Introduction
Growth differentiation factor 11 (GDF11), also known as bone morphogenetic protein 11, is a member of the TGF-β family playing pleiotropic roles in mammalian development [
1]. GDF11 is expressed in multiple tissues, including heart, kidney, skeletal muscle, nervous system, olfactory system, retina, pancreas, intestine. GDF11 has been considered as a rejuvenation factor capable of reversing aging-related dysfunctions in multiple organs including cardiac hypertrophy, skeletal muscle dysfunction, and cerebral vasculature dysfunction [
1‐
5] although conflict results have been reported [
6‐
9]. The correlation of blood circulation level of GDF11 with aging was conflicting in humans showing increase, decrease, or no change with aging [
6,
10‐
14]. Some reports showed that the circulation level of GDF11 increased in type 2 diabetes (T2D) or obesity in humans and mice [
15,
16]. However, some reports are contrary, showing no change with type 2 diabetes and obesity in humans [
13,
14].
More recent studies have suggested the potential of GDF11 for the treatment of metabolic diseases such as atherosclerosis, type 2 diabetes, and diabetes-related vascular dysfunctions [
15,
17,
18]. Administration of recombinant GDF11 protein (rGDF11) improved β cell function and attenuated the symptom of T2D in diabetic mice [
17]. Mei and colleagues demonstrated a reduction of atherosclerotic plaques and improvement of endothelial injury using AAV-mediated
Gdf11 gene transfer in
ApoE−/− mice [
15]. Employing the psoriasis-like skin inflammation model and high-fat diet (HFD) induced obese mice, Wang and colleagues demonstrated the anti-inflammatory activity of GDF11 [
15,
19]. Also, GDF11 activated multiple signal pathways, such as Smad, Akt, and p38 MAPK, which were involved in the development of obesity, fatty liver, and insulin resistant [
1,
20‐
22], implying that GDF11 may play an important role in obesity and obesity-related metabolic disorders. However, little is known about the function of GDF11 in the development of obesity and obesity-related metabolic disorders.
The focus of the current study is to investigate the functions of GDF11 in regulating metabolic homeostasis and energy balance in high fat diet-induced obesity mice and animals with STZ-induced diabetes. We demonstrate that GDF11 overexpression via gene transfer leads to significant improvement of metabolic homeostasis in obese mice and mice with STZ-induced diabetes, and blockade of high fat diet-induced weight gain, hyperglycemia, insulin resistance, and fatty liver development. These results support the notion that GDF11 plays a critical role in regulating metabolic homeostasis and could be considered as a therapeutic agent for the treatment of metabolic disorders.
Materials and methods
Materials
The pLIVE empty vector and pLIVE-SEAP plasmid (carrying secreted alkaline phosphatase gene) were purchased from Mirus Bio (Madison, WI, USA). The cDNA of the mouse Gdf11 gene was amplified by PCR and inserted into pLIVE vector at BamH1 and Sac1 sites to make pLIVE-GDF11 plasmid. The new plasmid construct was amplified in E. coli and extracted using endotoxin-free maxi plasmid kits from Tiangen Biotech (Beijing, China). The inserted Gdf11 gene sequence in the plasmid was verified by DNA sequencing. PrimeScript™ RT reagent kit was from Takara Bio. (Dalian, China). SYBR Green kit for real time-PCR was from Qiagen (Duesseldorf, Germany). BCA Quantitation kit for proteins was purchased from Applygen Technologies Inc. (Beijing, China). ELISA kit for GDF11 protein was from MEIMIAN (Cat. MM-44346M1, Wuhan, China). Primary antibodies against AKT (#4691), p-AKT (Thr 308, #13038), SMAD2 (#5339), p-SMAD2 (Ser 465/467, #3108), FOXO1 (#2880), AMPK (#5831), and p-AMPK (Thr172, #2535) were from Cell Signaling Technology (Danvers, CO, USA). Anti-p-FOXO1 antibody (Ser 256, ab131339) was from Abcam (Cambridge, UK). Primary antibodies against UCP1 (#23673-1-AP), UCP2 (#11081-1-AP), and β-actin (#66009-1-lg) antibodies were from Proteintech (Chicago, USA). The HRP-linked anti-mouse (Cat. ZB-2305) and anti-rabbit (Cat. ZB-2301) second antibodies were from ZSGB BIO Inc. (Beijing, China). Human insulin (Humulin) was from Eli Lilly (Indianapolis, IN, USA). Mouse Insulin ELISA kit was from Shanghai Enzyme-linked Biotechnology Co., Ltd. (Cat. ml001983, Shanghai, China). H&E staining kit was from Yulu (Cat. L11020102, Nanchang, China). The streptozotocin (STZ) was purchased from Sigma Aldrich (St. Louis, MO, USA). The high-fat diet (60% kJ/fat, 20% kJ/carbohydrate, 20% kJ/protein) and regular Chow were from Research Diets, Inc. (Cat. D12492,NJ,USA) and Keao Xieli Feed Cooperation (SPF-level mice maintaining Chow, Beijing, China), respectively.
Animal procedure
C57BL/6 mice (male, ~ 25 g) were purchased from Charles River Laboratories China (Beijing, China). All animals were group-housed under standard conditions at 25 ± 2 °C with a 12 h light–dark cycle with free access to food and water. The animal protocol used was approved by the Animal Ethics Committee of the Nanchang University. In studies designed to examine the effect of
Gdf11 gene transfer on preventing HFD-induced obesity, mice were divided into 3 groups (5 mice each). Two groups were fed an HFD and the other with regular Chow. Each mouse was hydrodynamically injected with 25 µg (dose: 1 mg/kg) of pLIVE-GDF11 or pLIVE-SEAP control plasmid DNA, respectively, according to the previously published procedure [
21]. Body weight and food intake were measured weekly. The rectal temperature of the animals was measured by an electric rectal thermometer weekly. Blood was collected at predetermined time points, and serum was prepared and stored at − 20 °C until use.
To examine the effects of Gdf11 gene transfer on obese mice, mice were fed an HFD to establish obesity, and then divided into two groups (5 mice each). pLIVE-GDF11 or pLIVE-SEAP plasmids were hydrodynamically injected via tail vein, respectively. The volume of injected saline solution with 25 µg plasmid DNA was adjusted to 6% body weight for obese mice.
For the STZ-induced type 2 diabetic model, 15 mice were fed an HFD for 4 weeks, and injected intraperitoneally with a single dose of STZ (100 mg/kg). Blood glucose levels were measured 2 weeks later using Ultra One Touch glucosemeter (Johnson & Johnson, USA). The mice were considered diabetic when non-fasting blood glucose level was higher than 13.9 mmol/l for at least 2 consecutive days. The diabetic mice were divided into 2 groups (5 mice each) and 25 µg of plasmid DNA were hydrodynamically injected with an injection volume equal to 8% body weight. Animals were continued on an HFD for additional 4 weeks, and non-fasting blood glucose levels were determined and serum biochemistry assays were performed.
Glucose tolerance test (GTT) and insulin tolerance test (ITT)
GTT and ITT were performed during the last week of the study on the same animals with a 2-day interval for recovery. For GTT, mice were fasted for 8 h and intraperitoneally injected with glucose in saline (0.2 g/ml, 2 g/Kg). Blood glucose was determined using glucose strips at 0, 30, 60, and 120 min after injection. For ITT, animals were fasted for 6 h and then intraperitoneally injected with insulin (100 IU/ml, 0.75 U/Kg). Blood glucose level was measured at the same time points as in GTT. Total blood was collected 2 days later for serum biochemistry. Insulin levels in the blood were determined using the Mouse Insulin ELISA Kit. Insulin resistance (HOMA-IR) was calculated using the formula as previous described [
23]: HOMA-IR = (fasting insulin (ng/ml) × fasting blood glucose (mg/dl)/405).
Histochemistry analysis
Collected tissue samples were fixed in 10% neutrally buffered formalin for at least 24 h and embedded in paraffin following the instruction of H&E staining kit. Tissue sections were made at a thickness of 5 μm followed by H&E staining. For Oil-Red O staining, freshly collected liver samples were embedded in OCT medium and frozen in liquid nitrogen. The frozen tissues were equilibrated in a cryostat and sectioned at 8 μm in thickness. The slide sections were fixed in 10% neutrally buffered formalin for 30 min, stained with 0.5% Oil Red O in 60% isopropanol for 15 min, counterstained with hematoxylin, washed three times, and then sealed in neutral gum. Histological photographs were taken under an optical microscope equipped with a Nikon camera.
Immunohistochemistry
White adipose tissue was embedded in paraffin and cut at 5 μm thickness. The tissue sections were placed in citrate antigen repair buffer (pH 6.0) in the thermostatic water bath for antigen retrieval. Slices were then put into 3% BSA buffer and blocked at room temperature for 30 min. After blocking, the slices were incubated with primary antibody against F4/80 (dilution 1:500, Cat. GB11027, Wuhan Servicebio Technology Co., Ltd., Wuhan, China) at 4 °C overnight and HRP-linked secondary antibody (dilution 1:200, Cat. G1213, Wuhan Servicebio Technology Co., Ltd., Wuhan, China) for 50 min at room temperature. Then color development was performed with freshly prepared DAB coloring solution (Cat. G1212-200, Wuhan Servicebio Technology Co., Ltd., Wuhan, China). After counterstained nuclei with hematoxylin for 3 min, the slices were dehydrated and sealed.
Biochemical analysis
Serum concentrations of aspartate aminotransferase (Cat. AST03, NingBo PureBio Biotechnology, Ningbo, China), alanine aminotransferase (Cat. ALT03, NingBo PureBio Biotechnology, Ningbo, China), triglycerides (E1013, Applygen Technologies Inc.), total cholesterol (E1015, Applygen Technologies Inc.), and free fatty acids (#15781, Diasy Diagnostic Systems, Frankfurt, Germany) were determined using ELISA kits. To assess the lipid contents in the liver, liver samples (100 mg) were homogenized in a mixture of chloroform and methanol (2:1, volume ratio) and incubated at 70 °C for 10 min. The homogenates were centrifuged at 2000 rpm for 5 min, and the supernatants were dried and re-dissolved in 5% Triton X-100. The amounts of cholesterol, triglyceride and free fatty acids were determined following the manufacturers’ instructions.
The individual mouse was placed in a gas-tight metabolic cage and acclimated for 1 day before the parameters of energy metabolism were determined using a TSE-PhenoMaster system (TSE Systems, Germany) as previously described [
24]. The parameters monitored include water and food consumption, total animal activity, the volume of O
2 (VO
2) and CO
2 (VCO
2), and respiratory exchange ratio (RER). VO
2 was calculated by the equation: “VO
2” [ml/h/kg] = FlowML [ml/h] * (V1 [%^2] + V2 [%^2])/(N2Ref[%] * BodyWeight[kg] * 100.0 [%]). VCO
2 was calculated by the equation: “VCO
2” [ml/h/kg] = FlowML [ml/h]*dCO
2 [%]/(BodyWeight [kg] * 100.0 [%]). RER was calculated by the equation: RER = VCO
2 [ml/h/kg]/VO
2 [ml/h/kg]. Energy expenditure (EE) was calculated by the equation: EE = 3.941 × VO
2 + 1.106 × VCO
2.
Analysis of gene expression
Real-time PCR was performed to determine the expression level of selected genes. Total RNA was isolated from the mouse liver, white and brown adipose tissues using TRIzol reagents from Invitrogen. One μg of total RNA was reverse-transcribed to generate the first-strand cDNA using PrimeScript™ RT reagent kit. RT-PCR was performed using SYBR Green kit. GAPDH RNA served as an internal control and data were normalized using GAPDH RNA level as 1. All primer sequences employed are summarized in Additional file
1: Table S1.
Western blotting
Proteins in brown and white adipose tissue were extracted by adipose tissue protein extraction kit (BB-312262, BestBio, Shanghai, China). Briefly, 100 mg adipose tissue was homogenized in 500 μl protein extraction buffer with protease inhibitors. The homogenate was incubated for 30 min at 4 °C and centrifuged for 15 min at 12,000g at 4 °C. The supernatant was collected and centrifuged one more time using the same condition. The concentration of total proteins was detected by BCA Quantitation Kits. The same amount of sample (100 µg total protein) was loaded in each well and resolved by SDS-PAGE. The protein bands after electrophoretic separation were transferred to Immobilon-P PVDF Membrane by Mini-PROTEAN® Tetra system at 200 mA for 90 min with transfer buffer (25 mM Tris, 189 mM Glycine, 200 ml MeOH, 800 ml ddH2O). Primary antibodies (dilution 1:2000) were added and incubated overnight at 4 °C. After washing, the HRP-linked anti-mouse second antibody (dilution 1:2500) or anti-rabbit second antibody (dilution 1:2500) were added and incubated for 120 min. The specific protein bands were visualized using a Chemiluminescence imager. The relative levels of protein were scaled to the level of β-actin. The relative levels of the phosphorylated protein were normalized to the signal of their total abundance of that protein.
Statistical analysis
Statistical analysis was performed by using the Student’s t-test, one-way ANOVA, or nonparametric test. Normality and homogeneity of variances were analyzed using Shapiro–Wilk test and Levene’s test respectively. Post-hoc comparisons were performed using LSD and SNK test in oneway ANOVA. If the data was still not a normal distribution after log transformation, Kruskal–Wallis H test and Mann–Whitney U test were performed. Results were expressed as the mean ± SEM. P < 0.05 was considered significantly different.
Discussion
In this study, we demonstrate that hydrodynamic transfer of Gdf11 gene prevented HFD induced obesity, hyperglycemia, insulin resistance and fatty liver in mice. Gdf11 gene transfer also improved metabolic homeostasis in obese mice and mice with STZ-induced diabetes.
The finding that GDF11 is capable of improving glucose metabolism and metabolic homeostasis is consistent with the results of previous studies using
db/db and STZ-induced diabetic mice as an animal model [
15,
17]. Two mechanisms might be involved. First of all, it is possible that overexpression of GDF11 promoted β-cell differentiation and development of pancreas through Smad2 and PI3K/AKT/FoxO1 signal pathways [
17,
26,
27] as previously shown that administration of rGDF11 protein or AAV-GDF11 vectors increased the survival of β-cells in diabetic mice [
17]. Results in Fig.
9 confirmed that GDF11 activated TGF-β/Smad2 and PI3K/AKT/FoxO1 pathways in WAT. The second possibility is that GDF11 blocks pathways that lead to metabolic disorders that involve inflammatory cytokines and activation of macrophages [
21,
22,
28]. It was previously shown that GDF11 has anti-inflammatory effects [
15,
19,
29]. Treatment with rGDF11 attenuated inflammation in psoriasis-like skin inflammation mice and obese mice [
15,
19]. rGDF11 or AAV-mediated
Gdf11 gene transfer reduced inflammatory cytokines and suppressed the expression of inflammatory genes in aortas, and increased anti-inflammation cytokine IL10 in
apoE−/− mice [
15]. The conclusions of these earlier studies are supported by our results demonstrating that GDF11 suppressed the expression of inflammatory genes in WAT including
Tnfα and
Ccl2 and macrophage marker genes such as
F4/80, Cd68, Cd11b, and
Cd11c in adipose tissue (Fig.
8a). GDF11 also reduced the development of crown-like structures in adipose tissue, a sign of macrophage infiltration and migration into WAT (Figs.
2c,
8b). As inflammation is known for its role in inducing obesity and metabolic complications, the anti-inflammation function of GDF11 is likely one of the reasons for the beneficial effects seen in animals employed in our study (Fig.
9d). In addition, GDF11-induced decrease in expression of gluconeogenesis gene
G6P (Fig.
8g), the downstream of PI3K/AKT/FoxO1 signal pathway, in the liver, may also play critical role in reducing blood glucose level and reestablishing glucose homeostasis (Fig.
9d). Moreover, AMPK activity was induced by GDF11 (Fig.
9b), which may also play an important role in glucose uptake, and glucose homeostasis [
30,
31].
The fact that no significant weight gain was seen in GDF11-treated animals while there was no difference in their food intake (Fig.
1f, g) and activity compared to the HFD-fed control animals suggests that GDF11-treated animals consume more energy than HFD-fed control animals, which was confirmed by metabolic cage assay that the oxygen consumption and energy expenditure were significantly increased after GDF11 treatment (Fig.
7). The higher level of energy consumption is likely achieved by converting food energy to heat which is released. The rectal temperature was a little bit higher in GDF11-treated mice comparing with HFD-fed control and Chow animals (Fig.
7i). GDF11-induced increase in Ucp1 and Ucp2 expression seen in BAT (Fig.
8c, d) supports the notion that heat release is the mechanism that prevents the animals with
Gdf11 gene transfer from gaining weight. Increased thermogenesis is also supported by elevated expression of
Elovl3 which is an important component for lipid recruitment in BAT [
32].
Elovl3 gene expression is significantly increased during cold stimulation linking
Elovl3 to the thermogenic process [
33]. These suggest that GDF11-mediated thermogenesis and energy consumption play important roles in prevention of HFD-induced obesity and metabolic disorders.
Previous studies have demonstrated that GDF11 binds to activin type I/II receptors and activates the canonical TGF-β/Smad signaling pathway. TGF-β/Smad pathway was involved in improving islet β-cell function, glucose homeostasis, and lipid metabolism [
17,
34‐
36]. In the current study, we also revealed that GDF11 significantly increased the phosphorylation of SMAD2 in WAT (Fig.
9a). In addition, GDF11 also increased the phosphorylation of AKT and FoxO1 (Fig.
9c) which play an important role in the PI3K/AKT/FoxO1 signal cascade critical for regulating lipid metabolism and glucose homeostasis in adipose tissue, liver, pancreas, and skeletal muscle [
37,
38]. Moreover, the phosphorylation of AMPK was also upregulated by GDF11 (Fig.
9b). It is known that PI3K/AKT/FoxO1 and AMPK pathways, the noncanonical GDF11 signal cascades, have cross-talk with canonical TGF-β/Smad cascade of GDF11 influencing metabolic homeostasis [
31,
39,
40]. These results suggest that GDF11 function in preventing metabolic disorders is achieved through TGF-β/Smad2, PI3K/AKT/FoxO1, and AMPK pathways.
Hydrodynamic injection of
Gdf11 gene generated the sustained blood circulating level of GDF11 and long-term anti-diabetic effects, indicating that GDF11 has therapeutic potential in treating diabetes and obesity-related metabolic diseases. However, the detailed mechanisms of GDF11 in preventing obesity and fatty liver remain elusive, as well as the mechanisms of GDF11 in regulating oxygen consumption, thermogenesis, and inflammation. Further studies are needed to investigate the mechanisms of GDF11 in regulating adipocyte development and metabolic homeostasis. Moreover, the hydrodynamic tail vein injection is a physical method of gene delivery to hepatocytes in the liver [
41]. The overexpressed GDF11 was secreted into the blood circulation and reached multiple tissues. Additional work is needed to investigate the long term effect of GDF11 in bone, skeletal muscle, and other tissues where GDF11 activities have been demonstrated [
16,
42,
43]. Differences reported in different animal models about GDF11 functions need additional investigation. For example, the role of GDF11 on the development of skeletal muscle was different in young and aged mice [
44]. AAV-mediated
Gdf11 gene transfer blocks the growth and development of skeletal muscle in neonatal mice [
42], but supplementary of GDF11 in old mice restored skeletal muscle dysfunctions, enhanced muscle fibers, and increased the number of multinucleated myotubes [
4]. Controversial results have also been reported about the function of GDF11 in aging-related cardiac hypertrophy, cerebral vasculature dysfunctions, and cachexia [
1,
15,
45,
46]. The effects of GDF11 on aging have been the focus of GDF11-related research in recent years, and future studies will surely shine light on these seeming controversy results.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.