Refined assessment of the impact of cell shape on human mesenchymal stem cell differentiation in 3D contexts

Acta Biomaterialia 87 (2019) 166–176

Contents lists available at ScienceDirect

Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat

Refined assessment of the impact of cell shape on human mesenchymal stem cell differentiation in 3D contexts Andrea C. Jimenez-Vergara a, Rodrigo Zurita a, Abigail Jones a, Patricia Diaz-Rodriguez b, Xin Qu b, Kenneth L. Kusima a, Mariah S. Hahn b,⇑, Dany J. Munoz-Pinto a,c,⇑ a b c

Department of Engineering Science, Trinity University, San Antonio, TX 78212, United States Biomedical Engineering Department, Rensselaer Polytechnic Institute, Troy, NY 12180, United States Neuroscience Program, Trinity University, San Antonio, TX 78212, United States

a r t i c l e

i n f o

Article history: Received 7 September 2018 Received in revised form 4 January 2019 Accepted 24 January 2019 Available online 25 January 2019 Keywords: Interpenetrating networks Cell shape Human mesenchymal stem cell Differentiation

a b s t r a c t Numerous studies have demonstrated that the differentiation potential of human mesenchymal stem cells (hMSCs) can be modulated by chemical and physical cues. In 2D contexts, inducing different cell morphologies, by varying the shape, area and/or curvature of adhesive islands on patterned surfaces, has significant effects on hMSC multipotency and the onset of differentiation. In contrast, in vitro studies in 3D contexts have suggested that hMSC differentiation does not directly correlate with cell shape. However, in 3D, the effects of cell morphology on hMSC differentiation have not yet been clearly established due to the chemical and physical properties being intertwined in 3D matrices. In this work, we studied the effects of round or elongated cell morphologies on hMSC differentiation independently of scaffold composition, modulus, crosslink density and cell-mediated matrix remodeling. The effects of cell shape on hMSC lineage progression were studied using three different cell culture media compositions and two values of scaffold rigidity. Differences in cell shape were achieved using interpenetrating polymer networks (IPNs). The mechanical and diffusional properties of the scaffolds and cell-matrix interactions were characterized. In addition, cell responses were evaluated in terms of cell spreading via gene and protein expression of differentiation markers. Cumulative results support, and extend upon previous work indicating that cell shape alone in 3D contexts does not significantly modulate hMSC differentiation, at least for the scaffold chemistry, range of modulus and culture conditions explored in this study. Statement of Significance In 2D contexts, inducing different cell shapes, by varying the curvature, area size and shape of a patterned surface, has significant effects on hMSC multipotency and the onset of cell differentiation. In contrast, in vitro studies in 3D contexts have suggested that hMSC differentiation does not directly correlate with cell shape. However, in 3D, the effects of cell morphology on hMSC differentiation have not yet been clearly established due to the chemical and physical properties being intertwined in 3D matrices. In this work, we studied the effects of round or elongated cell morphologies on the differentiation of hMSCs independently of scaffold composition, modulus, crosslink density and cell mediated matrix remodeling. Cumulative results support, and extend upon previous work indicating that cell shape alone in 3D contexts does not significantly modulate hMSCs differentiation commitment. Ó 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction ⇑ Corresponding authors at: Dept of Engineering Science, Neuroscience Program, Center for the Sciences and Innovation, CSI 470C, Trinity University, One Trinity Place, San Antonio, TX 78212, United States (D. J. Munoz-Pinto). Dept of Biomedical Engineering, Biotech 2121, Rensselaer Polytechnic Institute, 110 8th Street, Troy, NY 12019, United States (M.S. Hahn). E-mail addresses: [email protected] (M.S. Hahn), [email protected] (D.J. Munoz-Pinto). https://doi.org/10.1016/j.actbio.2019.01.052 1742-7061/Ó 2019 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Human mesenchymal stem cells (hMSCs) are recognized as a viable cell source for treating a number of human diseases, including Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis and multiple sclerosis, among others [1]. In addition, their high proliferative capacity and their potential to differentiate into multiple cell lineages makes them a promising cell source for

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

tissue engineering applications in connective tissue [2–5]. The multipotency and proliferative capacity of this cell source have been modulated using chemical and physical cues. For instance, hMSC differentiation has been successfully directed using small chemical functional groups, growth factors, extracellular matrix (ECM) ligands and cell–cell interactions [6,7]. In addition to chemical cues, physical properties such as substrate topography, modulus, matrix mesh structure, traction forces, cytoskeletal tension and induced cell shape have also been shown to be powerful factors that regulate hMSC behavior [8–11]. The success of tissue engineering strategies using hMSCs depends highly on the accurate understanding of the relative effects of substrate properties and soluble factors on hMSC fate. An ideal bio-inspired material should closely mimic the chemical composition of native tissue while providing close matching mechanical and diffusional performance to those exhibited in vivo. In addition, it is also highly desirable that a functional artificial tissue promotes the development of physiologically relevant cell morphologies. While cells in cartilage or adipose tissue exhibit round shapes, cells in muscle or nerve tissue natively display elongated shapes. These differences in cell shape are correlated with tissue function and appear to be independent of tissue stiffness. Thus, both round and elongated cells can be observed over a broad range of mechanical properties. Specifically, round cells are present in lower elastic modulus (adipose) or high stiffness (cartilage) tissue. Similarly, elongated cells are found in tissues with higher elastic modulus (muscle tissue) or low stiffness (nerve tissue). In two-dimensional (2D) contexts, inducing different cell shapes by varying the shape, area and/or curvature of adhesive islands on patterned surfaces has significant effects on hMSC multipotency and the onset of differentiation [9,12–16]. The modulation of cell phenotype has been linked to the regulation of the contractile machinery of the cytoskeleton and the primary regulation of the RhoA pathway [12]. Although these findings are very significant for understanding the hMSC differentiation process, most cells in vivo experience a three-dimensional (3D) milieu rather than a 2D surface and, therefore, the direct extrapolation of these observations has to be exerted with caution [17]. In contrast to the 2D findings, in vitro studies in 3D contexts have suggested that hMSC differentiation does not directly correlate with cell shape. Instead, cell fate appears to be modulated by scaffold viscoelastic property dynamics, integrin binding, and the exertion of traction forces [18–20]. In addition, it has proven challenging to study the effects of cell morphology on hMSC fate in 3D contexts in isolation. In 3D, there is a strong entanglement among chemical composition, matrix rigidity, matrix mesh structure, diffusional constraints and the capacity of cells to interact with and/or to reorganize the matrix. Using alginate-based scaffolds, Huebsch et al. observed that hMSC fate was not correlated with cell morphology in 3D. Instead, cell response was induced by changes in traction dependent forces and scaffold rigidity [18]. Similarly, Khetan et al. demonstrated that hMSC differentiation was linked with hydrogel permissivity, or capacity of cells to remodel the surrounding microenvironment [19]. Moreover, Chaudhuri et al. also demonstrated that the rate of scaffold stress relaxation modulates cell spreading, which variably activates cell contractile machinery leading to the control of osteogenic differentiation [20]. In each of these studies, results suggested that in 3D, cell shape alone was not a determining factor in triggering adipogenic or osteogenic differentiation. However, these findings were drawn using scaffolds where the rigidity and cross-link density (mesh structure) were not consistent across different groups [18,19]. In summary, the effects of cell morphology were not studied using scaffolds exhibiting similar modulus, average mesh size and/or levels of cell-matrix remodeling. Although in each study, controls were incorporated to try to exclude the impact of variable cross-

167

link density on cell outcome, the question regarding the impact of morphology on hMSC differentiation in 3D contexts remains. In this work, we evaluated the effect of cell morphology on hMSC differentiation in the absence of significant differences in matrix rigidity (storage modulus, E’), diffusional constraints (average mesh size, f) or cell-mediated matrix remodeling. To achieve these conditions, we used collagen-poly(ethylene glycol) diacrylate interpenetrating networks (collagen-PEGDA IPNs) fabricated by sequential infiltration [8]. We first utilized different culture media formulations to evaluate the effects of cell shape on hMSC phenotypic progression in the presence of soluble differentiation factors at a modulus of
37 kPa. The mechanical and diffusional properties of the scaffolds were characterized using dynamic mechanical analysis and swelling behaviors, respectively. hMSC lineage progression was characterized at the gene and protein levels. We subsequently conducted a more in depth study of the effects of cell shape on differentiation using a single medium formulation but a scaffold modulus of
15 kPa. The cumulative set of results demonstrated and corroborated that in 3D, differences in cell morphology (round versus elongated) are not alone sufficient to direct the differentiation of hMSCs, at least for the scaffold chemistry, range of modulus and culture conditions examined herein.

2. Materials and methods 2.1. PEGDA synthesis and characterization PEGDA (6.0 kDa and 10.0 kDa) was prepared by mixing dry PEG (Fluka) with acryloyl chloride (Sigma) in anhydrous dichloromethane using a molar ratio 1:4 or 1:8, respectively. Substitution of the terminal OH groups in longer PEG chains is more difficult than for shorter PEG molecules. Due to the reduction in the relative OH group abundance coupled with the increased viscosity of the reaction mixture with increasing PEG chain length, the PEG to acryloyl chloride ratio was adjusted to ensure >90% acrylation for both PEG chain lengths utilized. The mixtures were stirred under argon overnight in the presence of trimethylamine (1:2 M ratio) as the catalyst [21]. The product of the reaction was purified by washing it with a 2 M K2CO3 aqueous solution and drying the organic phase with MgSO4. The final product was precipitated in diethyl ether, filtered and dried. The level of end group substitution was evaluated using 1H NMR. The average achieved acrylation level was above 92%.

2.2. IPN fabrication Collagen-PEGDA IPNs were fabricated following the protocol and directions previously developed by Munoz-Pinto et al. [8]. To fabricate the double network hydrogels, collagen type I (collagen, BD Biosciences) was neutralized and diluted to achieve a final concentration of 3 mg/mL. Approximately 7.5  105 hMSCs/mL were suspended in the neutralized collagen solution. Using 12 well plate cell culture inserts (Corning), 300 lL aliquots of the collagen solution were then cured for 30 min at 37 °C and allowed to swell in DMEM without phenol red or additional supplements for an additional 30 min. The degree of cell spreading was controlled by the formation of the PEGDA network. To achieve round or elongated cell morphologies, the initial collagen hydrogel was placed in contact with 1.7 mL of 11.7% w/w PEGDA (6.0 kDa or 10.0 kDa) solution in DMEM immediately or 6 h after the complete curing of the collagen hydrogel, respectively. Cells that were allowed to spread in the collagen network for 6 h before being placed in contact with the PEGDA solution exhibited an elongated shape, while

168

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

cells placed in contact with PEGDA immediately following encapsulation exhibited a round morphology. In each case, the PEGDA solution was allowed to infiltrate into the collagen network for 1 h, after which excess of the PEGDA solution was removed. The infiltrated PEGDA was then polymerized by 6 min exposure to longwave UV light (Spectroline,
6 mW/cm2, 365 nm), a process that has previously been demonstrated to be cytocompatible [22–24] and to not alter gene expression [22]. The molecular weight of the infiltrating PEGDA solution dictated the modulus of the resulting IPNs, with the 6.0 kDa PEGDA IPNs exhibiting higher modulus than the 10.0 kDa PEGDA IPNs. The PEGDA network of the IPN renders the IPN a non-permissive structure in that cells are unable to proliferate within the IPN or alter the IPN structure due to the covalent crosslinking, nanoscale mesh structure and slow degradation rate of PEGDA networks [25]. That said, hMSC viability within collagen-PEGDA IPNs formed by the sequential infiltration procedure has previously been shown to be >90% [8]. 2.3. Mechanical characterization To mechanically characterize the IPNs, the viscoelastic properties of each IPN formulations were evaluated using dynamic mechanical analysis (DMA). Briefly, four 8 mm discs of approximately 1.1 mm in thickness were tested for each treatment group using a 3300 TA Instruments mechanical tester equipped with a 250 g (
2.45 N) load cell. All samples were tested using 1 Hz frequency, a 0.1 mm amplitude displacement and an initial indentation of 0.05 mm. The storage modulus (E0 ) was reported as the average plus or minus the standard deviation of the measurement. 2.4. Hydrogel mesh size assessments and PEGDA concentration The average mesh size of each IPN was evaluated using swelling ratio measurements of four independent specimens per treatment group. To estimate the average mesh size, we first calculated the molecular weight between crosslinks, Mc, using a semi-empirical correlation developed by Jimenez-Vergara et al. [26].

1 ¼ 0:109 Mc




m2;r m2;s



  h m

V1

Inð1  m2;s Þ þ m2;s þ vm22;s  1   3 1 m2;s m2;r mm2;s  2 m2;s 2;r

i ð1Þ

2.5. Cell studies Human mesenchymal stem cells (hMSCs, Lonza) from a 22-year-old African American female donor were purchased at passage 2 and expanded on cell culture treated polystyrene at 37 °C and 5% CO2 using MesenPro RS medium (Gibco, Life Technologies). The stemness of these cells was assessed by confirming expression of the markers CD105, CD166, CD29, and CD44 by over 90% of the cell population and expression of CD14, CD34, and CD45 by <10% of the cell population. In addition, the multipotency of these cells was confirmed by the expression of bone, cartilage or adipose tissue markers following 14 or 21 days of differentiation stimulation. Following the initial cell expansion, hMSCs at passage 4 were encapsulated in the proposed IPNs. The response of round or spread hMSCs to different cell culture media conditions (Table 1) was studied for the high modulus (6.0 kDa PEGDA) IPN and for a single medium condition for the low modulus (10.0 kDa PEGDA) IPN. The cell culture media was changed every other day. 2.6. Cell shape assessments The morphology of encapsulated hMSCs was quantitatively evaluated using laser confocal microscopy. Twenty-four hours post IPN fabrication, three independent specimens per treatment group were fixed with formalin and stained using rhodamine phalloidin (Life Technologies, 1:100 in 1 DPBS) following the protocol from the manufacturer. Ten independent regions per specimen were then imaged using a Nikon A1 laser confocal microscope equipped with a 20 magnification objective. The degree of spreading was characterized in terms of circularity and roundness using Image J software, where circulatory values approaching a numerical value of one correspond to round morphologies, and values approaching zero correspond to cells exhibiting a high degree of elongation [8]. hMSCs have previously been confirmed to maintain these ‘‘time zero” round or elongated cell shapes through at least 14 days of culture in the collagen-PEGDA IPNs [8]. 2.7. Cell matrix interactions



where m is the specific volume of the polymer (0.893 cm3/g for PEG), V1 is the molar volume of the solvent (18 cm3/mole), t2,s and t2,r are the polymer volume fractions after equilibrium and at the relaxed state, respectively, and v is the polymer solvent interaction parameter (0.426 for PEG-water systems). The mesh size was then calculated using Eq. (2):

 2 1=2 n ¼ m2;s 1=3 r 0


1  2 12 2Mc 2 12 r0 ¼l Cn Mr

ð3Þ

with l = 1.50 Å, Mr = 44 g/mol and Cn = 4. The swelling ratio measurements were also used to estimate the PEGDA concentration in the IPNs. The PEGDA concentration was calculated using Eq. (4):

md x100 mi

In the collagen-PEGDA IPNs, cell-matrix interactions are dominated by the interactions of cell integrin complexes and the collagen network. Adult stem cells are characterized by the expression of the integrin a subunits 1, 2, 3, 4, 5, and v, and b subunits 1, 3 and 5. The integrin subunit b1 interacts to form integrin complexes with a range of the a subunits [27]. Therefore, although collagen

ð2Þ

The end-to-end distance of the unperturbed (solvent-free) state of  2 1=2 was computed using Eq. (3): the polymer r 0

%w=w ¼

where md and mi are the dry and initial masses of the IPN, respectively. The impact of the collagen mass on the calculation of the percent PEGDA was considered negligible since it only accounts for 0.3% of the initial IPN mass, or 3% of its dry mass.

ð4Þ

Table 1 Cell culture media composition. Medium*

Supplements/Composition

Growth medium (GM) Adipogenic maintenance medium (AMM) Adipogenic induction medium (AIM)

1% antibiotic/antimycotic Recombinant h-insulin, 1% antibiotic/ antimycotic Recombinant h-insulin, dexamethasone, indomethacin, 3-isobuty-l-methyl-xanthine, 1% antibiotic/antimycotic Dexamethasone, ascorbate, b-glycerophosphate, 1% antibiotic/antimycotic 1:1 vol of AIM to OIM 1:1 vol of AMM to OIM

Osteogenic induction medium (OIM) AIM:OIM AMM:OIM

* All cell culture media was prepared using DMEM and 10% v/v MSC-qualified FBS.

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

I supports primarily a1b1 and a2b1 integrin interactions, results for the integrin b1 subunit alone would have been challenging to interpret and attribute to particular a subunits [28]. This observation is significant given that integrin a1b1 and integrin a2b1 have differential effects on internal cell signaling. We therefore chose to characterize the most abundant hMSC a subunit as a direct measurement of the interaction with the collagen network. Toward this end, hMSCs were encapsulated at 2.5  106 cells/ mL in a 3 mg/mL collagen hydrogel. Cells exhibiting round or elongated morphologies (0 h or 6 h spreading time) were lysed using lysis buffers of different strength and following a modified version of the protocol described by Jones et al. [29]. Briefly, the 150 mL collagen hydrogels were homogenized in 400 mL of RIPA buffer (Sigma) using a pestle and 1.5 mL Eppendorf tubes. The samples were then incubated at room temperature for 20 min. Following the incubation period, the samples were centrifuged at 10,000 rpm and 4 °C for 10 min. The supernatant (fraction 1) was then carefully collected and stored at 80 °C. A second protein extraction was performed on the remaining pellet using RIPAEDTA buffer (25 mM EDTA in RIPA buffer). The pellet was homogenized, incubated at room temperature, and centrifuged at 10,000 rpm and 4 °C. The second supernatant (fraction 2) was stored at 80 °C until use. Western blot analysis was performed on the two protein fractions following the protocol described in Section 2.8.2 and the antibodies listed in Supplementary Table 2. The relative expression levels of integrin a2 were reported as the ratio between fraction 2 (strong) and fraction 1 (weak). This nomenclature is a result of the differences between the two fractions. RIPA buffer solubilizes only integrin complexes not bound or only weakly bound to a ligand, leaving strongly bound integrin-ligand complexes intact. Thus, fraction 1 represents integrin complexes that were in the unbound/weakly bound state at the time of extraction [18]. The addition of EDTA to the RIPA buffer increases its strength and since the removal of divalent cations by EDTA completely inhibits integrin-ligand binding [30], fraction 2 will represent integrin complexes that were in the bound state at the time of extraction. To further assess the initial interaction of round or elongated hMSCs with the collagen network, constructs were stained for collagen 24 h post-fabrication as reported by Munoz-Pinto et al. [8]. The cytoskeleton of hMSCS was stained using rhodamine phalloidin (Invitrogen, 1:100 dilution in DPBS). Confocal microscopy was performed using a Zeiss LSM 510 META confocal microscope equipped with a 40 water immersion objective. Four randomly selected regions in three independent samples per treatment group were evaluated. Qualitative assessment of collagen fiber arrangement around round or elongated cells was performed and indicated no difference between treatment groups. 2.8. IPN endpoint analyses To assess hMSC lineage progression under the different cell culture conditions, the cell response was evaluated at the gene and protein levels. After 14 days of culture, samples from each IPN (n = 6 per formulation) were harvested for gene expression, DNA and western blot analyses (for both 6.0 kDa and 10.0 kDa PEGDA IPNs) as well as for multiplex immunoassay assessments (in the case of the 10.0 kDa PEGDA IPNs). A 6 mm disc was cored from each of the IPN constructs and transferred to an RNase-free conical tube, flash frozen in liquid nitrogen and stored at 80 °C until use. In addition, a portion of each 6 mm disc was collected for histological analysis for the 10.0 kDa PEGDA IPN formulations and the 6.0 kDa PEGDA IPN GM controls. PolyA-mRNA, DNA and proteins were extracted using the Dynabeads mRNA direct kit (Ambion, Life Technologies) as previously described by Munoz et al. [2]. In brief, each IPN sample specimen

169

was immersed in 330 mL of the provided lysis binding buffer. The samples were then homogenized using a plastic RNase-free pestle (Kimble Chase) and incubated at room temperature for 10 min. Following incubation, the samples were centrifuged for 5 min at 10,000 rpm, and the polyA-mRNA in the supernatant was harvested using 20 mL of Dynabeads oligo (dT)25 magnetic beads. The remaining supernatant, containing the DNA and proteins, was collected. PolyA-mRNA was eluted from the magnetic beads into TrisHCl buffer by heating at 80 °C for 2 min. The supernatant was collected and stored at 80 °C until qRT-PCR analysis. The hydrogel pellet was placed in contact again with 330 mL of lysis buffer, and a second protein extraction was performed. After incubating the pellet for 10 min at room temperature, the specimens were exposed to three freeze–thaw cycles. The samples were then centrifuged for 5 min at 10,000 rpm. The supernatant was then removed and combined with the first extraction and stored at 20 °C for DNA measurements, western blotting and multiplex protein analysis. 2.8.1. Gene expression analyses To evaluate the differentiation of hMSCs toward osteogenic, adipogenic, chondrogenic or smooth muscle lineages, verified qRT-PCR primers for human runt-related transcription factor 2 (RUNX2), tissue nonspecific alkaline phosphatase (TNAP), osteopontin (SPP1), peroxisome proliferator-activated receptor c (PPARc), complement factor D (CFD), adipocyte fatty acid-binding protein (AFABP), SRYbox 9 (SOX9), aggrecan (ACAN), collagen type II (Coll 2), transgelin (SM22a), smooth muscle alpha-actin 2 (ACTA2), and calponin 1 (CNN1) were used to evaluate the relative expression of the selected differentiation markers (Supplementary Table 1). qRTPCR was performed on each sample using a 7500 real-time PCR system (Life Technologies) and the SuperScript III Platinum One-Step qRT-PCR kit (Invitrogen, Life Technologies). Approximately, 3 ng of polyA-mRNA and 5 lL of 0.001 mM primer were added per 25 lL of reaction mixture. Levels of mRNA for each gene of interest were tested in duplicate from 4 to 6 independent constructs per experimental group. The amplification process of the PCR products was recorded by measuring the increase in fluorescence of SYBR green dye. ROX dye was used as a passive reference. The amplification cycle at which the fluorescence signal reached a selected threshold value was evaluated and recorded using 7500 software v2.3. The relative expression of each gene was calculated with respect to reference genes b-actin and GAPDH using the DDCt method. Melting temperature analysis was used to verify the identity and selectivity of the PCR amplicon product. 2.8.2. Western blot analysis for protein expression The initial results from the qRT-PCR screening were further investigated at the protein level using western blotting. Semiquantitative analysis was employed to compare levels of protein associated with the adipogenic markers CFD and AFABP, and the osteogenic markers RUNX2, Osterix (OSX) and TNAP. The levels of these proteins were detected using the antibodies listed in Supplementary Table 2. DNA content from samples was determined using the Quant-iTTM PicoGreen dsDNA Assay (Invitrogen) following the protocol from the manufacturer. For each antibody examined, protein solution volumes representing equal amounts of DNA (850 ng per IPN) were concentrated using 3000 MWCO Amicon filter units (Millipore) to a final volume of 20 lL. Protein samples were denatured by the addition of b-mercaptoethanol and by heating at 95 °C for 10 min. Proteins were subsequently loaded into separate wells of an 8 or 12% polyacrylamide gel, separated by electrophoresis and transferred to a nitrocellulose membrane (Thermo Scientific). The membrane was blocked with a 5% BSA solution (Fraction V, Fisher Scientific) in Tris-buffered saline (TBST; 25 mM Tris-HCl, pH. 7.5,

170

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

137 mM NaCl, 0.1% Tween 20) for 1 h at room temperature. Primary antibody diluted in TBST containing 5% BSA was then applied overnight at 4 °C. The primary antibodies bound to each blot were detected using an appropriate horse radish peroxidase (HRP)- or alkaline-phosphatase (AP)- conjugated secondary antibody (Jackson Immunoresearch) diluted in TBST containing 5% BSA for 1 h at room temperature. The luminescent signal was detected using a luminol reagent (Santa Cruz Biotechnology, SCBT) or AP chemiluminescent solution (Novex). Sample imaging was performed using the molecular imager Chemidoc XRS system (Biorad) with exposure times set manually to prevent signal saturation. The integrated optical density of each band was quantified using Adobe Photoshop. The levels of each target protein were normalized to b-actin. 2.8.3. MAGPIX immunoassay analyses To confirm and expand upon the western blot analyses, relative protein levels of AFABP, Leptin, Osteonectin (ON) and SPP1 were quantified for the 10.0 kDa PEGDA IPNs using MAGPIX immunoassay multiplexing (Luminex). Briefly, 25 mL of sample lysates were pipetted into appropriate wells of 96-well plates, after which they were reacted with the components of a premixed magnetic bead analyte kit (R&D Systems) per the manufacturer’s instructions. Protein concentrations were determined from the median fluorescence intensity (MFI) values in comparison to an analyte-specific standard curve. The obtained values were normalized to the DNA content of each construct (Invitrogen). 2.8.4. Histological analyses To complement the qRT-PCR and protein assessments, the expression of late osteogenic and adipogenic markers was evaluated at the histological level for the 10.0 kDa PEGDA IPNs and the 6.0 kDa PEGDA IPN GM controls. Toward this end, samples were fixed with 10% formalin for 30 min, embedded and frozen in OCT medium (Tissue-Tek), and cut into 30 lm sections using a cryomicrotome. Mineralization and lipid droplets were evaluated. To detect lipid deposits, IPN sections were stained using a 3 mg/ ml Oil red O (Fisher) solution. Rehydrated sections were exposed to 10% formalin for 10 min and then rinsed with double distilled H2O (ddH2O). After 5 min of exposure to Oil Red O, stained sections were rinsed with ddH2O and mounted. To detect mineralization, formalin fixed samples of each group were stained using a 20 mg/ml Alizarin Red (Acros Organics) solution. In brief, sections were rehydrated with distilled H2O (dH2O), followed by a rinse with 50% ethanol 4 times. Sections were then exposed to Alizarin Red solution for 5 min and briefly rinsed with dH2O, acetone, and xylene. Sections were mounted with polymount (Polysience, Inc). Additionally, von Kossa (American MasterTech Scientific) staining was performed to detect and visualize calcium deposits. In brief, rehydrated sections were rinsed with ddH2O, after which a 5% silver nitrate solution was applied. Sections were then exposed to full-spectrum light for 1 h. After rinsing with dH2O, sections were exposed to 5% sodium thiosulfate for 2.5 min, briefly rinsed and mounted. All staining procedures were performed on 4 independent samples per treatment group. Cell counts for positively stained cells relative to total cells in the stained sections were performed by an observer blinded to outcome. Stained sections were imaged using an Eclipse TS100 Nikon microscope (Nikon) and representative images were selected to illustrate the overall expression of the late osteogenic and adipogenic markers in the IPNs cultured in AIM:OIM and GM.

3. Results 3.1. Mechanical properties, average mesh size, and PEGDA concentration The mechanical and diffusional properties of proposed collagen-PEGDA IPNs can be modulated by changing the collagen concentration, the PEGDA concentration, or the PEGDA molecular weight [26]. As anticipated, dynamic mechanical analysis (DMA) showed that the IPN formulations containing the 6.0 kDa PEGDA exhibited a storage modulus, E0
37 kPa, while the formulations fabricated with 10.0 kDa PEGDA displayed a lower average storage modulus, E0
15 kPa (Fig. 1A) [26,31]. Furthermore, the 6.0 kDa and 10.0 kDa IPNs exhibited values of storage modulus
25.2 and
10.3 fold greater than the pure collagen control hydrogel, respectively (p  0.001). In terms of differences in mechanical performance between the two IPN groups, the 6.0 kDa and 10.0 kDa IPNs showed significantly different storage moduli (p < 0.001). In addition, significant differences between the bulk storage modulus of the IPNs of a given PEGDA molecular weight containing elongated or rounded cells were not observed (p  0.303). Estimation of the average mesh size correlated well with the modulus data. In brief, while modulus was reduced as PEGDA chain length increased, the average mesh size was significantly increased from approximately 24 nm in the 6.0 kDa PEGDA IPN to 36 nm in the 10.0 kDa PEGDA IPN (Fig. 1B, p < 0.001). Importantly, significant differences in the average mesh size of IPN formulations of a given PEGDA molecular weight containing round and spread cells were not observed (p > 0.366). To further confirm that the round and elongated cells within IPNs of a given PEGDA molecular weight were experiencing similar initial environments, the concentration of PEGDA within the IPNs was calculated using IPN swelling data (Table 2). The PEGDA concentration in the 6.0 kDa PEGDA IPN was
10.3% and 10.4% in the constructs supporting round or elongated morphologies, respectively. In the 10.0 kDa PEGDA series, the concentration of PEGDA in the formulations supporting round or spread cells was
8.4%. In short, significant differences between IPNs of a given PEGDA molecular weight supporting round or spread cells were not observed (p  0.6), although the PEGDA concentration in IPNs containing 10.0 kDa PEGDA was significantly lower than that in the 6.0 kDa PEGDA series (p < 0.001). To study the isolated effects of cell shape on hMSC differentiation, the sequential fabrication of the IPNs has an advantage over other proposed platforms [18,19] since it allows for the control of cell shape without changes in scaffold chemistry, modulus or crosslinking density. For instance, Khetan et al. developed a 3D platform that regulated cell shape by allowing encapsulating hMSCs within a hydrogel network [19]. In a subset of gels, rounded morphologies were preserved by inducing secondary crosslinks at day 0 of culture. In contrast, hMSCs within the elongated cell treatment group were allowed to spread over a period of 7 days prior to induction of the secondary crosslinks [19]. This design implies that the rounded versus elongated cells experienced different diffusional constraints, scaffold rigidities, and degrees of cell-matrix remodeling for the 7 days permitted for cell elongation. In our design, we have reduced this delay in induction of secondary crosslinks to 6 h.

3.2. Cell morphology assessment 2.9. Statistical analyses Data are reported as mean ± standard deviation. Comparison of sample means was performed using ANOVA followed by Tukey’s post-hoc test (SPSS software), p < 0.05.

Representative images of spread and round cells 24 h following encapsulation are shown in Fig. 2A. As a quantitative measurement of the degree of spreading, cell circularity and roundness were evaluated (Fig. 2B and 2C). The circularity and roundness were

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

171

Fig. 1. Storage modulus and average mesh size in IPNs as a function of PEGDA molecular weight and round versus elongated cell morphology. (A) Storage modulus, (B) average mesh size. Results are reported as mean ± standard deviation, n = 4. $Significantly different from 3 mg/mL collagen I (Col I) control, p < 0.05; *significantly different from 6.0 kDa PEGDA IPN supporting round cells, p < 0.05; #significantly different from 6.0 kDa PEGDA IPN supporting spread cells, p < 0.05.

Table 2 Concentration of PEGDA in Collagen type I-PEGDA IPNs. PEGDA MW

Cell Morphology

Concentration [% w/w]

6.0 kDa

Round Spread

10.28 ± 0.29 10.43 ± 0.30

10.0 kDa

Round Spread

8.42 ± 0.22a,b 8.42 ± 0.34a,b

The concentration of PEGDA in the different IPN formulations was calculated using d swelling data as follows: %w=w ¼ m mi x100, where md and mi are the dry mass and initial mass of the IPN respectively. a Significantly different from 6.0 kDa PEGDA round IPN, p < 0.001. b Significantly different from 6.0 kDa PEGDA spread IPN, p < 0.001, n = 4.

each reduced
1.4–1.7 fold in the elongated versus round IPN formulations for each PEGDA molecular weight (p < 0.001). In addition, for cells exhibiting round or elongated morphologies, the use of 6.0 kDa PEGDA and 10.0 kDa PEGDA in the IPNs appeared to have only partial influence on cell shape. Specifically, differences in circularity between the round and elongated treatment groups of the 6.0 kDa and 10.0 kDa PEGDA IPNs were not statistically significant (p = 0.370). However, the roundness measure appeared to capture small differences in cell shape between the two round morphology groups (p < 0.001) with the 10.0 kDa IPN being associated with reduced roundness. It is possible that the more open, less rigid network of the 10.0 kDa IPN resulted an increased capacity of the rounded cells to partially spread within the crosslinked network. Furthermore, once the IPNs have been placed in cell culture medium, the 10.0 kDa PEGDA network swells to a greater extent than the 6.0 kDa PEGDA network, which can induce additional stretching of the cells. The current results and numerical values

for circularity and roundness are in close agreement with previously reported data for this system [8], in which it was also demonstrated that round and elongated morphologies were maintained for at least 14 days [8]. 3.3. Cell-matrix interactions To achieve different degrees of cell elongation, the fabrication process of the proposed scaffolds relies on the ability of the cells to increasingly spread in the collagen hydrogel with time. Round cells were observed when PEGDA infiltration and curing was performed within 2 h of initial cell encapsulation within the collagen network. Spread cells, on the other hand, were obtained after 6 h of spreading time in the collagen scaffold before the PEGDA hydrogel was formed. This approach was similar to the strategy used by Khetan et al. [19], but instead of allowing the cells to spread for 7 days, we limited the initial cell-matrix contact difference to 6 h. The reduction in the spreading time from 7 days to 6 h prevents significant cell-mediated remodeling of the scaffold. However, since there was still a difference in the cell contact time with the permissive collagen network prior to formation of nonpermissive PEGDA network, we evaluated the potential of initial differences in cell-matrix interactions between the round and spread treatment groups. Initial insight into similarity in cellmatrix integrin interactions were obtained by measuring the levels of strongly bound versus weakly bound/unbound integrin subunit a2. The data demonstrated that the ratio of strongly bound to weakly bound/unbound integrin subunit a2 was not significantly different between round and spread cells (Supplementary Fig. 1; p = 0.684). Furthermore, qualitative analysis of collagen fibers

Fig. 2. Cell shape descriptors. (A) Representative images of round and spread hMSCs in the 6.0 kDa PEGDA and 10.0 kDa PEGDA IPNs. Scale bar = 100 lm. (B) Circularity. (C) Roundness. Results are reported as mean ± standard deviation, with the shape of at least 100 cells from four independent specimens per experimental group being evaluated. * Significantly different from 6.0 kDa PEGDA IPN supporting round cells, p < 0.001; #significantly different from 10.0 kDa PEGDA IPN supporting round cells, p < 0.001.

172

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

around individual round or elongated cells suggested that there were not significant differences in collagen fiber organization between these two groups. In particular, the images did not indicate an increase in the density of collagen fibers surrounding the cells in either of the two experimental groups. Representative images of individual cells in the IPN matrix are shown in Supplementary Fig. 2. 3.4. Gene expression analyses To evaluate the effects of cell shape on the differentiation of hMSCs exhibiting round and spread morphologies, an initial screening of several differentiation markers was performed using qRT-PCR. At the gene expression level, the response of hMSCs to the presence of soluble factors (1:1 adipogenic inductionosteogenic medium or 1:1 adipogenic maintenance-osteogenic medium) was evaluated in terms of the expression of early, intermediate and late differentiation markers for adipose (PPARc, CFD and AFABP) and bone (RUNX2, TNAP and SPP1) cell markers. In the 6.0 kDa PEGDA IPN, and in the presence of 1:1 adipogenic induction-osteogenic medium (AIM:OIM), there was a consistent and significant upregulation in the expression of the adipogenic markers PPARc, CFD and AFABP relative to the initial time point (day 1). The average increase in gene expression was 10, 100 and 650-fold with p  0.045, p  0.013 and p  0.001, respectively (Fig. 3A). There were not significant differences between the round and spread groups for the early and intermediate adipogenic markers PPARc and CFD. Relative to the day 1 group, AFABP increased over 778-fold in the round group and 519-fold in the spread group. Relative to round cells, AFABP expression in spread cells was significantly decreased by 33% (p = 0.034). In addition, it appeared that the osteogenic markers RUNX2, TNAP and SPP1 exhibited a slight increase in their expression relative to the day 1 group. However the upregulation of these markers was not statistically significant (p  0.243, p  0.070 and p  0.074 respectively). Furthermore, no significant differences in RUNX2, TNAP or SPP1 expression levels were observed between the round and spread treatment groups. Cumulatively, the qRT-PCR results suggested that encapsulated hMSCs in the 6.0 kDa PEGDA IPN (E0
37 kPa) were preferentially driven toward an adipogenic cell lineage when cultured in AIM:OIM, but the effects of cell shape appeared to be limited. To continue studying the effects of cell morphology on hMSC differentiation, a separate experiment was performed in which 6.0 kDa PEGDA IPNs were cultured in the presence of 1:1 adipogenic maintenance-osteogenic medium (AMM:OIM). We hypothesized that perhaps the strong effects of the adipogenic soluble factors were overriding the influence of cell shape on hMSC differentiation. In contrast to AIM, AMM only contained insulin but lacked dexamethasone, indomethacin, and 3-isobuty-lmethyl-xanthine (Table 1). Therefore, a reduced load in soluble factors could allow cell shape to play a more relevant role in controlling hMSC differentiation. The relative gene expression profile of adipogenic and osteogenic markers for this set is shown in Fig. 3B. As expected, the relative expression of the selected adipogenic markers was reduced relative to cells cultured in AIM: OIM (Fig. 3A versus Fig. 3B). Relative to the day 1 control, differences in the expression of PPARc in the round or spread cells were not significantly different (p = 0.631). Moreover, CFD expression was significantly increased by 13.8-fold relative to the day 1 control (p  0.042), but significant differences in the gene expression of CFD between round and elongated cells were not observed (p = 0.878). That said, AFABP expression in spread cells was significantly decreased relative to round cells (p < 0.050). Finally, since the AMM:OIM contained less adipogenic soluble factors than the AIM:OIM, we initially expected the upregulation of osteogenic markers in the AMM:OIM group relative to the AIM:OIM group.

However, none of the osteogenic markers were enhanced when compared with the day 1 control (p  0.295). Furthermore, the expression of the late osteogenic marker SPP1 was under the detection limit. These results can potentially be explained by the reduction of dexamethasone in the overall concentration of the AMM:OIM mixture, which is a common soluble induction factor for osteogenic and adipogenic lineages. To gain additional insight on the effects of cell shape on hMSC differentiation in 3D, we evaluated the expression of differentiation markers in the absence of any soluble induction factors. The spontaneous differentiation of hMSCs cultured in growth medium (GM) was assessed in terms of the relative expression of adipogenic, osteogenic, chondrogenic and smooth muscle cell markers. As shown in Fig. 4A, the adipogenic markers, PPARc, CFD and AFABP, and the osteogenic markers, TNAP and SPP1 were below the detection limit, and the relative expression of RUNX2 was indistinguishable among the experimental groups. Similarly, Fig. 4B shows that the chondrogenic markers, SOX9 and Coll 2, and the smooth muscle marker, SM22a, were also below the detection limit. Moreover, the expression of the chondrogenic marker, ACAN, was downregulated 2.2-fold in the spread cells and relative to the day 1 control group (p = 0.027), but there were not statistical differences when compared with cells exhibiting round morphologies (p = 0.282). No significant differences were observed in the expression of the smooth muscle markers, ACTA2 and CNN1 (p  0.167). Combined, the qRT-PCR results across these 3 media conditions suggest that hMSCs encapsulated in the 6.0 kDa PEGDA IPN (E0
37 kPa) display minimal differences in gene expression with cell morphology. 3.5. Protein expression analyses Gene expression analysis is an important source of cell phenotypic information due to its rapid, sensitive, and broad screening capabilities. However, the interpretation of the data must be exerted with caution since qRT-PCR results represent a snapshot of the possible current or future production of cellular proteins. To confirm the gene expression profile data for the 6.0 kDa PEGDA IPNs, we evaluated the cumulative response of round and spread cells at the protein level. Western blot analyses of the AIM:OIM (Fig. 5A) or AMM:OIM (Fig. 5B) treatment groups indicated that the protein levels of the adipogenic markers, CFD and AFABP, and of the osteogenic markers, RUNX2, TNAP and OSX, were not significantly impacted by differences in cell morphology (p  0.095). With regards to the group of round and spread hMSCs cultured in GM, western blot analyses revealed that only RUNX2 and TNAP were above the detection limit (Fig. 5C). Furthermore, differences between the round and spread treatment groups were not statistically significant (p = 0.632 for RUNX2 and p  0.089 for TNAP). Representative western blot images are provided in Supplementary Fig. 3. 3.6. Effect of modulus It is widely accepted that matrix rigidity and viscoelastic properties have a strong impact in directing hMSC differentiation [32–35]. In screening hMSCs encapsulated within 6.0 kDa PEGDA IPNs, qRT-PCR analyses suggested that cell shape appeared to only significantly modulate the gene expression of the late adipogenic differentiation marker AFABP, although this effect was not seen at the protein level (Fig. 5A). We hypothesized that the use of the less rigid 10.0 kDa PEGDA IPNs (15 kPa versus 37 kPa) could potentially enhance the effects of cell shape. Fig. 6A and Fig. 6B show the expression of adipogenic and osteogenic markers at the gene and protein level, respectively, in the 10.0 kDa PEGDA IPNs. To evaluate the effects of cell shape on the differentiation potential

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

173

Fig. 3. Relative gene expression profile of the adipogenic and osteogenic markers, PPARc, CFD, AFABP, RUNX2, TNAP and SPP1, by hMSCs in 6.0 kDa PEGDA IPNs. (A) hMSCs cultured in AIM:OIM after 14 days in culture. (B) hMSCs cultured in AMM:OIM after 14 days in culture. Results are reported as mean ± standard deviation, n = 4–6. * Significantly different from day 1 controls, p  0.045; #significantly different from 6.0 kDa PEGDA IPN supporting round cells, p  0.034; ND, non-detected.

Fig. 4. Relative gene expression profile of adipogenic, osteogenic, chondrogenic and smooth muscle markers of hMSCs in 6.0 kDa PEGDA IPNs after 14 days in culture in GM. (A) Adipogenic and osteogenic markers. (B) Chondrogenic and smooth muscle markers. Results are reported as mean ± standard deviation, n = 4–6. *Significantly different from day 1 controls, p = 0.027; ND, non-detected.

Fig. 5. Relative protein expression levels of the adipogenic markers CFD and AFABP and the osteogenic markers RUNX2, TNAP and OSX for hMSCs cultured in 6.0 kDa PEGDA IPNs. (A) hMSCs exhibiting round and spread morphologies after 14 days in culture in AIM:OIM. (B) hMSCs exhibiting round and spread morphologies after 14 days in culture in AMM:OIM. (C) hMSCs exhibiting round and spread morphologies after 14 days in culture in GM. Results are reported as mean ± standard deviation, n = 4; ND, nondetected.

of hMSCs in 10.0 kDa PEGDA IPNs, we used the traditional mixture of induction media, AIM:OIM [12,13,18], for 14 days. At the gene expression level, the adipogenic markers, PPARc, CFD and AFABP, were each increased relative to day 1 controls (p < 0.001). However, no differences in the levels of any of the adipogenic markers were observed between the round and spread treatment groups (p  0.09). Similarly, no differences in expression were observed between round and spread hMSCs for the osteogenic markers, RUNX2 and TNAP (p  0.06), although TNAP was significantly downregulated relative to day 1 controls for both groups (p < 0.001). In addition, western blot analysis indicated a

lack of significant differences between the groups in the expression of the adipogenic markers, CFD and AFABP and the osteogenic markers, RUNX2, TNAP and OSX. In general, these results indicate that cell morphology did not significantly affect the expression of adipogenic or osteogenic markers at the gene expression or protein expression level for hMSCs cultured for 14 days in 10.0 kDA PEGDA IPNs (E0
15 kPa). To gain further resolution of the end point differentiation state of the hMSCs at the protein level, multiplex immunoassays for additional adipogenic and osteogenic markers were performed. These assays have greater sensitivity than western blot analyses

174

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

Fig. 6. Relative hMSC expression of adipogenic and osteogenic markers in 10.0 kDa PEGDA IPNs after 14 days in culture in AIM:OIM. (A) Gene expression analysis using qRTPCR. (B) Relative protein levels using western blot analyses. Results are reported as mean ± standard deviation, n = 4–6; *Significantly different from day 1 controls, p  0.001; ND, non-detected.

and therefore could potentially detect subtle differences in differentiation between the round and elongated treatment groups. As shown in Fig. 7, the levels of the adipogenic markers, AFABP and leptin, and the osteogenic markers, osteonectin (ON) and SPP1, were not significantly different between round and elongated cells (p  0.075) based on multiplex analysis. As a further confirmation of the qRT-PCR and protein level results, we performed staining of lipid droplets and calcium deposits, which are late markers of adipogenic and osteogenic lineages (Figs. 8 and 9). In the staining, hMSCs cultured for 14 days in GM media served as negative controls (Supplementary Fig. 4).The overall quantitative and qualitative assessments of these stainings complemented and confirmed the observations of cell behavior at the gene and protein expression levels. In summary, the histological analysis showed that round or spread cell morphologies did not significantly impact the deposition of lipids or calcium minerals.

Fig. 8. Cell counts for Oil red O and Alizarin red stainings of hMSCs after 14 d in culture in 10.0 kDa PEGDA IPNs in AIM:OIM. Experimental data is reported as mean ± standard deviation, 1230–1810 cells per sample specimen were counted from 4 independent samples.

4. Discussion For the fabrication of a functional tissue substitute using tissue engineering approaches, it is important that the resulting material has the flexibility to promote the appropriate cell behavior associated with the target tissue. However, the isolated effects of cell morphology on the acquisition of appropriate phenotypes are difficult to study in 3D contexts due to the interconnection of scaffold properties such as chemical composition, structural organization, matrix rigidity and diffusional constraints. In this work, we demon-

strated that the proposed collagen-PEGDA IPNs, fabricated via sequential infiltration, can be used to expose hMSCs to microenvironments that display similar crosslink densities, matrix structures, average mesh size, and elastic properties while promoting distinct cell morphologies. This platform offers significant advantages over other previously explored strategies which could not be used directly to fully isolate the effect of cell shape from elastic modulus [18], crosslink densities, or prolonged cell-mediated

Fig. 7. Relative expression of adipogenic and osteogenic markers using immunoassay multiplexing after 14 days in culture in 10.0 kDa PEGDA IPNs in AIM:OIM. (A) AFABP, (B) leptin, (C) osteonectin (ON) and (D) SPP1. Results are reported as mean ± standard deviation, n = 6.

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

175

significantly impact the differentiation potential of hMSCs in 3D contexts. The overall advantage of our approach is the fact that we were able to demonstrate this observation directly without the influence of differences in matrix crosslink density, remodeling, or the use of small molecules to inhibit cytoskeleton tension [12,19]. 5. Conclusions

Fig. 9. Representative images of lipid droplets and calcium deposit stainings in hMSCs after 14 days in culture in 10.0 kDa PEGDA IPNs in AIM:OIM. Lipid droplet formation was assessed using Oil red O staining while calcium deposition was evaluated using Alizarin red and von Kossa staining. Scale bar = 50 lm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

matrix reorganization [19]. To our understanding, this is the first study in which the effects of cell shape on hMSC differentiation have been studied independently of scaffold chemical and physical properties. In the collagen-PEGDA IPNs, distinct cell morphologies were obtained by controlling the spreading time within a time frame of 6 h. Via this approach, round or elongated cells can be observed over a broad modulus range without altering the biochemical landscape of the scaffolds. Since elongated or round cell shapes can be found in both hard or soft tissues, we successfully evaluated the effects of shape using two different sets of IPNs that exhibited 37 kPa or 15 kPa average storage modulus, within similar range of previously published studies [18]. The differences in elastic properties of the scaffolds were achieved by changing the PEGDA molecular weight from 6.0 kDa to 10.0 kDa. For the 6.0 kDa PEGDA IPNs, we explored different cell culture media compositions beyond the traditional mixtures involving adipogenic and osteogenic induction factors. In each medium condition, cumulative qRT-PCR and western blot results indicated similar lineage progression between round and elongated cells. To confirm that these observations were not a result of the selected matrix rigidity, the differentiation of round versus spread hMSCs was also examined in lower modulus 10.0 kDa PEGDA IPNs. In terms of adipogenic or osteogenic differentiation there was strong agreement among the gene expression results, the protein expression assessments by western or multiplex immunoassays, and the histological evaluation of lipid droplets or calcium deposits. Overall, the 6.0 kDa PEGDA IPN and the 10.0 kDa PEGDA IPN data demonstrated that cell shape alone did not significantly influence the differentiation of hMSCs, at least in the range modulus, matrix composition and soluble factors explored. In our approach, round or elongated cells experience similar cell-matrix interactions while the presence of a PEGDA network provides a pure elastic behavior with time invariant mechanical properties for the time frame of the experiment. Our current findings are in close agreement and complement previously published data suggesting that differences in cell morphology alone do not

In this work, we demonstrated that the sequential fabrication of collagen-PEGDA IPNs resulted in a platform that can be used to control the degree of cell spreading of hMSCs independently of matrix composition, matrix structure, crosslink density or elastic modulus. Round or elongated cell morphologies were achieved by controlling the spreading time over a 6 h time period. In these scaffolds, the initial cell-matrix interactions of round or elongated cells were assessed by analysis of bound integrin a2 and collagen fiber organization were also consistent between the two groups. The cumulative set of data revealed that after 14 days in culture, the end point protein expression levels of adipogenic or osteogenic markers were not significantly different between the round and spread cells. Under the experimental conditions included in this research work—two elastic moduli and 3 culture media conditions—we conclude that cell shape is insufficient to modulate on its own the differentiation of hMSCs in 3D contexts. Despite these findings, cell shape is and must be a relevant parameter to consider when designing biomaterials given its presence as a fundamental physiological feature/organizer of functional tissue. The sequential fabrication process of the IPN is potentially an initial approach toward the fabrication of new scaffolds in which cell shape can be precisely controlled. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.actbio.2019.01.052. References [1] E. Dantuma, S. Merchant, K. Sugaya, Stem cells for the treatment of neurodegenerative diseases, Stem Cell Res. Ther. 1 (5) (2010) 37, https://doi. org/10.1186/scrt37. [2] M. Rodrigues, L.G. Griffith, A. Wells, Growth factor regulation of proliferation and survival of multipotential stromal cells, Stem Cell Res. Ther. 1 (4) (2010) 32, https://doi.org/10.1186/scrt32. [3] M.F. Pittenger, A.M. Mackay, S.C. Beck, R.K. Jaiswal, R. Douglas, J.D. Mosca, M.A. Moorman, D.W. Simonetti, S. Craig, D.R. Marshak, Multilineage potential of adult human mesenchymal stem cells, Science 284 (5411) (1999) 143–147, https://doi.org/10.1126/science.284.5411.143. [4] M.L.B.K. Dominici, K. Le Blanc, I. Mueller, I. Slaper-Cortenbach, F.C. Marini, D.S. Krause, R.J. Deans, A. Keating, D.J. Prockop, E.M. Horwitz, Minimal criteria for defining multipotent mesenchymal stromal cells, The International Society for Cellular Therapy position statement, Cytotherapy 8 (4) (2006) 315–317, https://doi.org/10.1080/14653240600855905. [5] G. Chamberlain, J. Fox, B. Ashton, J. Middleton, Concise review: mesenchymal stem cells: their phenotype, differentiation capacity, immunological features, and potential for homing, Stem Cells 25 (11) (2007) 2739–2749, https://doi. org/10.1634/stemcells.2007-0197. [6] D.S. Benoit, M.P. Schwartz, A.R. Durney, K.S. Anseth, Small functional groups for controlled differentiation of hydrogel-encapsulated human mesenchymal stem cells, Nat. Mater. 7 (10) (2008) 816–823, https://doi.org/10.1038/ nmat2269. [7] D.E. Discher, D.J. Mooney, P.W. Zandstra, Growth factors, matrices, and forces combine and control stem cells, Science 324 (5935) (2009) 1673–1677, https:// doi.org/10.1126/science.1171643. [8] D.J. Munoz-Pinto, A.C. Jimenez-Vergara, T.P. Gharat, M.S. Hahn, Characterization of sequential collagen-poly (ethylene glycol) diacrylate interpenetrating networks and initial assessment of their potential for vascular tissue engineering, Biomaterials 40 (2015) 32–42, https://doi.org/ 10.1016/j.biomaterials.2014.10.051. [9] D.J. Munoz-Pinto, X. Qu, L. Bansal, H.N. Hayenga, J. Hahn, M.S. Hahn, Relative impact of form-induced stress vs. uniaxial alignment on multipotent stem cell myogenesis, Acta Biomater. 8 (11) (2012) 3974–3981, https://doi.org/10.1016/ j.actbio.2012.06.044.

176

A.C. Jimenez-Vergara et al. / Acta Biomaterialia 87 (2019) 166–176

[10] K. Metavarayuth, P. Sitasuwan, X. Zhao, Y. Lin, Q. Wang, Influence of surface topographical cues on the differentiation of mesenchymal stem cells in vitro, ACS Biomater. Sci. Eng. 2 (2) (2016) 142–151, https://doi.org/10.1021/ acsbiomaterials.5b00377. [11] F. Guilak, D.M. Cohen, B.T. Estes, J.M. Gimble, W. Liedtke, C.S. Chen, Control of stem cell fate by physical interactions with the extracellular matrix, Cell Stem Cell 5 (1) (2009) 17–26, https://doi.org/10.1016/j.stem.2009.06.016. [12] R. McBeath, D.M. Pirone, C.M. Nelson, K. Bhadriraju, C.S. Chen, Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment, Dev. Cell 6 (4) (2004) 483–495, https://doi.org/10.1016/S1534-5807(04)00075-9. [13] K.A. Kilian, B. Bugarija, B.T. Lahn, M. Mrksich, Geometric cues for directing the differentiation of mesenchymal stem cells, Proc. Natl. Acad. Sci. U.S.A. 107 (11) (2010) 4872–4877, https://doi.org/10.1073/pnas.0903269107. [14] X. Qu, W. Zhu, S. Huang, Y.S. Li, S. Chien, K. Zhang, S. Chen, Relative impact of uniaxial alignment vs. form-induced stress on differentiation of human adipose derived stem cells, Biomaterials 34 (38) (2013) 9812–9818, https:// doi.org/10.1016/j.biomaterials.2013.09.009. [15] 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 (4) (2015) 218–226, https://doi.org/10.1021/ab500003s. [16] D. Zhang, K.A. Kilian, The effect of mesenchymal stem cell shape on the maintenance of multipotency, Biomaterials 34 (16) (2013) 3962–3969, https://doi.org/10.1016/j.biomaterials.2013.02.029. [17] M.W. Tibbitt, K.S. Anseth, Hydrogels as extracellular matrix mimics for 3D cell culture, Biotechnol. Bioeng. 103 (4) (2009) 655–663, https://doi.org/10.1002/ bit.22361. [18] N. Huebsch, P.R. Arany, A.S. Mao, D. Shvartsman, O.A. Ali, S.A. Bencherif, J. Rivera-Feliciano, D.J. Mooney, Harnessing traction-mediated manipulation of the cell/matrix interface to control stem-cell fate, Nat. Mater. 9 (6) (2010) 518–526, https://doi.org/10.1038/NMAT2732. [19] S. Khetan, M. Guvendiren, W.R. Legant, D.M. Cohen, C.S. Chen, J.A. Burdick, Degradation-mediated cellular traction directs stem cell fate in covalently crosslinked three-dimensional hydrogels, Nat. Mater. 12 (5) (2013) 458–465, https://doi.org/10.1038/NMAT3586. [20] O. Chaudhuri, L. Gu, D. Klumpers, M. Darnell, S.A. Bencherif, J.C. Weaver, N. Huebsch, H.P. Lee, E. Lippens, G.N. Duda, D.J. Mooney, Hydrogels with tunable stress relaxation regulate stem cell fate and activity, Nat. Mater. 15 (3) (2016) 326–334, https://doi.org/10.1038/NMAT4489. [21] D.J. Munoz-Pinto, B. Grigoryan, J. Long, M. Grunlan, M.S. Hahn, An approach for assessing hydrogel hydrophobicity, J. Biomed. Mater. Res. Part A 100 (10) (2012) 2855–2860, https://doi.org/10.1002/jbm.a.34289. [22] S.J. Bryant, K.S. Anseth, Hydrogel properties influence ECM production by chondrocytes photoencapsulated in poly (ethylene glycol) hydrogels, J. Biomed. Mater. Res. 59 (1) (2002) 63–72, https://doi.org/10.1002/jbm.1217. [23] S.J. Bryant, C.R. Nuttelman, K.S. Anseth, Cytocompatibility of UV and visible light photoinitiating systems on cultured NIH/3T3 fibroblasts in vitro, J.

[24]

[25]

[26]

[27]

[28]

[29]

[30] [31]

[32]

[33]

[34]

[35]

Biomater. Sci. Polym. Ed. 11 (5) (2000) 439–457, https://doi.org/10.1163/ 156856200743805. C.G. Williams, A.N. Malik, T.K. Kim, P.N. Manson, J.H. Elisseeff, Variable cytocompatibility of six cell lines with photoinitiators used for polymerizing hydrogels and cell encapsulation, Biomaterials 26 (11) (2005) 1211–1218, https://doi.org/10.1016/j.biomaterials.2004.04.024. D.J. Munoz-Pinto, A.C. Jimenez-Vergara, L.M. Gelves, R.E. McMahon, V. GuizaArguello, M.S. Hahn, Probing vocal fold fibroblast response to hyaluronan in 3D contexts, Biotechnol. Bioeng. 104 (4) (2009) 821–831, https://doi.org/10.1002/ bit.22436. A.C. Jimenez-Vergara, J. Lewis, M.S. Hahn, D.J. Munoz-Pinto, An improved correlation to predict molecular weight between crosslinks based on equilibrium degree of swelling of hydrogel networks, J. Biomed. Mater. Res. B Appl. Biomater. 106 (3) (2018) 1339–1348, https://doi.org/10.1002/jbm. b.33942. S. Dånmark, A. Finne-Wistrand, A.C. Albertsson, M. Patarroyo, K. Mustafa, Integrin-mediated adhesion of human mesenchymal stem cells to extracellular matrix proteins adsorbed to polymer surfaces, Biomed. Mater. 7 (3) (2012), https://doi.org/10.1088/1748-6041/7/3/035011 035011. S.M. Becerra-Bayona, V.R. Guiza-Arguello, B. Russell, M. Höök, M.S. Hahn, Influence of collagen-based integrin a1 and a2 mediated signaling on human mesenchymal stem cell osteogenesis in three dimensional contexts, J. Biomed. Mater. Res. Part A 106A (10) (2018) 2594–2604, https://doi.org/10.1002/jbm. a.36451. M.C. Jones, J.D. Humphries, A. Byron, A. Millon-Frémillon, J. Robertson, N.R. Paul, D.H. Ng, J.A. Askari, M.J. Humphries, Isolation of integrin-based adhesion complexes, Curr. Protoc. Cell Biol. 66 (1) (2015), https://doi.org/10.1002/ 0471143030.cb0908s66. 9.8.1–9.8.15. K. Zhang, J. Chen, The regulation of integrin function by divalent cations, Cell Adhes. Migrat. 6 (1) (2012) 20–29, https://doi.org/10.4161/cam.18702. S. Lin, N. Sangaj, T. Razafiarison, C. Zhang, S. Varghese, Influence of physical properties of biomaterials on cellular behavior, Pharm. Res. 28 (6) (2011) 1422–1430, https://doi.org/10.1007/s11095-011-0378-9. A.J. Engler, S. Sen, H.L. Sweeney, D.E. Discher, Matrix elasticity directs stem cell lineage specification, Cell 126 (4) (2006) 677–689, https://doi.org/10.1016/ j.cell.2006.06.044. M. Guvendiren, J.A. Burdick, Stiffening hydrogels to probe short-and long-term cellular responses to dynamic mechanics, Nat. Commun. 3 (2012) 792, https:// doi.org/10.1038/ncomms1792. O. Chaudhuri, L. Gu, M. Darnell, D. Klumpers, S.A. Bencherif, J.C. Weaver, N. Huebsch, D.J. Mooney, Substrate stress relaxation regulates cell spreading, Nat. Commun. 6 (2015) 6365, https://doi.org/10.1038/ncomms7365. K.H. Vining, D.J. Mooney, Mechanical forces direct stem cell behaviour in development and regeneration, Nat. Rev. Mol. Cell Biol. 18 (12) (2017) 728– 742, https://doi.org/10.1038/nrm.2017.108.