Characterization

lable at ScienceDirect

Biomaterials 29 (2008) 3781–3791

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

Synthesis and characterization of biodegradable elastomeric polyurethanescaffolds fabricated by the inkjet technique

Changhong Zhang, Xuejun Wen, Naren R. Vyavahare, Thomas Boland*

Department of Bioengineering, Clemson University, 420 Rhodes Research Centre, Clemson, SC 29634, USA

a r t i c l e i n f o

Article history:Received 15 February 2008Accepted 7 June 2008Available online 3 July 2008

Keywords:BiodegradationCell adhesionCell proliferationCell–material interactionDegradationMechanical properties

* Corresponding author. Tel.: þ1 864 656 7639; faxE-mail address: [email protected] (T. Boland).

0142-9612/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.biomaterials.2008.06.009

a b s t r a c t

Biodegradable polyurethanes (PUs) were synthesized from methylene di-p-phenyl-diisocyanate (MDI),polycaprolactone diol (PCL-diol) and N,N-bis (2-hydorxyethyl)-2-aminoethane-sulfonic acid (BES),serving as a hard segment, soft segment and chain extender, respectively. MDI was chosen due to itsreactivity and wide application in synthesis of biomedical polyurethanes due to its reactivity; PCL-diolwas chosen because of its biodegradability; and BES was chosen because it allowed the introductionsulfonic acid groups onto the polymer chains. We evaluated the polyurethanes’ degradation rate, me-chanical properties, hydrophilicity, antithrombogenecity, and ability to support fibroblast cell attachmentand growth by comparing with polymers having a 2,2-(methylimino)diethanol (MIDE) chain extender.Mechanical testing demonstrated that the PU containing BES has tensile strengths of about 17 MPa andelongations up to 400%, about three times the strength and four times the elongation than the MIDEbased PUs. The polymers showed decreased in vitro degradation rates, lower glass transition temperature(Tg) and hydrophilicity possibly due to enhanced microphase separation. Preliminary cytocompatibilitystudies showed that all the PUs are non-toxic, but PU containing BES exhibited much lower cell at-tachment and proliferation than the MIDE chain extended polymers. An in vitro platelet adhesion assayshowed lower platelet attachment on BES containing PU. Additionally, due to the existence of sulfonicacid groups, the BES extended PU became water soluble in basic condition and insoluble in acidiccondition, a phenomenon that is reversible at pH value of 8.7, making this a pH sensitive polymer at-tractive for bioprinting applications. By adding acetic acid into an inkjet cartridge and printing it ontoa PU solution with pH above 8.7, precision fabricated scaffolds can be obtained, suggesting that BES basedPUs are promising candidates as synthetic inks used for customizable fabrication of tissue engineeringscaffolds.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

For decades, polyurethanes (PUs) have been a biomaterial ofchoice for many applications [1,2] because of their biocompatibilityand excellent mechanical properties. However, as most poly-urethanes are non-degradable, they may cause long-term foreignbody reaction, may fail to integrate or exhibit material fatigue andhave found little use in the field of tissue engineering [3]. Degradablebiomaterials, on the other hand, whether natural or synthetic, havepoor mechanical properties; the majorities are either too stiff withlow flexibility or too soft with relatively low strength [4–6]. Toovercome those limitations, a number of degradable PUs have beenintroduced for a range of biomedical applications varying fromcardiovascular repair, cartilage implant, ligament regeneration,bone replacement to controlled drug/gene delivery [7–9].

: þ1 864 656 4466.

All rights reserved.

It has been shown that incorporation of sulfonic groups intonon-degradable PUs will improve the polymers’ blood compati-bility, an effect explained by principle that the electrostatic re-pulsion between the sulfonic groups and the blood proteins lessensprotein adsorption and lowers platelet adhesion and activation[10,11]. Several studies found that the incorporation of sulfonicgroups also affected other physical properties, such as the glasstransition, tensile strength, modulus, melt viscosity, relaxationbehavior, and solution behavior. Generally, these polymers showa higher degree of water absorption than their non-sulfonatedcounterparts [12], and exhibit reduced mechanical properties inaqueous solutions due to their water absorption. Some researchersfound that the increased sulfonic ion content in the PU backboneresulted in the water soluble polymers; other groups reported thatthe sulfonated PUs became water soluble in basic environment butwater insoluble in acidic environment [13,14]. However, there areno reports about how to utilize these pH sensitive polymers tofabricate scaffolds. Furthermore, although much research work hasbeen done incorporating sulfonic groups into non-degradable PUs,

(PCL530)

+

MP530B

2OCN

(MP530M)

NH CO

NHCO

(CH2)5CO

[ ]n

O CH2 CH2O O CH2NH CO

NHCO

CH2O CH2)2( CH2)2( O{ }

nN

CH3

CH2 NH CO

NHCO

(CH2)5 CO

[ ]n

O CH2O O CH2NH CO

NHCO

CH2{ O CH2)2( N

SO3H

CH2)2( O}n

(MP530B)

+CH2 NCO HO (CH2)5 CO

[ ]n

O CH2 CH2 OH

(MDI)

+

DMSO, 60 °C, 2-3 hStannous octoate

DMSO, room temperature, 12 h

(CH2)2 OH(CH2)2 N

SO3H

HO

(BES)

(CH2)2 (CH2)2N

CH3

OHHO

(MIDE)

MP530M

Fig. 1. Synthesis scheme for the preparation of MDI, PCL530, MIDE or BES based PUs. Chain extender was changed to obtain materials with different functionalities.

C. Zhang et al. / Biomaterials 29 (2008) 3781–37913782

there is, to our knowledge, no report about incorporating thosegroups into degradable PUs that may find applications in the tissueengineering arena.

Precision fabrication of biological scaffolds that provide supportto the cells during growth and development into complex three-dimensional structures is one goal of tissue engineering. Because ofthe limitation of the traditional fabrication to control and predefinemicrostructure of the final scaffold, several techniques hithertotermed ‘solid freeform fabrication’ (SFF), were adapted to customfabricate scaffolds [15–18]. These include stereo lithography [19],wax printers [20], and inkjet printing [18,19]. The main advantagesof these techniques are their capability to precisely control matrixarchitecture and their ability to be interfaced with computer im-aging techniques, which make it possible to produce constructswith customized size and shape for tailor-specific tissue engi-neering applications [21]. Among them, the low-cost inkjet printingtechnique was recently shown to possess particular advantages ongenerating three-dimensional scaffolds and cellular structures[15,22] because it allows simultaneous and precise deposition ofcontrolled quantities of scaffolding materials, nutrients, therapeu-tic drugs, growth factors and other bioactive components to formcells/scaffold constructs for in vitro and in vivo growth [23]. How-ever, the biomaterials available to date for SFF in general and inkjetprinting in particular are very limited and restricted to a very fewnatural polymers such as alginic acid. Therefore most investigationon the inks used for precision scaffolds have been focused on very,which is known to crosslink in calcium chloride solution throughstatic force interaction among ions and polymer chains [24–26].The availability of cell compatible synthetic polymers with good

mechanical properties for use in SFF will prove beneficial for manybiomedical applications.

In order to obtain a pH sensitive, degradable polymer with highstrength, we chose to employ a methylene di-p-phenyl-diisocya-nate (MDI) hard segment, a polycaprolactone diol (PCL-diol) softsegment and an N,N-bis (2-hydorxyethylhydroxyethyl)-2-amino-ethane-sulfonic acid (BES) chain extender because MDI based PUshave proven to possess excellent mechanical properties, PCL isa biodegradable and biocompatible polymer and BES chainextenders have shown to yield pH sensitive PUs. The effects ofsulfonic acid incorporation were evaluated by comparing witha non-sulfonic acid containing polyurethane, which had a 2,2-(methylimino) diethanol (MIDE) chain extender. MIDE was usedbecause it owns a similar chemical structure to BES but contains nosulfonic acid group as shown in Fig. 1. Tecoflex�, a commerciallyavailable non-degradable biomedical-grade PU, was also used forcomparison.

2. Materials and methods

Physiochemical property characterization methods were used to characterizethe bulk polymers, specifically attenuated total reflective-Fourier transform infraredspectroscopy (ATR-FTIR) to characterize the structures, gel permeation chroma-tography (GPC) to measure the molecular weights, differential scanning calorimetry(DSC) to characterize the thermal properties, water swelling rate analysis to measurehydrophilicity, tensile testing to measure the mechanical properties, in vitro deg-radation rate analysis to measure the degradability and scanning electron micro-scope (SEM) to observe the porous scaffold structure. In vitro platelet adhesionassays were conducted for the preliminary investigation of the PUs’ blood-con-tacting response. The cell viability and proliferation condition were evaluated by 3T3fibroblasts culturing on both flat polymer membranes and fabricated scaffolds.

C. Zhang et al. / Biomaterials 29 (2008) 3781–3791 3783

Compared to the PU containing MIDE, PU containing BES was expected to exhibithigher mechanical properties, lower platelet attachment and water solubility inbasic solution.

2.1. Materials

MDI, BES, MIDE, acetic acid, dichloromethane (CH2Cl2) and N,N-dime-thylformamide (DMF) were obtained from Acros Organics Fine Chemicals (Geel,Belgium). Dimethyl sulfoxide (DMSO) was purchased from Fisher Scientific In-ternational. Stannous octoate (Sn(oct)2, Sn[CH3(CH2)3CH(C2H5)COO]2) and poly-caprolactone diol with Mn¼ 530 (PCL530) were purchased from Sigma–Aldrich (St.Louis, MO). MDI was purified by vacuum distillation. BES was vacuum dried at 60 �Cfor 48 h before use. MIDE was distilled with calcium hydride (CaH2) to eliminatemoisture. DMSO was distilled over CaH2 at atmospheric pressure under nitrogenprotection. PCL530 was dehydrated in a vacuum oven at 60 �C for 48 h. Sn(oct)2 waspurified over 4 Å molecular sieve with overnight stirring to eliminate trace waterprior to use. Acetic acid, CH2Cl2 and DMF were used as received. NIH 3T3 mousefibroblasts (CRL-1658, ATCC, Manassas, VA) were used as a model cell type for cellviability and proliferation assay. Dulbecco’s Modified Eagles Media (DMEM) aug-mented by 10% fetal bovine serum (FBS) and 1% antibiotics was used as culturemedia, all obtained from Sigma Chemical Company.

2.2. PU synthesis and membrane preparation

The polymers used in this study were synthesized by the traditional two-stepmethod under nitrogen protection as described in Fig. 1 [27]. Briefly, the stoichi-ometry of the reaction was approximately 2:1:1 MDI/PCL530/BES or MIDE. The MDIwas dissolved in DMSO in a four-neck flask and PCL530/DMSO solution containing1 wt% Sn(oct)2 as catalyst was added dropwise at 60 �C. This mixture was allowed toreact for 2.5–3 h and then cooled down to room temperature. Following that, 5% w/vchain extender in DMSO was added dropwise to the prepolymer solution underconstant stirring for another 12 h. After completion of the reaction, the polyurethanesolution was precipitated into deionized water for at least 48 h, washed thoroughlywith ethanol for 6 h at room temperature, and then dried in a vacuum oven at 60 �Cfor 48 h before further use and characterization. The final polymers were designatedMP530B and MP530M depending whether the chain extender was BES or MIDE. AllPU membranes were prepared by casting 8% (W/V) DMF solution onto a Teflon moldand drying at 60 �C for 24 h; Tecoflex� was dissolved in CH2Cl2. All the cast filmswere removed from the mold and further dried in a vacuum oven at 60 �C for 48 h toremove residual solvent. The membranes had an average thickness of about0.17� 0.03 mm and were stored in a desiccator at room temperature. The mem-branes were cut into appropriate size and used for mechanical testing, waterswelling rate analysis, in vitro degradation assay and platelet adhesion experiment.

2.3. Bulk property characterization

GPC was conducted using polystyrene solutions in DMF (EasiCal PS-1, Poly-merLabs, Amherst, MA) with molecular weights in a range of 580–7,500,000 Da ascalibration standards. The polymers were dissolved at 0.25% (W/V) in the GPC carriersolvent and 20 ml samples were injected. The samples for ATR-FTIR (Nicolet IR200)were prepared by solution casting of 8% (W/V) polymer in DMF directly onto KBrcrystal plates and vacuum dried at 60 �C for 24 h prior to characterization. Thermalanalysis was performed in Mettler DSC analyzer (DSC 822e), with a heating rate of20 �C/min under constant nitrogen flow. Polymer samples (70–90 mg) were heatedto 50 �C for 10 min, quenched to �100 �C, maintained at this temperature for10 min, then tested over the range from �50 �C to 180 �C at 20 �C/min. Water ab-sorption analysis was performed by measuring the swelling rate of the polymermembranes after immersion in deionized water at room temperature for 10 min,25 min, 1 h, 1.5 h, 7 h and 48 h. The surface water on each sample was blotted witha filter paper before weighing. The swelling ratio was calculated by dividing theweight of the swollen sample by the weight of the dry sample. Three samples of eachPU were measured to obtain the mean water swelling ratio and standard deviation.For tensile tests, PU samples were cut from membranes into approximate5.5�12 mm2 rectangular strips. Tensile test were conducted using an Instron 4502mechanical tester (Instron, Norwood, MA) at a crosshead speed of 25 mm/min withmaximum load of 10 kN. Due to the application of polymers in water-rich envi-ronment such as culture media and live tissues, measurements of the mechanicalproperties were carried out in both dry and wet conditions. Wet samples wereprepared by saturating them in phosphate-buffered saline (PBS) for 12 h at roomtemperature to reach the equilibrium. Three samples of each condition were mea-sured to get an average tensile strength, elongation and standard deviation values.

2.4. In vitro degradation assay

In vitro degradation of the PUs was evaluated by recording the samples’ weightloss, molecular weight changes, and mechanical properties changes over time in PBSbuffer solution at 67 �C. Each PU sample cut from membrane was with the weight ofabout 14 mg and the size of approximate 5.5 mm in width, 12 mm in length and0.16 mm in thickness. The samples were placed in small vials separately filled with1.5 ml PBS buffer solution containing 0.5% (W/V) sodium azide as antimicrobial

agent. Those tubes were placed in a 67 �C water bath with gentle shake of ap-proximate 70–80 rpm to simulate dynamic in vivo tissue environment. A highertemperature was used to accelerate the degradation rate. An established relation-ship with different temperatures is available to convert the degradation profile to37 �C [7]. At each time point of week 1, 2, 3 and 4, three vials of each type of materialwere sampled, rinsed for 1 h by deionized water and vacuum dried for three daysbefore analysis of weight loss and molecular weight change. The PBS solution in eachvial was collected for pH value measurement. After that, the PU stripes were thensaturated by PBS buffer solution at room temperature for the tensile test in wetcondition.

2.5. In vitro cytocompatibility assay

PU solutions were evenly coated on 18 mm diameter coverglass (n¼ 6 for eachtime point and each material). After being dried at 60 �C in the oven for 12 h,samples were further dried in vacuum oven at 60 �C for 24 h to eliminate the re-sidual solvent. The samples were sterilized in 75% ethanol for 15 min, washed withsterile PBS three times, and then put into 12-well tissue culture plates. Fibroblastcells were seeded at concentration of 1.5�106 cells/ml on the polymer surfaces,after 3 h, when cells were attached on the polymer surfaces, 2 ml media was addedto each well to continue the culture for 9 days observation. Cell adhesion and pro-liferation on the sample surfaces were examined at 3, 5, 7, and 9 days by phasecontrast microscopy and total cell populations; culture dishes and Tecoflex� coatedcover glasses were used as negative controls.

2.6. Platelet adhesion characterization

Fresh pig blood was obtained from a local slaughter house, collected into 8.5 mlvenous blood collection tubes (Vacutainer�, BD Company) that are pre-filled with1.5 ml anticoagulant containing 0.22% trisodium citrate, 0.08% citric acid and 0.24%dextrose. Platelet-rich plasma (PRP) was obtained from the blood by centrifugationat 750g for 10 min; the upper layer was collected and added to an equal volume ofPBS. The platelet concentration of PRP was measured by Coulter counter (Z2COULTER COUNTER�, Beckman Coulter, Inc.) and adjusted to be about5.2�107 platelets/ml by dilution. Polyurethane membranes were punched intoround shaped films with diameter of 8 mm (n¼ 6 for each sample) and equilibratedin PBS at 37 �C overnight. After removing the samples from the PBS solution, thesamples were immersed in PRP and incubated in a cell culture incubator at 37 �C and5% CO2 atmosphere for 1 h. Following incubation, the samples were gently rinsedwith PBS solution three times to wash away any non-attached platelets, moved toa new 12-well plate; 400 ml of 1% Triton-X100 solution was added into each well andincubated at 37 �C for 1 h to lyse the adherent platelets. Lactate dehydrogenase(LDH) activity was measured using a CytoTox 96� Non-Radioactive Assay kit(Promega Corp.). Briefly, the stable LDH released from lysed platelets was coupled toa tetrazolium salt (2-p-iodophenyl-3-p-nitrophenyl-5-phenyl tetrazolium chloride,INT) and caused the conversion of INT into a red formzan product. The concentrationof red product was indirectly obtained by measuring the light absorbance at 490 nmand the value of light absorbance was proportional to the number of platelets.Tecoflex� was used as control. Six samples of each PU were measured to obtain themean light absorbance value and standard deviation.

2.7. PU neutralization and scaffold fabrication

PU neutralization was carried out by first dissolving MP530B in DMF to forma 12% (W/V) solution. A calculated stoichiometric amount of 1 mol/L sodium hy-droxide water solution was added and reacted at room temperature for 1 h as de-scribed in Fig. 7C; sulfonic acid groups on chain extender were reacted to sulfonicions and the neutralized polymer became water soluble. The neutralized polymer/DMF solution was further diluted by deionized water to 2.5% (W/V) for scaffoldfabrication. The HP desktop 3900 printer and HP 21 black ink cartridge was modifiedto fabricate single-layer patterns as described elsewhere [28]. The 50% (V/V) aceticacid (HAc) water solution was filled into the cartridge. The 2% (W/V) polymer so-lution was dropped on the glass slides to form an even liquid layer on the standardglass slide surface, this glass slides was then placed onto the print stage mountedunder the print heads. The patterns that consisted of letters (CLEMSON, font: Timesand New Roman, size 16) or rowed rings were designated using Microsoft Words toprogram the printer. The HAc solution was printed out in one cartridge pass to theglass slide and reacted with the polymer solution (Fig. 7C), thus the final expectedpatterns on the glass slide were obtained. These scaffolds on the glass slides weredehydrated by immersion in a series of ethanol/water solution (50, 70, 85, 90, 100%V/V) before critical point drying with liquid CO2 as transitional liquid medium. Aftersputter coating with gold, samples were observed by (Hitachi s7400) scanningelectron microscope (SEM 3500, Hitachi Ltd) at magnifications of 400, 2.5K and 20K.

2.8. Fibroblast seeding on printed scaffold

Dehydrated scaffolds were sterilized with 75% ethanol for 24 h and vacuumdried for another 24 h. Before cell seeding, scaffolds were rehydrated in sterilized0.1 M PBS for 1 h. Fibroblast cells were seeded at concentration of 1.5�108 cells/mlon the printed O-ring shaped polymer scaffold on the slide in a Petri dish

do

th

erm

M530B, Tg= 8°C, Cp = 0.48 J/g°C

C. Zhang et al. / Biomaterials 29 (2008) 3781–37913784

(100 mm� 20 mm), after 3 h, when cells were attached to the polymer, 10 ml mediawas added into the Petri dish to fully cover the scaffold, and the study continued for5 days observation. At day 5, the slide with scaffold was transferred to a clean Petridish, 10 ml prepared calcein AM (Molecular Probes, Eugene) PBS solution (5 ml/10 ml) was added onto scaffold to stain the cells for 45 min in incubator. The samplewas then examined by UV light fluorescence microscopy (Nikon Diaphot 300) tocheck the cell morphology on printed scaffold.

-50 -20 10 40 70 100 130 160Temperature (°C)

En

M530M, Tg= 19°C,Cp = 0.52 J/g°C

Fig. 3. DSC thermograms of two types of biodegradable PUs of MP530B and MP530M.The curve on the top shows the thermal behavior of MP530B with Tg at 8 �C and DCp of0.48 J/g �C. The curve on the bottom shows the thermal behavior of MP530M with Tg at19 �C and DCp of 0.52 J/g �C.

3. Results and discussion

3.1. Bulk characterization

The FTIR absorption spectra of the PUs at room temperature areshown in Fig. 2. The two synthesized PUs exhibited similar spectrabetween 400 and 4000 cm�1 with the exception of a peak at1043 cm�1 observed for the BES containing polymer, which wasassigned to the –SO3– symmetric stretch [29,30] and indicates thesuccessful synthesis of this PU. The ratio of the hydrogen bondedcarbonyl absorbance of C]O stretching vibration at about1704 cm�1 to the free carbonyl absorbance at about 1731 cm�1

provides a qualitative estimation of hydrogen bonds and phaseseparation in PUs [31]. As estimated from Fig. 2, the ratio of bondedcarbonyl to free carbonyl for the BES extended polymer is about0.87 and 0.95 for the MIDE extended polymer. The lower ratio in-dicates the formation of the hydrogen bonds between the pendentsulfonic acid groups and urethane groups, which most likely willcontribute to better mechanical properties in those polymers. InPUs, hydrogen bonding can occur between the urethane groups,hard segment carbonyl oxygen and soft segment carbonyl oxygen.In addition the BES containing polyurethanes can form hydrogenbonds using the sulfonic acid groups, thus affecting the hard seg-ment hydrogen bonding and packing among carbonyl and ure-thanes groups. This will result in the change in physical propertiessuch as lower glass transition temperature, and higher mechanicalproperties compared to the control.

The DSC results for two types of PUs are shown in Fig. 3. Glasstransition temperatures (Tgs) of 8 �C and 19 �C were observed forMP530B and MP530M, respectively (Table 1). Both Tgs are lowerthan 37 �C, indicating that two types of PUs can maintain elasticityat body temperature. These Tg values are also substantially higherthan that of pure PCL (Tg¼�58 �C), indicating phase mixing of hardsegments and soft segments in the PCL based PUs [32,33]. Thelower Tg value of MP530B than MP530M suggests that the BESchain extended PU has higher phase separation degree than itsMIDE chain extended analog, which is consistent with the resultsfrom FTIR observation. The enhanced phase separation in MP530B

Wavenumbers(cm-1

)

Tran

sm

ittan

ce

M530B-SO3H

C=O(Bonded)C=O(Free)

80010001200140016001800

M530M

Fig. 2. FTIR spectra of the PUs with different chain extenders. The band at 1042 cm�1

corresponds to the symmetric stretching mode of –SO3–, the bands at 1704 cm�1 and1731 cm�1 correspond to the stretching mode of the hydrogen bonded and free car-bonyl group, respectively. The ratio of bonded carbonyl absorbance to free carbonylabsorbance of MP530B is about 0.87 and same ratio of MP530M is about 0.95, in-dicating higher percentage of free carbonyl in MP530B.

can be explained by the significantly high polarity difference be-tween the sulfonic acid containing chain extenders. It is generallythought that the phase separation typical in all PUs is attributed tothe polarity difference of the polar hard segment and un-polar softsegment; the larger the difference between hard segment and softsegment polarities, the higher the phase separation will be [27]. Inour study, the PCL in polymer soft segment can be considered to beof very low polarity, the sulfonic acid in the chain extender ofMP530B has much higher polarity than methyl group of MIDE inMP530M. Therefore, there is a higher polarity difference of thedifferent segments of MP530B, which can cause more phaseseparation.

The water absorption kinetic profiles in Fig. 4 show that theamount of water in PUs initially increases rapidly, and then grad-ually reaches the plateau after 7 h. The final swelling rates aredocumented in Table 1. In this study, MP530B exhibited much lowerwater absorption rate and lower equilibrium swelling rate than thatof MP530M, which is mainly attributed to the high phase separa-tion of BES chain extended polymers. Although the nitrogen atomsand pendent sulfonic acid groups in MP530B have certain degree ofhydrophilicity, the high degree of phase separation limits thepolymer chains’ mobility, thus limits the formation of hydrogenbonds between water and polymer chains. Water absorption inMP530B took longer time to reach the plateau value (Fig. 4), whichreflects the lower mobility of the polymer chains and limited abilityof polymer chains to combine. Compared to MP530B, MP530Mexhibited higher water absorption rates and higher final swellingrates, which can be explained by the lower polarity of chain ex-tender MIDE and the presence of nitrogen atoms. The polymerchains of MP530M have higher phase mixing and polymer chainmobility, which further resulted in the higher possibility of thehydrogen bond formation among water molecules and polymerchains. As also seen in the figure, the shorter time for MP530M toreach the plateau value also reflects the higher polymer chainmobility and higher microphase mixing. Therefore, the degree ofmicrophase separation in both PUs appears to be the primarydriving force determining the PU’s water absorption properties.

3.2. Mechanical properties

The elastomeric properties of the polyurethane are derived fromthe phase separation of the hard and soft segments, this phaseseparation occurs because of the incompatibility of a mainly non-polar soft segment with a polar hard segment. The stiff andimmobile urethane hard segment domains serve as crosslinks

Table 1Composition and properties of biodegradable PUs made from MDI, PCL530 and chain extender of BES or MIDE

Materials Composition molar ratio Mechanical propertiesin percentage to Tecoflex SG60

GPC Mn (g/mol) Swelling ratioat 48 h (%)

Tg (�C)

Tensile strength(%)

Elongation (%)

Dry Wet Dry Wet

MP530B MDI/PCL/BES 2/1/1 59.1 11.3 52.7 109 162,000 4.71� 1.85 8MP530M MDI/PCL/MIDE 2/1/1 9.6 3.5 26.8 7.3 66,000 11.01� 0.86 19

C. Zhang et al. / Biomaterials 29 (2008) 3781–3791 3785

between the uncoiled and flexible soft segment domains. Uponmechanical deformation the hard segment become aligned whilesoft segment become uncoiled, which bring about the high tensilestrength and elongation to the polyurethanes [27]. In this study, themechanical properties of MP530B and MP530M are reported inTable 1 as a percentage of the Tecoflex control which has reportedultimate tensile strength of 41 MPa and 400% elongation at break.MP530B exhibits three times the tensile strength of MP530M in dryconditions and approximately 2/3 that of Tecoflex. When saturatedwith PBS, the ultimate tensile strength of MP530B decreases toapproximately 10% of the Tecoflex strength (w4 MPa); MP530Mdecreases to approximately 3% of the Tecoflex strength. The highermechanical properties of MP530B can be explained by two reasons,first, the incorporation of the polar sulfonic acid in the polymerchains formed ion aggregates, increased hard segment polarity andserved as extra physical crosslinks in the polymer, thereby in-creasing microphase separation and reinforcing the polymerchains’ interaction; second, the lower molecular weight ofMP530M, which is about 40% of the Mn of MP530B, decreased thepolymer chains’ integrity and interaction.

As shown in Table 1, elongation at break of MP530B was abouthalf that of Texoflex and about the same when saturated with PBS.These changes are attributed to the interaction between the sul-fonic groups and water molecules through hydrogen bonding,which outcompete the polymer–polymer hydrogen bonds, therebyweakening the material. On the other hand, water can also increasethe polymer elongation by serving as inter-chains’ ‘lubricant’ or‘plasticizer’ in the polymer matrix [34], which was in favor of thelong polymer chains’ realignment at high deformation, therebyenhanced the polymer matrix integrity in wet condition. MP530Mexhibited both decreases of tensile strength and elongation in wetcondition. Due to the low molecular weight and short polymerchain length of MP530M used in this study, a high percentage ofpolymer–polymer hydrogen bonds was replaced by water–polymerhydrogen bonds, and as a result, the polymer chains’ interactionwas seriously interrupted, resulting in decreased elongation.

MP530M

0

2

4

6

8

10

12

14

0 1 2 3 4 5 6 7 8Time (hours)

Sw

ellin

g ra

te

(%

)

MP530B

Fig. 4. Swelling rate of two types of PU samples in water at room temperature.MP530B exhibited much lower swelling rate than MP530M in 7 h assay.

Mechanical properties of the control MP530M have beenreported previously [35]. However, that study did not compare theproperties to a known control. As absolute values for tensilestrength vary considerably depending on specimen size, prepara-tion and testing parameters such as pulling speed, we chose here tocompare the polymers to Tecoflex as it has been well characterized.A direct comparison between MP530B and MP530M of the pre-vious study is hence not possible. The high elasticity and tensilestrength of MP530B are essential for engineering tissues with highelasticity, such as muscle, tendon, ligament, cardiovascular andvocal cord. In this study, although the tensile strength of MP530B inwet condition is lower than that of Tecoflex, the strength is stillsufficient for most tissue engineering applications [36].

3.3. In vitro degradation assay

The degradation behavior of MP530B and MP530M was exam-ined by measuring the molecular weight, mass loss, tensile strengthand pH value. To speed up the degradation procedure to a measur-able range, environmental temperature of 67 �C was employed. It hasbeen reported that the degradation rate of MDI and PCL based PUs at37 �C is about 1/20 of the degradation rate at 77 �C [7]; thereby, thedegradation rate of MP530B and MP530M at 37 �C will be approxi-mate 1/9 of the degradation rate at 67 �C according to Arrhenius’equation and assuming similar activation energies and pre-expo-nential factors. However, the rates should be taken as a guide and willneed to be verified by in vivo experiments. As seen from Fig. 5,MP530B exhibited much lower mass loss and molecular weight lossthan MP530M. The molecular weight of MP530B decreased by 20% atweek 3 and that of MP530M decreased by 70% at week 2; the pHvalue of degradation solution of MP530B decreased from 7.4 to7.09� 0.02 and that of MP530M decreased from 7.4 to 7.30� 0.02after 4 weeks. The degradation rate of MP530M was about 4 and 10times faster than that of MP530B. This result can be explained by thedegree of microphase separation and initial molecular weight ofpolymers. The high phase mixing of MP530M caused the moreflexibility of polymer chain, the low initial molecular weight ofMP530M also resulted in the lower polymer chains’ interaction; bothfacilitating easy attack of water molecules to the backbone, whichfurther increased the rate of hydrolysis. Although MP530B has thesulfonic acid groups that are hydrophilic, the high microphase sep-aration can increase the polymer chains’ interaction and resist theattack of water molecules to ester groups, hence slow down thehydrolysis procedure. Interestingly, because MP530B released sul-fonic acid groups from chain extender and caproic acid from softsegment, while MP530M only release caproic acid during hydrolysis,the MP530B degradation solution was more acidic than that ofMP530M degradation solution. Considering this, the lower pH en-vironment was expected to accelerate the hydrolysis procedure.However, in comparison of MP530M, the more acidic MP530Bexhibited lower degradation rate in the assay. Thus, we may concludethat during MP530B degradation, the microphase separation playeda dominant role over the self-catalysis by acidic environment. On theother hand, compared to the pH value change from 7.4 to 3.9 duringpolylactide acid (PLA) degradation at 77 �C in 4 days [35], the

0 1 2

MP530M

3 492

94

96

98

100

Weeks in PBS solution at 67 C

Re

ta

in

ed

w

eig

ht in

pe

rc

en

ta

ge

(%

)

MP530B

Weeks in PBS solution at 67°C

0

20

40

60

80

100

0 1 2 3 4

Re

ta

in

ed

m

ole

cu

la

r

we

ig

ht in

p

erc

en

ta

ge

(%

)

MP530B

MP530M

MP530B

MP530M

0

20

40

60

80

100

0 1 2 3 4Time (weeks)

Reta

in

ed

te

ns

ile

stren

gth

a

t b

re

ak

(%

)

A B

C

Fig. 5. Degradation behavior of two types of PUs in PBS solution at 67 �C. (A) The average retained weight of MP530B and MP530M changes as a function of weeks in PBS. (B) Theaverage retained molecular weight of MP530B and MP530M changes as a function of weeks in PBS. (C) The change of tensile strength of MP530B and MP530M as a function ofweeks in PBS.

C. Zhang et al. / Biomaterials 29 (2008) 3781–37913786

degradation products of MP530B and MP530M are not expected tosignificantly influence pH in the culture media in vitro at 37 �C.

The tensile strength as a function of in vitro degradation time isshown in Fig. 5C. MP530B lost only about 15% tensile strength after4 weeks while MP530M lost about 85% compared to a 38 and 85%loss of molecular weight, respectively. Obviously, the influence ofwater on the tensile strength change of MP530B during degrada-tion is much weaker than that on MP530M. Moreover, MP530Bshowed the lower tensile strength loss than molecular weight loss,indicating that the higher microphase separation and ionic forcestill maintained the polymer’s integration. The property thatMP530B can maintain the mechanical properties during degrada-tion may be an advantage for the tissue engineering application ofMP530B as long-term degrading scaffolds.

3.4. Cytocompatibility and platelet adhesion

The difference in cell proliferation rates on different materialcan be attributed to the cell–material interactions. Some recentstudies have compared effects of surface charges on cell adhesion.Some demonstrated greater cell attachment and spreading onhydrophilic, positively charged glass surfaces versus hydrophobicsurfaces [37]. Other work has demonstrated greater cell attach-ment and spreading on negatively charged and hydrophilic glasssurface relative to hydrophilic glass surfaces [38]; yet other workdemonstrated lower cell attachment and normal cell spreading onnegatively charged glass surface [39]. Looking specifically at sul-fonation, increased cell attachment and proliferation were docu-mented with increasing surface sulfonation density of polystyrene/poly(methyl methacrylate) copolymers [40]. To examine the cyto-compatibility of the PUs in this study, fibroblasts were seeded onthe membrane samples of MP530B, MP530M, Tecoflex and tissueculture dishes. Tecoflex and non-coated Petri dishes were used ascontrol. As shown in Fig. 6, fibroblasts were attaching and pro-liferating on all the samples with varying densities with normalflattened appearance, indicating the existence of the different cell–

material interactions for each material but all of them were cyto-compatible. Fibroblasts proliferated fastest on MP530M and Petridish controls; there were no statistical differences in cell numbersfor cell populations grown on MP530M and culture dishes. Pro-liferation of fibroblasts on both MP530B and the Tecoflex controlswas significantly lower than that of MP530M or culture dishes; nostatistical difference was found between cell proliferation rates onthe MP530B and Tecoflex. As qualitatively shown in Fig. 6A, the lowcell proliferation rate of fibroblasts on MP530B is most likely at-tributed to the existence of negative sulfonic acid groups. In cellculture, the sulfonic acid on the materials’ surface can be ionized byculture media (pH¼ 7.4), and because the flexibility of the polymerchains, the ionized sulfonic groups can form a thin negativelycharged layer on the substrate surface. This layer may repel thefibroblast attachment by the electrostatic interaction. Protein ad-sorption on these charged polymers may also be low, but has notbeen investigated in this study. The similar reason may also beused to explain the low cell proliferation on Tecoflex controls,which is made from MDI, poly(tetramethylene oxide) (PTMO) andbutane diol (BD). The soft segment PTMO in Tecoflex containssufficient amount of oxygen atoms that provide rich negativeelectrons, the electron-rich soft segment can accumulate on thesubstrate surface due to the polymer chains’ flexibility, thus formthe similar electrostatic interaction as that of MP530B. As a com-parison, MP530M exhibits good cell proliferation rate during 9days culture, this can also be explained by cell–material electro-static interaction, chain flexibility and hydrophilicity. In MP530M,the tertiary nitrogen atoms in chain extender MIDE can form cat-ionic groups in the water and give a slightly positive polarity on thepolymer chains. In culture media the polymer chains can changethe conformation to form a positively charged layer on the sub-strate, which will interact with negatively charged cell membranesurface and most serum proteins, thus promote cell attachmentand proliferation.

Blood compatibility of a biomaterial can be reflected by theability to prevent or reduce the formation of thromboemboli in

0

25000

50000

75000

100000

125000

150000

Day 3 Day 5 Day 7 Day 9

MP530BMP530MPertridishTecoflex

Ce

lls

n

um

be

r (/w

ell)

A

100 um

MP530B MP530M TecoflexPetridish

Day 5

Day 7

Day 9

B

Fig. 6. Cytocompatiblity assay for the MP530B and MP530M membranes. Petri dish and Tecoflex membranes were employed as control. (A) Quantitative assays of cell attachmentand proliferation on MP530B, MP530M, Petri dish control and Tecoflex control. Data represent the mean of three samples (P< 0.05). (C) Phase contrast micrographs of NIH 3T3fibroblasts grown on the monolayer of each type of polymers at three time points. Cells on four types of polymer membranes exhibited spread patterns at each time point, whileMP530M and Petri dish exhibited higher cell attachment and proliferation than that of MP530B and Tecoflex.

C. Zhang et al. / Biomaterials 29 (2008) 3781–3791 3787

cardiovascular environment, which is largely influenced by plateletadhesion [41,42]. Therefore, a platelet adhesion study was con-ducted to investigate one aspect of the material’s blood compati-bility. In this study, the amount of LDH released from the lysedplatelets that adhered on the surface of each material was mea-sured. The LDH assay for the Tecoflex and MP530B exhibited similarabsorbances with 0.013� 0.002 and 0.014� 0.004, respectively,and the LDH assay for the MP530M exhibited significantly higherUV absorbance (0.028� 0.007). These results indicate that MP530Band Tecoflex show much lower platelet adhesion than that of

MP530M. Platelet adhesion on materials is a complicated phe-nomenon including protein adsorption, protein deformation andreaction, and therefore protein adsorption is thought to regulate allthe subsequent blood–material interactions [43]. Protein adsorp-tion experiments will be carried out in another study to furtherreveal the blood compatibility of MP530B. Tecoflex has been usedas a cardiovascular biomaterial for decades, and we rationalize thatdue to the similarity in the elasticity and anti-platelet adhesionproperty MP530B may find potential applications in cardiovasculartissue engineering.

C

A

4mm

4mm

pH > 8.7

pH < 8.7

PU

SO3HPU

SO3-

(Water insoluble) (Water soluble)

B

Fig. 7. Photograph of the patterns printed on glass slides by inkjet printing technique.A 50% (V/V) acetic acid water solution in HP 21 cartridge was printed out by one passon MP530B solution coated glass slide. Patterns immediately formed in 1 s afterprinting. (A) Separate letters printed on the stand glass slides; (B) contacted O-ringsprinted on the standard glass slides. All the patterns were computer designed byMicrosoft Office software. The letters fixed on the right side of the standard glass slideswere used as size control; (C) schematic diagram shows that MP530B in the solutionwas neutralized and became soluble with pH higher than 8.7, this reaction is revisablewith solution pH value adjusted lower than 8.7. This reaction was utilized for thescaffold fabrication by inkjet printer.

C. Zhang et al. / Biomaterials 29 (2008) 3781–37913788

3.5. Scaffold fabrication by inkjet printing technology and cellculture on the scaffold

To build scaffolds by inkjet printing technique, a HP inkjetprinter was modified as described previously [21]. The standardglass slides in the chamber of the printer were evenly coated witha 2.5% (W/V) ionized PU solution. The ink cartridge was filled with50% (V/V) HAc solution, which can be printed onto selected areas ofthe PU solution according to a pre-designed pattern. Upon impactof the droplets the hydrogen ions in HAc solution reacted withpendent sulfonic ions on the polymer chains according to thechemical reaction shown in Fig. 7C, causing the polymer to pre-cipitate. This reaction is reversible, as the pattern was observed tobe gradually re-ionized and dissolved into water at pH higher than8.7. However, at physiological pH the patterns remain solid. Asshown in Fig. 7, different letters (CLEMSON) and rings could beprinted on identically computer-designed patterns, indicating thatthe MP530B can be easily used for the fabrication of the scaffold.MP530B can also be adapted to fabricate three-dimensional pat-terns using the method described earlier [21], which will bereported separately. Fig. 8 shows the microstructure of the printedpatterns by SEM observation. The patterns exhibited porous three-dimensional morphology. The primary structure at low magnifi-cation showed the number of pores with size between 10 mm and30 mm, which could allow cell penetration. On the inner wall ofthose large size pores, the secondary structure at high magnifica-tion showed numerous pores with average size less than 100 nmevenly distributed and interconnected. This interconnection rep-resents a great advantage for cell growth in the scaffold as it wouldbe able to help the nutrients’ and growth factors’ transportation,

Fig. 8. SEM images of the PU scaffold showing, (A) and (B) primary structures at different magnifications show the porous structure with pore size ranging from 10 mm to 30 mm. (C)Secondary structure shows the porous structure with uniform and inter-connective pores, the average size of pores is less than 1 mm. (A) Magnification 400�, (B) magnification2500�, (C) magnification 20,000�.

Fig. 9. Microscopic examination of the fibroblast proliferation on the printed O-ring scaffold after 5 days culture, (A) Micrograph of the hydrated O-ring scaffold by phase contrast microscopy, the picture exhibited the porous structure. Thepore size observed by optical microscopy was between 10 mm and 150 mm; (B) Fluorescent micrograph of the fibroblast proliferation on the porous O-ring scaffold. Cells were stained in green by calcein AM, and the red area indicated theporous scaffold. The penetration and the proliferation of the fibroblast into the porous scaffold were observed.

C.Zhanget

al./Biom

aterials29

(2008)3781–3791

3789

C. Zhang et al. / Biomaterials 29 (2008) 3781–37913790

gas penetration, signal transmission among cells distributed in thelarge size pores. The formation of this particular primary and sec-ondary porous scaffold can be explained by surface gelling mech-anism [21], in which the surface of the HAc droplets immediatelyreacted with PU solution, forming a capsule or the first shell aroundthe droplets, thus the big size pores formed at this step. The HAcinside the pores diffused across the capsule membrane and con-tinued to react with the solution causing more PU precipitation;thus, the diffusion of the HAc through the first shell then resulted inthe second shell. As this shell formation and ion diffusion contin-ued, porous and layered structures of shells were formed. It hasbeen reported that size of ink droplet can be influenced duringprinting by cartridge orifice size, cartridge–substrate working dis-tance, the ink surface tension and the surface ink interfacial prop-erties [15,21,44,45]. Therefore, the large pore size in the primarystructure of the scaffold can potentially be adjusted by changing thedrop size using any of the above methods; while the small pore sizein the secondary structure can be adjusted by changing the HAcand/or polymer solution concentration. Our lab is studying therelationship among those factors to obtain more functionalscaffolds.

Fig. 9 shows the microstructure of the printed O-ring patterns byoptical microscopic examination after 5-day fibroblast culture.Compared to the SEM observation at high magnification, at 10�magnification a number of pores with size between 10 mm and150 mm can be observed, this higher average pore size may becaused by the hydration of the scaffold during cell culture and un-even distribution of the pores. After 5 days culture, fibroblasts werefound to attach and proliferate on the porous scaffold, and a numberof the cells penetrated into the pores with size 20 mm or higher.

4. Conclusion

Our purpose was to develop a cell compatible, degradable andsynthetic polymer for the novel bioprinting area that shows goodmechanical properties and biocompatibility. We successfully syn-thesized a water-soluble polyurethane elastomer that showeda rapid phase change by incorporation of PCL and sulfonic acid intoPU chains. The polymer proved to be degradable, possesses goodanticoagulation properties, and pH sensitive in water solution. Dueto its pH sensitivity, this PU becomes water soluble at pH higherthan 8.7 and insoluble at pH less than 8.7. This property was utilizedto fabricate two-dimensional scaffolds by printing image-basedpatterns. The scaffolds exhibited primary and secondary porousmicrostructures, which may allow the cell attachment and pro-liferation. Furthermore, the good elasticity and anti-platelet adhe-sion property of this PU, combined with inkjet printing technique,may further extend the inkjet printing technique to vascular tissueengineering area.

Acknowledgements

This work is partially supported by NSF grant no. EEC 0609035.The authors greatly acknowledge Dr. John Desjardin, Dr. Robert A.Latour, Ms. Cassie Gregory, Mr. Fenghai Guo, Mr. Sahil Jalota and Mr.Sivaraman Balakrishnan for technical support, Mr. Nathan Brownfor manuscript proofreading.

References

[1] Lamba NM, Woodhouse KA, Cooper SL. Polyurethanes in biomedical applica-tions. Boca Raton, USA: CRC Press; 1998.

[2] Patrick V, Hans JG, Laroche Gaetan, Robert G. Biomedical applications ofpolyurethanes. Austin, TX: Landes Bioscience; 2001.

[3] Engelberg I, Kohn J. Physico-mechanical properties of degradable polymersused in medical applications: a comparative study. Biomaterials 1991;12(3):292–304.

[4] Ambrosio AM, Allcock HR, Katti DS, Laurencin CT. Degradable poly-phosphazene/poly(alpha-hydroxyester) blends: degradation studies. Bio-materials 2002;23(7):1667–72.

[5] Leong KW, Brott BC, Langer R. Bioerodible polyanhydrides as drug-carriermatrices. I: characterization, degradation, and release characteristics. J BiomedMater Res 1985;19(8):941–55.

[6] Ma L, Gao C, Mao Z, Zhou J, Shen J, Hu X, et al. Collagen/chitosan porousscaffolds with improved biostability for skin tissue engineering. Biomaterials2003;24(26):4833–41.

[7] Gisselfalt K, Edberg B, Flodin P. Synthesis and properties of degradablepoly(urethane urea)s to be used for ligament reconstructions. Bio-macromolecules 2002;3(5):951–8.

[8] Yang TF, Chin WK, Cherng JY, Shau MD. Synthesis of novel biodegradablecationic polymer: N,N-diethylethylenediamine polyurethane as a gene carrier.Biomacromolecules 2004;5(5):1926–32.

[9] Zhang J, Doll BA, Beckman EJ, Hollinger JO. A biodegradable polyurethane-ascorbic acid scaffold for bone tissue engineering. J Biomed Mater Res A 2003;67(2):389–400.

[10] Chen KY, Kuo JF, Chen CY. Synthesis, characterization and platelet adhesionstudies of novel ion-containing aliphatic polyurethanes. Biomaterials 2000;21(2):161–71.

[11] Silver JH, Lewis KB, Ratner BD, Cooper SL. Effect of polyol type on the surfacestructure of sulfonate-containing polyurethanes. J Biomed Mater Res 1993;27(6):735–45.

[12] Silver JH, Marchant JW, Cooper SL. Effect of polyol type on the physicalproperties and thrombogenicity of sulfonate-containing polyurethanes.J Biomed Mater Res 1993;27(11):1443–57.

[13] Nojiri C, Kuroda S, Saito N, Park KD, Hagiwara K, Senshu K, et al. In vitrostudies of immobilized heparin and sulfonated polyurethane using epifluor-escent video microscopy. ASAIO J 1995;41(3):M389–94.

[14] Silver JH, Hart AP, Williams EC, Cooper SL, Charef S, Labarre D, et al. Anticoag-ulant effects of sulphonated polyurethanes. Biomaterials 1992;13(6):339–44.

[15] Boland T, Ovsianikov A, Chickov BN, Doraiswamy A, Narayan RJ, Yeong WY.Rapid prototyping of artificial tissues and medical devices. Adv Mater Process2007;165(11):51–3.

[16] Yeong WY, Chua CK, Leong KF, Chandrasekaran M. Rapid prototyping in tissueengineering: challenges and potential. Trends Biotechnol 2004;22(12):643–52.

[17] Yeong WY, Chua CK, Leong KF, Chandrasekaran M, Lee MW. Comparison ofdrying methods in the fabrication of collagen scaffold via indirect rapid pro-totyping. J Biomed Mater Res B Appl Biomater 2007;82(1):260–6.

[18] Yeong WY, Chua CK, Leong KF, Chandrasekaran M, Lee MW. Indirect fabrica-tion of collagen scaffold based on inkjet printing technique. Rapid PrototypingJ 2006;12(4):229–37.

[19] Dhariwala B, Hunt E, Boland T. Rapid prototyping of tissue-engineering con-structs, using photopolymerizable hydrogels and stereolithography. TissueEng 2004;10(9–10):1316–22.

[20] Hollister SJ. Porous scaffold design for tissue engineering. Nat Mater 2005;4(7):518–24.

[21] Boland T, Xu T, Damon B, Cui X. Application of inkjet printing to tissue engi-neering. Biotechnol J 2006;1(9):910–7.

[22] Varghese D, Deshpande M, Xu T, Kesari P, Ohri S, Boland T. Advances in tissueengineering: cell printing. J Thorac Cardiovasc Surg 2005;129(2):470–2.

[23] Xu T, Gregory CA, Molnar P, Cui X, Jalota S, Bhaduri SB, et al. Viability andelectrophysiology of neural cell structures generated by the inkjet printingmethod. Biomaterials 2006;27(19):3580–8.

[24] Fedorovich NE, De Wijn JR, Verbout AJ, Alblas J, Dhert WJ. Three-dimensionalfiber deposition of cell-laden, viable, patterned constructs for bone tissueprinting. Tissue Eng Part A 2008;14(1):127–33.

[25] Khalil S, Nam J, Sun W. Multi-nozzle deposition for construction of 3D bio-polymer tissue scaffolds. Rapid Prototyping J 2005;11(1):9–17.

[26] Liu CZ, Xia ZD, Han ZW, Hulley PA, Triffitt JT, Czernuszka JT. Novel 3D collagenscaffolds fabricated by indirect printing technique for tissue engineering.J Biomed Mater Res B Appl Biomater 2008;85(2):519–28.

[27] Vermette P. Biomedical applications of polyurethanes. Georgetown, TX; Aus-tin, TX: Landes Bioscience. Available from, Eurekah.com; 2001.

[28] Xu T, Petridou S, Lee EH, Roth EA, Vyavahare NR, Hickman JJ, et al. Con-struction of high-density bacterial colony arrays and patterns by the ink-jetmethod. Biotechnol Bioeng 2004;85(1):29–33.

[29] Jyoji I. Formation and reaction of polyenesulfonic acid. I. Reaction of poly-ethylene films with SO3. J Polym Sci Part A Polym Chem 1988;26(1):167–76.

[30] Anders F, Greger O, Per J. Ionic interactions and transport in a low-molecular-weight model polymer electrolyte. J Chem Phys 1998;108(17):7426–33.

[31] Okkema AZ, Visser SA, Cooper SL. Physical and blood-contacting properties ofpolyurethanes based on a sulfonic acid-containing diol chain extender. J Bi-omed Mater Res 1991;25(11):1371–95.

[32] Koberstein J, Leung L. Compression-molded polyurethane block copolymers. II.Evaluation of microphase compositions. Macromolecules 1992;23:6205–13.

[33] Couchman P. Prediction of glass-transition temperature for compatible blendsformed from homopolymers of arbitrary degree of polymerization. Compo-sitional variation of glass-transition temperatures. Macromolecules 1980;13:1272–6.

[34] Blasi P, D’Souza SS, Selmin F, DeLuca PP. Plasticizing effect of water on poly-(lactide-co-glycolide). J Controlled Release 2005 Nov 2;108(1):1–9.

[35] Zhang C, Zhang N, Wen X. Improving the elasticity and cytophilicity of bio-degradable polyurethane by changing chain extender. J Biomed Mater Res BAppl Biomater 2006;79(2):335–44.

C. Zhang et al. / Biomaterials 29 (2008) 3781–3791 3791

[36] Dunn MG, Avasarala PN, Zawadsky JP. Optimization of extruded collagen fibersfor ACL reconstruction. J Biomed Mater Res 1993;27(12):1545–52.

[37] Healy KE, Thomas CH, Rezania A, Kim JE, McKeown PJ, Lom B, et al. Kinetics ofbone cell organization and mineralization on materials with patterned surfacechemistry. Biomaterials 1996;17(2):195–208.

[38] Clark P, Connolly P, Moores GR. Cell guidance by micropatterned adhesivenessin vitro. J Cell Sci 1992;103(Pt 1):287–92.

[39] Webb K, Hlady V, Tresco PA. Relative importance of surface wettability andcharged functional groups on NIH 3T3 fibroblast attachment, spreading, andcytoskeletal organization. J Biomed Mater Res 1998;41(3):422–30.

[40] Kowalczynska HM, Nowak-Wyrzykowska M. Modulation of adhesion,spreading and cytoskeleton organization of 3T3 fibroblasts by sulfonic groupspresent on polymer surfaces. Cell Biol Int 2003;27(2):101–14.

[41] Sefton MV, Gemmell CH, Gorbet MB. What really is blood compatibility?J Biomater Sci Polym Ed 2000;11(11):1165–82.

[42] Kim YH, Han DK, Park KD, Kim SH. Enhanced blood compatibility of polymersgrafted by sulfonated PEO via a negative cilia concept. Biomaterials 2003;24(13):2213–23.

[43] Lelah MD, Lambrecht LK, Young BR, Cooper SL. Physicochemical character-ization and in vivo blood tolerability of cast and extruded biomer. J BiomedMater Res 1983;17(1):1–22.

[44] Mironov V, Boland T, Trusk T, Forgacs G, Markwald RR. Organ printing:computer-aided jet-based 3D tissue engineering. Trends Biotechnol 2003;21(4):157–61.

[45] Wilson Jr WC, Boland T. Cell and organ printing 1: protein and cell printers.Anat Rec A Discov Mol Cell Evol Biol 2003;272(2):491–6.