Acta Biomaterialia 7 (2011) 3277–3284

Contents lists available at ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /actabiomat

Melt electrospinning of biodegradable polyurethane scaffolds

Ari Karchin a, Felix I. Simonovsky a, Buddy D. Ratner a,b, Joan E. Sanders a,⇑a Department of Bioengineering, Box 355061, University of Washington, Seattle, WA 98195, USAb Department of Chemical Engineering, Box 351750, University of Washington, Seattle, WA 98195, USA

a r t i c l e i n f o

Article history:Received 28 February 2011Received in revised form 26 April 2011Accepted 16 May 2011Available online 20 May 2011

Keywords:Aliphatic polyurethanesBiodegradable polymerMelt electrospinningTissue engineeringScaffolds

1742-7061/$ - see front matter � 2011 Acta Materialdoi:10.1016/j.actbio.2011.05.017

⇑ Corresponding author. Tel.: +1 206 221 5872; faxE-mail addresses: jsanders@uw.edu, jsanders@u.w

a b s t r a c t

Electrospinning from a melt, in contrast to from a solution, is an attractive tissue engineering scaffoldmanufacturing process as it allows for the formation of small diameter fibers while eliminatingpotentially cytotoxic solvents. Despite this, there is a dearth of literature on scaffold formation via meltelectrospinning. This is likely due to the technical challenges related to the need for a well-controlledhigh-temperature setup and the difficulty in developing an appropriate polymer. In this paper, abiodegradable and thermally stable polyurethane (PU) is described specifically for use in melt electros-pinning. Polymer formulations of aliphatic PUs based on (CH2)4-content diisocyanates, polycaprolactone(PCL), 1,4-butanediamine and 1,4-butanediol (BD) were evaluated for utility in the melt electrospinningprocess. The final polymer formulation, a catalyst-purified PU based on 1,4-butane diisocyanate, PCL andBD in a 4/1/3 M ratio with a weight-average molecular weight of about 40 kDa, yielded a nontoxic poly-mer that could be readily electrospun from the melt. Scaffolds electrospun from this polymer containedpoint bonds between fibers and mechanical properties analogous to many in vivo soft tissues.

� 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

A number of polymers have been used in melt electrospinning.However, the use of polyurethanes (PUs) in melt-electrospunscaffolds is not well represented in the literature despite theattractive properties of this class of polymers [1]. The seminal meltelectrospinning work of Larrondo and Manley [2–4] focused onpolyethylene and polypropylene. Other polymers that have beenmelt-electrospun include polyester [5], poly(lactic acid) [6],poly(ethylene glycol) [7], and a blend of poly(ethylene oxide-block-caprolactone) with polycaprolactone (PCL) [8]. PUs containproperties that are superior to many commonly used polymersfor tissue engineering scaffolds [9]. They can be designed such thatthe finished material is thermally stable, degradable, nontoxic andwith tunable mechanical properties [10–12]. A subclass of PUs,segmented linear elastomeric biodegradable urethane blockcopolymers consisting of alternating backbone hard and softsegments, are attractive for tissue engineering, primarily due totheir molecular design flexibility, which provides a broad rangeof properties, including biodegradability. Most commerciallyavailable linear PU block copolymers are based on aromatic diiso-cyanates such as 4,40-diphenylmethane diisocyanate and toluenediisocyanate [11]. Although several medical devices composed ofPU elastomers containing aromatic cycles in their hard segmentshave been approved by the regulatory authorities in past years

ia Inc. Published by Elsevier Ltd. A

: +1 206 616 2509.ashington.edu (J.E. Sanders).

for medical applications, these are not biodegradable formulationsand the products of decomposition (if, indeed, decomposition oc-curs) may be toxic [13,14]. Some groups have worked to developless toxic degradable aliphatic diisocyanate-based PUs [15–17];however, such a polymer has yet to be melt electrospun.

The goal of the current work was to develop a novel degradableand biocompatible aliphatic PU that could be formed into scaffoldsvia melt electrospinning. The PU formulations investigated in thisstudy were based on various combinations of PCL diol, with anumber-average molecular weight (MW) of about 1250 Da, 2,6-diisocyanato methyl caproate (lysine diisocyanate, LDI), 1,4-butanediisocyanate (BDI), 1,4-butanediamine (BDA) (putrescine)and 1,4-butanediol (BD). PU copolymers based on (CH2)4-contentdiisocyanates, such as LDI and BDI, are expected to be nontoxicsince the product of hydrolytic degradation, BDA, is nontoxic[18]. Similarly, the use of PCL as the soft segment is common dueto its biocompatibility and biodegradability [10]. The polymercomposition and weight-average MW (Mw) were optimized formelt electrospinning. The final polymer was evaluated for cytotox-icity, and the mechanical properties of electrospun scaffolds weredetermined.

2. Materials and methods

2.1. Polymer synthesis

All chemicals were purchased from Aldrich Chemical Company,Inc. (Milwaukee, WI), unless otherwise noted. The synthesized

ll rights reserved.

3278 A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284

polymers were made via the standard procedure of multi-stepaddition polymerization in N,N-dimethylacetamide (DMAc) sol-vent, described elsewhere [22,23]. This general method allowsPUs with a variety of compositions to be synthesized with preci-sion and reproducibility [19,20]. The polymerization pathway canbe briefly described by three main steps:

� The reaction between PCL and a variable molar excess of isocy-anate (NCO) groups of either LDI or BDI yields intermediatesend-capped with isocyanate groups (prepolymers).� Either BDA or BD with a variable molar excess of amino (NH2) or

hydroxyl (OH) groups was used as chain extenders to provide apolymerization with the isocyanate prepolymer in DMAc to cre-ate a low MW PU end-capped with either NH2 or OH groups. Areaction between aliphatic NCO groups of the prepolymer andNH2 groups of BDA was carried out at room temperature inDMAc. Dibutyltin dilaurate (DBTDL) served as the catalystaccelerating the reaction between aliphatic NCO groups of theprepolymer and OH groups of BD at approximately 80 �C inDMAc.� The low MW PU was then combined with a monomeric diisocy-

anate in a gradual approach to stoichiometry [21,22] to yield ahigh MW polymer of specific Mw.

The simplified scheme of the urethane polymerization processbased on PCL, BDI and BD is presented in Fig. 1.

2.2. Polymer purification

In order to remove methanol-soluble catalyst after the last syn-thesis step, the polymer was precipitated in an excess of methanolat room temperature. After filtration, remaining polymer waswashed three more times in fresh methanol at room temperature(12 h washing steps). These steps were followed by two 12 h rinsesin fresh, circulating deionized (DI) water to remove the methanol

Fig. 1. Simplified synthesis scheme of a thermally s

and DMAc. The polymer was finally dried under vacuum at approx-imately 60–65 �C for at least 12 h.

2.3. Electrospinning

Scaffolds were electrospun using a custom apparatus describedpreviously [23]. Briefly, the apparatus consists of a melt chambermounted on an enclosure that housed a collecting surface andelectrode. The melt chamber was made up of a metal sleeve thatterminated in a 0.5 mm diameter nozzle. A pair of band heaterswere secured around the melt chamber to heat the as-receivedpurified and dried polymer. A grounding rod was placed insidethe polymer which terminated at the external surface of a cus-tom-made poly(ether ether ketone) cap screwed onto the meltchamber. The positive electrode was mounted below the collectionsurface. The collection surface, an adhesive-backed copper stripaffixed to a glass microscope slide, was mounted on a three-axisstage (404150XR, Parker Daedal, Harrison City, PA) that movedthe collector in concentric circles beneath the electrospinningnozzle. A high-voltage power supply (RP-50-100, Del, Valhalla,NY) supplied a potential of 30 kV between the grounding rod andthe positive electrode.

The distance between the positive and ground electrodes wasset at 13 cm, and the melt temperature set at 220–240 �C for allruns. The reason a 30 kV voltage and a 13 cm electrode distancewere used was that they provided an electric field of sufficientstrength to form fibers into an electrospun mat on the collector.A lower electric field strength was not strong enough to overcomethe surface tension of the polymer droplet [24], while a higherelectric field strength increased the bending instability of the elec-trospun fiber. For a given applied voltage, a larger distance be-tween electrodes did not result in the formation of a Taylor coneand subsequent fiber formation, while a shorter electrode distancecaused the fiber to miss the target collector.

After electrospinning, samples were immediately transferred toa 60 mm Petri dish, sealed with Parafilm and placed in a desiccator

table polyurethane based on BDI, PCL and BD.

Table 1Electrospun fiber appearance related to polymer composition.

Polymer Code Composition and molar ratio Hard segment %wt. Soft segment %wt. Appearance

PU-1 PCL/LDI/BDA = 1/2/1 29.05 70.95 Weak, yellowed fibersPU-2 PCL/LDI/BDA = 1/4/3 46.95 53.05 Robust, slightly yellowed fibersPU-3 PCL/LDI/BDA = 1/8/7 64.78 35.22 Electrospray and dripsPU-4 PCL/BDI/BDA = 1/4/3 39.38 60.62 Unable to melt polymerPU-5 PC 1 BDI BD = 1/4/3 3522 64.78 Good quality fibers

A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284 3279

until cytotoxicity testing began. Electrospun scaffolds werephotographed with a 6.1 megapixel digital camera (D70S, Nikon,Melville, NY).

2.4. Cytotoxicity

Cytotoxicity screening was performed on electrospun mats toevaluate their suitability as a biomaterial. Cytotoxicity testingwas based on the United States Pharmacopeia in vitro biologicalreactivity ‘‘elution test’’ [25]. This test determines the reactivityof leachates from polymeric materials to mammalian cell cultures.Included in the test is a description of standardized grades, 0–4,that correlate cell morphology to a description of reactivity.According to the test specifications, a grade of 0 correlates to an ab-sence of cell lysis, while a grade of 4 correlates to near-totaldestruction of the cell layer.

Prior to cytotoxicity testing, electrospun samples were washedand sterilized. Briefly, samples were placed in a dilute solution ofRBS 35 detergent and sonicated for 30 min. The samples were thenrinsed with DI water and sonicated in DI water several times usingfresh water each time to remove most or all of the detergent. Sam-ples were dried overnight in a biological safety cabinet and, oncedried, sterilized using ultraviolet (UV) light. Samples were irradi-ated for 30 min on each side at a distance of 14 cm beneath thelamp, which was previously shown to be sufficient for sterilizationof electrospun fibers [26]. Subsequently, samples were incubatedat 37 �C and 5% CO2 for 24 h in Dulbecco’s modified Eagle’s med-ium (DMEM) at a weight-to-volume ratio of 4 g per 20 ml to obtainthe polymer extract. High-density polyethylene and latex surgicalgloves served as the negative and positive reference controls,respectively.

Human fetal foreskin fibroblast cells (SCRC-1041, ATCC, Manas-sas, VA) were cultured at 37 �C and 5% CO2. Cells were plated at2 � 105 cells well–1 in 6-well tissue culture plates or 5 � 104 cellswell–1 in 24-well tissue culture plates. When a cell monolayerwas achieved, the medium was aspirated and replaced with com-plete DMEM containing extracts from the test sample and controlpreparations. For the 6- and 24-well culture plates, 0.8 and 0.1 mlof extracts were added, respectively. After 48 h in culture, cellmonolayers were examined microscopically (Microphot-SA, NikonInc., Tokyo, Japan) for the presence of morphological changes, areduction in cell density or lysis. Samples were tested in triplicate.

2.5. Melt electrospinning and molecular weight

The change in Mw from the melt electrospinning process wasdetermined by computing the percent change in Mw values be-tween the as-received polymer and the electrospun scaffold usinggel permeation chromatography (GPC). A series of four new poly-mers with variable Mw based on the component molar ratio ofPU-5 (Table 1) were made via the standard procedure of solutionpolymerization in DMAc described above. During the synthesis,four aliquots of the polymer solutions were withdrawn from thereaction flask at the last step in various precalculated equivalent

OH/NCO group ratios: 1.15/1 (PU-5-1), 1.10/1 (PU-5-2), 1.05/1(PU-5-3) and 1.00/1 (PU-5-4). Each sample was purified separatelyfollowing the procedure described above.

Electrospinning was performed at the temperature required tomelt each polymer. In a pilot test, the temperature in the meltchamber was raised from 200 �C until a droplet of polymer formedand was held in place at the nozzle. The surface tension in the poly-mer maintained the droplet’s shape and position. The maintenanceof the droplet shape and position was a critical precursor to Taylorcone and fiber formation once the electric field was applied. Thetemperature necessary to melt the polymer and form a dropletwas considered the appropriate melt electrospinning temperature.

Three electrospun samples for each polymer with different Mw

values were used for MW measurements. Using statistical software(SPSS, Chicago, IL), a one-way independent analysis of variance andpost hoc tests (Bonferroni, Tukey, and Games-Howell) were per-formed to determine the statistical significance (p < 0.05) of differ-ences in MW between polymers.

2.6. Gel permeation chromatography

In preparation for GPC analysis, polymer samples were dis-solved in HPLC-grade N,N-dimethylformamide (DMF) then filteredwith a disposable 0.2 lm syringe filter. GPC analysis was per-formed using three Tosoh TSK-GEL columns (TSK-a ± 3000,a ± 3000, a ± 4000) connected in series to a Viscotek GPCmax(VE2001, Viscotek, Houston, TX) and Viscotek RI and UV detectors(VE3580 and VE3210, respectively, Viscotek). The mobile phasewas DMF containing 0.1 wt.% LiBr. Mw distributions were deter-mined relative to a series of poly(methyl methacrylate) standards.

2.7. Fiber diameter

Mean fiber diameter was determined by measuring individualfibers based on unbiased stereological techniques. To accomplishthis measurement, five equidistant 12 lm sections were cut fromscaffolds embedded in optimal cutting temperature embeddingmedium in triplicate. Sections were digitized with an uprightmicroscope at � 2 by taking random 0.5 mm2 images. Images wereimported into Image J software (National Institutes of Health,Bethesda, MD) for the analysis of the fiber diameter with a dissec-tor probe. This method yielded more than 200 measurements perscaffold.

2.8. Cell seeding

A primary cell culture was established from anterior cruciateligament fibroblasts. Ligaments were aseptically harvested fromfreshly sacrificed adolescent (3–5 months) Yorkshire pigs. In a bio-safety cabinet, ligaments were minced to facilitate cell outgrowthaccording to Protocol 11.4 of Freshney [14]. Minced tissue was cul-tured in cell culture medium in an incubator at 37 �C with ahumidified atmosphere of 5% CO2. The medium consisted of DMEM(D5796, Sigma–Aldrich, St. Louis, MO) supplemented with 10%

3280 A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284

fetal bovine serum (SH3008802, Fisher Scientific, Pittsburg, PA), 1%L-glutamine (G7513, Sigma–Aldrich), 1% 100X antibiotic/antimi-crobial (15240–062, Invitrogen, Carlsbad, CA) and Hepes buffer(H0887, Sigma–Aldrich). Fibroblast migration typically took 5–7 days, during which the medium was changed every 3 days. Oncecells approached 70% confluence, the minced tissue was removedand then cells were replated onto tissue culture flasks. Cells frompassage 3 were resuspended in medium containing 10% dimethylsulfoxide and stored at -70 �C in liquid N2 until use.

Prior to cell seeding, electrospun scaffolds were washed andsterilized as described above. Fibroblasts from passage 6 wereseeded onto fibronectin-coated scaffolds at a concentration of1 � 106 cells ml–1. Seeded scaffolds were cultured for 4 weeks priorto fixation with Karnovsky’s fixative (15720, Electron MicroscopySciences, Hatfield, PA).

2.9. Scanning electron microscopy

High-resolution scanning electron microscopy (HRSEM; Sirion,FEI Company, Eindhoven, The Netherlands) was used to visualizethe sample morphology and fiber bonding. The HRSEM micro-graphs were taken using a voltage of 5 kV for the electrospun sam-ples. Prior to visualization, samples were sputter-coated for 30 s toform an approximately 10 nm thick gold layer in order to preventcharging.

2.10. Mechanical testing

The scaffolds were tested following ASTM standard D5035(Standard Test Method for Breaking Force and Elongation of TextileFabrics – Strip Method). Samples were cut into rectangular teststrips with a width of 5 mm and a length of 26 mm. Five verticallyoriented samples were subjected to a tensile test in an Instron test-ing machine (5543, Instron, Norwood, MA) at room temperature ata strain rate of 300 mm min–1 until sample failure. For strain calcu-lations, the initial gauge length was taken as the grip-to-grip dis-tance. For stress calculations, the bulk thickness of each samplewas measured in three locations with a digital micrometer whilethe sample was at zero displacement. Measurements were takenat both ends and the center of the sample gauge length. The meanof these three measurements constituted the bulk scaffold thick-ness. To compute cross-sectional area accurately, the bulk widthand thickness measurements were multiplied by the fiber areafraction. The fiber area fraction was determined with a samplingtechnique similar to that used to measure the fiber diameters.The area fraction was determined using Image J. The elastic modu-lus was computed from the linear region of the stress–strain curve,defined as the portion that gave the best fit linear regression(R2 > 0.99).

Fig. 2. Electrospun polymers: (A) PU-1; (B) PU-2; (C) PU-3; (D) PU-4; (E) PU-5. Scalebar = 10 mm.

3. Results and discussion

The ability to form tissue engineering scaffolds with PUs isadvantageous due to their wide acceptance as biomedical poly-mers, excellent mechanical properties and biocompatibility [27].In the present study, biodegradable segmented linear PU blockcopolymers with a backbone consisting of two types of segmentsbased on three constituents, oligodiol, diisocyanate and a difunc-tional chain extender, were synthesized.

3.1. Polymer formulation and fiber diameter

In order to determine an optimal biodegradable polymer com-position with a segmental concentration suitable for melt electros-pinning, three polymers were synthesized and electrospun in a

pilot study. The ideal polymer must possess an appropriate meltingtemperature, melt viscosity at the temperature of electrospinningand high thermal stability, and also be a good fiber-forming com-position under electrospinning conditions.

To melt electrospin successfully, the polymer compositionshould be designed to have an appropriate temperature ‘‘gap’’ be-tween the melting temperature (Tm) and temperature of thermaldegradation (Td). In other words, the macromolecular structureshould be engineered for a Tm high enough to break hydrogenbonds between hard segments in adjacent chains but lower thanthe Td so that covalent chemical bonds in the polymer backboneare not cleaved.

In a pilot study, the polymers spanned a broad range of compo-sitions with increasing hard segment concentrations and corre-sponding molar ratios of PCL/LDI/BDA: 1/2/1 (PU-1), 1/4/3 (PU-2)and 1/8/7 (PU-3) (Table 1). It was expected that the PCL segmentcontributed both biodegradability and elasticity. We assumed thatdegradation products of such a polymer would be nontoxic due tothe polymer composition consisting of ester linkages (hydrolyzingto carboxylic acid and hydroxyl) in the PCL molecular chain, ureagroups in the LDI–BDA hard segment, and lysine-like structuresfor both LDI and BDA. We hypothesized that optimal mechanicalproperties of the polymers based on chosen components could befound in the concentration range of the hard segment betweenabout 30 and 65 wt.% and the corresponding concentration range

A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284 3281

of the soft segment between about 35 and 70 wt.%. PU-1, PU-2 andPU-3 were electrospun at melt temperatures of 220, 230 and240 �C, respectively.

Each polymer formulation yielded a different result. The elec-trospun mat from PU-1 was composed of yellowed fibers(Fig. 2A) that were clumped together. Minor mechanical manipula-tion readily separated the fibers and broke them apart. The signif-icant color change in this polymer was likely a result of thermaldegradation during the melt electrospinning process, attributedto the relatively high soft segment concentration. ElectrospunPU-3 was devoid of fibers, containing only drops on the collectionsurface (electrospray) (Fig. 2C), attributed to the high hard seg-ment fraction, thus making the polymer quite brittle. Of the threeinitial polymers, electrospun PU-2 was the most promising(Fig. 2B), though the yellowed polymer appearance was taken aspossible evidence of thermal degradation.

LDI was initially chosen because there is experimental evidencethat the pendant methyl ester group in LDI is rapidly hydrolyzed,followed by slower hydrolysis of urethane bonds in the backbonechain [28], and the product of its degradation has a nontoxic ly-sine-like structure. However, at the temperatures required for elec-trospinning, 220–240 �C, degradation of the LDI-based PUoccurred, as evidenced by the yellowing of the polymer (Fig. 2A–C).

This result prompted further exploration for a more thermallystable formulation. PU-4 was made with the same component mo-lar ratio as PU-2 and a similar segmental weight concentration, butbased on the use of BDI instead of LDI. with the assumption thatthe structure of the BDI–BDA hard segment would provide in-creased thermal stability. After 25 min at 300 �C, PU-4 did not melt,resulting in a polymer that was a deep orange/yellow color(Fig. 2D). The inability to melt was likely due to the structure ofthe BDI–BDA hard segments that are closely packed in the hard do-mains because BDI offers no steric hindrance from a pendant unitas in the LDI structure. This leads to stronger hydrogen bonding inthe BDI–BDA domains compared to LDI–BDA.

To improve the melt processing, a polymer with a lower Tm wassynthesized with BD replacing BDA as the chain extender, since thehydrogen bonding associated with BDI–BD urethane groups is

Fig. 3. Cytotoxicity results from: (A) positive control; (B) negative control; (C) electrospbar = 100 lm.

typically reduced compared to the urea groups of the BDI–BDA do-main. Thus, the final polymer formulation, PU-5, was based on PCLas an elastic segment, BDI as a fragment of the hard segment andBD as the chain extender (molar component ratio 1/4/3). Electro-spun PU-5 yielded a homogeneous fibrous scaffold with little orno visual evidence of thermal degradation (Fig. 2E). With the appli-cation of an electrostatic field, a continuous single fiber ejectedfrom the Taylor cone. No beads were observed within the scaffoldas a result of dripping or electrospray during electrospinning. Thispolymer was further evaluated in the remainder of this study.

The higher viscosities associated with melt electrospinning typ-ically yield fiber diameters that are orders of magnitude larger thanthose produced by solution electrospinning, which are mostly inthe submicron range. The relationship between viscosity anddiameter can be attributed to the interplay of forces involved inelectrospinning. The applied electrostatic field energizes the poly-mer, causing bending instability that violently whips the fiber,resulting in a reduction of the cross-section area. A more viscouspolymer will resist the bending instability due to increased fiberstiffness, thereby stabilizing the cross-sectional area. While ourmelt-electrospun fibers are relatively large compared to those fromsolution electrospinning, the mean diameter of melt-electrospunPU-5 fibers, 11.2 ± 2.3 lm, would be suitable for scaffolds intendedto replace many soft tissue structures. For example, the anteriorcruciate ligament contains collagen fibers in the 1–20 lm range[29].

3.2. Cytotoxicity

All positive controls resulted in a grade of at least 3 (moderatereactivity), while all negative controls resulted in a grade of 0 (noreactivity). Representative images of positive and negative controlsare shown in Fig. 3A and B, respectively. The sheet of cells exposedto the negative control eluent is intact and continuous, with spin-dle-shaped cells. In contrast, cells exposed to the positive controleluent form a discontinuous sheet. Many of the cells are non-viableand appear to be detached.

un PU-5 with DBTDL catalyst; (D) electrospun PU-5 without DBTDL catalyst. Scale

Table 2USP in vitro biological reactivity grades of as-received and electrospun PU polymers.

Polymer code Polymerstate

Samplegrade

Negativecontrol grade

Positivecontrol grade

PU-5 with DBTDLcatalyst

As-received 3 0 3Electrospun 3

PU-5 withoutDBTDL catalyst

As-received 0 0 4Electrospun 0

Fig. 4. Effect of electrospinning on polymer Mw. Mean + 1 standard deviation.

3282 A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284

Since a preliminary test for the cytotoxicity of one of our poly-mers was positive, residual DBTDL catalyst, having poor solubilityin water, was the suspect toxic agent. Before leaching out of thecatalyst, PU-5 with DBTDL in both the as-received and electrospunstates was found to be cytotoxic, achieving grade 3 reactivity. Cellsexposed to an eluent of pre-washed PU-5 (Fig. 3C) became roundedand detached, attaining a morphology similar to the positivecontrols. Based on this result, the polymer purification proceduredescribed in the materials and methods section was developed.PU-5 purified of catalyst was found to be nontoxic. Cells exposedto the eluent from this polymer exhibited no observable changein morphology between these samples and the negative controls(Fig. 3D). A summary of the graded results is shown in Table 2.

In vitro cytotoxicity testing has limitations. While in vitro cyto-toxicity results typically hold in the in vivo environment [27], thetest used in the present study should be considered an initialscreening for cytotoxicity. The United States Pharmacopeia (USP)in vitro elution protocol specifies cell exposure for 48 h, therebygiving no indication of long-term cell viability. The short-term testdoes not allow for assessment of degradation products from theelectrospun polymer. However, a longer-term cell culture alsoshowed no indication of cytotoxicity; it is described in more detailbelow.

3.3. Melt electrospinning and molecular weight

Ideally, polymers for melt electrospinning should not changeMw as a function of the processing temperature. The number ofchain entanglements is proportional to the polymer’s Mw, as isthe melt viscosity [30]. Chain entanglement affects the morphol-

Fig. 5. Scanning electron micrographs of electrospun scaffold using

ogy of fibers since viscosity is important to the formation of uni-form electrospun fiber. For example, a lower Mw can lead to theformation of beads on fibers [31].

Melt electrospinning had a significant effect on the Mw, withPU-5-1 showing the smallest change. The as-synthesized polymershad Mws of 39.7, 49.0, 54.0 and 66.8 kDa for PU-5-1, PU-5-2, PU-5-3 and PU-5-4, respectively. PU-5-4 could not be adequatelyelectrospun at any temperature up to 300 �C, just yielding beadsof yellowed polymer. Polymers PU-5-1, PU-5-2 and PU-5-3 weremelt-electrospun at 235, 240 and 247 �C, respectively. A signifi-cantly lower change (p < 0.05) in Mw after melt electrospinning ofPU-5-1 compared to PU-5-2 and PU-5-3 was noted (Fig. 4). There-fore, despite the relatively high Mw of as-synthesized PU-5-2 andPU-5-3, the greater reduction in Mw during melt electrospinningwill unfavorably impact the fiber morphology at longer electros-pinning times.

3.4. Morphological characterization and cell interaction

The fibers produced by melt electrospinning PU-5-1 were smoothand uniform, and lacked the beads-on-the-string morphology

PU-5-1: (A) top view; (B) cross-section; (C) top view magnified.

A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284 3283

sometimes observed in solution-electrospun fibers (Fig. 5A).Macroscopically, scaffolds are highly porous (Fig. 5B). The electro-spun fibers were neither focused on a particular spot, forming amound, nor did they form looped coils, which have been observedin melt-electrospun scaffolds [32]; rather, the fibers were evenlydistributed throughout the scaffold. No obvious bonds betweenfibers were apparent in the sample (Fig. 5C). However, manualmanipulation of the scaffolds and microscope observation leadsto the conclusion that fiber point bonding exists since deformationof an individual fiber would affect adjacent fibers [33]. If therewere no point bonding, individual fibers would pull apart uponapplication of strain. Point bonding is difficult to quantify in SEMimages since a true point bond will show virtually no alterationof fiber morphology at the junction.

To assess cell infiltration, and to further assess the potential forcytotoxic breakdown products, fibroblasts were grown on scaffoldsfor up to 4 weeks. Fibroblasts seeded onto electrospun scaffoldsadhered and were observed to assume different morphologies(Fig. 6). Some cells assumed a spindly morphology, elongatingand wrapping partway around fibers. Others stretched betweenadjacent fibers, extending multiple processes between them. Cellsinfiltrated throughout the 200–400 lm thick scaffolds.

Fig. 6. Fibroblasts in an electrospun scaffold.

Fig. 7. Representative stress–strain curve for electrospun PU-5-1.

3.5. Mechanical testing

A representative stress–strain curve for electrospun PU-5-1scaffolds is shown in Fig. 7. Tensile test data reveal mechanicalperformance similar to many in vivo soft tissues. The averageelastic modulus, ultimate tensile strength (UTS) and ultimatestrain were 1.8 ± 0.0 MPa, 36.7 ± 9.9 MPa and 221.5 ± 95.1%,respectively. These values fall between those of collagen andelastin [34]. This result is relevant since these proteins are themain contributors to the mechanical properties of soft tissue.The elastic modulus, UTS and ultimate strain of PU-5-1 scaffoldscould be adjusted in future formulations by altering the constitu-ents of the copolymer and/or the fiber architecture. Alignment offibers along different axes has been shown to induce changes inthe mechanical properties of an electrospun scaffold [35]. Futurestudies will modulate scaffold mechanical properties through thecontrol of fiber alignment.

4. Conclusion

A biodegradable, linear, segmented block polyurethane copoly-mer was designed specifically for preparing scaffolds via melt elec-trospinning. Polycaprolactone diol, 1,4-butane diisocyanate and1,4-butanediol in a 1/4/3 M ratio and an Mw of about 40 kDa (PU-5) yielded a useful polymer for this processing method. ExtractingDBTDL catalyst resulted in a nontoxic polymer suitable for tissueengineering applications. The resulting nontoxic scaffold is com-posed of smooth, robust fibers that lack beads and have mechanicalproperties comparable to many native tissues. This polymer isworth pursuing in future investigations on scaffold alignmentand its effect on cell and extracellular matrix alignment, with theultimate goal being the tailoring of the mechanical and biologicalproperties of a tissue engineering scaffold.

Author disclosure statement

No competing financial interests exist.

Acknowledgements

The authors are grateful to Dr. Weiping Zhang for assistancewith the cytotoxicity study. This study was supported by Univer-sity of Washington Engineered Biomaterials (UWEB) and NIHRO1 HD038554. In addition, A.K. was funded through the NIHEngineered Biomaterials Training Program (EBTP).

Appendix A. Figures with essential colour discrimination

Certain figures in this article, particularly Figures 2 and 3, aredifficult to interpret in black and white. The full colour imagescan be found in the on-line version, at doi:10.1016/j.actbio.2011.05.017.

References

[1] Bhardwaj N, Kundu SC. Electrospinning: a fascinating fiber fabricationtechnique. Biotechnol Adv 2010;28(3):325–47.

[2] Larrondo L, Manley RS. Electrostatic fiber spinning from melts I. Experimentalobservations on fiber formation and properties. Journal of Polymer Science1981;19:909–20.

[3] Larrondo L, Manley RS. Electrostatic fiber spinning from melts II. Examinationof the flow field in an electrically driven jet. Journal of Polymer Science1981;19:921–32.

[4] Larrondo L, Manley RS. Electrostatic fiber spinning from melts III. Electrostaticdeformation of pendant drop of polymer melt. Journal of Polymer Science1981;19:933–40.

[5] Kim J-S, Lee DS. Thermal properties of electrospun polyesters. Polymer Journal2000;32(7):616–8.

3284 A. Karchin et al. / Acta Biomaterialia 7 (2011) 3277–3284

[6] Zhou H, Green TB, Joo YL. The thermal effects on electrospinning of polylacticacid melts. Polymer 2006;47:7497–505.

[7] Dalton PD, Lleixa Calvet J, Mourran A, Klee D, Moller M. Melt electrospinning ofpoly-(ethylene glycol-block-epsilon-caprolactone). Biotechnol J2006;1(9):998–1006.

[8] Dalton PD, Klinkhammer K, Salber J, Klee D, Moller M. Direct in vitroelectrospinning with polymer melts. Biomacromolecules 2006;7(3):686–90.

[9] Rockwood DN, Woodhouse KA, Fromstein JD, Chase DB, Rabolt JF.Characterization of biodegradable polyurethane microfibers for tissueengineering. J Biomater Sci Polym Ed 2007;18(6):743–58.

[10] Heijkants RG, van Calck RV, van Tienen TG, de Groot JH, Buma P, Pennings AJ,et al. Uncatalyzed synthesis, thermal and mechanical properties ofpolyurethanes based on poly(epsilon-caprolactone) and 1, 4-butanediisocyanate with uniform hard segment. Biomaterials 2005;26(20):4219–28.

[11] Xie R, Bhattacharjee D, Argyropoulos J. Polyurethane elastomers based on 1, 3and 1, 4-bis(isocyanatomethyl)cyclohexane. Journal of Applied PolymerScience 2009;113:839–48.

[12] Santerre JP, Woodhouse K, Laroche G, Labow RS. Understanding thebiodegradation of polyurethanes: from classical implants to tissueengineering materials. Biomaterials 2005;26(35):7457–70.

[13] Marcos-Fernandez A, Abraham GA, Valentin JL, San Roman J. Synthesis andcharacterization of biodegradable non-toxic poly(ester-urethane-urea)s basedon poly(epsilon-caprolactone) and amino acid derivatives. Polymer2006;47(3):785–98.

[14] Freshney RI. Culture of Animal Cells: a Manual of Basic Technique andSpecialized Applications, 6th edn. Hoboken, NJ: John Wiley & Sons; 2010.

[15] Bruin P, Veenstra GJ, Nijenhuis AJ, Pennings AJ. Design and synthesis ofbiodegradable poly(ester-urethane) elastomer networks composed of non-toxic building-blocks. Makromolekulare Chemie-Rapid Communications1988;9(8):589–94.

[16] Skarja GA, Woodhouse KA. Synthesis and characterization of degradablepolyurethane elastomers containing and amino acid-based chain extender. JBiomater Sci Polym Ed 1998;9(3):271–95.

[17] Skarja GA, Woodhouse KA. In vitro degradation and erosion of degradable,segmented polyurethanes containing an amino acid-based chain extender. JBiomater Sci Polym Ed 2001;12(8):851–73.

[18] Benjamin M, Evans EJ, Copp L. The histology of tendon attachments to bone inman. J Anat 1986;149:89–100.

[19] Simonovsky FI, Wu Y, Golledge SL, Ratner BD, Horbett TA. Poly(etherurethane)s incorporating long alkyl side-chains with terminal carboxylgroups as fatty acid mimics: synthesis, structural characterization andprotein adsorption. J Biomater Sci Polym Ed 2005;16(12):1463–83.

[20] Wu Y, Simonovsky FI, Ratner BD, Horbett TA. The role of adsorbed fibrinogen inplatelet adhesion to polyurethane surfaces: a comparison of surfacehydrophobicity, protein adsorption, monoclonal antibody binding, andplatelet adhesion. J Biomed Mater Res A 2005;74(4):722–38.

[21] Kafengaus AP, Kazarinova NV, Nepyshnevsky VM, Simonovsky FI, GladkayaMA. Method for producing polyurethane solutions. SU Patent 640547; 1978.

[22] Simonovsky F, Zaplatin A, Samigullin F, Khudyak E. Method for producingpolyetherurethanes in solution. Soviet Union Patent 1746688; 1989.

[23] Mitchell SB, Sanders JE. A unique device for controlled electrospinning. JBiomed Mater Res A 2006;78(1):110–20.

[24] Zong XH, Kim K, Fang DF, Ran SF, Hsiao BS, Chu B. Structure and processrelationship of electrospun bioabsorbable nanofiber membranes. Polymer2002;43(16):4403–12.

[25] USP, Biological reactivity tests in vitro <87>. Rockville, MD: US Pharmacopeia,2000: p. 1832.

[26] Neumann T, Hauschka SD, Sanders JE. Tissue engineering of skeletal muscleusing polymer fiber arrays. Tissue eng 2003;9(5):995–1003.

[27] Ratner B, Hoffmann A, Schoen F, Lemons J, editors. Biomaterials Science: AnIntroduction to Materials in Medicine, 2nd edn. London: Elsevier AcademicPress, 2004. 879.

[28] Yamamoto N, Nakayama A, Oshima M, Kawasaki N. Aiba S-i. Enzymatichydrolysis of lysine diisocyanate based polyurethanes and segmentedpolyurethane ureas by various proteases. Reactive and Functional Polymers2007;67(11):1338–45.

[29] Duthon VB, Barea C, Abrassart S, Fasel JH, Fritschy D, Ménétrey J. Anatomy ofthe anterior cruciate ligament. Knee Surgery, Sports Traumatology,Arthroscopy 2005; 14(3): 204-213.

[30] Tan S-H, Inai R, Kotaki M, Ramakrishna S. Systemic parameter study for ultra-fine fiber fabrication via electrospinning process. Polymer 2005;46:6128–34.

[31] Gupta P, Elkins C, Long TE, Wilkes GL. Electrospinning of linear homopolymersof poly(methyl methacrylate): exploring relationships between fiberformation, viscosity, molecular weight and concentration in a good solvent.Polymer 2005;46(13):4799–810.

[32] Dalton PD, Grafahrend D, Klinkhammer K, Klee D, Moller M. Electrospinning ofpolymer melts: Phenomenological observations. Polymer 2007;48:6823–33.

[33] Johnson J, Ghosh A, Lannutti J. Microstructure-property relationships in atissue-engineering scaffold. J Appl Polym Sci 2007;104(5):2919–27.

[34] Bronzio J, editor The Biomedical Engineering Handbook, vol. 2. Boca Raton, FL:CRC Press LLC, 1999. 1656.

[35] Ayres CE, Bowlin GL, Pizinger R, Taylor LT, Keen CA, Simpson DG. Incrementalchanges in anisotropy induce incremental changes in the material propertiesof electrospun scaffolds. Acta Biomater 2007;3(5):651–61.