Twenty-four female Sprague-Dawley (SD) rats (180–250 g) were obtained from the Center of Laboratory Animal Science, Hubei, China. The rats were group-housed in a room with a regular 12-h light/dark cycle, and had free access to food and water. All animals were treated in accordance with the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the study protocol was approved by the Animal Ethics Committee of Hubei University of Arts and Science.

The PCOS model was developed in female rats according to the method described by Poretsky et al (33). Female rats (70 to 85 days old) were subcutaneously injected with increasing doses of insulin (0.5 IU, 5.0 IU and 5.0 IU/day) for the first 10 days, and then with 5 IU insulin + 1.5 IU HCG from day 11 through day 22. The rats were fed a high fat diet for 6 weeks, and their weights were measured weekly. Rats in the control groups were fed a normal diet every day. Finally, the PCOS model rats displayed typical characteristics of PCOS, such as hyperinsulinemia, insulin resistance, weight gain, and polycystic degeneration in the ovaries (33). Fasting blood glucose and fasting insulin levels were assessed and the homeostatic model assessment-insulin resistance (HOMA-IR) of each rat was calculated. Next, the rats were assigned to four different groups: a non-PCOS/non-IR group, a non-PCOS/IR group, a PCOS/non-IR group, and a PCOS/IR group. The rats were then anesthetized with 10% chloral hydrate, and laparotomies were performed to collect ovarian tissue specimens for further processing. Insulin resistance was assessed by the HOMA-IR index, and rats with a HOMA-IR ≥2.5 were considered insulin resistant (IR).

The ovaries of each rat were removed, processed, and sectioned at 4-mm thickness; after which, the sections were stained with H&E and examined for evidence of morphological changes. Five sections were obtained from each ovary. Immunohistochemistry (IHC) was then used to examine the ovarian tissues for SREBF1, HMGA2 and GLUT4 expression. Tissue sections used for immunohistochemistry were first deparaffinized and hydrated, and endogenous peroxidase activity was quenched with 0.3% hydrogen peroxide in PBS. Next, the tissue sections were incubated in blocking buffer for 60 min, and then incubated overnight at 4°C with primary antibodies that included GLUT4 antibody (1:500; rabbit polyclonal; cat. no. PB0143; Boster Biological Technology, Wuhan, China), SREBF1 antibody (1:500; rabbit polyclonal; cat. no. 14088–1-AP; Proteintech; Wuhan Sanying Biotechnology, Wuhan, China), and HMGA2 antibody (1:500; rabbit polyclonal; cat. no. bs-0556R; Bioss Antibodies, Beijing, China). They were then incubated with horseradish peroxidase-conjugated secondary antibodies [1:5,000; goat anti-rabbit IgG H&L (HRP); cat. no. ab205718; Abcam, Cambridge, MA, USA] as per the manufacturer's instructions. Positively-labeled cells were visualized using a streptavidin-HRP conjugate and DAB chromogen. The tissue sections were then counterstained with hematoxylin, dehydrated, and mounted in DPX.

Human adipocyte cells (ScienCell Research Laboratories, Inc., San Diego, CA, USA) were incubated with adipocyte maintenance medium at 37°C in a humidified incubator containing 5–10% CO₂. The adipocyte cells were then cultured in normal adipocyte maintenance medium (control group) or treated with different concentrations of high-glucose adipocyte maintenance medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) (0, 3.15, 9.45 or 28.35 g/l) for 24 h. After culture, the adipocytes were collected and their levels of miR-33b-5p, SREBF1, HMGA2 and GLUT4 were examined using real-time polymerase chain reaction (PCR) or and western blotting methods.

Human adipocytes were incubated with adipocyte maintenance medium at 37°C in a humidified incubator containing 5–10% CO₂. The adipocyte cells were then treated with different doses of insulin (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) (0, 5, 50 or 500 nM) for 24 h; after which, they were collected and tested for their levels of miR-33b-5p, SREBF1, HMGA2 and GLUT4 expression using real-time PCR or western blotting methods.

Human adipocytes were cultured as aforementioned; after which, they were seeded into 6-well plates and grown to 60–80% confluence for use in transfection studies. A 9-µl volume of Lipofectamine 2000 was mixed with 150 µl of Opti-MEM medium (both from Thermo Fisher Scientific, Inc.). A miR-33b-5p inhibitor or NC inhibitor (Shanghai JiKai Gene Chemical Technology Co., Ltd., Shanghai, China) was diluted with Opti-MEM medium and then mixed with Lipofectamine-Opti-MEM medium at a ratio of 1:1 to achieve miR-33b-5p inhibitor concentrations of 0, 25, 50 and 100 nM, respectively. The sequence of miR-33b-5p inhibitor: 5′-GCAAUGCAACAGCAAUGCAC-3′. After allowing the cells to incubate at room temperature for 5 min, the miR-33b-5p inhibitor was added, and the cells were incubated with the different concentrations of miR-inhibitor complex for 48 h. Next, the adipocytes were collected, and their levels of miR-33b-5p, SREBF1, HMGA2 and GLUT4 were determined using real-time PCR or western blotting methods.

A miRACLE isolation kit (Jinfiniti Biotech, LLC, Augusta, GA, USA) was used to extract total RNA from the ovarian tissues of rats and cultured adipocytes. The various cDNAs of mRNAs or miRNAs were synthesized using a cDNA reverse transcription kit. Real-time PCR was carried out using SYBR-Green Supermix and an Applied Biosystems 7300 Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). The primers used for PCR are listed in Table I. U6 and GAPDH were used as internal control genes. Real-time PCR was performed on target genes under the following conditions: 35 cycles each containing 94°C for 30 sec, 58°C for 45 sec and 72°C for 35 sec. The relative levels of mRNA and miR-33b-5p were calculated using the 2^−ΔΔCT method.

Proteins were extracted from cultured adipocytes, and then separated using 10% SDS-PAGE. The separated protein bands were transferred onto a PVDF membrane, which was then probed with rabbit monoclonal antibodies against SREBF1 (1:500; rabbit polyclonal; cat. no. 14088-1-AP; Proteintech); HMGA2 (1:1,000; rabbit polyclonal; cat. no. bs-0556R; Bioss Antibodies); GLUT4 (1:1,000; rabbit polyclonal; cat. no. PB0143; Boster Biological Technology) or GAPDH (1:2,000; mouse monoclonal; cat. no. ab8245; Abcam) as an internal control overnight at 4°C. After incubation at 37°C for 1 h with horseradish peroxidase-conjugated secondary antibodies (1:5,000; goat polyclonal to HRP; cat. no. ab181658; Abcam), an Immobilon™ Western Chemiluminescent ECL kit was used to detect bound antibodies. Protein staining intensity was assessed using Image-Pro Plus 6.0 software.

Wild-type and mutant type HMGA2 3-UTRs were amplified and inserted into psiCHECK2 luciferase vectors (Promega Corporation, Madison, WI, USA). Next, the vectors with either the miR-33b-5p mimics or negative mimics (GenePharma Co., Ltd., Shanghai, China) (1.5 mg) were used to transfect 293T cells. Following 48 h of transfection, cellular luciferase activity was detected with a Microplate reader (Tecan GENios Pro; Tecan Group, Ltd., Mannedorf, Switzerland) according to the manufacturer's protocol.

HMG2 transcription factor binding sites for GLUT4 and SREBF1 5′-promoter luciferase vectors were amplified and inserted into pGL3 luciferase vectors (Promega Corporation). Next, the vectors with either the pcDNA3.0-HMGA2 overexpression plasmid or empty plasmid (0, 0.5, 1.5, 2.5 and 5.0 µg) were transfected into 293T cells. Luciferase activity was examined after 48 h of transfection. SREBF1 transcription factor binding sites were used to establish GLUT4 5′-promoter wild-type and mutant type luciferase vectors. Vectors with either the pcDNA3.0-SREBF1 overexpression plasmid or empty plasmid were transfected into 293T cells, and cellular luciferase activity was assessed.

To analyze the binding of SREBF1/HMGA2 protein to the GLUT4 gene promoter, adipocyte cells transfected with either SREBF1 or HMGA2 antibodies (Abcam) were used for ChIP analyses that were performed using a Pierce Agarose Chip Kit (EpiGentek Group, Inc., Farmingdale, NY, USA). Normal rabbit IgG was used as a negative control. Non-precipitated genomic DNA input was amplified as an input control. After purification, the DNAs were used for PCR analyses that were performed using primers (Sangon Biotech Co., Ltd., Shanghai, China) that encompassed the GLUT4 promoter region. The conditions used for PCR were as follows: denaturation at 95°C for 2 min, followed by 40 cycles at 95°C for 20 sec, 58°C for 20 sec, and 72°C for 20 sec.

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 19.0 (IBM Corporation, Armonk, NY, USA). The levels of miR-33b-5p and gene expression are shown as the mean ± SD, and differences between groups were analyzed using Student's t-test. One-way ANOVA and Tukey's test were used to evaluate the levels of miR-33b-5p and gene expression in human adipocytes treated with different concentrations of glucose or insulin. The association between miR-33b-5p and SREBF1/HMGA2 expression was assessed using Pearson's correlation test. P-values <0.05 were regarded as statistically significant.

The PCOS rat model was successfully developed using the method described by Poretsky et al (33). Insulin resistance in PCOS rats was indicated by their significantly increased insulin resistance index (HOMA-IR) values when compared with those of control rats (Table II). Based on their HOMA-IR score, the rats were then further classified into four groups: a non-PCOS/non-IR group, a non-PCOS/IR group, a PCOS/non-IR group, and a PCOS/IR group. The levels of miR-33b-5p expression in the ovarian tissues of non-PCOS/IR rats were significantly higher than those in the ovarian tissues of non-PCOS/non-IR rats (P<0.05). The PCOS/IR group also exhibited upregulated levels of miR-33b-5p expression when compared with the PCOS/non-IR group (Fig. 1A). Conversely, the levels of HMGA2, SREBF1, and GLUT4 expression in the ovarian tissues of non-PCOS/IR rats were significantly lower than those in the non-PCOS/non-IR rats (both P<0.05). PCOS/IR rats had lower HMG2 and GLUT4 levels when compared with PCOS-non-IR rats (Fig. 1B-D). Furthermore, the levels of HMG2 and GLU4 expression were lower in the PCOS/non-IR group than in the non-PCOS/non-IR group. miR-33b-5p levels were negatively associated with SREBF1 and HMGA2 expression in the ovarian tissues of rats in all four groups (P<0.01), and miR-33b-5p levels were negatively correlated with GLUT4 levels (P<0.01) (Fig. 1E-G).

H&E staining of rat ovarian tissues revealed that ovary oocytes in the non-PCOS/non-IR group were clear, and large numbers of granulosa cells were present. Ovarian tissues in the other three groups contained significantly fewer granulosa cells. The granulosa cells were loosely arranged, and the cortical stromal cells displayed obvious signs of hyperplasia (Fig. 2). Furthermore, immunohistochemistry was conducted to detect and compare the expression of GLUT4, SREBF1 and HMGA2, which validated the data by real-time PCR in Fig. 1. Consistently, obvious positive expression of GLUT4, SREBF1 and HMGA2 (yellow color) were found in non-PCOS/non-IR group compared to that in the other three groups. Immunohistochemistry data revealed that GLUT4, SREBF1 and HMGA2 were all highly expressed in the ovarian tissues of rats in the non-PCOS/non-IR group, however, their expression levels were significantly decreased in the PCOS/IR group (Fig. 2).

As shown in Fig. 3, excess extracellular glucose increased miR-33b-5p mRNA expression and reduced GLUT4, SREBF1, and HMGA2 expression in adipocytes (Fig. 3A-D). These findings were further supported by western blotting data (Fig. 3E-F). These results revealed that excess glucose can alter the level of miR-33b-5p expression, which in turn, may further alter the expression of mRNAs for GLUT4, SREBF1 and HMGA2 in adipocytes.

We also assessed the effect of extracellular insulin on miR-33b-5p, GLUT4, SREBF1 and HMGA2 expression in adipocytes. We found that insulin produced a similar effect as glucose; it upregulated miR-33b-5p levels and downregulated GLUT4, SREBF1 and HMGA2 mRNA and protein levels (Fig. 4). These results revealed that GLUT4, SREBF1 and HMGA2 expression in human adipocytes may also be regulated by changes in miR-33b-5p levels induced by excess insulin.

After determining that increased miR-33b-5p levels produced by treatment with either excess extracellular glucose or insulin suppressed GLUT4, SREBF1 and HMGA2 expression, we examined the effect of miR-33b-5p on GLUT4, SREBF1 and HMGA2 levels in adipocytes by reducing miR-33b-5p expression. Our data revealed that the expression levels miR-33b-5p gradually decreased as the concentration of miR-33b-5p inhibitor increased. In contrast, the GLUT4, SREBF1 and HMGA2 mRNA and protein levels increased in a concentration-dependent manner, suggesting that miR-33b-5p negatively regulates the levels of GLUT4, SREBF1 and HMGA2 (Fig. 5).

Transient co-transfection of the SREBF1 5′-promoter luciferase vector with either the pcDNA3.0-HMGA2 overexpression or empty plasmid revealed that pcDNA3.0-HMGA2 significantly increased luciferase activity in a dose-dependent manner, suggesting that HMGA2 could target SREBF1 and promote its expression (Fig. 6A). Chip data (Fig. 7A) revealed that HMGA2 directly binds to the 5′-promoter region of SREBF1 and promotes its expression.

Transient co-transfection of the GLUT4 5′-promoter luciferase vector with either the pcDNA3.0-HMGA2 overexpression or empty plasmid revealed that pcDNA3.0-HMGA2 also increased luciferase activity in a dose-dependent manner (Fig. 6B). Chip experiments also demonstrated that HMGA2 could bind to the 5′-promoter region of GLUT4 (Fig. 7B). These findings revealed that HMGA2 can directly bind to the 5′-promoter region of GLUT4 and promote its expression.

Transient co-transfection of the wild-type GLUT4 5′-promoter or mutant type vectors with either the pcDNA3.0-SREBF1 overexpression or empty plasmid revealed that SREBF1 could significantly increase luciferase activity in the wild-type GLUT4 5′-promoter group (Fig. 6C). Chip experiments also demonstrated that SREBF1 could bind to the 5′-promoter region of GLUT4 (Fig. 7C). These findings indicated that SREBF1 can directly bind to the 5′-promoter region of GLUT4 and promote its expression.

Transient co-transfection of the wild-type luciferase HMGA2 3′-UTR and mutant vectors with either miR-33b-5p mimics or NC mimics into 293T cells demonstrated direct binding of miR-33b-5p to the 3′-UTR of HMGA2, and resulted in significant reductions in luciferase activity (Fig. 6D). These results revealed that miR-33b-5p could target the 3′-UTR of HMGA2 and inhibit its expression.