Matrix stiffness determines the phenotype of vascular smooth muscle cell in vitro and in vivo: Role of DNA methyltransferase 1

Biomaterials 155 (2018) 203e216

Contents lists available at ScienceDirect

Biomaterials journal homepage: www.elsevier.com/locate/biomaterials

Matrix stiffness determines the phenotype of vascular smooth muscle cell in vitro and in vivo: Role of DNA methyltransferase 1 Si-An Xie a, Tao Zhang b, Jin Wang a, Feng Zhao c, Yun-Peng Zhang a, Wei-Juan Yao a, Sung Sik Hur d, Yi-Ting Yeh d, Wei Pang a, Li-Sha Zheng c, Yu-Bo Fan c, Wei Kong a, Xian Wang a, Jeng-Jiann Chiu e, Jing Zhou a, * a

Department of Physiology and Pathophysiology, School of Basic Medical Sciences, Peking University, Key Laboratory of Molecular Cardiovascular Sciences, Ministry of Education, Beijing 100191, PR China Department of Vascular Surgery, Peking University People's Hospital, Beijing 100044, PR China c School of Biological Science and Medical Engineering, Beihang University, Key Laboratory for Biomechanics and Mechanobiology of Ministry of Education, Beijing 100191, PR China d Department of Bioengineering and Institute of Engineering in Medicine, University of California, San Diego, La Jolla, CA 92093, USA e Institute of Cellular and System Medicine, National Health Research Institutes, Miaoli 350, Taiwan b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 24 July 2017 Received in revised form 23 October 2017 Accepted 21 November 2017 Available online 21 November 2017

Cells perceive the physical cues such as perturbations of extracellular matrix (ECM) stiffness, and translate these stimuli into biochemical signals controlling various aspects of cell behavior, which contribute to the physiological and pathological processes of multiple organs. In this study, we tested the hypothesis that during arterial stiffening, vascular smooth muscle cells (SMCs) sense the increase of ECM stiffness, which modulates the cellular phenotype through the regulation in DNA methyltransferases 1 (DNMT1) expression. Moreover, we hypothesized that the mechanisms involve intrinsic stiffening and deficiency in contractility of vascular SMCs. Substrate stiffening was mimicked in vitro with polyacrylamide gels. A contractile-to-synthetic phenotypic transition was induced by substrate stiffening in vascular SMCs through the down-regulation of DNMT1 expression. DNMT1 repression was also observed in the tunica media of mice aortas in an acute aortic injury model and a chronic kidney failure model, as well as in the tunica intima of human carotid arteries with calcified atherosclerotic lesions. DNMT1 inhibition facilitates arterial stiffening in vivo and promotes osteogenic transdifferentiation, calcification and cellular stiffening of vascular SMCs in vitro. These effects may be attributable, at least in part, to the role of DNMT1 in regulating the promoter activities of Transgelin (SM22a) and a-smooth muscle actin (SMA) and the functional contractility of SMCs. We conclude that DNMT1 is a critical regulator that negatively regulates arterial stiffening via maintaining the contractile phenotype of vascular SMCs. This research may facilitate elucidation of the complex crosstalk between vascular SMCs and their surrounding matrix in healthy and in pathological conditions and provide new insights into the implications for potential targeting of the phenotypic regulatory mechanisms in material-related therapeutic applications. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Arterial stiffening DNA methylation DNMT1 Smooth muscle cell calcification Cellular stiffening

1. Introduction Physical properties of the extracellular matrix (ECM) and mechanical forces are integral to tissue homeostasis. Cells have evolved sophisticated systems to perceive both their native

* Corresponding author. Department of Physiology and Pathophysiology, Basic Medical College of Peking University, Beijing 100191, PR China. E-mail address: [email protected] (J. Zhou). https://doi.org/10.1016/j.biomaterials.2017.11.033 0142-9612/© 2017 Elsevier Ltd. All rights reserved.

microenvironments and the material properties of artificial implants in terms of ECM stiffness, translate these stimuli into biochemical signals controlling various aspects of cell behavior, with the consequent modulation in the physiological and pathological processes of natural organs as well as in biocompatibility of artificial organs [1,2]. Matrix stiffness directs the lineage specification of embryonic and adult stem cells with pluripotency and self-renewing properties. For example, soft matrices that mimic brain are neurogenic, stiffer matrices that mimic muscle are

204

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

myogenic, and comparatively rigid matrices that mimic bone are osteogenic [3]. Matrix stiffness also acts solely or synergistically (with some biochemical factors) to affect the phenotypic modulation of highly-differentiated/mature cells that retain lineage plasticity and multipotential capability; that is, cell-fate control [4e6]. Vascular smooth muscle cells (SMCs) are able to undergo modulation in their phenotypic continuum, ranging from a differentiated contractile and quiescent state, to a dedifferentiated synthetic, secretory, and proliferative state [7]. These states differ in the expression of vascular SMC-restricted contractile marker genes and synthetic or osteoblastic marker genes, and cell behavior [8]. Several in vitro studies have shown that increase of substrate stiffness downregulates the expression of a-smooth muscle actin (SMA) in cultured vascular SMCs [9], promotes SMC migration [10] and proliferation [11], and exacerbates SMC response to plateletderived growth factor (PDGF) [12]. Such findings implicate that vascular SMC phenotype and behavior could be changed due to increased vessel stiffness during the development of vascular diseases. However, the underlying mechanisms of how matrix stiffness affects SMC phenotype and function are largely elusive. Arterial stiffening, occurring as a consequence of aging, hypertension, arteriosclerosis, atherosclerosis, diabetes, and chronic kidney failure, is featured with an increase in elastic modulus of the arteries [13e16]. The mean elastic modulus of thoracic aortas from Apolipoprotein E (ApoE)-deficient mice (which are susceptible to atherosclerosis), examined by atomic force microscope, reached up to 15 kPa in comparison with 5 kPa from the age-matched wildtype mice [15]; the elastic modulus of aortas from Lewis polycystic kidney rat, a rodent model of chronic kidney disease, was 56% higher compared with the wild-type Lewis rat [16]. The increasing of arterial stiffness is assumed to be mainly attributed to changes in both the composition and the organization of the ECM, i.e., deposition of excess ECM and calcium phosphate salts, fragmentation and loss of arterial elastin, and crosslinking of adjacent collagen fibers [17,18]. However, arterial stiffness may also increase with changes in phenotype and structural properties of vascular SMC. Emerging evidences have revealed that arterial stiffening is associated with increased vascular SMC proliferation, migration, apoptosis [19], and osteochondrogenic transformation [20], accompanied by changes in intrinsic mechanical properties of the cells [21,22]. An interplay exists between the phenotypic modulation of SMCs and the structure and content of ECM. The key molecules responsible for sensing arterial stiffness and mediating the SMC responses instructed by the cellular microenvironment remain incompletely understood. Covalent methylation of DNA cytosine occurs almost exclusively in the context of CpG dinucleotides and in most cases promotes transcriptional repression or sometimes results in activation of genes and noncoding genomic regions [23]. The importance of DNA methylation in maintaining normal development and biological functions is reflected by the development of many diseases due to hyper- or hypo-methylation of DNA with improper temporal or spatial regulations [24,25]. Maintenance of methylation patterns during cell replication is mediated by DNA methyltransferase 1 (DNMT1), which catalyzes the transfer of a methyl group from S-adenosyl methionine to hemi-methylated DNA [23]. Aberrant DNMT1expression and DNA methylation have been observed in cultured aortic SMCs in propathogenic conditions. For instance, recent data demonstrated a global DNA hypomethylation and a decrease of DNMT1 expression in aortic SMCs in proliferating and replicative aging [26]. Additional reports documented that treatment of aortic SMCs with 5-Aza-20 -deoxycytidine (5-Aza), a DNA methyltransferase inhibitor, increases the expression of genes related to osteogenesis and facilitates the inorganic phosphorusinduced mineralization of the cells [27]. These findings suggest a

contribution of smooth muscle DNMT1 dysfunction to arterial stiffening. We have previously showed that DNMT1 is a mechanosensitive molecule, as evidenced by its shear stress-regulated expression and activation in vascular endothelial cells in vivo and in vitro [28]. Thus, we wondered whether DNMT1 mediates the regulation of ECM stiffness on SMC phenotype and how DNMT1 is involved in arterial stiffening. In this study, we elucidated whether ECM stiffness modulates smooth muscle phenotype through DNMT1. Substrate stiffening is mimicked in vitro with polyacrylamide (PA) gels in which the concentration of bis-acrylamide crosslinking sets the elasticity, and adhesion is provided by coating the gels with fibronectin. We observed a contractile-to-synthetic phenotypic transition induced by substrate stiffening in vascular SMCs through a down-regulation of DNMT1 expression. By using an acute aortic injury mice model and a chronic kidney failure mice model, both of which show a phenotype of arterial stiffening, we provided the first evidence that DNMT1 is repressed in stiffening arteries. Data of DNMT1 expression in calcified atherosclerotic lesions of human carotid arteries are in line with the results obtained in the mice models. We also showed that DNMT1 inhibition facilitates arterial stiffening in vivo and promotes osteogenic transdifferentiation, calcification and cellular stiffening of vascular SMCs in vitro. Finally, we suggested a mechanism that DNMT1 responds to and regulates arterial stiffness, at least in part, by increasing the promoter activities of smooth muscle contractile genes, Transgelin (SM22a) and SMA, and by promoting SMC contractility. 2. Materials and methods 2.1. Cell culture Primary human umbilical artery SMCs (HUASMCs) at passages 3 to 8 were used for all experiments. HUASMCs were maintained in Nutrient Mixture F12 Ham Kaighn's Modification (F12K, SigmaAldrich) supplemented with 20% FBS (Gemini) and 10% SMC Growth Medium (Cell Applications). To inhibit DNMT1 expression and activity, the cells were either incubated in 5-aza-20 -deoxycytidine (5-Aza) in culture medium at a concentration of 10 mmoL/L for three days with daily replacement of the medium, or infected with recombinant adenovirus expressing shRNA specifically targeting human DNMT1 genes (ad-shDNMT1) three days before the experiments. To inhibit myosin ATPase and microtubule assembly, cells were incubated respectively with 2,3-Butanedione 2-monoxime (BDM, 10 mmol/L) or nocodazole (5 mmoL/L) in culture media for 5 h after being cultured on gels for 6 h. 2.2. Antibodies, reagents, plasmids and adenovirus Rabbit polyclonal antibody (pAb) against DNMT1 and rabbit pAb against fibronectin were purchased from ABclonal. Goat pAb against proliferating cell nuclear antigen (PCNA), rabbit pAb against Cyclin A, rabbit pAb against smooth muscle myosin heavy chain (SMMHC), rabbit pAb against SM22a were purchased from Santa Cruz Biotechnology. Rabbit pAb against RunX2 and rabbit pAb against bone morphogenetic protein 2 (BMP2) were from Abcam. Rabbit pAb against GAPDH was from Easybio. Mouse mAb against 5-methlcytosine (5-meC) was from Eurogentec. 5-aza-20 -deoxycytidine was from Aladdin. BDM was obtained from Abcam. Nocodazole was from Sigma. pcDNA3/Myc-DNMT1, in which the full-length cDNA for human DNMT1 was cloned into EcoRI and NotI sites of pcDNA3/Myc, was a gift from Arthur Riggs (Addgene plasmid # 36939). Ad-shDNMT1 and the control adenovirus expressing GFP (ad-GFP) were obtained from Vigene Biosciences.

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

2.3. PA gel manufacturing PA gels were prepared as previously described [29]. Briefly, 40% w/v acrylamide and 2% w/v bis-acrylamide stock solutions (Bio-Rad) were combined to prepare PA gel solution. Gels with different stiffness were obtained by varying the final concentrations of bis-acrylamide cross-linker (0.048% and 0.264%) for the corresponding stiffness of 2.61 ± 0.82 and 19.66 ± 1.19 kPa [29]. The mixtures containing acrylamide and bis-acrylamide were degassed for 20 min to remove oxygen from the solutions. To polymerize the mixtures, 4.5 ml of 10% w/v ammonium persulfate (Bio-Rad) and 3 ml of N,N,N9,N9-tetramethylethylenediamine (TEMED; Bio-Rad) were added to yield a final volume of 500 mL PA solution. A 50-70 mm-thick gel was cast, and the thickness was confirmed through imaging of embedded fluorescent beads (Invitrogen) in the gel by confocal microscope. To cross-link extracellular matrix proteins onto the gel surface, the gels were activated by exposing the heterobifunctional crosslinker sulfo-SANPAH (Pierce, 0.5 mg/ml in PBS) to UV 264 nm light and subsequently were coated with 400 ml of fibronectin (0.05 mg/ml, from human plasma, Corning), collagen Type I(0.1 mg/ml, from rat tail, Corning), or laminin (0.05 mg/mL, from Engelbreth-Holm-Swarm mouse tumor, Corning) for each gel at room temperature for 1 h.

205

and rehydrated in an ethanol series and antigen retrieval was performed in a steamer. For immunostaining of attached cells, the cells were fixed in 4% PFA and permeabilized with 0.5% Triton X-100 in PBS. Nonspecific binding was blocked by 5% BSA in PBS. The sections/cells were probed with primary antibody, washed and then probed with secondary antibody including Alexa Fluor 488conjugated donkey/goat anti-rabbit/-mouse IgG (1:500, ThermoFisher) or Alexa Fluor 555-conjugated donkey anti-rabbit/-goat IgG (1:500, ThermoFisher). F-Actin was staining with Rhodamine Phalloidin (Cytoskeleton). Nuclei were counterstained with DAPI. IHC staining were developed using DAB (Vector Laboratories) followed by Hematoxylin counterstaining (Sigma). After mounting, the slips were visualized by fluorescence microscopy. Verification of immunofluorescence cytochemistry of fibronectin, DNMT1, and 5-methylcytosine (5-meC) was conducted by using the respective isotype controls (Fig S1 and S2). For morphometric analysis, the cell adhesion area and the aspect ratio (the ratio of the cell's long axis to its short axis) were calculated by NIH ImageJ software. For quantification of fluorescence, the regions of cell nucleus or cytoplasm were manually determined and measurements of fluorescence intensity were conducted by using NIH ImageJ software. 2.7. Characterization of calcium deposition by Alizarin Red S staining

2.4. RNA isolation and quantitative RT-PCR RNA was extracted from cultured cells by using TRIzol reagent (Life Technologies) according to the manufacturer's instructions. Isolated RNAs were reversed-transcribed into complementary DNA with M-MLV RT system (Invitrogen) by using Oligo(dT) primers. Real-time PCR was performed with the 2X RealStar power SYBR Mixture (Genestar) by using the specific primer pairs (Table S1). Gene expressions were normalized against GAPDH. 2.5. Immuno-dot blot assays For immuno-dot blot of DNA, genomic DNA was purified from cells with QIAamp DNA mini kit (Qiagen) according to manufacturer's instructions. DNA was denatured with 0.4 M NaOH, 10 mM EDTA at 95  C for 10 min, and then neutralized with 2 M ammonium acetate (pH 7.0). Denatured DNA samples with an amount of 5.0 mg and 2.5 mg were spotted on a nitrocellulose membrane in a Bio-Dot SF apparatus (Bio-Rad). The membrane was baked at 80  C for 2 h. For Western blot, cells were lysed in the RIPA lysis buffer: 25 mM HEPES, pH 7.4, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 125 mM NaCl, 5 mM EDTA, 50 mM NaF, protease inhibitor cocktail tablets (Roche). Equal amounts of protein were separated on SDS-PAGE, transferred to nitrocellulose membranes. Non-specific binding was blocked in 5% skimmed milk in TBS containing 0.1% Tween 20, and then the membrane was incubated with 1:1000 dilution of 5-meC antibody or other antibodies overnight at 4  C, followed by its detection using donkey anti-rabbit/-mouse/-goat IgG (H&L) antibody IRDye 800/700 Conjugated (Rockland). Visualization was performed with an Odyssey Fluorescent Western blot imaging systems (LI-COR Biosciences). For the staining of total DNA, the blot membrane was hybridized with 0.02% methylene blue in 0.3 M sodium acetate (pH 5.2). 2.6. Immunohistochemistry and immunofluorescence Tissues were fixed in 4% paraformaldehyde (PFA) and embedded in paraffin or were incubated in 2 moL/L of sucrose before being frozen in TissueTek cutting medium (Sakura Finetek). Six-mm sections were processed for immunofluorescent (IF) or immunohistochemistry (IHC) analyses. Paraffin sections were deparaffinized

Frozen/paraffin sections or cells in 6-well plates were washed with ddH2O, fixed with 4% formaldehyde for 10 min, washed with ddH2O and then exposed to 2% Alizarin Red S for 10 or 30 min followed by washing with 0.2% acetic acid. Positively stained cells displayed a reddish/purple color. 2.8. Transient transfection and luciferase reporter assay For gain-of-function studies of DNMT1, cells at 80% confluence were transfected with pcDNA3/Myc-DNMT1 or control plasmid pcDNA3 using lipofectamine 2000 transfection agent (ThermoFisher) according to the manufacturer's instructions. Luciferase reporter constructs were gifts from Qingbo Xu (King's College London BHF Center, London, United Kingdom). For luciferase assay, the pSV-b-galactosidase plasmid was co-transfected with the luciferase reporter vectors to normalize the transfection efficiency. Twenty-four hours post-transfection, luciferase activity was measured using the Luciferase assay system (Beyotime Biotechnology) and normalized to the b-galactosidase activity assessed using o-nitrophenyl-b-D-galactopyranoside (Amresco). 2.9. Animal models All animal studies were performed in accordance with the guidelines of the Animal Care and Use Committee of Peking University and approved by the Ethics Committee of Peking University Health Science Center (LA2015017). 12-week-old C57/BL6 wildtype mice were obtained from the Experimental Animal Center at Peking University Health Science Center (Beijing, China) and were housed in a temperature-controlled room with 12-h light-dark cycles with free access to food and water. In the CaCl2-induced aortic stiffening model, animals were divided into CaCl2 (n ¼ 24) and control groups (n ¼ 24), anaesthetized and subjected to a laparotomy for exposure of infrarenal abdominal aortas. To induce acute aortic injury, sterile cotton gauze soaked in 0.5 moL/L CaCl2 or normal saline (control) was periadventitially applied to the infrarenal abdominal aortas of mice and then incubated with the aortas for 15 min. Throughout the operation, the level of anesthesia was monitored by testing corneal reflexes and motor responses to tail pinch. Each animal's temperature

206

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

was maintained using a heating pad. The aortas were dissected 7 days post-surgery. The operated arteries (approximately 1 cm in length for each animal) were collected for all the assays. To induce chronic kidney failure, mice were randomly divided into chow diet (n ¼ 18) and high adenine diet (n ¼ 17) groups and all the groups were given a 1-week chow diet followed by the 6week distinct diets (chow vs. adenine (0.3% adenine)) respectively. Body weight and blood pressure were recorded by end of week 0, week 3 and week 6 after the starting of adenine diet. The abdominal arteries between renal arteries and diaphragm were dissected from mice to assay the vessel rigidity and the calcium deposition. The rest of abdominal arteries and the thoracic aortas were harvested to extract protein. For investigating the effects of DNMT1 inhibition on arterial stiffening, mice received repetitive intraperitoneal injection of 0.2 mg/kg body weight daily of DNMT1 inhibitor, 5-Aza in normal saline (n ¼ 11) or the control reagents (DMSO in normal saline) (n ¼ 11) for 14 days. The abdominal arteries between renal arteries and diaphragm were dissected to conduct all the assays.

2.10. Human specimen Internal mammary arteries and atherosclerotic plaques were obtained from patients undergoing coronary artery bypass grafting or carotid endarterectomy. All samples (endarterectomy specimens, n ¼ 5; internal mammary arteries, n ¼ 3) were obtained with the agreement of the patients and approved by the Peking University People's Hospital Medical Ethics Committee (2015PHB024). Immunohistochemistry, immunofluorescence, and measurement of rigidity in human specimen were carried out in accordance with the approved guidelines.

E¼

iL  G A

where A is the cross-sectional area of the specimen that derived from D and d. 2.12. Measurement of material stiffness with nanoindentation The mechanical properties of cells, tissues, and substrates were determined by a newly-developed ferruled optical fiber-based nanoindenter. We used the ferrule-top nanoindenter setup together with the PIUMA controller/drive (Optics11, Amsterdam; The Netherlands) [30]. By combining fiber-optical Fabry-Perot interferometry with a monolithical cantilever-based probe, local micro-elasticity can be examined with high accuracy and precision [31]. A probe with a 0.18 N/m spring constant and a 9 mm spherical indentation tip was used. All measurements were performed by fixing the specimen or cells-on-slide onto the bottom of a Petri dish and then submerging them in buffer at room temperature with the nanoindenter tip remaining well below the surface of buffer at all times. During indentation the tip was brought into contact with the material surface and load-indentation and load-time data was recorded. The indents were depth controlled (10 mm) and the loading and unloading period was set to be 2 s. Based on the loaddisplacement curves the reduced Young's modulus (RedYM) was calculated using the Hertz spherical indentation model. For each gel specimen, the mean Young's modulus was generated from 20 to 30 single measurements in three independent tests. For arterial tissue analysis, 25e30 measurements were made on each tissue specimen. One individual cell derived one measurement. 2.13. Gel contraction assay

2.11. Tensile testing All vessels were isolated on the day of testing and maintained in basal cell culture media until analysis. Any specimen that appeared to sustain damage was discarded. Mechanical analysis was conducted using a uniaxial tensile testing rig (AG-IS, Shimadzu Corporation). Vessel specimens were secured longitudinally within the testing rig using stainless clamps designed for soft biological tissues. In order to acquire reproducible stress-strain curves, the force loading protocol included two following subsequent steps: (I) five loadingeunloading cycles to stabilize the specimens; and (II) uniaxial tensile extension to characterize the mechanical response of the aortic tissues. Uniaxial tensile extension was performed at a speed of 1 mm/min until specimen failure occurred, with force (F) and displacement (L) in terms of strain recorded over time. The internal (d) and external diameters (D) of specimen were measured at both ends of the vessel using a vernier caliper. Measurements of vessel lengths were taken using a micrometer. The distance between the closer edges of the clamps was measured as the initial length (iL) of the specimen. The current length of the specimen was computed by adding the initial length of the specimen and the displacement of the jaw during the test. The Material rigidity (G) was calculated from the force-displacement curve to derive the elastic modulus or Young's Modulus (E). The slope of the linear region prior to failure was used for the calculation. It follows that

G¼

DF DL

whereDF is the increase of force loading applied by the testing rig, DL represents the deformation displacement of the specimen. The elastic modules (E) is calculated as

SMCs were subjected to the treatments listed below before being suspended in the collagen solution: (1) being transfected with pcDNA3/Myc-DNMT1or the pcDNA 3.1 empty vectors; (2) being treated with 5-Aza at 10 mmoL/L or with the control solvent, DMSO, for three days; or (3) being infected with ad-shDNMT1 or ad-GFP. The pretreated cells at a density of 250,000 cells/ml, Type I collagen (4.42 mg/mL), 0.1 moL/L NaOH, 2  F12K, and FBS were mixed on ice in the volume ratio of 4:5:1:8:2 (final concentration of type I collagen, 1.0 mg/ml; final cell density, 5  104 cells/mL) [32], seeded in 12-well plates (400 ml of mixture per well), and placed at 37  C for at least 1 h to polymerize. Following overnight equilibration in F12K medium containing 2% FBS, gels were detached carefully from well wall using a spatula and the extent of gel contraction was monitored on an imaging workstation. The relative gel area was obtained by dividing the area of the collagen gel at each time point by the initial are of the gel. 2.14. In vitro calcification of SMC SMCs at 60% confluent state were inoculated in F12K medium supplemented with 10% FBS in the presence of 1.3 mmoL/L sodium dihydrogen phosphate and 1.3 mmoL/L disodium hydrogen phosphate for 7 days (calcification medium). The medium was replaced with fresh medium every 2 days. Cells that were selected to be used as controls were inoculated in F12K medium supplemented with 10% FBS for 7 days(normal medium), The medium was replaced with fresh medium every 2 days. 2.15. Statistical analysis Data are expressed as mean ± SEM from at least 3 independent

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

experiments. Statistical analysis for single cell experiments was performed by nonparametric Mann-Whitney U test and that for other experiments was performed by t-test for 2 groups of data and by 2-way ANOVA for multiple comparisons. Statistical significance among multiple groups was determined by post hoc analysis (Tukey honestly significant difference test). Values of P < 0.05 were considered statistically significant. 3. Results 3.1. Substrate stiffness regulates the phenotype of cultured vascular SMCs via DNMT1 To investigate how ECM stiffness contributes to the cell fate determination and the phenotypic modulation of vascular SMCs, PA gels with different stiffness were prepared and coated with fibronectin (0.05 mg/ml). Measuring the stiffness of the substrates with a nanoindenter indicated mean elastic moduli of 2.16 and 16.75 kPa, respectively (Fig. 1A). The gels have elastic moduli within the physiological/pathological range of natural solid tissues, which is from roughly 0.1 kPa (for soft brain tissue) to >30 kPa (for rather rigid, calcifying bone) [33]. The consistency of fibronectin immobilization on the gels with different stiffness was visualized by immunofluorescence (Fig S1). HUASMCs were grown on gels for 24 h and were then subjected to assays for gene expression. We observed a decreased expression of DNMT1, but not DNMT3A or 3B (other DNMTs family members) in cells on the stiffer versus softer substrate (Fig. 1B and C). The modulation of DNMT1 expression was probably independent of the type of matrix proteins, since growing the cells on stiffer substrate resulted in DNMT1 down-regulation no matter that the gels were coated with fibronectin, collagen type I, or laminin (Fig. 1D). We therefore used only fibronectin in the rest of the on-gel experiments. Immunofluorescent staining and its semiquantification indicated a predominantly nuclear localization and a higher expression of DNMT1 in the cells on soft substrate in contrast to that on stiff substrate (Fig. 1E and Fig S2A). DNMT1 catalyzes the post-replication methylation of DNA (cytosine-5) and regulates expression of many critical genes through cell differentiation and dedifferentiation. The staining of 5-methylcytosine was strong and clearly in nuclei on soft substrate but became evenly distributed on stiff substrate (Fig. 1F and Fig S2B). Immuno-dot blot assay confirmed the down-regulation of global DNA methylation by substrate stiffening (Fig. 1G). Moreover, treatment of the cells with nocodazole (a rapidly-reversible inhibitor of microtubule polymerization) but not with BDM (an inhibitor of myosin ATPases) after seeding the cells on gels eliminated the down-regulation of DNMT1 expression by substrate stiffening (Fig S3), suggesting a potential role of microtubule cytoskeleton in mediating the regulation of DNMT1 expression by substrate stiffness. Next, we assayed the expressions of smooth muscle phenotypic markers in cells grown on stiff and soft PA gels by quantitative RTPCR and Western blot. The mRNA levels of contractile/differentiation markers, including SMMHC, SM22a, Calponin and Smoothelin in cells on stiff substrate were significantly lower than on soft substrate, whereas the mRNA levels of proliferative/dedifferentiation markers Cyclin A and PCNA and osteoblastic markers MMP3, BMP2 and RunX2 were dramatically higher in cells on stiff as compared with that on soft substrate (Fig. 2A). We confirmed most of the results by Western blot (Fig. 2B). These data indicate that the smooth muscle phenotype is indeed regulated by ECM stiffness. To test the potential role of DNMT1 in this regulation, we utilized gainand loss-of-function approaches to manipulate DNMT1 in SMCs. For DNMT1 overexpression, cells were transfected with plasmids harboring the human full-length mRNA of DNMT1 24 h before being seeded on gels. For DNMT1 inhibition, cells were treated with

207

5-Aza, a DNMTs inhibitor, for three days or infected with adenoviruses carrying shRNAs specifically targeting DNMT1 (adshDNMT1) three days before being seeded on gels. Substrate stiffening resulted in decreased expressions of DNMT1, SMMHC, and SM22a, and increased expressions of RunX2 and BMP2 (Fig. 2C); these alterations in gene expression could be reversed by overexpression of DNMT1 on stiff substrate (Fig. 2C). The increases of Cyclin A and PCNA on stiff substrate won't be blocked by DNMT1 overexpression (Fig. 2C). These findings suggest that DNMT1 mediates the substrate stiffness-regulated synthetic-contractile phenotypic switch and osteoblastic transdifferentiation; substrate stiffening-induced smooth muscle proliferation is not likely dependent on DNMT1. Results from DNMT1 inhibitions by 5-Aza or by ad-shDNMT1 were in line with the DNMT1 overexpression experiments (Fig S4A and S4B). In addition, the immunofluorescent study of SM22a in SMCs grown on soft or stiff substrate with an infection with ad-shDNMT1 or the control virus confirmed the results from Western blot experiments (Fig. 2D). SMC architecture, in particular, cell shape has been linked to functional contractile output. SMCs grown on soft substrate were less spreaded and were elongated with an in vivo-like spindle shape, as indicated by F-action staining (Fig. 2E). In contrast, cells on stiff substrate had a higher extent of cell spreading as quantified with cell adhesion area and were characterized by a hypertrophic polygonal shape (Fig. 2E). DNMT1 inhibition in cells on soft substrate increased the area of cell adhesion, whereas decreased the aspect ratio, implicating impairment in the functional contractile property (Fig. 2E). Collectively, these results indicated that substrate stiffening induces a contractile-to-synthetic phenotypic transition in vascular SMCs through its down-regulation in DNMT1 expression. 3.2. DNMT1 expression and global DNA methylation are repressed in stiffened arteries of mice To gain insight into the in vivo relevance of these issues, we examined whether mechanical stimuli conveyed by ECM stiffness affect DNMT1 expression and DNA methylation in animal models. For this, we utilized two arterial stiffening models, both of which were postulated to have a phenotype of arterial stiffening but their pathological mechanisms differ. In an acute aortic injury and calcification model, CaCl2 at 0.5 moL/L was applied periadventitially to the infrarenal abdominal aorta of C57BL/6 mice and the aortas were harvested 7-days after the surgery. For measuring the mechanical properties of the aortas, we started with a uniaxial rig in consideration of the complexity and heterogeneous nature of vessel wall. Tensile tests indicated that local aortic stiffening was induced with CaCl2 application, as evidenced by an increase of arterial rigidity from 0.119 to 0.256 N/mm, and an increase of elastic modulus from 0.348 to 0.700 MPa (Fig. 3A, left and middle panels). Moreover, we examined the stiffness of the aortas with nanoindentation. In this experiment, arterial strips were cut from the aortas and then the adventitia and media layers were mechanically separated. The outer surface of tunica media was accessed by a nanoindentation probe. The results indicated an increase of elastic modulus from 2.44 kPa to 9.29 kPa (Fig. 3A, right panel). Representative results of force-displacement (load-indentation) curves for nanoindentation and stress-strain curves for tensile tests are included in the Supporting Information (Fig S5 and S6). Results from nanoindentation were typically lower than those from tensile tests. Nevertheless, both tests demonstrate a successful induction of arterial stiffening in the mice model. Alizarin red S staining (Fig. 3B) and immunohistochemistry of BMP2 (Fig. 3C) indicated calcium deposition and smooth muscle osteogenesis. Strikingly, CaCl2 application decreased the expression of DNMT1 and contractile markers

208

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

Fig. 1. Expression of DNA methyltransferase 1 (DNMT1) and global DNA methylation decrease with substrate stiffness in cultured vascular smooth muscle cells (SMCs). (A) Elastic modulus (EM) of the polyacrylamide (PA) gels was verified from 20 to 30 single measurements in three independent tests by a nanoindenter. (B) Western blot assay in SMCs grown on soft and stiff fibronectin (FN)-coated PA gels for 24 h to determine the protein level of DNMT1. Semi-quantification was derived from 8 independent experiments. (C) Realtime PCR analysis to measure gene expressions of DNMTs (n ¼ 7). (D) DNMT1 expressions in cells grown on soft and stiff PA gels coated with collagen type I (Co I) or laminin (Ln). Semi-quantification was derived from 3 independent experiments. (E) and (F) Representative immunofluorescence images of DNMT1, cytosine methylation (5-meC) and nuclei (DAPI) in SMCs. Quantification was derived from 40 single cells (in each group) in triplicate experiments. (F) Left: Immuno-dot blot assay in SMCs to determine global DNA methylation (n ¼ 3). Right: The blot membrane was stained with methylene blue to indicate equal loading of DNA. *P < 0.05 vs Soft. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(SMMHC and SM22a) (Fig. 3C and D). The DNMT1 fluorescence in tunica media layer, as indicated by SM22a was much lower in mice with CaCl2 application than that with sham operation (Fig. 3E). CaCl2 application decreased methylation of cytosine-5 in arterial smooth muscle (Fig. 3E), in parallel with its effects on DNMT1 expression. In consideration of the mechanistic diversity underlying acute versus chronic arterial stiffening, we next sought to determine whether DNMT1 expression and global DNA methylation alter with arterial stiffness over long durations. Clinical observations have

suggested that increased arterial stiffness is a hallmark of vasculopathy in chronic kidney disease patients. Therefore we employed another arterial stiffening model, in which mice were fed high adenine diet for 6 weeks to induce chronic kidney failure. A significant reduction in body weight was observed in the adenine group compared with controls after 3 weeks (Fig S7A). No significant increase in the mean blood pressure was observed (Fig S7B). Adenine diet resulted in increases in arterial rigidity and elastic modulus (Fig. 4A), calcium deposition (Fig. 4B) and BMP2 expression (Fig. 4C). Expression of DNMT1 and contractile markers

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

209

Fig. 2. DNMT1 mediates the regulation of substrate stiffness on the phenotype of SMCs. (A) Real-time PCR analysis in SMCs grown on soft and stiff PA gels for one day to measure expressions of differentiation and dedifferentiation marker genes (n ¼ 4e8). (B) Western blot assay in SMCs grown on soft and stiff PA gels to measure the protein level of SMMHC, SM22a, Cyclin A, PCNA, RunX2 and BMP2 (n ¼ 4e5). *P < 0.05 vs Soft. (C) Western blot assay to measure protein levels of the indicated genes. Cells were transfected with control plasmids (Vector) or plasmids expressing full-length human DNMT1 gene, plated on soft and stiff PA gels, and then harvested 24 h later. Semi-quantification was derived from 4 independent experiments. *P < 0.05 vs Soft/Vector. #P < 0.05 vs Stiff/Vector. (D and E) Representative immunofluorescent staining of SM22a (n ¼ 3) and F-actin in SMCs. Cells were infected with shDNMT1 or control virus, and then plated on soft and stiff PA gels. Cell adhesion area and cell aspect ratio were measured manually by using ImageJ. Quantification of cell adhesion area and aspect ratio was derived from 40 to 50 single cells (in each group) in triplicate experiments. *P < 0.05 vs Soft or the indicated groups.

(SMMHC and SM22a) were markedly repressed by adenine diet (Fig. 4C and D). Moreover, adenine diet inhibited DNMT1 expression and cytosine-5 methylation in tunica media (Fig. 4EeG). Taken together, the results from these two animal models indicate a negative correlation between arterial stiffening and DNMT1 expression, suggesting an involvement of DNMT1 and DNMT1-mediated DNA methylation in this pathological process.

3.3. DNMT1 expression and DNA methylation are repressed in calcified atherosclerotic lesions of human carotid arteries Atherosclerosis-associated intima calcification is characterized by changes in biochemical and physical properties of ECM in the thickened intima. We measured the elastic moduli from three typical regions of the carotid endarterectomy specimens

210

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

Fig. 3. DNMT1 expression, global DNA methylation, and expressions of smooth muscle contractile markers are decreased in aorta from a mice model of CaCl2-induced aortic stiffening. (A) Tensile testing (left and middle) and nanoindentation (right) in infrarenal abdominal aortas from mice with periadventitial application of CaCl2 or with the sham operation to calculate their mechanical properties (material rigidity and elastic modules). Each dot represents one animal. For nanoindentation, one dot was derived from 20 to 30 single measurements in different regions of each aorta. EM, elastic modulus. (B) Representative Alizarin red S staining of the CaCl2-treated or the sham-operated infrarenal abdominal aortas (n ¼ 6 in each group). (C) Representative immunohistochemistry of BMP2 and DNMT1 in the indicated aortas (n ¼ 6 in each group). (D) Western blot assay to measure the protein level of DNMT1, SMMHC and SM22a in the indicated aortas (n ¼ 6 in each group). (E-G) Representative immunofluorescence images of DNMT1, 5-meC, SM22a and nuclei in the indicated aortas (n ¼ 6 in each group). *P < 0.05 vs the sham-operation with normal saline application (Sham). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

from patients with atherosclerotic occlusive diseases by nanoindentation and found they ranged from 2.30 kPa (for comparatively healthy regions) to 23.94 kPa (for severely calcified regions) (Fig. 5A and Fig S8). The severely calcified regions had stronger staining of Alizarin red S and BMP2 immunohistochemistry, and decreased staining of contractile protein SM22a, in comparisons with the non-diseased internal mammary arteries (Fig. 5B). We then examined whether ECM stiffening conveyed by intima calcification leads to changes in DNMT1 expression. We observed that in the lesions the expression of DNMT1 and global methylation were weaker than

in the normal arteries (Fig. 5C). To address more directly the relevance of the changes in DNMT1 expression and cytosine-5 methylation in calcified lesions to intimal SMCs, we performed immunofluorescent staining to analyze DNMT1 expression and its colocalization with smooth muscle marker SM22a. As compared with the normal control, SM22acolocalized DNMT1 was repressed in calcified intima (Fig. 5D), paralleled by a reduced level of cytosine-5 methylation in SM22apositive cells (Fig. 5E). Collectively, these findings suggest the relevance of DNMT1 in atherosclerotic intimal calcification.

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

211

Fig. 4. DNMT1 expression, global DNA methylation, and expressions of smooth muscle contractile markers are decreased in aorta from mice with adenine diet-induced arterial stiffening. (A) Tensile testing (left and middle) and nanoindentation (right) in thoracic aortas from mice fed adenine diet or chow diet to calculate their mechanical properties (material rigidity and elastic modules). Each dot represents one animal. For nanoindentation, one dot was derived from 20 to 30 single measurements in different regions of each aorta. EM, elastic modulus. (B) Representative Alizarin red S staining of thoracic aortas from mice with adenine or chow diet (n ¼ 3 in each group). (C) Representative immunohistochemistry of BMP2 and DNMT1 in the indicated aortas (n ¼ 3 in each group). (D) Western blot assay to measure the protein level of DNMT1, SMMHC and SM22a in the indicated aortas (n ¼ 4 in each group). (E-G) Representative immunofluorescence images of DNMT1, 5-meC, SM22a and nuclei in the indicated aortas (n ¼ 3 in each group). *P < 0.05 vs chow diet. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. DNMT1 inhibition facilitates arterial stiffening We then postulated that DNMT1 repression mediates the pathological development of arterial stiffening. We analyzed the mechanical properties of abdominal aortas from mice with repetitive intraperitoneal injection of 0.2 mg/kg body weight daily of DNMT1 inhibitor (5-Aza, dissolved in normal saline) or the control reagent for 14 days. Arterial rigidity and elastic modulus were significantly increased by DNMT1 inhibition (Fig. 6A). We validated the 5-Aza-induced arterial stiffening by Alizarin red S staining and immunohistochemistry of BMP2 (Fig. 6B and C).

Immunohistochemistry of DNMT1 verified the DNMT1 inhibition by 5-Aza injection in the abdominal aortas (Fig. 6C). Arterial stiffening is accompanied by osteogenic transdifferentiation and calcification of SMCs in tunica media. Is then DNMT1 repression responsible for the above changes of SMCs during arterial stiffening? To answer this question, we used loss-offunction study to test whether DNMT1 mediates arterial stiffening by managing smooth muscle phenotype in vitro. In comparison with the treatment by control reagent or infection by control virus, inhibition of DNMT1 by 5-Aza treatment or by ad-shDNMT1 infection decreased the mRNA expression of DNMT1 while

212

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

Fig. 5. DNMT1 expression and global DNA methylation are repressed in the smooth muscle cells of calcified atherosclerotic lesion in diseased human carotid arteries. (A) Elastic modulus (EM) measured by nanoindentation in three typical regions of the human carotid endarterectomy specimens. The data were derived from 20 to 30 single measurements in each region. (B) Representative Alizarin red S staining and immunohistochemistry of BMP2 and SM22a in the calcified atherosclerotic intimal lesions of human carotid arteries (Lesion, n ¼ 4) and the non-diseased control internal mammary arteries (Normal, n ¼ 3). (C) Representative immunohistochemistry of DNMT1 and 5-meC in the diseased and control arteries. (D) and (E) Representative immunofluorescence images of DNMT1, 5-meC, SM22a and nuclei in the diseased and control arteries. *P < 0.05 vs the indicated groups. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

increased the mRNA expressions of osteoblastic markers MMP3, BMP2 and RunX2 in SMCs grown on plastic substrate (Fig. 6D). We verified the changes in expressions of DNMT1, BMP2, and RunX2 by Western blot (Fig. 6E). To determine the contribution of DNMT1 repression to SMC calcification, we utilized a high-phosphateinduced calcification model. Alizarin red S staining indicated mineralization of SMCs with osteogenic media containing 2.6 mmoL/L inorganic phosphates (Fig. 6F). Strikingly, DNMT1 inhibition by 5-Aza or by shDNMT1 aggravated SMC mineralization (Fig. 6F). Complementary experiments indicated that the highphosphate-stimulated smooth muscle calcification could be prohibited by overexpression of DNMT1 (Fig S9). Taken together, these results imply a role of DNMT1 in protecting arteries from stiffening. 3.5. DNMT1 regulates cellular stiffness Increased arterial stiffness is attributable not only to changes in ECM but also to changes in intrinsic mechanical properties of vascular SMCs. We next examined the cellular stiffness of SMCs

with overexpression or inhibition of DNMT1. Single cell stiffness was measured using nanoindenter. Notably, DNMT1 inhibition by 5-Aza treatment or by ad-shDNMT1 infection increased the elastic modulus from 6.59 kPa to 7.05 kPae9.38 kPa and 10.30 kPa, respectively (Fig. 7A, left and middle panels; Fig S10). Transfection of the cells with DNMT1-overexpression plasmids decreased the elastic modulus of the cell body from 6.49 kPa to 4.35 kPa (Fig. 7A, right panel; Fig S10). Soft substrate produced softer cells in comparison with the stiff substrate (Fig. 7BeD and Fig S11-13). Inhibition of DNMT1 in cells on soft substrate eliminated the effect of soft substrate on softening the cells (Fig. 7B, C and Fig S11, 12) while overexpression of DNMT1 in cells on stiff substrate ameliorated the effect of stiff substrate on stiffening the cells (Fig. 7D and Fig S13). These results demonstrate that matrix stiffness modulates SMC stiffness via its regulation on DNMT1. 3.6. DNMT1 maintains SMC contractility SMCs from stiff arteries express low levels of contractile marker

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

213

Fig. 6. DNMT1 inhibition facilitates arterial stiffening in vivo and promotes osteogenic transdifferentiation and calcification of vascular SMCs in vitro. (A) Tensile testing in abdominal aortas from mice with repetitive intraperitoneal injection of DNMT1 inhibitor (5-Aza) or control reagents (DMSO) to calculate their mechanical properties. Each dot represents one animal. (B) Representative Alizarin red S staining of abdominal aortas from mice received 5-Aza or control reagents (n ¼ 4 in each group). (C) Representative immunohistochemistry of BMP2 and DNMT1 in the indicated aortas (n ¼ 4 in each group). (D) and (E) Real-time PCR (n ¼ 4e7) and Western blot assays (n ¼ 4e7) to detect the expressions of DNMT1 and osteoblastic markers in SMCs with the indicated treatments. Cells on plastic dishes were treated with 5-Aza, or infected with adenovirus expressing shDNMT1. (F) Representative Alizarin red S staining (n ¼ 3) of SMCs with the indicated treatments. Cells were treated with 5-Aza or infected with adenovirus expressing shDNMT1 followed by high-phosphate (Pi) stimulation for 7 days. *P < 0.05 vs DMSO, control virus (Ad-CL) or the indicated groups. #P < 0.05 vs Ad-CL. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

proteins and have decreased cell contractility. SMC contractility is generated by actomyosin interactions and actin polymerization and is enhanced by expression of contractile proteins such as SM22a and SMA. To further study the mechanisms of DNMT1-mediated protection for arteries from stiffening, we tested the regulation of DNMT1 on the promoter activities of SM22a and SMA. Luciferase reporter plasmids harboring the promoter fragments of SM22a or SMA control vectors were transfected into SMCs. Gain- and loss-of-

function of DNMT1 were achieved by co-transfection of the cells with DNMT1 overexpression plasmids, or by treatment with 5-Aza, or by infection with ad-shDNMT1. DNMT1 overexpression significantly increased the reporter luciferase activities, by 169% and 174%, as compared with the control vector (Fig. 8A, upper panel). In contrast, 5-Aza treatment or ad-shDNMT1 infection markedly decreased the reporter luciferase activities, by 73%, 74% and 72%, 85% (Fig. 8A, middle and lower panels). We then measured SMC

214

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

Fig. 7. DNMT1 softens SMCs. Cellular stiffness by means of elastic modulus was measured in SMCs with the indicated treatments. (A) Cells in plastic dishes were treated with 5-Aza, or infected with adenovirus expressing shDNMT1, or transfected with plasmids overexpressing DNMT1. (B-D) Cells pre-treated with 5-Aza (B), infected with shDNMT1 (C), or transfected with plasmids expressing DNMT1 (D), and the cells were seeded on soft or stiff PA gels. The mean Young's modulus was generated from 20 to 30 single measurements (in each group) in three independent tests. One individual cell derived one measurement. *P < 0.05 vs DMSO, control virus (Ad-CL), or control plasmids (Vector).

Fig. 8. DNMT1 modulates the promoter activities of SM22a and SMA and SMC contractility. (A) Luciferase activity assay (n ¼ 4e7) in SMCs with the transfection of reporter constructs containing the promoter regions of SM22a or SMA. Cells were treated with 5-Aza, or infected with adenovirus expressing shDNMT1, or transfected with plasmids overexpressing DNMT1, accompanied by a transfection with the luciferase reporter constructs. (B) Left: Representative gel contraction images. Right: The quantification of gel contraction with respect to the indicated treatments (n ¼ 4). *P < 0.05 vs DMSO, control virus (CL) or control plasmids (Vector).

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

contractility with gel contraction assay. In this assay SMCs were seeded within hydrated collagen lattices and contractility of SMC was analyzed by measuring alterations in diameter of the lattices. The cell-driven gel contraction in forms of the reduction in surface area of the lattices was enhanced by DNMT1 overexpression but was weakened by DNMT1 inhibition in the embedded SMCs (Fig. 8B). Collectively, these data demonstrates that expression of DNMT1 maintains SMC contractility. 4. Discussion Researchers have embraced the concept that vascular SMCs are remarkably plastic. They can switch their phenotype between contractile and synthetic, depending on the environmental cues they sense. Under physiological conditions, the SMCs within adult blood vessels exhibit a contractile/quiescent phenotype, characterized by the expressions of contractile proteins, low proliferation rates, low rates of proteins synthesis and secretion. SMCs can undergo phenotypic modulation in response to physical or chemical signals and de-differentiate into activated/dedifferentiated cells, that expressing relatively few contractile proteins, re-entering the cell cycle, becoming migratory, and undergoing apoptosis and cell death sometimes. In recent years, emerging evidence has suggested a key role of epigenetic mechanisms (DNA methylation and demethylation, histone modification, and RNAbased mechanisms) in mediating the environmental signalsinduced modulation of SMC phenotype [34,35]. Several studies have reported that certain SMC genes, e.g., cytokines signaling 3(SOCS3) [36], collagen, type XV, alpha 1(COL15A1) [26], alkaline phosphatase (ALP) [27], and SM22a [37], are regulated by DNA methylation that could be consequently associated with SMC phenotypic modulation and the pathogenesis of vascular diseases. However, the understanding of mechanical stimuli-elicited epigenetic regulation of SMC phenotype is limited. Here we report a strong and previously undocumented influence of microenvironment stiffness on SMC phenotype modulation that is mediated by DNMT1. DNMT1 was firstly identified as a mechanosensitive protein in our early study, in which we found that the expression and activity of DNMT1 could be induced by atheroprone flow shear stress in vascular endothelial cells [28]. In the current study, we further demonstrated the fundamental role of DNMT1 as downstream element of mechanosensor in how vascular SMCs perceive their physical microenvironment. When the matrix elasticity gets stiffer the cells produce less DNMT1 and the nuclear localization of DNMT1 becomes less predominant, resulting in a decline of DNA methylation level (Fig. 1E and F). The detailed mechanotransduction mechanisms by which ECM stiffness regulates DNMT1 await further characterization, but it is tempting to speculate that microtubule but not the myosinfilament cytoskeletal networks may be involved (Fig S3). Moreover, the modulation of DNMT1 expression by the mechanical cues is independent on the ligands (fibronectin, collagen type I, or laminin) that are used to link the cells to the substrates (Fig. 1B and D). These findings have potential implications for the study of mechanotransduction and highlight the importance of DNMT1 in mediating the interactions between SMCs and matrix stiffness, where the influences of mechanics on DNMT1 expression is not conditioned on ligand biochemistry. Functionally, we showed in substrate-stiffness model that cells read substrate stiffness as level of DNMT1, such that experimental manipulations of DNMT1 levels can instruct cell phenotype in terms of expression of phenotype-specific gene (i.e., SMMHC, SM22a, BMP2, and RunX2), overruling the mechanical inputs (Fig. 2). The responses of SMCs to substrate stiffness here may potentially include remodeling the microenvironment; for

215

example, MMP3 production is greater in SMCs on stiff substrate than that in cells on soft substrate (Fig. 2A), consistent with the secretory phenotype. Interestingly, our results not only showed that SMCs adapted to and remodeled the exogenous mechanomicroenvironment by producing more matrix proteins, but also demonstrated that SMCs change their own cellular stiffness to contribute to the gross mechanical properties (Fig. 7). Indeed, stiffer matrices producing stiffer cells have been indicated in various cell types in previous reports [3,38]. Our study proposes an explanation for the underlying mechanism by showing that when mechanosensitive protein, DNMT1 in our case, receives signals from the mechano-microenvironment, cells respond by altering their stiffness. Contractile-to-synthetic phenotypic transition of SMCs has also been observed in our in vivo studies, in which the SMCs within stiffer arteries express fewer contractile markers and DNMT1 than within the controls (Figs. 3 and 4). Conversely, expression of osteogenesis marker, BMP2, is greater in the stiffening arteries (Figs. 3C and 4C). It is possible that changes in both the composition and the organization of ECM induced by CaCl2 application or by adenine diet impairs the expression of DNMT1, whose impairment in turn facilitates SMC stiffening and the consequent arterial stiffening. Such an implication is contrary to current concepts that increased vascular stiffness of diseases is simply attributable to deposition of calcium phosphate salts and ECM changes, primarily collagen and elastin. However, our findings are in line with recent reports revealing that increased aortic stiffness in spontaneously hypertensive rats [22] and aged monkeys [21] is mainly attributable to the increased vascular SMC stiffness. Additionally, decreased smooth muscle expression of DNMT1 was also revealed in calcified atherosclerotic lesions of human carotid arteries, which was accompanied by increased expression of BMP2 (Fig. 5), reinforcing the implication that impairment in DNMT1 expression may be both a leading cause and/or a major consequence of arterial stiffening. Functional cellular contractility is the most robust indicator of contractile SMC phenotype and is defected during arterial stiffening [39,40]. Contractility of SMCs overall can be assessed quantitatively by gel contraction assay. We found that the contraction exerted by the embedded SMCs was enhanced by DNMT1 overexpression and were reduced by DNMT1 inhibition of the SMCs (Fig. 8B). Mechanistically, we demonstrated that DNMT1 elicits its effects on contractility by regulating the promoter activities of SMA and SM22a (Fig. 8A). SMA is not required for the induction of contraction, but its expression leads to contractility upregulation that involves stress fibers modification [41,42]. SM22a is physically associates with cytoskeletal actin filament bundles in contractile SMCs and its deficiency decreases aortic contractility in vivo [43]. Our findings suggest a compelling speculation for the mechanisms of DNMT1 regulation of SMC function as well as arterial stiffness. In addition to the mechanistic exploration of DNMT1-mediated modulation on SMC contractility, a key feature of our study is the identification of DNMT1 as a novel regulator of arterial stiffness, as evidenced by the inhibition of DNMT1 in vivo induces aortic stiffening (Fig. 6AeC). 5. Conclusion Our data indicate that substrate stiffening induces a contractileto-synthetic phenotypic transition of vascular SMCs through downregulation of DNMT1 and global DNA methylation. DNMT1 is deficient in stiffening arteries in mice with acute aortic injury or chronic kidney failure, as well as in calcified atherosclerotic lesions of human carotid arteries. Our results also reveal that pharmacological inhibition of DNMT1 promotes osteogenic

216

S.-A. Xie et al. / Biomaterials 155 (2018) 203e216

transdifferentiation, calcification and cellular stiffening of vascular SMCs and hence arterial stiffening. Our results suggested an underlying mechanism that DNMT1 responds to and modulates arterial stiffness by regulating SMC contractility. This research may pave the way for a better understanding of the complex crosstalk of vascular SMCs and their matrix during the development of vascular disease, as well as suggest novel therapeutic strategies for arterial stiffening. The wider implications include potential targeting of the phenotypic regulatory mechanisms in material-related therapeutic applications, such as in-stent restenosis following the implantation of the metal material/device in the atherosclerotic vessels. Acknowledgements This work was funded by the National Natural Science Foundation of the P.R. China 91539116, 31522022, 81470590 (to J. Z.), and 31570938 (to W. J. Y.); Beijing Natural Science Foundation 7152081 (to J. Z.). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.biomaterials.2017.11.033. References [1] D.E. Jaalouk, J. Lammerding, Mechanotransduction gone awry, Nat. Rev. Mol. Cell Biol. 10 (1) (2009) 63e73. [2] B. Suki, Assessing the functional mechanical properties of bioengineered organs with emphasis on the lung, J. Cell Physiol. 229 (9) (2014) 1134e1140. [3] A.J. Engler, et al., Matrix elasticity directs stem cell lineage specification, Cell 126 (4) (2006) 677e689. [4] T.J. Kim, et al., The regulation of beta-adrenergic receptor-mediated PKA activation by substrate stiffness via microtubule dynamics in human MSCs, Biomaterials 35 (29) (2014) 8348e8356. [5] J. Seong, et al., Distinct biophysical mechanisms of focal adhesion kinase mechanoactivation by different extracellular matrix proteins, Proc. Natl. Acad. Sci. U. S. A. 110 (48) (2013) 19372e19377. [6] Chaterji, S., et al., Synergistic effects of matrix nanotopography and stiffness on vascular smooth muscle cell function. Tissue Eng. Part A. 20(15e16): p. 2115e2126. [7] N. Shi, S.Y. Chen, Smooth muscle cell differentiation: model systems, regulatory mechanisms, and vascular diseases, J. Cell Physiol. 231 (4) (2016) 777e787. [8] K. Sobue, K. Hayashi, W. Nishida, Expressional regulation of smooth muscle cell-specific genes in association with phenotypic modulation, Mol. Cell Biochem. 190 (1e2) (1999) 105e118. [9] S. Chaterji, et al., Synergistic effects of matrix nanotopography and stiffness on vascular smooth muscle cell function, Tissue Eng. Part A 20 (15e16) (2014) 2115e2126. [10] S.R. Peyton, A.J. Putnam, Extracellular matrix rigidity governs smooth muscle cell motility in a biphasic fashion, J. Cell Physiol. 204 (1) (2005) 198e209. [11] S.R. Peyton, et al., The effects of matrix stiffness and RhoA on the phenotypic plasticity of smooth muscle cells in a 3-D biosynthetic hydrogel system, Biomaterials 29 (17) (2008) 2597e2607. [12] X.Q. Brown, et al., Effect of substrate stiffness and PDGF on the behavior of vascular smooth muscle cells: implications for atherosclerosis, J. Cell Physiol. 225 (1) (2010) 115e122. [13] J. Blacher, et al., Arterial calcifications, arterial stiffness, and cardiovascular risk in end-stage renal disease, Hypertension 38 (4) (2001) 938e942. [14] C. Palombo, M. Kozakova, Arterial stiffness, atherosclerosis and cardiovascular risk: pathophysiologic mechanisms and emerging clinical indications, Vasc. Pharmacol. 77 (2016) 1e7. [15] D. Kothapalli, et al., Cardiovascular protection by ApoE and ApoE-HDL linked to suppression of ECM gene expression and arterial stiffening, Cell Rep. 2 (5) (2012) 1259e1271.

[16] K. Ng, et al., Aortic stiffness is associated with vascular calcification and remodeling in a chronic kidney disease rat model, Am. J. Physiol. Ren. Physiol. 300 (6) (2011) F1431eF1436. [17] M. Kanbay, et al., Arterial stiffness in dialysis patients: where are we now? Int. Urol. Nephrol. 42 (3) (2010) 741e752. [18] K. Amann, Media calcification and intima calcification are distinct entities in chronic kidney disease, Clin. J. Am. Soc. Nephrol. 3 (6) (2008) 1599e1605. [19] M.R. Bennett, Apoptosis of vascular smooth muscle cells in vascular remodelling and atherosclerotic plaque rupture, Cardiovasc Res. 41 (2) (1999) 361e368. [20] Y. Du, et al., Upregulation of a disintegrin and metalloproteinase with thrombospondin motifs-7 by miR-29 repression mediates vascular smooth muscle calcification, Arterioscler. Thromb. Vasc. Biol. 32 (11) (2012) 2580e2588. [21] H. Qiu, et al., Short communication: vascular smooth muscle cell stiffness as a mechanism for increased aortic stiffness with aging, Circ. Res. 107 (5) (2010) 615e619. [22] N.L. Sehgel, et al., Increased vascular smooth muscle cell stiffness: a novel mechanism for aortic stiffness in hypertension, Am. J. Physiol. Heart Circ. Physiol. 305 (9) (2013) H1281eH1287. [23] A. Portela, M. Esteller, Epigenetic modifications and human disease, Nat. Biotechnol. 28 (10) (2010) 1057e1068. [24] P. Sharma, et al., Detection of altered global DNA methylation in coronary artery disease patients, DNA Cell Biol. 27 (7) (2008) 357e365. [25] A.V. Shirodkar, P.A. Marsden, Epigenetics in cardiovascular disease, Curr. Opin. Cardiol. 26 (3) (2011) 209e215. [26] J.J. Connelly, et al., Epigenetic regulation of COL15A1 in smooth muscle cell replicative aging and atherosclerosis, Hum. Mol. Genet. 22 (25) (2013) 5107e5120. [27] T. Azechi, et al., 5-aza-2'-Deoxycytidine, a DNA methyltransferase inhibitor, facilitates the inorganic phosphorus-induced mineralization of vascular smooth muscle cells, J. Atheroscler. Thromb. 21 (5) (2014) 463e476. [28] J. Zhou, et al., Epigenetic mechanism in regulation of endothelial function by disturbed flow: induction of DNA hypermethylation by DNMT1, Cell Mol. Bioeng. 7 (2) (2014) 218e224. [29] J.R. Tse, A.J. Engler, Preparation of hydrogel substrates with tunable mechanical properties, Curr. Protoc. Cell Biol. (2010) (Chapter 10): p. Unit 10.16.1e10.16.16. [30] G. Mattei, et al., The nano-epsilon dot method for strain rate viscoelastic characterisation of soft biomaterials by spherical nano-indentation, J. Mech. Behav. Biomed. Mater 50 (2015) 150e159. [31] D. Chavan, et al., Ferrule-top nanoindenter: an optomechanical fiber sensor for nanoindentation, Rev. Sci. Instrum. 83 (11) (2012) 115110. [32] E. Bell, B. Ivarsson, C. Merrill, Production of a tissue-like structure by contraction of collagen lattices by human fibroblasts of different proliferative potential in vitro, Proc. Natl. Acad. Sci. U. S. A. 76 (3) (1979) 1274e1278. [33] G.C. Reilly, A.J. Engler, Intrinsic extracellular matrix properties regulate stem cell differentiation, J. Biomech. 43 (1) (2010) 55e62. [34] R. Liu, K.L. Leslie, K.A. Martin, Epigenetic regulation of smooth muscle cell plasticity, Biochim. Biophys. Acta 1849 (4) (2015) 448e453. [35] Liu, R., et al., Ten-eleven translocation-2 (TET2) is a master regulator of smooth muscle cell plasticity. Circulation. 128(18): p. 2047e2057. [36] K. Dhar, et al., SOCS3 promotor hypermethylation and STAT3-NF-kappaB interaction downregulate SOCS3 expression in human coronary artery smooth muscle cells, Am. J. Physiol. Heart Circ. Physiol. 304 (6) (2013) H776eH785. [37] A. Montes de Oca, et al., High-phosphate-induced calcification is related to SM22alpha promoter methylation in vascular smooth muscle cells, J. Bone Min. Res. 25 (9) (2010) 1996e2005. [38] C. Collins, et al., Haemodynamic and extracellular matrix cues regulate the mechanical phenotype and stiffness of aortic endothelial cells, Nat. Commun. 5 (2014) 3984. [39] A. Ramachandran, et al., JunB mediates basal- and TGFbeta1-induced smooth muscle cell contractility, PLoS One 8 (1) (2013) e53430. [40] R.M. Fitch, et al., Synergistic effect of angiotensin II and nitric oxide synthase inhibitor in increasing aortic stiffness in mice, Am. J. Physiol. Heart Circ. Physiol. 290 (3) (2006) H1190eH1198. [41] J.H. Wang, J.S. Lin, Cell traction force and measurement methods, Biomech. Model Mechanobiol. 6 (6) (2007) 361e371. [42] J. Chen, et al., Alpha-smooth muscle actin expression enhances cell traction force, Cell Motil. Cytoskelet. 64 (4) (2007) 248e257. [43] H.D. Je, U.D. Sohn, SM22alpha is required for agonist-induced regulation of contractility: evidence from SM22alpha knockout mice, Mol. Cells 23 (2) (2007) 175e181.