Colloids and Surfaces B: Biointerfaces Microscale control over ...

Colloids and Surfaces B: Biointerfaces 67 (2008) 210–215Contents lists available at ScienceDirectColloids and Surfaces B: Biointerfacesjournal homepage: www.elsevier.com/locate/colsurfbMicroscale control over collagen gradient on poly(l-lactide) membranesurface for manipulating chondrocyte distributionHuaping Tan, Li Wan, Jindan Wu, Changyou Gao ∗Key Laboratory of Macromolecular Synthesis and Functionalization, Ministry of Education, and Department of Polymer Science and Engineering,Zhejiang University, Hangzhou 310027, ChinaarticleinfoabstractArticle history:Received 1 March 2008Received in revised form 26 July 2008Accepted 21 August 2008Available online 30 August 2008Keywords:Poly(l-lactide) membraneCollagen gradientAminolysisMicroinfusionChondrocytesA collagen gradient was constructed to interrogate cell adhesion on a poly(l-lactide) (PLLA) membranesurface. Utilizing a microinfusion pump, gradients of amino groups were generated on the PLLA surfaceby an aminolysis method. Immobilization of collagen onto the gradient surfaces was performed by glutaraldehyde(GA) coupling to form the collagen gradients. The –NH 2 and immobilized collagen densityprofiles on the PLLA membranes were quantitatively determined by ninhydrin and hydroproline (Hyp)analysis, respectively. By using fluorescein isothiocyanate (FITC) labeled collagen (FITC-Col), the profileof the as-prepared collagen gradient was directly monitored by a fluorescence microscope. The scanningforce microscopy and water contact angle studies revealed that the morphology and wettability of themodified membranes changed progressively as a function of position along the gradient surface. Rabbitauricular chondrocytes were cultured on the collagen gradient membranes to test their cellular response.The attachment and spreading behaviors of the chondrocytes were dependent on the surface collagendensity. These results indicate a surface on which the variation of collagen gradient strongly modifies thebiological response of chondrocytes.© 2008 Elsevier B.V. All rights reserved.1. IntroductionControl over the properties of biological interfaces is essentialfor developing novel biomedical materials and devices. Inparticular, precisely designed surface chemistries such as surfacechemical gradients possess great potentials for applications infields of tissue engineering and regenerative medicine [1,2].Surfaces with these gradients can regulate cell–surface interactionand control cellular behaviors such as attachment and migration,which are essential for understanding the intrinsic mechanismof cell–material interaction, and in turn instructive for designsof biomaterials with cell and tissue-inductive properties. It iswell known that a variety of cellular events are regulated by theinteraction between integrin in cell membranes and extracellularmatrix (ECM) [3]. Collagen is one of the main ECM proteins andcontains many active domains such as RGD peptide sequence [4,5].An artificial collagen gradient may then control cell adhesion andmigration, and materials bearing the collagen gradient can thusbe expected to have an ability to spatially induce and guide tissueregeneration. Therefore, the materials with these properties are inhigh demand for the restoration of damaged tissues.∗ Corresponding author. Tel.: +86 571 87951108; fax: +86 571 87951108.E-mail address: cygao@mail.hz.zj.cn (C. Gao).Some methods have been developed to immobilizebiomolecules either physically or chemically on the materialsurfaces to create the surface chemistry gradients [1]. Typicalexamples of physical adsorption include controlled diffusion [6–8],microfluidic mixing [4,9–13], deposition of protein–nanoparticleconjugate [14], and self-assembled monolayers (SAMs) [15–17].These physical methods are generally easier to handle, yet thegradient stability is susceptible to environmental conditions andmay undergo changes along with the time. On the contrary, thegradients created by covalent chemical conjunction are morestable. These methods include microstamping [18], corona dischargetreatment [19], photochemical immobilization [1,20,21]and chemical grafting [22–24]. Most of the aforementioned methods,however, were not deliberately developed for biomaterials,especially not for biomaterials used for regenerative medicine.Thus various limitations are inevitable in terms of the creationof micron scale gradients, controllability and gradient quality,shape and size of the materials, biocompatibility, and suitabilityfor biodegradable polymers.In this work, a method to design a collagen gradient on abiodegradable film is developed. Utilizing a microinfusion pump, agradient of amino groups is generated on a biodegradable polymerof PLLA membrane by an aminolysis method [25–27], followedby the immobilization of collagen onto the surface to form acollagen gradient by GA coupling. Rabbit auricular chondrocytes0927-7765/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.colsurfb.2008.08.019

212 H. Tan et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 210–215Scheme 1. Schematic illustration of fabricating process of collagen gradient on a PLLA membrane. The hexanediamine/propanol solution is continuously injected into atube, in which a PLLA membrane is vertically fixed. The aminolysis degree is increased with the contacting time, thus creates the –NH 2 gradient, by which covalent collagengradient is obtained by GA coupling.First, for the convenience of gradient confirmation and theoreticalanalysis, a series of PLLA membranes were homogenouslyaminolyzed for 2, 4, 5, 6, 8 and 10 min, respectively. These timepoints correspond to positions of ∼97, ∼194, ∼242, ∼291, ∼388and ∼484 m fromthetop(0m) of the PLLA membrane, respectively,supposing a membrane is gradually aminolyzed with aninfusion rate of 0.4 ml/h. The free NH 2 density on the homogeneouslyaminolyzed membrane is shown in Fig. 1a. The NH 2 densitywas indeed increased gradually, displaying sigmoidal regime, i.e.slow in the initial and final stages while fast in between. Scanningforce microscopy images (SFM Fig. 1a, inset) reveal that fluctuationexisted on all the membrane surfaces, regardless of theaminolyzing time. However, smaller roughness (100 nm) over anarea of 400 m 2 was recorded on the 5 min aminolyzed PLLA membranethan that of the control (400 nm) and the 8 min (200 nm) or10 min (400 nm) aminolyzed ones. It is known that the aminolysisFig. 1. (a) –NH 2 density, (b) collagen density, (c) water contact angle, and (d) chondrocyte density as a function of aminolyzing time. The PLLA membranes were homogeneouslyaminolyzed in 0.06 g/ml 1,6-hexanediamine/propanol solution at 50 ◦ C for different time (a), which were then immobilized with collagen (b) and used for cell culture. Thesolid line in (a) was a sigmoidal simulation using Eq. (1). Insets are SFM images in (a) and (b) (numbers are heights in z-direction) and CLSM images in (d) (scale bar 20 m)with the pointed out aminolyzing time. Cells were cultured for 24 h with a seeding density of 10 × 10 4 /ml.

H. Tan et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 210–215 213breaks the ester bonds and produces aminohexyl amides(–CONH(CH 6 )NH 2 ) and –OH groups [25,26] on termini of thechains, leading to a dissolution or a dissociation of the oligomericfragments when their size is small enough. Therefore, the final platformshould be the result of equilibrium between the loss and thegeneration rates. By contrast, the initial slow increasing rate impliesthat the aminolysis requires an induction period.Fig. 1b reveals that the immobilized collagen amount followeda similar sigmoidal regime as well, and matched perfectly with the–NH 2 density. The original fluctuation was almost preserved, butthe overall surface roughness was decreased (Fig. 1b inset). Thegradual alteration of the surface chemistry of both the aminolyzedand collagen immobilized membranes was further confirmed bywater contact angle (Fig. 1c). The steady decrease of the water contactangle illustrates that the surface became more hydrophilic afterlonger aminolysis. After collagen immobilization, still a smallerwater contact angle was measured. The linear decrease profiles,which are different from that of their surface NH 2 or collagen density,were caused by the interplay of the surface wettability and themorphology. In vitro chondrocyte culture on the PLLA membranesrevealed that the cell number exactly followed a same regime asthe collagen density (Fig. 1d). Moreover, more spreading morphologywas observed on the membrane with higher collagen density(Fig. 1d inset). All these results ensure that control over the chondrocytedistribution and morphology by the collagen gradient ispossible.The –NH 2 density (C NH2 ) profile was quantitatively analyzedbefore fabrication of the continuous collagen gradient. Mathematically,the “S” shaped profile can be described by a well-knownlogistic function [32]:C NH2 (t) = C max × (1 + e −k(t−tm) ) −1 (1)where t is the aminolysis time, t m (a constant) is the time atwhich the maximum “apparent rate” of aminolysis reaches, k isa factor determining the curvature of the profile, and C max (a constant)is the maximum NH 2 density on the membrane [33]. A LeastSquares Fitting was applied to give the values of C max , k and t m as1.77 × 10 −7 mol/cm 2 , 1.08 min −1 and 5.5 min, respectively. The simulatedline matched perfectly with the experimental data (Fig. 1a),with a relative standard deviation of 2.5%.Since l = vt/r 2 (here, l is defined as the distance from one placeof the membrane to liquid level, 0) for a gradient film, on the prerequisitethat the same reaction was performed, Eq. (1) can betransformed into:C NH2 (l) = 1.77 × (1 + e −1.08(r2 (l/v)−5.5) ) −1 (2)Thus, the –NH 2 density at a given position on the gradient film canbe obtained. Moreover, the total –NH 2 content (Q NH2 ) on a gradientsurface can be obtained by the integral of Eq. (2):∫ lQ NH2 | = C NH2 dl = 1.64vr02 (ln(e1.08(r2 (l/v)−5.5) + 1) − 0.0027)(3)Atafixedr (0.662 cm), infusion time (10 min) and an infusion rateof 1v (8.3 ml/h), 2v or 4v, the Q NH2 is calculated by Eq. (3) and shownin Fig. 2 (the solid lines). Experimental results matched well withthe theoretical calculation with relative standard deviations of 7.7%,2.4% and 1.6% for 1v, 2v and 4v, respectively. The correctness andapplicability of these equations on the –NH 2 density on the gradientmembranes were thus confirmed.According to l = vt/r 2 , 10 min aminolysis with injection ratesof 1v, 2v and 4v gave the l lengths of 1, 2 and 4 cm, respectively.From a simple calculation of the data shown in Fig. 2a byEq.(3),Fig. 2. (a) Total –NH 2 and (c) collagen amounts on a PLLA gradient surface as a function of distance, respectively. (b) –NH 2 and (d) collagen densities calculated from (a)and (c), respectively as a function of position. To illustrate the controllability, three infusion rates were applied. The solid lines in (a) were drawn by Eq. (3). Here the wholegradient surfaces were regarded as five segments with same length. The total –NH 2 content of every segment could be obtained from experimental results (a), then theaverage –NH 2 density of every segment was calculated, as shown in b. Position in b was the midpoint of every segment. (d) Was calculated from c similarly.

214 H. Tan et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 210–215Fig. 3. (a) A fluorescence microscopic image of a collagen gradient created on a PLLA membrane. (b) A line profile recorded from (a) to illustrate the collagen distribution.FITC conjugated collagen was used. (c) A photograph of the water droplets sitting on a collagen gradient at different position. (d) A fluorescent image to show the chondrocytedistribution and morphology controlled by the collagen gradient. The image was taken after cell seeding for 24 h with a seeding density of 10 × 10 4 /ml. (e) Integratedfluorescence intensity from (d) within a distance of 50 m as a function of position. The solid line is a sigmoidal simulation to the data. The gradient membrane with a sizeof ∼500 m was prepared using an infusion rate of 0.4 ml/h at a fixed tube radius (0.662 cm) and infusion time (10 min).the formed –NH 2 gradients can be clearly visualized in Fig. 2b. Theyfollow the exact same profiles shown in Fig. 1a. Immobilizationof collagen on the aminolyzed membranes of Fig. 2a resulted inthe same collagen content profiles (Fig. 2c), from which the sameprofiles of collagen gradients (Fig. 2d) as that of the –NH 2 gradientswere also derived.The obtained collagen gradient was further confirmed by directobservation via CLSM by using FITC-Col. The successive enhancementof the fluorescence intensity (Fig. 3a) demonstrates clearlythat a continuous collagen gradient has been formed. A line profileacross the fluorescent image (Fig. 3b) also has a sigmoidal shape asthat shown in Fig. 2d. To investigate the surface wettability, a collagengradient membrane with a size of ∼14.5 mm was preparedusing an infusion rate of ∼12 ml/h. Fig. 3c shows that the water contactangle was decreased gradually along the gradient direction. Inconclusion, by a continuous infusion technique, a stable collagengradient with a controllable size can be conveniently fabricated.Chondrocytes were seeded onto the collagen gradient membrane.24 h later, a fluorescent microscopy image was taken afterstaining the viable cells with FDA (Fig. 3d). At the region of a lowercollagen density, few dot-like cells were observed. As the collagendensity increased, more cells adhered and their morphologybecame more spreading. Many cell clusters were also formed atthe higher collagen region. Integrated fluorescence intensity of cellsfrom Fig. 3d within a distance of 50 m gave a similar profile as thatof the collagen gradient (Fig. 3e). These results demonstrated thata collagen gradient can effectively manipulate the chondrocytesattachment and spreading, which is of great significance for tissueinduction.4. ConclusionsIn conclusion, a stable collagen gradient on biodegradablepolymers, illustrated with PLLA, is successfully produced bymicroinfusion of a diamine solution, following by GA coupling. Formationof the –NH 2 gradient is first simulated and confirmed byhomogeneous aminolysis for desired time. The “S” shaped increaseof –NH 2 density can be described by the well-known logistic function[32], which can predict the density and total content of –NH 2and match perfectly with the real –NH 2 gradient created on aPLLA membrane. The immobilized collagen density followed theexact profile of –NH 2 density. The water contact angle on thecollagen gradient was steadily decreased along with the collagendensity increase. The collagen gradient effectively manipulatedchondrocyte attachment and spreading, e.g. the number of culturedchondrocytes and their degree of spreading were increasedwith the increase of collagen density. Besides this stable feature,the length of the gradient can be conveniently modulated within amicron scale by the experimental parameters. Moreover, the techniquecan be extended to other biofunctional species and polymersubstrates with variable shape and microstructure. For example,further application of this method on 3D materials may result intissue inductive scaffolds used for tissue regeneration.AcknowledgmentsThis study is financially supported by the Major StateBasic Research Program of China (2005CB623902), the NationalHigh-tech Research and Development Program (2006AA03Z442),the Science and Technology Program of Zhejiang Province(2006C13022), and the National Science Fund for DistinguishedYoung Scholars of China (No. 50425311).References[1] B. Li, Y.X. Ma, S. Wang, P.M. Moran, Biomaterials 26 (2005) 1487.[2] H.J. Song, G.L. Ming, M.M. Poo, Nature 388 (1997) 275.

H. Tan et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 210–215 215[3] B.P. Harris, J.K. Kutty, E.W. Fritz, C.K. Webb, K.J.L. Burg, A.T. Metters, Langmuir22 (2006) 4467.[4] N.L. Jeon, S.K.W. Dertinger, D.T. Chiu, I.S. Choi, A.D. Stroock, G.M. Whitesides,Langmuir 16 (2000) 8311.[5] S.L. Rowe, J.P. Stegemann, Biomacromolecules 7 (2006) 2942.[6] X. Cao, M.S. Shoichet, Neuroscience 103 (2001) 831.[7] X. Zhao, S. Jain, H.B. Larman, S. Gonzalez, D.J. Irvine, Biomaterials 26 (2005)5048.[8] K. Mougin, A.S. Ham, M.B. Lawrence, E.J. Fernandez, A.C. Hillier, Langmuir 21(2005) 4809.[9] J.A. Burdick, A. Khademhosseini, R. Langer, Langmuir 20 (2004) 5153.[10] K.A. Fosser, R.G. Nuzzo, Anal. Chem. 75 (2003) 5775.[11] I.Y. Tsai, M. Kimura, T.P. Russell, Langmuir 20 (2004) 5952.[12] N. Zaari, P. Rajagopalan, S.K. Kim, A.J. Engler, J.Y. Wong, Adv. Mater. 16 (2004)2133.[13] R.C. Gunawan, J. Silvestre, H.R. Gaskins, P.J.A. Kenis, D.E. Leckband, Langmuir 22(2006) 4250.[14] S. Krämer, H. Xie, J. Gaff, J.R. Williamson, A.G. Tkachenko, N. Nouri, D.A. Feldheim,D.L. Feldheim, J. Am. Chem. Soc. 126 (2004) 5388.[15] J.T. Smith, J.K. Tomfohr, M.C. Wells, T.P. Beebe, T.B. Kepler, W.M. Reichert, Langmuir20 (2004) 8279.[16] X. Yu, Z. Wang, Y. Jiang, X. Zhang, Langmuir 22 (2006) 4483.[17] N.V. Venkataraman, S. Zürcher, N.D. Spencer, Langmuir 22 (2006) 4184.[18] S.M. Bhangale, V. Tjong, L. Wu, N. Yakovlev, P.M. Moran, Adv. Mater. 17 (2005)809.[19] J.H. Lee, H.B. Lee, J. Biomed. Mater. Res. 41 (1998) 304.[20] J. Higashi, Y. Nakayama, R.E. Marchant, T. Matsuda, Langmuir 15 (1999) 2080.[21] B.P. Harris, A.T. Metters, Macromolecules 39 (2006) 2764.[22] R.R. Bhat, B.N. Chaney, J. Rowley, A. Liebmann-Vinson, J. Genzer, Adv. Mater. 17(2005) 2802.[23] Y. Mei, T. Wu, C. Xu, K.J. Langenbach, J.T. Elliott, B.D. Vogt, K.L. Beers, E.J. Amis,N.R. Washburn, Langmuir 21 (2005) 12309.[24] S.A. DeLong, J.J. Moon, J.L. West, Biomaterials 26 (2005) 3227.[25] Y.B. Zhu, C.Y. Gao, Y.X. Liu, T. He, J.C. Shen, Tissue Eng. 10 (2004) 53.[26] Y.B. Zhu, C.Y. Gao, X.Y. Liu, J.C. Shen, Biomacromolecules 3 (2002) 1312.[27] L.Y. Santiago, R.W. Nowak, J.P. Rubin, K.G. Marra, Biomaterials 27 (2006) 2962.[28] A. Schindler, D. Harper, J. Polym. Sci. Polym. Chem. Ed. 17 (1979) 2593.[29] J.S. Pieper, A. Oosterhoff, P.J. Dijkstra, J.H. Veerkamp, T.H.V. Kuppevelt, Biomaterials20 (1999) 847.[30] L. Ma, C.Y. Gao, Z.W. Mao, J. Zhou, J.C. Shen, X.Q. Hu, C.M. Han, Biomaterials 24(2003) 4833.[31] Y.H. Gong, Q.L. Zhou, C.Y. Gao, J.C. Shen, Acta Biomater. 3 (2007) 531.[32] P.F. Verhulst, Correspondence Mathematiques et Physiques 10 (1838) 113.[33] X.Y. Yin, J. Goudriaan, E.A. Lantinga, J. Vos, H.J. Spiertz, Ann. Bot. 91 (2003) 361.