|Home | About | Journals | Submit | Contact Us | Français|
Both type 1 and type 2 diabetes are associated with increased risk of cardiovascular disease. This is in part attributed to the effects of hyperglycemia on vascular endothelial and smooth muscle cells, but the underlying mechanisms are not fully understood. In diabetic animal models, hyperglycemia results in hypercontractility of vascular smooth muscle possibly due to increased activation of Rho-kinase. The aim of the present study was to investigate the regulation of contractile smooth muscle markers by glucose and to determine the signaling pathways that are activated by hyperglycemia in smooth muscle cells. Microarray, quantitative PCR, and Western blot analyses revealed that both mRNA and protein expression of contractile smooth muscle markers were increased in isolated smooth muscle cells cultured under high compared with low glucose conditions. This effect was also observed in hyperglycemic Akita mice and in diabetic patients. Elevated glucose activated the protein kinase C and Rho/Rho-kinase signaling pathways and stimulated actin polymerization. Glucose-induced expression of contractile smooth muscle markers in cultured cells could be partially or completely repressed by inhibitors of advanced glycation end products, L-type calcium channels, protein kinase C, Rho-kinase, actin polymerization, and myocardin-related transcription factors. Furthermore, genetic ablation of the miR-143/145 cluster prevented the effects of glucose on smooth muscle marker expression. In conclusion, these data demonstrate a possible link between hyperglycemia and vascular disease states associated with smooth muscle contractility.
Diabetes confers a 2–4-fold excess risk for a wide range of cardiovascular diseases, including macrovascular complications leading to coronary heart disease and ischemic stroke, as well as microvascular diseases, such as nephropathy and retinopathy (1, 2). Based on current trends, the rising incidence of diabetes (expected to reach 333 million people worldwide by 2025) will undoubtedly equate to increased cardiovascular mortality. Chronic hyperglycemia has long been recognized as an independent risk factor for cardiovascular disease (3, 4). Importantly, the progressive relationship between glucose levels and cardiovascular risk extends below the threshold for diabetes diagnosis (fasting plasma glucose ≥7.0 mmol/liter or 2-h plasma glucose ≥11.1 mmol/liter (2, 3)), and more recently, even transient hyper- and hypoglycemia have emerged as important determinants of cardiovascular disease (5). Despite the vast clinical and epidemiological experience linking blood glucose and poor glucose control to the development and progression of cardiovascular disease, the underlying molecular mechanisms leading to vascular dysfunction and disease are poorly understood (6).
It has been well established that hyperglycemia results in vascular hyperreactivity in diabetic patients (7) and animal models (8,–10). Part of this effect may be attributed to a decrease in nitric oxide (NO) bioavailability as well as a reduced response of vascular smooth muscle cells to NO (11). However, hyperglycemia can also cause an endothelium-independent hypercontractility of smooth muscle cells (8, 10, 12). The underlying mechanism behind this effect is in part due to activation of the PKC/Rho/Rho-kinase pathway by glucose, which results in calcium sensitization by inhibiting myosin light chain phosphatase (12,–15). The activation of PKC and Rho in smooth muscle by hyperglycemia is thought to be dependent on the formation of advanced glycation end products (AGEs)2 (16, 17).
In addition to calcium sensitization, activation of the Rho/Rho-kinase pathway in vascular smooth muscle cells promotes actin polymerization, an effect that plays a major role in the regulation of smooth muscle gene expression and activates myocardin-related transcription factors (MRTFs, also known as MKL1/2), co-factors to serum response factor (SRF) (18, 19). Monomeric actin sequesters MRTFs and inhibits their activity by preventing interaction with SRF. When actin is polymerized, MRTFs are released and can associate with SRF in the nucleus. SRF binds to genetic elements referred to as CArG boxes (CC(A/T)6GG), and this binding is antagonized by the transcription factor Krüppel-like factor 4 (20). We have previously demonstrated that actin polymerization is crucial for stretch-induced contractile differentiation in vascular smooth muscle (21,–24). Further, we reported that important regulators of smooth muscle differentiation, such as the small non-coding microRNA miR-145, promotes actin polymerization and that this effect is required for the regulation of contractile smooth muscle genes (25,–27). However, the impact of actin polymerization on smooth muscle gene expression in the context of diabetic vascular disease has not yet been explored.
Herein, we aimed to determine the effects of extracellular glucose on vascular smooth muscle contractile differentiation. This was investigated using isolated smooth muscle cells in culture, arteries from hyperglycemic Akita mice, and mammary arteries from diabetic patients.
MicroRNA-143/145 knock-out (KO) mice were generated at the Max-Plank-Institute for Heart and Lung research as described (28). Wild type (WT) littermates were used as controls.
Akita type 1 diabetic mice (C57BL/6-Ins2C96Y/+/J) were obtained from the Jackson Laboratory and bred at Lund University. Mice were euthanized by exsanguination through cardiac puncture under deep anesthesia. Mesenteric arteries from adult 12-week-old male heterozygous Akita and littermate wild-type non-diabetic control mice were placed in RNAlater (catalog no. 76104, Qiagen) and dissected free from fat and surrounding tissue.
All experiments were approved by the Malmö/Lund animal ethics committee. This investigation conforms to Directive 2010/63/EU of the European Parliament.
Left internal mammary arteries from patients undergoing bypass surgery were obtained in collaboration with surgeons at Skåne University Hospital in Lund. The vessels were immediately placed in ice-cold HEPES buffer, dissected free from fat and surrounding tissue, and snap-frozen in liquid nitrogen. All participants gave their informed consent. The study protocols conformed to the Declaration of Helsinki and were approved by the local Human Ethics Committee.
Vascular smooth muscle cells were isolated from miR-143/145 KO and WT mouse aorta by enzymatic digestion and maintained in culture as described (27). Mouse aortic smooth muscle cells from passage 3 were cultured in DMEM (Gibco, catalog no. 11966-025) supplemented with low glucose (1.7 mm glucose), normal glucose (5.5 mm glucose), high glucose (25 mm glucose), or low glucose plus mannitol (1.7 mm glucose + 23.3 mm mannitol) for 1–6 weeks. The medium was routinely exchanged every 48 h. All media were supplemented with 10% fetal bovine serum (Biochrom, catalog no. S0115) and 50 units/50 μg/ml penicillin/streptomycin (Biochrom, catalog no. A2212). Cells were passaged approximately every 10 days during glucose treatment. Cells were treated with Latrunculin B (250 nm; Calbiochem, catalog no. 76343-94-7), Y-27632 (10 μm; Tocris, catalog no. 1254), CCG-1423 (10 μm; Calbiochem, catalog no. 555558), GF-109203X (10 μm; Sigma-Aldrich, catalog no. B6292), verapamil (1 μm; Sigma-Aldrich, catalog no. V4629), aminoguanidine hydrochloride (100 μm; Sigma-Aldrich, catalog no. A7009), or 0.1% DMSO as a vehicle control for the last 24 h of a 1-week incubation. The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) cell viability assay was used to exclude any cytotoxic effects of the inhibitors.
siRNA experiments were performed using Silencer® Select predesigned siRNA (catalog no. 4390771, Thermo Scientific). Cells were transfected with 20 nm MKL1 siRNA (s73019), 20 nm MKL2 siRNA (s109074), 20 nm ROCK1 siRNA (s104595), 20 nm ROCK2 siRNA (s73020), or 20 nm control siRNA using Oligofectamine transfection reagent (Invitrogen) for 96 h. In some experiments, cells were transfected with 10 nm MKL1 + 10 nm MKL2 siRNA or 10 nm ROCK1 + 10 nm ROCK2 siRNA. Each siRNA caused a ~50% reduction in the expression of its respective target gene.
Total RNA was isolated using the miRNeasy minikit (Qiagen, catalog no. 217004), as described previously (29). Purity and concentration were determined with an ND-1000 spectrophotometer (Nanodrop Technologies Inc., Wilmington, DE). The expression of mRNA was analyzed by real-time qPCR (StepOnePlus qPCR cycler, Applied Biosystems) using the QuantiFast SYBR Green RT-PCR kit (Qiagen, catalog no. 204156) and QuantiTect primer assays: mouse, Mm_Kcnmb1, QT00101500; Mm_Tagln, QT00165179; Mm_Cnn1, QT00105420, Mm_GADPH, QT01658692; Mm_Dmd, QT00161336; Mm_Des, QT00102333; Mm_Dcn, QT00131068, Mm_Ptgs2 (COX-2), QT00165347; Mm_Tnf, QT00165347; Mm_Mglap, QT00145775; Mm_Spp1, QT02524536. Primer sequences are proprietary of Qiagen.
RNA was extracted using the RNeasy minikit (Qiagen, catalog no. 74106). Purity and concentration were determined with an ND-1000 spectrophotometer (Nanodrop Technologies). After passing quality control by an Agilent Bioanalyzer, total RNA was analyzed by an Affymetrix GeneChip® Mouse Gene 2.0 ST Array at the Swegene Center for Integrative Biology at Lund University. The microarray data are accessible via the Gene Expression Omnibus (accession number GSE66280, scheduled release on 1 July 2015).
Transcription factor binding site analysis was made using the online tool oPOSSUM. q = 0 was used as cut-off, and default search parameters were chosen throughout.
Cells grown on 6-well plates were washed twice with ice-cold PBS and lysed on ice directly in the wells with 75 μl of 1× Laemmli sample buffer (60 mm Tris-HCl, pH 6.8, 2% SDS, 10% glycerol). Frozen tissue was pulverized using liquid nitrogen as described previously (30). Proteins were detected using commercially available primary antibodies: α-actin (1:2000; Sigma-Aldrich, catalog no. A5528), calponin (1:1000; Abcam, catalog no. ab46784), SM22α (1:5000; Abcam, catalog no. ab14106), phospho-ERM (ezrin (Thr-567)/radixin (Thr-564)/moesin (Thr-558)) (1:1000; Cell Signaling, catalog no. 3141S), total ERM (Cell Signaling, catalog no. 3142S), total and phosphorylated cofilin-2 (Ser-3) (1:500; Merck Millipore, catalog nos. 07-300 and 07-326), phospho-LIMK1 (Thr-508) (1:1000; Cell Signaling, catalog no. 3841S), LIMK-1 (1:1000; Cell Signaling, catalog no. 3842s), phospho-MYPT (Thr-696) (1:1000; Cell Signaling, catalog no. 5163S), MYPT (1:1000; Cell Signaling, catalog no. 2634s), HSP90 (1:1000; BD Transduction Laboratories, catalog no. 610418), β-actin (1:2000; Sigma-Aldrich, catalog no. A5441), phospho-(Ser) PKC substrate (1:1000; Cell Signaling, catalog no. 2261S), AMP kinase α (AMPKα) (1:1000; Cell Signaling, catalog no. 2603), Thr-172-phospho-AMPKα (1:1000; Cell Signaling, catalog no. 2535), Ser-79-phosphoacetyl-CoA carboxylase (1:1000; Cell Signaling, catalog no. 3661), acetyl-CoA carboxylase (1:1000; Cell Signaling, catalog no. 3662). Secondary mouse or rabbit HRP-conjugated antibodies (1:5000 or 1:10,000; Cell Signaling, catalog nos. 7074 and 7076) or fluorescently labeled DyLight800 and DyLight680 secondary antibodies (1:5000; Cell Signaling Technology, catalog nos. 5257, 5366, 5470, and 5151) were used. Bands were visualized using ECL (Pierce West Femto), and images were acquired using the Odyssey Fc Imager (LI-COR Biosciences).
Mouse aortic smooth muscle cells grown in low or high glucose medium for 7 days. Medium was replaced every 48 h. At the end of the last 48-h incubation of a 1-week culture, medium was collected and evaluated using a glucose assay kit (GAGO-20, Sigma-Aldrich) according to the manufacturer's instructions.
Mouse aortic smooth muscle cells were cultured in low or high glucose medium for 7 days as described above. Before uptake, cells were rinsed in serum-free DMEM and incubated for 15 min in 1 μCi/ml [3H]2-deoxyglucose at room temperature. The uptake was terminated on ice, after which cells were solubilized in 1 n NaOH, dissolved in scintillation fluid, and subjected to scintillation counting. Cytochalasin B (1 μm) was used to measure nonspecific binding.
Mouse aortic smooth muscle cells from passage 3 were plated at a density of 3 × 103 cells/well in 96-well plates in low glucose (1.7 mm glucose) or high glucose (25 mm glucose) for 1 week. The media were subsequently replaced with 100 μl of serum-free DMEM and 10 μl of MTT solution (5 mg/ml MTT in PBS) for 1 h. The MTT-containing medium was removed, and the dark blue formazan crystals that formed in intact cells were solubilized with DMSO (100 μl), and the absorbance at 540 nm was measured with a Multiscan GO microplate spectrophotometer (Thermo Scientific). The percentage of cell viability was calculated based on the relative absorbance of treated cells versus cells exposed to the control vehicle.
Mouse aortic smooth muscle cells were incubated for 1 week in low (1.7 mm) or high (25 mm) glucose conditions. The culture medium was replaced with HEPES-buffered Kreb's solution (135.5 mm NaCl, 5.9 mm KCl, 2.5 mm CaCl2, 1.2 mm MgCl2, 11.6 mm HEPES, 11.5 mm glucose, pH 7.4, at 25 °C). Following a 90-min incubation at room temperature with 5 μm Fluo-4 (Life Technologies, F-14201) and 0.05% Pluronic-F127, the cells were washed and incubated for a further 30 min in HEPES-buffered solution to allow complete de-esterification of intracellular acetoxymethyl esters. Cells were then stimulated in room temperature with 60 mm KCl for 7 min. Relative changes in calcium levels were recorded using a Zeiss LSM 510 laser-scanning confocal microscope with excitation at 488 nm and emission at 505 nm. Images were acquired every 0.8 s, before and after each stimulation, and changes in global fluorescence (F/F0) were calculated in confluent cells at the peak and plateau (100–200 s after stimulation) response. Baseline fluorescence was evaluated using arbitrary units.
Aortic rings (1.5–2 mm in length) were mounted in a Mulvany myograph (610M, Danish Myo Technology) as described previously (31). After an equilibration period, cumulative concentration-response curves for serotonin (5-hydroxytryptamine; 5-HT) were generated. A 20 μm concentration of the Rho-kinase inhibitor Y-27632 was added on top of the highest concentration of 5-HT to evaluate the influence of Rho-kinase. The whole experiment was performed in the presence of the nitric-oxide synthase inhibitor l-NG-nitroarginine methyl ester (300 μm).
Mouse aortic smooth muscle cells (passage 3–5) from wild type and miR-143/145 knock-out mice were grown on glass coverslips in 10% FBS, DMEM low or high glucose medium for 1 week. The cells were transduced using cytomegalovirus promoter-driven adenoviral construct, with 100-multiplicity of infection Ad-hMKL1/eGFP (catalog no. ADV-215499, Vector Biolabs) for 96 h. Following fixation with 4% formaldehyde and permeabilization with 0.1% Triton X-100, the nuclei were stained with NucRed Dead 647 (catalog no. R37113, Molecular Probes). GFP and nuclear stains were visualized using a Zeiss LSM 510 laser-scanning confocal microscope with excitation at 488 and 633 nm. A fixed threshold was set to identify positive nuclei, and the ratios of positive versus total nuclei were then quantified from each condition.
Aortas from WT and miR-143/145 KO mice were fixed and sectioned as described previously (32). Sections were stained using calponin primary antibody (Abcam, catalog no. ab46784). The staining intensity was determined using a computerized image analysis system (CellSens dimension software, Olympus).
Actin polymerization was evaluated using the G-actin/F-actin In Vivo Assay Biochem Kit (Cytoskeleton, Inc., catalog no. BK037) according to the manufacturer's instructions. Briefly, mouse aortic smooth muscle cells grown in low or high glucose medium were collected in LAS02 lysis buffer (containing ATP and protease inhibitor mixture). F-actin was pelleted by centrifugation at high speed (100,000 × g) using a Beckman ultracentrifuge at 37 °C for 1 h. G-actin was transferred to fresh test tubes. F-actin pellet was dissolved in F-actin depolymerizing buffer and incubated on ice for 1 h. Equal volumes of filamentous and globular actin fraction (10 μl) were loaded on the gel. Proteins were transferred to nitrocellulose membrane using Trans Blot Turbo (Bio-Rad) for 10 min at 2.5 A. The membrane was then incubated with rabbit smooth muscle α-actin antibody (Cytoskeleton, catalog no. AAN01). Anti-rabbit HRP-conjugated secondary antibody (Cell Signaling, catalog no. 7076; 1:10,000) was used. Bands were visualized using ECL (Pierce West Femto), and images were acquired using the Odyssey Fc Imager (LI-COR Biosciences).
Values are presented as mean ± S.E. unless otherwise stated. Significance analysis of microarrays was done using TMEV version 4.0 software and returned a q value (33). A q value of ≤0.1 was considered significant. p values were calculated by Student's test with one- or two-way analysis of variance followed by Bonferroni's post hoc testing using GraphPad Prism version 5 (GraphPad Software Inc.). p < 0.05 was considered statistically significant. *, p < 0.05; **, p < 0.01; ***, p < 0.001.
To evaluate the effects of glucose on smooth muscle differentiation, aortic smooth muscle cells were cultured in low (1.7 mm), normal (5.5 mm), and high (25 mm) glucose conditions for 1, 3, or 6 weeks. Following 1 week of culture, during which the medium was exchanged every 48 h, the glucose level in the medium was slightly reduced to 1.1 ± 0.001 mm in LG and 21.1 ± 0.001 mm in HG. In accordance with previous observations (34, 35), we found that uptake of 2-deoxyglucose glucose uptake was significantly reduced in smooth muscle cells cultured under elevated glucose conditions for 1 week (-fold change in HG versus LG: 0.37 ± 0.11, p < 0.001, n = 5). High versus low glucose levels did not affect cell proliferation or viability measured by the MTT assay (-fold change: 1.01 ± 0.02 in HG versus LG, p > 0.05, n = 8). Elevating the extracellular glucose concentration for 1 week resulted in increased mRNA expression of several myocardin/MRTF/SRF-regulated contractile markers, including calponin (Cnn1), dystrophin (Dmd), desmin (Des), SM22 (Tagln), and the β1-subunit of the BK-channel (Fig. 1, A–E). All of these genes are positively regulated by actin polymerization in smooth muscle cells (36). Coincidentally, the expression of the proteoglycan decorin, which is reduced by actin polymerization (36), was decreased by high glucose (Fig. 1F).
To test the effect of glucose on genes associated with vascular calcification, we analyzed mRNA expression of osteopontin (Spp1) and matrix Gla protein (Mgp). Expression of Mgp was unchanged in HG versus LG conditions (-fold change: 0.96 ± 0.05), whereas Spp1 expression was significantly increased (-fold change: 3.82 ± 0.4, p < 0.001). Hyperglycemia has previously been demonstrated to activate inflammatory gene expression in endothelial cells (37). However, we did not observe any statistically significant increase in the mRNA expression of either TNF-α (LG, 1.330 ± 0.41-fold; HG, 1.965 ± 0.19-fold, p > 0.05, n = 6) or COX-2 (LG, 1.226 ± 0.43-fold; HG, 2.291 ± 0.38-fold, p > 0.05, n = 6) in smooth muscle cells.
Low glucose with 23.3 mm mannitol had no effect on gene expression, ruling out potential osmotic effects of high glucose. Similar to the effect on mRNA, glucose stimulated protein expression of smooth muscle contractile markers, including SM22, smooth muscle α-actin, and calponin (Fig. 1, G–J). The effect of glucose on these markers was still evident after 3 weeks in culture (data not shown). Furthermore, Cnn1 and Kcnmb1 were significantly elevated at a 6-week time point, whereas no significant change was observed for Dmd (data not shown). Shorter culture periods (48 and 96 h) had no impact on the expression of contractile smooth muscle markers, suggesting that the effects of glucose primarily occur during the end of the 1-week period (data not shown).
To further define the glucose-dependent transcriptome in smooth muscle cells, cells were cultured in low and high glucose for 2 weeks, and Affymetrix Gene ST mRNA arrays were performed. 491 genes were differentially expressed at the significance level q = 0. Four of five of the most highly up-regulated genes (Fig. 2A) were known myocardin/MRTF/SRF-regulated contractile markers. Among the five most highly repressed genes (Fig. 2B), two (Sdf2l1 and Derl3) play roles in the endoplasmic reticulum stress response; one is a component of the respiratory chain in mitochondria, and one plays a key role in the synthesis of vasoactive H2S (Cth, cystathionine γ-lyase). Genes involved in glucose metabolism via the glycolysis, sorbitol, and hexokinase pathways were also evaluated from the microarray data set. However, this analysis did not reveal any clear evidence of up- or down-regulation of any of these pathways (data not shown). Using the online tool oPOSSUM, we next identified transcription factors potentially involved in the regulation of the most differentially expressed genes. This in silico approach ranks the rate of occurrence of transcription binding sites in these genes, assigning z-scores (number of S.D. values above the mean) to each transcription factor, which is normally plotted versus the GC content of the specified genomic region. As shown in Fig. 2C, 15 transcription factors had z-scores higher than 10, and four clearly stood out, suggesting potential functional relevance. These four were Krüppel-like factor 4 (Klf4), SRF, SP1, and NFYA. The Affymetrix data also revealed that Klf4 expression was reduced (3-fold, q = 0), whereas Srf expression was increased, suggesting that the effects of glucose on SRF-dependent transcription may be in part due to Klf4 repression and Srf induction (Fig. 2D).
In further support of a key role of SRF for the glucose-dependent transcriptional response, we found that 18 of the 50 most up-regulated transcripts are confirmed or hypothetical SRF/myocardin/MRTF-regulated genes (Table 1). When comparing the effects of glucose with previous results using the actin-stabilizing agent jasplakinolide, we found that 40 of the 100 most up-regulated genes were common for both treatments. In all, these analyses show that myocardin/MRTF/SRF targets constitute a major transcriptomic signature in vascular smooth muscle cells exposed to high glucose.
Although it is known that glucose can stimulate the activity of Rho, leading to inhibition of myosin phosphatase-targeting protein (MYPT) and calcium sensitization, the duration of this response has not been thoroughly investigated. In accordance with previous observations (13), we found increased phosphorylation of MYPT after 1 week of high glucose exposure (Fig. 3, A and B). Furthermore, additional Rho-kinase targets, such as LIM kinase and ERM, were phosphorylated by glucose (Fig. 3, C and D). The effect was mainly observed between normal and high glucose levels, and the phosphorylation persisted at least for 3 weeks. In addition to Rho-kinase, phosphorylation of MYPT, LIM kinase, and ERM can be regulated by PKC. To determine the activation of PKC by glucose, we used a phospho-(Ser) PKC substrate antibody and found that glucose increased the activity of PKC in smooth muscle (Fig. 3F).
LIM kinase can phosphorylate and inactivate the actin depolymerizing factor cofilin, leading to stabilization of actin filaments. Phosphorylation of cofilin at Ser-3 was significantly increased in high compared with low glucose (Fig. 3E). Furthermore, using an F/G-actin assay, we found increased actin polymerization after 1 week of high glucose exposure (Fig. 3G).
To confirm the metabolic effects caused by altering extracellular glucose conditions, we next examined the effects of glucose on the AMPK pathway, which is a well known energy sensor in smooth muscle. As expected, the activating phosphorylation of AMPK and the inhibitory phosphorylation of its downstream substrate acetyl-CoA carboxylase were reduced in high glucose conditions. Collectively, these results support a role for PKC, Rho/Rho-kinase and actin polymerization in the regulation of smooth muscle contractile markers by glucose.
In order to identify the importance of the different mediators of the PKC/Rho-actin-MRTF signaling pathway in the response to glucose, we used a number of pharmacological inhibitors, including the actin-depolymerizing agent Latrunculin B, the Rho-kinase inhibitor Y-26732, the PKC inhibitor GF-109203X, and the MRTF inhibitor CCG-1423. These inhibitors were used during the last 24 h of a 1-week incubation in high glucose. We found that the inhibitors completely or partially reversed the high-glucose-driven changes in calponin, desmin, and dystrophin mRNA (Fig. 4, A–C). Furthermore, all of the PKC/Rho-actin-MRTF inhibitors reversed the glucose-induced changes in calponin, SM22α, and α-actin protein expression (Fig. 4, D–G). Similar results were obtained after silencing of Rho-kinase or MRTFs using siRNA (Fig. 4H).
The miRNAs miR-145 and miR-143 are highly expressed in smooth muscle and transcribed as a bicistronic cluster. MiR-145 is known to regulate smooth muscle contractile differentiation, in part via an effect on Rho signaling and actin polymerization. We therefore hypothesized that this miRNA cluster could play a role in the effects of glucose in smooth muscle (27, 38). To test this hypothesis, we analyzed the effect of glucose in smooth muscle cells from WT and miR-143/145 KO mice. Compared with the response in WT cells, the induction of smooth muscle marker expression was essentially lost in miR-143/145 KO smooth muscle cells, both at the mRNA and protein levels (Fig. 5, A–G). Furthermore, this effect correlated with a reduced phosphorylation of mediators in the Rho-signaling pathway (Fig. 5, H–K).
As demonstrated previously (28), basal expression of smooth muscle marker calponin was slightly reduced in the vascular wall of miR-143/145 KO mice in situ (Fig. 5, L and M). Using GFP-tagged MKL1, we found that HG caused increased nuclear localization (Fig. 5N, MKL1 (green) and nuclei (red)). This was not seen in KO mice, arguing that impaired actin regulation by HG in KO mice associates with reduced MKL1 activation.
Calcium influx via L-type calcium channels is an important regulator of Rho activity and expression of contractile smooth muscle markers (39, 40). We therefore analyzed calcium levels in cells under high and low glucose conditions and found that baseline fluorescence of the Fluo-4 calcium indicator was significantly increased in cells subjected to hyperglycemia (Fig. 6A). However, although the plateau response tended to be increased in HG versus LG conditions, we did not observe a statistically significant difference in either peak (Fig. 6B) or plateau responses (Fig. 6C) to depolarization with high KCl.
To test the importance of L-type channels, we used the calcium channel inhibitor verapamil, which significantly decreased glucose-induced smooth muscle marker expression (Fig. 6, D–F). Incubation of smooth muscle cells in low glucose with the KATP channel inhibitor glibenclamide resulted in the induction of dystrophin mRNA expression to a similar level as with high glucose (Fig. 6G). A similar trend was observed with the expression of calponin and desmin, but these changes did not reach statistical significance (data not shown). The effect of KATP channel activation in high glucose conditions was then tested using the KATP channel opener cromakalim. However, cromakalim had no effect on the glucose-induced mRNA expression of dystrophin (Fig. 6G).
Activation of the PKC/Rho pathway by hyperglycemia can be mediated by glycated and oxidized proteins and lipids called AGEs. To determine the importance of this process for glucose-mediated activation of Rho signaling and expression of smooth muscle markers, we treated smooth muscle cells incubated under high and low glucose conditions with aminoguanidine, an inhibitor of AGE formation. Although aminoguanidine hydrochloride had no significant effect in low glucose conditions, it completely prevented the increased expression of contractile smooth muscle markers in high glucose medium (Fig. 7, A–D). Furthermore, this effect correlated with a reduced phosphorylation of mediators in the Rho-signaling pathway, including MYPT, ERM, and LIMK (Fig. 7, E–G).
To confirm that glucose can promote smooth muscle differentiation in vivo, where baseline expression of smooth muscle markers is already relatively high, we used hyperglycemic Akita mice and analyzed the smooth muscle marker expression 8 weeks after the onset of diabetes. As shown in Fig. 8A, mRNA expression of the three smooth muscle markers calponin, dystrophin, and desmin was significantly higher in hyperglycemic mice when compared with normoglycemic littermate control WT mice.
We next tested whether contractile function was altered in the Akita mice using isometric force measurements of aortas stimulated with the contractile agonist 5-HT. In accordance with previous observations in diabetic animal models (12, 41), we found an increased maximal contraction to 5-HT in Akita mice (Fig. 8B). This effect was not dependent on altered NO production because the nitric oxide inhibitor l-NG-nitroarginine methyl ester was present during the experiment. Approximately 75% of the contraction to 5-HT was mediated by Rho-kinase, as demonstrated by relaxation to the Rho-kinase inhibitor Y-27632 (Fig. 8C).
To determine whether type-2 diabetes is associated with increased expression of smooth muscle contractile markers, we collected left internal mammary arteries from diabetic (n = 6, 100% male) and non-diabetic patients (n = 6, 67% male) undergoing coronary artery bypass grafting. Their mean age was 68.5 ± 2.6 and 67.8 ± 6.9 years, for diabetic and non-diabetic, respectively. The gender imbalance reflects the predominance of males undergoing coronary artery bypass grafting and a greater prevalence of type 2 diabetes in male subjects. Other parameters, such as blood pressure, LDL cholesterol, and serum creatinine, were similar between the groups, whereas blood glucose levels were significantly increased in the diabetic subjects (Table 2). Elevated blood pressure was treated pharmacologically in both diabetic and non-diabetic patients. The expression of several smooth muscle markers at both the mRNA and protein levels was higher in the vessels from diabetic patients (Fig. 8, D, E, and G). Furthermore, a significant increase was found in the phosphorylation levels of ERM, LIMK, and cofilin, suggesting activation of the PKC and/or Rho/Rho-kinase pathway (Fig. 8, F and G). Collectively, these data suggest that diabetic conditions, possibly via increased extracellular glucose levels, can induce contractile differentiation of smooth muscle cells.
In the present study, we have demonstrated that PKC/Rho activation by glucose can promote actin polymerization and expression of contractile and cytoskeletal markers in cultured smooth muscle cells. We also found that this effect is dependent on miR-143/145 expression, PKC/Rho activation, and L-type calcium channel signaling. Interestingly, the increased contractile differentiation induced by hyperglycemia appears to be a general effect because the in vitro results could be largely recapitulated in hyperglycemic Akita mice and diabetic patients.
Extensive research has suggested that the vascular smooth muscle from diabetic patients and animal models acquires a hypercontractile phenotype (7,–10, 12, 42, 43). Similar effects have been observed in detrusor and airway smooth muscle, suggesting a general mechanism for some smooth muscle tissues (44, 45). In many cases, the cause of glucose-induced smooth muscle hypercontractility has been ascribed to an increased PKC/Rho-kinase activity (10, 13, 14, 43, 45, 46). Increasing levels of glucose cause an elevation in cytosolic calcium levels in vascular smooth muscle (10, 47), and both PKC and Rho activation are in part dependent on intracellular calcium levels (48, 49).
In agreement with previous studies, we found that elevated glucose levels increase phosphorylation of specific downstream targets of Rho and/or PKC, such as MYPT, ERM, LIMK, and cofilin, and caused a general increase in PKC substrate phosphorylation in vascular smooth muscle cells. Glucose-induced signaling events are maintained over several weeks in cell culture, which is an important consideration because this can then result in long lasting effects on protein expression and remodeling of vascular smooth muscle (18, 50). However, an increased PKC/Rho-kinase activity can also directly regulate smooth muscle contractility by inhibition of myosin phosphatase, thereby causing calcium sensitization (51, 52). Naturally, other signaling pathways are regulated by glucose in smooth muscle, and we cannot rule out their importance for the effects on contractile gene expression in smooth muscle.
High intracellular glucose levels can inhibit the activity of the AMPK-signaling pathway via an increased ATP/AMP ratio. For this to occur, it requires an increased glucose uptake and glucose metabolism in the cells. Because smooth muscle cells mainly express insulin-insensitive GLUT1 transporter (53), glucose transport is limited by the amount of GLUT1 in the cell membrane. Interestingly, elevated glucose levels resulted in decreased uptake of glucose after 1 week in culture. This is in accordance with previous observations and can in part be explained by a decreased GLUT1 expression (34, 35). However, the decrease in GLUT1 expression is not sufficient to normalize intracellular glucose levels in smooth muscle cells subjected to hyperglycemia (34). To confirm an increased glucose metabolism under hyperglycemic conditions in our model, we analyzed the regulation of AMPK/acetyl-CoA carboxylase signaling and found a significant decrease in AMPK activation under hyperglycemic conditions. However, in both cultured cells and intact blood vessels, activation of AMPK by AICAR results in a significant induction of smooth muscle marker expression (29). Thus, our previous data indicate that the effect of glucose on AMPK would rather counteract glucose-induced expression of contractile smooth muscle markers.
Downstream of Rho/PKC, glucose stimulation increases the F/G-actin ratio in cultured smooth muscle cells, which suggests that actin polymerization is stimulated by hyperglycemia. This may in part be due to a glucose-induced phosphorylation of cofilin. Cofilin belongs to a family of actin-depolymerizing factors and is inactivated by phosphorylation by LIM kinase, leading to increased accumulation of actin filaments (54). Although slower than calcium sensitization, this effect can contribute to acute smooth muscle contractility following Rho activation (55, 56). In accordance with our observation in smooth muscle, F-actin remodeling has been reported to be involved in glucose-induced insulin secretion in pancreatic β-cells, partly via inhibition of cofilin (57).
Concurring with an increased Rho activation and actin polymerization, we found that glucose promotes the expression of contractile smooth muscle markers at both mRNA and protein levels. If functional contractile filaments are formed from these proteins, this can potentially lead to a sustained augmentation of vascular contractile function and may be an additional mechanism for diabetes-associated hypertension.
The mechanism behind actin-dependent transcription involves nuclear translocation of the transcriptional co-activator MRTF, which is released from monomeric G-actin upon increased actin polymerization (19). MRTF binds to SRF, and the resulting ternary complex activates genes that play roles in cytoskeletal organization and contractility (58). SRF binds to genetic elements referred to as CArG boxes, and SRF binding to these genetic elements is inhibited by Klf4 (20). Here we found that Klf4 is repressed by glucose at the mRNA level, arguing that SRF-dependent transcription is not only activated via actin polymerization but also relieved from Klf4-mediated repression. This dual regulation of SRF-dependent transcription is a plausible reason why SRF has such a large, if not dominating, impact on the glucose-dependent transcriptome in smooth muscle. Further work is required to elucidate the mechanism of repression of Klf4 by glucose and its importance in vascular disease. However, it is notable that conditional KO of Klf4 in smooth muscle accelerates neointima formation while maintaining SMC marker expression (59).
A controversial aspect of this study is that diabetes and hyperglycemia are known to exacerbate vascular diseases associated with a synthetic vascular smooth muscle phenotype (i.e. atherosclerosis or restenosis after balloon angioplasty). However, the contractile and synthetic phenotypes of smooth muscle cells are not mutually exclusive because they are regulated by separate intracellular mechanisms (59,–61). Thus, pathways leading to increased expression of both contractile and synthetic markers can be activated simultaneously, which seems in line with our finding that osteopontin is simultaneously up-regulated with the smooth muscle differentiation markers. Osteopontin is a marker of the synthetic phenotype of smooth muscle, which has been shown to be up-regulated in the context of vascular disease, including diabetic vascular complications. In agreement with our results, previous reports have demonstrated that osteopontin is induced by hyperglycemia in smooth muscle cells. Several mechanisms have been proposed, including activation of the PKC/Rho-kinase pathway (62) and activation of the transcription factor NFATc3 (63, 64). Thus, hyperglycemia in diabetic patients may result in simultaneous activation of the contractile and synthetic gene programs, leading to increased risk of neointimal hyperplasia as well as hypertension and vasospasm. The extent to which these gene programs may be activated may also vary depending on the vascular bed (i.e. a synthetic program dominance in regions prone to atherosclerosis or a contractile dominance in resistance arteries).
We have previously demonstrated that stretch-induced contractile differentiation in vascular smooth muscle is mediated by increased actin polymerization, which is dependent on L-type calcium channel activation and small non-coding RNAs, such as miR-145 (23,–26, 40, 65). The regulation of L-type calcium channel expression by miR-145 is in part mediated by direct interaction of miR-145 with its target CamKIIδ (26, 66), which in turn affects the translocation of a transcriptional repressor called DREAM (67). In earlier studies, we found that knockdown of miR-145 or genetic deletion of the miR-143/145 cluster results in increased CamKIIδ expression and reduced expression of L-type calcium channels (26, 31, 65). Considering the sensitivity to L-type calcium channel verapamil, it is likely that reduced L-type calcium channel expression is involved in the diminished glucose-mediated transcription of contractile smooth muscle markers in miR-143/145 KO cells.
It is known that L-type calcium influx can promote (40, 69) Rho signaling and contractile gene expression in smooth muscle. Recent reports suggest that hyperglycemia and diabetes can promote persistent calcium sparklet activity via activation of L-type calcium channels in arterial smooth muscle (70, 71). Interestingly, sparklet activity is controlled by PKC and underlies a sustained calcium entry through persistent L-type channel activity (72). In accordance with these findings, we found an increased basal Fluo-4 fluorescence following hyperglycemia, which is an indicator of elevated baseline calcium levels in these cells. Relative to baseline levels, the voltage-activated calcium influx was not significantly elevated in hyperglycemic conditions. However, it is likely that the slow and persistent effects that we observe in smooth muscle cells during hyperglycemia require persistent elevations of intracellular calcium levels. This is further supported by our finding that inhibition of L-type calcium channels by verapamil can inhibit glucose-induced smooth muscle differentiation.
The activation of Rho by calcium can be mediated via PKC and/or the calcium-sensitive proline-rich tyrosine kinase 2 (39, 73). In addition to Rho/Rho-kinase, mechanical stress has been demonstrated to activate PKC (74). Thus, glucose and mechanical stimuli share many of the signaling mechanisms leading to increased smooth muscle differentiation. Accordingly, we found that inhibition of PKC/Rho/MRTF signaling and genetic KO of the miR-143/145 cluster reduced glucose-induced contractile gene expression. We cannot rule out some off-target effects of the inhibitors used in this study. However, because siRNAs against Rho-kinases or MRTFs had similar effects as the inhibitors on glucose-induced calponin expression, our results strongly suggest that the PKC/Rho-MRTF pathway is involved in this process.
The most proximal effects of glucose are not fully understood but are thought to involve the formation of AGEs. These glycated and oxidized lipids and proteins can be formed both intra- and extracellularly and interact with AGE receptors (reviewed in Ref. 75). In vascular cells, this interaction results in the activation of a number of intracellular pathways, including PKC/Rho (16, 17, 76). The prototype therapeutic agent aminoguanidine is known to prevent AGE formation by reacting with derivatives of early glycation products, such as 3-deoxyglucosone (reviewed in Ref. 77). Herein, we found that aminoguanidine inhibited activation of Rho-signaling and smooth muscle marker expression induced by glucose but had no effect in low glucose conditions. Our results thus support a major role of AGEs in glucose-induced contractile differentiation.
Although AGE formation is a plausible explanation for the effects of glucose on smooth muscle, we also tested other possible explanations, such as an increased glucose metabolism, which in turn can elevate the levels of ATP in the cells. In pancreatic β-cells, it is well known that glucose-induced calcium influx via voltage-gated calcium channels is activated by inhibition of KATP channels, which is due to an increased ATP production following glucose metabolism. To test if a similar mechanism is possible in smooth muscle cells, we used the KATP channel inhibitor glibenclamide and the KATP channel opener cromakalim in the presence of low and high glucose, respectively. Glibenclamide mimicked the effects of high glucose on contractile marker expression, suggesting that inhibition of KATP channels is sufficient to induce expression of contractile proteins. However, the lack of any effect of cromakalim in high glucose conditions argues against an important role of KATP channels in the effects of high glucose on contractile gene expression.
Isolation and culture of vascular smooth muscle cells rapidly results in a phenotypic shift of the cells, which may affect the response observed to a wide variety of stimuli. We thus found it important to test whether some of our findings could be recapitulated in vivo, where baseline expression of contractile smooth muscle markers is severalfold higher compared with the in vitro situation. The increased expression of differentiation markers in Akita mice and diabetic patients points toward a general mechanism that may, at least in part, contribute to the hypercontractile phenotype of smooth muscle cells in diabetic conditions. This is in line with previous work that demonstrated increased contractile responses in the vasculature from diabetic patients and hyperglycemic Akita mice (7, 78). Our results support these previous observations by demonstrating increased smooth muscle contractions to the vasoactive agonist 5-HT in Akita mice. An increased contraction to 5-HT has been shown in other diabetic animal models and is known to depend on increased activation of the Rho-signaling pathway (12, 14, 41). However, increased expression of smooth muscle markers does not necessarily result in increased contractility, and the increased contractile responses in diabetic animal models can also be mediated by myosin phosphatase inhibition and calcium sensitization. Further studies are thus needed to dissect the exact contribution of glucose-induced transcriptional regulation of contractile genes to force development in smooth muscle in vivo.
In diabetic patients, we found that the increase in smooth muscle marker expression correlated with an increased activation of the Rho signaling pathway. This supports the idea that activation of Rho-kinase may be a contributing factor in the development of diabetic vascular disease and that inhibitors of this pathway can be beneficial against cardiovascular complications of diabetes (68, 79, 80). However, the mechanism of Rho activation in smooth muscle may differ in vivo compared with the in vitro situation. In vivo, the influence of endothelial cells cannot be ruled out, and it is thus possible that endothelial dysfunction can contribute to the effects of diabetes that we observe in smooth muscle. Another limitation of this study is the low sample number in the patient study. A larger cohort is needed to verify the changes observed herein and to limit the effects of confounding factors. However, despite a low n value, we were able to detect several significant changes in diabetic patients compared with non-diabetic controls. The low variability between samples can be partly explained by the fact that the patients were relatively well matched regarding age, sex, and blood pressure.
In conclusion, the results of the present study provide evidence for a role of glucose in contractile gene expression in smooth muscle via activation of AGE-, PKC-, and Rho-dependent signaling pathways, actin polymerization, and Klf4 repression. Altogether, these data suggest a novel possible mechanism for the development of diabetic vascular diseases and provide further support for the potential therapeutic use of Rho-kinase inhibitors in these conditions.
All authors critically revised the manuscript and approved its submission. T. T. H., K. Swärd, M. F. G., K. Stenkula, and S. A. contributed to the conception and design of the experiments, drafting the manuscript, and interpretation of data. T. T. H., K. M. T., D. D., M. E., M. G., J. S., J. N., T. Braun, T. Boettger, K. Stenkula, K. Swärd, S. A., and E. G. V. contributed to the acquisition and analysis of data.
We acknowledge the Swegene Center for Integrative Biology at Lund University and Dr. S. Veerla for carrying out the microarray analyses.
*This work was supported by grants from the Novo Nordisk Foundation, the Per-Eric and Ulla Schyberg Foundation, and the Albert Påhlsson Foundation (to S. A.). This work was also supported by Swedish Heart and Lung Foundation Grant HLF 20130700, Swedish Research Council Grant 2014-3352, and Swedish Diabetes Association (Diabetesfonden) (to M. F. G. and K. Stenkula). This work was also supported by Innovative Medicines Initiative Joint Undertaking 115006, comprising funds from the European Union's Seventh Framework Programme (FP7/2007–2013) and EFPIA companies' in kind contribution. The authors declare that they have no conflicts of interest with the contents of this article.
2The abbreviations used are: