Studies on nerve cell affinity of chitosan-derived materials

Gong Haipeng, Zhong Yinghui, Li Jianchun, Gong Yandao, Zhao Nanming, Zhang XiufangDepartment of Biological Sciences and Biotechnology, State Key Lab of Biomembrane and Membrane Biotechnology,Tsinghua University, Beijing 100084, China

Received 23 August 1999; accepted 18 February 2000

Abstract: Reparation of the central nervous system (CNS)is important because when it is impaired its recovery is dif-ficult and concomitant malfunction of other parts of bodyoccurs. In our previous studies, chitosan was found to be agood material supporting nerve repair. The purpose of thisarticle was to study the ability of chitosan and some chito-san-derived materials to facilitate the growth of nerve cells.Those materials were chitosan, glutaraldehyde-crosslinkedchitosan, glutaraldehyde-crosslinked chitosan-gelatin conju-gate, a chitosan-gelatin mixture, chitosan coated with poly-lysine (CAP), and a chitosan-polylysine mixture (CPL).Gelatin and polylysine were used as controls. After nervecells (gliosarcoma cells and normal cerebral cells) weregrown on those materials, their attachment, spread, andgrowth were observed. The adsorption of some extracellularmatrix molecules such as laminin and fibronectin on the

materials and the role the molecules play in nerve cell at-tachment and spreading were also studied by enzyme-linked immunosorbent assay and MTT method. We foundthat both CAP and CPL have excellent nerve cell affinity,defined as the ability to promote nerve cell to grow andfunction normally. Those two materials may be promisingfor the repair of the nervous system. Materials precoatedwith laminin, fibronectin, and serum were analyzed for theirnerve cell affinity. Results suggest that after being precoatedwith laminin and fibronectin solution or serum, all materialhave better nerve cell affinity. © 2000 John Wiley & Sons,Inc. J Biomed Mater Res, 52, 285–295, 2000.

Key words: chitosan; biocompatibility; nerve cell affinity;neuron; protein adsorption

INTRODUCTION

More than any other form of trauma, nerve injuriescomplicate successful rehabilitation because matureneurons (like many other cells in the body) do notreplicate; that is, they do not undergo cell division.1

Once the nervous system is impaired, its recovery isdifficult and malfunction of other parts of the bodyoccurs. The central nervous system (CNS), consistingof brain and spinal cord, is important to the normalfunction of the whole body. Its repair has become thefocus of a popular medical topic. A lot of resorbablebiomaterials are used in the repair of nervous system,such as polyglycolic acid, polylactic acid, polyester,copolymers of L-lactide, and «-caprolactone.1–4 How-ever, chitin and chitosan are seldom studied for nerverepair although they have good biocompatibilities.Chitin is a natural polysaccharide found particularly

in the shell of crustaceans, the cuticles of insects, andthe cell walls of fungi. It has b(1–4)-linked N-acetyl-D-glucosamine repeat units and is the second most abun-dant form of polymerized carbon found in nature.5, 6

Chitosan, the fully or partially deacetylated form ofchitin, is the most extensively used material derivedfrom chitin. It has b(1–4)-linked D-glucosamine repeatunits and has been shown useful as chelating agents,drug carriers, membranes, water treatment additives,biodegradable pressure-sensitive adhesive tapes,wound-healing agents, and in a number of other im-portant applications.7,8 Chitosan is reported to pos-sess excellent biocompatibility. Although little atten-tion has been paid to its function in nerve repair, ourprevious study showed that neurons cultured on thechitosan membrane can grow well and that chitosanconduit can greatly promote the repair of the periph-eral nervous system (PNS).9 Studies conducted byDongxu et al. also showed good nerve affinity of chi-tosan.10

The goals of our experiment were to devise somechitosan-based materials promising for nerve repairand to evaluate their ability to support the attachment,spread, and growth of nerve cells in vitro (defined asnerv e cell affinity). Chitosan (Chi) and some of its de-

Correspondence to: Z. Xiufang; e-mail: zxf-dbs@mail.tsinghua.edu.cn

Contract grant sponsor: National Research and Develop-ment Projects of High Technology; Contract grant number:819-07-01

© 2000 John Wiley & Sons, Inc.

rived materials, including glutaraldehyde-crosslinkedchitosan (GC), glutaraldehyde-crosslinked chitosan-gelatin conjugate (CG), a chitosan-gelatin mixture(CAG), chitosan coated with polylysine (CAP), and achitosan-polylysine mixture (CPL), were prepared,with gelatin, (Gel) and polylysine (PL) used as con-trols. GC was selected because the intra- and intermo-lecular crosslinking formed via glutaraldehyde is ex-pected to improve the mechanical strength of chitosan.The gelatin-related materials (CG and CAG) and thepolylysine-related materials (CAP and CPL) weremade because of the excellent biocompatibility of gela-tin and polylysine. The fetal mouse cerebral cortex(FMCC) cells were cultured on all the materials andobserved for 6 days. The result of this experiment wasthe main criterion to evaluate the nerve cell affinity ofmaterials. Despite normal nerve cells, fetal mouse ce-rebral cortex (FMCC) cells include several kinds ofcells, with neurons and glial cells being the main com-ponents, and it is hard to analyze them quantitatively.Hence, gliosarcoma cells (9L), the monoclonal cancerglial cells, were selected to assist in evaluation of thenerve cell affinity of materials.

According to previous studies, the rate of growth,proliferation, and differentiation of cells on a materialdepend on the successful initial attachment andspread of the cells on the surface of the substratum.11

There are several factors affecting cell attachment andspread, the first of which is the characteristics of thematerials. Using a series of different culture sub-strates, previous studies have shown that initial celladhesion and spread depend on surface physicochem-ical properties such as wettability and charge; for ex-ample, the adhesion and spread of neurons were ob-served to be better on hydrophilic surfaces than onhydrophobic ones.12 Another factor is the extracellularmatrix (ECM) molecules such as laminin and fibronec-tin. Those molecules contain specific amino acid se-quences that bind to cell surface integrin receptors,thereby influencing cell behavior and gene expres-sion.13 Fibronectin is thought to be involved in a widevariety of cellular activities.13 The simplest unifyinginterpretation of these findings was that the fibronec-tin function as adhesive proteins binding cells to othercells or to the substratum. This conclusion is based onexperiments in vitro using a model system of cellularadhesion, in which purified fibronectin was found toincrease the adhesion of cells to cells or to the substra-tum on which they are grown.13 Serum factors arerequired for the attachment of cells to collagen-coateddishes or for cells to attach and spread on tissue cul-ture dishes. These cell-attachment and cell-spreadingfactors are supposed to be connected with serum fi-bronectin which can bind directly to collagen or totissue cultural dishes in the absence of cells.13 Lamininis typically known to be an extremely large moleculecomposed of several different subunits that can be as-

sociated in different combinations with distinct forms.In the developing and maturing of the central nervoussystem (CNS), laminin plays a crucial role, for ex-ample, in cell migration, differentiation, and axonalgrowth.14,15 To study the mechanism of the interactionbetween materials and nerve cells and understand theresults of the nerve cell culture and MTT experiment,material properties such as water contact angle andthe surface charge of all eight materials were mea-sured or analyzed, and the adsorption amounts of theECM molecules (laminin and fibronectin) on the dif-ferent membranes were compared by enzyme-linkedimmunosorbent assay (ELISA). In addition, thegrowth and morphology of neurons on the materialsprecoated with laminin, fibronectin, and serum werealso observed, to explore the improvement in nervecell affinity of materials after adsorption of these ad-hesive proteins.

MATERIALS AND METHODS

Preparation of materials

Chitosan-derived materials were prepared with Gel andPL used as controls. The three basic materials are chitosan(degree of deacetylation > 85%; Sigma), gelatin (BBL), andpolylysine (Sigma). The Chi solution was prepared as fol-lows: 1 g chitosan powder was added to 0.2 N acetic acidsolution. The mixture was stirred until most chitosan wasdissolved. Then, it was filtered to remove the undissolvedsubstance. The final concentration of chitosan was 1%. Tomake the GC, 2.5% glutaraldehyde was added into 1% chi-tosan solution until its final concentration reached 0.025%.To prepare the CG, 100 mL of 1% chitosan solution wasmixed with 10 mL 10% gelatin. Then, 2.5% glutaraldehydewas added until its final concentration reached 0.00125%.The CAG was obtained by a simple process: mixing 100 mLof 1% chitosan solution with 10 mL of 10% gelatin. Theconcentrations of the Gel and PL solutions were 0.2% and 0.1mg/mL respectively. We mixed 60 mL of 1% Chi solutionwith 40 mL of 0.1 mg/mL PL solution to produce CPL.

Preparation of membranes

The Chi, GC, CG, CAG, Gel, and CPL membranes weremade with the same method. The solutions of the corre-sponding materials were added to the wells of tissue culturecluster or ELISA cluster, dried under 37°C, and washed with80% ethanol until the membranes were neutral (pH ≈7). Inthe process of CAP and PL membrane preparation, wellswith and without Chi membrane were incubated with 0.1mg/mL polylysine for 5 min. After the supernatant wasremoved, both membranes were washed three times withdistilled water. Finally, all eight kinds of membranes were

286 HAIPENG ET AL.

balanced with high-glucose Dulbecco’s modified Eagle’smedium (DMEM; Gibco) for 2 h.

Measurement of wettability of membranes

The static contact angles of all membranes were measuredusing a contact angle goniometer (Model JY-82; ChengdeExperimental Machine Plant, China). Deionized distilledwater was dropped onto the surface of the membrane beforemeasuring.

Preculture of 9L cells

The gliosarcoma cells (9L), obtained from Capital MedicalUniversity, were cultured in DMEM containing 50 U/mLpenicillin and streptomycin and 10% fetal bovine serum(FBS; Hyclone) under 5% CO2 and 95% air at 37°C in thetissue culture flask (25 cm2; Costar).

Laminin, fibronectin, and serum solutions

Because structures of fibronectin are conservative amongall mammals,16 we used human fibronectin and human se-rum. The laminin used was extracted from mice. The humanserum (Beijing North TZ-Biotech Development Co.) was di-luted to the concentration of 5%. Fibronectin and laminin(Beijing Medical University) were diluted to 20 mg/mL, aconcentration comparable to that of fibronectin in 5% se-rum.17

Laminin and fibronectin adsorption (ELISA)

The adsorption of fibronectin and laminin was detectedby means of ELISA. First, diluted laminin, fibronectin, orserum was poured into a 96-well ELISA cluster (Costar)coated with all eight membranes and incubated at 37°C for1 h. After we removed the supernatant, each well was rinsedwith washing buffer (0.02M Tris-HCl buffer plus 0.05%Tween-20) three times for 15 min and the surfaces wereblocked with 0.5% egg albumin (Sigma) solution for 2 h at37°C. After rinsing, the antibody (rabbit anti-human fibro-nectin or laminin; Beijing Medical University) was addedand the tissue culture cluster was incubated at 37°C for 2 h.The rinsing process was repeated; then, horseradish perox-ide (HRP)-conjugated secondary antibody, goat anti-rabbitimmunoglobulin 6 (Jackson ImmunoResearch Laboratories),was incubated on each surface for 1.5 h. After being rinsed,the wells were added with O-phenylenediamine (Sigma) so-lution (including H2O2) and placed in dark for 10 min. Fi-nally, 2M H2SO4 was added to stop the reaction and theamount of adsorption was measured at 490 mm by an ELISAreader (Model 550; BioRad).

MTT experiment

The 9L cells were detached from tissue culture flask with0.25% trypsin in D-Hank’s solution. After the cell suspensionwas centrifuged, the supernatant was removed and mediumwas added to suspend the cells again. The cells werecounted by a hemacytometer and the cell suspension wasdiluted with fresh medium to the final concentration of 5 ×104 cells/mL. Then, the diluted suspension was added to a96-well tissue culture cluster with 200 mL/well. Controlwells included 200 mL medium. The cluster was incubated at37°C and 5% CO2 for about 20 h. We then added 20 mL ofMTT solution (5 mg/mL) to each well and incubated thecluster for another 4 h. Finally, supernatant was removedand 150 mL dimethylsulfoxide was added to each well todissolve the blue substance returned from MTT by the suc-cinate dehydrogenase of cells. The absorbency of light at 490nm was measured by an ELISA reader (BioRad).

Culture of FMCC cells on membranes

The eight kinds of membranes were made on the surfaceof a 24-well tissue culture cluster (Costar). A pregnantWistar mouse was killed by ethyl ether and fetal mice weretaken out. The cerebral cortex tissue of each fetal mouse wastaken out, washed with D-Hank’s solution supplementedwith glucose and sucrose, and sheared into fragments. Weadded 0.25% trypsin solutions to these tissue fragments andallowed it to act on them for 30 min under 37°C. After weremoved the supernatant, we added medium [high-glucoseDMEM (Gibco) containing 50 U/mL penicillin and strepto-mycin, 0.6 mg/mL insulin (Sigma), and 10% fetal bovineserum (Hyclone)]. After pipetting these tissue fragments,they were filtrated to obtain the cell suspension. After count-ing the cells, the suspension was diluted with medium to aconcentration of 5 × 105 cells/mL. We poured 1 mL of thediluted cell suspension on membranes in each well. The cellswere incubated at 37°C and 5% CO2. One day later, thesupernatant was removed and D-Hank’s solution was addedto rinse the membrane. Fresh medium was added to incu-bate again for 2 days. Then, the medium was discarded andnew medium with 50 mg/mL cytosine b-D-arabino-furanoside was added and incubated for another 2 days toinhibit the growth of glial cells. Finally, the medium waschanged to that without cytosine b-D-arabino-furanoside.

Culture of FMCC cells on membranes precoatedwith proteins

The laminin, fibronectin, and serum were diluted with0.01M PBS to the concentration of 20 mg/mL, 20 mg/mL, and5% respectively. Then those solutions were incubated withall eight materials at 37°C for 1 h. Finally, FMCC cell sus-pension was poured on the precoated materials and cul-tured by the means mentioned above.

287NERVE CELL AFFINITY OF CHITOSAN-DERIVED MATERIALS

RESULTS

Wettability of and surface charges of membranes

Static water contact angles on the membranes andtissue culture dish are shown in Figure 1. All eightmembranes had contact angles <90°. The lower con-tact angle indicated higher hydrophilicity. As a con-sequence, Gel and PL were the two most hydrophilicmaterials. Chi and GC had moderate levels of wetta-bility, whereas CAP and CPL were relatively hydro-phobic, like the tissue culture dish. CG and CAG werethe most hydrophobic of all materials.

Adsorption amount of ECM moleculeson membranes

Results are shown in Figures 2–4. In laminin adsorp-tion (Fig. 2), PL had the highest A490, much higherthan the other materials. CPL ranked second; Chi, GC,and CAP (A490 of the three materials was not signifi-cantly different) had moderate values; CG and CAGranked next; and Gel, a frequently used biomaterial,had the lowest A490. In fibronectin adsorption (Figs. 3and 4), the final absorbencies in the fibronectin solu-tion experiment and the serum experiment had a simi-lar order among different materials, although therewere a lot of other components in the serum. Gel andPL had the highest light absorbency in both the fibro-nectin solution and serum, and the other materials allshowed similar final values.

The A490 in the adsorption of fibronectin in fibro-nectin solution and serum on the same material cannotbe compared, for two reasons: (a) Although the fibro-nectin solution is diluted to the concentration compa-rable to that of fibronectin in 5% serum, the exact con-centrations of fibronectin in the solution and serum

are not equal; (b) other proteins in the serum competewith fibronectin in adsorption onto the membranes.

MTT experiment

Results of the MTT experiment are shown in Figures5 and 6. Because cells on PL-related materials attachand spread much faster than those on Gel-related ma-terials, those two series of materials were tested inseparate experiments with both Chi and GC as con-trols. Cells on Gel-related materials (Fig. 5) were cul-tured for more than 20 h, whereas cells on PL-relatedmaterials (Fig. 6) were cultured for only 18 h. PL hadthe highest value of A490, with CAP and CPL rankingsecond, and the absorbency values of those three ma-terials were far higher than those of others. Althoughmuch lower than those of PL-derived materials, thevalues of Chi and GC were significantly higher thanthose of CG and Gel (CAG’s value was not signifi-cantly different from that of Chi). Gel had the lowestvalue, whereas CG and CAG had moderate valuesbetween those of Chi and Gel. After the 9L cells were

Figure 3. Adsorption of fibronectin in fibronectin solutionon all materials. n = 6 surfaces. *p < .05 relative to Chi.

Figure 1. Water contact angle of all eight materials andtissue culture dish (TCD). n > 6 surfaces. *p < .05 relative toChi.

Figure 2. Adsorption of laminin in laminin solution on allmaterials. n = 6 surfaces. *p < .05 relative to Chi.

288 HAIPENG ET AL.

cultured on the materials, we observed their shapechanges with the passage of time. After 3 h cells on PLchanged shape from spherical to flat, revealing attach-ment. Many cells even protruded pseudopodia, whichmeant that spreading had begun. When cells were cul-tured for 18 h, cells on PL all had polygonal flatshapes, whereas most cells on CPL and CAP had be-gun to spread. After 20 h, small amounts of cells beganto spread on Chi and Gel. Cell on CG, CAG, and GCdid not change their spherical shapes during the 1-dayperiod.

Culture of neurons on materials

Figures 7 and 8 show neurons cultured for 1 and 6days, respectively, on all kinds of materials. The pho-tographs of 1-day culture reflect the status of nervecell attachment and spreading, whereas photographsof 6-day culture reflect the conditions of nerve cellgrowth. After 1 day, most cells on Chi, CAP, and CPLchanged from spherical to flat. This meant that mostcells had finished attachment. Many cells presenting a

bipolar shape like normally growing neurons were inthe process of spreading. Among those three materi-als, CAP and CPL had a much higher ratio of bipolar-shaped cells than Chi, indicating their greater abilityto promote neuron spreading than Chi. The number offlat cells on Gel-derived materials (CG, CAG, and Gel)was even less than that on Chi, CAP, and CPL, show-ing that nerve cell attachment was less efficient onGel-derived materials. Although some glial cells fin-ished spreading (in polygonal shapes) on CG, CAG,and Gel, most cells on CG, CAG and Gel had notbegun spreading and remained round. Cells on GCwere all round and the cell number was large. FMCCcells on PL did not show as well as expected, becausetheir number was lower than Chi, CAP, and CPL.

After being cultured for 6 days, cells on all materialsexcept GC grew well. The typical shape of neuronsand the axons of neurons could be seen clearly; someaxons even began to cluster into fascicules. On GC,cells remained round after 6-day culture.

Culture of neurons on materials precoatedwith proteins

Photographs of neurons cultured on Chi precoatedwith laminin, fibronectin, and serum for 1 day areshown in Figure 9. Other photographs of cells on othermaterials precoated with proteins are not shown.

From results of 1 day, laminin, serum, and fibronec-tin all greatly promoted neuron attachment andspread. On the materials precoated with laminin, thisphenomenon was the most remarkable. Materials ad-sorbed with laminin had a far higher cell number thancorresponding bare materials. All cells on Chi, CAP,and CPL had begun to spread. Even some cells on GCbegan to change their shapes. However, cells on CG,CAG, and Gel did not change their shape greatly. Cellspreading on materials precoated with serum was not

Figure 4. Adsorption of fibronectin in 5% serum on allmaterials. n = 6 surfaces. *p < .05 relative to Chi.

Figure 5. Growth of gliosarcoma cells on chitosan- andgelatin-derived materials. n = 6 surfaces. *p < .05 relative toChi.

Figure 6. Growth of gliosarcoma cells on chitosan- andpolylysine-derived materials. n = 6 surfaces *p < .05 relativeto Chi.

289NERVE CELL AFFINITY OF CHITOSAN-DERIVED MATERIALS

Figure 7. Photographs of FMCC cells cultured for 1 day on (a) Chi, (b) GC, (c) CG, (d) CAG, (e) Gel, (f) PL, (g) CAP, and(h) CPL.

290 HAIPENG ET AL.

Figure 8. Photographs of FMCC cells cultured for 6 days on (a) Chi, (b) GC, (c) CG, (d) CAG, (e) Gel, (f) PL, (g) CAP, and(h) CPL.

291NERVE CELL AFFINITY OF CHITOSAN-DERIVED MATERIALS

as great as that on materials precoated with laminin,but it was better that that on materials precoated withfibronectin. There were similar tendencies of cell ad-hesion and growth on all materials precoated withfibronectin and serum, respectively. When CG, CAG,and Gel were precoated with fibronectin or serum,some cells on them began to spread. Cells on Chi,CAP, and CPL spread well, but not as well as those onmaterials precoated with laminin. It is strange thatprecoating PL with laminin, serum, or fibronectin can-not strengthen its ability to promote cell adhesion, al-though PL can adsorb highest amount of laminin andfibronectin in both solution and serum.

After 6-day culture, on all materials precoated withall proteins except GC, neurons grew as well as thoseon bare materials. The axon of neurons grew longerand the fascicles were thicker than those on bare ma-terials. However, this change was not obvious.

DISCUSSION

The purpose of this study was to compare the nervecell affinity of all eight materials, which was mainlyevaluated by nerve cell attachment, spread, andgrowth in this work. For the evaluation, results ofFMCC cells were used as the major criteria, with those

of MTT as the minor criteria, because the gliosarcomacells used in MTT experiment are not normal nervecells. Cell attachment, spread, and growth could beanalyzed from photographs of cell culture and MTTresults. The results of 1-day cell culture (FMCC 1-dayculture and MTT experiment) reflected mainly the at-tachment and spread of cells, whereas the photo-graphs of 6-day FMCC culture reflected mainly cellgrowth. On the photographs of 1-day cell culture, flatand round cells had finished their attachment process;cells with flat and anomalous shapes (bipolar shapesin neurons and polygonal shapes in glial cells or 9Lcells) and cells with stretched pseudopodia had initi-ated spreading. CAP and CPL were the two materialswith the greatest nerve cell affinity. They are the mostpromising materials for nerve repair. Other materialsexcept GC are all good materials suitable for nerveregeneration.

The results of nerve cell attachment and spread(FMCC cell culture and MTT experiment) may be ex-plained by the physicochemical properties of materi-als and the adsorption of the ECM molecule. There aretwo phases in the process of cell attachment. The firstis nonspecific adsorption of cells on the materialsmainly mediated by the physicochemical interactionbetween cells and materials. Material properties suchas hydrophilicity and positive surface charges contrib-ute to this physicochemical interaction. According to

Figure 9. (a–c) Photographs of FMCC cells cultured for 1 day on Chi precoated with laminin, serum, and fibronectin,respectively.

292 HAIPENG ET AL.

the research of Webb et al., when material surfaces areexposed to dilute serum, cell attachment, spread, andcytoskeletal organization are greater on hydrophilicsurfaces relative to hydrophobic surfaces.17 Positivelycharged surface functional groups further enhance cellattachment.22 In addition, other factors such as rough-ness, porosity, and surface topography all may influ-ence cell behavior on a material.19,23,24 The secondphase is the specific adsorption of cells on the mate-rials. ECM molecules such as laminin and fibronectinparticipate in the process of specific cell attachmentand cell spread. After nonspecific adhesion, which ismostly dependent on material properties, cells will se-crete some ECM molecules which can adsorb onto thematerial surface, bind the integrins on the surface ofnormal nerve cells including neurons and glial cells,and mediate the specific interaction between cells andmaterials. Fibronectin in medium will also be ad-sorbed onto the materials and take part in the interac-tion between cells and materials.26 Successful attach-ment is conducive to the later spreading process.Hence, the results of the cell culture experiment maybe explained comprehensively by the material prop-erties and the amount of adsorption of ECM moleculeson the materials.

Hydrophilicity could be obtained form the watercontact angles on the materials, and the surfacecharges of all the materials could be estimated fromtheir components. Chi, CAP, CPL, and PL all have apositive charge on their surfaces owing to their com-ponent Chi and PL. The surface charges of CG andCAG are far less than those of Chi and PL because inCG and CAG the amino groups which carry positivecharges crosslink with the side chains of gelatin evenwithout the existence of glutaraldehyde. (According tounpublished data from our laboratory, the infraredspectra of CG and CAG were measured and similarvibration peaks in both spectra were found). GC haseven fewer surface positive charges because the aminogroups of chitosan are crosslinked by the glutaralde-hyde. Gel may be thought of as a neutral material.

The adsorption amount of ECM molecules on thematerials were obtained from the results of the ELISAexperiment. In the ELISA experiment, the absorbencyof light at 490 nm did not represent the absoluteamount of adsorbed protein, but instead the amountof adsorbed antigen with a functional antibody recog-nition site. High A490 needs the abilities of materialsboth to adsorb large amount of proteins and to pre-serve the structure of the adsorbed proteins. Accord-ing to the studies of Lampin et al., hydrophobic ma-terials can adsorb more proteins than hydrophilicones, but it seems that hydrophilic materials promotea weak and reversible adsorption of proteins, thuspreserving the structure of adsorbed proteins.19 Pro-teins on hydrophilic materials are more likely to pre-serve the structure of their epitopes than hydrophobic

ones, and thus are more liable to be detected by theirantibodies. On the other hand, positive charges on thesurface of a material can also help to strengthen theinteraction between adsorbed proteins and materials,thereby enlarging the adsorption amount of proteins.PL’s highest hydrophilicity and positive charges onits surface would explain its highest value of A490 inthe adsorption of both fibronectin and laminin. Earlystudies also proved that PL cannot only adsorb lami-nin, but can also protect the structure of laminin fromchanging during adsorption.12 Similarly, Chi’s con-siderably high value of A490 during adsorption of bothfibronectin and laminin may result from its high wet-tability (only less than that of PL and Gel) and surfacepositive charges. Although the glutaraldehyde in GCmay crosslink some of the amino groups of chitosanwhich carry positive charges in neutral solution, thereare some aldehyde groups on the surface of GC whichmay bind proteins via their amino groups by the for-mation of an azomethine bond.20 This compensationled to the similarity between the value of A490 of Chiand GC in fibronectin and laminin adsorption. Sur-prisingly, CPL, the material with a wettability a littleless than Chi and surface positive charges comparableto Chi, had higher A490 in both laminin and fibronectinadsorption than Chi. This result shows that theamount of adsorbed proteins with little structuralchange on a material is determined by factors otherthan its wettability and surface charge, such as bio-logical activities. The high ability of CPL to adsorbfibronectin and laminin and preserve their structuremay be acquired from that of PL, one of its compo-nents. CAP and PL are the same except for their dif-ferent substrata—chitosan and tissue culture cluster,respectively. Hence CAP’s far lower A490 than PLsuggests that the characteristic of materials’ substratais important to the adsorption of proteins onto them.CG and CAG, the most hydrophobic of the eight ma-terials, had the lowest light absorbencies at 490 nm. Itis strange that Gel, a hydrophilic material, had thelowest value of A490 in laminin adsorption. Probably,material properties other than wettability and surfacecharges play a role in protein adsorption. Detailed ex-planation needs further study. Gel’s high value infibronectin adsorption may result from the differentadsorption mechanism—specific adsorption. It hasbeen reported that fibronectin has several function do-mains, some of which can directly bind with collagenI–IV, and that its ability to bind with gelatin (the de-natured collagen) is much stronger than that to bindwith native collagen.18

With the obtained properties of materials and theadsorption amount of ECM molecules on the materi-als, results of FMCC cell culture and MTT experimentmay be explained. Corresponding with predictiondrawn from the rules, Chi, CAP, and CPL, materialswith moderate wettability, a great number of positive

293NERVE CELL AFFINITY OF CHITOSAN-DERIVED MATERIALS

surface charges, and relatively high ability to adsorbECM molecules, could promote the attachment andspreading of gliosarcoma cells and FMCC cells effi-ciently. CAP and CPL (with a lower wettability thanChi) is even better than Chi in facilitating nerve cellattachment of spreading because they can adsorblarger amount of ECM molecules than Chi. It suggeststhat ECM adhesive proteins play a more importantrole than material properties, especially in cell spread.CG and CAG are both the most hydrophobic materialsand their surface positive charges are far less thanthose of Chi, CAP, and CPL. The low positive chargesand high hydrophobicity, plus their poor ECM mol-ecule adsorption ability, result in the poor ability ofCG and CAG to promote both gliosarcoma and FMCCcell attachment and spreading. As a hydrophilic ma-terial with the highest fibronectin adsorption capabil-ity, Ge’s low ability to promote gliosarcoma cells andFMCC cell attachment and spreading may be due toits poor laminin adsorption ability. It seems that lam-inin plays an even more important role in nerve cellspreading than fibronectin.12 GC is different fromother materials in that there are more glutaraldehydein it than in other materials, including CG (stated inMaterials and Methods). It was reported that glutar-aldehyde is poisonous to cells.25 Nerve cells on GC didnot spread, although their number was high becauseof the effect of aldehyde groups on its surfaces. In theMTT experiment, PL, the material with the highestwettability, positive surface charges, and the ability toadsorb proteins and to protect their structure, wasable to promote gliosarcoma cell attachment andspread efficiently. What surprised us most is thatFMCC cells did not attach and spread most efficientlyon PL. Although this is hard to explain, it is probablyrelated to the substratum of PL, the tissue culture clus-ter, as FMCC cells can not grow on the bare tissueculture cluster.

Because MTT is a substrate which can be convert toa blue substance by the succinate dehydrogenase, theA490 represents not only the number but also the av-erage vitality of cells on materials. Appropriate cellshape is critical for cell vitality. In previous studies, ascells of various lines were brought from an extremelyflat shape to a spherical conformation, fewer cells in-corporated [3H]thymidine.21 In our experiment, flatcells had a higher average vitality than spherical cells.The order of cell average vitality of materials is asfollows: PL > CAP, CPL > Chi, Gel > CAG, CG, andGC. Considering the fact that cell numbers are similaron different materials except GC and Gel, MTT resultscan primarily be explained. Larger cell number on GCthan on Chi can explain GC’s higher A490 than Chi.Similarly GE’s lowest A490 results from the low num-ber of cells attached. In the MTT experiment on Gel-derived materials (Fig. 5), cells were cultured for morethan 20 h, while in the experiment on PL-derived ma-

terials (Fig. 6), cells were cultured for 18 h. Cells on GCchanged little from the 18th to the 20th h. On the con-trary, the spreading of the cells on Chi changed obvi-ously within this period of culture. This is why the Chiin Figure 5 is closer to the GC than the Chi in Figure 6(the Chi in Fig. 6 is significantly lower than the GC).

Nerve cell growth was analyzed from the results ofthe 6-day FMCC cell culture. All materials except GCwere able to facilitate neuron growth efficiently. Thisproves our former supposition that the glutaralde-hyde in GC is toxic to nerve cells. Although glial cellsare inhibited by b-D-arabino-furanoside, some glialcells spreading on CG, CAG, Gel, PL, CAP, and CPLcould be found under the microscope. On those ma-terials, some neurons tended to grow on the layer ofglial cells. However, there were few glial cells on Chiand all neurons on it were in direct contact with thematerial surface. This difference between Chi andother materials indicates that Chi is not in favor of thegrowth of normal glial cells.

Generally speaking, however, precoating ECM mol-ecules on materials may improve their nerve cell af-finity. Less improvement in CG, CAG, and Gel thanChi, CAP, and CPL after the laminin precoating mayhave resulted from the lower ability of them to adsorblaminin than other materials. Serum’s greater abilityto improve cell attachment and spreading than fibro-nectin may have resulted from other components inserum such as albumin, the protein reported to be ableto promote cell adhesion.17 The similar tendency in theimprovement in nerve cell affinity of materials pre-coated with fibronectin and serum suggests that fibro-nectin is the most important ingredient in serum topromote nerve cell adhesion. CG, CAG, and GC allhave binding sites for fibronectin. Hence, precoatingfibronectin or serum can greatly improve their abilityto promote cell attachment and spread.

Neurons on GC cannot grow well, meaning that theprotein layer on GC cannot nullify its toxicity to neu-rons. By the 6th day cells secreted laminin and fibro-nectin and the amount of those secreted proteins wassupposed to be far more than that of precoated pro-teins owing to the large initial cell concentration (5 ×105 cells/mL). Hence, the nerve cell growth on mate-rials precoated with ECM molecules was not signifi-cantly more efficient than bare materials.

CONCLUSIONS

Chitosan has excellent nerve cell affinity besides itsgood biocompatibility. Chitosan-derived materials,CAP and CPL, are even better materials than chitosanin nerve cell affinity. They are promising materials fornerve repair. Precoating materials with ECM mol-ecules, especially laminin, can greatly improve their

294 HAIPENG ET AL.

nerve cell affinity. The ECM molecules adsorbed onthe materials and the physicochemical properties ofmaterials, together with other factors such as bioactiv-ity, jointly influence the attachment and spread ofnerve cells on materials.

The authors greatly appreciate the opinion of ProfessorJiang Bo. Discussion with him greatly inspired the authors.The cooperation of the operating room staff at the Depart-ment of Materials, Tsinghua University, is also appreciated.

References

1. Heath CA, Rutkowski GE. The development of bioartificialnerve grafts for peripheral-nerve regeneration. TIBTECH 1998;16:163–168.

2. Langer R, Vacanti JP. Tissue engineering. Science 1993;260:920–926.

3. Kiyotani T, Teramachi M, Takimono Y, Nakamura T, ShimizuY, Endo K. Nerve regeneration across a 25-mm gap bridged bya polyglycolic acid-collagen tube: A histological and electro-physiological evaluation of regenerated nerves. Brain Res 1996;740:66–74.

4. den Dunnen WFA, van der Lei B, Robinson PH, Holwerda A,Pennings AJ, Schakenraad JM. Biological performance of a de-gradable poly(lactic acid-«-caprolactone) nerve guide: Influ-ence of tube dimensions. J Biomed Mater Res 1995;29:757–766.

5. Sato H, Mizutani S, Tsuge S. Determination of the degree ofacetylation of chitin/chitosan by pyrolysis-gas chromatogra-phy in the presence of oxalic acid. Anal Chem 1998;70:7–12.

6. Kurita K. Chemistry and application of chitin and chitosan.Polym Degradation Stability 1998;59:117–120.

7. Xu J, McCarthy SP, Gross RA. Chitosan film acylation andeffects on biodegradability. Macromolecules 1996;29:3436–3440.

8. Tokura S, Nishimura S-I, Sakairi N, Nishi N. Biological activi-ties of biodegradable polysaccharide. Macromol Symp 1996;101:389–396.

9. Jianchun L, Yinghui Z, Haipeng G, Yandao G, Nanming Z,Xiufang Z. A primary study of using chitosan for nerve repairconduit. In: 5th IUMRS Int Conf Adv Mater, June 13–18, 1999,Beijing, China.

10. Dongxu P, Xiaodong C, Lijiang M, Yuanjie H, Wenshu X,Ruihuan S, Ying Z. Study on nerve cell affinity of polymerhydrogel films. In: 3rd Far-Eastern Symp Biomed Mater, July15–17, 1997, Chengdu, China.

11. Elghannam A, Ducheyne P, Shapiro IM. Bioactive materialtemplate for in vitro synthesis of bone. J Biomed Mater Res1995;29:359–370.

12. Matsuzawa M, Kazutoshi K, Sugioka K, Knoll W. A biocom-patible interface for the geometrical guidance of central neu-rons in vitro. J Colloid Interface Sci 1998;202:213–221.

13. Lin F-H, Liao C-J, Liu H-C, Chen K-S, Sun JS. Behavior of fetalrat osteoblasts cultured in vitro on the DP-bioactive glass sub-stratum. Mater Chem Phys 1997;49:270–276.

14. Huber M, Heiduschka P, Kiele S, Pavlidis C, Mack J, Walk T,Jung G, Thanos S. Modification of glass carbon surfaces withsynthetic laminin-derived peptides for nerve cell attachmentand neurite growth. J Biomed Mater Res 1998;41:278–288.

15. Agius E, Cochard P. Comparison of neurite outgrowth inducedby intact and injured sciatic nerves: A confocal and functionalanalysis. J Neurosci 1998;18:328–338.

16. Jun, C. The molecular biology and clinical diseases of extracel-lular matrix. Beijing: Publishing Company of Beijing MedicalUniversity, 1999. p 29–38.

17. Webb K, Hlady V, Tresco PA. Relative importance of surfacewettability and charged functional groups on NIH 3T3 fibro-blast attachment, spreading, and cytoskeletal organization. JBiomed Mater Res 1998;41:422–430.

18. Borkenhgen M, Clemence JF, Sigrist H, Aebischer P. 3-Dimen-sional extracellular matrix engineering in the nervous system.J Biomed Mater Res 1998;40:392–400.

19. Lampin M, Warocquier-Clerout R, Legris C, Degrange M,Sigot-Luizard MF. Correlation between substratum roughnessand wettability, cell adhesion, and cell migration. J BiomedMater Res 1997;36:99–108.

20. Welle A, Grunze M, Tur D. Plasma protein adsorption andplatelet adhesion on poly[bis(trifluoroethoxy)phosphazene]and reference material surface. J Colloid Interface Sci 1998;197:263–274.

21. Folkman J, Moscona A. Role of cell shape in growth control.Nature 1978;273:345–349.

22. Hamano T, Suzuki H, Teramoto A, Iizuka E, Abe K. Functionalculture of human periodontal fibroblast (HPLF) on polyelec-trolyte complex (PEC). Pure Appl Chem 1998;A35:439–455.

23. Curtis A, Wilkinson C. Topographical control of cells. Bioma-terials 1997;18:1573–1583.

24. Hallab NJ, Bundy KJ, O’Connor K, Clark R, Moses RL. Celladhesion to biomaterials: Correlations between surface charge,surface roughness, adsorbed protein, and cell morphology. JLong-Term Effects Med Implants 1995;5:209–231.

25. Schweikl H, Schmalz G. Glutaraldehyde-containing dentinbonding agents are mutagens in mammalian cells in vitro. JBiomed Mater Res 1997;36:284–288.

26. Greenberg JH, Seppa S, Seppa H, Hewitt AT. Role of collagenand fibronectin in neural crest cell adhesion and migration.Dev Biol 1981;87:259–266.

295NERVE CELL AFFINITY OF CHITOSAN-DERIVED MATERIALS