RESEARCH ARTICLE

nature biotechnology • VOLUME 20 • JUNE 2002 • http://biotech.nature.com602

A tough biodegradable elastomerYadong Wang1, Guillermo A. Ameer1,2, Barbara J. Sheppard3, and Robert Langer1*

Biodegradable polymers have significant potential in biotechnology and bioengineering. However, for someapplications, they are limited by their inferior mechanical properties and unsatisfactory compatibility with cellsand tissues. A strong, biodegradable, and biocompatible elastomer could be useful for fields such as tissueengineering, drug delivery, and in vivo sensing. We designed, synthesized, and characterized a toughbiodegradable elastomer from biocompatible monomers.This elastomer forms a covalently crosslinked, three-dimensional network of random coils with hydroxyl groups attached to its backbone. Both crosslinking and thehydrogen-bonding interactions between the hydroxyl groups likely contribute to the unique properties of theelastomer. In vitro and in vivo studies show that the polymer has good biocompatibility. Polymer implantsunder animal skin are absorbed completely within 60 days with restoration of the implantation sites to theirnormal architecture.

Biodegradable polymers have significant potential in various fieldsof bioengineering, such as tissue engineering, drug delivery, and in vivo sensing1,2. Because many biomedical devices are implanted ina mechanically dynamic environment in the body, the implants mustsustain and recover from various deformations without mechanicalirritations to the surrounding tissues. In many cases, the matricesand scaffolds of these implants would ideally be made of biodegrad-able polymers whose properties resemble those of the extracellularmatrix (ECM), a soft, tough, and elastomeric proteinaceous networkthat provides mechanical stability and structural integrity to tissuesand organs. Hence a soft biodegradable elastomer that readily recov-ers from relatively large deformations is advantageous for maintain-ing the implant’s proper function without mechanical irritation tothe host. Three classes of biodegradable elastomers have beenreported: hydrogels3,4, elastin-like peptides5–7, and polyhydroxyalka-noates (PHAs)8,9. Here we report a tough and inexpensivebiodegradable elastomer with excellent biocompatibility. This elas-tomer is much tougher than most hydrogels3,4. Compared withelastin-like peptides made by bacterial fermentation10–12, it is inex-pensive, nonimmunogenic, and endotoxin-free. The elastomer iscapable of much larger reversible deformations than a reportedlyelastomeric PHA (see Results and discussion), poly-4-hydroxybu-tyrate (P4HB).

This biodegradable elastomer is analogous to vulcanized rubber inthat it forms a crosslinked, three-dimensional network of randomcoils (it is this characteristic that is considered to give vulcanizedrubber its elasticity)13. Such a strategy to achieve tough and elas-tomeric materials is also found in nature: collagen and elastin, themajor fibrous protein components of ECM, are both crosslinked14.In addition to covalent crosslinking, hydrogen-bonding interactionsthrough hydroxyproline hydroxyl groups also contribute to themechanical strength of collagen14,15. This relationship is manifestedin various diseases in which the strength of collagen fibers decreasesmarkedly when crosslink density decreases considerably, or when theproduction of hydroxyproline is interrupted. Collagen fibers canusually recover from deformations of ∼ 20% (refs 16, 17), far greater

than the most prevalent biodegradable polymers: poly(lactide),poly(glycolide), and their copolymers (PLGA)18,19. The deformationis also greater than that of P4HB (∼ 10%). The design of our polymeris based on two hypotheses: (i) Good mechanical properties could beobtained through covalent crosslinking and hydrogen-bondinginteractions (comparison of the in vivo degradation profiles of thispolymer and PLGA will be published elsewhere); (ii) rubberlike elas-ticity could be obtained by building a three-dimensional network ofrandom coils through copolymerization in which at least onemonomer is trifunctional.

To realize this design, we considered five criteria: (i) as a degra-dation mechanism, we chose hydrolysis to minimize individual dif-ferences in degradation characteristics caused by enzymes2; (ii) toprovide a hydrolyzable chemical bond, we chose ester for its estab-lished and versatile synthetic methods20; (iii) because a high degreeof crosslinking usually leads to rigid and brittle polymers, we want-ed a low density of crosslinking; (iv) we chose crosslink chemicalbonds to be hydrolyzable and identical to those in the backbone tominimize the possibility of heterogeneous degradation; (v) werequired that specific monomers be nontoxic, at least one be tri-functional, and at least one provide hydroxyl groups for hydrogenbonding. We chose glycerol (CH2(OH)CH(OH)CH2OH), thebasic building block for lipids, as the alcohol monomer because itsatisfies all three requirements. From the same toxicological and polymer chemistry standpoints, we chose sebacic acid(HOOC(CH2)8COOH) as the acid monomer. Sebacic acid is thenatural metabolic intermediate in ω-oxidation of medium- tolong-chain fatty acids21–24. It has been shown to be safe in vivo25.The US Food and Drug Administration has approved glycerol and polymers containing sebacic acid for medical applications.Furthermore, sebacic acid has the appropriate chain length. Short-chain dicarboxylic acids are more acidic and more likely to cyclizeduring polymerization reactions, whereas long-chain dicarboxylicacids are more hydrophobic and mix poorly with glycerol. Bothmonomers are inexpensive, an advantage for large-scale applica-tions. We anticipated that this approach would yield biodegradable

1Department of Chemical Engineering, 77 Massachusetts Avenue, Massachusetts Institute of Technology, Cambridge, MA 02139. 2Current address: BiomedicalEngineering Department, Northwestern University, Evanston, IL 60208. 3Division of Comparative Medicine, 77 Massachusetts Avenue, Massachusetts Institute of

Technology, Cambridge, MA 02139. *Corresponding author (rlanger@mit.edu).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

RESEARCH ARTICLE

http://biotech.nature.com • JUNE 2002 • VOLUME 20 • nature biotechnology 603

polymers with improved mechanical properties and biocompati-bility. The polymer is referred to as poly(glycerol-sebacate) (PGS).

Results and discussionPolycondensation of glycerol and sebacic acid yields a transparent,almost colorless elastomer (note that under different conditions, arigid, totally crosslinked polymer has been synthesized from glyceroland sebacic acid in the molar ratio 2:3)26. The resulting polymer fea-tures a small number of crosslinks and hydroxyl groups directlyattached to the backbone. The intense C=O stretch at 1,740 cm–1 inFourier-transformed infrared (FTIR) spectrum confirms the forma-tion of ester bonds (Supplementary Fig. 1 online).

FTIR also shows a broad, intense OH stretch at 3,448 cm–1, indicat-ing that the hydroxyl groups are hydrogen bonded. The polymersurface is very hydrophilic because of the hydroxyl groups attachedto its backbone. Its water-in-air contact angle is 32.0°, almost iden-tical to that of a flat 2.7-nm-thick type I collagen film (31.9°)27. Thepolymer is insoluble in water, and swells 2.1 ± 0.33% after soaking

in water for 24 h. Elemental analysis confirms the composition ofthe polymer as approximately 1:1 glycerol/sebacic acid (calculatedfor C13H22O5: C, calculated 60.47—found 60.46; H, calculated8.53—found 8.36). The crosslinking density is expressed by n(moles of active network chains per unit volume), which is 38.3 ±3.40 mol/m3, and Mc, the relative molecular mass betweencrosslinks, which is 18,300 ± 1,620, calculated from the followingequation28:

n = E0/3RT = ρ/Mc

where E0 is Young’s modulus, R is the universal gas constant, T isthe temperature, and ρ is the density. Differential scanningcalorimetery (DSC) measurement showed two crystallization tem-peratures at –52.14°C and –18.50°C, and two melting temperaturesat 5.23°C and 37.62°C, respectively. No glass transition tempera-ture was observed above –80°C, which is the lower detection limitof the instrument. The DSC results indicate that the polymer istotally amorphous at 37°C. Similar to vulcanized rubber, this elas-tomer is a thermoset polymer. However, the uncrosslinked pre-polymer can be processed into various shapes, because it can bemelted into liquid and is soluble in common organic solvents,such as 1,3-dioxolane, tetrahydrofuran, ethanol, isopropanol, andN,N-dimethylformamide. We have prepared PGS sheets and foamswith these methods. Briefly, a mixture of NaCl particles of appro-priate size and an anhydrous 1,3-dioxolane solution of the prepoly-mer is poured into a PTFE mold. The polymer is cured in the mold

Figure 1. Stress–strain curves of PGS, vulcanized rubber, and P4HB.Both PGS and vulcanized rubber are marked by low modulus and largeelongation ratio, indicating elastomeric and tough materials. The highmodulus and low yield strain (∼ 10%) of P4HB indicate a stiffer material.UTS, ultimate tensile strength.

Figure 2. Comparison of NIH 3T3 fibroblast cell morphology and number.PGS sample wells (A) and PLGA control wells (B), six days after seeding.The PGS wells had more adherent cells and the cell morphologyappeared normal, while those in the control well adopted a long, thin,threadlike shape. Scale bar = 200 µm.

Figure 3. Comparison of growth rate of NIH 3T3 fibroblast cells in PGS(�) wells and PLGA (�) wells. MTT absorption measured at 570 nm,normalized value shown.

Figure 4. Change of thickness of the immune responses with time forPGS and PLGA. The inflammatory response decreased with time for bothpolymers, which had similar inflammatory-zone thickness. However, thethickness of fibrous capsules surrounding PLGA was consistently andsignificantly larger than that of PGS. Inflammatory zone: PGS (�); PLGA(�). Fibrous capsule: PGS (�); PLGA (�).

A B

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

RESEARCH ARTICLE

nature biotechnology • VOLUME 20 • JUNE 2002 • http://biotech.nature.com604

in a vacuum oven at 120°C and 100 mtorr. A porous scaffold isobtained after salt leaching in deionized water.

Tensile tests on thin strips of PGS reveal a stress–strain curvecharacteristic of an elastomeric and tough material (Fig. 1). Thenonlinear shape of the tensile stress–strain curve is typical for elas-tomers, and resembles those of ligament29–31 and vulcanized rub-ber32. Compared with hard and brittle materials, which have highmodulus (the initial slope of the stress–strain curve) and low strain(relative deformation), PGS can be elongated repeatedly to at leastthree times its original length without rupture. The total elongationis unknown, because grip breaks occurred at 267 ± 59.4% strain.The tensile Young’s modulus of the polymer is 0.282 ± 0.0250 MPa,indicating a soft material. The ultimate tensile strength is >0.5 MPa,the point at which the PGS strips broke from the grip of themechanical tester. P4HB, a reportedly elastomeric degradable PHA(refs 8,9), has a strain to failure value of 11.1 ± 0.491% and aYoung’s modulus of 253 ± 5.29 MPa, similar to that of low-densitypolyethylene. The ultimate tensile strength is 10.4 ± 0.554 MPa.Overall, P4HB has a much higher modulus (stiffer) and much lowerstrain to failure compared with either PGS or vulcanized rubber.The value of the Young’s modulus of PGS is between those of liga-ments (kPa scale)29–31, which contain a large amount of elastin inaddition to collagen, and tendon (GPa scale)16,17,33, which is mainlymade of collagen. The strain to failure of PGS is similar to that ofarteries and veins (up to 260%)34, and much larger than that of ten-dons (up to 18%)35. After soaking for 24 h in water, the weight ofPGS barely changes, and the mechanical properties are virtually thesame as dry polymer.

PGS appears to be biocompatible both in vitro and in vivo. NIH3T3 fibroblast cells were seeded homogeneously on PGS-coatedglass Petri dishes with PLGA-coated dishes as controls. The cells inPGS sample wells are viable and showed normal morphology with

higher growth rate than the control, as tested by MTT assay36

(Figs 2A, 3). Cells in PLGA wells tend to form clusters, and there is agreater number of floating cells; furthermore, most of the attachedcells adopted a long, thin, threadlike morphology (Fig. 2B). Theseexperiments suggested that PGS is at least as biocompatible as PLGAin vitro.

Subcutaneous (s.c.) implantation in Sprague–Dawley rats wasused to compare the in vivo biocompatibility of PGS and PLGA.These PGS and PLGA implants have the same surface area/volumeratio (1.33 ± 0.04). Both PGS and PLGA samples were implantedsymmetrically into the back of the same animal. The inflammatoryresponses subsided with time for both polymer implants. In the firstthree weeks, the inflammatory response of PLGA implantation siteswas ∼ 16% thinner than that of PGS sites (Figs 4, 5). The thickness ofthe inflammatory zone in both implantation sites was approximate-ly the same at weeks 4 and 5. Fibrous capsules (thick avascular colla-gen layer) surrounding PLGA implants developed within 14 days,and their thickness remained ∼ 140 µm. Collagen deposition did notappear around PGS implants until 35 days post implantation. Thecollagen layer was highly vascularized and was only ∼ 45 µm thick.The inflammatory response and fibrous capsule formation observedfor PLGA are similar to those reported in the literature37,38. Thickfibrous capsules block mass transfer between the implants and sur-rounding tissues, which can impair implant functions. In an in vivostudy with PGS alone, the s.c. implantation sites were undetectabledespite repeated sectioning of the specimens at multiple levels in 60days (two implantation sites each in three animals). The implantswere completely absorbed without granulation or formation of scartissues, and the implantation site was restored to its normal histolog-ic architecture. Overall, the inflammatory response of PGS is similarto that of PLGA. However, unlike PLGA, PGS induces little, if any,fibrous capsule formation.

Figure 5. Photomicrographsof rat skin. Comparisons oflumen wall characteristics(H&E, 10×) and fibrouscapsule thickness (insets,MTS, 5×) at implantation sitesacross time. (A, C, E) PGS;(B, D, F) PLGA. (A, B) At 7 days postimplantation (p.i.),the lumenal wall wasmarkedly thickened by a zoneof dense vascular proliferationand mild inflammation withoutdetectable collagendeposition. (C, D) At 21 daysp.i., the lumenal wall wassignificantly thinner with amodest degenerativeinflammatory infiltrateimmediately adjacent to thepolymer. The PLGAimplantation site (D) wasmarked by a significantcollagen fibrous capsule,which was absent in PGS.(E, F) At 35 days p.i., thelumenal wall was reduced to athin zone of cell debris with novascular proliferation.Collagen deposition in PGSimplantation site was muchthinner than that surroundingthe fragmented PLGAimplant. Top, skin; blank area,implantation site; scale bar =200 µm.

A B

D

C

E F

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

RESEARCH ARTICLE

http://biotech.nature.com • JUNE 2002 • VOLUME 20 • nature biotechnology 605

We examined the degradation characteristics of PGS both in vitroand in vivo. Agitation for 60 days in PBS solution at 37°C caused thepolymer to degrade 17 ± 6%, as measured by change of dry sampleweight. In contrast, s.c. implants in rats were totally absorbed in 60 days.For the in vivo experiment, enzymes, and perhaps macrophages aswell, may have caused differences in degradation rate. In vivo degra-dation thinned the polymer implants, with the explants maintainingtheir square shape and relatively sharp edges up to at least 35 days.Preliminary data indicate that mechanical strength likely decreaseslinearly with mass loss (comparison of the in vivo degradation pro-files of this polymer and PLGA will be published elsewhere). Bothresults suggest that PGS disks degrade predominantly through sur-face erosion. The preservation of integrity during the degradationprocess can be important for certain types of tissue-engineeredimplants, drug delivery devices, and in vivo sensors.

Compared with existing biodegradable elastomers, PGS appearsto be tougher, inexpensive, and more flexible. In the models tested,the material is biocompatible both in vitro and in vivo. The polymer’sproperties, such as hydrophilicity, and degradation rate and patterncan potentially be tailored by grafting hydrophobic moieties to thehydroxyl groups39,40. To provide further control or regulation ofpolymer interaction with cells, biomolecules could be coupled to thehydroxyl groups or integrated into the polymer backbone41–43.

Experimental protocolSynthesis and characterization of the polymer. The polymer was synthesizedby polycondensation of 0.1 mol each of glycerol (Aldrich, Milwaukee, WI)and sebacic acid (Aldrich) at 120°C under argon for 24 h before the pressurewas reduced from 1 torr to 40 mtorr over 5 h. The reaction mixture was keptat 40 mtorr and 120°C for 48 h. A KBr pellet of newly prepared polymer wasused for FTIR analysis on a Nicolet Magna-IR 550 spectrometer. A Perkin-Elmer DSC differential scanning calorimeter was used for DSC measure-ment. Elemental analysis on vacuum-dried samples was performed by QTI(Whitehouse, NJ). The water-in-air contact angle was measured at roomtemperature using the sessile-drop method and an image analysis of the dropprofile with VCA2000 video contact angle system on slabs of polymer fixedon glass slides.

Mechanical properties. Tensile tests were conducted on six 25 × 5 × 0.7 mmpolymer strips cut from polymer sheets according to ASTM standard D 412-98a on an Instron 5542 mechanical tester equipped with a 50 N load cell.Vulcanized rubber and P4HB (Metabolix, Cambridge, MA) strips (25 × 5 × 0.5mm) were cut from polymer sheets. Deflection rate was kept at 50 mm/min.The samples were elongated to failure.

In vitro degradation. Slabs of dry polymer (5 × 5 × 2 mm) were weighed andtransferred to 15 ml centrifuge tubes (Falcon, Bedford, MA) filled with PBS(pH 7.4; Gibco, Carlsbad, CA). After 60 days, the samples were removed andwashed with deionized water. The surface water was removed with Kimwipe,

and the samples were weighed after drying at 40°C in an oven for seven days.The degree of degradation was determined by dry-weight change.

In vitro biocompatibility. Nine glass Petri dishes (60 mm diameter) werecoated with 1,3-dioxolane solution of the prepolymer (1%). The coateddishes were transferred into a vacuum oven after evaporation of the solventin air. The prepolymer was crosslinked into the elastomer after 24 h at120°C and 120 mtorr. Nine control dishes were coated with 1% CH2Cl2

solution of PLGA (50:50, carboxyl ended, relative molecular mass 15,000(Mr 15K), Boehringer Ingelheim, Ingelheim, Germany), and the solvent wasevaporated for 24 h in air. The coated dishes were sterilized by UV radiationfor 15 min. Each dish was soaked in growth medium for 4 h, replaced withfresh medium, and soaked for 4 h before cell seeding to remove any unre-acted monomers or residual solvents. Each dish was seeded with 100,000NIH 3T3 fibroblast cells and 8 ml of growth medium. The cells were incu-bated at 37°C with 5% CO2. Cell density was measured by MTT assay44.Medium exchange was carried out every 48 h. At day 6, phase contrastimages were taken for both the polymer wells and the control wells on aZeiss Axiovert 200 microscope equipped with a Dage 240 digital camera.

In vivo biocompatibility. Autoclaved PGS slabs of approximately 6 × 6 × 3 mm,and ethylene oxide–sterilized PLGA (50:50, carboxyl ended, Mr 15K,Boehringer Ingelheim) disks (2 mm thick, 12.5 mm diameter) were implant-ed s.c. in 15 seven-week-old female Sprague–Dawley rats (Charles RiverLaboratories, Wilmington, MA) by blunt dissection under deep isoflu-rane–O2 general anesthesia. The animals were cared for in compliance withthe regulations of MIT and the NIH45. The surface area/volume ratio was keptthe same for both PGS and PLGA implants. Two implants each of PGS andPLGA were implanted symmetrically on the upper and lower back of thesame animal. Every implantation site was marked by two tattoo marks 2 cmaway from the implantation center. The animals were randomly divided intofive groups. At each predetermined time point (7, 14, 21, 28, and 35 days), onegroup of rats was killed and tissue samples (∼ 15 × 15 mm) surrounding theimplants were harvested with the intact implant. The samples were fixed in10% formalin for 24 h and embedded in paraffin after a series of dehydrationsteps in ethanol and xylenes. The slides were stained with hematoxylin andeosin (H&E) and Masson’s trichrome stain (MTS). At each time point, 12 slidesfor each polymer were obtained. All histologic preparations were assessed bya pathologist (B.J.S.), who was not informed of the identity of the polymerimplant in each slide. The thickness of the inflammatory zone (H&E) andcollagen deposition (MTS) for each polymer implant is expressed as the aver-age value of three readings per slide of six slides at each time point.

Note: Supplementary information is available on the Nature Biotechnologywebsite.

AcknowledgmentsThe authors thank Dr. David LaVan and Dr. Daniel Anderson for advice anddiscussions. We appreciate comments from Dr. David LaVan, Prof. HiroyukiIjima, and Ms. Sheryl Villa on the manuscript. This work was supported byNIH grant 5-R01-HL60435.

Received 21 September 2001; accepted 25 February 2002

1. Peppas, N.A. & Langer, R. New challenges in biomaterials. Science 263,1715–1720 (1994).

2. Langer, R. Biomaterials: status, challenges, and perspectives. AIChE J. 46,1286–1289 (2000).

3. Lee, K.Y. et al. Controlling mechanical and swelling properties of alginate hydro-gels independently by cross-linker type and cross-linking density.Macromolecules 33, 4291–4294 (2000).

4. Temenoff, J.S., Athanasiou, K.A., LeBaron, R.G. & Mikos, A.G. Effect of poly(eth-ylene glycol) molecular weight on tensile and swelling properties ofoligo(poly(ethylene glycol)fumarate) hydrogels for cartilage tissue engineering. J.Biomed. Mater. Res. 59, 429–437 (2002).

5. van Hest, J.C.M. & Tirrell, D.A. Protein-based materials, toward a new level ofstructural control. Chem. Comm. 19, 1897–1904 (2001).

6. Welsh, E.R. & Tirrell, D.A. Engineering the extracellular matrix: a novel approachto polymeric biomaterials. I. Control of the physical properties of artificial proteinmatrices designed to support adhesion of vascular endothelial cells.Biomacromolecules 1, 23–30 (2000).

7. Urry, D.W. et al. Elastic protein-based polymers in soft tissue augmentation andgeneration. J. Biomater. Sci., Polym. Ed. 9, 1015–1048 (1998).

8. Poirier, Y., Nawrath, C. & Somerville, C. Production of polyhydroxyalkanoates, afamily of biodegradable plastics and elastomers, in bacteria and plants.Bio/Technology 13, 142–150 (1995).

9. Sodian, R. et al. Fabrication of a trileaflet heart valve scaffold from a polyhydrox-yalkanoate biopolyester for use in tissue engineering. Tissue Eng. 6, 183–187(2000).

10. Katsarava, R. et al. Amino acid–based bioanalogous polymers. Synthesis andstudy of regular poly(ester amide)s based on bis(α-amino acid) α,ω-alkylenediesters, and aliphatic dicarboxylic acids. J. Polym. Sci. Part A: Polym. Chem. 37,391–407 (1999).

11. Kemnitzer, J. & Kohn, J. In Handbook of biodegradable polymers (eds Domb, A.J.,Kost, J. & Wiseman, D.M.) 251–272 (Harwood Academic Publishers, Amsterdam;1998).

12. Petsch, D. & Anspach, F.B. Endotoxin removal from protein solutions. J.Biotechnol. 76, 97–119 (2000).

13. Erman, B. & Mark, J.E. In Science and technology of rubber (eds Mark, J.E.,Burak, E. & Eirich, F.R.) 189–210 (Academic Press, San Diego, CA; 1994).

14. Voet, D. & Voet, J.G. Biochemistry (John Wiley & Sons, New York, 1995).15. Stryer, L. Biochemistry (W.H. Freeman and Company, New York; 1995).16. Fratzl, P. et al. Fibrillar structure and mechanical properties of collagen. J. Struct.

Biol. 122, 119–122 (1998).17. Wang, J.L., Parnianpour, M., Shirazi-Adl, A. & Engin, A.E. Failure criterion of col-

lagen fiber: viscoelastic behavior simulated by using load control data. Theor.Appl. Fract. Mech. 27, 1–12 (1997).

18. Storey, R.F., Warren, S.C., Allison, C.J. & Puckett, A.D. Methacrylate-endcapped

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om

RESEARCH ARTICLE

nature biotechnology • VOLUME 20 • JUNE 2002 • http://biotech.nature.com606

poly(d,l-lactide-co-trimethylene carbonate) oligomers. Network formation by ther-mal free-radical curing. Polymer 38, 6295–6301 (1997).

19. Helminen, A., Korhonen, H. & Seppala, J.V. Biodegradable crosslinked polymersbased on triethoxysilane terminated polylactide oligomers. Polymer 42,3345–3353 (2001).

20. March, J. Advanced organic chemistry (John Wiley & Sons, New York; 1992).21. Liu, G., Hinch, B. & Beavis, A.D. Mechanisms for the transport of α,ω-dicarboxy-

lates through the mitochondrial inner membrane. J. Biol. Chem. 271,25338–25344 (1996).

22. Grego, A.V. & Mingrone, G. Dicarboxylic acids, an alternate fuel substrate in par-enteral nutrition: an update. Clin. Nutr. 14, 143–148 (1995).

23. Mortensen, P.B. & Gregersen, N. The biological origin of ketotic dicarboxylicaciduria. In vivo and in vitro investigations of the omega-oxidation of C6-C16-monocarboxylic acids in unstarved, starved and diabetic rats. Biochim. Biophys.Acta 666, 394–404 (1981).

24. Mortensen, P.B. C6–C10-dicarboxylic aciduria in starved, fat-fed and diabetic ratsreceiving decanoic acid or medium-chain triacylglycerol. An in vivo measure ofthe rate of β-oxidation of fatty acids. Biochim. Biophys. Acta 664, 349–355 (1981).

25. Tamada, J. & Langer, R. The development of polyanhydrides for drug deliveryapplications. J. Biomater. Sci. Polym. Ed. 3, 315–353 (1992).

26. Nagata, M. et al. Synthesis, characterization, and enzymatic degradation of net-work aliphatic copolyesters. J. Polym. Sci. Part A: Polym. Chem. 37, 2005–2011(1999).

27. Dupont-Gillain, C.C., Nysten, B. & Rouxhet, P.G. Collagen adsorption onpoly(methyl methacrylate): net-like structure formation upon drying. Polym. Int.48, 271–276 (1999).

28. Sperling, L.H. Introduction to physical polymer science (John Wiley & Sons, NewYork; 1992).

29. Yamaguchi, S. Analysis of stress–strain curves at fast and slow velocities of load-ing in vitro in the transverse section of the rat incisor periodontal ligament follow-ing the administration of β-aminopropionitrile. Arch. Oral Biol. 37, 439–444 (1992).

30. Komatsu, K. & Chiba, M. The effect of velocity of loading on the biomechanicalresponses of the periodontal ligament in transverse sections of the rat molar invitro. Arch. Oral Biol. 38, 369–375 (1993).

31. Chiba, M. & Komatsu, K. Mechanical responses of the periodontal ligament in thetransverse section of the rat mandibular incisor at various velocities of loading invitro. J. Biomech. 26, 561–570 (1993).

32. Nagdi, K. Rubber as an engineering material: guideline for users (Hanser,Munich;1993).

33. Misof, K., Rapp, G. & Fratzl, P. A new molecular model for collagen elasticity basedon synchrotron X- ray scattering evidence. Biophys. J. 72, 1376–1381 (1997).

34. Lee, M.C. & Haut, R.C. Strain rate effects on tensile failure properties of the com-mon carotid artery and jugular veins of ferrets. J. Biomech. 25, 925–927 (1992).

35. Haut, R.C.The effect of a lathyritic diet on the sensitivity of tendon to strain rate. J.Biomech. Eng. 107, 166–174 (1985).

36. Northup, S.J. & Cammack, J.N. In Handbook of biomaterials evaluation (ed. vonRecum, A.F.) 325–339 (Taylor & Francis, Philadelphia; 1999).

37. Cadee, J.A., Brouwer, L.A., den Otter, W., Hennink, W.E. & Van Luyn, M.J.A. Acomparative biocompatibility study of microspheres based on crosslinked dextranor poly(lactic-co-glycolic)acid after subcutaneous injection in rats. J. Biomed.Mater. Res. 56, 600–609 (2001).

38. van der Elst, M., Klein, C.P.A.T., de Blieck-Hogervorst, J.M., Patka, P. & Haarman,H.J.T.M. Bone tissue response to biodegradable polymers used for intramedullaryfracture fixation: a long-term in vivo study in sheep femora. Biomaterials 20,121–128 (1999).

39. Jayachandran, K.N. & Chatterji, P.R. Synthesis of dense brush polymers withcleavable grafts. Eur. Polym. J. 36, 743–749 (2000).

40. Laschewsky, A., Rekai, E.D. & Wischerhoff, E. Tailoring of stimuli-responsivewater- soluble acrylamide and methacrylamide polymers. Macromol. Chem. Phys.202, 276–286 (2001).

41. Barrera, D.A., Zylstra, E., Lansbury, P.T., Jr. & Langer, R. Synthesis and RGD pep-tide modification of a new biodegradable copolymer: poly(lactic acid-co-lysine). J.Am. Chem. Soc. 115, 11010–11011 (1993).

42. West, J.L. & Hubbell, J.A. Polymeric biomaterials with degradation sites for pro-teases involved in cell migration. Macromolecules 32, 241–244 (1999).

43. Mann, B.K., Gobin, A.S., Tsai, A.T., Schmedlen, R.H. & West, J.L. Smooth musclecell growth in photopolymerized hydrogels with cell adhesive and proteolyticallydegradable domains: synthetic ECM analogs for tissue engineering. Biomaterials22, 3045–3051 (2001).

44. Hansen, M.B., Nielsen, S.E. & Berg, K. Re-examination and further developmentof a precise and rapid dye method for measuring cell growth/cell kill. J. Immunol.Methods 119, 203–210 (1989).

45. National Institutes of Health. Principles of Laboratory Animal Care (NIH pub. no.85-23, rev. 1985).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://b

iote

ch.n

atu

re.c

om