased from Aladdin (Aladdin Technology Co., Shanghai, China) and Selleck Chemicals (Selleck Chemicals, Houston, TX, USA) separately. All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise noted.

The experimental protocol is shown in Supplementary Figure S1. Briefly, male C57BL/6J mice aged 6 weeks with weights of 20–22 g were obtained from the Charles River (Beijing, China). Mice were allowed to acclimatize in the laboratory environment for 1 week before the beginning of the experiment. All mice had free access to food and water. Then mice were randomly divided into four groups (n = 10 per group): normal-chow diet (NCD)-fed group (control group) (D12450B, Research Diets Inc., New Brunswick, NJ, USA), HFD-fed group (model group) (D12492, Research Diets Inc.), HFD plus Metf group (Metf group), and HFD plus Cana group (Cana group). Throughout the establishment of T2DM mice model period of 24 weeks, most individuals with HFD feeding appeared to be significantly hyperglycemic compared with the NCD feeding mice, while a few mice maintained a relatively normal blood glucose level. Thus, we randomly selected six mice with hyperglycemia for the following treatment trial. The sample size of each group was calculated by the G*Power software. In a one-way ANOVA with an effect size of f = 0.7, α of 0.05, power level of 80%, and group number of 4, it was calculated that a number of six mice per group would be sufficient to reach statistical significance. After that, the Metf and the Cana groups were gavaged with Metf (200 mg/kg dissolve in saline solution containing 0.025% Tween-20 and 0.5% carboxymethyl cellulose) or Cana (50 mg/kg dissolve in saline solution containing 0.025% Tween-20 and 0.5% carboxymethyl cellulose) once a day for six more weeks, while the control group and the model group were given equal amounts of solvents as a control vehicle. The body weight was weighed weekly. At the end of the experiment, mice were fasted for 12 h and were euthanized by pentobarbital sodium (100 mg/kg, ip) (Arjinajarn et al., 2016; Thongnak et al., 2017; Jaikumkao et al., 2018).

Blood samples were collected in 1.5-ml K₃EDTA-containing tubes (Sarstedt, Nümbrecht, Germany) via the eyeball method, then analyzed by an automated veterinary hematology analyzer (Mindray, Shenzhen, China).

Cardiac tissues were collected, then partial cardiac tissues were fixed in a 4% paraformaldehyde solution or 2.5% glutaraldehyde for 48 h, while frozen tissues were used for Oil Red O staining. Then tissues were dehydrated and embedded, and cut into sections. The tissue sections were analyzed by Oil Red O, H&E, or Masson’s trichrome staining (Goldner, 1938; Fischer et al., 2008; Wang et al., 2016).

FBG was measured after fasting for 12 h. Oral glucose tolerance test (OGTT) and intraperitoneal insulin tolerance test (IPITT) assays were conducted (Fraulob et al., 2010; Fei and Zhao, 2013). The increment of blood glucose after glucose loading was expressed in terms of area under the curve (AUC). Blood was collected and centrifuged at 375 × g for 10 min to obtain serum. Hemoglobin A1c (HbA1c) and insulin in serum were determined using HbA1c and insulin ELISA Kits (OM549127 and OM451187; Omnimabs, Alhambra, CA, USA), respectively. Homeostasis model assessment-insulin resistance (HOMA-IR) was calculated as fasting insulin × fasting glucose / 22.5. Serum triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) levels were detected by assay kits (F001-1-1, F002-1-1, A112-1-1, and A113-1-1; Jiancheng Bioengineering Institute, Nanjing, China). IL-6, monocyte chemotactic protein 1 (MCP-1), and TNF-alpha levels were determined using ELISA kits (M6000B, 479-JE-010/CF, DY410-05; R&D systems, Minneapolis, MN, USA). Soluble cluster of differentiation 40 ligand (sCD40L) and cardiac troponin I (cTn I) contents were determined using ELISA kit (OM448329 and OM453121; Omnimabs, Alhambra, CA, USA). Cardiac tissues (200 mg) were homogenized in cold phosphate buffer, then centrifuged to collect the supernatant. Glutathione (GSH), superoxide dismutase (SOD), malondialdehyde (MDA), and catalase (CAT) contents in cardiac tissue were determined (A006-2-1, A001-3-2, A003-4-1, and A007-1-1; Nanjing Jiancheng Bioengineering Institute, Nanjing, China). In addition, ROS content in cardiac tissue was detected by fluorescent dye dihydroethidium (DHE) as previously described (Zheng et al., 2018).

Cluster of differentiation 31 (CD31) (AF6191, Affinity Biosciences, Cincinnati, OH, USA) and heme oxygenase 1 (HO-1) (86806S, Cell Signaling Technology, Danvers, MA, USA) in cardiac tissues were determined as previously described (El-Azab et al., 2012). After blocking with goat serum, sections were incubated with primary antibody at 4°C for 12 h and then incubated at room temperature for 1 h with fluorescent secondary antibody. Then the sections were stained with DAPI for 30 min, and images were acquired.

TdT-mediated fluorescein nucleotide (cyanine 3-dUTP) nick-end labeling (TUNEL) was conducted using the TUNEL Assay Kit (ab66110, Abcam, Cambridge, UK) to evaluate the apoptosis of cardiac tissues. the incorporated fluorescein-labeled dUTP was analyzed.

Transmission electron microscopy (TEM) of cardiac tissues was conducted as previously described (Das et al., 2017). Cardiac tissues were fixed in 2.5% glutaraldehyde for 48 h at room temperature. Tissues were embedded in araldite, then cardiac microvasculature and mitochondria were observed by an electron microscope. The vascular basement membrane thickness of each sample was determined by the average value of six measurements at various position.

Total bacterial DNA was isolated from colonic contents by the QIAamp DNA Stool Mini Kit (51604, Qiagen, Hilden, Germany). The colonic bacterial 16S rDNA gene was amplified by PCR, then the PCR products were purified by the QIAquick Gel Extraction Kit (28706) and sequenced by Realbio (Realbio Technology Co., Ltd., Shanghai, China). Linear discriminate analysis effect size (LEfSe) was used to characterize the significant differences in the abundance of microbial taxa between different treated groups. Heatmap was analyzed by R Cytoscape. In addition, β-diversity was analyzed by the QIIME software.

Western blot analysis was conducted as in our previous study (Wang et al., 2021). In addition, the membranes were incubated with primary antibodies directed against pAMPK (2,531), AMPKα (2,532), and β-actin (4,970) from Cell Signaling Technology (Danvers, MA, USA). The ImageJ software was used for densitometry analysis.

Data are presented as the mean ± standard error of the mean (SEM). The differences between groups were analyzed by GraphPad Prism software using Student’s t-test, one-way, or two-way ANOVA as appropriate (San Diego, CA, USA). Differences were statistically significant when p < 0.05. In addition, microbial community β- and α-diversity were calculated using Chao1, Shannon, and Simpson indices, and Bray–Curtis distances were visualized by PCoA. The LEfSe characterized a value of p < 0.05 and an LDA score > 2.

First, we demonstrated that Cana significantly improved blood glucose and insulin homeostasis in mice with T2DM (Supplementary Figure S2). Moreover, Metf treatment remarkedly decreased body weight, body weight gain, and BMI after fed with HFD (Supplementary Figure S3) (p < 0.05), while Cana treatment could not. In addition, evidence demonstrates that increased concentrations of cholesterol and TG are causal risk factors for CVD (Nordestgaard and Varbo, 2014). Hyperlipidemia is a vital factor for T2DM-related CVD (Carr and Brunzell, 2004). Thus, managing hyperlipidemia is an effective way to prevent CVD. Our results showed that HFD feeding markedly elevated the levels of TC, TG, and LDL-C in serum (p < 0.05) (Figures 1A–C) compared with the control group. However, serous TG, TC, and LDL-C levels in the Cana group were markedly reduced (p < 0.05) (Figures 1A–C). In addition, serum HDL-C level increased in the Cana and Metf groups compared with the Model group (Figure 1D) (p < 0.05). Atherogenic index of plasma (AIP) and atherogenic index (AI) values increased with the elevated CVD risk, which are also highly sensitive markers of lipoprotein profiles in CVD patients (Dobiásová, 2006). Cana decreased the AIP and AI levels in T2DM mice (Figures 1E, F), which demonstrated that Cana relieved CVD-induced hyperlipidemia in mice with T2DM. In addition, Oil Red O staining indicated that abnormal lipid accumulation was observed in the model group, while treatment with Cana alleviated the lipid accumulation in cardiac tissues (Figure 1G). In addition, we demonstrated the activation of AMPK in cardiac tissues after Cana treatment (Supplementary Figure S4), which may contribute to decrease the lipid levels. Taken together, the above results demonstrated that Cana can perform an antiCVD effect via suppressing lipid accumulation in serum and cardiac tissue in T2DM mice.

Oxidative stress was increased in T2DM, and this appeared to underlie the development of T2DM and diabetic complications such as CVD (Tangvarasittichai, 2015). Oxidative stress is known as a major contributor to endothelial dysfunction (Magenta et al., 2013), which is a key precursor to the development of CVD (Rocha et al., 2010). The area and intensity of fluorescence from DHE oxidation were notably increased in the model group (p < 0.05) (Figure 2A), while Cana or Metf treatments markedly reversed this situation (p < 0.05). Notable decreases in the activities of SOD and GSH, while increasing MDA content were observed in the hearts of T2DM mice compared with the normal mice (Figures 2B–D). However, the above situation was conspicuously reversed by Cana supplementation, implying that Cana can effectively relieve myocardial oxidative stress (Figures 2B–D). Importantly, TEM revealed marked muscular fiber twisting, cardiomyocyte dissolution, intercalated disc blurred, and Z line disappearance. Meanwhile, some myofilaments were flocky in appearance and showed no cross striations in the myocardial tissues of diabetic mice. In addition, mitochondrial crista deformation and damage, mitochondrial autophagy (red arrows presented in Figure 2E, vacuoles with black residue), and fat drop (yellow arrows presented in Figure 2E, vacuoles without black spots) were remarkedly presented in the model group, the effects of which were greatly attenuated by Cana or Metf treatment (Figure 2E), which are beneficial for the suppression of mitochondrial ROS production and subsequently oxidative stress (Zhou et al., 2018).

Chronic inflammation is a mediator of vascular dysfunction in T2DM patients (de Jager et al., 2006; Du et al., 2015; Zhang et al., 2017). Through inflammatory processes, the initial lesion of atherosclerosis and CVD is formed (Libby, 2006). We found that HFD feeding elevated inflammation state by increasing serum levels of TNF-alpha, MCP-1, and IL-6 compared with the control group, whereas Cana administration markedly suppressed this increase in proinflammatory cytokines (Figures 2F–H). Besides, HO-1 level was remarkedly decreased in the model group, while Cana or Metf treatment effectively elevated the HO-1 level in T2DM mice (Figure 2I). Upregulation of HO-1 expression plays an important role in the protective response against oxidative injury and inflammatory effects (Lee and Chau, 2002), which proposes that HO-1 is a promising target protein for therapeutic intervention in CVD. In general, the above results indicate that Cana can effectively alleviate oxidative stress and systematic inflammation by protecting myocardial structural and mitochondrial homeostasis.

Studies demonstrate the strong association between hematological parameters and risk of CVD during systemic inflammation, and hypoxemia is implicated in the pathophysiology mechanisms of CVD (Fan et al., 2018). Our results (Table 1) showed that the white blood cell (WBC) count, platelets, and mean platelet volume (MPV) levels were conspicuously increased in T2DM mice compared with the normal mice, while Cana or Metf treatment effectively reversed this adverse situation. There is a significant association of increased MPV and platelet counts with diabetes related to endothelial dysfunction, coronary artery disease, and its vascular complications (Jabeen et al., 2013; Lippi et al., 2015). Moreover, elevated WBC is a classical inflammatory marker and associated with several CVD risk factors (Yayla et al., 2017). The above results indicate that Cana can conduct an anti-CVD effect via regulating the disorderly hematological parameters.

Normal vasculature is crucial to cardiac function and CVD, which is susceptible to hyperglycemia (Pinter et al., 1999). CD31 plays an important role in endothelial protection, which alleviates the apoptosis of vascular endothelial cells (Evans et al., 2001). HFD-induced T2DM triggered an overt drop in the number of CD31-positive microvessels compared with the normal mice (Figure 3A), while the unfavorable state was reversed by Cana or Metf treatment so as to play a good role in protecting the vascular endothelium. In addition, T2DM-induced basement membrane thickening and microvascular morphology including fibrosis (Figures 3B, C) were largely ameliorated by Cana administration. These data demonstrated a cardioprotective property of Cana by maintaining cardiovascular structural homeostasis.

Damage to microvessel integrity and cardiac microvascular endothelial cell are considered to be the initial step in vascular complications (Thygesen et al., 2018). Diabetes-induced effects (such as turbulent blood flow and capillary blockage) raised the accumulation of erythrocytes in the microvessel (Zhou et al., 2018), while Cana considerably retarded this negative situation (white arrows presented in Figure 4A). Additionally, Metf or Cana treatment inhibited the diabetes-induced TUNEL-positive cells from increasing (white arrows presented in Figure 4B), supporting the prosurvival capacity of Cana on hyperglycemia-mediated cardiac microvascular endothelial cell apoptosis. The above results indicated the beneficial effects of Cana on microvessel integrity, favoring a marked reduction in the risk of cardiac microvascular endothelial cell dysfunction or death in diabetic hearts.

Cardiac troponin is used to diagnose myocardial infarction, and the increased concentration of cardiac troponin indicates the elevated risk for adverse outcomes in individuals (Eggers and Lindahl, 2017). In addition, cTn I, an important subunit of cardiac troponin, is recognized as a marker of myocardial damage (Eggers and Lindahl, 2017). Studies also demonstrated that high levels of sCD40L indicate enhanced inflammatory responses, heightened risk of death, and myocardial infarction in patients (Schönbeck et al., 2001; Heeschen et al., 2003; Cipollone et al., 2003). In our study, levels of serous cTn I and sCD40L in T2DM mice were markedly elevated compared with the control group (Figures 4C, D), while Cana treatment markedly improved the situation. The above results demonstrate the salient effects of Cana on alleviating myocardial injury in T2DM mice.

As shown in Supplementary Table S1, α-diversity (e.g., Good’s coverage, Chao1, Shannon, and Simpson indices) of the colonic flora in the Cana-treated group was not markedly different compared with the control and model groups. In addition, evaluation of β-diversity based on nMDS ordination plot (Supplementary Figure S5A), weighted UniFrac heatmap (Supplementary Figure S5B), and weighted PCoA analysis (Figure 5A) showed that there were significant differences in colonic flora between the Cana treatment group and model group, and microbial community of the Cana group was similar to the control group. At the phylum level (Figure 5B), the Cana group had a higher abundance of Bacteroidetes with a lower abundance of Firmicutes, and an increased ratio of Firmicutes/Bacteroidetes (Figure 5C) compared with that of the model group. A significant elevation in the relative abundance of Proteobacteria was induced in the model group compared with control group and was remarkedly decreased in the Cana group, however, not reversed by Metf treatment (Figure 5D). In addition, administration of Cana remarkably enriched the abundance of Actinobacteria and decreased the abundance of Deferribacteres in mice with T2DM (Figures 5E, F). The abundance analysis at the genus level indicated that there were conspicuous differences in flora types and abundance in T2DM mice after Cana treatment (Figure 5G). Furthermore, Cana regulated the disorder of microbial community in the genus level as follows: 1. increasing the relative abundance of Olsenella, Alistipes, and Alloprevotella (Figures 5I–K), which may be associated with a healthier microbiome, as suggested previously (Kong et al., 2019); 2. decreasing the relative abundance of Helicobacter and Mucispirillum (Figures 5H, L), which was positively correlated with TNF-alpha and LPS contents (Li et al., 2019). Altogether, the above results demonstrated that intestinal ecosystem in mice with diabetic CVD were adjusted to a relatively normal state by oral administration of Cana.

To further elucidate the effects of Cana treatment on the regulation of various bacterial taxa in the colon, a pairwise comparison between the model and Cana groups was conducted by LEfSe analysis. It indicated that Cana reduced the level of Firmicutes, including Sporacetigenium, Veillonella, and Clostridium XI, while remarkedly elevating Bacteroidetes, including short-chain fatty acid (SCFA) producers such as Alloprevotella and Bacteroides (p < 0.05) (Figures 6A,B). The Cana treatment increased the amount of Porphyromonadaceae, Rikenellaceae, and Corynebacteriaceae, while a decrease in Desulfovibrionaceae and Veillonellaceae. At the genus level, the Cana group was marked by a dramatic increase in Alistipes, Roseburia, and Corynebacterium (Figure 6B), which can increase the production of short-chain fatty acids (SCFAs) (Franzosa et al., 2018; Amabebe and Anumba, 2020). Meanwhile, Paraprevotella, Veillonella, Sporacetigenium, and Intestinimonas were significantly reduced after Cana treatment compared with the model group (Figure 6B). In general, the taxonomical distribution within groups at phylum, family, and genus levels for the colonic samples revealed the divergent composition of communities after Cana treatment in mice with diabetic CVD.

Using Spearman’s correlation analysis, relationships of physiological index and bacterial abundance were clarified (Figures 6C–E). At the genus level, Parabacteroides and Alistipes were positively correlated with SOD and CAT, while negatively correlated with IL-1β, MDA, ROS, TNF-alpha, MCP-1, and LBP levels, and their abundances were increased after Cana treatment. In addition, Clostridium XI, Veillonella, and Paraprevotella were negatively correlated with CAT and SOD, while positively related to MDA, ROS, and the above inflammatory cytokines levels. Meanwhile, Alistipes and Roseburia were negatively correlated with LDL-C, TC, and TG contents, while positively related to HDL-C level. Cana treatment caused an increase in the relative abundances of Alistipes and Roseburia in mice with diabetic CVD. Moreover, Paraprevotella, Veillonella, Clostridium XI, and Barnesiella presented a positive correlation with LDL-C, TC, and TG contents, while negatively related to HDL-C level. These findings indicate that alterations in the gut microbiota after Cana treatment lead to the reduction of systemic inflammation, oxidative stress, and lipid accumulation in mice with diabetic CVD. Collectively, these data reinforce the link between gut dysbiosis and diabetic CVD, and indicate that Cana can manipulate gut microbiota and attenuate the CVD complications of T2DM.