Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Biochem Biophys Res Commun. Author manuscript; available in PMC 2010 April 17.
Published in final edited form as:
PMCID: PMC2703609

Adapting collagen/CNT matrix in directing hESC differentiation


The lineage selection in human embryonic stem cell (hESC) differentiation relies on both the growth factors and small molecules in the media and the physical characteristics of the micro-environment. In this work, we utilized various materials, including the collagen-carbon nanotube (collagen/CNT) composite material, as cell culture matrices to examine the impact of matrix properties on hESC differentiation. Our AFM analysis indicated that the collagen/CNT formed rigid fibril bundles, which polarized the growth and differentiation of hESCs, resulting in more than 90% of the cells to the ectodermal lineage in Day 3 in the media commonly used for spontaneous differentiation. We also observed the differentiated cells followed the coarse alignment of the collagen/CNT matrix. The research not only revealed the responsiveness of hESCs to matrix properties, but also provided a simple yet efficient way to direct the hESC differentiation, and imposed the potential of forming neural-cell based bio-devices for further applications.

Keywords: Human embryonic stem cell, collagen, CNT, differentiation, ectodermal lineage


Human embryonic stem cells (hESC) have the potential to remain undifferentiated in a tight colony format or differentiate into all three primary germ layers including endoderm, mesoderm and ectoderm cells that form all tissues in human body [,1]. Both the scenarios occur depending on the growth factors and small molecules available in the media. Due to the immense potential of hESCs in regenerative medicine for many hard-to-treat-diseases, tremendous amount of research has been dedicated to harness the pluripotency of hESCs mainly by fine-tuning the media composition. Such methods rely heavily on our understanding of traditional mammalian developmental biology. Nevertheless, recent studies have revealed that stem cells are extremely responsive to their immediate physical environment [2,3], the extracellular matrix (ECM). ECM regulates the cell fate by the ECM molecules in concert with its physical properties [49], which have been frequently reported to be critical in cell lineage specification. For instance, cell shape, controlled by matrix geometry, acted as a cue in the commitment of human mesenchymal stem cells to adipocyte or osteoblast [4]; murine neural progenitor cells aligned with fibrous scaffolds, and the differentiation with long neurofilament was more preferentially shown on the nanoscale scaffolds than on the microscale scaffolds [6]; mesenchymal stem cells were shown to specify lineage and commit to neurogenic, myogenic and osteogenic phenotypes in response to matrix elasticity [7]. These physical effects of ECM are imposed by affecting the cytoskeletal tension, altering the cell adhesion, and deforming the cell membrane. In conjunction with stimulants added in the media, they are now deemed to be the foremost players in directing lineage specification of stem cells [10,11].

Our previous studies on the spontaneous differentiation of hESCs used a high serum concentration with DMEM media in the presence of non-essential amino acids without additional growth factors [12]. When gelatin or matrigel were used as a matrix, the medium allowed rapid spontaneous differentiation into all three lineages with no specification. According to our atomic force microscopic (AFM) analysis, such matrices are structure-less and soft. To investigate the impact of structured matrix on hESC differentiation, we chose type I collagen and its composite with carbon nanotube (collagen/CNT) as the cell culture matrices in this work. With the same medium for spontaneous differentiation, we endeavored to monitor the effects of collagen and collagen/CNT on cell lineage specification at the early stage of hESC differentiation.

Type I collagen is the most abundant collagen in human body and is a component of tendons, skin and artery walls. Upon gelation, it forms a network of interconnected fibers. A single collagen fibril is composed of three monomers that twist to form a single left handed superhelix with a well-ordered periodic structure termed the D period. Type I collagen is an ECM protein, and is known to support the growth of many neuronal cell types [13] as well as the neural differentiation of hESCs in a neural differentiation medium [5]. In this study, we attempted the use of type I collagen as a substrate to direct the differentiation of hESCs in the medium of spontaneous differentiation. In order to justify the effects of physical factors, we further manipulated the substrate properties by preparing the collagen/CNT composite material. CNTs, a one-dimensional nano-material, are frequently used in biological studies due to their high tensile strength and electrical properties [14,15]. Properly functionalized CNTs are non-toxic to cells at a low concentration in the medium or in the matrix scaffolds [15,16]. We utilized the carboxyl modified single wall CNTs to make a composite material with type I collagen, and the CNTs were found to greatly impact the collagen’s stiffness and structure, resulting in the early differentiation of more than 90% of the cells into nestin-positive cells. Additionally, when we roughly aligned the collagen by dip-coating preparation, the differentiated cells lined up with the pre-aligned collagen and collagen/CNT fibril bundles. Besides the understanding of the high sensitivity of hESCs to matrix properties, the research also revealed a novel strategy of forming neural cell based bio-devices for further applications.

Materials and Methods

Cell matrices preparation

Collagen type I from rat tails was purchased from Sigma (St. Louis, MO). According to the manufacturer’s protocol, 1.1 mg/ml of collagen was prepared under sterile conditions in 10× PBS buffer (from Fisher, Fair Lawn, NJ) containing 1 N NaOH to achieve the pH of 7.4. Plastic slides were dip-coated with the collagen solution along one direction (to promote the alignment of fibrous scaffolds). The slides were then incubated at 37 °C for 1.5–2 hrs for gelation before being using for cell culture.

High purity single-walled carbon nanotubes (SWCNTs) were purchased from HELIX Material Solution (Richardson, TX). Raw SWCNTs were functionalized with carboxyl groups by sonicating SWCNTs in an aqueous solution containing 8 M of H2SO4 and 8 M of HNO3 at 70°C for 3 hours [1719]. After centrifugation at 5000 RPM for 10 minutes, the supernatant was extracted, and the CNTs were collected on a filter membrane. The CNTs were further washed by DI-H2O until the pH was around 6~7, then dried in an oven at 60 °C for 3 hours to obtain the powder. The oxidation of SWCNTs was confirmed by FTIR (Thermo Nicolet, NEXUS 470) with peaks at 3400 cm−1 and 1700 cm−1, characteristic for the carboxyl groups [20]. The higher dispersion of functionalized CNTs in water also distinguished them from the raw CNTs.

A low dose of CNT is less toxic to cells [15,16]. In the preparation of the collagen/CNT composite material, we added the oxidized SWCNTs to a 1.1 mg/ml collagen solution at a final CNT concentration of 0.005 mg/ml. This was followed by the same gelation procedure as for pure collagen, except a longer gelation time (2.5–3 hrs).

hESC culture

H9 cells from WiCell (Madison, WI) were grown according to standard Wicell protocol. In brief, cells were routinely cultured on feeder cells derived from mitotically inactivated mouse embryo fibroblast (MEFS). Cells were maintained in a DMEM/F12 (Mediatech, Herndon, VA) medium, containing 20% knockout serum replacement, 1% nonessential amino acid, 1 mM l-glutamine, 100 μM beta-mecaptoethanol, and 4 ng/ml bFGF (Invitrogen, Carlsbad, CA). Confluent hESCs were split using 0.5% trypsin EDTA and plated on the prepared matrices at a ratio of 1:6. The cells were grown in a DMEM (high glucose) medium, containing10% fetal bovine serum and 1% non essential amino acids, commonly used for spontaneous differentiation of hESCs.


Day 3 cells grown on the collagen, collagen/CNT and gelatin matrices were fixed and permeabilized with pre-cold methanol for 5 minutes and rinsed with PBS. The cells were then incubated with the primary antibody against nestin (purchased from Millipore, Phillipsburg, NJ) with 1% BSA in PBST for one hour at room temperature followed by the incubation with alexa fluor 594 fluorescent conjugated secondary antibody for one hour at room temperature. The nuclei were stained with DAPI for 5 minutes. A Nikon TE-U 2000 fluorescence microscope was used in the study.

AFM measurements

Collagen, collagen/CNT or gelatin coated substrates were thoroughly rinsed with DI-H2O before the surfaces were gently dried with nitrogen while retaining the moisture on the surfaces. The AFM imaging was carried out in air-tapping mode using a multimode Nanoscope IIIa AFM (Veeco Metrology, Santa Barbara, CA), equipped with a J-scanner. Single crystal silicon tips at a resonance frequency of 300–350 kHz were used in the studies.

The substrate stiffness measurements were conducted by using Si3N4 tips operated in fluid contact mode in DI-H2O. AFM nano-indentation method was applied to obtain the Young’s Modulus using the formula [2123]

Eq. 1

where E is the young’s modulus, k is the spring constant of the AFM tip, νis the poisson’s ratio (commonly takes 0.5 for biological samples), α(= 35°) is the half angle of the tip, Δd is the relative cantilever deflection and Δz is the relative sample displacement. The spring constant of the tips was 0.056 ± 0.002 N/m, as calibrated by using reference cantilevers with known spring constants [22,24]. From a force spectrum, which illustrates the tip deflection-displacement relationship during the tip approaching and retraction from a substrate, Δd and Δz can be experimentally determined and thus Young’s modulus can be evaluated. The average E value of each matrix type was estimated based on at least 25 separate measurements at five different areas and on different samples with different tips.


hESC differentiation on various matrices

Undifferentiated hESCs grow in compact colonies, with well-defined round-shaped boundaries as shown in Figure 1a. Three days after they were re-seeded on the substrates without the MEFS feeder layer or the conditioned media, the cells on gelatin showed random cell morphologies (Fig. 1b), deviating from the small spherical shape of undifferentiated hESCs. However, the cells barely expressed the markers of sox17 (endoderm), brachury (mesoderm) and nestin (ectoderm) for early stage differentiation. With collagen or collagen/CNT as the matrices, more than 90% of the cells elongated and well spread with long filaments (Fig. 1e,h).

Figure 1
Cell morphology and nestin expression of hESCs after cultured in the medium of spontaneous differentiation for three days on various matrices. The cells were fixed by methanol for 5 minutes. (a) Phase-contrast image of a colony of undifferentiated cells; ...

The cells and their filamental extensions preferentially aligned along one direction due to the dip-coating preparation of collagen or collagen/CNT cell culture matrices, which will be analyzed later. The same alignment was also obeyed by most nuclei, evident in Fig. 1g and j .

The phenotypic feature of the cells implies some degree of ectodermal lineage specification. When the differentiated cells at Day 3 were stained by the antibody against nestin, an early neural progenitor marker [25], to our surprise, cells on the collagen/CNT matrix (Fig. 1i) demonstrated significantly higher level of nestin expression than those on the pure collagen matrix (Fig. 1f), regardless of the similarity in cell shape in the two cases. This was reproducibly observed with no exceptions. Note that Fig. 1e–j illustrate the regions with relatively high cell densities. The same cell morphology and the marker expression characteristics were shown at regions of low cell densities. Over a longer term (Day 6) of culture in the same medium, we observed increased nestin expression level on pure collagen while the expression level on collagen/CNT remained higher. Meanwhile, a low level of nestin expression showed up on cells cultured on gelatin. Taken together, while hESCs spontaneously differentiate into various cell lineages in the absence of the MEFS feeder and the conditioned media, the collagen and collagen/CNT matrices specify the differentiation towards nestin-expression cells, which are likely neural progenitor cells. The collagen/CNT matrix stimulates the hESC differentiation to the ectodermal lineage at a much faster pace, making the composite material a superior matrix for producing neural progenitors.

Matrix surface characterization

To unveil the difference in physical property among the gelatin, collagen and collagen/CNT matrices, we carried out AFM characterization. As illustrated in Fig. 2, the gelatin surface is structure-less, while one-dimensional fibrous nano-scaffolds are dominant on the collagen and the collagen/CNT surfaces. Notably, thinner, better spread, flexible and tangled collagen fibrils (5–15 nm in diameter) are frequently seen on the surface of pure collagen, in contrast to the thicker, rigid fibril bundles (35–55 nm in diameter) with less surface coverage in collagen/CNT samples. At high resolution (see insets of Fig. 2b and c), the D-periods of collagen are clearly visible, and were measured at 68 ± 1 nm and 70 ± 1 nm, respectively, on the pure collagen and the collagen/CNT samples. The difference is small yet conclusive according to the statistical analysis (based on more than 30 measurements for each matrix type).

Figure 2
AFM characterization (amplitude images) of the surface structures of gelatin (a), collagen (b) and collagen/CNT (c) matrices prepared under the cell culture conditions. Image size: 8.5 × 8.5 μm2, inset 2 × 2 μm2.

Besides the structural modification, the addition of CNT to collagen altered the matrix stiffness. The stiffness of the three matrices was compared via the elasticity measurement by the nano-indentation method. The Young’s Moduli of gelatin, collagen and collagen/CNT were calculated using Eq. 1, and were 0.07 MPa,0.59 MPa, 1.88 MPa and, respectively, in contrast to ~700 MPa measured on blank tissue culture dishes (see Table 1). Gelatin provides a much softer surface for cell culture. Apparently, CNT confers additional stiffness to the collagen fibrils. Note that CNTs were not visualized on the collagen/CNT samples, though CNTs can be well resolved on samples from pure CNT preparation. Since both CNT and collagen are one-dimensional materials, and the –COOH functionalized CNTs can strongly interact with the amine and acid groups of collagen via the interactions such as hydrogen bond, electrostatic interaction and the hydrophilic interactions, we infer that the CNTs may be “woven” into the collagen fibers, forming an integrated structure at the molecular level. The increased D-period in collagen/CNT supports such an assumption. Further study of the mechanism of collagen-CNT interaction is undertaken in the lab.

Table 1
Physical Properties of cell culture matrices characterized by AFM

Cell – matrix interaction

Collagen fibril bundles were not visualized in Fig. 1 on the methanol fixed samples. Most likely, the fixation distorted the collagen structures. To understand the cell-matrix interaction, we performed live cell imaging. Figure 3 shows the phase contrast images of live hESCs at Day 3 of differentiation. A region of low cell density was chosen to highlight the relative locations of the cells and the collagen fibers. Consistent with the result of AFM study, we observed pure collagen formed thin, flexible and tangled fibril bundles and were well spread across the substrate. In the collagen/CNT sample, the fibril bundles were thicker with less coverage on the substrate. Coarse collagen alignment is visible on both samples, with a better alignment in collagen/CNT due to its thick and rigid fibril bundles. Consequently, cells on collagen/CNT mostly grew long bipolar filaments along the side of the well-oriented thick fibril bundles; whereas cells on pure collagen grew filaments frequently branching and extending along variable directions. Occasionally, cells were found at the collagen-free regions (e.g., the cell in the yellow circle at the lower portion of Fig. 3b). These cells exhibit random geometry, and differ from those grown at the collagen-rich areas. They barely stain nestin according to the immunofluorescence. Taken together, we have shown that the growth and the development of individual hESCs are remarkably responsive to the structure and the stiffness of matrices, indicating a strong cell-matrix interaction.

Figure 3
Phase-contrast images of live hESCs cultured on collagen (a) and collagen/CNT (b) for three days in the medium of spontaneous differentiation. The yellow circles refer to cell bodies; the red arrows indicate the variable directions of filament development ...


hESCs can spontaneously differentiate to cells in endoderm, mesoderm and ectoderm lineages. When bio-factors remained the same in the media, our results have shown that the microarchitecture of a cell matrix plays a critical role in controlling the overall growth and differentiation pattern of hESCs. A structure-less and soft gelatin matrix results in the hESC differentiation to cells in all three lineages. A collagen or a collagen/CNT matrix, characterized by one-dimensional fibril structures, causes preferential development of hESCs to elongated cells with long filaments. Both the cell nuclei and the extensions showed coarse alignment when the collagen was pre-aligned (upon dip-coating). The observation is consistent with those by others [4,6], suggesting that the microstructure of a cell matrix can provide the guidance for cell shape development by exerting anisotropic local stresses to the cells to influence the positions in which cells assemble their focal adhesions and the orientation in which they spread. The cells are also sensitive to the difference between collagen and collagen/CNT. The addition of CNTs not only modified the nanostructure of collagen (increase of D-period), but also enhanced the collagen-collagen interaction, leading to the elimination of flexible, thin, tangled fibrils and the development of rigid, long, straight and thick fibril bundles (Fig. 2). Cells were responsive to such matrix properties by growing extremely long, bipolar filaments closely adhered to the collagen/CNT fibril bundles (Fig. 3b).

It is not a surprise that a large majority of hESCs grown on the collagen and the collagen/CNT matrices developed into elongated cells with long filaments, resembling ectodermal cells. Strikingly, the cells on a collagen/CNT matrix expressed nestin at a very high level at as early as Day 3 in the medium for spontaneous differentiation, suggesting a remarkable polarization in the early differentiation choices toward the ectodermal lineage. To our knowledge, this is the first report of differentiating hESCs to neural progenitor cells in three days in a simple yet efficient way while bypassing the formation of embryoid bodies. Pure collagen also specified the ectodermal lineage at a later stage, evident by the increased level of nestin expression at Day 6. The difference in the rate of ectodermal lineage commitment of hESCs on the collagen and the collagen/CNT matrices is ascribed to their distinctions in both the microstructure and the stiffness. In particular, cells are very sensitive to ECM stiffness, which determines the level of tensional forces in the cell cytoskeleton. These forces, mediated by the transmembrane adhesion proteins, e.g., integrins and cadherins, can substantially affect the expression of the signaling proteins in stem cells [68]. Collagen type I is an ECM ligand for integrins. The cytoplasmic tail of these integrins is bound by integrin linked kinase (ILK) [26], a focal adhesion protein that resides next to actins. It was reported that ILK activation is implicated in accumulation of cytoplasmic beta-catenin, thus positively affects the wnt-beta catenin signaling which is essential for neuronal differentiation [27]. We speculate that the increased tensional forces offered by the stiffer collagen/CNT fibrils modulate the stresses at the actin structure, which are then transduced into the wnt-beta catenin pathway via ILK, promoting the neuronal differentiation.

This research explored the use of a collagen/CNT composite material as an in vitro cell culture matrix to direct the hESCs’ early differentiation to neural progenitor cells at a high yield. In our future work, incorporation of biochemical studies will be pursued to unveil the signal pathway of the matrix regulated differentiation events. A detailed time-dependent study is expected to provide the insight into the differentiation routes of hESCs on various cell culture matrices. The development of methods for quantitative spatial control of collagen/CNT alignment will render the patterned hESC differentiation to ectodermal, and further neural cells, which naturally form well confined nano/micro bio-devices for further applications.


This research was supported by NIH (R01 NS047719) and a seed grant from the Pritzker Institute of Biomedical Science and Engineering at IIT.


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


1. Itskovitz-Eldor J, Schuldiner M, Karsenti D, Eden A, Yanuka O, Amit M, Soreq H, Benvenisty N. Differentiation of the human embryonic stem cells into embryoid bodies into the three embryonic germ layers. Mol Med. 2000;6:88–95. [PMC free article] [PubMed]
2. Kurpinski K, Chu J, Hashi C, Li S. Anisotropic mechanosensing by mesenchymal stem cells. Proc Natl Acad Sci. 2006;103:16095–100. [PubMed]
3. Ruiz SA, Chen CS. Emergence of patterned stem cell differentiation within multicellular structures. Stem cells. 2008;26:2921–7. [PMC free article] [PubMed]
4. McBeath R, Pirone DM, Nelson CM, Bhadriraju K, Chen CS. Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Dev Cell. 2004;6:483–495. [PubMed]
5. Ma W, Tavakoli T, Derby E, Serebryakova Y, Rao MS, Mattson MP. Cell-extracellular matrix interactions regulate neural differentiation of human embryonic stem cells. BMC Dev Biol. 2008;8:90. [PMC free article] [PubMed]
6. Jones ENB, Mallapragada SK. Directed growth and differentiation of stem cells towards neural cell fates using soluble and surface-mediated cues. J Biomater Sci Polymer Edn. 2007;18:999–1015. [PubMed]
7. Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell. 2006;126:677–89. [PubMed]
8. Forte G, Carotenuto F, Pagliari F, Pagliari S, Cossa P, Fiaccavento R, Ahluwalia A, Vozzi G, Vinci B, Serafino A, Rinaldi A, Traversa E, Carosella L, Minieri M, Di Nardo P. Criticality of the biological and physical stimuli array inducing resident cardiac stem cell determination. Stem Cells. 2008;26:2093–103. [PubMed]
9. Rowlands AS, George PA, Cooper-White JJ. Directing osteogenic and myogenic differentiation of MSCs: interplay of stiffness and adhesive ligand presentation. Am J Physiol Cell Physiol. 2008;295:C1037–44. [PubMed]
10. Vunjak-Novakovic G. Patterning stem cell differentiation. 2008;3:362–63. [PMC free article] [PubMed]
11. Burdick JA, Vunjak-Novakovic G. Review: Engineered microenvironments of controlled stem cell differentiation. Tissure Eng Part A. 2008;14
12. Qiu D, Xiang J, Li Z, Krishnamoorthy A, Chen L, Wang R. Profiling TRA-1–81 antigen distribution on a human embryonic stem cell. Biochem Biolphys Res Com. 2008;369:735–740. [PMC free article] [PubMed]
13. Lin HJ, O’Shaughnessy TJ, Kelly J, Ma W. Neural stem cell differentiation in a cell-collagen-bioreactor culture system. Brain Res Dev Brain Res. 2004;153:163–73. [PubMed]
14. Zanello LP, Zhao B, Hu H, Haddon RC. bone cell proliferation on carbon nanotubes. Nano letters. 2006;5:562–7. [PubMed]
15. MacDonald RA, Laurenzi BF, Viswanathan G, Ajayan PM, Stegemann JP. Collagen-Carbon Nanotube composite materials as scaffolds in tissue engineering. J Biomed Mater Res Part A. 2005;74:489–96. [PubMed]
16. Mooney E, Dockery P, Greiser U, Murphy M, Barron V. Carbon nanotubes and mesenchymal stem cells: Biocompatibility. proliferation and differentiation. Nano lett. 2008;8:2137–43. [PubMed]
17. Xing Y. Synthesis and electrochemical characterization of uniformly- dispersed high loading Pt nanoparticles on sonochemically-treated carbon nanotubes. J Phys Chem B. 2004;108:19255–19259.
18. Hirsch A. Functionalization of single-walled carbon nanotubes. Angew Chem, Int, Ed. 2002;41:1853–59. [PubMed]
19. Balasubramanian K, Burghard M. Chemically functionalized carbon nanotubes. Small. 2004;1:180–192. [PubMed]
20. Lee TM, Huang CC. High thermal efficiency carbon nanotubes-resin matrix for thermal interface materials. J Phys Chem Part B. 2003;107:55–59.
21. Sneddon IN. The relation between load and penetration in the axisymmetric boussinesq problem for a punch of arbitrary profile. Int J Eng Sci. 1965;3:47–57.
22. Radmacher M, Fritz M, Kacher CM, Cleveland JP, Hansma PK. Measuring the viscoelastic properties of human platelets with the atomic force microscope. Biophys J. 1996;70:556–567. [PubMed]
23. Hutter JL, Chen J, Wan WK, Uniyal S, Leabu M, Chan BMC. Atomic force microscopy investigation of the dependence of cellular elastic muduli on glutaraldehyde fixation. J Microscopy. 2005;219:61–68. [PubMed]
24. Tortonese M, Kirk M. Characterization and application of specific probes for SPMs. SPIE. 1997;3009:53–60.
25. Lendahl U, Zimmerman LB, Mckay RDG. CNS stem cells express a new class of intermediate filament protein. Cell. 1990;60:585–595. [PubMed]
26. Hannigan GE, Leung-Hagesteijn C. Regulation of cell adhesion and anchorage-dependent growth by a new beta 1-integrin-linked protein kinase. Nature. 1996;379:91–96. [PubMed]
27. Otero JJ, Fu W, Kessler JA. b-Catenin signaling is required for neural differentiation of embryonic stem cells. Development. 2004;131:3545–3557. [PubMed]