MicroRNA‐92a promotes vascular smooth muscle cell proliferation and migration through the ROCK/MLCK signalling pathway

Abstract To identify the interaction between known regulators of atherosclerosis, microRNA‐92a (miR‐92a), Rho‐associated coiled‐coil‐forming kinase (ROCK) and myosin light chain kinase (MLCK), we examined their expressions during proliferation and migration of platelet‐derived growth factor‐BB (PDGF‐BB)‐regulated vascular smooth muscle cells (VSMCs), both in vivo and in vitro. During the formation of atherosclerosis plaque in mice, a parallel increase in expression levels of MLCK and miR‐92a was observed while miR‐92a expression was reduced in ML‐7 (an inhibitor of MLCK) treated mice and in MLCK‐deficient VSMCs. In vitro results indicated that both MLCK and miR‐92a shared the same signalling pathway. Transfection of miR‐92a mimic partially restored the effect of MLCK's deficiency and antagonized the effect of Y27632 (an inhibitor of ROCK) on the down‐regulation of VSMCs activities. ML‐7 increased the expression of Kruppel‐like factor 4 (KLF4, a target of miR‐92a), and siRNA‐KLF4 increased VSMCs' activity level. Consistently, inhibition of either MLCK or ROCK enhanced the KLF4 expression. Moreover, we observed that ROCK/MLCK up‐regulated miR‐92a expression in VSMCs through signal transducer and activator of transcription 3 (STAT3) activation. In conclusion, the activation of ROCK/STAT3 and/or MLCK/STAT3 may up‐regulate miR‐92a expression, which subsequently inhibits KLF4 expression and promotes PDGF‐BB‐mediated proliferation and migration of VSMCs. This new downstream node in the ROCK/MLCK signalling pathway may offer a potential intervention target for treatment of atherosclerosis.

and migration, apoptosis, matrix degradation, oxidative stress and inflammation. Although researches have identified some key signalling and molecular regulatory pathways involved in the initiation and progression of AS plaques, the pathophysiological mechanisms of AS have yet to be illuminated, 1,2 hence the prevention and treatment options for AS remain limited.
Endothelial cells (ECs) and VSMCs are the main cell types of within the vasculature and closely related in structure and function.
ECs that cover the interior surface of blood play an important role in the regulation of the vascular tone by releasing vasoactive agents controlling VSMCs proliferation or migration. [3][4][5] VSMCs that provide structural integrity to the vessel wall are fine-tuned by adjacent ECs. 6,7 Aberrant proliferation and migration of VSMCs are the most studied key pathological processes in the initiation and development of AS. [8][9][10][11] Among the various factors associated with the development of AS, the high expression of platelet-derived growth factor-BB (PDGF-BB) has been detected in nearly all cell types of the atherosclerotic aortic wall and in the infiltrating inflammatory cells. 12 PDGF-BB is a known potent mitogen and chemoattractant for VSMCs and is found in atherosclerotic lesions. 13 Moreover, PDGF-BB can activate Rho-associated coiled-coil-forming kinase (ROCK) and myosin light chain kinase (MLCK), [14][15][16][17] both of which regulate phosphorylation of myosin light chain (MLC). 18 The phosphorylation of MLC promotes the cell contraction and cell motility thereby leading changes in actin cytoskeleton. 18,19 The rearrangement of the actin cytoskeleton, in turn, may greatly influence inflammatory signalling. 20 Therefore, blocking the PDGF-BB-induced ROCK/MLCK signalling pathway could potentially prevent the dysregulation of VSMCs, and consequently attenuate the progression of AS. Exploring the novel regulatory mechanisms of the PGDF-BB signalling pathway could be of great scientific and therapeutic interest for AS.
MicroRNAs (miRNAs) are evolutionarily conserved, non-coding small RNAs that can regulate gene expression at post-transcriptional level, which means that one miRNA usually targets 3′-untranslated regions of various mRNAs that are involved in different steps of one precise metabolic/signalling pathway. Therefore, changes in the levels of one key miRNA affect various steps of one pathway, which is thereby promoted or inhibited.
This makes miRNAs potent future diagnostic and even therapeutic tools for personalized medicine. 21 Recent findings have revealed a key role for miRNAs in the pathophysiological processes of cardiovascular disease, such as miR-126, miR-146, miR-143/145 and others, have been identified as relevant mediators by modulating ECs and VSMCs function in angiogenesis, AS and in-stent restenosis, 1,5 miR-27a/b, miR-33, miR-122, miR-144 or miR-223 involved in lipid metabolism. 21,22 miR-92a, a member of the miR-17-92 cluster, is highly expressed in ECs of blood vessel walls. [23][24][25] The role of miR-92a in the development of AS in vivo has been welldocumented. 23,26 Specifically, miR-92a is highly expressed in athero-prone areas of the aortic arch compared with athero-resistant regions. 25,27 Up-regulation of miR-92a by oxidized low-density lipoproteins (oxLDL), present in athero-prone areas, enhances endothelial activation and atherosclerotic lesions' progression. 23 As both the PDGF-BB-induced ROCK/MLCK signalling pathway and miR-92a-mediated post-transcriptional effects are important aspects of the AS lesion formation, it is of interests to identify any potential connection between the two pathways. In this study, we observed that the expression levels of MLCK and miR-92a were significantly increased in parallel during atherosclerotic plaque formation in mice. The inhibition of MLCK with its inhibitor ML-7 reduced lipid deposition lesions, as well as miR-92a expression, which was also found decrease in MLCK-deficient VSMCs. We therefore highly speculated that MLCK could be involved in the regulation of miR-92a in VSMCs, and tested our hypothesis in an in vitro model. Our attempt to unveil the complicated signalling network in AS lesion progression may provide clues in the development of novel clinical biomarkers or therapeutic targets.

| AS model
Mice were randomly divided into control group (n = 50, standard chow fed) and high-fat diet fed group (n = 50, high-fat diet consisting of 78% common chow, 10% lard oil, 10% yolk powder, 1% cholesterol and 0.2% bile salt from pig). The mice from control and high-fat fed group at 6, 9, 12, 15 and 18 weeks were dissected to examine the extent of AS.

| ML-7 model
The AS mice were randomly divided into the untreated group (n = 16) and the ML-7 treatment group (n = 16). All mice in both groups were fed with a high-fat diet. ML-7 group were treated with ML-7 by injecting into the veins of the tails twice a week at a dose of 1 mg/ kg from 6 weeks. The mice of untreated group were injected with the same volume of saline as a control. The body weights of all mice were measured at 6, 9, 12, 15 and 18 weeks. All mice at 18 weeks of age were dissected. Thoracic aorta was excised, snap frozen in liquid nitrogen and stored at −80°C for subsequent RT-qPCR analysis.

| Serum lipid analysis
Serum total cholesterol (T-CHO), triglyceride (TG), High-density lipoprotein (HDL) and low-density lipoprotein (LDL) measurements were performed at 6, 9, 12, 15 and 18 weeks. Whole blood was obtained by retro-orbital bleeding and centrifuged at 2000 g for 15 minutes at 4°C. Blood lipid analyses were measured using the commercial kits (Nanjing Jiancheng Bioengineering Institute, China) according to the manufacturers' instructions.

| Haematoxylin and eosin (H&E) staining
Thoracic aortas were fixed with 4% paraformaldehyde, embedded in paraffin and sliced into 4 μm sections. The sections were baked at 70°C for 4 hours, dewaxed, hydrated in distilled water, stained with haematoxylin for 1 minute, differentiated in hydrochloric acid alcohol, blued in ammonia water, counter-stained with eosin (7 seconds), dehydrated with ethanol, transparentized with xylene I and xylene II, and finally mounted in neutral gum.

| Immunofluorescence staining
Paraffin-sectioned slides from thoracic aortas tissues were used. First, the slides were deparaffinized and rehydrated by dimethylbenzene and ethanol. Antigen retrieval was performed by incubating slides in 0.01 mol/L citrate buffer (pH 6.0) at 95°C for 20 minutes. The samples were then blocked for 30 minutes, followed by an overnight incubation with the KLF4 antibodies (1:100). Next, the slides were rinsed with PBS and incubated with the Rhodamine conjugated goat anti-rabbit IgG (H+L) (Thermo) for 30 minutes at 37°C. After being rinsed with PBS, the samples were incubated with FITC-conjugated α-smooth muscle actin antibody (1:100) for 40 minutes at 37°C. Finally, cell nuclei were counter-stained with DAPI. Digital images were captured with a fluorescence microscope (BX-51, TR32000; Olympus, Tokyo, Japan).
A7r5 or Gba cells were fixed in 4% formaldehyde, permeabilized with 0.1% Triton X-100 for 10 minutes and blocked with 5% BSA for 20 minutes. The cells were incubated with the rabbit anti-human KLF4 antibodies overnight at 4°C and washed, followed by a 1 hour incubation with the appropriate antibodies at room temperature, including Rhodamine conjugated goat anti-rabbit IgG (H+L) (Thermo) and FITC-conjugated F-action α-smooth muscle actin antibody (1:100). Nuclei were counter-stained with DAPI and then observed under a fluorescence microscope.

| Oil-Red-O staining
Thoracic aorta was washed with PBS and fixed with 78% methyl alcohol twice after the removal of aortic peripheral adipose tissue. The staining method is as follows: the fixed thoracic aorta samples were rinsed with 78% methyl alcohol for 5 minutes and stained in 0.5% Oil-Red-O solution for 1 hour. Thoracic aorta was differentiated in a 78% methyl alcohol solution for 5 minutes, and then was sliced longitudinally to expose the intimal surface. The stained thoracic aorta was spread on a black charpie for photographing using a digital camera under identical light conditions.

| Isolation and culture of primary rat aortic SMCs
Sprague-Dawley rats (250-300 g, from the animal experiment center of Dalian Medical University) were killed by diethyl ether.
Thoracic aorta was dissected to remove adhering periadventitial tissue and the endothelium was denuded with a catheter. After removing the adventitial layer, the remaining medial layer was minced into small pieces for digestion with Collagenase I (Catalog No. 17100-017, Gibco, Langley, OK) for 5 hours at 37°C. Then the small pieces of aorta were digested with 0.125% trypsin (Gibco) for 10 minutes at 37°C. Following the removal of digestion solution and re-suspending in 10% FBS F-12/DMEM medium, cells were gently transferred into culture dishes and incubated at 37°C. Every batch of VSMCs was tested by smooth muscle marker α-smooth muscle actin staining to ensure the purity of primary VSMCs to be above 95%.

| Cell culture
A7r5 is originally derived from the embryonic rat aorta and were pur-  Table 1.

| CCK-8 assay
Cells were transferred into 96-well plates at 24 hours after transfection. The medium was changed to serum-free medium with or without 10 ng/mL PDGF-BB for 24 hours. Ten-microlitre of CCK-8 solution (TransGen Biotech) was added to each well and absorbance was measured at 450 nm. The results are presented as a percentage of cell proliferation (the optical density [OD] of the experiment samples/control group).

| Wound-healing assay
Cells were seeded in 6-well plates and cultured until 90% confluence after 24 hours of transfection. Then cell layer were scratched with a 200-μL sterile pipette tip. The cells were washed three times with serum-free media to remove detached cells. After 10 hours incubation, cell migration was viewed and photographed using an optical microscope (BX-51, TR32000; Olympus).  Transfected cells were used at 24 or 48 hours.

| Statistical analysis
Statistical analysis was performed with GraphPad Prism 6 and presented as mean ± SD from at least three independent TA B L E 1 Primers used in qRT-PCR

| MLCK and miR-92a are involved in the formation of AS plaque
The development of AS is a dynamic process in which key signalling and molecular regulatory pathways are involved in the initiation and progression of AS plaques. 28 We generated an AS model using ApoE −/− mouse 29 and monitored the plasma lipid levels and the changes of aortic walls in mice during AS formation every 3 weeks from 6 weeks of age. The levels of triglyceride (TG), total cholesterol (T-CHO) and low-density lipoprotein (LDL) were significantly increased at 9 weeks, and reached its peak threshold at 9 and 12 weeks. HDL significantly decreased and reached its lowest point at 15 weeks ( Figure S1A-D). Histologically, the thickening of blood vessel wall was observed at 15 weeks whereas clear plaque was generated at 18 weeks ( Figure S1E). During the time-course of AS formation, the expression levels of MLCK and miR-92a were significantly increased at 9 and 12 weeks, and both reached its peak at 15 weeks (P < 0.01, Figure 1A,B), in line with the thickening of blood vessel wall, suggesting that both MLCK and miR-92a are involved in the formation of AS.

| Inhibition of MLCK activity down-regulated miR-92a expression and reduced lipid deposition lesions in AS mice
This parallel increase in both the expression of MLCK and miR-92a indicates that there is a possible connection between MLCK and miR-92a in AS formation process. To verify this possibility, we performed a series of experimental procedures, both in vivo and in vitro.
We first inhibited MLCK activity with its specific inhibitor ML-7.
ML-7 was injected to AS mice via tail vein twice per week. No significant difference of body weight between ML-7-treated and untreated mice were found ( Figure S1F). However, after ML-7 treatment, the level of LDL was significantly decreased compared with untreated mice (P < 0.05, Figure S1G), and aortic lipid deposition lesions were significantly reduced as well (P < 0.05, Figure 1C). Strikingly, miR-92a expression in aortic wall also declined significantly at the same time (P < 0.05, Figure 1D), suggesting that miR-92a may share the same pathway with MLCK. In line with these in vivo data, the expression of miR-92a was also found to be down-regulated in ML-7 treated human aortic smooth muscle cells (HASMCs) (P < 0.01, Figure 1E).

| Absence of MLCK gene in VSMCs leads to reduced expression of miR-92a
To further explore the connection between MLCK and miR-92a, we examined miR-92a expression in a guinea pig basilar artery smooth muscle cell line GbaSM-4 (wild-type) and its MLCK-deficient format (MLCK − /Gba). The expression level of MLCK in MLCK − /Gba was significantly down-regulated when compared to GbaSM-4 (P < 0.01, Figure S2). miR-92a was identified as the most dysregulated miRNA in GbaSM-4 vs MLCK − /Gba by Volcano Plot analysis ( Figure 1F). Next, we validated the microarray data by RT-qPCR.
The expression level of miR-92a was significantly lower in MLCK − / Gba compared to that of wild-type GbaSM-4 (P < 0.01, Figure 1G).
In contrast, after up-or down-regulating the miR-92a expression in VSMCs with miR-92a mimic or miR-92a inhibitor, respectively, MLCK expression remained at similar levels ( Figure 1H). These findings suggest that MLCK could be the upstream modulator of miR-92a in VSMCs.

| MLCK and miR-92a both regulate the migration and proliferation of VSMCs
It is well-known that dysregulated proliferation and migration of VSMCs in response to environmental stimuli play key roles in the development of AS. 8,9 PDGF-BB was originally identified as the platelet and serum mitogen for regulating proliferation and migration of VSMCs. 12,30 To further determine the relationship between MLCK and miR-92a, we studied their roles in proliferation and migration of PDGF-BB stimulated VSMCs. The expression of MLCK and miR-92a was elevated by the stimulation of VSMCs with PDGF-BB ( Figure S3A-C). The role of MLCK was then studied in this system by inhibiting either its function or expression.
The proliferation of HASMCs was inhibited by ML-7 at a concentration ranging from 10 to 100 μmol/L in a dose-dependent manner ( Figure 2A). Meanwhile, ML-7 also inhibited the migration of HASMCs (P < 0.01, Figure 2B). Compared to wild-type GbaSM-4 cells, the proliferation and migration of MLCK − /Gba cells were significantly decreased ( Figure 2C,D). To study the role of miR-92a in this system, we altered miR-92a expression with miR-92a inhibitor fragments in HASMCs (P < 0.01, Figure S4A). The results indicated that miR-92a inhibitor also suppressed the proliferation and migration of HASMCs regardless of the PDGF-BB presence ( Figure 2E,F).

| miR-92a mimic partially restored the MLCK deficiency induced impairment of proliferation and migration of VSMCs
To further explore the connection between MLCK and miR-92a in VSMCs, we transfected miR-92a inhibitor or miR-92a mimic into GbaSM-4 and MLCK − /Gba cells separately ( Figure S4B,C). In
Transfection of miR-92a inhibitor almost completely abolished the migration of GbaSM-4 cells through Boyden Chamber assay (P < 0.01) while miR-92a mimic transfection dramatically promoted the migration of MLCK − /Gba cells (P < 0.01, Figure 3C). These results indicated that both MLCK and miR-92a could be components of the same signalling pathway regulating proliferation and migration of VSMCs with the MLCK likely being an upstream regulator of miR-92a.

| Changes in miR-92a expression alter the inhibition of ROCK induced down-regulation of cell proliferation and migration
It is known that PDGF-BB induced VSMCs dysfunction is crucially dependent upon the ROCK/MLCK signalling pathway. 14-16 As our results identified MLCK as an upstream regulator of miR-92a, it is of interest to find out whether ROCK possesses similar role as MLCK in regulating miR-92a expression in VSMCs. We found that transfection with miR-92a mimic or miR-92a inhibitor did not alter the ROCK expression at the transcriptional level in A7r5 cells ( Figure   S5A). The inhibition of ROCK activity with its inhibitor Y27632 led to decreased expression of miR-92a in A7r5 cells (P < 0.01, Figure S5B) with subsequent reduction in A7r5 cell proliferation ( Figure 4A,B) and reduced migration of HASMCs ( Figure 4C). However, transfection of miR-92a mimic showed no effect on proliferation of PDGF-BB treated A7r5 cells, although it significantly antagonized the inhibitory effect of Y27632 on proliferation of A7r5 cells (P < 0.001, Figure 4A) and on migration of HASMCs (P < 0.05, Figure 4C).
Transfection of miR-92a inhibitor did not further decrease the proliferation of Y27632-treated A7r5 cells ( Figure 4B). Nevertheless, the μm. E, The proliferation rates of HASMCs were measured by CCK8 assay with or without 10 ng/mL PDGF-BB after transfected. F, The cell migration was assessed by Boyden Chamber assay with or without 10 ng/mL PDGF-BB after transfected. Migrated cells in each high-power field (HPF, 400×) were quantitated and the results are shown on the right (n = 5). Data are presented as mean ± SD from at least three independent experiments. *P < 0.05, **P < 0.01 transfection of miR-92a inhibitor did significantly further inhibit the migration of HASMCs treated with Y27632 (P < 0.001, Figure 4C). We founded that, consistent with the result from oil-red staining ( Figure 1C), there was no significant plaque formation in the aorta of ML-7-treated mice. Interestingly, the KLF4 protein staining was enhanced with ML-7-treated mice compared to that of AS mice, and the KLF4-positive cells were mainly distributed in the SMCs from the blood vessel walls ( Figure 5A). On the other hand, in our in vitro system, transfection of siRNA-KLF4 increased the proliferation (P < 0.001) and migration (P < 0.01) of rat primary aortic SMCs ( Figure 5B,C). Migration of A7r5 cells was also significantly increased by blocking KLF4 expression with siRNA (P < 0.001, Figure 5D). siRNA-KLF4 could partially rescue the effects of inhibitor miR-92a on PDGF-BB-mediated proliferation and migration of HASMCs ( Figure 5E,F).

| Expression of KLF4 was increased in MLCK − and ROCK − VSMCs
Next, we examined the effect of MLCK and ROCK on KLF4 expression. Western blot analyses showed that both miR-92a inhibitor transfection (P < 0.01) and Y27632 treatment (P < 0.05) increased KLF4 expression ( Figure 6A,B). In MLCK − /Gba cells and siRNA-ROCK transfected A7r5 cells, immunofluorescence labeling results showed that KLF4 protein staining was enhanced ( Figure 6C,D). Data are presented as mean ± SD from at least three independent experiments. *P < 0.05, **P < 0.01

| ROCK/MLCK up-regulated miR-92a expression in VSMCs through STAT3 activation
Given the noticeable data that ROCK/MLCK up-regulated miR-92a expression in VSMCs, we sought to further explore the mechanistic relationship between ROCK/MLCK and miR-92a expression.
Because the promoter region of the miR-92a gene contains a con-

| D ISCUSS I ON
MicroRNAs have been proven to regulate a wide range of biological processes, some of which are associated with AS. 33 Loyer et al clearly demonstrated that in vivo inhibition of miR-92a in ldlr −/− mice restricted the development of AS. 23 We found that miR-92a   were mediated through a Sp1-dependent mechanism that involves the direct binding of Sp1 to the KLF4 promoter and requires three consensus Sp1 sites. 53 But the signal pathway provided in this paper is only one of the signal pathways that PDGF-BB regulates KLF4, that is the activation of ROCK and/or MLCK may up-regulate miR-92a expression, which subsequently inhibits the KLF4 expression and promotes PDGF-BB-mediated proliferation and migration of VSMCs.
The novelty of the current study rests on coupling the ROCK and MLCK pathways as well as finding the role of miR-92a in AS.
However, how are MLCK and ROCK as kinases involved in regulating the expression of miR-92a? We know that the control of epigenetic regulation, transcriptional regulation, post-transcriptional regulation and degradation level regulation, are the four major mechanisms of miRNAs expression. 54,55 A few papers showed that the transcription factor, STAT3, is involved in the regulation of miR-92a expression. STAT3 can up-regulate miR-92a to inhibit PTEN or RECK target gene to promote cholangiocarinoma growth or lung cancer cells invasiveness. 58,59 Herein, we showed that ROCK/MLCK up-regulated miR-92a expression in VSMCs through STAT3 activation. Thus, the activation of ROCK/STAT3 or MLCK/ STAT3 signalling may be an important factor for the induction of miR-92a expression in VSMCs.
In summary, we identified miR-92a as a critical regulator in F I G U R E 8 Schematic illustration of the proposed mechanism for the proliferation and migration of vascular smooth muscle cells (VSMCs). PDGF-BB promotes the proliferation and migration of VSMCs through ROCK/STAT3 or MLCK/STAT3 up-regulated microRNA-92a (miR-92a) which targets KLF4