Available online at www.sciencedirect.com
Acta Biomaterialia 4 (2008) 218–229 www.elsevier.com/locate/actabiomat
Engineering cell de-adhesion dynamics on thermoresponsive poly(N-isopropylacrylamide) Beiyi Chen a, F.J. Xu b,c, Ning Fang a, K.G. Neoh b,*, E.T. Kang b, Wei Ning Chen a, Vincent Chan a,* a
Center of Biotechnology, School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 639798, Singapore b Department of Chemical and Biomolecular Engineering, National University of Singapore, Singapore 119260, Singapore c Division of Bioengineering, National University of Singapore, Singapore 119260, Singapore Received 11 July 2007; received in revised form 31 August 2007; accepted 10 September 2007 Available online 25 September 2007
Abstract Poly(N-isopropylacrylamide) (PIPAAm) has been demonstrated as an effective thermoresponsive polymer for non-invasive cell regeneration/recovery. However, little is known about the intricate relationship between the biophysical response of cells and physiochemical properties of PIPAAm during cell recovery. In this study, the de-adhesion kinetics of smooth muscle cell (SMC) on PIPAAm surfaces is probed with unique biophysical techniques. Water-immersion atomic force microscope (AFM) first showed that the nanotopology of PIPAAm surfaces is dependent on the polymerization time and collagen coating. It is found that the initial rate of cell de-adhesion increases with the increase in polymerization time. Moreover, the degree of cell deformation (a/R) and average adhesion energy are reduced with the increase of grafted PIPAAm density during 40 min of cell de-adhesion. It has been shown that collagen coating regulates cell adhesion on biomaterial surface. Interestingly, lower collagen density on PIPAAm leads to higher adhesion energy per cell during the initial 20 min compared with as-prepared PIPAAm, while the initial rate of cell de-adhesion remains unchanged. In contrast, higher collagen density leads to 50% reduction in the initial rate of cell de-adhesion and higher adhesion energy per cell during the entire 90 min. Furthermore, immunostaining of actin provides supporting evidence that the de-adhesion kinetics is correlated with the cytoskeleton transformation during cell de-adhesion below the lower solution critical temperature (LCST). 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: De-adhesion; Kinetics; SMC; Poly(N-isopropylacrylamide); Collagen
1. Introduction Thermoresponsive polymer (TRP) has emerged as a promising class of biomaterial for the regeneration of various cells such as hepatocytes, endothelial cells, urothelial cells and fibroblasts [1,2]. The application of TRP for the non-invasive recovery of cells/tissues is brought about by its acute switching of physiochemical properties across the lower critical solution temperature (LCST). In particu*
Corresponding authors. Tel.: +65 67904040 (V. Chan). E-mail addresses:
[email protected] (K.G. Neoh), mvchan@ntu. edu.sg (V. Chan).
lar, this technique eliminates the need for proteolytic enzymes or physical scraping for recovering cells from a tissue culture dish, and preserves the intracellular junctions as well as native tissue organization. One of the earliest members of the TRP family, poly(Nisopropylacrylamide) (PIPAAm), forms a hydrogel which is transformed from a swelling state to a collapsed state across its LCST [3]. Moreover, the LCST of PIPAAm can be engineered with the design of block copolymers containing PIPAAm and other polymers. To date, PIPAAm and its copolymers have been widely exploited for applications in bioseparation, drug delivery and other thermoresponsive devices [4,5]. Okano and co-workers have
1742-7061/$ - see front matter 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2007.09.002
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
pioneered the chemical grafting of PIPAAm onto tissue culture polystyrene (TCPS) by electron beam irradiation [6]. Alternatively, PIPAAm films may also be prepared by physical adsorption of the polymer chains onto glass [7], plasma polymerization of N-isopropylacrylamide (IPAAm) onto TCPS [8], photografting [9], gamma radiation [10], plasma immobilization [11] and plasma-induced graft polymerization [12]. It has been shown that the acute increase of hydrophilicity on PIPAAm films with the reduction in temperature drives the detachment of a confluent cell layer. Moreover, the time required for complete cell sheet detachment from PIPAAm films is highly dependent on cell types [13] and physiochemical properties of TRP. The biocompatibility of PIPAAm is less ideal than conventional TCPS for inducing strong cell adhesion, which is critical for effective tissue regeneration [14]. The biocompatibility of PIPAAm is improved by the incorporation of hydrophobic copolymer into the PIPAAm chain [15,16] or coating with extracellular matrix (ECM) proteins [17]. Collagen is the most abundant ECM protein found in mammalian tissues, forms a fibrous fiber with great tensile strength [18,19] and plays a key role in the physiological regulation of smooth muscle cell (SMC). It has also been shown that cell adhesion and migration on biomaterial surfaces are dependent on the concentration of adsorbed collagen [20]. SMC is an interesting model system for TRP because the non-invasive recovery of engineered tissues containing native SMC is a critical area of vascular tissue engineering [21]. The recent development of integrated biophysical techniques has elucidated the adhesion contact dynamics of cells on biomaterials during the initial cell seeding [22,23]. Recently, our unique approach has been successfully applied to probe the de-adhesion kinetics of single SMC from hydroxybutyl chitosan [24]. Moreover, our group has developed an improved two-step reaction to couple PIPAAm chains to silicon coverslips [25]. To the best of our knowledge, there is currently a lack of quantitative correlation between the physiochemical properties of PIPAAm films and the resulting de-adhesion kinetics of cells upon temperature reduction. In this study, the dynamics of SMC de-adhesion on PIPAAm surfaces under different polymerization times and collagen coating are probed with our unique biophysical techniques. Specifically, confocal reflectance interference contrast microscopy (C-RICM) in combination with phase contact microscopy determined key biophysical parameters of cell de-adhesion including initial de-adhesion rate, degree of deformation, average adhesion energy and adhesion energy per cell of SMCs. In addition, the cytoskeleton structure of SMC during the course of thermal-induced de-adhesion is detected by confocal fluorescence microscopy through immunostaining of actin. It is demonstrated that the de-adhesion dynamics and cytoskeleton transformation of SMC can be engineered by the polymerization time and collagen modification of PIPAAm surfaces. The elucidation of the kinetics of cell de-adhesion from different PIPAAm surfaces would
219
likely provide important insights for designing highly tailored process for cell regenerations. 2. Materials and methods 2.1. Materials High glucose Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), streptomycin, penicillin, 10· Trypsin–EDTA (0.5%) and 10· phosphate-buffered saline (PBS) (pH 7.4) were purchased from Gibco (Singapore). CuCl2, CuCl, NIPAAm, 1,1,4,7,10,10-hexamethyl-triethylenetetramino, paraformaldehyde, Triton X-100, fluorescein isothiocyanate (FITC)–phalloidin and anti-vinculin–FITC-conjugated antibody were purchased from Sigma Chemical Pte. Ltd. (Singapore). Type I collagen was bought from Benton Dickinson (Singapore). The reagents were diluted with 1· PBS. Highly purified 18.2 MX water was obtained from water purification system (Sartorius, Germany). 4(Chloromethyl)phenyltrichlorosilane (97%) was obtained from Alfa Aesar (Singapore). 2.2. Preparation of PIPAAm surface Glass coverslips were cleaned with piranha solution (30% H2O2 with 70% H2SO4) for 1 h, rinsed thoroughly with 18.2 MX water and then dried under vacuum at room temperature. The procedures for the preparation of PIPAAm-grafted surfaces using the atom transfer radical polymerizations (ATRP) reaction have been described elsewhere [25,26]. In brief, the glass coverslips were first immersed in 30 ml of chloroform followed by 0.5 ml of triethylamine and 2 ml of 4-(chloromethyl)phenyltrichlorosilane. After reaction for 24 h, the coverslips were washed thoroughly with acetone and kept in acetone for another 30 min to remove the unreacted silane. After the coverslips were dried in air, they were added to 15 ml of 0.23 g ml1 NIPAAm solution. Then 20 mg of CuCl and 4 mg of CuCl2 were added and the tube was degassed with argon for 20 min. Finally, 50 ll of 1,1,4,7,10,10-hexamethyl-triethylenetetramino was added to the mixture and the tube was sealed for the interfacial polymerization of NIPAAm. To terminate the polymerization, the samples were removed from the reaction mixture and washed sequentially with copious amounts of DMSO and double-distilled water, prior to being dried under reduced pressure. Throughout this report, NIPAAm surfaces obtained from different polymerization times of IPAAm are represented by the time following the polymer notation; for example, the sample with a polymerization time of 0.5 h is represented by PIPAAm-0.5 h. 2.3. Preparation of collagen-coated surface The original collagen solution (2.9 mg ml1 in 0.12 N HCl) was diluted to 0.1· (0.29 mg ml1) and 0.04· (0.12 mg ml1). A 200 ll quantity of solution at each
220
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
concentration was poured onto the PIPAAm surface and incubated at 4 C for 24 h. The surfaces were then neutralized with 1 N NaOH for 20 min and thoroughly washed with deionized water to remove the excess ions. Finally, the surfaces were sterilized with 70% ethanol and stored in refrigerator before further experiments. Throughout this report, collagen was coated on PIPAAm-1.5 h, and surfaces obtained using 0.29 and 0.12 mg ml1 of collagen in the coating process are represented by 0.1· and 0.04·, respectively, before the polymer notation. 2.4. X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) The XPS measurements were performed on a Kratos AXIS HSi spectrometer using a monochromatized Al Ka X-ray source (1486.6 eV photons) and procedures similar to those described earlier [21,22]. An atomic force microscope (Model: MFP-3D, Asylum Research, CA) with water-immersion capability was used for all nanotopographical measurements with contact mode under water at 37 and 18 C, which is above and below the LCST of PIPAAm, respectively. A scanning rate of 1 Hz was used in all experiments. An electronic fluid cell was applied to control the temperature of the PIPAAm surface under water. Each AFM topographic image was representative of at least 10 images on three samples and is highly reproducible using the same procedures of sample preparation. In certain cases, AFM measurements of PIPAAm or collagen-coated PIPAAm surfaces were carried under the dry mode at 18 C. 2.5. Cell culture A7r5 smooth muscle cells (ATCC: CRL 1444) were originally derived from embryonic rat aorta and exhibit the typical phenotypes of adult SMC. The cells were cultured in high glucose DMEM supplemented with 10% FBS, 5 mg ml1 penicillin and 5 mg ml1 streptomycin (Gibco, Singapore). Cell cultures were maintained at 37 C under a humidified atmosphere of 5% CO2 in air. The culture medium was changed every 2 days, and the cells were passaged at least once a week. Cells were detached from the culture flask by the addition of 0.5% 5.3 mM Trypsin– EDTA solution in PBS. Before each experiment, the coverslip grafted with different density of PIPAAm or collagen-coated PIPAAm was maintained above the LCST (at 37 C) in order to induce cell adhesion. SMCs at a density of 5200 cell cm2 were then seeded onto the PIPAAms samples and incubated at 37 C in a 5% CO2 environment for 3 h. Afterwards, the samples were immediately transferred to an online CO2 incubator at 37 C under a humidified atmosphere of 5% CO2 in air. After equilibration at 37 C for 20 min, the temperature ramping system was turned on and the temperature was rapidly decreased to 18 C within another 2 min for subsequent biophysical measurements.
2.6. Confocal reflection interference contrast microscopy (C-RICM) The system is based on a laser scanning confocal microscope (Pascal 5, Carl Zeiss, Germany) and is integrated with an online CO2 incubator (Carl Zeiss, Germany). The illumination source is an argon ion laser with a maximum power of 1 mW and excitation wavelength of 488 nm. A 20· objective (NA: 1.25) was used in this study. The strong contact zone of the adhering cells appears as dark region on the image. Simultaneously, a transmitted light analyzer (Carl Zeiss, Germany) attached to the confocal microscope was used for performing phase contrast microscopy of the adherent cells. The cells were cultured on PIPAAm surface for 3 h at 37 C and 5% CO2 incubation, and then refreshed with serum-free medium at 37 C. The sample was then transferred to a online CO2 incubator on the microscope stage at 37 C. The online incubator has the capability of rapid cooling (PE94, Linkam, UK). After 20 min of incubation at 37 C, the temperature was ramped down to 18 C within 2 min in a 5% CO2 atmosphere. In order to monitor the de-adhesion/detachment dynamics of SMC from various PIPAAm surfaces below the LCST, a series of CRICM images on a selected region of the sample was taken against time during incubation at 18 C. At each time point, 60 or more cells on at least four identical samples were imaged. The contact area and projected area of each cell were measured by drawing a region of interest with a PC mouse at the periphery of the dark region (representing strong adhesion contact) with the ZSM 5 software (Carl Zeiss, Germany). The average radius of adhesion contact area a and projected area R of each cell were calculated from the adhesion contact area and projected area, respectively. The normalized adhesion contact area is defined as the ratio of adhesion contact area at any time during cell de-adhesion (a) and adhesion contact area before the temperature reduction at 37 C (a0). The degree of deformation of each cell, which represents the cell geometry, is simply the ratio a/R. 2.7. Immunostaining and fluorescence microscopy The cells before de-adhesion at 37 C or cells after 30 min of de-adhesion at 18 C from PIPAAm surface were stained with FITC–phalloidin for labeling the actin. In brief, SMCs were washed with cold PBS and fixed with 3% formaldehyde in 1· PBS at room temperature for 10 min. The cells were then permeabilized with 0.2% Triton X-100 in PBS for 10 min at room temperature following by a wash with 1· PBS and blocking with 10% FBS in 0.1% Triton X-100 for another 10 min. Then the fixed and permeabilized SMCs were incubated with FITC–phalloidin for 1 h. Lastly, the cells are washed with 0.1% Triton X100 for 15 min twice. Fluorescence images of SMCs immunostained with FITC–phalloidin were obtained using a confocal
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
fluorescence microscope (Pascal 5, Carl Zeiss, Germany) fitted with a 40· oil immersion objective and an argon ion laser as the excitation source (488 nm), at a resolution of 1024 · 1024 pixels and with a slow scan speed. The excitation light was set at 488 nm and fluorescence emissions between 505 and 530 nm were recorded.
achieved. It has been shown previously that the adhesion energy, W, is as follows: W ¼ ð1 cos hÞCe þ Ce2
The details of the contact mechanics model of adherent cells have been described previously [27]. Briefly, the equilibrated geometry of a cell adhering to a non-deformable substrate is modeled as a truncated sphere with a mid-plane radius R. The degree of deformation, sin h = (a/R) = a, is an experimentally measurable parameter, where R and a are measured by C-RICM and phase contrast microscopy, respectively. The cell wall is assumed to be under a uniform equi-biaxial stress, r = Ce. C is the stress equivalent and is equal to Eh/(1 m) in a linear system under small strain, where E, h and m are the elastic modulus, membrane thickness and Poisson’s ratio, respectively. The average biaxial strain, e, is directly calculated from R and a as follows: " # 1=2 1 2 þ 2ð1 a2 Þ e¼ 1 ð1Þ 2 4=R2 a2
3. Results and discussion The presence of PIPAAm layer and collagen coating was first confirmed by X-ray photoelectron spectroscopy (XPS) analysis. Fig. 1 shows the C 1s and N 1s core-level spectra of (a, a 0 ) PIPAAm-0.5 h, (b, b 0 ) PIPAAm-1.5 h,
In the absence of external force, the cell spontaneously adjusts its distance towards the substrate until equilibrium is
(a') N 1s
(b') N 1s
(b) C 1s
PIPAAm-0.5hr
PIPAAm-1.5hr
396
399
402
405
C-N
C-H
Intensity (Arb. Units)
O=C-N
284
286
(c) C 1s
288
290
396
C-H
C-H
282
ð2Þ
By taking the square root of the ratio of projected area (from phase contrast microscope) and adhesion contact area (from C-RICM), W can be found by Eqs. (1) and (2). For the case of an elliptical cell shape, a/R represents the ratio of the root mean square (RMS) radius for adhepffiffiffiffiffiffiffiffiffiffiffi sion contact ð a pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bÞ and the RMS radius of the projected area ð R1 R2 Þ, where a(R1) and b(R2) are the semi-major and semi-minor axis of the elliptical adhesion contact (projected shape), respectively. Thus a/R of an ellipse as mentioned above is useful for comparing the size of adhesion contact with that of the projected area. Despite the presence of an elliptical cell shape during the early stage of cell de-adhesion, our approach to quantifying the extent of adhesion contact formation against the projected cell area as mentioned above remains valid. The elastic modulus E of rat thoracic aortic smooth muscle cells has been determined as 9.3 ± 2.8 kPa [28].
2.8. Determination of adhesion energy
(a) C 1s
221
282
284
402
405
C-N O=C-N
286
(d) C 1s
(c') N 1s
399
288
290
(d') N 1s
0.1x-PIPAAm
0.04x-PIPAAm
C-H C-H C-N 396
399
402
C-N C-O O=C-N O=C-O
282
284
286
288
290
396
405
C-H
399
402
405
C-O O=C-N O=C-O
282
284
286
288
290
Binding Energy (eV) 0
Fig. 1. XPS C 1s and N 1s core-level spectra of (a, a ) PIPAAm-0.5 h, (b, b 0 ) PIPAAm-1.5 h, (c, c 0 ) 0.04·-PIPAAm and (d, d 0 ) 0.1·-PIPAAm.
222
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
(c, c 0 ) 0.04·-PIPAAm and (d, d 0 ) 0.1·-PIPAAm. The C 1s core-level spectra of PIPAAm-0.5 h and PIPAAm-1.5 h can be curve-fitted into three peak components with binding energies (BEs) of about 284.6, 285.7 and 287.4 eV, attributable to the C–H, C–N and O@C–N species of PIPAAm, respectively [22]. The increase of polymerization time (PIPAAm-0.5 h vs. PIPAAm-1.5 h) has led to a higher surface density of grafted PIPAAm, as shown by the slight increase of the intensities of C–N, and O@C–N peaks. The C 1s core-level spectra of 0.04·-PIPAAm and 0.4·-PIPAAm can be curve-fitted into five peak components with BEs at about 284.6, 285.7, 286.2, 287.5 and 288.4 eV, attributable to the C–H, C–N, C–O, O@C–N and O@C– O species, respectively. The C–N and O@C–N peak components are associated with the peptide bonds in collagen,
as well as the components of PIPAAm. The C–O and O@C–O peak components are solely associated with collagen. With the increase in collagen density on PIPAAm surface (0.04·-PIPAAm vs. 0.1·-PIPAAm), the intensities of the C–N, C–O, O@C–N and O@C–O components increase substantially. The results as mentioned above are consistent with the PIPAAm layer and the collagen coating being successfully prepared on the substrate. Moreover, the formation of PIPAAm layer and collagen coating was validated by AFM. Fig. 2 shows typical 2 · 2 lm AFM topographic images of PIPAAm-0.5 h (a, b), PIPAAm-1.5 h (c, d) and 0.04·-PIPAAm at either 18 C (a, c, e) or 37 C (b, d, f) under water. At 18 C, the morphology of PIPAAm-0.5 h is homogeneous below the LCST, as shown by the low value of RMS roughness
Fig. 2. A series of AFM images of PIPAAm samples at 18 C (a, c, e) and 37 C (b, d, f) in a water environment. They are PIPAAm-0.5 h (a, b), PIPAAm1.5 h (c, d) and 0.04·-PIPAAm (e, f).
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
of 0.9 nm (Fig. 2a). Similar nanotopology has been found in other polymer thin films immobilized on glass substrate [29]. In contrast, a shorter polymerization time leads to highly heterogeneous topology due to incomplete PIPAAm coverage (data not shown). When the temperature moves above the LCST to 37 C in situ, the nanotopology of the same region on PIPAAm-0.5 h remains unchanged except that a small number of agglomerates with average diameter of 85 nm emerges (Fig. 2b). The result as mentioned above leads to slight increase of RMS roughness, from 0.9 to 1.1 nm, and is linked to the aggregations of PIPAAm chains which are sparsely distributed on the surface above the LCST [30–32] (Table 1). The increase in polymerization time from 0.5 to 1.5 h leads to the formation of massive number of circular domains with an average diameter of 50 nm at 18 C (Fig. 2c). The surface of PIPAAm-1.5 h is rougher than that on PIPAAm-0.5 h, as shown by the greater RMS roughness of 1.7 nm. When temperature increases from 18 to 37 C on the same region of interest, a large number of circular domains disappear and the surface becomes smoother (Fig. 2d). The result is supported by the reduction of RMS roughness from 1.7 to 1.1 nm, and is caused by the collapse of PIPAAm layer above LCST. Upon collagen coating, dense collagen fibers are formed on the PIPAAm-1.5 h surface (Fig. 2e). A similar structure of collagen fiber is found on other substrates, e.g. the self-assembled monolayer on gold [33]. Moreover, the surface feature and morphology of 0.04·-PIPAAm remains unchanged during heating across LCST (Fig. 2f). The result suggests that the adsorbed collagen is stable during the thermal transition of PIPAAm layer. C-RICM has been proven to be an effective tool for probing the real-time change in cell–biomaterial interactions [34]. C-RICM result indicates that PIPAAm-5 min is ineffective to trigger the cell de-adhesion even if temperature falls below the LCST. Most cells adhere directly to the silane layer on top of the glass surface, which is not thermosensitive. Fig. 3 shows a set of (a) phase contrast images and (b) C-RICM images of a typical SMC in real-time on PIPAAm-0.5 h from 0 to 60 min after incubation at 18 C and 5% CO2. At 37 C, the cell adopts a diamond shape with the formation of lamellipodium (Fig. 3a), and establishes strong adhesion contact (dark region) after 3 h culture on the surface (Fig. 3b). The initial projected and adhesion contact areas are 650 and 525.3 lm2, respectively, at 0 min and 37 C. From 0 to 10 min of 18 C incubation, the projected area and the contact area are Table 1 Root-mean-square (RMS) roughness of PIPAAm surfaces RMS (nm)
PIPAAm-0.5 h PIPAAm-1.5 h 0.04·-PIPAAm
18 C
37 C
0.9 ± 0.4 1.7 ± 1.8 1.2 ± 0.5
1.1 ± 0.6 1.1 ± 0.8 1.6 ± 0.5
223
significantly reduced by 20.9% and 19.6%, respectively. The rate of recession of both areas is reduced after 30 min. After 60 min, the cell becomes more spherical due to the active retraction of the cell body. Overall, the projected and contact areas are reduced to 205 lm2 (by 68.5%) and 143.2 lm2 (by 72.7%). It has been recently shown that the change in osmotic pressure alters the adhesion contact area of liposome on a rigid substrate [27]. In our current study, the DMEM used for culturing cells on PIPAAm had a similar salt concentration as to the cytoplasm of SMC. Under the iso-osmotic condition between cell and medium, the effect of osmotic pressure on the cell de-adhesion kinetics is negligible. The change in polymerization time of NIPAAm may trigger a change in cell de-adhesion kinetics. Fig. 4 shows a series of (a) phase contrast images and (b) C-RICM images of a typical SMC on PIPAAm-1.5 h from 0 to 60 min after incubation at 18 C and 5% CO2. The initial projected and adhesion contact areas are 556.8 and 431.2 lm2, respectively, at 0 min and 37 C. From 0 to 10 min, the cell rounds up from its originally elongated shape as the projected area and adhesion contact area are reduced by 45.7% and 52.3%, respectively. This result shows that the extent of cell retraction on PIPAAm-1.5 h is significantly greater than that on PIPAAm-0.5 h during the initial 10 min. Moreover, the result demonstrates that the increase of polymerization time accelerates the adhesion contact recession. The general driving force of cell sheet detachment has been discussed previously by Okano et al. for PIPAAm-grafted TCPS [34]. First of all, the hydrophobic to hydrophilic transition of PIPAAm chains during the temperature drop below the LCST has been known to induce passive de-adhesion/detachment of cells. This is followed by active detachment triggered by cellular metabolism in order to complete cell detachment. Our result demonstrates that sufficient coverage of PIPAAm on the substrate provided a passive driving force that induced de-adhesion of single cells within a relatively short time interval of low-temperature incubation. However, single cells fail to detach completely from the PIPAAm surface because of the absence of intracellular contacts in cell sheet as shown in a previous study [35]. The role of ECM protein on cell de-adhesion remains to be elucidated. Fig. 5 shows a series of C-RICM and phase contrast images of a typical SMC on 0.04·-PIPAAm (a, b) and 0.1·-PIPAAm (c, d) at different times after 18 C incubation. This result shows that SMC spontaneously shrinks by retracting its lamellipodium within the initial 10 min of low-temperature incubation on 0.04·-PIPAAm (Fig. 5a and b). The adhesion contact area and projected area on 0.04·-PIPAAm are reduced by 67.2% and 64.3%, respectively, during the initial 20 min, and are reduced by 71.1% and 69.8% during the entire period of 1 h. This trend is likely to be the result of the sparse concentration of collagen which triggered cell traction through the activation of integrin–collagen interaction below the LCST. In contrast, the recession of adhesion contact and retraction of cell
224
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
Fig. 3. A series of (a) phase contrast images and (b) C-RICM images for a typical SMC on PIPAAm-0.5 h surface which has been pre-cultured at 37 C for 3 h at different times during 18 C incubation. The scale-bar represents 10 lm.
Fig. 4. A series of (a) phase contrast images and (b) C-RICM images for a typical SMC on PIPAAm-1.5 h surface which has been pre-cultured at 37 C for 3 h at different times during 18 C incubation. The scale-bar represents 10 lm.
body are significantly delayed on 0.1·-PIPAAm (Fig. 5c and d). The cell remains to be elongated towards the end of the de-adhesion, as shown by the phase contrast image. The adhesion contact area and projected area are reduced by only 30.1% and 39.2% during the first 20 min, and by 34.2% and 49.1% during the entire 1 h period (Fig. 5c and d). This result shows that the high concentration of collagen has prevented the cells from de-adhered from PIPAAm-1.5 h. In our study, two concentrations of collagen solution, 0.1· (0.29 mg ml1) and 0.04· (0.12 mg ml1), were used to coat the PIPAAm surface. The higher colla-
gen concentration (0.29 mg ml1) was rationalized from our previous work of coating collagen on several types of polymeric biomaterial films, which demonstrates the formation of a highly saturated layer of adsorbed collagen under this higher range of collagen concentration [34]. The lower concentration of collagen has actually been demonstrated as the transitional level below which the cell de-adhesion kinetics becomes insensitive to the further reduction of collagen concentration, i.e. 0.029 mg ml1. Furthermore, our XPS data validate the significant difference in the concentration of adsorbed collagen under the
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
225
Fig. 6. (a) Average a/a0 for SMCs pre-incubated on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm against time during low-temperature incubation at 18 C. The error bar is the standard error of at least 60 cells on three identical samples. (b) Average a/R for SMCs pre-incubated on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm against time during low-temperature incubation at 18 C. The error bar is the standard error of at least 60 cells on three identical samples. Fig. 5. A series of (a) phase contrast images and (b) C-RICM images for a typical SMC on 0.04·-PIPAAm surface; (c) phase contrast images and (d) C-RICM images for a typical SMC on 0.1·-PIPAAm surface which has been pre-cultured at 37 C for 3 h at different times during 18 C incubation. The scale-bar represents 10 lm.
two concentrations of collagen solution used. The trend of cell de-adhesion against the change of collagen concentration agrees with the recent report on the minimized fibroblast spreading at intermediate concentration of collagen [20]. The normalized adhesion contact area a/a0 quantifies the rate of recession in adhesion contact area during cell de-adhesion below the LCST. Fig. 6a shows a plot of a/ a0 against time during SMC de-adhesion on PIPAAm0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm at 18 C. The error bar is the standard error of at least 60 cells on three identical samples. As the adhesion contact area shrinks during de-adhesion, a/a0 for SMCs on all four
samples follows an exponential decay until it reaches a steady-state after a long time. By fitting the experimental data with a typical exponential decay function, y = y0 + menx (fitted lines in Fig. 6a), the kinetic coefficient n is obtained, as shown in Table 2. The fitted parameter n for cells on PIPAAm-0.5 h and PIPAAm-1.5 h is 0.023 and 0.051 min1, respectively. This result, as mentioned above, indicates that the initial rate of de-adhesion (decay rate of a/a0) is doubled by the increase of polymerization time of
Table 2 Fitted parameters in the de-adhesion kinetics of smooth muscle cells on PIPAAm and collagen-coated PIPAAm surfaces n PIPAAm-0.5 h PIPAAm-1.5 h 0.04·-PIPAAm 0.1·-PIPAAm
0.023 0.051 0.051 0.031
226
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
IPAAm. Moreover, a recent study has shown the existence of a concentration threshold of PIPAAm chains on TCPS for achieving effective cell detachment [36]. After collagen coating, n is 0.051 and 0.031 min1 on 0.04·-PIPAAm and 0.1·-PIPAAm, respectively. The result indicates that the initial decay rate of a/a0 on 0.04·-PIPAAm is similar to that on PIPAAm-1.5 h (Table 2). After 20 min of deadhesion, a/a0 of cells on 0.04·-PIPAAm is lower than that of cells on PIPAAm-1.5 h. Interestingly, the 50% reduction of n on 0.1·-PIPAAn compared with that on 0.04·-PIPAAm indicates that the higher collagen concentration is required for reducing the initial rate of cell de-adhesion. Moreover, the steady-state level of a/a0 on 0.1·-PIPAAm is 15% and 65% higher than that on PIPAAm-1.5 h and 0.04·-PIPAAm, respectively. The overall trend shows that the adhesion contact recession from PIPAAm is highly dependent on the concentration of coated collagen. Since a/a0 does not provide any insight into the actual geometry of cells during de-adhesion, the degree of cell deformation (a/R) is used for quantifying the geometric transformation of the entire cell [27]. Fig. 6b shows the average a/R for SMCs against time on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm at 18 C. The error bar is the standard error of at least 60 cells on three identical samples. The result indicates that the initial rate of reduction (during the initial 10 min) of a/R for PIPAAm-0.5 h and PIPAAm-1.5 h is 0.005 and 0.01 min1, respectively. Analogous to the trend of a/a0, the increase of polymerization time of NIPAAm leads to a doubling of the initial rate of reduction of a/R. The result indicates that cells initially recover faster from a highly deformed shape to rigid geometry on PIPAAm-1.5 h than on PIPAAm0.5 h since the adhesion contact area reduces faster than the projected area. During the entire period of 2 h de-adhesion, a/R is reduced from 0.85 to 0.56 (by 34.1%) and 0.88 to 0.67 (by 23.8%) on PIPAAm-0.5 h and PIPAAm-1.5 h, respectively. The result indicates that the concentration of PIPAAm has more prominent effect on the initial stage of geometric transformation of cells. Moreover, the initial rate in reduction of a/R on 0.04·-PIPAAm is the same as that on PIPAAm-1.5 h, while a higher value of a/R is found during the initial 20 min. This result indicates that the lower concentration of collagen does not induce a distinct mechanism of initial cell geometry transformation compared with unmodified PIPAAm. In contrast, a/R remains unchanged at around 0.92 during the entire 2 h period of de-adhesion instead of the slow exponential decay found in a/a0 on 0.1·-PIPAAm. The unique trend mentioned above is brought about by the similar rate of retraction of cell body compared with that of adhesion contact recession under higher collagen concentration. Most importantly, the biophysical responses of SMC during de-adhesion are demonstrated to be dependent on the amount of collagen coated on the PIPAAm. The average adhesion energy is a biophysical parameter for characterizing the cell–substrate interaction during cell de-adhesion [27]. Fig. 7a shows the average adhesion
energy of SMCs on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm during de-adhesion (at 18 C) at 37 C and 5% CO2. This result shows that the initial rate of decay in average adhesion energy on PIPAAm1.5 h is higher than that on PIPAAm-0.5 h. On the other hand, the overall reduction in average adhesion energy from 1.8 · 107 to 1.5 · 108 J m2 on PIPAAm-0.5 h is higher than that on PIPAAm-1.5 h (from 1.8 · 107 to 6 · 108 J m2). The result supports the finding that the polymerization time of IPAAm directly modulates the initial decay of average adhesion energy. For cells on 0.04·PIPAAm, the initial decay in average adhesion energy is lower than that on PIPAAm-1.5 h. After 1.5 h, the adhesion energy on 0.04·-PIPAAm is reduced to that on PIPAAm-1.5 h (5 · 108 J m2). Interestingly, the average adhesion energy of cells on 0.1·-PIPAAm remains constant at 1.8 · 107 J m2 against time and is three times higher than that on PIPAAm-1.5 h after 1.5 h of cell
Fig. 7. (a) Average adhesion energy for SMCs pre-incubated on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm against time during low-temperature incubation at 18 C. The error bar is the standard error of at least 60 cells on three identical samples. (b) Adhesion energy per cell for SMCs pre-incubated on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm against time during low-temperature incubation at 18 C. The error bar is the standard error of at least 60 cells on three identical samples.
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
de-adhesion. The result strongly indicates that higher collagen concentration is required for preventing the cells from detaching from PIPAAm through the perseverance of strong cell–substrate adhesiveness. The non-ideal geometry of cells which deviate from a spherical and perfect elliptical shape is occasionally found during the de-adhesion of SMC and is likely caused by the complex cytoskeletal features of cells. On the other hand, there is currently no analytical solution of adhesion energy that would account for the non-ideal geometry of adherent cells. To quantify the collective adhesive force between cell and substrate, the adhesion energy per cell, which is the product of the average adhesion energy and the adhesion contact area, is used. Fig. 7b shows the adhesion energy per cell for SMCs on PIPAAm-0.5 h, PIPAAm-1.5 h, 0.04·-PIPAAm and 0.1·-PIPAAm during de-adhesion at 18 C and 5% CO2. During the initial 60 min, the trend of the adhesion energy per cell against time on PIPAAm0.5 h is similar to that on PIPAAm-1.5 h. However, the overall extent in the reduction of adhesion energy per cell from 0 to 90 min on the PIPAAm-0.5 h surface is five times higher than that on PIPAAm-1.5 h. These results suggest that the polymerization time of NIPAAm does not moderate the initial decay of adhesion energy per cell directly. On the other hand, the lower concentration of collagen (0.04·PIPAAm) led to a higher adhesion energy per cell than that on PIPAAm-1.5 h during the initial 30 min of cell de-adhesion. This result shows that a moderate amount of collagen concentration improves the PIPAAm’s biocompatibility without causing any apparent undesirable effect on the dynamics of cell de-adhesion. Interestingly, the initial decay of adhesion energy per cell on 0.1·-PIPAAm is significantly slower than that on 0.04·-PIPAAm. Moreover,
227
the adhesion energy per cell on 0.1·–PIPAAm is two orders of magnitude higher than those on 0.04·-PIPAAm or PIPAAm-1.5 h after 90 min. This indicates that the higher collagen concentration on PIPAAm maintains the strong interaction between the entire cell and the substrate during cell de-adhesion. Previously, it has been shown that cells go through cytoskeletal remodeling during interaction with biomaterials. Fig. 8 shows the immunofluorescence images of actin for representative SMCs on (a) PIPAAm-0.5 h, (b) PIPAAm-1.5 h, (c) 0.04·-PIPAAm and (d) 0.1·-PIPAAm at 0 and 30 min after low-temperature incubation at 18 C. Before cell de-adhesion, the actins are concentrated at the periphery of the elongated SMCs on both PIPAAm-0.5 h and PIPAAm-1.5 h. At the same time, the formation of microfilaments which isotropically orient along the long axis of cell is detected in the cytoplasm. The extensive spreading combined with the polarization of the cytoskeleton indicates that the cells spread well onto PIPAAm above LCST. The typical organization of the actin cytoskeleton as mentioned above has been reported in SMC adhering on other biomaterials [37]. After 30 min of cell de-adhesion, the concentration of actin at the cell periphery on PIPAAm-1.5 h is obviously higher than that on PIPAAm-0.5 h. The result indicates that cytoskeleton remodeling is correlated with the faster initial decay of a/a0 and a/R against the increase of polymerization time. This result is consistent with the recently reported role of actin cytoskeleton in the contractility of the cell through the increase in traction at the cell boundary [38]. It would also be of future interest to analyze the role of microtubules in the process of de-adhesion, as the dynamics of microtubules has been shown to affect cell shape changes [39].
Fig. 8. The fluorescence image of a typical SMC (actin staining) pre-cultured on (a) PIPAAm-0.5 h, (b) PIPAAm-1.5 h, (c) 0.04·-PIPAAm, and (d) 0.1·PIPAAm at 37 C at 0 and 30 min after low-temperature incubation at 18 C. The scale-bar represents 1 lm.
228
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229
Compared with SMCs on PIPAAm-1.5 h, the cells on 0.04·-PIPAAm contain a higher concentration of actin at the cell periphery and lamellipodium before cell deadhesion (Fig. 8c). Moreover, stress fibers are formed inside the cell cytoplasm on collagen-coated surfaces. After 30 min of de-adhesion, the actin remains concentrated in the cell periphery after cell contraction but the stress fibers in the cytoplasm disappear. At higher collagen concentration, cells adopt a more elongated shape (Fig. 8d), with a higher concentration of isotropically aligned stress fibers in the cytoplasm compared with that on 0.04·-PIPAAm. The high degree of initial polarization of stress fibers mentioned above is directly correlated with the slowest rate of de-adhesion on 0.1·-PIPAAm. Extensive transformation of actin cytoskeleton is found on 0.1· PIPAAm towards 30 min of cell de-adhesion compared with that on PIPAAm-1.5 h. A ligand-modified surface may not only encourage cells to maintain their differentiated state by simulating the in vivo microenvironment but also facilitate cell attachment and proliferation. It is known that smooth muscle cells are dispersed in connective tissue throughout the body, where they secrete an ECM that is rich in type I collagen and other proteins. For instance, collagen is a widely known ECM protein for the regulation of cellular behaviors such as adhesion, spreading, proliferation and migration, and thus has been used extensively to enhance cell–material interactions for both in vivo and in vitro applications [40]. Our result on 0.04·-PIPAAm agrees with recent examples of the faster detachment of whole cell sheets of human foreskin fibroblasts from the methylcellulose (MC)/PBS/collagen hydrogel matrix below the LCST compared with nonECM protein-coated gel [41]. 4. Conclusion In summary, we have successfully demonstrated that the polymerization time for PIPAAm grafting directly affects the initial rate of cell de-adhesion. First, it is shown that a homogeneous layer of PIPAAm film on glass is necessary for triggering cell de-adhesion. Initial de-adhesion kinetics is enhanced on the surface with the denser PIPAAm layer induced by higher polymerization time of NIPAAm. The effect of collagen coating on cell de-adhesion is highly dependent on the concentration of collagen dispersed on the PIPAAm surface. Actin staining confirms the cell viability and demonstrates that the difference in deadhesion kinetics on collagen-coated PIPAAm is related to cytoskeletal remodeling. Overall, our study shows the possibility of engineering cell de-adhesion kinetics with the change of physiochemical properties of PIPAAm. Acknowledgement The research was supported by Academic Research Fund.
References [1] Shiroyanagi Y, Yamato M, Yamazaki Y, Toma H, Okano T. Urothelium regeneration using viable cultured urothelial cell sheets grafted on demucosalized gastric flaps. BJU Int 2004;93:1069–75. [2] Kushida A, Yamato M, Konno C, Kikuchi A, Okano T. Temperature-responsive culture dishes allow non-enzymatic harvest of differentiated Madin–Darby canine kidney (MDCK) cell sheets. J Biomed Mater Res 2000;51:216–23. [3] Stile RA, Burghardt WR, Healy KE. Synthesis and characterization of injectable poly(N-isopropylacrylamide)-based hydrogels that support tissue formation in vitro. Macromolecules 1990;32:7370–9. [4] Wei H, Zhang XZ, Cheng C. Self-assembled, thermosensitive micelles of a star block copolymer based on PMMA and PNIPAAm for controlled drug delivery. Biomaterials 2007;28:99–107. [5] Gao CY, Chen B, Mohwald H. Thermosensitive poly(allylamine)-gpoly(N-isopropylacrylamide) copolymers: salt-tuned phase separation, particle formation and their applicability on curved surface. Colloid Surface A 2006;272:203–10. [6] Okano T, Yamada N, Sakai H, Sakurai Y. A novel recovery-system for cultured-cells using plasma-treated polystyrene dishes grafted with poly(N-isopropylacrylamide). J Biomed Mater Res 1993;27:1243–51. [7] Callewaert M, Grandfils C, Boulange-Petermann L, Rouxhet PG. Adsorption of poly(N-isopropylacrylamide) on glass substrata. J Colloid Interf Sci 2004;276:299–305. [8] Canavan HE, Cheng XH, Graham DJ, Ratner BD, Castner DG. Surface characterization of the extracellular matrix remaining after cell detachment from a thermoresponsive polymer. Langmuir 2005;21:1949–55. [9] Curti PS, Moura MRd, Veiga W, Radovanovic E, Rubira AF, Muniz EC. Characterization of PNIPAAm photografted on PET and PS surfaces. Appl Surf Sci 2005;245(1–4):223–33. [10] Lin Z, Xu T, Zhang L. Radiation-induced grafting of N-isopropylacrylamide onto the brominated poly(2,6-dimethyl-1,4-phenylene oxide) membranes. Radiat Phys Chem 2006;75(4):532–40. [11] Schmaljohann D, Beyerlein D, Nitschke M, Werner G. Thermoreversible swelling of thin hydrogel films immobilized by low-pressure plasma. Langmuir 2004;20:10107–14. [12] Huang J, Wang XL, Chen XZ, Yu XH. Temperature-sensitive membranes prepared by the plasma-induced graft polymerization of N-isopropylacrylamide into porous polyethylene membranes. J Appl Polym Sci 2003;89(12):3180–7. [13] Morikawa N, Matsuda T. Thermoresponsive artificial extracellular matrix: N-isopropylacrylamide-graft-copolymerized gelatin. J Biomater Sci-Polym Ed 2002;13:167–83. [14] Yarmush ML, Toner M, Dunn JC, Rotem A, Hubel A, Tompkis RG. Hepatic tissue engineering: development of critical technologies. Ann NY Acad Sci 1992;665:238–52. [15] Kim YS, Lim JY, Donahue HJ, Lowe TL. Thermoresponsive terpolymeric films applicable for osteoblastic cell growth and noninvasive cell sheet harvesting. Tissue Eng 2005;11:30–40. [16] Liu HC, Ito Y. Gradient micropattern immobilization of a thermoresponsive polymer to investigate its effect on cell behavior. J Biomed Mater Res Part A 2003;67A:1424–9. [17] Moran MT, Carroll WM, Selezneva I, Gorelov A, Rochev Y. Cell growth and detachment from protein-coated PNIPAAm-based copolymers. J Biomed Mater Res Part A. 2007;81(4):870–6. [18] Alberts B, Bray D, Lewis J, Raff M, Roberts K, Watson JD. Mol Biol Cell. 3rd edn. New York: Garland Publishing, Inc.; 1994. [19] Lee CH, Singla A, Lee Y. Biomedical applications of collagen. Int J Pharm 2001;221:1–22. [20] Gaudet C, Marganski WA, Kim S, Brown CT, Gunderia V, Dembo M, et al. Influence of type I collagen surface density on fibroblast spreading, motility, and contractility. Biophys J 2003;85:3329–35. [21] Yim EK, Reano RM, Pang SW, Yee AF, Chen CS, Leong KW. Nanopattern-induced changes in morphology and motility of smooth muscle cells. Biomaterials 2005;26:5405–13.
B. Chen et al. / Acta Biomaterialia 4 (2008) 218–229 [22] Zhu AP, Fang N, Chan-Park MB, Chan V. Adhesion contact dynamics of 3T3 fibroblasts on poly (lactide-co-glycolide) surface: regulation by chitosan and gelatin surface immobilization. Biomaterials 2006;27(12):2566–76. [23] Feng ZQ, Chian KS, Mhaisalka PS, Chan V, Ratner BD. The dual requirements of ECM protein and chitosan for inducing adhesion contact evolution of esophageal epithelia. J Biomed Materi Res Part A 2007;82A:788–801. [24] Chen BY, Dang JY, Tan TL, Fang N, Chen WN, Leong KW, et al. Dynamics of smooth muscle cell deadhesion from thermosensitive hydroxybutyl chitosan. Biomaterials 2007;28:1503–14. [25] Xu FJ, Zhong SP, Yung LYL. Thermoresponsive comb-shaped copolymer-Si(100) hybrids for accelerated temperature-dependent cell detachment. Biomaterials 2006;27(8):1236–45. [26] Xu FJ, Kang ET, Neoh KG. pH- and temperature-responsive hydrogels from crosslinked triblock copolymers prepared via consecutive atom transfer radical polymerizations. Biomaterials 2006;27(14):2787–97. [27] Liu KK, Wan KT. Contact mechanics of a thin-walled capsule adhered onto a rigid planar substrate. Med Biol Eng Comput 2001;39(5):605–8. [28] Nagayama K, Nagano Y, Sato M, Matsumoto T. Effect of actin filament distribution on tensile properties of smooth muscle cells obtained from rat thoracic aortas. J Biomech 2006;39(2):293–301. [29] Cheng X, Canavan HE, Stein MJ, Hull JR, Kweskin SJ, Wagner MS, et al. Surface chemical and mechanical properties of plasma-polymerized N-isopropylacrylamide. Langmuir 2005;21(17):7833–41. [30] Chee CK, Rimmer S, Soutar I, Swanson L. Fluorescence investigations of the thermally induced conformational transition of poly(Nisopropylacrylamide). Polymer 2001;42:5079–87. [31] Zhu PW, Napper DH. Longer time kinetics of collapse transition of polymer–surfactant complexes at interfaces near to collapse temperatures. Phys Rev E, Part B 2000;61(6):6866–71.
229
[32] Crowther HM, Vincent B. Swelling behavior of poly-N-isopropylacrylamide microgel particles in alcoholic solutions. Colloid Polym Sci 1998;276:46–51. [33] Gurdak E, Booth J, Roberts CJ. Influence of collagen denaturation on the nanoscale organization of adsorbed layers. J Colloid Interf Sci 2006;302(2):475–84. [34] Tan WJ, Teo GP, Liao K, Leong KW, Mao HQ, Chan V. Adhesion contact dynamics of primary hepatocytes on poly(ethylene terephthalate) surface. Biomaterials 2005;26(8):891–8. [35] Okano T, Yamada NO M, Sakai H, Sakurai Y. Mechanism of cell detachment from temperature-modulated, hydrophilic–hydrophobic polymer surfaces. Biomaterials 1995;16:297–303. [36] Yang J, Yamato M, Nishida K, Ohki T, Kanzaki M, Sekine H, et al. Cell delivery in regenerative medicine: the cell sheet engineering approach. J Control Release 2006;116(2):193–203. [37] Fultz ME, Li C, Geng W, Wright GL. Remodeling of the actin cytoskeleton in the contracting A7r5 smooth muscle cell. J Muscle Res Cell Motility 2000;21:775–87. [38] Burgstaller G, Gimona M. Actin cytoskeleton remodelling via local inhibition of contractility at discrete microdomains. J Cell Sci 2004;117:223–31. [39] Zhang DH, Sherwood J, Li L, Swartz DR. Unloaded shortening velocity in single permeabilized vascular smooth muscle cells is independent of microtubule status. J Muscle Res Cell Motility 2004;25:167–75. [40] Neff JA, Tresco PA, Caldwell KD. Surface modification for controlled studies of cell–ligand interactions. Biomaterials 2000;20:2377. [41] Chen CH, Tsai CC, Chen W, Mi FL, Liang HF, Chen SC, et al. Novel living cell sheet harvest system composed of thermoreversible methylcellulose hydrogels. Biomacromolecules 2006;7(3): 736–43.