Substrate elasticity regulates adipose-derived stromal cell differentiation towards osteogenesis and adipogenesis through β-catenin transduction

Accepted Manuscript Full length article Substrate elasticity regulates adipose-derived stromal cell differentiation towards osteogenesis and adipogenesis through β-catenin transduction Jing Xie, Demao Zhang, Chenchen Zhou, Quan Yuan, Ling Ye, Xuedong Zhou PII: DOI: Reference:

S1742-7061(18)30485-9 https://doi.org/10.1016/j.actbio.2018.08.018 ACTBIO 5622

To appear in:

Acta Biomaterialia

Received Date: Revised Date: Accepted Date:

25 December 2017 3 August 2018 17 August 2018

Please cite this article as: Xie, J., Zhang, D., Zhou, C., Yuan, Q., Ye, L., Zhou, X., Substrate elasticity regulates adipose-derived stromal cell differentiation towards osteogenesis and adipogenesis through β-catenin transduction, Acta Biomaterialia (2018), doi: https://doi.org/10.1016/j.actbio.2018.08.018

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

First page information Full length scientific article Substrate elasticity regulates adipose-derived stromal cell differentiation towards osteogenesis and adipogenesis through β-catenin transduction

Jing Xie1, Demao Zhang1, Chenchen Zhou1, Quan Yuan1, Ling Ye1, Xuedong Zhou1*

1.

State Key Laboratory of Oral Diseases, West China Hospital of Stomatology, Sichuan University, Chengdu,

China

*Corresponding author: Professor Xuedong Zhou, Director of State Key Laboratory of Oral Diseases, Academic Dean of West China School of Stomatology, Sichuan University Chengdu Sichuan 610064, CHINA Tel: 86-28-85501469; Fax: 86-28-85582167; Email: [email protected]

Abstract It is generally recognised that mesenchymal stem cells (MSCs) can differentiate into multiple lineages through guidance from the biophysical properties of the substrates. However, the precise biophysical mechanism that enables MSCs to respond to substrate properties remains unclear. In the current study, polydimethylsiloxane (PDMS) substrates with different stiffnesses were fabricated and the way in which the elastic modulus of the substrate regulated differentiation towards osteogenesis and adipogenesis in adipose-derived stromal cells (ASCs) was explored. Initially, a cell morphology change by SEM was observed between the stiff and soft substrates. The cytoskeleton stains including microfilament by F-actin and microtubule by α- and β-tubulin further showed a larger cell spreading area on the stiff substrate. Then the expression of vinculin, in charge for the linkage of adhesion molecules to the actin cytoskeleton, was enhanced on the stiff substrate. This change in focal adhesion plaque further triggered intracellular β-catenin signaling and promoted its nuclear translocation especially on the stiff substrate. The influence of β-catenin signaling on direct differentiation to osteogenic lineages was through direct binding between its downstream protein, Lef-1, and the osteogenic transcriptional factors, Runx2 and Osx, while on differentiation to adipogenic lineages was through modulating the expression of PPARγ. The imbalance of stiffness-induced β-catenin signaling finally induced a stronger osteogenesis and a weaker adipogenesis on the stiff substrate relative to those on the soft substrate. This study indicates the importance of stiffness on ASC differentiation and could help to increase understanding of the mechanism underlying molecular signal transduction from mechanosensing, mechanotransducing to stem cell differentiation.

*Statement of Significance

Mesenchymal stem cells can differentiate into multiple lineages, such as adipogenesis, myogenesis, neurogenesis, angiogenesis and osteogenesis, through influence of biophysical properties of the extracellular matrix. However, the precise bio-mechanism that triggers stem cell differentiation in response to matrix biophysical properties remains unclear. In the current study, we provide a series of experiments involving the characterization of cell morphology, microfilament, microtubule and adhesion capacity of adipose-derived stromal cells (ASCs) in response to substrate stiffness, and further elucidation of cytoplasmic β-catenin-dependent signal transduction, nuclear translocation and resultant promoter activation of transcriptional factors for osteogenesis and adipogenesis. This study provides an explanation on deeper understanding of bio-mechanism underlying substrate stiffness-triggered β-catenin signal transduction from active mechanosensing, mechanotransducing to stem cell differentiation.

Keywords: Substrate elasticity, Adipose-derived stromal cells, Mechanosensing, Mechanotransducing, Cell differentiation

1. Introduction Adipose-derived stromal cells (ASCs), isolated from subcutaneous fat, are fibroblast-like cells which are shown to have the similar phenotype to mesenchymal stem cells (MSCs), isolated from bone marrow and umbilical cord blood [1, 2], and are believed to have the great potential in tissue engineering and regenerative medicine [3]. The advantages of ASCs, including a less invasive harvesting approach, a larger number of stem cell progenitors yielded from an equivalent amount of other tissues harvested, a better differentiation towards angiogenesis, neurogenesis, osteogenesis and adipogenesis, and a stronger immunomodulatory property, make them a promising alternative to the multi-functional BMSCs [4, 5]. Besides, their application in tissue engineering and regenerative medicine is not limited to mesoderm-derived tissue but extends to both ectoderm- and endoderm-derived tissues and organs, although ASCs originate from mesoderm lineages [6, 7]. Thus, the use of autologous ASCs as both research tool and cellular therapy is feasible and has been shown to be effective in preclinical and clinical studies of injuries and diseases. Although the ASCs have gained much attention in regenerative medicine, the exact mechanism directing ASC differentiation in response to micro-environmental biophysical properties of the extracellular matrix (ECM) is much exclusive. Previous reports have elucidated that interface between stem cells and their ECM was a dynamic and complex micro-environment where the cells were in contact with the ECM through sensing the biophysical properties such as stiffness and geometry and responded to these cues in multiple ways [8-12]. The biophysical

properties of the interface, in spite of biochemical factors in the ECM, have been confirmed to play a critical role in regulation of cell biology [9]. For example, for adult cells, the evidence indicated that substrate elasticity, mimicking the micro-environment of different tissues, such as neuron, adipose tissue, cartilage, cancellous bone, and blood vessel, was capable of influencing the phenotype and functionalization of terminal cell types [10]. For stem cells, the data confirmed that substrate elasticity could induce the differentiation of MSCs into various cell end points [11, 12]. Engler et al. demonstrated that exposure of BMSCs to materials of low, intermediate, and high stiffness directed cell commitment into adipogenic, myogenic, and osteogenic lineages, respectively [12]. ASC lineage specification has also been shown to be sensitive to biophysical and biochemical cues such as matrix stiffness, cell geometry and adhesion ligands, which exist in the surrounding microenvironment [13-19]. For instance, Duan et al. demonstrated that ASCs could directly differentiate towards osteogenesis by activating gene expressions, i.e., alkaline phosphatase (ALP), osteoprotegerin (OPG), osteocalcin (OCN), cannabinoid receptors type I (CNR1) and II (CNR2) and receptor activator of nuclear factor kappa β ligand (RANKL), and ECM protein components, i.e., collagens, elastins and sulfated glycosaminoglycan [13]. Lechner et al. confirmed that ASCs could differentiate into adipose tissue through CCAAT/enhancer binding protein beta (C/EBPβ), besides the adipogenic key regulators, peroxisome proliferator activated receptor gamma (PPARγ) and C/EBPα [14]. Differentiation of ASCs is believed to occur in two stages [15]. The first stage is commitment of ASCs to primary differentiation which could be regulated by biophysical and biochemical cues such as substrate elasticity [16], cell shape [17], and matrix ligand [18]. The second stage is terminal differentiation to mature cells by transcriptional factors [19]. For example, in terminal differentiation to mature osteoblasts, activation of runt related transcription factor 2 (Runx2) and Osterix (Osx) is necessary, while in adipocytes, activation of PPARγ and C/EBPα was essential. Although the multitude of literature on their differentiation capacity is rapidly advancing, further investigations of the precise biophysical mechanisms that enable stem cells to respond to matrix biophysical properties remain necessary. In this study, based on the silicon-based elastomer polydimethylsiloxane (PDMS) substrates with stiff/soft stiffnesses but with the indistinctive surface topologies, we explored the changes of cellular morphology, cytoskeleton, adhesion ability, mechanotransducing signal, and final differentiations of ASCs. The process from active mechanosensing, mechanotransducing, to cell fate decision in response to stiffness could help to increase understanding the mechanism underlying the interface between cells and their ECM micro-environment.

2. Methods and Materials 2.1 Preparation of PDMS substrates PDMS has been used as a principal candidate in many biomaterial studies because of its flexibility, optical clarity and elastic tenability [20]. PDMS elastomer was prepared by mixing Sylgard 184 (Corning, NY, USA) as

curing agent in two different ratios to oligomeric base (i.e., 1 : 5 and 1 : 45), which was subsequently cast onto a single-well petri dish (35 × 10 mm, Corning) and cured at 60 °C for 3 h. This was followed by a sterilization step, carried out by exposing samples to UV radiation for 2 h, and then cell culturing was performed. 2.2 Young’s modulus measurements of PDMS substrates Young’s modulus measurement was carried out by the spherical indentation method to characterize the stiffness of different PDMS substrates, as described previously [21]. The ElectroForce® 3100 test instrument (Bose, Shanghai, China) was used for the indentation tests. A spherical indenter 3 mm in radius was used, while the measurements were performed in ambient conditions. Loading procedures were undertaken with displacement controlled, and the loading rate was set at 2 mm/s, while the maximum indentation depth was 3 mm. Samples had a diameter of 55 mm and height 15 mm. Six measurements at different positions of the sample were conducted, and the depth-indentation load curves were recorded. The initial shear modulus was determined by fitting the load curves up to different ratios of h/R using the Hertzian solution and the Hyperelastic solution [22], as follows 16 h P  E Rhh (1  0.15 ) 9 R Where E represents the Young’s modulus, P is the indentation load, h is the indentation depth and R the indenter radius. 2.3 Surface topographic characterization of PDMS substrates The PDMS membranes with different stiffness were imaged by atomic force microscopy (AFM) (Nanoscope IIIa, Digital Instruments, Santa Barbara, CA) in tapping mode with 512 × 512 pixel data acquisition. The scan speed was 1 Hz at ambient conditions. The topographic images were recorded with a standard silicon tip on a cantilever beam. The spring constant of the cantilever was 50 pN/nm, and its length was 125 μm, with a resonant frequency of 300 kHz. 2.4 Scanning electron microscope (SEM) test For observation of morphological changes, adipose-derived stromal cells (ASCs) were seeded onto PDMS substrates with different stiffnesses at a lower density (30% confluence) for 3 days. The cells were then fixed by 2.5% glutaraldehyde for 2 h and then dehydrated in a graded ethanol concentration from 50%, 60%, 70%, 80%, 90%, 95%, to 100%. After dehydration, the samples were mounted on specimen holders, coated with a thin layer of gold, and then scanned by scanning electron microscope (SEM). 2.5 Cell culture The animal materials used for this study were obtained according to ethical principles and the protocol was reviewed and approved by our Institutional Review Board (Institutional Review Board at the West China Hospital of Stomatology, No.WCHSIRB-D-2017-029). ASCs were isolated from the subcutaneous adipose tissue located in inguen of 3 weeks old female mice (C57 mice). Briefly, the adipose tissues were cut into small pieces in 15 ml tube (Corning) and digested in 0.075% type I collagenase solution (in α-MEM) for 60 min in an incubator at 37 °C at 40 rpm. Then α-MEM media (0.1 mM

non-essential amino acids, 4 mM L-glutamine, 1% penicillin–streptomycin solution) with 10% heat-activated fatal bovine serum (FBS) were added into the digested buffer at 1 : 1 ratio. Then the tube was shaken to neutralize the mixture and then centrifuged at 1000 rpm for 5 min. After removal of supernatants, remaining ASCs were re-suspended in 10% FBS α-MEM and seeded into plates or T25 flasks, then incubated at 37 °C in a 5% CO2 incubator. Cells were then used to prepare for further experiments until next 2 passages as previously described [23]. The characterization of ASCs was confirmed by stem cell markers [24] and differentiation capacity analyses [25] in our lab. In the current study, we used the purified ASCs after flow sorting based on positive expressions of CD34, CD146, Sca-1 and CD44. 2.6 ALP stain and oil red stain The ASCs were allowed 24 h to seed onto PDMS substrates with stiff/soft stiffnesses. The culture media were then removed and replaced by differentiation media. For osteogenic differentiation, the 10% FBS α-MEM medium should be added with dexamethasone (D8893, Sigma-Aldrich, St. Louis, MO, final concentration: 10-2 μM), β-glycerol phosphate (G9891, Sigma-Aldrich, final concentration: 8 mM) and ascorbic acid (PHR1008, Sigma-Aldrich, final concentration: 50 µg/ml). For adipogenic differentiation, the 10% FBS α-MEM medium should be added with dexamethasone (final concentration: 2.5 × 10-2 μM), indomethacin (I8280, Sigma-Aldrich, final concentration: 5 μM), 3-isobutyl-1-methylxanthine (I7018, Sigma-Aldrich, final concentration: 50 μM) and bovine insulin (I6634, Sigma-Aldrich, final concentration: 10 μg/ml). After 7 day’s induction, the alkaline phosphatase (ALP) stain for osteogenic differentiation and Oil red stain for adipogenic differentiation were performed. ALP stain is based on the Kit from Solarbio (G2450, Shanghai, China). The cells were prepared and stained following the manufacturer’s recommendations. Briefly, after 4% PFA fixation for 10 min, the cells stained by using the ALP reagents for about 30 min at 37 ºC, then washed with ddH2O to remove the rest ALP solution. The cells were then protected in 10% glycerol solution till image taking. Oil red stain is also based on kit from Solarbio (G1260, Shanghai, China). After 7 days’ culture, cells were fixed with 4% PFA for 10 min. Oil red stain was performed according to the manufacturer’s recommendations. 10% glycerol solution was also used for protection. The images were taken by inverted light microscopy (Olympus-IX71, Japan). 2.7 Confocal laser scanning microscope (CLSM) For visualization of cytoskeleton and protein changes in response to different substrate stiffnesses, ASCs were seeded on PDMS substrates with different stiffnesses and equilibrated for 72 h. Then the cells in each group were washed three times using 1 × PBS and fixed in 4% (w/v) paraformaldehyde for 1 h. The fixed cells were washed with 1 × PBS until the smell of paraformaldehyde vanished. Bovine serum albumin (BSA, 5% (w/v)) was applied for 1 h in order to achieve blockage, and staining with FITC-phalloidin (Invitrogen, CA, dissolution at 6.6 μM and incubation at 45 : 1000) or protein antibodies for 1 ~ 2 h at RT (or incubated overnight at 4 °C; the antibodies from Abcam (Cambridge, UK) included anti-alpha tubulin (ab64503), anti-beta tubulin (ab7792), anti-vinculin

(EPR8185), goat anti-mouse IgG (ab150115) and goat anti-mouse IgG (ab150113), and were all incubated at 1 : 200. The antibody from Thermo Fisher Scientific (Waltham, MA) was β-catenin (CAT-5H10, ChiP grade) and incubated at 1 : 400. The next step was the removal of FITC-phalloidin and antibodies and sealing with the block buffer (Invitrogen). The cytoskeletons of cell samples were observed through a confocal laser scanning microscope (A1R MP+, Nikon, Tokyo, Japan, and parameter: 20 ×, Nikon Microsystems original image: 1024 × 1024; and FV3000, Olympus, Tokyo, Japan, and parameter: 40 ×, Nikon Microsystems original image: 1024 × 1024). Cell spreading areas were calculated using IPP (Image Pro Plus 6.0). 2.8 Quantitative real-time PCR (qPCR) RNA samples were isolated with a genomic DNA eliminator. Isolated RNA was dissolved in RNAse-free water and quantified by measuring the absorbance at 260 and 280 nm with a spectrophotometer. The RNA samples were then treated with DNAse I and cDNA was prepared for each sample, using 0.5 mg of total RNA and the cDNA synthesis kit in a final volume of 20 µl. To evaluate the expression levels of adipogenic and osteogenic markers in ASCs in response to substrate stiffnesses, qPCR was performed with SYBR Green PCR Kit using iCycler (Bio-Rad). qPCR primer pairs of these genes were designed using Primer 5.0 (Supplementary table 1). The basic local alignment search tool (BLAST, NCBI) was used to search for all primer sequences to ensure gene specificity. PCR reactions were performed at 0.5 µM for each primer in a 25 µl volume containing 1 µl cDNA sample. The reaction was initiated by activating the polymerase with a 5-min pre-incubation at 95 ºC. Amplification was achieved with 45 cycles of 15 sec denaturation at 94 ºC, 20 sec annealing at 65 ºC and 10 sec extension at 72 ºC. The program was concluded by a melting curve analysis. All experiments were performed in triplicates. The copy numbers of each gene were determined with cycle threshold (△△CT) methods. The means of the copy numbers of GAPDH were used as internal controls. Standard curves of all primers were prepared from total normal cDNA, amplified by semi-quantitative PCR, and cloned using TOPO II TA Cloning Kit following the manufacturer’s recommendations. 2.9 Western blot Protein samples were prepared by mixing one part of tissue sample with one part of Bio-Rad Laemmli sample buffer and then boiled at 100 ºC for 5 min. Proteins were separated in 8 - 12% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) (according to the molecular weights) and transferred to a PVDF membrane at 200 mA for 1 h at room temperature (RT). The blots were blocked with 5% non-fat dry milk suspended in 1 × TBST for 2 h at RT. The resulting blots were incubated with 1 : 500 - 5,000 antibodies (The antibodies from Santa Cruz Biotech (Delaware Avenue, CA) were β-actin (sc-47778, incubation concentration: 1 : 1000), m-IgGκ BP-HRP(sc-516102, incubation concentration: 1 : 4000) and mouse anti-rabbit IgG-HRP(sc-2357, incubation concentration: 1 : 2000. The antibodies from Abcam including anti-vinculin (ab73412), Osx (ab22552), PPARγ (ab45036) and C/EBPα (ab40764) were incubated at 1:2000. The antibodies from Cell Signaling Technology was Lef-1 (C18A7, Rabbit mAb #2286 (ChiP grade), Boston, MA). The antibodies from Thermo

Fisher Scientific (β-catenin, CAT-5H10) was incubated at 1 : 4000 for 2 h at RT or overnight at 4 ºC, followed by incubation with secondary adntibodies from Santa Cruz Biotech for 2 h at RT. Signals from blots were obtained using Santa Cruz Western Blotting Luminol Reagent Kit. Proteins were visualized via chemiluminescence with hydrogen peroxide using Kodak X-AR and luminol as substrate. 2.10 Bioinformatics analysis The Gene information such as Genebank ID and the promoter sequence (~ 2000 bp) before transcriptional starting sites of runt related transcription factor 2 (Runx2), Osterix (Osx, or Sp7 transcription factor 7), peroxisome proliferator activated receptor gamma (PPARγ) and CCAAT/enhancer binding protein alpha (C/EBPα) C/EBPα were

all

came

from

NCBI

resources

(https://www.ncbi.nlm.nih.gov/)

and

BioGPS

(http://biogps.org/#goto=welcome). The binding site sequences predicted at the promoters of genes were achieved through the online tool - PROMO (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3). The detailed binding site information was supplemented in Supplementary material - 3. 2.11 Chromatin Immunoprecipitation (ChIP) Chromatin immunoprecipitation was performed using PierceTM Agarose ChIP Kit (Lot#TA265476). Briefly, ASCs were lysated according to manufacturer. Protein-DNA complexes are stabilized and then extracted following the standard protocol in the Kit. Crosslinking performed directly in cells locks in the protein-DNA complexes, trapping these unstable and sometimes transient interactions. The DNA fragments which was pull out by protein-DNA complexes after overnight incubation with Lef-1 antibody was used as the template for qPCR. The primers targeted for binding sites were designed and the specificity was checked by BLAST tool in NCBI (Supplementary table 2).

2.12 Statistical analysis The analyzed results were all presented as mean ± SEM. Experimental points were performed in triple with a minimum of three independent experiments (n = 3). Statistical analysis was performed by one-way analysis of variance (ANOVA) to determine whether differences existed among groups. Post-hoc analysis utilized Fisher’s protected least significant differences (PLSD). In each analysis, critical significance level was set to be p < 0.05.

3. Results 3.1 Substrate stiffness alters the morphology of adipose-derived stem cells (ASCs) We fabricated two polydimethylsiloxane (PDMS) substrates with curing agent-to-oligomeric base at 1 : 5 and 1 : 45 ratio as previously described [21]. The surface topologies of these two substrates were characterized by AFM (Fig. 1A). The surface roughness was then measured by Ra parameter and it was found the substrates showed significant lower roughness compared to that of normal petri dish (Fig. 1B). However the two PDMS substrates showed significant different stiffnesses by measuring Young's modulus [fig. 1C]. For the stiff substrate the Young's modulus was 1.014 ± 0.15 MPa, while for the soft substrate the Young's modulus was 0.046 ± 0.02 MPa. Because the

spreading area of ASCs in monolayer culture was far broader (>>1μm), we ignored the impact of surface topographic factor and only considered the stiffness of PDMS substrate as the primary physical mechanical stimulation in the current study. We next used SEM to compare the surface changes of the two PDMS substrates (Fig. 2A, upper lane). Also, the SEM results about PDMS surface topologies between the two substrates at micro-scale showed no difference just as the AFM results at nano-scale in Fig. 1A & B. We seeded the ASCs onto the PDMS substrates and found the cell morphological changes after 3 day’s culture (Fig. 2A, lower lane). ASCs on the stiff substrate showed larger spreading area compared to that on the soft substrate. Specifically, the cell body became broader and cell cilia showed more and longer on the stiff substrate relative to those on the soft substrate. The cell spreading area was quantified and shown to be 14% reduction on soft substrate compared to the stiff one (Fig. 2B). 3.2 Substrate stiffness induced the cytoskeleton change of ASCs We then found the changes of cytoskeletons by characterizing the distribution of microfilament and microtubule in ASCs. For microfilament, we found the different distributions of F-actin in ASCs in response to substrate stiffness by CLSM (Fig. 3A). On the stiff substrate, F-actin was found to be brighter and displayed a bundle-like distribution at random directions. But on the soft substrate, F-actin was inclined to be scattered. Through fluorescence optical density (OD) assay, it was found that F-actin showed a higher expression along the border of cell membrane on the stiff substrate. This cell type showed a high population ratio (up to 80%, fig. 3B (1)). But this high expression shown on the stiff substrate was reduced to a moderate level on the soft substrate and the reduction in this cell type on the soft substrate was shown to be a high ratio (up to 85%, fig. 3B (2)). The fluorescent intensity was totally reduced to 68% on the soft substrate relative to that on the stiff substrate (Fig. 3C). For microtubule, we found the different distributions of α- and β-tubulin in ASCs in response to substrate stiffness by CLSM (Fig. 4A). The distributions of α- and β-tubulin in ASCs on the stiff substrate were shown to be much broader due to the larger cell spreading area and were even great brighter than those on the soft substrate. Fluorescence OD assay further confirmed the distribution changes of α- and β-tubulin between the two PDMS substrates (Fig. 4B). The fluorescent intensity was totally reduced to 53% on the soft substrate relative to that on the stiff substrate (Fig. 4C). 3.3 Substrate stiffness enhances adhesion force by promoting high vinculin expression in ASCs on the stiff substrate relative to the soft substrate We next explored the expressions of vinculin, an important focal adhension protein, in ASCs in response to the different substrate stiffnesses. By CLSM, it was found that vinculin was shown to be spot-like expression and distributed along the border of cell membrane especially at the end of bundle-like actin microfilament (fig. 5A). On the stiff substrate, the spot-like expression was shown to be more and brighter compared to that on the soft substrate. Fluorescence OD measurement quantified the different distributions of vinculin in nucleus in ASCs on the stiff and soft PDMS substrates (Fig. 5B). To further confirm the changes of vinculin, western blot was

performed and it was indicated that expression of vinculin in ASCs in response to the stiff substrate was higher than that in response to the soft substrate (Fig. 5C). The higher expression of vinculin on the stiff substrate was up to 144% relative to that on the soft substrate (Fig. 5D). 3.4 Substrate stiffness triggered the expression of cytoplasmic β-catenin and promotes its nuclear translocation. Substrate stiffness not only triggered cell-matrix molecular variations and cytoskeleton changes but also induced the alteration of intracellular events. We had screened the cytoplasmic signal proteins in response to substrate stiffness and found Wnt/β-catenin was dictated by extracellular matrix stiffness through integrin/FAK pathway in chondrocyte and bone marrow mesenchymal stem cells (BMSCs) [26]. In ASCs, we deduced that β-catenin might be a potential target signal followed by the changes of focal adhension proteins in response to substrate stiffness. ASCs were allowed 24 h to seed onto the stiff and soft substrates with 10% FBS α-MEM. Then the culture medium was changed to fresh osteogenic differentiation medium. After 3 day’s induction, it was found that β-catenin was expressed in all cytoplasm of ASCs; moreover, on the stiff substrate the increased β-catenin was mainly accumulated in nuclear region (Fig. 6A). The further fluorescence OD quantification confirmed these results (Fig. 6B). The enhanced distributions of vinculin in nucleus in ASCs was shown on the stiff PDMS substrate and this enhanced distribution showed a high cell population ratio (up to 71%, fig. 3B (1)), while, on the soft substrate the distributions of vinculin in nucleus was decreased and this decrease in cell type also held a high population ratio (up to 76%, fig. 3B (2)). The cells with increase of β-catenin in nuclei through translocation totally had a higher population ratio on the stiff substrate than these on the soft one (Fig. 6C). The cell population with nuclear translocation of β-catenin was up to 86% on the stiff substrate but down to 27% on the soft substrate. We used western blot and further confirmed the increase of β-catenin in ASCs in response to the stiff substrate relative to the soft substrate (Fig. 6D). The increase on the stiff substrate was up to 35% compared to the soft substrate after quantification (fig. 6E). In addition, when we used adipogenic differentiation media, the data about nuclear translocation of β-catenin showed the same trend. 3.5 The stiff substrate enhances osteogenic differentiation but impairs adipogenic differentiation of ASCs through transcriptional factors. Finally, we explored the differentiation towards osteogenesis and adipogenesis of ASCs in response to different substrate stiffnesses. For osteogenesis, we found the stiff substrate could enhance the directed differentiation to osteogenic lineages by ALP stain (Fig. 7A). This enhancement was accompanied with the up-regulated transcriptional factors, i.e., Runx2 and Osx, which were recognized to play an important role in directing the osteogenic lineage commitment [27] (Fig. 7B). In order to elucidate the influence of β-catenin signaling on osteogenic differentiation, the activator (LiCl) and inhibitor (DKK1) of canonical β-catenin signaling were used. It was found that the activator group showed enhanced osteogenesis by ALP while the inhibitor group showed decreased osteogenesis in both the stiff and soft substrates (Fig. 7C), but on the stiff substrate the

osteogenesis was shown to be stronger relative to that on the soft substrate. We then found that the activator induced the protein increases of Runx2 and Osx, while the inhibitor induced the protein decreases of Runx2 and Osx (Fig. 7D). In order to establish the direct influence of β-catenin signaling on osteogenesis, we next use bioinformatics and found the potential binding sites of β-catenin signaling-induced downstream protein, lymphoid enhancer-binding factor (lef-1), was at the promoter regions of both Runx2 and Osx (Fig. 7E). ChIP qPCR was performed to find out the actual binding site of Lef-1 at -450~-442 bp before the transcriptional starting site (TSS) of Runx2 (Fig. 7F) and at -1279~-1271bp before the TSS of Osx (Fig. 7G). The agarose gels from the final products of ChIP qPCR were further confirmed the results (Fig. 7 H & I). For adipogenesis, we found the soft substrate could enhance the directed differentiation to adipogenic lineages by Oil red stain (Fig. 8A). This enhancement on the soft substrate was accompanied with the up-regulated transcriptional factors, i.e., PPARγ and C/EBPα, which were considered to direct the adipogenic lineage commitment [28] (Fig. 8B). We then found the activator (LiCl) of β-catenin signaling could reduce adipogenic differentiation and the inhibitor (DKK1) could increase adipogenic differentiation of ASCs in both stiff and soft substrates (Fig. 8C). The enhanced adipogenic differentiation by DKK1 on the soft substrate was accompanied with the increase of PPARγ but not C/EBPα, and also the reduction of adipogenic differentiation of ASCs by LiCl on the stiff substrate was accompanied with the decrease of PPARγ but not C/EBPα (Fig. 8D). To further try to establish the direct influence of β-catenin signaling on adipogenesis, we next found the two potential binding sites of lef-1 at the promoter regions of both PPARγ and C/EBPα by bioinformatics (Fig. 8E). But the ChIP qPCR did not detected the direct binding sites (Fig. 8 F&G). The agarose gels from the final products of ChIP qPCR were further confirmed the negative results (Fig.8 H&I).

4. Discussion It is generally accepted that (A) stem cells can differentiate into multiple lineages through biophysical properties of the extracellular matrix including substrate elasticity and geometry; (B) adult cells can modulate their phenotypes and cell functionalizations in response to these biophysical properties. Although the multitude of literature on their differentiation capacity is accumulating, little is known about the precise biophysical signal mechanisms that enable stem cell differentiation triggered by matrix biophysical properties. In the current study, we found the changes of microfilaments and microtubules, and focal adhension adaptor, vinculin, in charge for the linkage of adhesion molecules to the actin cytoskeleton, in ASCs when the matrix elasticity were closed to in vivo trabecular bone and even cortical bone. The change in focal adhesion plaque triggered intracellular β-catenin signaling and promoted its nuclear translocation especially on the stiff substrate. Nuclear translocation of β-catenin signaling up-regulates its downstream protein, Lef-1, expression. Lef-1 binds at -450~-442 bp before the transcriptional starting site (TSS) of Runx2 and at -1279~-1271bp before the TSS of Osx, thus enhancing the directed differentiation to osteogenic lineages. Meanwhile, impaired nuclear translocation of β-catenin signaling on the soft substrate reduced Lef-1

expression but increased PPARγ expression, which resultantly enhance the directed differentiation to adipogenic lineages. This study provided an explanation on the mechanism underlying β-catenin signal transduction from mechanosensing, mechanotransducing to stem cell differentiation. The polydimethylsiloxane (PDMS) has been considered to be a principal candidate in many biomaterial studies due to its biocompatibility, flexibility, optical clarity and elastic tenability [20]. PDMS has been integrated into microtechnology biological studies, for example, geometric control of cell life and death [29], high-throughput bioreactor assays based on microfluidic cell culture [30], Subcellular positioning of small molecules under biochemical stimuli [31], single cell assays by microfluidic device [32], immunoassays [33] and stem cell research [34]. In the current study, PDMS were fabricated at 1 : 5 and 1 : 45 ratio (curing agent to oligomeric base) to mimic the stiff and soft substrates in testing the ASC response. There are some major factors, such as material surface roughness and hydrophilicity, influencing the cell responses. For the surface roughness, previous published papers have been shown that surface roughness can influence cell functions, such as cell migration [35], cell proliferation [63] and differentiation [64]. Particularly, for osteogenesis, the surface roughness showed to be one of the most important parameters governing osteointegration [65]. In that regard, roughness topography could further mimic the physical cue on the bone surface and showed great importance in osteoclast activity during bone resorption [66]. However, these surface roughnesses above are based on the range from the sub-micron to the micrometer (Ra, from at least 0.3 μm to 10 μm). In addition, for the researches about surface roughness gradients, Tobias et al used the surface roughness at ≥ 3μ [67] and Faia-Torres et al used the surface roughness at ≥ 500 nm [68]. In 2011, Bigerelle et al began to try to find out the threshold of surface roughness on influencing mesenchymal stem cells, and the range of roughnesses was limited to 1.2 μm ~ 21 μm (Ra) [69]. At nano scale, surface steps as small as 11 nm (Ra) could lead to contact guidance [70] and surface nanoposts as small as 13 nm (Ra) could increase cell spreading, proliferation and cytoskeletal formation [71]. But there has been no report involving in Ra below 10 nm (< 10 nm). It is supposed that the surface roughness as small as 10 nm could have little effect on cell behavior [35, 72]. In our Fig. 1B, we first showed the surface roughness was ~10 nm (Ra) in the petri dish and then found the surface roughness in soft and stiff PDMS substrates was smaller than that in the petri dish. It means that the impact of surface roughness in our manuscript could be ignored. As to the surface hydrophilicity, it also plays a determinant role in the supramolecular organization and function of adsorbed protein layers [73]. Although the atomically smooth substrate of PDMS were widely used for the cell-interface (material) interaction, it was actually more hydrophobic compared to most kinds of hydrogels. Thus, the substrates with varied surface roughness but the same stiffness could potentially differ with respect to the bioactivity of an adsorbed protein layer. This important confounding factor appears to be either universally unappreciated or discounted without appropriate controls. A recent report showed that the enhanced osteogenic behavior was observed on more hydrophilic surfaces of PDMS, with an associated increase in activation of α1β1 integrins and DDR1 receptors that ultimately affects the ERK/MAPK pathway [74]. In the current study, we did not consider the surface hydrophilicity of stiff and soft

substrates by using the magnetron sputtering method (physical approach), although the surface hydrophilicity may show impact on ASCs. The interface between cells and their extracellular matrix (ECM) is a complex environment where cells can sense the biophysical properties of ECM such as stiffness and geometry and respond to these cues by changing their behaviors such as adhesion, migration, proliferation, and differentiation [27, 36]. For adult cells, accumulating evidence indicated that substrate stiffness mimicking the microenvironment of different tissues, including neurons, adipose tissue, and cartilage, could modulate the phenotype changes and cell functionalization [37]. For example, Smith et al. revealed that embryonic myocardial cells were able to enhance their self-renewal on matrices whose elastic modulus matches the stiffness of cardiac tissue [38]. Further, Gilbert et al. reported that stiffness cues increased the expression of endothelial markers and enhanced the survival of cardiosphere derived cells [39]. For stem cells, they could differentiate into multiple lineages through biophysical stiffness. Engler et al. demonstrated that exposure of bone marrow mesenchymal stem cells (BMSCs) to materials of low, intermediate, and high stiffness directed cell commitment into adipogenic, myogenic, and osteogenic lineages, respectively, based on polyacrylamide gels [12]. In the current study, we used PDMS substrates, which were shown to have broad range of Young's moduli, instead of polyacrylamide gel substrates. The Young's modulus of “soft” PDMS substrate was close to that of the “stiff” polyacrylamide gel in the Engler paper. In this case, we also showed the “soft” PDMS substrate (0.046 ± 0.02 MPa) could enhance directed differentiation to osteogenic lineages (the red stain in Fig. 7 A-soft substrate) by induction culture media. Furthermore, in a far stiffer substrate (1.014 ± 0.15 MPa), of which the Young's modulus could be reached to the trabecular bone (0.8 ~ 2.5 MPa) and even cortical bone (up to 10 MPa) [58, 59], the enhancement was further significantly shown. As to adipogenesis, on the “stiff” PDMS substrate the oil particles were shown to be fewer, and in the “soft” PDMS substrate the oil particles were shown to be more and bigger. Thus, based on these selections, it inferred that a stiff or a soft PDMS substrate may enhance directed differentiation to osteogenic or adipogenic lineages, respectively. With the rapid advance of knowledge concerning the relationship between matrix stiffness and cell phenotype, to further dig out the precise biophysical mechanisms that enable cells to respond to matrix stiffness seems to be of great significance. The mechanism from active mechanosensing, mechanotransducing, to final cell fate decision is a complicated processing [40]. Once the mechanical stimulus or its lack thereof is sensed by cells, a cascade of signaling events could take place in order to generate biological responses eventually leading to the changes of cell fate and function [41]. Previous literature focused on the scattered aspects or components in this process, such as the individual proteins, e.g., focal adhesions (FAs) which were located at discrete regions of a cell that provide sites for mechanical attachment to the ECM [42] and contractile stress fibers (SFs) [43]. Coupling between the extracellular environment and the cellular interior events occurs via multiprotein transmembrane complexes that are mainly based on integrins, receptor tyrosine kinases, and cadherins (Fig. 9) [44]. The attachment of FAs to the substrate is mediated by integrins. Then integrin-mediated-adhesions forms multiprotein complexes that link the

extracellular matrix to the actin cytoskeleton [45]. This is one way to directly achieve the changes of cell morphology and cytoskeleton in response to substrate stiffness. Further, formation of stable adherens junctions requires coupling of the cadherin intracellular domain via a cytoplasmic protein complex to the actin cytoskeleton [46]. It forms another way to directly achieve the changes of cytoskeleton in response to substrate stiffness. In the cytoplasmic events, the β-catenin at inner surface of cell membrane binds to α-catenin, which also links the complex to the actin cytoskeleton and recruits actin-remodeling molecules [47]. It forms the third way to directly achieve the changes of cytoskeleton. Meanwhile, the β-catenin, as an active signal transducer in cell cytoplasm, can further dissociate from the cadherin/catenin cell membrane complex and translocates into cell nuclei where it acts as a transcriptional coactivator binding with the members of the T cell factor/lymphoid enhancer factor (TCF/LEF) transcription factor family [48]. This process of β-catenin translocation has been verified in chemical biology such as adult cell function [49], tissue regeneration [50] and tumor metastasis [51]. For mechanotransducing in adult chondrocytes, we had previously shown that activation of β-catenin by stiff ECM was not dependent on Wnt signals but was elevated by the activation of integrin/focal adhesion kinase (FAK) pathway. The accumulated β-catenin then bound to the wnt1 promoter region to up-regulate the gene transcription, thus constituting a positive feedback of the Wnt/β-catenin pathway [26]. Herein in the current study, we first observed the changes of cell spreading area by SEM (Fig. 2A) and cell cytoskeleton of ASCs cultured on substrates with different stiffnesses by CLSM through characterizing microfilament and microtubule (Fig. 3 & 4). These results were consistent with other reports in MSCs [28, 52]. We then detected the changes of vinculin, a membrane-cytoskeletal protein in focal adhesion plaques that is involved in linkage of integrin adhesion molecules to the actin cytoskeleton [53] and in intracellular signal transduction [54]. And more importantly, cytoplasmic β-catenin signaling was consequently activated and resulted in partial nuclear translocation (Fig. 6). We provided the direct evidence that downstream protein, Lef-1, of β-catenin signaling could bind to the promoters of transcriptional factor genes, Runx and Osx, responsible for osteogenesis (Fig. 7 C-I). On the other hand, although we did not detect the direct binding between Lef-1 and transcriptional factors, PPARγ and C/EBPα, we found the expression of PPARγ could be induced by the activator and inhibitor of β-catenin signaling, and thus modulate direct differentiation to adipogenic lineages (Fig. 8 C-I) (The direct relationship between β-catenin signaling and transcriptional factors, PPARγ and C/EBPα, may not be from Lef-1 but TCF-1 as some references inferred [60-62]). Finally, from this signal cascade (Fig. 9) we herein provide a potential mechanism through β-catenin transduction in the explanation for differentiation of ASCs towards osteogenesis and adipogenesis in response to different substrate stiffnesses. We admit some limitations in this study. Firstly, as the properties of ASCs are shown to be depot-specific differences in genetic, biochemical, and metabolic endpoints [55], the ASCs in this study were isolated from subcutaneous adipose of mouse groin, thus cannot be fully represent the general properties of ASCs and might show different capacity towards osteogenesis and adipogenesis relative to the adipose tissues from other parts of mouse body. The second limitation is that primary ASCs are isolated from subcutaneous adipose and cannot reach

absolutely high purified ASCs. The other residents such as preadipocytes, fibroblasts, immune cells and other multipotent stem cells might be presented. The third limitation is that we did not detect the influence of surface hydrophilicity of stiff and soft substrates on ASCs. The fourth limitation is that we provide a signal cascade from primary mechanosensing to final differentiation towards osteogenesis and adipogenesis through cytoplasmic β-catenin signaling in mouse ASCs. In the literature, there is conflicting evidence for the effect of Wnt signaling on osteogenic differentiation in human adult stem cells, while this study is performed exclusively in rodent cells. Regarding to the applicability/significance of this work which shows potential in translating to that of human ASCs, the further confirmation work has to be considered. The fifth limitation is that there are several important multiprotein transmembrane complexes formed by substrate stiffness stimulus (fig. 9). The mechanotransducing signals are not limited to Wnt/β-catenin but to the other signal pathways such as Notch [56] and RhoA/ROCK pathway [57]. The differentiation of ASCs towards osteogenesis and adipogenesis are believed to be modulated through a complicated signaling network. These limitations have to be considered in the further work when the differentiation of ASCs is involved.

Acknowledgements This study was supported by the funding of NSFC grants (81371136, 81430011), JCPT2011-9 to Dr. Zhou XD, the funding of NSFC grant (81600840, 81771047) to Dr. Jing Xie. We acknowledged Dr Chenghui Li in the Analytical & Testing Center of Sichuan University for her Support in CLSM imaging (A1R MP+, Nikon, Tokyo, Japan).

Contributions Jing Xie and Xuedong Zhou designed the study; Jing Xie and Demao Zhang collected data; Jing Xie, Lin Ye and Quan Yuan analyzed and interpreted the data; Jing Xie, Demao Zhang and Chenchen Zhou drafted the manuscript; Xuedong Zhou made a critical revision and final approved the manuscript.

Competing interests The authors declare that no competing interests exist.

References [1] X. Jiang, S. Fitch, C. Wang, C. Wilson, J Li, G.A. Grant, F. Yang, Nanoparticle engineered TRAIL-overexpressing adipose-derived stem cells target and eradicate glioblastoma via intracranial delivery, Proc.

Natl. Acad. Sci. USA. 113 (2016) 13857-13862. [2] R.D. González-Cruz, V.C. Fonseca, E.M. Darling, Cellular mechanical properties reflect the differentiation potential of adipose-derivedmesenchymal stem cells, Proc. Natl. Acad. Sci. USA. 109 (2012) E1523-E1529. [3] L. Moradi, M. Vasei, M.M. Dehghan, M. Majidi, S. Farzad Mohajeri, S. Bonakdar, Regeneration of meniscus tissue using adipose mesenchymal stem cells-chondrocytes co-culture on a hybrid scaffold: In vivo study, Biomaterials 126 (2017) 18-30. [4] X. Zhang, M.P. Bendeck, C.A. Simmons, J.P. Santerre, Deriving vascular smooth muscle cells from mesenchymal stromal cells: Evolving differentiation strategies and current understanding of their mechanisms, Biomaterials 145 (2017) 9-22. [5] P. Gentile, M.G. Scioli, A. Bielli, A. Orlandi, V. Cervelli, Concise Review: The Use of Adipose-Derived Stromal Vascular Fraction Cells and Platelet Rich Plasma in Regenerative Plastic Surgery, Stem. Cells. 35 (2017) 117-134. [6] H. Mizuno, M. Tobita, A.C. Uysal, Concise review: Adipose-derived stem cells as a novel tool for future regenerative medicine, Stem. Cells. 30 (2012) 804-810. [7] P. Gentile, M.G. Scioli, A. Bielli, A. Orlandi, V. Cervelli, Concise Review: The Use of Adipose-Derived Stromal Vascular Fraction Cells and Platelet Rich Plasma in Regenerative Plastic Surgery, Stem. Cells. 35 (2017) 117-134. [8] J. Du, X. Chen, X. Liang, G. Zhang, J. Xu, L. He, Q. Zhan, X.Q. Feng, S. Chien, C. Yang, Integrin activation and internalization on soft ECM as a mechanism of induction of stem cell differentiation by ECM elasticity, Proc. Natl. Acad. Sci. USA. 108 (2011) 9466-9471. [9] I. Dinulovic, R. Furrer, C. Handschin, Plasticity of the Muscle Stem Cell Microenvironment, Adv. Exp. Med. Biol. 1041 (2017) 141-169. [10] J.L. Allen, M.E. Cooke, T. Alliston, ECM stiffness primes the TGFβ pathway to promote chondrocyte differentiation, Mol. Biol. Cell. 23 (2012) 3731-3742. [11] V.V. Hiew, S.F.B. Simat, P.L. Teoh, The Advancement of Biomaterials in Regulating Stem Cell Fate, Stem. Cell. Rev. 14(2018) 43-57. [12] A.J. Engler, S. Sen, H.L. Sweeney, D.E. Discher, Matrix elasticity directs stem cell lineage specification, Cell 126 (2006) 677-689. [13] W. Duan, M. Haque, M.T. Kearney, M.J. Lopez, Collagen and Hydroxyapatite Scaffolds Activate Distinct Osteogenesis Signaling Pathways in Adult Adipose-Derived Multipotent Stromal Cells, Tissue. Eng. Part. C. Methods. 23(2017) 592-603. [14] S. Lechner, M.C. Mitterberger, M. Mattesich, W. Zwerschke, Role of C/EBPβ-LAP and C/EBPβ-LIP in early adipogenic differentiation of human white adipose-derived progenitors and at later stages in immature adipocytes, Differentiation 85 (2013) 20-31. [15] A.G. Cristancho, M.A. Lazar, Forming functional fat: a growing understanding of adipocyte differentiation. Nat. Rev. Mol. Cell. Biol. 12 (2011) 722-734.

[16] Y.S. Choi, L.G. Vincent, A.R. Lee, M.K. Dobke, A.J. Engler, Mechanical derivation of functional myotubes from adipose-derived stem cells, Biomaterials 33(2012):2482-2491. [17] K.A. Kilian, B. Bugarija, B.T. Lahn, M. Mrksich, Geometric cues for directing the differentiation of mesenchymal stem cells, Proc. Natl. Acad. Sci. USA. 107 (2010) 4872–4877. [18] K.A. Kilian, M. Mrksich, Directing stem cell fate by controlling the affinity and density of ligand-receptor interactions at the biomaterials interface, Angew. Chem. Int. Ed. Engl. 51 (2012) 4891-4895. [19] J. Lee, A.A. Abdeen, A.S. Kim, K.A. Kilian, Influence of biophysical parameters on maintaining the mesenchymal stem cell phenotype, ACS Biomater. Sci. Eng. 1 (2015) 218–226. [20] S.Y. Chou, C.M. Cheng, P.R. Leduc, Composite polymer systems with control of local substrate elasticity and their effect on cytoskeletal and morphological characteristics of adherent cells, Biomaterials 30(2009) 3136-3142. [21] J. Xie, Q. Zhang, T. Zhu, Y. Zhang, B. Liu, J. Xu, H. Zhao, Substrate stiffness-regulated matrix metalloproteinase output in myocardial cells and cardiac fibroblasts: implications for myocardial fibrosis, Acta. Biomater. 10 (2014) 2463-2472. [22] M.G. Zhang, Y.P. Cao, G.Y. Li, X.Q. Feng, Spherical indentation method for determining the constitutive parameters of hyperelastic soft materials, Biomech. Model. Mechanobiol. 13(2014) 1-11. [23] Q. Xie, J. Xie, J. Zhong, X. Cun, S. Lin, Y. Lin, X. Cai, Hypoxia enhances angiogenesis in an adipose-derived stromal cell/endothelial cell co-culture 3D gel model, Cell. Prolif. 49 (2016) 236-245. [24] Y. Yue, X. Yang, X. Wei, J. Chen, N. Fu, Y. Fu, K. Ba, G. Li, Y. Yao, C. Liang, J. Zhang, X. Cai, M. Wang, Osteogenic differentiation of adipose-derived stem cells prompted by low-intensity pulsedultrasound, Cell. Prolif. 46 (2013) 320-327. [25] X. Cai, J. Xie, Y. Yao, X. Cun, S. Lin, T. Tian, B. Zhu, Y. Lin, Angiogenesis in a 3D model containing adipose tissue stem cells and endothelial cells is mediated by canonical Wnt signaling, Bone. Res. 5 (2017) 17048. [26] J. Du, Y. Zu, J. Li, S. Du, Y. Xu, L. Zhang, L. Jiang, Z. Wang, S. Chien, C. Yang, Extracellular matrix stiffness dictates Wnt expression through integrin pathway, Sci. Rep. 6 (2016) 20395. [27] J. Lee, A.A. Abdeen, X. Tang, T.A. Saif, K.A. Kilian, Matrix directed adipogenesis and neurogenesis of mesenchymal stem cells derived from adiposetissue and bone marrow, Acta. Biomater. 42 (2016) 46-55. [28] R. Izadpanah, C. Trygg, B. Patel, C. Kriedt, J. Dufour, J.M. Gimble, B.A. Bunnell, Biologic properties of mesenchymal stem cells derived from bone marrow and adipose tissue. J. Cell. Biochem. 99 (2006) 1285-1297. [29] C.S. Chen, M. Mrksich, S. Huang, G.M. Whitesides, D.E. Ingber, Geometric control of cell life and death, Science 276 (1997) 1425-1428. [30] P.J. Hung, P.J. Lee, P. Sabounchi, R. Lin, L.P. Lee, Continuous perfusion microfluidic cell culture array for high-throughput cell-based assays, Biotechnol. Bioeng. 89 (2005) 1-8. [31] S. Takayama, E. Ostuni, P. LeDuc, K. Naruse, D.E. Ingber, G.M. Whitesides, Subcellular positioning of small molecules, Nature 411 (2001) 1016.

[32] A.R. Wheeler, W.R. Throndset, R.J. Whelan, A.M. Leach, R.N. Zare, Y.H. Liao, K. Farrell, I.D. Manger, A. Daridon, Microfluidic device for single-cell analysis, Anal. Chem. 75 (2003) 3581-3586. [33] K. Sato, M. Yamanaka, H. Takahashi, M. Tokeshi, H. Kimura, T. Kitamori, Microchip-based immunoassay system with branching multichannels for simultaneousdetermination of interferon-gamma, Electrophoresis 23 (2002) 734-739. [34] J.Y. Park, S.K. Kim, D.H. Woo, E.J. Lee, J.H. Kim, S.H. Lee, Differentiation of neural progenitor cells in a microfluidic chip-generated cytokine gradient, Stem Cells 27 (2009) 2646-2654. [35] A. Mittal, R. Kumar, D. Parsad, N. Kumar, Cytomodulin-functionalized porous PLGA particulate scaffolds respond better to cell migration, actin production and wound healing in rodent model, J. Tissue. Eng. Regen. Med. 8 (2014) 351-363. [36] W.L. Murphy, T.C. McDevitt, A.J. Engler, Materials as stem cell regulators, Nat. Mater. 13 (2014) 547-557. [37] R.C. Andresen Eguiluz, K.B. Kaylan, G.H. Underhill, D.E. Leckband, Substrate stiffness and VE-cadherin mechano-transduction coordinate to regulate endothelialmonolayer integrity, Biomaterials 140 (2017) 45-57. [38] A.W. Smith, J.D. Hoyne, P.K. Nguyen, D.A. McCreedy, H. Aly, I.R. Efimov, S. Rentschler, D.L. Elbert, Direct reprogramming of mouse fibroblasts to cardiomyocyte-like cells using Yamanaka factorson engineered poly(ethylene glycol) (PEG) hydrogels, Biomaterials 34 (2013) 6559-6571. [39] P.M. Gilbert, K.L. Havenstrite, K.E. Magnusson, A. Sacco, N.A. Leonardi, P. Kraft, N.K. Nguyen, S. Thrun, M.P. Lutolf, H.M. Blau, Substrate elasticity regulates skeletal muscle stem cell self-renewal in culture, Science 329 (2010) 1078-1081. [40] Z. Razinia, T. Mäkelä, J. Ylänne, D.A. Calderwood, Filamins in mechanosensing and signaling, Annu. Rev. Biophys. 41 (2014) 227-246. [41] M. Maycas, P. Esbrit, A.R. Gortázar, Molecular mechanisms in bone mechanotransduction, Histol. Histopathol. 32 (2017) 751-760. [42] A.P. Gilmore, K. Burridge, Molecular mechanisms for focal adhesion assembly through regulation of protein-proteininteractions, Structure 4 (1996) 647-651. [43] P. Hotulainen, P. Lappalainen, Stress fibers are generated by two distinct actin assembly mechanisms in motile cells, J. Cell. Biol. 173 (2006) 383-394. [44] Y.L. Dorland, S. Huveneers, Cell-cell junctional mechanotransduction in endothelial remodeling. Cell. Mol. Life. Sci. 74(2017):279-292. [45] B. Chen, B. Ji, H. Gao, Modeling Active Mechanosensing in Cell-Matrix Interactions. Annu. Rev. Biophys. 44 (2015) 1-32. [46] D. Gulino-Debrac, Mechanotransduction at the basis of endothelial barrier function, Tissue. Barriers. 1(2013) e24180. [47] F. Meng, Y. Rui, L. Xu, C. Wan, X. Jiang, G. Li, Aqp1 enhances migration of bone marrow mesenchymal stem

cells through regulation of FAK and β-catenin, Stem. Cells. Dev. 23 (2014) 66-75. [48] M. Polette, M. Mestdagt, S. Bindels, B. Nawrocki-Raby, W. Hunziker, J.M. Foidart, P. Birembaut, C. Gilles, Beta-catenin and ZO-1: shuttle molecules involved in tumor invasion-associated epithelial-mesenchymal transition processes, Cells. Tissues. Organs. 185 (2007) 61-65. [49] S.P. Monga, Role and regulation of β-catenin signaling during physiological liver growth, Gene. Expr. 16 (2014):51-62. [50] B.T. MacDonald, K. Tamai, X. He, Wnt/beta-catenin signaling: components, mechanisms, and diseases. Dev. Cell. 17 (2009) 9-26. [51] A. Berthon, A. Martinez, J. Bertherat, P. Val, Wnt/β-catenin signalling in adrenal physiology and tumour development, Mol. Cell. Endocrinol. 351 (2012) 87-95. [52] M. Maldonado, L.Y. Wong, C. Echeverria, G. Ico, K. Low, T. Fujimoto, J.K. Johnson, J. Nam, The effects of electrospun substrate-mediated cell colony morphology on the self-renewal of human induced pluripotent stem cells, Biomaterials 50(2015) 10-19. [53] P. Atherton, B. Stutchbury, D. Jethwa, C. Ballestrem, Mechanosensitive components of integrin adhesions: Role of vinculin, Exp. Cell. Res. 343(2016) 21-27. [54] J. Zhang, C.H. Wong, W. Xia, D.D. Mruk, N.P. Lee, W.M. Lee, C.Y. Cheng, Regulation of Sertoli-germ cell adherens junction dynamics via changes in protein-protein interactions of the N-cadherin-beta-catenin protein complex which are possibly mediated by c-Src and myotubularin-related protein 2: an in vivo study using an androgen suppression model, Endocrinology 146 (2005) 1268-1284. [55] A. Cignarelli, S. Perrini, R. Ficarella, A. Peschechera, P. Nigro, F. Giorgino, Human adipose tissue stem cells: relevance in the pathophysiology of obesity and metabolicdiseases and therapeutic applications, Expert. Rev. Mol. Med. 14 (2012) e19. [56] W.R. Jiang, G. Cady, M.M. Hossain, Q.Q. Huang, X. Wang, J.P. Jin, Mechanoregulation of h2-calponin gene expression and the role of Notch signaling. J. Biol. Chem. 289 (2014) 1617-1628. [57] T. Ebata, Y. Mitsui, W. Sugimoto, M. Maeda, K. Araki, H. Machiyama, I. Harada, Y. Sawada, H. Fujita, H. Hirata, K. Kawauchi, Substrate Stiffness Influences Doxorubicin-Induced p53 Activation via ROCK2 Expression, Biomed. Res. Int. 2017 (2017) 5158961. [58] X.S. Liu, P. Sajda, P.K. Saha, F.W. Wehrli, X.E. Guo, Quantification of the roles of trabecular microarchitecture and trabecular type in determining the elastic modulus of human trabecular bone, J. Bone. Miner. Res. 21 (2006) 1608-16017. [59] G. Unnikrishnan, C. Xu, K.L. Popp, J.M. Hughes, A. Yuan, K.I. Guerriere, S. Caksa, K.E. Ackerman, M.L. Bouxsein, J. Reifman, Regional variation of bone density, microarchitectural parameters, and elastic moduli in the ultradistal tibia of young black and white men and women, Bone. 112 (2018) 194-201. [60] N.P. Hoverter, M.D. Zeller, M.M. McQuade, A. Garibaldi, A. Busch, E.M. Selwan, K.J. Hertel, P. Baldi, M.L.

Waterman, The TCF C-clamp DNA binding domain expands the Wnt transcriptome via alternative target recognition, Nucleic. Acids. Res. 42 (2014) 13615-13632. [61] E. Lambertini, T. Franceschetti, E. Torreggiani, L. Penolazzi, A. Pastore, S. Pelucchi, R. Gambari, R. Piva, SLUG: a new target of lymphoid enhancer factor-1 in human osteoblasts, BMC. Mol. Biol. 11 (2010) 13. [62] M. Dose, A.O. Emmanuel, J. Chaumeil, J. Zhang, T. Sun, K. Germar, K. Aghajani, E.M. Davis, S. Keerthivasan, A.L. Bredemeyer, B.P. Sleckman, S.T. Rosen, J.A. Skok, M.M. Le Beau, K. Georgopoulos, F. Gounari, β-Catenin induces T-cell transformation by promoting genomic instability, Proc. Natl. Acad. Sci. U S A. 111 (2014) 391-396. [63] T.P. Kunzler, T. Drobek, M. Schuler, N.D. Spencer, Systematic study of osteoblast and fibroblast response to roughness by means of surface-morphology gradients, Biomaterials 28 (2007) 2175-2182. [64] A. Martins, M.L. Alves da Silva, Faria S, A.P. Marques, R.L. Reis, N.M. Neves, The influence of patterned nanofiber meshes on human mesenchymal stem cell osteogenesis, Macromol. Biosci. 11 (2011) 978-987. [65] C.N. Elias, Y. Oshida, J.H. Lima, C.A. Muller, Relationship between surface properties (roughness, wettability and morphology) of titanium and dental implant removal torque, J. Mech. Behav. Biomed. Mater. 1 (2008) 234-342. [66] T. Hefti, M. Frischherz, N.D. Spencer, H. Hall, F. Schlottig, A comparison of osteoclast resorption pits on bone with titanium and zirconia surfaces, Biomaterials 31 (2010) 7321-7331. [67] P.K. Tobias, D. Tanja, M.S. Christoph, S. Martin, D.S. Nicholas, Fabrication of material-independent morphology gradients for high-throughput applications, Appl. Surf. Sci. 253 (20064) 2148-2153. [68] A.B. Faia-Torres, S. Guimond-Lischer, M. Rottmar, M. Charnley, T. Goren, K. Maniura-Weber, N.D. Spencer, R.L. Reis, M. Textor, N.M. Neves, Differential regulation of osteogenic differentiation of stem cells on surface roughness gradients, Biomaterials 35 (2014) 9023-9032. [69] M. Bigerelle, S. Giljean, K. Anselme, Existence of a typical threshold in the response of human mesenchymal stem cells to a peak and valley topography, Acta. Biomater. 7 (2011) 3302-3311. [70] A. Curtis, C. Wilkinson, New depths in cell behaviour: reactions of cells to nanotopography. Biochem. Soc. Symp. 65 (1999) 15-26. [71] M.J. Dalby, M.O. Riehle, D.S. Sutherland, H. Agheli, A.S. Curtis, Use of nanotopography to study mechanotransduction in fibroblasts--methods and perspectives. Eur. J. Cell. Biol. 83 (2004) 159-169. [72] V. Vogel, M. Sheetz, Local force and geometry sensing regulate cell functions. Nat. Rev. Mol. Cell. Biol. 7 (2006) 265-275. [73] C.C. Dupont-Gillain, I. Jacquemart, P.G. Rouxhet, Influence of the aggregation state in solution on the supramolecular organization of adsorbed type I collagen layers. Colloids. Surf. B Biointerfaces. 43 (2005) 179-186. [74]. T. Razafiarison, U. Silván, D. Meier, J.G. Snedeker, Surface-Driven Collagen Self-Assembly Affects Early Osteogenic Stem Cell Signaling. Adv. Healthc. Mater. 5 (2016) 1481-1492.

Figure legends Figure1. Characterization of PDMS substrates. (A) Representative surface topographic AFM images. Stiff substrate: 1:5 PDMS and Soft substrate: 1:45 PDMS. Each topographic image of PDMS film was collected in six independent experiments (n = 6). (B) Surface roughness by Ra parameter. The analysis was based on the six independent experiments (n = 6), *Significant difference with respect to petri dish control (p < 0.05). (C) Young’s moduli of PDMS substrates. The data were the mean of six independent experiments (n = 6), *p < 0.05.

Figure2. Cell morphologies of adipose-derived stromal cell (ASCs) cultured on PDMS substrates with different stiffnesses. (A) Representative SEM scanning images of ASCs cultured on PDMS substrates with different stiffnesses for 3 days. The cell images were collected in six independent experiments (n = 6). (B) Histogram showing the changes of ASC spreading area on stiff/soft PDMS substrates. The statistics was based on three independent experiments (n = 3), *, p < 0.05. The numbers shown in parenthesis indicated cell numbers for statistics of cell spreading area examined in each case.

Figure3. Microfilament changes of ASCs on PDMS substrates with different stiffnesses. (A) Representative microfilament changes by F-actin fluorescent stain. The ASCs were cultured on PDMS for 3 days. The microfilament fluorescent images were collected in six independent experiments (n = 6). (B) Representative

distribution of F-actin in ASCs cultured on stiff substrate (1) and soft substrate (2). Fluorescence optical density (OD) by Image Pro Plus 6.0 was used to measure the F-actin distribution. Histograms in (1) and (2) showing the cell population ratio of representative ASCs on the stiff and soft PDMS substrates. The statistics was based on three independent experiments (n = 3), *, p < 0.05. (C) The total F-actin fluorescent intensities per cell area were shown on the stiff and soft PDMS substrates. The statistics was based on three independent experiments (n = 3), *, p < 0.05. The numbers shown in parenthesis indicated cell numbers for statistics of cell spreading area examined in each case.

Figure4. Microtubule changes of ASCs on PDMS substrates with different stiffnesses. (A) Representative microtubule changes by fluorescence co-staining of α- and β-tubulin. The ASCs were cultured on PDMS for 3 days. The microtubule fluorescent images were collected in six independent experiments (n = 6). (B) Representative distribution of α- and β-tubulin in ASCs cultured on stiff substrate (1) and soft substrate (2). Fluorescence optical density (OD) by Image Pro Plus 6.0 was used to measure the tubulin distribution. Histograms in (1) and (2) showing the cell population ratio of representative ASC on the stiff and soft PDMS substrates. The statistics was based on three independent experiments (n = 3), *, p < 0.05. (C) The total tubulin fluorescent intensities per cell area were shown on the stiff and soft PDMS substrate. The statistics was based on three independent experiments (n = 3), *, p < 0.05. The numbers shown in parenthesis indicated cell numbers for statistics of cell spreading area examined in each case.

Figure5. The changes of vinculin in ASCs cultured on PDMS substrates with different stiffnesses. (A) Representative IF staining by CLSM showing the different distributions of vinculin in ASCs in response to different PDMS substrates. The fluorescent images were collected in six independent experiments (n = 6). (B) Fluorescence OD measurement showing the amount of vinculin in ASCs on different PDMS substrates. The analytic statistics was based on the eight independent experiments (n = 8). *Significant difference presents between the two groups (p < 0.05). (C) Western blot showing the protein changes of vinculin in ASCs on the substrates with different stiffnesses. The blot images were collected in three independent experiments (n = 3). (D) Quantification was performed to confirm the protein changes of (C). The statistics was performed at three independent experiments (n = 3). *, p < 0.05.

Figure6. The nuclear translocation of β-catenin on the PDMS substrates with different stiffnesses. (A) Representative IF staining by CLSM showing the protein changes of β-catenin in response to PDMS substrates with different stiffnesses. The IF staining images were based on the three independent experiments (n = 3). (B) Representative distribution of β-catenin in ASCs cultured on stiff (1) and soft (2) substrates. Fluorescence OD by Image Pro Plus 6.0 was used to measure the β-catenin distribution. Histograms in (1) and (2) showing the cell

population ratio of representative ASCs on the stiff and soft PDMS substrates. The statistics was based on three independent experiments (n = 3), *, p < 0.05. (C) Histogram showing the ratio of cell population with nuclear translocation of β-catenin in response to different stiffnesses. *, p < 0.05. The numbers shown in parenthesis indicate statistical data examined in each case. (D) Western blot showing the protein changes of β-catenin in response to different PDMS substrates. The blot images were collected in three independent experiments (n = 3). (E) Quantification was performed to confirm the protein changes of (D). The statistics was performed based on three independent experiments (n = 3). *, p < 0.05.

Figure7. Differentiation of ASCs towards osteogenesis cultured on PDMS substrates with different stiffnesses. (A) Representative ALP staining showing differential osteogenesis capacity of ASCs seeded on PDMS substrates with different stiffnesses for 7 days’ osteogenic induction medium. (B) q-PCR further confirmed gene changes of osteogenic transcriptional factors, Runx2 and Osx, in response to different stiffnesses. q-PCR results were based on the three independent experiments (n = 3). *Significant difference presents between the two groups (p < 0.05). (C) Representative ALP staining showing the influence of activator (LiCl) and inhibitor (DKK1) of wnt/β-catenin on the differential osteogenesis capacity of ASCs seeded on PDMS substrates with different stiffnesses for 7 days’ osteogenic induction medium. (D) Western blot showing the activator (LiCl) and inhibitor (DKK1) of wnt/β-catenin induced different protein expressions of Runx2 and Osx (n = 3). (E) Bioinformatics showing the binding sites of the downstream protein, Lef-1, of β-catenin at the promoter regions of Runx2 and Osx. (F-G) ChIP qPCR showing specific binding sites of Lef-1 at the promoter regions of Runx2 and Osx. The statistics was performed at three independent experiments (n = 3). *, p < 0.05. (G-I) Agarose gel image using ChIP qPCR products in (F-G).

Figure8. Differentiation of ASCs towards adipogenesis cultured on PDMS substrates with different stiffnesses. (A) Representative Oil red staining showing differential adipogenesis capacity of ASCs seeded on PDMS substrates with different stiffnesses for 7 days’ adipogenic induction medium. (B) q-PCR further confirmed gene changes of osteogenic transcriptional factors, PPARγ and C/EBPα, in response to different stiffnesses. q-PCR results were based on the three independent experiments (n = 3). *Significant difference presents between the two groups (p < 0.05). (C) Representative Oil red staining showing the influence of activator (LiCl) and inhibitor (DKK1) of wnt/β-catenin on the differential adipogenesis capacity of ASCs seeded on PDMS substrates with different stiffnesses for 7 days’ osteogenic induction medium. (D) Western blot showing the activator (LiCl) and inhibitor (DKK1) of wnt/β-catenin induced different protein expressions of PPARγ and C/EBPα (n = 3). (E) Bioinformatics showing the binding sites of the downstream protein, Lef-1, of β-catenin at the promoter regions of PPARγ and C/EBPα. (F-G) ChIP qPCR showing no specific binding sites of Lef-1 at the promoter regions of PPARγ and C/EBPα. The experiments were performed at three independent experiments (n = 3). (G-I) Agarose gel image using

ChIP qPCR products in (F-G).

Figure9. The schematic diagram elucidating matrix stiffness modulates cytoskeleton rearrangements and cell differentiation of ASCs. The grey parts are involved in the cytoskeleton rearrangements in published papers but not in this study; and the green and purple contexts indicate the stiffness-modulated differentiation towards osteogenesis and adipogenesis through β-catenin in this study.