- Email: [email protected]
Three-dimensional extracellular matrix-directed cardioprogenitor differentiation: Systematic modulation of a synthetic cell-responsive PEG-hydrogel
Biomaterials 29 (2008) 2757–2766
Contents lists available at ScienceDirect
Biomaterials journal homepage: www.elsevier.com/locate/biomaterials
Three-dimensional extracellular matrix-directed cardioprogenitor differentiation: Systematic modulation of a synthetic cell-responsive PEG-hydrogel Thomas P. Kraehenbuehl a, Prisca Zammaretti a, Andre´ J. Van der Vlies a, Ronald G. Schoenmakers a, Matthias P. Lutolf b, Marisa E. Jaconi c, Jeffrey A. Hubbell a, * a b c
Institute of Bioengineering and Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fe´de´rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland Institute of Bioengineering, Ecole Polytechnique Fe´de´rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland Department of Pathology and Immunology, Faculty of Medicine, University of Geneva, 1 Rue Michel-Servet, 1211 Geneva 4, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 January 2008 Accepted 16 March 2008 Available online 7 April 2008
We show that synthetic three-dimensional (3D) matrix metalloproteinase (MMP)-sensitive poly(ethylene glycol) (PEG)-based hydrogels can direct differentiation of pluripotent cardioprogenitors, using P19 embryonal carcinoma (EC) cells as a model, along a cardiac lineage in vitro. In order to systematically probe 3D matrix effects on P19 EC differentiation, matrix elasticity, MMP-sensitivity and the concentration of a matrix-bound RGDSP peptide were modulated. Soft matrices (E ¼ 322 64.2 Pa, stoichiometric ratio: 0.8), mimicking the elasticity of embryonic cardiac tissue, increased the fraction of cells expressing the early cardiac transcription factor Nkx2.5 around 2-fold compared to embryoid bodies (EB) in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa, stoichiometric ratio: 1.2) decreased the number of Nkx2.5-positive cells significantly. Further indicators of cardiac maturation were promoted by ligation of integrins relevant in early cardiac development (a5b1, avb3) by the RGDSP ligand in combination with the MMP-sensitivity of the matrix, with a 6-fold increased amount of myosin heavy chain (MHC)-positive cells as compared to EB in suspension. This precisely controlled 3D culture system thus may serve as a potential alternative to natural matrices for engineering cardiac tissue structures for cell culture and potentially therapeutic applications. Ó 2008 Elsevier Ltd. All rights reserved.
Keywords: Extracellular matrix Biomimetic hydrogel Poly(ethylene glycol) Cardiac tissue engineering P19 embryonal carcinoma cells Stem cells
1. Introduction Nearly eight million persons in the United States suffer from myocardial infarction, with an estimated 800,000 new incidences every year [1]. Current therapies are limited by the restricted intrinsic regeneration capacity of the myocardium and by the lack of organs for transplantation. In recent attempts to regenerate functional heart muscle, adult cells such as bone marrow-derived cells have been injected into the ventricular wall or coronary vessels in animal models with myocardial infarction, as well as in human patients in first clinical trials [2,3]. Although improved cardiac performance has been reported, it remains unclear if paracrine signaling effects of the implanted cells, endogenous stem cell recruitment or even trans-differentiation are responsible for the improvement of cardiac function [4,5]. Furthermore, efficacy of the cell engraftment is very low: less than 10% of the injected cells typically engraft, mainly due to cell death, and less than 2% appear * Corresponding author. Tel.: þ41 21 693 96 81; fax: þ41 21 693 96 85. E-mail address: jeffrey.hubbell@epfl.ch (J.A. Hubbell). 0142-9612/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2008.03.016
to take on the in vivo fate of cardiomyocytes [6,7]. To better retain the cells at the site of infarction and to better control growth and differentiation, cardiac grafts have been engineered in vitro employing biodegradable materials as cell carriers or as cell ingrowth matrices. Both naturally derived [8,9,10] and synthetic materials [11,12] have been demonstrated to support embryonic and adult cell-derived cardiac tissue development containing contracting cardiomyocytes. However, mechanisms controlling 3D cardiac development are still poorly understood. We have recently developed a 3D PEG-based synthetic hydrogel material mimicking key biochemical characteristics of natural collagenous matrices [13], the major constituent of cardiac extracellular matrices [14]. This artificial extracellular matrix system allows the nearly independent control of matrix elasticity, integrinstimulating ligands and protease-sensitivity and is thus a potentially powerful tool to direct 3D cardiac development. We have previously reported that matrix metalloproteinase (MMP)-sensitive peptides crosslinking branched PEG chains enable cell-mediated proteolytic matrix degradation and remodeling [15]. These characteristics are also important in the heart, both under physiological
2758
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
and pathological conditions, where increased MMP-expression and activation has been observed [16,17]. Furthermore, matrix-bound RGDSP peptide promotes cell adhesion and stimulation of integrins relevant in early cardiac development (a5b1, aVb3) [18,19,20]. Here, we examined the ability of these cell-responsive PEGbased hydrogels for directing cardiac differentiation of pluripotent P19 embryonal carcinoma cells, a well-accepted model system for embryonic stem cell-based cardiac differentiation [21,22]. We systematically modulated matrix elasticity, the concentration of the matrix-bound RGDSP oligopeptide and MMP-sensitivity to dissect biophysical and biochemical matrix parameters potentially involved in cardiac differentiation. 2. Materials and methods 2.1. Synthesis of poly(ethylene glycol)–vinylsulfone and peptide precursors (RGDSP, MMP-substrate) PEG-vinylsulfone was synthesized adapting a previous protocol [23]. Branched 4-arm PEG-OH (Mw ¼ 20,000 g/mol) (Shearwater Polymers, Huntsville, AL) was dried by azeotropic distillation in toluene (VWR, Nyon, Switzerland) for 4 h using a Dean–Stark trap. Toluene was distilled off and the residue dissolved in dichloromethane (Fisher Scientific, Wohlen, Switzerland). To the clear solution, sodium hydride (Sigma–Aldrich, Buchs, Switzerland) was added at 20-fold molar excess over OH-groups. After hydrogen evolution ceased, divinylsulfone (Fluka, Buchs, Switzerland) was added at a 50-fold molar excess over OH-groups. The reaction was carried out at room temperature under argon with constant stirring for 3 d. After the addition of acetic acid (Fluka, Buchs, Switzerland) to neutralize excess sodium hydride, the mixture was filtered over a filter cell cake and concentrated by rotary evaporation. The polymer was then isolated by precipitation in ice-cold diethylether (Brunschwig, Basel, Switzerland) and filtered. Finally, the product was dried under vacuum to give a white solid (yield: 85%). The degree of PEG functionalization with vinylsulfone was determined by proton NMR spectroscopy (in CDCl3) using a Bruker 400 spectrometer (Bruker BioSpin, Faellanden, Switzerland). Characteristic vinylsulfone peaks were observed at 6.1, 6.4, and 6.8 ppm. The degree of end group conversion was found to be z95%. The polymer was stored under argon at 20 C until used. RGDSP and the substrates for matrix metalloproteinases (MMP) were synthesized by solid phase peptide synthesis using NovaSyn TGR resin (Merck Biosciences, Laeufelfingen, Switzerland) with an automated peptide synthesizer (Chemspeed, Augst, Switzerland) with standard Fmoc chemistry. Purification was performed by mass-directed reverse phase-C18 HPLC using a Waters Autopurification System. Separation and collection were performed by UV with broad wavelength detection (210–400 nm) (Waters PDA 996 UV photodiode array) and mass directed software (Waters Masslynx software). Peptide sequences were confirmed by ion trap ESI mass spectrometry (all Waters, Baden-Daettwil, Switzerland).
2.3. Cell culture and differentiation within PEG-based hydrogels P19 embryonic carcinoma (EC) cells (LGC Promochem, Molsheim, France) were grown in monolayer cultures in alpha-minimum essential medium (Invitrogen, Basel, Switzerland) supplemented with 7.5% bovine calf serum and 2.5% fetal bovine serum (LGC Promochem, Molsheim, France) (Fig. 2A). For differentiation in suspension (control cells), embryoid body (EB) formation and differentiation were induced by dissociating the undifferentiated P19 EC cells colonies with trypsin and resuspending them in differentiation medium containing 0.8% dimethylsulfoxide (DMSO) (Sigma–Aldrich, Buchs, Switzerland) in non-treated Petri dishes for 4 d. The EB were then transferred into tissue culture-treated Petri dishes containing medium without DMSO. For differentiation in the PEG-based hydrogels, undifferentiated single cells were mixed into the gel precursor solution at a concentration of 100,000 or 500,000 cells/ml gel. After the crosslinking reaction at 37 C, the cell-seeded gels of 25 mL volume (before swelling) were cultured in 24-well plates with 1.5 ml of differentiation medium for 4 d followed by culture in medium without DMSO. Medium was exchanged every third day. In some experiments, broad-range MMP-inhibitor GM6001 (Chemicon International, Temecula, CA) was added to the medium at 25 mM concentration 12 h after seeding the cells into the PEG-based hydrogels. The duration of the experiments was 14 d in general, and exceptionally 28 d for qualitatively analyzing cell behavior in the 3D PEG-hydrogels.
2.4. Cell proliferation Cell proliferation was measured with Alamar Blue (Lucerna-Chem, Lucerne, Switzerland). This non-toxic assay allowed the repetitive monitoring of the same cell samples [25,26]. Gels were overlaid with fresh medium containing 4% Alamar Blue solution and incubated for 6 h at 37 C. The supernatant was then collected and analyzed by measuring the fluorescence emission at 605 nm (excitation wavelength: 575 nm) (Safire, Tecan, Maennedorf, Switzerland).
2.5. FACS analysis For quantitative analysis by flow cytometry, gels were digested at 14 d after seeding, and single cells were extracted as described above. The cells were washed with ice-cold washing buffer (HBSS containing 0.1% w/v BSA) and fixed with 1% formaldehyde for 15 min. Cells were then incubated with the primary antibodies at 4 C overnight in washing buffer containing 0.3% (w/v) Saponin (Axon Lab, Le Montsur-Lausanne, Switzerland). Primary antibodies were monoclonal mouse antimyosin heavy chain (Abcam, Cambridge, UK, 1:100), polyclonal rabbit anti-Nkx2.5 (1:100), monoclonal mouse anti-MyoD (1:100) and monoclonal mouse anti-Oct-4 (all from Santa Cruz, Heidelberg, Germany, 1:100). After washing, the cells were incubated with the secondary antibodies goat anti-rabbit IgG (Alexa Fluor 488, 1:400) and goat anti-mouse IgG1 and IgG2b (Alexa Fluor 647, 1:400, both Invitrogen, Basel, Switzerland) at 4 C for 30 min. Non-conjugated isotype controls were used as negative controls (IgG and IgG2b from Santa Cruz, IgG1 from Chemicon International, Temecula, CA). The samples were analyzed by flow cytometry using a CyAn ADP (Dako, Baar, Switzerland). Between 40,000 and 100,000 cell events were counted per sample. Processing of the data was performed using FlowJo software (TreeStar, Ashland, OR).
2.2. Formation and characterization of PEG-based hydrogels Gel formation was done under physiological conditions as described elsewhere [23]. Briefly, the synthesis was carried out through Michael-type addition reaction of thiol-containing peptides onto vinylsulfone-functionalized PEG. Four-arm PEG-vinylsulfone was dissolved in triethanolamine buffer (0.3 M, pH 8.0) (Fluka, Buchs, Switzerland) to give a 10% (w/v) solution. A solution of the integrin ligand peptide (Ac-GCGYGRGDSPG-NH2) in the same buffer was added to the 4-arm PEG-vinylsulfone solution. After 10 min, a suspension of cells and MMP-sensitive peptide (Ac-GCRDGPQGIWGQDRCG-NH2) in the triethanolamine buffer was added. Although gel formation occurred within minutes, the crosslinking reaction was continued for around 30 min. Gels were characterized by their storage and loss modulus in small-strain oscillatory shear using a Bohlin CVO 120 high resolution rheometer (Malvern Instruments, Worcestershire, UK) with plate–plate geometry at 25 C under a humidified atmosphere. Storage and loss modulus as well as phase angle were measured as a function of frequency (from 0.05 to 5 Hz) in a constant strain mode (0.05). Young’s modulus (E ) was then calculated from the elastic shear modulus G0 using the formula E ¼ 3G0 . The swelling ratio of the gels was determined by the weight ratio of the gel before and after swelling in water overnight. Mesh size was determined as previously described [24]. The stoichiometric ratio was defined as the ratio between PEG-arms and crosslinker. To enable cell analysis at the gene level by RT-PCR and at the protein level through flow cytometry, the cells were extracted from the matrix after controlled degradation using collagenase IV. Gel digestion was carried out on ice for 1 h with 2 ml collagenase IV (Invitrogen, Basel, Switzerland) of 0.5 mg/ml in Hanks balance salt solution (Invitrogen, Basel, Switzerland) containing 0.1% bovine serum albumin (Sigma-Aldrich, Buchs, Switzerland) (¼washing buffer). The remaining gravimetric weight was measured (Supplementary Fig. 1).
2.6. Immunostaining and image analysis with confocal microscopy For confocal imaging, the gels were fixed in 2% paraformaldehyde for 30 min. After washing, the cells were permeabilized with a PBS solution containing 0.25% Triton-X (Sigma–Aldrich, Buchs, Switzerland) for 45 min at room temperature. The samples were then incubated with the primary antibodies at 4 C overnight in washing buffer. After three washes for 1 h each, cells were incubated with the secondary antibodies at 4 C overnight. The samples were washed 3 for 1 h. Cell nuclei were stained with DAPI (Invitrogen, Basel, Switzerland) at 300 nM concentration for 3 min. After mounting the samples with Vectashield (Reactolab, Servion, Switzerland), cells were stored at 4 C for imaging analysis. All washing and incubation steps were performed on a XY-shaker. For imaging, LSM510 Meta confocal microscope (Carl Zeiss, Feldbach, Switzerland) was used. Central layers through the cell clusters were analyzed using Imaris software.
2.7. Immunohistochemistry For immunohistochemistry, gels were fixed for 30 min with 2% paraformaldehyde 14 d after seeding, and embedded in Optimum Cutting Temperature (O.C.T.) formulation (Tissue Tek, Electron Microscopy Sciences, Hatfield, PA). The gels were frozen in isopentane (Acros Organics, Geel, Belgium) and sectioned with a cryostat (Microm Microtech, Francheville, France) at 18 C. Transverse sections (10 mm thickness) were placed on microscope slides and routinely processed for hematoxylin and eosin (Medite Medizintechnik, Nunningen, Switzerland) staining. Samples were then analyzed with a Axiovert 200M microscope (Carl Zeiss, Feldbach, Switzerland).
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766 2.8. RT-PCR analysis For isolating RNA from the cells, gels were digested as described above. Total RNA was isolated at 14 d after seeding with Trizol Reagent (Invitrogen, Basel, Switzerland) following the supplier’s protocol. Total RNA was quantified by a UV spectrophotometer (WPA, Cambridge, UK). Reverse Transcription was performed on 1 mg total RNA with equal volumes of oligo-dT primers and random hexamers, dNTPs (all from Promega, Wallisellen, Switzerland) and Superscript II Reverse Transcriptase (Invitrogen, Basel, Switzerland). PCR was carried out using Taq polymerase (Qiagen, Hombrechtikon, Switzerland). The amplified products were separated with 2% agarose gels containing ethidium bromide (both from Invitrogen, Basel, Switzerland). Primers’ sequences and reaction conditions are given in Supplementary Table 1. 2.9. Statistical analysis The results of all cell experiments are mean values (SD) of samples performed in triplicates in two independent experiments. Comparative analysis was performed with two-tailed Student’s t-test using S-PLUS software. Differences between two data sets were considered statistically different when *p < 0.05, **p < 0.001.
3. Results 3.1. Physicochemical characteristics of PEG-based hydrogels To modulate the biophysical matrix characteristics in a range possibly relevant for cardiac differentiation, the stoichiometric ratio between PEG and MMP-sensitive linkers was systematically altered (Fig. 1A). Cardiac cell differentiation was examined in a soft matrix (E ¼ 322 64.2 Pa) and in a stiffer one (E ¼ 4036 64.2 Pa) obtained at stoichiometric ratio of 0.8 and 1.2, respectively. The minimum value of the Young’s modulus (E ¼ 322 64.2 Pa) corresponds to the maximum value of the swelling ratio (Q ¼ 40 1.2) at a stoichiometric ratio of 0.8 (Fig. 1B). It is not possible to decouple matrix elasticity from the number of MMP-sensitive crosslinkers to assign cell behavior to one of these two matrix parameters. 3.2. Cell cluster formation in PEG-based hydrogels To reveal the potential of the cell-responsive hydrogel system for directing embryonal cardiac development, undifferentiated single cells were seeded into PEG-based hydrogels (Fig. 2B). After
2759
2 d, newly formed cell clusters appeared by cell proliferation, homogenously distributed as shown by phase contrast microscopy (Fig. 2C). Over time, the clusters grew concentrically, resulting in spherical shapes at 14 d with densely packed cells (Fig. 2D). The control EB produced and maintained in suspension were less compact than their counterparts in the gels. After 14 d in the 3D PEG-matrix, some cell clusters (<10%) started to converge by flattening the prospective contact area until cell–cell contact was established between neighboring cell clusters (Fig. 3A and B). In MMP-sensitive hydrogels containing the integrin ligand RGDSP at 100 mM, cells started to migrate out of the cell clusters after around 3 wk keeping their round-shaped morphology (Fig. 3C). Around 4 wk after seeding, we observed spindle-shaped cells radially growing out of the clusters forming well-connected networks (Fig. 3D). We next quantified cluster size from light microscopy images. Cluster size was affected by matrix elasticity, MMP-sensitivity and RGDSP concentration. Cells grown in stiffer matrices (E ¼ 4036 419.6 Pa) were smaller and more homogenous in size than clusters developed in soft matrices (E ¼ 322 64.2 Pa) (Fig. 4A). When inhibiting cell-mediated proteolytic degradation using a broadrange MMP-inhibitor (GM6001), the mean cluster diameter was significantly reduced compared to clusters grown without MMP-inhibition. Increasing the RGDSP concentration resulted in larger cell clusters compared to matrices without the integrin ligand (Fig. 4B). Absence of the differentiation-initiating DMSO in the culture medium affected neither cluster number nor size (data not shown). Notably, we did not detect spontaneously contracting cells within the gels. However, when we extracted cell clusters from the matrix at 14 d by collagenase treatment and plated the intact cell clusters in tissue culture-treated Petri dishes, we detected spontaneous contractions beginning at 3 d after seeding at a concentration of around 10 cell clusters per 10 cm Petri dish (Supplementary Fig. 2 and Supplementary movie). 3.3. Quantification of cell proliferation To assess 3D cell growth in the PEG matrices, cell proliferation was quantified within the sample construct every other day over a 14-d period. Gels with high cell seeding density (500,000 cells/ml gel) displayed a 2-fold faster proliferation rate compared to gels with lower cell seeding density (100,000 cells/ml gel) (Fig. 5A). The concentration of the RGDSP ligand significantly affected cell proliferation only between day 6 and 8 (Fig. 5B). The effect of matrix elasticity (E ¼ 322 64.2 Pa vs. E ¼ 4036 419.6 Pa) on cell proliferation was shown to be significant throughout the experiment (Fig. 5C). Addition of the MMP-inhibitor GM6001 decreased proliferation significantly (Fig. 5D). 3.4. Expression of cardiac markers on P19 EC cells in the PEG-matrix
Fig. 1. Young’s modulus and swelling properties of the cell-responsive PEG-based hydrogels. Young’s modulus (A) and swelling ratio (B) were modified by altering the stoichiometric ratio between PEG and MMP-sensitive crosslinkers. The maximum value of the Young’s modulus corresponds to the minimum value of the swelling ratio at r ¼ 1.2.
To analyze the influence of the extracellular matrix environment on gene and protein expression, the three matrix variables were systematically modulated as depicted in Table 1. Soft matrices (E ¼ 322 64.2 Pa) matching the elasticity of embryonic cardiac tissue [8,24], increased expression of the early cardiac transcription factor Nkx2.5 [27] by 2-fold compared to EB in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa) significantly decreased the number of Nkx2.5-positive cells as shown qualitatively at the gene (Fig. 6) and quantitatively at the protein level (Fig. 7 and Table 2). Addition of GM6001 did not significantly affect Nkx2.5 expression. For further cardiac maturation, the presence of RGDSP ligand to stimulate integrins involved in early cardiac development (a5b1, avb3) and the MMP-sensitivity of the matrix were shown to have a major effect as the expression of MHC, with an increase by 6-fold compared to EB in suspension. Addition of the broad-spectrum
2760
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Fig. 2. Culture and differentiation of P19 cells in 3D synthetic cell-responsive PEG-based hydrogels. (A) Undifferentiated P19 cells on tissue culture plastic. (B) Undifferentiated, single cells seeded into the PEG-based hydrogel (arrows). (C) Cell cluster formation after 2 d in the matrix in differentiation medium supplemented with 0.8% DMSO. (D) Cell cluster at 14 d after seeding into the matrix. Scale bar: 20 mm.
MMP-inhibitor resulted in a 50% decrease in MHC expression. Indeed, our MHC antibody detects both, cardiac MHC and skeletal muscle MHC, but this latter to at least five time lower specificity than the antibody binds to cardiac MHC according to the manufacturer. In addition, others [28] have shown that P19 EC cells aggregated in the
presence of DMSO and cultured for 10 d display only small amounts of skeletal muscle (<5%) assessed by the expression of embryonic MHC and the slow isoform of MHC, whereas cardiac muscle with MHC expression was detected even 3 d earlier and at a three time higher extent [29,30].
Fig. 3. Cell dynamics inside the PEG-based hydrogel. (A) Some cell clusters converge after 14 d by flattening the prospective contact area. (B) Neighboring cell clusters established cell–cell contacts after 14 d. (C) Around 3 wk after seeding undifferentiated single cells in hydrogels containing the RGDSP ligand (100 mM), the cells start to migrate out of the cell clusters. (D) Around 4 wk, some cells leave the cell clusters radially, showing spindle-shaped cell morphology. Scale bar: 20 mm. Scale bar insert image D: 10 mm.
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2761
Fig. 4. Size distribution of cell clusters as a function of the artificial extracellular matrix microenvironment. For each condition, the diameters of 100 randomly selected cell clusters per gel were analyzed at 14 d. (A) Different Young’s moduli: -, E ¼ 322 Pa; , E ¼ 4036 Pa; ,, E ¼ 4036 Pa, GM6001. (B) Different RGDSP concentrations: -, E ¼ 322 Pa; , E ¼ 4036 Pa; ,, E ¼ 4036 Pa, GM6001.
3.5. Self-renewal capacity and skeletal muscle differentiation of the P19 EC cells in the PEG-microenvironment To examine self-renewal capacity and other muscle differentiation of the PEG-matrix, the matrix environment was modulated as described above. The number of undifferentiated cells was determined on the gene (Fig. 6) and protein level by Oct-4 expression (Fig. 7 and Table 2), a marker for pluripotent embryonic stem cells typically down-regulated quickly upon differentiation. Cells in all PEG-matrix environments evaluated still expressed substantial amounts of Oct-4 compared to the control cells in suspension. However, quantitative differences were observed: whereas soft matrices (E ¼ 322 64.2 Pa) contained around 15% Oct-4-positive cells, stiffer matrices retained around 50% of the cells in an undifferentiated state (E ¼ 4036 419.6 Pa). MMP-inhibition did not significantly alter Oct-4 expression. Expression of skeletal muscle marker MyoD was significantly higher in stiffer matrices (E ¼ 4036 419.6 Pa), resembling embryonic skeletal muscle tissue [31], while soft matrices (E ¼ 322 64.2 Pa) with the same RGDSP concentration did not significantly promote MyoD expression (Fig. 7 and Table 2). Inhibition of MMP-activity did not significantly alter MyoD expression. The expression and localization of the self-renewal marker Oct-4, the cardiac marker Nkx2.5 and the skeletal muscle marker
MyoD were also assessed at 14 d by immunofluorescence staining and confocal imaging (Fig. 8). All three transcription factors were mainly located within the cell nuclei. With immunohistology of the 3D PEG-hydrogels with the embedded P19 EC cell clusters, we show a necrotic central area (arrow) in larger cell clusters, but not in smaller ones (Fig. 9). 4. Discussion Here, we explore the utility of a material system [13] to support and direct differentiation of a stem cell model along a targeted lineage, employing the pluripotent P19 EC model and targeting cardiomyocytes. Our data indicates the necessity to control several biophysical and biochemical features of the extracellular matrix to drive P19 EC cells, and presumably stem cells as cardioprogenitors, along the cardiac differentiation pathway. Compared to common matrices used in cardiac tissue engineering [32,33] our artificial extracellular matrix microenvironment allowed modulation of elasticity, adhesivity and proteolytic susceptibility nearly independently, and thus permitted dissection of the physiological complexity into experimentally amenable culture conditions. We show that initial cardiac muscle commitment was enhanced by the matrix elasticity. However, further maturation appeared to
2762
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Fig. 5. Cell proliferation in the 3D hydrogels measured with Alamar Blue. Cell proliferation was shown to be directed by the artificial extracellular matrix microenvironment and the cell density. (A) High cell density demonstrated significantly higher proliferation than low cell density: -, 100,000 cells/ml gel; , 500,000 cells/ml gel. (B) Presence of cell adhesion ligand (RGDSP): -, 0 mM; , 100 mM (C) Highest proliferation was demonstrated by matrices with low Young’s modulus: -, E ¼ 4036 Pa; , E ¼ 322 Pa. (D) Broad inhibition of cell-secreted MMP by GM6001 decreased proliferation: -, 0 mM; , 100 mM.
be influenced not by the matrix elasticity but also by the stimulation of integrins relevant in early cardiac development (a5b1, avb3) via the RGDSP ligand in combination with the MMP-sensitivity of the matrix. This limited effect of matrix stiffness in directing terminal differentiation was recently shown in experiments with human mesenchymal stem cells (hMSC) [34,35]. When seeded on collagen-coated polyacrylamide gels mimicking the elasticity of the respective natural microenvironment, hMSC were selectively directed toward myogenic, neurogenic and osteogenic programs, however, with limited expression of terminal differentiation. However, neither in our 3D model system nor in the abovementioned collagen-coated polyacrylamide gel, it is possible to assign the cell differentiation fully to either matrix elasticity or density of the mesh/fibers, respectively. However, we have shown the relation between gel elasticity and the number of cleavable crosslinker peptides (stoichiometric ratio) [23]. A stoichiometric ratio of less than 1 may lead to defects in the PEG-network due to the incomplete reaction and thus on the mechanical level decrease the Young’s modulus of the matrix, and on the cellular level provide additional room for migration or proliferation. Cell behavior
Table 1 Overview of the extracellular matrix conditions employed: matrix elasticity, concentration of RGDSP ligand, and the matrix sensitivity Condition
10 100 2 3 4 5
Description Age cells [d]
Young’s modulus [Pa]
RGDSP-concentration [mM]
Addition GM6001
0 14 14 14 14 14
0 (control, suspension) 0 (control, suspension) 322 64.2 322 64.2 4036 419.6 4036 419.6
0 (control, suspension) 0 (control, suspension) 0 100 100 100
No No No No No Yes
including differentiation may always be affected by the combined effects of matrix elasticity and number of MMP-sensitive crosslinkers without being able to completely decouple. In contrast to the recently discovered role of elasticity in regulation of stem cell fate, the importance of avb3 and a5b1- integrins during cardiac differentiation has been reported by studies in vitro and in vivo with knock-out models [36]. Deficiency of b1-integrin,
Fig. 6. Representative gene expression of P19 EC cells in the 3D PEG-based hydrogels at seeding (10 ) and 14 d (100 and 2–5) as assessed by RT-PCR. RNA was isolated by digestion of the matrix. For each sample, reverse transcription was performed with 1 mg RNA. Cells grown in suspension were used as a control. We show a fast downregulation of Oct-4 expression within 14 d of the control cells in suspension, but strong expression by the cells embedded in the 3D PEG-hydrogel. While GATA-4, as an endothelial marker and precursor of the cardiac pathway is expressed in all conditions at 14 d, Nkx2.5, which is considered as the first cardiac marker is expressed in the soft matrices (E ¼ 322 64.2 Pa) only, but is not supported by the stiffer matrix environment (E ¼ 4036 419.6 Pa). While MEF2C as a cardiac and skeletal transcription factor is expressed in all conditions at 14 d, beta MHC, which is know to be mainly a marker for mature cardiac cells is supported by the PEG-matrix.
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2763
Fig. 7. Representative protein expression of P19 EC cells in the 3D synthetic PEG-based hydrogels as assessed by FACS. Cells were extracted from the matrix at 14 d and subjected to flow cytometric analysis. Samples are shown in black, respective isotype controls in grey.
2764
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Table 2 Quantification of protein expression from cells extracted from the PEG-based hydrogel at 14 d as measured by flow cytometry Protein
Matrix parameters
Average
S.D.
p-Value
Significance
Oct-4
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
91.2 3.5 14.5 14.8 58.8 44.3
4.62 2.88 4.59 5.49 9.56 11.38
2.6E-12 Ref. 0.0006 0.0012 9.1E-08 6.7E-06
** Ref. ** ** ** **
Nkx2.5
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 13.3 18.2 21.0 9.5 9.3
N.A 2.80 5.78 5.25 2.59 4.13
N.A Ref. 0.0950 0.0103 0.0337 0.0781
N.A Ref. – * * –
MyoD
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 12.2 12.0 14.0 19.2 16.8
N.A 3.60 5.06 4.38 2.99 1.72
N.A Ref. 0.9489 0.4469 0.0044 0.0168
N.A Ref. – – * *
MHC
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 6.2 11.3 37.2 31.5 15.0
N.A 1.94 2.58 8.86 5.01 3.41
N.A Ref. 0.0029 0.0000 0.0000 0.0003
N.A Ref. * ** ** *
S.D.: standard deviation, N.A.: not applicable, Ref.: reference for p-value. Self-renewal capacity as evident from Oct-4- expression was sensitive to matrix elasticity rather than to the RGDSP ligand concentration. Softer matrices (E ¼ 322 64.2 Pa), mimicking the elasticity of embryonic cardiac tissue, increased the fraction of cells expressing the early cardiac transcription factor Nkx2.5 as compared to cells in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa) decreased the expression of Nkx2.5 significantly. Further cardiac maturation was promoted by the stimulation of integrins relevant in early cardiac development (avb3 a5b1) by the RGDSP ligand in combination with the MMP-sensitivity of the matrix by a 6-fold increased number of myosin heavy chain (MHC)-positive cells as compared to cells in suspension. Expression of skeletal muscle marker MyoD was significantly higher in stiffer matrices (E ¼ 4036 419.6 Pa), resembling embryonic skeletal muscle tissue, while softer matrices (E ¼ 322 64.2 Pa) did not significantly support MyoD expression. The averages (in percentage) represent the mean values of positive stained samples performed in triplicates in two independent experiments. Differences between two data sets were considered statistically different when *p < 0.05, **p < 0.001.
the predominant cardiomyocyte b-integrin subunit, displayed defective myofibrillogenesis and disturbed electrical properties in embryonic stem cell-derived cardiomyocytes in vitro [37]. In vivo, loss of b1-integrin function resulted in retardation of cardiac differentiation as demonstrated by the delayed expression of a- and b-MHC compared to their wild-type counterparts [38,39]. Mice with deletion of the a5-integrin gene showed a defect in posterior mesoderm formation [40,41]. To examine the potential of the artificial extracellular matrix not only for guiding differentiation but also for influencing selfrenewal, we analyzed Oct-4 expression of the P19 EC cells at the gene and protein level. Our data demonstrates that Oct-4 expression depends on matrix elasticity and independent of changes in integrin ligand concentration (Figs. 6 and 7 and Table 2). To our knowledge, an influence of extracellular matrix mechanical characteristics on self-renewal has not yet been reported. Rather than matrix elasticity, it is known that a5b1-integrins plays a crucial role in keeping embryonic stem cell undifferentiated [42,43,44]. That is, a fibronectin-coated surface was shown to be essential to maintain self-renewal in human embryonic stem cells grown on a welldefined feeder layer and in animal serum-free conditions supplemented with growth factors. b1-integrin signaling is also required for the in vitro self-renewal of other stem cells such as epidermal stem cells [45,46]. The observed co-expression of Oct-4 and Nkx2.5 even after 14 d is in line with recent studies using mouse embryonic stem cells that show a distinct correlation between cardiac lineage commitment and dose-dependent Oct-4 expression [47]. Transgene- or TGFb-induced increase in Oct-4 was shown to trigger the expression of mesodermal and cardiac-specific genes through Smad2/4 pathway in the undifferentiated ES cells as well as in early differentiation [48].
To determine whether the size of the cell clusters affects cardiac differentiation, we analyzed the cell cluster diameters in the modular PEG matrices. We show that cardiac differentiation occurs mainly in larger cell clusters grown in soft matrices (Fig. 5C), which clusters sometimes contain a necrotic area in the center, possibly caused by insufficient oxygen delivery which creates a hypoxic center (Fig. 9A and B). The importance of cluster formation in cardiac and skeletal muscle differentiation in P19 EC cells is well known [49,50]. Cardiac muscle differentiation was, for example, demonstrated in large aggregates with a necrotic/hypoxic center area and a DMSO-induced epithelial surface coat identified as primitive endoderm. We then further analyzed cluster formation by adding a broad-range MMP-inhibitor to the culture medium to inhibit cell migration within the PEG gels. We previously demonstrated that cell migration in these dense PEG networks (mesh size z25 nm) is possible only through the activity of cell-secreted MMPs and not through preexisting pores as occurs, for example, in collagen gels [24]. We detected cluster formation after MMPinhibition as well, suggesting that cluster growth was based on cell proliferation rather than migration. Thus, the cell clusters may be derived by clonal expansion of the single cells seeded in the matrix. No spontaneous cardiac cell contraction was detected in the 3D PEG-based hydrogel, in contrast to other studies using 3D culture systems [8,9,10,11,12]. We speculate that even the lowest matrix elasticity of the PEG-matrix (E ¼ 322 64.2 Pa) is too stiff to enable spontaneous cell contraction [8,9,10]. However, for therapeutic applications it is still controversial if tissue engineered cardiac patches must contract already before implantation to enable electro-mechanical coupling in vivo. While studies with contracting neonatal cardiomyocytes in a collagen/Matrigel mixture claim
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2765
Fig. 8. Co-existence of cardiac and skeletal muscle markers (Nkx2.5, MyoD) with self-renewal marker Oct-4 of P19 EC cells in PEG-hydrogels after 14 d as shown by immunostaining and confocal imaging. All three transcription factors are mainly nuclearly localized. (A1) DAPI, (A2) Oct-4, (A3) Merge, (B1) DAPI, (B2) Nkx2.5, (B3) Merge, (C1) DAPI, (C2) MyoD, and (C3) Merge. Scale bars: 10 mm.
proper electro-mechanical coupling between the implanted cardiac graft and the infarcted rat heart without evidence of arrhythmias [8], others show that only in 50% of infarcted pig hearts regular and stable beating rhythm are observed upon injection of spontaneously contracting embryoid bodies from human embryonic stem cells [51]. 5. Conclusions We present here a method for engineering 3D cardiac tissue clusters using a precisely controlled artificial extracellular matrix
system that may serve as an alternative to natural biopolymer matrices such as collagen. We show that soft PEG-based matrices mimicking the elasticity of embryonic cardiac tissue can direct cardiac commitment of pluripotent P19 EC cells. However, further cardiac maturation was promoted by the MMP-sensitivity of the matrix, allowing cell-triggered matrix remodeling, supported by covalently bound adhesion peptides, but not by the matrix elasticity. Our cell-sensitive matrix allows dissection of this physiologically complex system into experimentally amenable culture conditions and thus may help to further understand how to drive stem cells toward specific fates.
Fig. 9. Immunohistochemical staining of cell clusters at 14 d with hematoxylin and eosin. The arrow marks a necrotic center area sometimes observed in larger clusters (diameter > 50 mm) but not in smaller ones (diameter < 50 mm). However, we did not detect cavities in the larger cell clusters at 14 d. Scale bar: 20 mm.
2766
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Acknowledgement We thank Dr. Mayumi Mochizuki and Conlin P. O’Neil for help with peptide synthesis, Dr. Mathieu Hauwel for support with RT-PCR, Miriella Pasquier for assistance in tissue embedding and processing, Olivier Brun for supporting with confocal imaging and Sergei Startchik for helping with confocal image processing. We also thank Prof. Ilona Skerjanc, University of Western Ontario, Canada, and Dr. Marcel Van der Heyden, University Medical Center Utrecht, The Netherlands, for helpful discussions. This work was supported by grants of the Novartis Foundation for Medicine and Biology and of the European Union’s 6th framework program Expertissues.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biomaterials.2008.03.016. References [1] American Heart Association. Heart disease and stroke statistics: 2007 update. Dallas: American Heart Association; 2007. [2] Orlic D, Kajstura J, Chimenti S, Jakoniuk I, Anderson SM, Li B, et al. Bone marrow cells regenerate infracted myocardium. Nature 2001;410:701–5. [3] Wollert KC, Meyer GP, Lotz J, Ringes-Lichtenberg S, Lippolt P, Breidenbach C, et al. Intracoronary autologous bone-marrow cell transfer after myocardial infarction: the BOOST randomised controlled clinical trial. Lancet 2004;364:141–8. [4] Murry CE, Soonpaa MH, Reinecke H, Nakajima H, Nakajima HO, Rubart M, et al. Haematopoietic stem cells do not transdifferentiate into cardiac myocytes in myocardial infarcts. Nature 2004;428:664–8. [5] Balsam LB, Wagers AJ, Christensen JL, Kofidis T, Weissmann IL, Robbins RC. Haematopoietic stem cells adopt mature haematopoietic fates in ischaemic myocardium. Nature 2004;428:668–73. [6] Muller-Ehmsen J, Whittaker P, Kloner RA, Dow JS, Sakoda T, Long TI, et al. Survival and development of neonatal rat cardiomyocytes transplanted into adult myocardium. J Mol Cell Cardiol 2002;34:107–16. [7] Jackson KA, Majka SM, Wang H, Pocius J, Hartley CJ, Majesky MW, et al. Regeneration of ischemic cardiac muscle and vascular endothelium by adult stem cells. J Clin Invest 2001;107:1395–402. [8] Zimmermann WH, Melnychenko I, Wasmeier G, Didie M, Naito H, Nixdorff U, et al. Engineered heart tissue grafts improve systolic and diastolic function in infarcted rat hearts. Nat Med 2006;12:452–8. [9] Akhyari P, Fedak PW, Weisel RD, Lee TY, Verma S, Mickle DA, et al. Mechanical stretch regimen enhances the formation of bioengineered autologous cardiac muscle grafts. Circulation 2002;106:I137–42. [10] Leor J, Aboulafia-Etzion S, Dar A, Shapiro L, Barbash IM, Battler A, et al. Bioengineered cardiac grafts: a new approach to repair the infarcted myocardium? Circulation 2000;102:III56–61. [11] Zund G, Breuer CK, Shinoka T, Ma PX, Langer R, Mayer JE, et al. The in vitro construction of a tissue engineered bioprosthetic heart valve. Eur J Cardiothorac Surg 1997;11:493–7. [12] Caspi O, Lesman A, Basevitch Y, Gepstein A, Arbel G, Habib IH, et al. Tissue engineering of vascularized cardiac muscle from human embryonic stem cells. Circulation 2007;100:263–72. [13] Lutolf MP, Weber FE, Schmoekel HG, Schense JC, Kohler T, Muller R, et al. Repair of bone defects using synthetic mimetics of collagenous extracellular matrices. Nat Biotechnol 2003;21:513–8. [14] Werb Z. ECM and cell surface proteolysis: regulating cellular ecology. Cell 1997;91:439–42. [15] Lutolf MP, Lauer-Fields JL, Schmoekel HG, Metters AT, Weber FE, Fields GB, et al. Synthetic matrix metalloproteinase-sensitive hydrogels for the conduction of tissue regeneration: engineering cell-invasion characteristics. Proc Natl Acad Sci U S A 2003;100:5413–8. [16] Webb CS, Bonnema DD, Ahmed SH, Leonardi AH, McClure CD, Clark LL, et al. Specific temporal profile of matrix metalloproteinase release occurs in patients after myocardial infarction: relation to left ventricular remodeling. Circulation 2006;114:1020–7. [17] Heymans S, Luttun A, Nuyens D, Theilmeier G, Creemers E, Moons L, et al. Inhibition of plasminogen activators or matrix metalloproteinases prevents cardiac rupture but impairs therapeutic. angiogenesis and causes cardiac failure. Nat Med 1999;5:1135–42. [18] Sepulveda JL, Gkretsi V, Wu C. Assembly and signaling of adhesion complexes. Curr Top Dev Biol 2005;68:183–225. [19] Ross RS, Borg TK. Integrins and the myocardium. Circ Res 2001;88:1112–9. [20] Maitra N, Flink IL, Bahl JJ, Morkin E. Expression of alpha and beta integrins during terminal differentiation of cardiomyocytes. Cardiovasc Res 2000;47: 715–25.
[21] Van der Heyden MA, Van Kempen MJ, Tsuji Y, Rook MB, Jongsma HJ, Opthof T. P19 embryonal carcinoma cells: a suitable model system for cardiac electrophysiological differentiation at the molecular and functional level. Cardiovasc Res 2003;58:410–22. [22] McBurney MW, Jones-Villeneuve EM, Edwards MK, Anderson PJ. Control of muscle and neuronal differentiation in a cultured embryonal carcinoma cell line. Nature 1982;299:165–7. [23] Lutolf MP, Hubbell JA. Synthesis and physicochemical characterization of endlinked poly(ethylene glycol)-co-peptide hydrogels formed by Michael-type addition. Biomacromolecules 2003;4:713–22. [24] Raeber GP, Lutolf MP, Hubbell JA. Molecularly engineered PEG hydrogels: a novel model system for proteolytically mediated cell migration. Biophys J 2005;89:1374–88. [25] Voytik-Harbin SL, Brightman AO, Waisner B, Lamar CH, Badylak SF. Application and evaluation of the alamarBlue assay for cell growth and survival of fibroblasts. In Vitro Cell Dev Biol Anim 1998;34:239–46. [26] Ahmed SA, Gogal Jr RM, Walsh JE. A new rapid and simple non-radioactive assay to monitor and determine the proliferation of lymphocytes: an alternative to [3H] thymidine incorporation assay. J Immunol Methods 1994; 170:211–24. [27] Jamali M, Rogerson PJ, Wilton S, Skerjanc IS. Nkx2-5 activity is essential for cardiomyogenesis. J Biol Chem 2001;276:42252–8. [28] Ridgeway AG, Wilton S, Skerjanc IS. Myocyte enhancer factor 2C and myogenin up-regulate each other’s expression and induce the development of skeletal muscle in P19 cells. J Biol Chem 2000;275:41–6. [29] Rudnicki MA, McBurney MW. Cell culture methods and induction of differentiation of embryonal carcinoma cell lines. In: Robertson EJ, editor. Teratocarcinomas and embryonic stem cells. A practical approach. Oxford: IRL Press; 1987. p. 19–49. [30] Gre´pin C, Nemer G, Nemer M. Enhanced cardiogenesis in embryonic stem cells overexpressing the GATA-4 transcription factor. Development 1997;124: 2387–95. [31] Engler AJ, Griffin MA, Sen S, Bonnemann CG, Sweeney HL, Discher DE. Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological implications for soft or stiff microenvironments. J Cell Biol 2004; 166:877–87. [32] Zammaretti P, Jaconi M. Cardiac tissue engineering: regeneration of the wounded heart. Curr Opin Biotechnol 2004;15:430–4. [33] Leor J, Amsalem Y, Cohen S. Cells, scaffolds, and molecules for myocardial tissue engineering. Pharmacol Ther 2005;105:151–63. [34] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell 2006;126:677–89. [35] Discher DE, Janmey P, Wang YL. Tissue cells feel and respond to the stiffness of their substrate. Science 2005;310:1139–43. [36] Yang JT, Bader BL, Kreidberg JA, Ullmann-Cullere M, Trevithick JE, Hynes RO. Overlapping and independent functions of fibronectin receptor integrins in early mesodermal development. Dev Biol 1999;215:264–77. [37] Fassler R, Rohwedel J, Maltsev V, Bloch W, Lentini S, Guan K, et al. Differentiation and integrity of cardiac muscle cells are impaired in the absence of beta 1 integrin. J Cell Sci 1996;109:2989–99. [38] Fassler R, Meyer M. Consequences of lack of beta 1 integrin gene expression in mice. Genes Dev 1995;9:1896–908. [39] Rohwedel J, Guan K, Zuschratter W, Jin S, Ahnert-Hilger G, Furst D, et al. Loss of beta1 integrin function results in a retardation of myogenic, but an acceleration of neuronal, differentiation of embryonic stem cells in vitro. Dev Biol 1998;201:167–84. [40] Yang JT, Rayburn H, Hynes RO. Embryonic mesodermal defects in alpha 5 integrin-deficient mice. Development 1993;119:1093–105. [41] Goh KL, Yang JT, Hynes RO. Mesodermal defects and cranial neural crest apoptosis in alpha5 integrin-null embryos. Development 1997;124:4309–19. [42] Amit M, Shariki C, Marqulets V, Itskovitz-Eldor J. Feeder layer- and serum-free culture of human embryonic stem cells. Biol Reprod 2004;70:837–45. [43] Amit M, Itskovitz-Eldor J. Feeder-free culture of human embryonic stem cells. Methods Enzymol 2006;420:37–49. [44] Prudhomme W, Daley GQ, Zandstra P, Lauffenburger DA. Multivariate proteomic analysis of murine embryonic stem cell self-renewal versus differentiation signaling. Proc Natl Acad Sci U S A 2004;101:2900–5. [45] Watt FM, Hogan BL. Out of Eden: stem cells and their niches. Science 2000; 287:1427–30. [46] Zhu AJ, Haase I, Watt FM. Signaling via beta1 integrins and mitogen-activated protein kinase determines human epidermal stem cell fate in vitro. Proc Natl Acad Sci U S A 1999;96:6728–33. [47] Niwa H, Myazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet 2000; 24:372–6. [48] Zeineddine D, Papadimou E, Chebli K, Gineste M, Liu J, Grey C, et al. Oct-3/4 dose dependently regulates specification of embryonic stem cells toward a cardiac lineage and early heart development. Dev Cell 2006;11:535–46. [49] Smith SC, Reuhl KR, Craig J, McBurney MW. The role of aggregation in embryonal carcinoma cell differentiation. J Cell Physiol 1987;131:74–84. [50] Skerjanc IS, Slack RS, McBurney MW. Cellular aggregation enhances MyoDdirected skeletal myogenesis in embryonal carcinoma cells. Mol Cell Biol 1994;14:8451–9. [51] Kehat I, Khimovich L, Caspi O, Gepstein A, Shofti R, Arbel G, et al. Electromechanical integration of cardiomyocytes derived from human embryonic stem cells. Nat Biotechnol 2004;22:1282–9.
Contents lists available at ScienceDirect
Biomaterials journal homepage: www.elsevier.com/locate/biomaterials
Three-dimensional extracellular matrix-directed cardioprogenitor differentiation: Systematic modulation of a synthetic cell-responsive PEG-hydrogel Thomas P. Kraehenbuehl a, Prisca Zammaretti a, Andre´ J. Van der Vlies a, Ronald G. Schoenmakers a, Matthias P. Lutolf b, Marisa E. Jaconi c, Jeffrey A. Hubbell a, * a b c
Institute of Bioengineering and Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fe´de´rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland Institute of Bioengineering, Ecole Polytechnique Fe´de´rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland Department of Pathology and Immunology, Faculty of Medicine, University of Geneva, 1 Rue Michel-Servet, 1211 Geneva 4, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 January 2008 Accepted 16 March 2008 Available online 7 April 2008
We show that synthetic three-dimensional (3D) matrix metalloproteinase (MMP)-sensitive poly(ethylene glycol) (PEG)-based hydrogels can direct differentiation of pluripotent cardioprogenitors, using P19 embryonal carcinoma (EC) cells as a model, along a cardiac lineage in vitro. In order to systematically probe 3D matrix effects on P19 EC differentiation, matrix elasticity, MMP-sensitivity and the concentration of a matrix-bound RGDSP peptide were modulated. Soft matrices (E ¼ 322 64.2 Pa, stoichiometric ratio: 0.8), mimicking the elasticity of embryonic cardiac tissue, increased the fraction of cells expressing the early cardiac transcription factor Nkx2.5 around 2-fold compared to embryoid bodies (EB) in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa, stoichiometric ratio: 1.2) decreased the number of Nkx2.5-positive cells significantly. Further indicators of cardiac maturation were promoted by ligation of integrins relevant in early cardiac development (a5b1, avb3) by the RGDSP ligand in combination with the MMP-sensitivity of the matrix, with a 6-fold increased amount of myosin heavy chain (MHC)-positive cells as compared to EB in suspension. This precisely controlled 3D culture system thus may serve as a potential alternative to natural matrices for engineering cardiac tissue structures for cell culture and potentially therapeutic applications. Ó 2008 Elsevier Ltd. All rights reserved.
Keywords: Extracellular matrix Biomimetic hydrogel Poly(ethylene glycol) Cardiac tissue engineering P19 embryonal carcinoma cells Stem cells
1. Introduction Nearly eight million persons in the United States suffer from myocardial infarction, with an estimated 800,000 new incidences every year [1]. Current therapies are limited by the restricted intrinsic regeneration capacity of the myocardium and by the lack of organs for transplantation. In recent attempts to regenerate functional heart muscle, adult cells such as bone marrow-derived cells have been injected into the ventricular wall or coronary vessels in animal models with myocardial infarction, as well as in human patients in first clinical trials [2,3]. Although improved cardiac performance has been reported, it remains unclear if paracrine signaling effects of the implanted cells, endogenous stem cell recruitment or even trans-differentiation are responsible for the improvement of cardiac function [4,5]. Furthermore, efficacy of the cell engraftment is very low: less than 10% of the injected cells typically engraft, mainly due to cell death, and less than 2% appear * Corresponding author. Tel.: þ41 21 693 96 81; fax: þ41 21 693 96 85. E-mail address: jeffrey.hubbell@epfl.ch (J.A. Hubbell). 0142-9612/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2008.03.016
to take on the in vivo fate of cardiomyocytes [6,7]. To better retain the cells at the site of infarction and to better control growth and differentiation, cardiac grafts have been engineered in vitro employing biodegradable materials as cell carriers or as cell ingrowth matrices. Both naturally derived [8,9,10] and synthetic materials [11,12] have been demonstrated to support embryonic and adult cell-derived cardiac tissue development containing contracting cardiomyocytes. However, mechanisms controlling 3D cardiac development are still poorly understood. We have recently developed a 3D PEG-based synthetic hydrogel material mimicking key biochemical characteristics of natural collagenous matrices [13], the major constituent of cardiac extracellular matrices [14]. This artificial extracellular matrix system allows the nearly independent control of matrix elasticity, integrinstimulating ligands and protease-sensitivity and is thus a potentially powerful tool to direct 3D cardiac development. We have previously reported that matrix metalloproteinase (MMP)-sensitive peptides crosslinking branched PEG chains enable cell-mediated proteolytic matrix degradation and remodeling [15]. These characteristics are also important in the heart, both under physiological
2758
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
and pathological conditions, where increased MMP-expression and activation has been observed [16,17]. Furthermore, matrix-bound RGDSP peptide promotes cell adhesion and stimulation of integrins relevant in early cardiac development (a5b1, aVb3) [18,19,20]. Here, we examined the ability of these cell-responsive PEGbased hydrogels for directing cardiac differentiation of pluripotent P19 embryonal carcinoma cells, a well-accepted model system for embryonic stem cell-based cardiac differentiation [21,22]. We systematically modulated matrix elasticity, the concentration of the matrix-bound RGDSP oligopeptide and MMP-sensitivity to dissect biophysical and biochemical matrix parameters potentially involved in cardiac differentiation. 2. Materials and methods 2.1. Synthesis of poly(ethylene glycol)–vinylsulfone and peptide precursors (RGDSP, MMP-substrate) PEG-vinylsulfone was synthesized adapting a previous protocol [23]. Branched 4-arm PEG-OH (Mw ¼ 20,000 g/mol) (Shearwater Polymers, Huntsville, AL) was dried by azeotropic distillation in toluene (VWR, Nyon, Switzerland) for 4 h using a Dean–Stark trap. Toluene was distilled off and the residue dissolved in dichloromethane (Fisher Scientific, Wohlen, Switzerland). To the clear solution, sodium hydride (Sigma–Aldrich, Buchs, Switzerland) was added at 20-fold molar excess over OH-groups. After hydrogen evolution ceased, divinylsulfone (Fluka, Buchs, Switzerland) was added at a 50-fold molar excess over OH-groups. The reaction was carried out at room temperature under argon with constant stirring for 3 d. After the addition of acetic acid (Fluka, Buchs, Switzerland) to neutralize excess sodium hydride, the mixture was filtered over a filter cell cake and concentrated by rotary evaporation. The polymer was then isolated by precipitation in ice-cold diethylether (Brunschwig, Basel, Switzerland) and filtered. Finally, the product was dried under vacuum to give a white solid (yield: 85%). The degree of PEG functionalization with vinylsulfone was determined by proton NMR spectroscopy (in CDCl3) using a Bruker 400 spectrometer (Bruker BioSpin, Faellanden, Switzerland). Characteristic vinylsulfone peaks were observed at 6.1, 6.4, and 6.8 ppm. The degree of end group conversion was found to be z95%. The polymer was stored under argon at 20 C until used. RGDSP and the substrates for matrix metalloproteinases (MMP) were synthesized by solid phase peptide synthesis using NovaSyn TGR resin (Merck Biosciences, Laeufelfingen, Switzerland) with an automated peptide synthesizer (Chemspeed, Augst, Switzerland) with standard Fmoc chemistry. Purification was performed by mass-directed reverse phase-C18 HPLC using a Waters Autopurification System. Separation and collection were performed by UV with broad wavelength detection (210–400 nm) (Waters PDA 996 UV photodiode array) and mass directed software (Waters Masslynx software). Peptide sequences were confirmed by ion trap ESI mass spectrometry (all Waters, Baden-Daettwil, Switzerland).
2.3. Cell culture and differentiation within PEG-based hydrogels P19 embryonic carcinoma (EC) cells (LGC Promochem, Molsheim, France) were grown in monolayer cultures in alpha-minimum essential medium (Invitrogen, Basel, Switzerland) supplemented with 7.5% bovine calf serum and 2.5% fetal bovine serum (LGC Promochem, Molsheim, France) (Fig. 2A). For differentiation in suspension (control cells), embryoid body (EB) formation and differentiation were induced by dissociating the undifferentiated P19 EC cells colonies with trypsin and resuspending them in differentiation medium containing 0.8% dimethylsulfoxide (DMSO) (Sigma–Aldrich, Buchs, Switzerland) in non-treated Petri dishes for 4 d. The EB were then transferred into tissue culture-treated Petri dishes containing medium without DMSO. For differentiation in the PEG-based hydrogels, undifferentiated single cells were mixed into the gel precursor solution at a concentration of 100,000 or 500,000 cells/ml gel. After the crosslinking reaction at 37 C, the cell-seeded gels of 25 mL volume (before swelling) were cultured in 24-well plates with 1.5 ml of differentiation medium for 4 d followed by culture in medium without DMSO. Medium was exchanged every third day. In some experiments, broad-range MMP-inhibitor GM6001 (Chemicon International, Temecula, CA) was added to the medium at 25 mM concentration 12 h after seeding the cells into the PEG-based hydrogels. The duration of the experiments was 14 d in general, and exceptionally 28 d for qualitatively analyzing cell behavior in the 3D PEG-hydrogels.
2.4. Cell proliferation Cell proliferation was measured with Alamar Blue (Lucerna-Chem, Lucerne, Switzerland). This non-toxic assay allowed the repetitive monitoring of the same cell samples [25,26]. Gels were overlaid with fresh medium containing 4% Alamar Blue solution and incubated for 6 h at 37 C. The supernatant was then collected and analyzed by measuring the fluorescence emission at 605 nm (excitation wavelength: 575 nm) (Safire, Tecan, Maennedorf, Switzerland).
2.5. FACS analysis For quantitative analysis by flow cytometry, gels were digested at 14 d after seeding, and single cells were extracted as described above. The cells were washed with ice-cold washing buffer (HBSS containing 0.1% w/v BSA) and fixed with 1% formaldehyde for 15 min. Cells were then incubated with the primary antibodies at 4 C overnight in washing buffer containing 0.3% (w/v) Saponin (Axon Lab, Le Montsur-Lausanne, Switzerland). Primary antibodies were monoclonal mouse antimyosin heavy chain (Abcam, Cambridge, UK, 1:100), polyclonal rabbit anti-Nkx2.5 (1:100), monoclonal mouse anti-MyoD (1:100) and monoclonal mouse anti-Oct-4 (all from Santa Cruz, Heidelberg, Germany, 1:100). After washing, the cells were incubated with the secondary antibodies goat anti-rabbit IgG (Alexa Fluor 488, 1:400) and goat anti-mouse IgG1 and IgG2b (Alexa Fluor 647, 1:400, both Invitrogen, Basel, Switzerland) at 4 C for 30 min. Non-conjugated isotype controls were used as negative controls (IgG and IgG2b from Santa Cruz, IgG1 from Chemicon International, Temecula, CA). The samples were analyzed by flow cytometry using a CyAn ADP (Dako, Baar, Switzerland). Between 40,000 and 100,000 cell events were counted per sample. Processing of the data was performed using FlowJo software (TreeStar, Ashland, OR).
2.2. Formation and characterization of PEG-based hydrogels Gel formation was done under physiological conditions as described elsewhere [23]. Briefly, the synthesis was carried out through Michael-type addition reaction of thiol-containing peptides onto vinylsulfone-functionalized PEG. Four-arm PEG-vinylsulfone was dissolved in triethanolamine buffer (0.3 M, pH 8.0) (Fluka, Buchs, Switzerland) to give a 10% (w/v) solution. A solution of the integrin ligand peptide (Ac-GCGYGRGDSPG-NH2) in the same buffer was added to the 4-arm PEG-vinylsulfone solution. After 10 min, a suspension of cells and MMP-sensitive peptide (Ac-GCRDGPQGIWGQDRCG-NH2) in the triethanolamine buffer was added. Although gel formation occurred within minutes, the crosslinking reaction was continued for around 30 min. Gels were characterized by their storage and loss modulus in small-strain oscillatory shear using a Bohlin CVO 120 high resolution rheometer (Malvern Instruments, Worcestershire, UK) with plate–plate geometry at 25 C under a humidified atmosphere. Storage and loss modulus as well as phase angle were measured as a function of frequency (from 0.05 to 5 Hz) in a constant strain mode (0.05). Young’s modulus (E ) was then calculated from the elastic shear modulus G0 using the formula E ¼ 3G0 . The swelling ratio of the gels was determined by the weight ratio of the gel before and after swelling in water overnight. Mesh size was determined as previously described [24]. The stoichiometric ratio was defined as the ratio between PEG-arms and crosslinker. To enable cell analysis at the gene level by RT-PCR and at the protein level through flow cytometry, the cells were extracted from the matrix after controlled degradation using collagenase IV. Gel digestion was carried out on ice for 1 h with 2 ml collagenase IV (Invitrogen, Basel, Switzerland) of 0.5 mg/ml in Hanks balance salt solution (Invitrogen, Basel, Switzerland) containing 0.1% bovine serum albumin (Sigma-Aldrich, Buchs, Switzerland) (¼washing buffer). The remaining gravimetric weight was measured (Supplementary Fig. 1).
2.6. Immunostaining and image analysis with confocal microscopy For confocal imaging, the gels were fixed in 2% paraformaldehyde for 30 min. After washing, the cells were permeabilized with a PBS solution containing 0.25% Triton-X (Sigma–Aldrich, Buchs, Switzerland) for 45 min at room temperature. The samples were then incubated with the primary antibodies at 4 C overnight in washing buffer. After three washes for 1 h each, cells were incubated with the secondary antibodies at 4 C overnight. The samples were washed 3 for 1 h. Cell nuclei were stained with DAPI (Invitrogen, Basel, Switzerland) at 300 nM concentration for 3 min. After mounting the samples with Vectashield (Reactolab, Servion, Switzerland), cells were stored at 4 C for imaging analysis. All washing and incubation steps were performed on a XY-shaker. For imaging, LSM510 Meta confocal microscope (Carl Zeiss, Feldbach, Switzerland) was used. Central layers through the cell clusters were analyzed using Imaris software.
2.7. Immunohistochemistry For immunohistochemistry, gels were fixed for 30 min with 2% paraformaldehyde 14 d after seeding, and embedded in Optimum Cutting Temperature (O.C.T.) formulation (Tissue Tek, Electron Microscopy Sciences, Hatfield, PA). The gels were frozen in isopentane (Acros Organics, Geel, Belgium) and sectioned with a cryostat (Microm Microtech, Francheville, France) at 18 C. Transverse sections (10 mm thickness) were placed on microscope slides and routinely processed for hematoxylin and eosin (Medite Medizintechnik, Nunningen, Switzerland) staining. Samples were then analyzed with a Axiovert 200M microscope (Carl Zeiss, Feldbach, Switzerland).
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766 2.8. RT-PCR analysis For isolating RNA from the cells, gels were digested as described above. Total RNA was isolated at 14 d after seeding with Trizol Reagent (Invitrogen, Basel, Switzerland) following the supplier’s protocol. Total RNA was quantified by a UV spectrophotometer (WPA, Cambridge, UK). Reverse Transcription was performed on 1 mg total RNA with equal volumes of oligo-dT primers and random hexamers, dNTPs (all from Promega, Wallisellen, Switzerland) and Superscript II Reverse Transcriptase (Invitrogen, Basel, Switzerland). PCR was carried out using Taq polymerase (Qiagen, Hombrechtikon, Switzerland). The amplified products were separated with 2% agarose gels containing ethidium bromide (both from Invitrogen, Basel, Switzerland). Primers’ sequences and reaction conditions are given in Supplementary Table 1. 2.9. Statistical analysis The results of all cell experiments are mean values (SD) of samples performed in triplicates in two independent experiments. Comparative analysis was performed with two-tailed Student’s t-test using S-PLUS software. Differences between two data sets were considered statistically different when *p < 0.05, **p < 0.001.
3. Results 3.1. Physicochemical characteristics of PEG-based hydrogels To modulate the biophysical matrix characteristics in a range possibly relevant for cardiac differentiation, the stoichiometric ratio between PEG and MMP-sensitive linkers was systematically altered (Fig. 1A). Cardiac cell differentiation was examined in a soft matrix (E ¼ 322 64.2 Pa) and in a stiffer one (E ¼ 4036 64.2 Pa) obtained at stoichiometric ratio of 0.8 and 1.2, respectively. The minimum value of the Young’s modulus (E ¼ 322 64.2 Pa) corresponds to the maximum value of the swelling ratio (Q ¼ 40 1.2) at a stoichiometric ratio of 0.8 (Fig. 1B). It is not possible to decouple matrix elasticity from the number of MMP-sensitive crosslinkers to assign cell behavior to one of these two matrix parameters. 3.2. Cell cluster formation in PEG-based hydrogels To reveal the potential of the cell-responsive hydrogel system for directing embryonal cardiac development, undifferentiated single cells were seeded into PEG-based hydrogels (Fig. 2B). After
2759
2 d, newly formed cell clusters appeared by cell proliferation, homogenously distributed as shown by phase contrast microscopy (Fig. 2C). Over time, the clusters grew concentrically, resulting in spherical shapes at 14 d with densely packed cells (Fig. 2D). The control EB produced and maintained in suspension were less compact than their counterparts in the gels. After 14 d in the 3D PEG-matrix, some cell clusters (<10%) started to converge by flattening the prospective contact area until cell–cell contact was established between neighboring cell clusters (Fig. 3A and B). In MMP-sensitive hydrogels containing the integrin ligand RGDSP at 100 mM, cells started to migrate out of the cell clusters after around 3 wk keeping their round-shaped morphology (Fig. 3C). Around 4 wk after seeding, we observed spindle-shaped cells radially growing out of the clusters forming well-connected networks (Fig. 3D). We next quantified cluster size from light microscopy images. Cluster size was affected by matrix elasticity, MMP-sensitivity and RGDSP concentration. Cells grown in stiffer matrices (E ¼ 4036 419.6 Pa) were smaller and more homogenous in size than clusters developed in soft matrices (E ¼ 322 64.2 Pa) (Fig. 4A). When inhibiting cell-mediated proteolytic degradation using a broadrange MMP-inhibitor (GM6001), the mean cluster diameter was significantly reduced compared to clusters grown without MMP-inhibition. Increasing the RGDSP concentration resulted in larger cell clusters compared to matrices without the integrin ligand (Fig. 4B). Absence of the differentiation-initiating DMSO in the culture medium affected neither cluster number nor size (data not shown). Notably, we did not detect spontaneously contracting cells within the gels. However, when we extracted cell clusters from the matrix at 14 d by collagenase treatment and plated the intact cell clusters in tissue culture-treated Petri dishes, we detected spontaneous contractions beginning at 3 d after seeding at a concentration of around 10 cell clusters per 10 cm Petri dish (Supplementary Fig. 2 and Supplementary movie). 3.3. Quantification of cell proliferation To assess 3D cell growth in the PEG matrices, cell proliferation was quantified within the sample construct every other day over a 14-d period. Gels with high cell seeding density (500,000 cells/ml gel) displayed a 2-fold faster proliferation rate compared to gels with lower cell seeding density (100,000 cells/ml gel) (Fig. 5A). The concentration of the RGDSP ligand significantly affected cell proliferation only between day 6 and 8 (Fig. 5B). The effect of matrix elasticity (E ¼ 322 64.2 Pa vs. E ¼ 4036 419.6 Pa) on cell proliferation was shown to be significant throughout the experiment (Fig. 5C). Addition of the MMP-inhibitor GM6001 decreased proliferation significantly (Fig. 5D). 3.4. Expression of cardiac markers on P19 EC cells in the PEG-matrix
Fig. 1. Young’s modulus and swelling properties of the cell-responsive PEG-based hydrogels. Young’s modulus (A) and swelling ratio (B) were modified by altering the stoichiometric ratio between PEG and MMP-sensitive crosslinkers. The maximum value of the Young’s modulus corresponds to the minimum value of the swelling ratio at r ¼ 1.2.
To analyze the influence of the extracellular matrix environment on gene and protein expression, the three matrix variables were systematically modulated as depicted in Table 1. Soft matrices (E ¼ 322 64.2 Pa) matching the elasticity of embryonic cardiac tissue [8,24], increased expression of the early cardiac transcription factor Nkx2.5 [27] by 2-fold compared to EB in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa) significantly decreased the number of Nkx2.5-positive cells as shown qualitatively at the gene (Fig. 6) and quantitatively at the protein level (Fig. 7 and Table 2). Addition of GM6001 did not significantly affect Nkx2.5 expression. For further cardiac maturation, the presence of RGDSP ligand to stimulate integrins involved in early cardiac development (a5b1, avb3) and the MMP-sensitivity of the matrix were shown to have a major effect as the expression of MHC, with an increase by 6-fold compared to EB in suspension. Addition of the broad-spectrum
2760
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Fig. 2. Culture and differentiation of P19 cells in 3D synthetic cell-responsive PEG-based hydrogels. (A) Undifferentiated P19 cells on tissue culture plastic. (B) Undifferentiated, single cells seeded into the PEG-based hydrogel (arrows). (C) Cell cluster formation after 2 d in the matrix in differentiation medium supplemented with 0.8% DMSO. (D) Cell cluster at 14 d after seeding into the matrix. Scale bar: 20 mm.
MMP-inhibitor resulted in a 50% decrease in MHC expression. Indeed, our MHC antibody detects both, cardiac MHC and skeletal muscle MHC, but this latter to at least five time lower specificity than the antibody binds to cardiac MHC according to the manufacturer. In addition, others [28] have shown that P19 EC cells aggregated in the
presence of DMSO and cultured for 10 d display only small amounts of skeletal muscle (<5%) assessed by the expression of embryonic MHC and the slow isoform of MHC, whereas cardiac muscle with MHC expression was detected even 3 d earlier and at a three time higher extent [29,30].
Fig. 3. Cell dynamics inside the PEG-based hydrogel. (A) Some cell clusters converge after 14 d by flattening the prospective contact area. (B) Neighboring cell clusters established cell–cell contacts after 14 d. (C) Around 3 wk after seeding undifferentiated single cells in hydrogels containing the RGDSP ligand (100 mM), the cells start to migrate out of the cell clusters. (D) Around 4 wk, some cells leave the cell clusters radially, showing spindle-shaped cell morphology. Scale bar: 20 mm. Scale bar insert image D: 10 mm.
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2761
Fig. 4. Size distribution of cell clusters as a function of the artificial extracellular matrix microenvironment. For each condition, the diameters of 100 randomly selected cell clusters per gel were analyzed at 14 d. (A) Different Young’s moduli: -, E ¼ 322 Pa; , E ¼ 4036 Pa; ,, E ¼ 4036 Pa, GM6001. (B) Different RGDSP concentrations: -, E ¼ 322 Pa; , E ¼ 4036 Pa; ,, E ¼ 4036 Pa, GM6001.
3.5. Self-renewal capacity and skeletal muscle differentiation of the P19 EC cells in the PEG-microenvironment To examine self-renewal capacity and other muscle differentiation of the PEG-matrix, the matrix environment was modulated as described above. The number of undifferentiated cells was determined on the gene (Fig. 6) and protein level by Oct-4 expression (Fig. 7 and Table 2), a marker for pluripotent embryonic stem cells typically down-regulated quickly upon differentiation. Cells in all PEG-matrix environments evaluated still expressed substantial amounts of Oct-4 compared to the control cells in suspension. However, quantitative differences were observed: whereas soft matrices (E ¼ 322 64.2 Pa) contained around 15% Oct-4-positive cells, stiffer matrices retained around 50% of the cells in an undifferentiated state (E ¼ 4036 419.6 Pa). MMP-inhibition did not significantly alter Oct-4 expression. Expression of skeletal muscle marker MyoD was significantly higher in stiffer matrices (E ¼ 4036 419.6 Pa), resembling embryonic skeletal muscle tissue [31], while soft matrices (E ¼ 322 64.2 Pa) with the same RGDSP concentration did not significantly promote MyoD expression (Fig. 7 and Table 2). Inhibition of MMP-activity did not significantly alter MyoD expression. The expression and localization of the self-renewal marker Oct-4, the cardiac marker Nkx2.5 and the skeletal muscle marker
MyoD were also assessed at 14 d by immunofluorescence staining and confocal imaging (Fig. 8). All three transcription factors were mainly located within the cell nuclei. With immunohistology of the 3D PEG-hydrogels with the embedded P19 EC cell clusters, we show a necrotic central area (arrow) in larger cell clusters, but not in smaller ones (Fig. 9). 4. Discussion Here, we explore the utility of a material system [13] to support and direct differentiation of a stem cell model along a targeted lineage, employing the pluripotent P19 EC model and targeting cardiomyocytes. Our data indicates the necessity to control several biophysical and biochemical features of the extracellular matrix to drive P19 EC cells, and presumably stem cells as cardioprogenitors, along the cardiac differentiation pathway. Compared to common matrices used in cardiac tissue engineering [32,33] our artificial extracellular matrix microenvironment allowed modulation of elasticity, adhesivity and proteolytic susceptibility nearly independently, and thus permitted dissection of the physiological complexity into experimentally amenable culture conditions. We show that initial cardiac muscle commitment was enhanced by the matrix elasticity. However, further maturation appeared to
2762
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Fig. 5. Cell proliferation in the 3D hydrogels measured with Alamar Blue. Cell proliferation was shown to be directed by the artificial extracellular matrix microenvironment and the cell density. (A) High cell density demonstrated significantly higher proliferation than low cell density: -, 100,000 cells/ml gel; , 500,000 cells/ml gel. (B) Presence of cell adhesion ligand (RGDSP): -, 0 mM; , 100 mM (C) Highest proliferation was demonstrated by matrices with low Young’s modulus: -, E ¼ 4036 Pa; , E ¼ 322 Pa. (D) Broad inhibition of cell-secreted MMP by GM6001 decreased proliferation: -, 0 mM; , 100 mM.
be influenced not by the matrix elasticity but also by the stimulation of integrins relevant in early cardiac development (a5b1, avb3) via the RGDSP ligand in combination with the MMP-sensitivity of the matrix. This limited effect of matrix stiffness in directing terminal differentiation was recently shown in experiments with human mesenchymal stem cells (hMSC) [34,35]. When seeded on collagen-coated polyacrylamide gels mimicking the elasticity of the respective natural microenvironment, hMSC were selectively directed toward myogenic, neurogenic and osteogenic programs, however, with limited expression of terminal differentiation. However, neither in our 3D model system nor in the abovementioned collagen-coated polyacrylamide gel, it is possible to assign the cell differentiation fully to either matrix elasticity or density of the mesh/fibers, respectively. However, we have shown the relation between gel elasticity and the number of cleavable crosslinker peptides (stoichiometric ratio) [23]. A stoichiometric ratio of less than 1 may lead to defects in the PEG-network due to the incomplete reaction and thus on the mechanical level decrease the Young’s modulus of the matrix, and on the cellular level provide additional room for migration or proliferation. Cell behavior
Table 1 Overview of the extracellular matrix conditions employed: matrix elasticity, concentration of RGDSP ligand, and the matrix sensitivity Condition
10 100 2 3 4 5
Description Age cells [d]
Young’s modulus [Pa]
RGDSP-concentration [mM]
Addition GM6001
0 14 14 14 14 14
0 (control, suspension) 0 (control, suspension) 322 64.2 322 64.2 4036 419.6 4036 419.6
0 (control, suspension) 0 (control, suspension) 0 100 100 100
No No No No No Yes
including differentiation may always be affected by the combined effects of matrix elasticity and number of MMP-sensitive crosslinkers without being able to completely decouple. In contrast to the recently discovered role of elasticity in regulation of stem cell fate, the importance of avb3 and a5b1- integrins during cardiac differentiation has been reported by studies in vitro and in vivo with knock-out models [36]. Deficiency of b1-integrin,
Fig. 6. Representative gene expression of P19 EC cells in the 3D PEG-based hydrogels at seeding (10 ) and 14 d (100 and 2–5) as assessed by RT-PCR. RNA was isolated by digestion of the matrix. For each sample, reverse transcription was performed with 1 mg RNA. Cells grown in suspension were used as a control. We show a fast downregulation of Oct-4 expression within 14 d of the control cells in suspension, but strong expression by the cells embedded in the 3D PEG-hydrogel. While GATA-4, as an endothelial marker and precursor of the cardiac pathway is expressed in all conditions at 14 d, Nkx2.5, which is considered as the first cardiac marker is expressed in the soft matrices (E ¼ 322 64.2 Pa) only, but is not supported by the stiffer matrix environment (E ¼ 4036 419.6 Pa). While MEF2C as a cardiac and skeletal transcription factor is expressed in all conditions at 14 d, beta MHC, which is know to be mainly a marker for mature cardiac cells is supported by the PEG-matrix.
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2763
Fig. 7. Representative protein expression of P19 EC cells in the 3D synthetic PEG-based hydrogels as assessed by FACS. Cells were extracted from the matrix at 14 d and subjected to flow cytometric analysis. Samples are shown in black, respective isotype controls in grey.
2764
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Table 2 Quantification of protein expression from cells extracted from the PEG-based hydrogel at 14 d as measured by flow cytometry Protein
Matrix parameters
Average
S.D.
p-Value
Significance
Oct-4
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
91.2 3.5 14.5 14.8 58.8 44.3
4.62 2.88 4.59 5.49 9.56 11.38
2.6E-12 Ref. 0.0006 0.0012 9.1E-08 6.7E-06
** Ref. ** ** ** **
Nkx2.5
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 13.3 18.2 21.0 9.5 9.3
N.A 2.80 5.78 5.25 2.59 4.13
N.A Ref. 0.0950 0.0103 0.0337 0.0781
N.A Ref. – * * –
MyoD
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 12.2 12.0 14.0 19.2 16.8
N.A 3.60 5.06 4.38 2.99 1.72
N.A Ref. 0.9489 0.4469 0.0044 0.0168
N.A Ref. – – * *
MHC
Suspension/0 d Suspension/14 d E ¼ 322 64.2 Pa, CRGDSP ¼ 0 mM E ¼ 322 64.2 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM E ¼ 4036 419.6 Pa, CRGDSP ¼ 100 mM, GM6001
N.A 6.2 11.3 37.2 31.5 15.0
N.A 1.94 2.58 8.86 5.01 3.41
N.A Ref. 0.0029 0.0000 0.0000 0.0003
N.A Ref. * ** ** *
S.D.: standard deviation, N.A.: not applicable, Ref.: reference for p-value. Self-renewal capacity as evident from Oct-4- expression was sensitive to matrix elasticity rather than to the RGDSP ligand concentration. Softer matrices (E ¼ 322 64.2 Pa), mimicking the elasticity of embryonic cardiac tissue, increased the fraction of cells expressing the early cardiac transcription factor Nkx2.5 as compared to cells in suspension. In contrast, stiffer matrices (E ¼ 4036 419.6 Pa) decreased the expression of Nkx2.5 significantly. Further cardiac maturation was promoted by the stimulation of integrins relevant in early cardiac development (avb3 a5b1) by the RGDSP ligand in combination with the MMP-sensitivity of the matrix by a 6-fold increased number of myosin heavy chain (MHC)-positive cells as compared to cells in suspension. Expression of skeletal muscle marker MyoD was significantly higher in stiffer matrices (E ¼ 4036 419.6 Pa), resembling embryonic skeletal muscle tissue, while softer matrices (E ¼ 322 64.2 Pa) did not significantly support MyoD expression. The averages (in percentage) represent the mean values of positive stained samples performed in triplicates in two independent experiments. Differences between two data sets were considered statistically different when *p < 0.05, **p < 0.001.
the predominant cardiomyocyte b-integrin subunit, displayed defective myofibrillogenesis and disturbed electrical properties in embryonic stem cell-derived cardiomyocytes in vitro [37]. In vivo, loss of b1-integrin function resulted in retardation of cardiac differentiation as demonstrated by the delayed expression of a- and b-MHC compared to their wild-type counterparts [38,39]. Mice with deletion of the a5-integrin gene showed a defect in posterior mesoderm formation [40,41]. To examine the potential of the artificial extracellular matrix not only for guiding differentiation but also for influencing selfrenewal, we analyzed Oct-4 expression of the P19 EC cells at the gene and protein level. Our data demonstrates that Oct-4 expression depends on matrix elasticity and independent of changes in integrin ligand concentration (Figs. 6 and 7 and Table 2). To our knowledge, an influence of extracellular matrix mechanical characteristics on self-renewal has not yet been reported. Rather than matrix elasticity, it is known that a5b1-integrins plays a crucial role in keeping embryonic stem cell undifferentiated [42,43,44]. That is, a fibronectin-coated surface was shown to be essential to maintain self-renewal in human embryonic stem cells grown on a welldefined feeder layer and in animal serum-free conditions supplemented with growth factors. b1-integrin signaling is also required for the in vitro self-renewal of other stem cells such as epidermal stem cells [45,46]. The observed co-expression of Oct-4 and Nkx2.5 even after 14 d is in line with recent studies using mouse embryonic stem cells that show a distinct correlation between cardiac lineage commitment and dose-dependent Oct-4 expression [47]. Transgene- or TGFb-induced increase in Oct-4 was shown to trigger the expression of mesodermal and cardiac-specific genes through Smad2/4 pathway in the undifferentiated ES cells as well as in early differentiation [48].
To determine whether the size of the cell clusters affects cardiac differentiation, we analyzed the cell cluster diameters in the modular PEG matrices. We show that cardiac differentiation occurs mainly in larger cell clusters grown in soft matrices (Fig. 5C), which clusters sometimes contain a necrotic area in the center, possibly caused by insufficient oxygen delivery which creates a hypoxic center (Fig. 9A and B). The importance of cluster formation in cardiac and skeletal muscle differentiation in P19 EC cells is well known [49,50]. Cardiac muscle differentiation was, for example, demonstrated in large aggregates with a necrotic/hypoxic center area and a DMSO-induced epithelial surface coat identified as primitive endoderm. We then further analyzed cluster formation by adding a broad-range MMP-inhibitor to the culture medium to inhibit cell migration within the PEG gels. We previously demonstrated that cell migration in these dense PEG networks (mesh size z25 nm) is possible only through the activity of cell-secreted MMPs and not through preexisting pores as occurs, for example, in collagen gels [24]. We detected cluster formation after MMPinhibition as well, suggesting that cluster growth was based on cell proliferation rather than migration. Thus, the cell clusters may be derived by clonal expansion of the single cells seeded in the matrix. No spontaneous cardiac cell contraction was detected in the 3D PEG-based hydrogel, in contrast to other studies using 3D culture systems [8,9,10,11,12]. We speculate that even the lowest matrix elasticity of the PEG-matrix (E ¼ 322 64.2 Pa) is too stiff to enable spontaneous cell contraction [8,9,10]. However, for therapeutic applications it is still controversial if tissue engineered cardiac patches must contract already before implantation to enable electro-mechanical coupling in vivo. While studies with contracting neonatal cardiomyocytes in a collagen/Matrigel mixture claim
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
2765
Fig. 8. Co-existence of cardiac and skeletal muscle markers (Nkx2.5, MyoD) with self-renewal marker Oct-4 of P19 EC cells in PEG-hydrogels after 14 d as shown by immunostaining and confocal imaging. All three transcription factors are mainly nuclearly localized. (A1) DAPI, (A2) Oct-4, (A3) Merge, (B1) DAPI, (B2) Nkx2.5, (B3) Merge, (C1) DAPI, (C2) MyoD, and (C3) Merge. Scale bars: 10 mm.
proper electro-mechanical coupling between the implanted cardiac graft and the infarcted rat heart without evidence of arrhythmias [8], others show that only in 50% of infarcted pig hearts regular and stable beating rhythm are observed upon injection of spontaneously contracting embryoid bodies from human embryonic stem cells [51]. 5. Conclusions We present here a method for engineering 3D cardiac tissue clusters using a precisely controlled artificial extracellular matrix
system that may serve as an alternative to natural biopolymer matrices such as collagen. We show that soft PEG-based matrices mimicking the elasticity of embryonic cardiac tissue can direct cardiac commitment of pluripotent P19 EC cells. However, further cardiac maturation was promoted by the MMP-sensitivity of the matrix, allowing cell-triggered matrix remodeling, supported by covalently bound adhesion peptides, but not by the matrix elasticity. Our cell-sensitive matrix allows dissection of this physiologically complex system into experimentally amenable culture conditions and thus may help to further understand how to drive stem cells toward specific fates.
Fig. 9. Immunohistochemical staining of cell clusters at 14 d with hematoxylin and eosin. The arrow marks a necrotic center area sometimes observed in larger clusters (diameter > 50 mm) but not in smaller ones (diameter < 50 mm). However, we did not detect cavities in the larger cell clusters at 14 d. Scale bar: 20 mm.
2766
T.P. Kraehenbuehl et al. / Biomaterials 29 (2008) 2757–2766
Acknowledgement We thank Dr. Mayumi Mochizuki and Conlin P. O’Neil for help with peptide synthesis, Dr. Mathieu Hauwel for support with RT-PCR, Miriella Pasquier for assistance in tissue embedding and processing, Olivier Brun for supporting with confocal imaging and Sergei Startchik for helping with confocal image processing. We also thank Prof. Ilona Skerjanc, University of Western Ontario, Canada, and Dr. Marcel Van der Heyden, University Medical Center Utrecht, The Netherlands, for helpful discussions. This work was supported by grants of the Novartis Foundation for Medicine and Biology and of the European Union’s 6th framework program Expertissues.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biomaterials.2008.03.016. References [1] American Heart Association. Heart disease and stroke statistics: 2007 update. Dallas: American Heart Association; 2007. [2] Orlic D, Kajstura J, Chimenti S, Jakoniuk I, Anderson SM, Li B, et al. Bone marrow cells regenerate infracted myocardium. Nature 2001;410:701–5. [3] Wollert KC, Meyer GP, Lotz J, Ringes-Lichtenberg S, Lippolt P, Breidenbach C, et al. Intracoronary autologous bone-marrow cell transfer after myocardial infarction: the BOOST randomised controlled clinical trial. Lancet 2004;364:141–8. [4] Murry CE, Soonpaa MH, Reinecke H, Nakajima H, Nakajima HO, Rubart M, et al. Haematopoietic stem cells do not transdifferentiate into cardiac myocytes in myocardial infarcts. Nature 2004;428:664–8. [5] Balsam LB, Wagers AJ, Christensen JL, Kofidis T, Weissmann IL, Robbins RC. Haematopoietic stem cells adopt mature haematopoietic fates in ischaemic myocardium. Nature 2004;428:668–73. [6] Muller-Ehmsen J, Whittaker P, Kloner RA, Dow JS, Sakoda T, Long TI, et al. Survival and development of neonatal rat cardiomyocytes transplanted into adult myocardium. J Mol Cell Cardiol 2002;34:107–16. [7] Jackson KA, Majka SM, Wang H, Pocius J, Hartley CJ, Majesky MW, et al. Regeneration of ischemic cardiac muscle and vascular endothelium by adult stem cells. J Clin Invest 2001;107:1395–402. [8] Zimmermann WH, Melnychenko I, Wasmeier G, Didie M, Naito H, Nixdorff U, et al. Engineered heart tissue grafts improve systolic and diastolic function in infarcted rat hearts. Nat Med 2006;12:452–8. [9] Akhyari P, Fedak PW, Weisel RD, Lee TY, Verma S, Mickle DA, et al. Mechanical stretch regimen enhances the formation of bioengineered autologous cardiac muscle grafts. Circulation 2002;106:I137–42. [10] Leor J, Aboulafia-Etzion S, Dar A, Shapiro L, Barbash IM, Battler A, et al. Bioengineered cardiac grafts: a new approach to repair the infarcted myocardium? Circulation 2000;102:III56–61. [11] Zund G, Breuer CK, Shinoka T, Ma PX, Langer R, Mayer JE, et al. The in vitro construction of a tissue engineered bioprosthetic heart valve. Eur J Cardiothorac Surg 1997;11:493–7. [12] Caspi O, Lesman A, Basevitch Y, Gepstein A, Arbel G, Habib IH, et al. Tissue engineering of vascularized cardiac muscle from human embryonic stem cells. Circulation 2007;100:263–72. [13] Lutolf MP, Weber FE, Schmoekel HG, Schense JC, Kohler T, Muller R, et al. Repair of bone defects using synthetic mimetics of collagenous extracellular matrices. Nat Biotechnol 2003;21:513–8. [14] Werb Z. ECM and cell surface proteolysis: regulating cellular ecology. Cell 1997;91:439–42. [15] Lutolf MP, Lauer-Fields JL, Schmoekel HG, Metters AT, Weber FE, Fields GB, et al. Synthetic matrix metalloproteinase-sensitive hydrogels for the conduction of tissue regeneration: engineering cell-invasion characteristics. Proc Natl Acad Sci U S A 2003;100:5413–8. [16] Webb CS, Bonnema DD, Ahmed SH, Leonardi AH, McClure CD, Clark LL, et al. Specific temporal profile of matrix metalloproteinase release occurs in patients after myocardial infarction: relation to left ventricular remodeling. Circulation 2006;114:1020–7. [17] Heymans S, Luttun A, Nuyens D, Theilmeier G, Creemers E, Moons L, et al. Inhibition of plasminogen activators or matrix metalloproteinases prevents cardiac rupture but impairs therapeutic. angiogenesis and causes cardiac failure. Nat Med 1999;5:1135–42. [18] Sepulveda JL, Gkretsi V, Wu C. Assembly and signaling of adhesion complexes. Curr Top Dev Biol 2005;68:183–225. [19] Ross RS, Borg TK. Integrins and the myocardium. Circ Res 2001;88:1112–9. [20] Maitra N, Flink IL, Bahl JJ, Morkin E. Expression of alpha and beta integrins during terminal differentiation of cardiomyocytes. Cardiovasc Res 2000;47: 715–25.
[21] Van der Heyden MA, Van Kempen MJ, Tsuji Y, Rook MB, Jongsma HJ, Opthof T. P19 embryonal carcinoma cells: a suitable model system for cardiac electrophysiological differentiation at the molecular and functional level. Cardiovasc Res 2003;58:410–22. [22] McBurney MW, Jones-Villeneuve EM, Edwards MK, Anderson PJ. Control of muscle and neuronal differentiation in a cultured embryonal carcinoma cell line. Nature 1982;299:165–7. [23] Lutolf MP, Hubbell JA. Synthesis and physicochemical characterization of endlinked poly(ethylene glycol)-co-peptide hydrogels formed by Michael-type addition. Biomacromolecules 2003;4:713–22. [24] Raeber GP, Lutolf MP, Hubbell JA. Molecularly engineered PEG hydrogels: a novel model system for proteolytically mediated cell migration. Biophys J 2005;89:1374–88. [25] Voytik-Harbin SL, Brightman AO, Waisner B, Lamar CH, Badylak SF. Application and evaluation of the alamarBlue assay for cell growth and survival of fibroblasts. In Vitro Cell Dev Biol Anim 1998;34:239–46. [26] Ahmed SA, Gogal Jr RM, Walsh JE. A new rapid and simple non-radioactive assay to monitor and determine the proliferation of lymphocytes: an alternative to [3H] thymidine incorporation assay. J Immunol Methods 1994; 170:211–24. [27] Jamali M, Rogerson PJ, Wilton S, Skerjanc IS. Nkx2-5 activity is essential for cardiomyogenesis. J Biol Chem 2001;276:42252–8. [28] Ridgeway AG, Wilton S, Skerjanc IS. Myocyte enhancer factor 2C and myogenin up-regulate each other’s expression and induce the development of skeletal muscle in P19 cells. J Biol Chem 2000;275:41–6. [29] Rudnicki MA, McBurney MW. Cell culture methods and induction of differentiation of embryonal carcinoma cell lines. In: Robertson EJ, editor. Teratocarcinomas and embryonic stem cells. A practical approach. Oxford: IRL Press; 1987. p. 19–49. [30] Gre´pin C, Nemer G, Nemer M. Enhanced cardiogenesis in embryonic stem cells overexpressing the GATA-4 transcription factor. Development 1997;124: 2387–95. [31] Engler AJ, Griffin MA, Sen S, Bonnemann CG, Sweeney HL, Discher DE. Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological implications for soft or stiff microenvironments. J Cell Biol 2004; 166:877–87. [32] Zammaretti P, Jaconi M. Cardiac tissue engineering: regeneration of the wounded heart. Curr Opin Biotechnol 2004;15:430–4. [33] Leor J, Amsalem Y, Cohen S. Cells, scaffolds, and molecules for myocardial tissue engineering. Pharmacol Ther 2005;105:151–63. [34] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell 2006;126:677–89. [35] Discher DE, Janmey P, Wang YL. Tissue cells feel and respond to the stiffness of their substrate. Science 2005;310:1139–43. [36] Yang JT, Bader BL, Kreidberg JA, Ullmann-Cullere M, Trevithick JE, Hynes RO. Overlapping and independent functions of fibronectin receptor integrins in early mesodermal development. Dev Biol 1999;215:264–77. [37] Fassler R, Rohwedel J, Maltsev V, Bloch W, Lentini S, Guan K, et al. Differentiation and integrity of cardiac muscle cells are impaired in the absence of beta 1 integrin. J Cell Sci 1996;109:2989–99. [38] Fassler R, Meyer M. Consequences of lack of beta 1 integrin gene expression in mice. Genes Dev 1995;9:1896–908. [39] Rohwedel J, Guan K, Zuschratter W, Jin S, Ahnert-Hilger G, Furst D, et al. Loss of beta1 integrin function results in a retardation of myogenic, but an acceleration of neuronal, differentiation of embryonic stem cells in vitro. Dev Biol 1998;201:167–84. [40] Yang JT, Rayburn H, Hynes RO. Embryonic mesodermal defects in alpha 5 integrin-deficient mice. Development 1993;119:1093–105. [41] Goh KL, Yang JT, Hynes RO. Mesodermal defects and cranial neural crest apoptosis in alpha5 integrin-null embryos. Development 1997;124:4309–19. [42] Amit M, Shariki C, Marqulets V, Itskovitz-Eldor J. Feeder layer- and serum-free culture of human embryonic stem cells. Biol Reprod 2004;70:837–45. [43] Amit M, Itskovitz-Eldor J. Feeder-free culture of human embryonic stem cells. Methods Enzymol 2006;420:37–49. [44] Prudhomme W, Daley GQ, Zandstra P, Lauffenburger DA. Multivariate proteomic analysis of murine embryonic stem cell self-renewal versus differentiation signaling. Proc Natl Acad Sci U S A 2004;101:2900–5. [45] Watt FM, Hogan BL. Out of Eden: stem cells and their niches. Science 2000; 287:1427–30. [46] Zhu AJ, Haase I, Watt FM. Signaling via beta1 integrins and mitogen-activated protein kinase determines human epidermal stem cell fate in vitro. Proc Natl Acad Sci U S A 1999;96:6728–33. [47] Niwa H, Myazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet 2000; 24:372–6. [48] Zeineddine D, Papadimou E, Chebli K, Gineste M, Liu J, Grey C, et al. Oct-3/4 dose dependently regulates specification of embryonic stem cells toward a cardiac lineage and early heart development. Dev Cell 2006;11:535–46. [49] Smith SC, Reuhl KR, Craig J, McBurney MW. The role of aggregation in embryonal carcinoma cell differentiation. J Cell Physiol 1987;131:74–84. [50] Skerjanc IS, Slack RS, McBurney MW. Cellular aggregation enhances MyoDdirected skeletal myogenesis in embryonal carcinoma cells. Mol Cell Biol 1994;14:8451–9. [51] Kehat I, Khimovich L, Caspi O, Gepstein A, Shofti R, Arbel G, et al. Electromechanical integration of cardiomyocytes derived from human embryonic stem cells. Nat Biotechnol 2004;22:1282–9.