Biomaterials 20 (1999) 253—264

Tyrosine—PEG-derived poly(ether carbonate)s as new biomaterialsPart I: synthesis and evaluation

Chun Yu, Joachim Kohn*Department of Chemistry, Rutgers, The State University of New Jersey, New Brunswick, NJ 08903, USA

Received 26 March 1998; accepted 7 August 1998

Abstract

Tyrosine—PEG-derived poly(ether carbonate)s were prepared by condensation copolymerization with phosgene. The resultingpolymers were random copolymers with weight average molecular weights from 40 000 to 200 000 dalton. Chemical structure andpurity were confirmed by NMR and FTIR spectral analysis. General structure—property correlations were established. The glasstransition temperature decreased with increasing PEG content and increasing pendent chain length. When higher molecular weightPEG blocks were used, the glass transition temperature increased relative to identical polymers having shorter PEG blocks. Thetensile modulus increased with decreasing PEG content, decreasing pendent chain length, and when longer PEG blocks were used.Water uptake and the rate of backbone degradation increased with increasing PEG content. Microspheres could be prepared bysolvent evaporation techniques from copolymers with low PEG content. Release rate of pNA and FITC-dextran from themicrospheres increased with increasing PEG content. While tyrosine-derived polycarbonates were excellent substrates for cellattachment and growth, the presence of only 5 mol% of PEG

1000led to low or no cell attachment in short-term cell culture with both

rat lung fibroblasts and osteoblasts. The polymers were non-cytotoxic. ( 1999 Elsevier Science Ltd. All rights reserved

Keywords: Poly(ether carbonate); Biomaterial; Degradable; Structure—property correlations; Cell culture; Drug release

1. Introduction

Polymeric scaffolds for tissue reconstruction, develop-ment of artificial organs, degradable membranes for theprevention of surgical adhesions, and intra-arterial coat-ings for the prevention of thrombosis or restenosis areillustrative examples for new applications of medical im-plants [1—5]. These advanced applications require degra-dable biomaterials whose physicomechanical, chemical,and biological properties can be closely matched toa wide range of narrowly defined, application-specificrequirements.

So far, biomedical engineers and medical device de-signers have employed almost exclusively degradablepolyesters derived of a-hydroxy acids such as poly-(glycolic acid), poly(lactic acid), polydioxanone or vari-

*Correspondence address: Department of Chemistry, Rutgers Uni-versity, 610 Taylor Rd./Busch Campus, Piscataway, NJ 08854-8087,USA. Tel.: 732 445 3888; fax: 732 445 5006; e-mail: Kohn@rutchem.rut-gers.edu

ous copolymers thereof which are the only synthetic,degradable polymers with an extensive regulatory ap-proval history in the USA. Although the utility of thesematerials as sutures and in a number of drug deliveryapplications is well established, some material’s needscannot be satisfied by their use. For example, all poly-esters release acidic degradation products [6—8], limitingtheir utility to applications where acidity at the implantsite is not a concern. The above polyesters also tend to berelatively rigid, inflexible materials [9]. This can bea definite disadvantage when mechanical compliancewith soft tissue or blood vessels is required. Finally, noneof the above polyesters provides a chemically reactivependent chain for the easy attachment of drugs, cross-linkers, or biologically active moieties.

This laboratory has used tyrosine-derived diphenolicmonomers (Fig. 1) in the synthesis of polycarbonates[10, 11], polyiminocarbonates [12], and polyarylates[13, 14]. Among these new polymers, tyrosine-derivedpolycarbonates have been studied most extensively andhave been found to be tissue-compatible, strong, tough,hydrophobic materials that degrade slowly under

0142-9612/99/$—See front matter ( 1999 Elsevier Science Ltd. All rights reserved.PII: S 0 1 4 2 - 9 6 1 2 ( 9 8 ) 0 0 1 6 9 - 0

Fig. 1. Synthesis and structure of DTR-PEG-derived poly(ether carbonate)s: The tyrosine-derived diphenolic monomer DTR is reacted in thepresence of phosgene with a predetermined molar ratio of PEG to yield the target polymer, poly(DTR co fPEG

M8carbonate). R represents the type of

the alkyl ester pendent chain, ‘f ’ represents the percent molar fraction of poly(ethylene glycol) units present within the backbone.

physiological conditions [11, 15]. These materials appearto be most suitable for use in orthopaedic implants[16, 17].

Other medical applications such as drug delivery, non-thrombogenic coatings, vascular grafts, scaffolds forwound healing and artificial skin require materials thatare more hydrophilic, softer, and that degrade faster thanthe available tyrosine-derived polycarbonates. To ad-dress these needs, tyrosine-derived diphenolic monomerswere copolymerized with blocks of poly(ethylene glycol)(PEG), resulting in a new class of poly(ether carbonate)s.PEG has been copolymerized previously with difunc-tional monomers such as dicarboxylic acids [18—20],lactic acid [21, 22], and amino acids to develop polymersfor biomedical use [23, 24]. For example, aromaticpoly(ether ester)s prepared from PEG and ethyleneterephthalate (PEG—PET copolymer) [18] or butyleneterephthalate (PEG—PBT copolymer, ‘Polyactive'’)[19, 20] were investigated as blood compatible and bone-bonding biodegradable materials.

The investigation of poly(ether carbonate)s derivedfrom diphenols such as bisphenol-A (BPA) and PEGstarted in the 1960s. Merrill [25] introduced PEG blocksinto poly(bisphenol-A carbonate) by the interfacialcopolymerization of poly(bisphenol-A carbonate) andPEG bischloroformate. Later, Goldberg described theuse of PEG as a comonomer with bisphenol-A [26]. Theincorporation of PEG blocks into the aromatic polycar-bonate chains produced elastomers with remarkable ten-

sile strength and elongations and detailed studies oftheir properties were reported [27, 28]. Plasma proteinabsorption [29] and platelet adhesion [30] were investi-gated and the copolymers were suggested as hemodialy-sis membranes or plasma separators in the patentliterature [31, 32]. However, copolymers of diphenolsand PEG have so far not been studied as medical implantmaterials.

Here we report the introduction of PEG blocks intothe backbone of tyrosine-derived polycarbonates to formpoly(ether carbonate)s. The copolymers (Fig. 1) are refer-red to as poly(DTR co fPEG

M8carbonate) where R rep-

resents the type of alkyl ester pendent chain, ‘f ’ representsthe percent molar fraction of PEG units present withinthe backbone, and M

8represents the molecular weight of

the PEG blocks. Thus, poly(DTE co 5%PEG1000

car-bonate) refers to a copolymer prepared from the ethylester of desaminotyrosyl-tyrosine containing 5 mol% ofPEG blocks of average molecular weight of1000 gmol~1. This molecular design provides three inde-pendent variables for the optimization of the materialsproperties and in the present study, the effects of (i) thependent chain ‘R’, (ii) overall PEG content ‘f ’, and (iii)length of the PEG block on the properties of the resultingcopolymers were investigated. Correlations were estab-lished between the copolymer composition and its mate-rial properties such as glass transition temperature,mechanical strength, degradation, microsphere forma-tion, and drug release. In addition, preliminary data on

254 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264

1Abbreviations used: DSC, differential scanning calorimetry; DTE,desaminotyrosine tyrosyl ethyl ester; DTB, desaminotyrosine tyrosylbutyl ester; DTH, desaminotyrosine tyrosyl hexyl ester; DTO, de-saminotyrosine tyrosyl octyl ester; DTR, desaminotyrosine tyrosyl al-kyl ester; FITC, fluorescein isothiocyanate; GPC, gel permeationchromatography; PBS, phosphate buffered saline; PEG, poly(ethyleneglycol); pNA, p-nitroaniline; PVA, poly(vinyl alcohol); SEM, scanningelectron microscopy; ¹

$, decomposition temperature; ¹

', glass transi-

tion temperature; TGA, thermo-gravimetric analysis; THF, tetrahydro-furan; ¹

., melting temperature.

the in vitro interaction of cells with these polymers wereobtained. In a separate publication, the inverse temper-ature transitions observed for this class of polymers aredescribed [33].

2. Materials and methods1

2.1. Materials

L-Tyrosine, thionyl chloride, pyridine, methylenechloride, tetrahydrofuran (THF), ethanol, butanol, hexa-nol, octanol, 3-(4@-hydroxyphenyl)propionic acid (des-aminotyrosine, Dat), dicyclohexyl carbodiimide (DCC),and hydroxybenzotriazole (HOBt) were obtained fromAldrich, phosgene (solution in toluene) was obtainedfrom Fluka. All solvents were of HPLC grade and wereused as received.

¼arning. Phosgene is an extremely toxic substancethat must be used with extreme care and only in suitablehoods.

2.2. Copolymer synthesis and structure characterization

DTR (x mmol) and PEG (y mmol) were placed intoa round bottom flask. Methylene chloride (5 ml for eachgram of the monomer mixture) and anhydrous pyridine(3.75(x#y) mmol) were added. At room temperature,phosgene solution in toluene (1.25(x#y) mmol) wasadded over 90 min to the reaction mixture with overheadstirring. THF was then added to dilute the reactionmixture to a 5% (w/v) solution. The copolymer wasprecipitated by slowly adding the mixture into 10 vol-umes of ethyl ether. For further purification, copolymerswith lower PEG content ((70% by weight) were redis-solved in THF (5% w/v) and reprecipitated by slowlyadding the polymer solution into 10 volumes of water.Copolymers with higher PEG content ('70% by weight)were redissolved in THF (10% w/v) and reprecipitated byslowly adding the polymer solution into 10 volumes ofisopropanol. In each case, the precipitated copolymerwas collected and dried under vacuum.

The molecular weight of the copolymers was deter-mined by GPC as reported before [11] using THF as thesolvent. Molecular weights were calculated relative topolystyrene standards without further correction. Chem-

ical structure and polymer purity were monitored byFT-IR, 1H-NMR, and 13C-NMR. FT-IR spectra wererecorded on a Matson Cygnus 100 spectrometer. Poly-mer samples were dissolved in methylene chloride andfilms were cast directly onto NaCl plates. All spectra werecollected after 50 scans at 2 cm~1 resolution. UV/Visspectra were recorded on a Perkin-Elmer Lambda 3Bspectrophotometer. NMR spectra of polymer solutionsin deuterated chloroform were recorded on a VarianVXR-200 spectrometer (64 scans).

2.3. Sample preparation

Thin polymer films were prepared by compressionmolding. The processing temperature was 30—35°Cabove ¹

'for each polymer. To minimize polymer ad-

hesion to the metal plates of the mold, two Teflon sheetswere added between the polymer and metal plates. Forcell culture, the bottom glass slide of cell culture dualchamber units (d177380, Nunc, Inc.) was spin-coatedwith a solution of the test polymer following publishedprocedures [11]. Poly(bisphenol-A carbonate) andpoly(DTR carbonate)s (R"ethyl, butyl, hexyl, or octyl)were included as controls in the cell growth studies.

2.4. Thermal properties

¹', crystallinity, and melting points (¹

.) were deter-

mined by differential scanning calorimetry (DSC) [11]except that ¹

'was determined in the second DSC scan as

the midpoint of the transition. According to Graham andZulfiqar [34], ¹

.was determined as the onset temper-

ature, and percentage crystallinity was calculated basedon a standard heat of 100% crystalline PEG *H

(&64)"

219.24 J g~1. The decomposition temperature (¹$) was

determined by thermogravimetric analysis (TGA) andwas reported at 2% decrease in weight. The heating ratefor both DSC and TGA was 10°C min~1 and the averagesample size was 15 mg.

2.5. Mechanical properties

Thin (approx. 0.1 mm) compression molded copoly-mer films were tested on a Sintech 5/D tensile testeraccording to ASTM standard D882-91 at room temper-ature. For each polymer, four individual specimens wereused. Sample width (+5 mm) and thickness (+0.1 mm)were averaged from three measurements prior to analy-sis. The crosshead speed was 2 mmmin~1, allowing fora reliable calculation of the elastic modulus. The yieldpoint was determined based on the zero slope criterion.The reported values of modulus, strength, and elongationwere derived from the stress-strain curves and averagedfrom five separate runs. For each copolymer, four addi-tional specimens were incubated in PBS at 37°C for 24 h,then removed from the incubation media, wiped to

C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264 255

remove water from the surface of the sample, and testedimmediately by the same procedure described above.

2.6. Water uptake

A piece of copolymer (15—20 mg) (compression moldedfilm) was incubated in PBS for various time periods at37°C. After the removal of the sample from the incuba-tion medium, the sample was wiped to remove waterfrom the surface. Water content was determined by TGAat a heating rate of 10°C min~1 and was reported aspercentage weight lost over the heating range from 30 to200°C.

2.7. Hydrolytic degradation

Samples were cut from compression molded films andwere incubated at 37°C in PBS (0.1 M, pH 7.4) containing200 mg l~1 of sodium azide to inhibit bacterial growth.The degradation process was followed by weekly record-ing of the changes in the molecular weight of the polymer.Results are the average of two separate specimens perpolymer.

2.8. Preparation of microspheres

Microspheres were prepared by solvent evaporation asdescribed by Mathiowitz [35] using 0.05 g of copolymerin 1 ml of methylene chloride dispersed within 50 ml ofaqueous medium containing poly(vinyl alcohol) (PVA).After 4 h of stirring at 1300 rpm, the microspheres werecollected by membrane filtration and washed six timeswith water to remove as much PVA as possible. Micro-spheres were collected and dried to constant weight un-der high vacuum.

To incorporate pNA into microspheres, pNA was dis-solved in the polymer solution followed by microsphereformation as described above. pNA loading was deter-mined by UV spectroscopy (j"380 nm) after completedissolution of a weighed quantity of microspheres inmethylene chloride. To incorporate FITC-dextran intomicrospheres, FITC-dextran was dissolved in 50 ll ofwater and dispersed in the polymer solution by sonica-tion (w/o/w method) followed by microsphere formationas described above. To determine the FITC-dextranloading, a weighed quantity of microspheres was dis-solved in methylene chloride and FITC-dextran was ex-tracted into aqueous phosphate buffer solution (0.1 M,pH 7.4), followed by fluorescence spectrophotometry (ex-citation: 495 nm, emission: 520 nm).

The release of pNA or FITC-dextran from the micro-spheres was determined by placing a weighed quantity ofmicrospheres into a measured volume of phosphate buf-fer solution (0.1 M, pH 7.4) at 37°C in a water shaker bath.The amount of pNA or FITC-dextran released into thebuffer solution was determined at regular intervals.

2.9. Cell growth

Fetal rat lung fibroblasts were seeded and proliferatedon the polymer surface following a procedure reportedbefore [11]. Murine osteoblast-like MC3T3-E1 cells [36]were grown in RPMI 1640 medium with 10% fetal calfserum. After 1 and 5 days of incubation, the number ofcells attached to the polymeric substrate was determined.

2.10. Sterilization

Ethylene oxide treatment and c-radiation of dry sam-ples was carried out as reported before [37]. For c-radiation, doses of 0.3, 1.0, and 3.3 Mrad were used.

3. Results and discussion

3.1. Solution polycondensation

Conventional procedures for the synthesis of polycar-bonates by solution phosgenation [10] required onlyslight modifications to control undesirable color forma-tion. Optimized conditions (see Methods) resulted inoff-white polymers with weight-average molecularweights up to about 200 000 dalton and symmetricalmolecular weight distributions (Table 1). The final purifi-cation steps used depended on the overall PEG contentof the copolymers. Crude polymers with low PEG con-tent were purified by precipitation from water. Sincepolymers with high PEG content are water soluble, iso-propanol was used as the precipitation medium. Theprecipitation removed most impurities and some of thelow molecular weight fractions of the polymers, resultingin narrowing of the molecular weight distribution.

A wide range of different copolymers were prepared(Table 1). To study the effect of PEG content, a series ofpoly(DTE co PEG

1000carbonate)s was synthesized with

increasing PEG content (1, 5, 15, 30, 70 mol%). To studythe effect of the ‘R’ pendent chain, a series of fourpoly(DTR co 5%PEG

1000carbonate)s were synthesized

using a constant PEG content of 5 mol% but differentpendent chains (R"ethyl, butyl, hexyl, and octyl). Tostudy the effect of the length of the PEG blocks,PEG

20000was used instead of PEG

1000in several syn-

theses.As a general rule, lower molecular weights were ob-

tained when a higher molar fraction of PEG was used inthe polymerization, but no effect of PEG content, PEGmolecular weight, or DTR pendent chain on the molecu-lar weight distribution was observed. The polymerizationyields ranged from 65 to 82% (after purification). Alltyrosine—PEG derived poly(ether carbonate)s were sol-uble in common organic solvents such as methylenechloride, THF, and DMF. Polymers with high PEGcontent (*70 wt%) were also soluble in water.

256 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264

Table 1Representative yields and molecular weights

Polymer! PEG (wt%) Yield" (%) M8# (kda) M

/# (kda) M

8/M

/

poly(DTE co 1%PEG1000

carbonate) 2.60 79 101.0 60.0 1.7poly(DTE co 5%PEG

1000carbonate) 12.8 81 127.0 83.3 1.5

poly(DTE co 30%PEG1000

carbonate) 53.1 74 41.0 31.0 1.3poly(DTE co 70%PEG

1000carbonate) 86.6 65 46.0 36.8 1.3

poly(DTB co 5%PEG1000

carbonate) 11.5 78 54.8 35.9 1.5poly(DTH co 5%PEG

1000carbonate) 10.8 82 89.9 56.8 1.6

poly(DTO co 1%PEG1000

carbonate) 2.2 79 135.0 74.0 1.8poly(DTO co 5%PEG

1000carbonate) 10.2 77 120.0 62.0 1.9

poly(DTO co 70%PEG1000

carbonate) 83.5 70 51.7 22.3 1.8poly(DTE co 0.26%PEG

20000carbonate) 12.8 81 178.0 84.0 2.1

poly(DTE co 2.1%PEG20000

carbonate) 53.1 77 175.0 115.6 1.5

! Representative polymer batches."Yield after purification, based on the amount of tyrosine-derived diphenol (DTR) in the reaction mixture.# Calculated by GPC relative to polystyrene standards in THF.

Fig. 2. Representative 1H-NMR chemical shift assignments of tyrosine-PEG derived poly(ether carbonate)s as illustrated for poly(DTE co5%PEG

1000carbonate).

3.2. Structure characterization

Chemical structure and polymer purity were deter-mined by FT-IR, 1H-NMR (Fig. 2) and 13C-NMR(Fig. 3). The polymers had characteristic FT-IR absorp-tions at 3300—3310 cm~1 (NH), 2930 cm~1 (CH fromDTR), 2870 cm~1 (CH from PEG), 1774 cm~1 (C"Ocarbonate), 1750 cm~1 (C"O, ester), and 1650 cm~1

(C"O amide). In the FT-IR spectra, an increasing PEG(CH) absorbance at 2870 cm~1 was observed with in-creasing PEG content. The assignments of 1H-NMRpeaks are shown in Fig. 2 using poly(DTE co5%PEG

1000carbonate) as an example. By 1H-NMR

peak integration, the actual ratio of DTE and PEG

present in the polymers was found to be within $10% ofthe theoretical ratio.

The 13C-NMR spectra of representative poly(ethercarbonate)s were obtained to identify the backbonestructure of the polymer. Peak assignments were basedon the calculation of the chemical shift of each carbon[38]. The results are shown in Fig. 3 using poly(DTH co10%PEG

1000carbonate) as an example. Six different

carbonate groups were expected in the polymer back-bone, corresponding to the following conformations atthe carbonate bond: DTH head—DTH head, DTH head—DTH tail, DTH tail—DTH tail, DTH head—PEG, DTHtail—PEG, and PEG—PEG. Indeed, in the 13C-NMRspectrum of poly(DTH co 10%PEG

1000carbonate), six

C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264 257

Fig. 3. Representative 13C-NMR chemical shift assignments of tyrosine-PEG derived poly(ether carbonate)s as illustrated for poly(DTH co10%PEG

1000carbonate).

carbonate peaks were observed in the range of151.81—154.53 ppm. The integration values for thesepeaks corresponded roughly to the statistical probabilityof forming the different carbonate bonds in a randomcopolymer structure.

3.3. Thermal properties and morphology

Introduction of PEG into tyrosine-derived polycar-bonates had profound effects on thermal properties andmorphology. These effects were a function of PEG con-tent, length of PEG block, and the pendent chain ‘R’attached to diphenol. PEG content had the most influ-ence on the polymer properties compared with the otherstructural variables.

3.3.1. Glass transition and crystallinityThe effect of PEG content on glass transition and

crystallinity of the copolymers was analyzed by compar-

ing poly(DTE co PEG1000

carbonate)s with differentPEG contents (Table 2). A single glass transition wasobserved from the DSC curve for each of the polymers.As the PEG content was increased, ¹

'decreased in an

exponential fashion (Table 2). When the PEG1000

con-tent reached 70 mol%, a melting point was also observedat 39°C, one degree lower than the melting temperatureof PEG

1000. However, the copolymer crystallinity (32%)

was significantly lower than the crystallinity of PEG1000

(72%).The effect of the length of PEG blocks on thermal

properties was analyzed by comparing copolymers con-taining PEG

1000and PEG

20000blocks but having iden-

tical weight fractions of PEG. One such pair ofcopolymers is poly(DTE co 5%PEG

1000carbonate) and

poly(DTE co 0.26%PEG20000

carbonate). These twocopolymers had similar ¹

'values of 64 and 68°C

respectively (Table 2). With increasing PEG content, theeffect of different PEG block length became more

258 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264

Table 2Thermal properties of copolymers

Polymer PEG (wt%) ¹'

(°C) ¹.

(°C) Crystallinity (%) ¹$

(°C)

PEG1000

100 * 40! 72! *poly(DTE carbonate) 0 90 * * 290poly(DTE co 1%PEG

1000carbonate) 2.60 85 * * 322

poly(DTE co 5%PEG1000

carbonate) 12.8 64 * * 300poly(DTE co 15%PEG

1000carbonate) 31.8 17 * * 302

poly(DTE co 30%PEG1000

carbonate) 53.1 !17 * * 310poly(DTE co 70%PEG

1000carbonate) 86.6 !45 39 32 339

poly(DTB carbonate) 0 78 * * 290poly(DTB co 5%PEG

1000carbonate) 11.5 49 * * 302

poly(DTH carbonate) 0 65 * * 320poly(DTH co 5%PEG

1000carbonate) 10.8 38 * * 333

poly(DTO carbonate) 0 53 * * 300poly(DTO co 5%PEG

1000carbonate) 10.2 27 * * 335

poly(DTO co 70%PEG1000

carbonate) 83.5 !48 39 36 342PEG

20000100 * 66! 92! *

poly(DTE co 0.26%PEG20000

carbonate) 12.8 68 * * 335poly(DTE co 2.1%PEG

20000carbonate) 53.1 !49 54 24 305

*Not observed.! Comparable to the literature value.

Fig. 4. The effect of pendent chain length ‘R’ (ethyl, butyl, hexyl, octyl -corresponding to 2, 4, 6, and 8 carbons) on the glass transition temper-ature (¹

') in a series of poly(DTR carbonate)s (n) and in the same series

of corresponding copolymers containing 5 mol% of PEG1000

,poly(DTR co 5%PEG

1000carbonate)s (e).

profound: poly(DTE co 30%PEG1000

carbonate) andpoly(DTE co 2.1%PEG

20000carbonate) both contain

53.1% of PEG by weight, but these two copolymers haddramatically different ¹

'values (!17 vs. !49°C). Fur-

thermore, poly(DTE co 30%PEG1000

carbonate) (withshorter PEG blocks) was amorphous while poly(DTE co2.1%PEG

20000carbonate) (with longer PEG blocks) was

crystalline.DTE, DTB, DTH, and DTO containing copolymers

each with 5 mol% of PEG1000

were analyzed to studythe effect of the pendent chain on thermal properties(Table 2). Increasing the length of the hydrophobic pen-dent chain lowered ¹

'in a linear fashion from 64 to 27°C

(Fig. 4). The linear correlation between the length of thependent chain and ¹

'is noteworthy. Likewise, it is note-

worthy that the pendent chain length vs. ¹'

curve forcorresponding tyrosine-derived homopolymers is paral-lel to the curve obtained for the poly(DTR co5%PEG

1000carbonate) series (Fig. 4). The inclusion of

5 mol% of PEG1000

reduced the ¹'

of each copolymerby 27$2°C relative to the corresponding polycarbonatehomopolymer.

3.3.2. Decomposition temperatureDecomposition temperatures were measured in an in-

ert nitrogen atmosphere by TGA (Table 2) and recordedas the temperature at which the polymer lost 2% of itsoriginal weight. When using this criterion, all copolymersexhibited a high degree of thermal stability. Measureddecomposition temperatures varied from 290 to 339°Cwithout obvious correlation to the specific polymerstructure. There was no weight loss due to thermal de-composition prior to the above-mentioned range of de-composition temperatures.

3.4. Mechanical properties

The inclusion of PEG in the backbone of tyrosine-derived polycarbonates significantly changed the mech-anical properties. Five copolymers were compressionmolded and subjected to mechanical analysis both in thedry state and the wet state (Table 3). When the PEGcontent was low, the polymers were strong and toughand had tensile stiffness and strength within the rangeobserved for the corresponding tyrosine-derived homo-polymers. As the PEG content was increased, the poly-mers lost their stiffness and strength. In the dry state,copolymers having larger PEG blocks were stronger

C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264 259

Table 3Mechanical properties of poly(DTR co fPEG M

8carbonate)s

Polymers! Tensile modulus(GPa)

Tensile strength (yield)(MPa)

Tensile strength (break)(MPa)

Elongation(%)

Dry Statepoly(DTE co 5%PEG

1000carbonate) 1.2 43 37 590

poly(DTH co 5%PEG1000

carbonate) 0.77 30 35 890poly(DTO co 5%PEG

1000carbonate) 0.25 7.1 32 890

poly(DTE co 0.26%PEG20000

carbonate) 1.8 48 36 460poly(DTE co 2.1%PEG

20000carbonate) 0.35 18 21 980

¼et state (at equilibrium water content)poly(DTE co 5%PEG

1000carbonate) 0.58 15 19 500

poly(DTH co 5%PEG1000

carbonate) 0.037 1.6 23 1040poly(DTE co 0.26%PEG

20000carbonate) 0.97 29 29 820

poly(DTE co 2.1%PEG20000

carbonate) 0.023 n/a 2.8 230

! PEG content by weight as in Table 2.

Fig. 5. Molecular weight retention of poly(DTE carbonate) (0 wt% ofPEG) (n), poly(DTE co 5%PEG

1000carbonate) (12.8 wt% of PEG)

(e) and poly(DTE co 30%PEG1000

carbonate) (53.1 wt% of PEG) (s)as a function of incubation time at 37°C in phosphate buffered saline.

than copolymers having identical overall PEG contentbut shorter blocks.

According to water uptake studies (see below), after24 h of incubation, all copolymer films had reached theirequilibrium water content. As expected, copolymers be-came softer and lost much of their stiffness and strengthin the wet stage (Table 3). Generally, copolymers contain-ing more than 5 mol% of PEG were flexible and softelastomers in the wet state.

3.5. Water uptake

Poly(DTE co PEG1000

carbonate)s containing 5, 15,and 30 mol% of PEG

1000were used in a study of the

effect of PEG content on water uptake. The amount ofwater uptake was calculated as the percentage of waterabsorbed over the weight of the dry polymer. As the PEGcontent was increased, the rate of water uptake and theequilibrium water content (EWC) increased. Poly(DTEco 5%PEG

1000carbonate) reached its EWC of 10% over

a 5 h period, while poly(DTE co 15%PEG1000

carbon-ate) reached its EWC of 25% after only 1 h. Poly(DTE co30%PEG

1000carbonate) had a EWC of 92% and reach-

ed equilibrium within a few minutes. At PEG1000

con-tents above 15 mol%, the copolymers behavedincreasingly like hydrogels, and at PEG

1000contents

over 70 mol%, the polymers became water soluble. Forcomparison, the homopolymer, poly(DTE carbonate),absorbs only about 1—2% of water [11].

3.6. Degradation in vitro

Poly(DTE carbonate) and poly(DTE co PEG1000

car-bonate)s containing 5 and 30 mol% of PEG were studiedfor the effect of PEG content on polymer degradation.Compression molded films were incubated in PBS at37°C. Molecular weight retention of the polymers was

monitored by GPC weekly (Fig. 5). The degradation rateincreased with increasing amount of PEG. Apparently,the hydrolytic cleavage of the homopolymer backbone islimited by the low water content within the polymericmatrix. Thus, the introduction of PEG into thecopolymer structure increases the degradation rate byincreasing the availability of water within the matrix.

3.7. Microsphere formation and drug release

3.7.1. pNA-loaded microspherespNA was used as a model for low molecular weight

hydrophobic drugs. Free flowing microspheres wereformed from poly(DTB carbonate), poly(DTB co1%PEG

1000carbonate), and poly(DTB co 5%PEG

1000carbonate). When polymers with PEG content above5 mol% were used, microspheres adhered to each other

260 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264

Fig. 6. pNA release from microspheres prepared from poly(DTB car-bonate) (0 wt% of PEG) (n), poly(DTB co 1%PEG

1000carbonate)

(2.6 wt%) (e), and poly(DTB co 5%PEG1000

carbonate) (11.5 wt%)(s) measured as a function of incubation time at 37°C in phosphatebuffer. Experiments were performed in triplicate. Error bars indicate therange between the highest and lowest value measured for each datapoint.

during work up, resulting in aggregation. Generally,about 70—80% (by weight) of the polymer used wasrecovered in the form of microspheres. The size of micro-spheres ranged from 10—50 lm and was independent ofthe type of polymer used. Under the experimental condi-tions of this study, pNA incorporation was about19—25% (by weight).

The presence of PEG in the copolymer structure signif-icantly increased the release rate of pNA from the micro-spheres (Fig. 6). Plotting pNA release as a function of thesquare root of time resulted in a linear correlation (notshown), indicating that the release of pNA from thecopolymers was either controlled by swelling or diffusion[39, 40].

3.7.2. FITC-dextran loaded microspheresFITC-dextran was used as a model for high molecular

weight, water soluble drugs. Under our experimentalconditions, about 83—92% of the FITC-dextran was in-corporated into the microspheres. The release of FITC-dextran from the microspheres was characterized bya short burst effect during the first hour, followed bya long lag period of about 14 days during which verylittle additional FITC-dextran was released (data notshown).

3.8. Cell growth

Rat lung fibroblasts were seeded and grown on thesurfaces of three series of tyrosine-PEG-derived poly(ether carbonate)s: poly(DTE co PEG

1000carbonate)s,

poly(DTB co PEG1000

carbonate)s, and poly(DTH coPEG

1000carbonate)s. The corresponding tyrosine-de-

rived homopolymers were studied for comparison. Afterone day, all non-adhering cells were washed away andthe number of attached cells was documented. This wasfollowed by the measurement of cell proliferation overfour additional days. The experimental results for thetyrosine-derived homopolymers were in agreement withthe previous observation by Ertel [11]. Poly(DTE car-bonate) was the best substrate for cell attachment andproliferation (Table 4). As the length of the pendent chain‘R’ increased, cell attachment and proliferation de-creased. The data collected in Table 4 indicate that theincorporation of very small amounts of PEG (1 mol%)may be beneficial to cell attachment and growth. How-ever, when the copolymer contained 5 mol% or morePEG, the ability of rat lung fibroblasts to attach andgrow on these copolymer surfaces was dramatically re-duced. Very similar results were obtained for osteoblasts(Table 4). Viability tests using trypan blue and calceinAM showed that cells remained viable throughout theentire five day culture period. This demonstrated that theelimination of cell attachment and growth was not due tocytotoxicity of the PEG-containing copolymers, but wasa reflection of the changed surface properties. Tissue

culture polystyrene, glass, and poly(BPA carbonate) wereincluded as control surfaces.

3.9. Sterilization

Three copolymers of the poly(DTE co PEG1000

car-bonate) series, containing 5, 30, and 70 mol% of PEGand poly(DTE co 2.1%PEG

20000carbonate) were steril-

ized by both ethylene oxide and c-irradiation. No notice-able change in color or physical appearance wasobserved for any of the copolymers after their exposureto ethylene oxide and changes in molecular weight andpolydispersity were all within the instrumental error($10%) — except for poly(DTE co 70%PEG

1000carbon-

ate) (Table 5). For this copolymer, containing the highestweight fraction of PEG, an increase in the molecularweight after exposure to ethylene oxide was observed.The reason for this observation is currently not known.

Three doses of c-irradiation (0.3, 1.1, 3.9 Mrad) wereused to evaluate the effect of c-sterilization on tyrosine-PEG-derived poly(ether carbonate)s. No effect on thephysical appearance of the polymers was observed. Thefour copolymers retained their initial molecular weightafter exposure to 0.3 Mrad of irradiation. The observedminor variations are within the error of our instrumenta-tion. However, when the irradiation dose was increasedto 1.1 and 3.9 Mrad, the molecular weight decreased forall polymers. The higher the PEG content of thecopolymers, the more pronounced was the decrease inmolecular weight following irradiation. Considering thata dose of about 2.5 Mrad is usually used for the steriliza-tion of medical implants, our preliminary data indicatethat only copolymers with a PEG content below30 mol% should be sterilized by irradiation.

C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264 261

Table 4Cell attachment and proliferation on surfaces of copolymers

Polymer 1 day attachment(]100 cells cm~2)

5 day proliferation(]100 cells cm~2)

Results for fibroblastspoly(DTE carbonate) 46$13 596$100poly(DTE co 5%PEG

1000carbonate) 8$8 46$14

poly(DTE co 15%PEG1000

carbonate) 4$5 11$10poly(DTE co 30%PEG

1000carbonate) 3$5 11$10

poly(DTB carbonate) 56$17 410$79poly(DTB co 1%PEG

1000carbonate) 50$14 163$40

poly(DTB co 5%PEG1000

carbonate) 16$10 18$13poly(DTB co 10%PEG

1000carbonate) 9$9 7$7

poly(DTH carbonate) 32$10 268$46poly(DTH co 1%PEG

1000carbonate) 52$31 275$71

poly(DTH co 5%PEG1000

carbonate) 9$11 3$7poly(DTH co 10%PEG

1000carbonate) 9$11 11$14

Fibroblast experimental control surfacesGlass 50$16 555$91Poly(BPA carbonate) 17$10 123$37

Results for osteoblastspoly(DTE carbonate) 34$15 381$93poly(DTE co 5%PEG

1000carbonate) 9$8 12$9

poly(DTE co 15%PEG1000

carbonate) 9$5 9$9poly(DTE co 30%PEG

1000carbonate) 13$5 6$7

Osteoblast experimental control surfacesGlass 59$22 739$83Tissue culture polystyrene 68$23 762$126

Table 5Sterilization of poly(DTE co PEG

1000carbonate)s and poly(DTE co PEG

20000carbonate) by ethylene oxide and c-radiation

Polymer Molecular weight retention (%)!

ethylene oxide c-radiation

0.3 Mrad 1.0 Mrad 3.3 Mrad

poly(DTE co 5%PEG1000

carbonate) 95 102 n/a 92poly(DTE co 30%PEG

1000carbonate) 106 99 85 60

poly(DTE co 70%PEG1000

carbonate) 145 105 50 41poly(DTE co 2.1%PEG

20000carbonate) 88 92 67 59

! Molecular weight retention given in percent of the starting molecular weight prior to sterilization, as determined by GPC. The measurement erroris estimated to be about $5%.

4. Conclusions

While the previously investigated, poly(DTR carbon-ate)s were strong, stiff, and very hydrophobic materials,the incorporation of PEG blocks into the polymer back-bone resulted in significant changes in the polymer prop-erties. The copolymers based on DTR diphenols andPEG appear to be random copolymers as indicated by13C-NMR. At low PEG content (5 mol%) thecopolymers were tough and strong materials that hadcomparable mechanical properties to the DTR-derivedpolycarbonate homopolymers. At higher PEG content(10—50 mol%), the copolymers acted as hydrogels and

were soft and elastomeric materials whose EWC could becontrolled accurately by the PEG content. At very highPEG content ('70 mol%), the copolymers becamewater soluble and exhibited inverse temperaturetransitions, e.g. the copolymers were soluble in coldaqueous solutions and precipitated upon warming. A de-tailed account of the inverse temperature transitions ofthese polymers is provided elsewhere [33].

Studies performed here provided an initial indicationthat copolymers based on DTR diphenols and PEG canbe sterilized by exposure to ethylene oxide and thatcopolymers containing less than 30 mol% of PEG mayalso be sterilizable by c-irradiation. However, the current

262 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264

studies did not examine whether any toxic residues areintroduced into the copolymers by exposure to ethyleneoxide and whether the PEG blocks crosslink under c-irradiation. Such detailed studies are the subject of futureinvestigations.

Particularly noteworthy is the dramatic effect of thepresence of PEG blocks on the biological properties ofthe copolymers. The polycarbonate homopolymers, es-pecially poly(DTE carbonate), are excellent substratesfor cell growth [11]. However, even at very low PEGcontent, when the physical properties of the copolymerswere very similar to those of the corresponding poly-carbonate homopolymers, fibroblasts and osteoblastsfailed to attach and grow on PEG containing copolymersurfaces.

Since copolymers based on DTR diphenols and PEGcan be prepared with relative ease, can be processed,fabricated, and exhibit a wide range of useful materialproperties, these new polymers may find practical ap-plications in the formulation of degradable medical im-plants.

Acknowledgements

This work was supported by NIH grant GM39455 andby a seed grant from the New Jersey Center for Bio-materials and Medical Devices. The authors thank Dr.Kenneth S. James for assisting with the cell growth stud-ies and Ms. Carole Kantor for editorial assistance inpreparing this manuscript.

References

[1] Langer R, Vacanti J. Tissue engineering. Science 1993;260:920—6.[2] Hubbell J. Biomaterials in tissue engineering. Biotechnology

1995;13:565—76.[3] Hoffman AS. Present and emerging applications of polymeric

biomaterials. Clin Mater 1992;11:13—8.[4] Ratner BD. New ideas in biomaterials science—a path to engi-

neered biomaterials. J Biomed Mater Res 1993;27:837—50.[5] Eisenberger P. In: NIH workshop: biomaterials and medical

implant science. Bethesda MD, Bethesda, MD: National Insti-tutes of Health, 1995.

[6] Bostman O, Hirvensalo E, Vainionpaa S, Makela A, Vihtonen, K,Tormala P, Rokkanen P. Ankle fractures treated using biode-gradable internal fixation. Clin Orthop 1989;238:195—203.

[7] Vert M, Christel P, Chabot F, Leray J. Bioresorbable plasticmaterials for bone surgery. In: Hastings GW, Ducheyne P, edi-tors. Macromolecular biomaterials. Boca Raton: CRC Press,1984:119—42.

[8] Vert M, Li S, Garreau H. More about the degradation of LA/GA-derived matrices in aqueous media. J Control Res 1991;16:15—26.

[9] Engelberg I, Kohn J. Physico-mechanical properties of degrada-ble polymers used in medical applications: a comparative study.Biomaterials 1991;12(3):292—304.

[10] Pulapura S, Kohn J. Tyrosine derived polycarbonates: backbonemodified, ‘pseudo’-poly(amino acids) designed for biomedicalapplications. Biopolymers 1992;32:411—7.

[11] Ertel SI, Kohn J. Evaluation of a series of tyrosine-derived poly-carbonates for biomaterial applications. J Biomed Mater Res1994;28:919—30.

[12] Pulapura S, Li C, Kohn J. Structure-property relationships for thedesign of polyiminocarbonates. Biomaterials 1990;11:666—78.

[13] Fiordeliso J, Bron S, Kohn J. Design, synthesis, and preliminarycharacterization of tyrosine-containing polyarylates: new bio-materials for medical applications. J Biomater Sci (Polym Ed)1994;5(6):497—510.

[14] Brocchini S, James K, Tangpasuthadol V, Kohn J. A combina-torial approach for polymer design. J Amer Chem Soc1997;119(19):4553—4.

[15] James K, Kohn J. Applications of pseudo-poly(amino acid) bio-materials. Trends Polym Sci 1996;4(12):394—7.

[16] Ertel SI, Kohn J, Zimmerman MC, Parsons JR. Evaluation ofpoly(DTH carbonate), a tyrosine-derived degradable polymer, fororthopaedic applications. J Biomed Mater Res 1995;29(11):1337—48.

[17] Choueka J, Charvet JL, Koval KJ, Alexander H, James KS,Hooper KA, Kohn J. Canine bone response to tyrosine-derivedpolycarbonates and poly(L-lactic acid). J Biomed Mater Res1996;31:35—41.

[18] Wang SG, Chen CF, Li ZF, Li XF, Gu HQ. Mixed polyether-polyester multiblock copolymer and its blood compatibility.J Macromol Sci Chem 1989;A26:505—18.

[19] Bakker D, van Blitterswijk CA, Hesseling SC, Koerten HK,Kuijpers W, Grote JJ. Biocompatibility of a polyether urethane,polypropylene oxide, and a polyether-polyester copolymer.A qualitative and quantitative study of three alloplastic tympanicmembrane materials in the rat middle ear. J Biomed Mater Res1990;24(4):489—515.

[20] Radder AM, Leenders H, van Blitterswijk CA. Bone-bondingbehavior of poly(ethylene oxide)—poly(butylene terephthalate)copolymer coatings and bulk implants: a comparative study.Biomaterials 1995;16(7):507—13.

[21] Jedlinski Z, Kurcok P, Walach W, Janeczek H, Radecka I, Poly-merization of lactones 17: synthesis of ethylene glycol-L-lactideblock copolymers. Makromol Chem 1993;194:1681—9.

[22] Ferruti P, Penco M, D’Addato P, Ranucci E, Deghenghi R.Synthesis and properties of novel block copolymers containingpoly(lactic-glycolic acid) and poly(ethylene glycol) segments. Bio-materials 1995;16(18):1423—8.

[23] Nishimura T, Sato Y, Yokoyama M, Okuya M, Inoue S. Ad-hesion behavior of rat lymphocytes on poly(c-benzyl L-glutamate)derivatives having hydroxyl groups or poly(ethylene glycol)chains. Makromol Chem 1984;185:2109—16.

[24] Yokoyama M, Anazawa H, Takahashi A, Inoue S. Synthesis andpermeation behavior of membranes from segmented multiblockcopolymers containing poly(ethylene oxide) and poly(b-benzylL-aspartate) blocks Makromol Chem 1990;191:301—11.

[25] Merrill SH. Block copolymers based on 2,2-bis(4-hydroxy-phenyl)-propane polycarbonate. I. Preparation and properties ofcopolymers from a variety of homopolymers. J Polym Sci1961;55:343—52.

[26] Goldberg EP. Elastomeric polycarbonate block copolymers.J Polym Sci Part C 1964;4:707—30.

[27] Tanjsugi H, Ohnuma H, Kotaka T. Swelling behavior ofbisphenol—A polycarbonate-polyoxyethylene multiblock copoly-mers in ethanol/water mixtures. Polym J 1984;16(8): 633—40.

[28] Suzuki T, Chihara H, Kotaka T. Sorption of water by bisphenol—A polycarbonate and polyoxyethylene multiblock copolymerswith varying composition and block length. Polym J1984;16(2):129—38.

[29] Mandenius CF, Ljunggren L. Ellipsometric studies of plasmaprotein adsorption on membrane polymers for blood purification.Biomaterials 1991;12(4):369—73.

[30] Cho C-S, Kotaka T, Akaike T. Cell adhesion onto block copolymerLangmuir—Blodgett films. J Biomed Mater Res 1993;27:199—206.

C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264 263

[31] Behnke J, Pitowski HJ. Polycarbonate—polyether copolymermembranes for hemodialysis, German Patent Application: DE79-2932761, issued: 1994, assigned to Akzo GmbH (Germany).

[32] Higley WS, Cantor PA, Fisher BS. Relatively thick polycarbonatemembranes for use in hemodialysis US Patent 4,308,145, issued:1981, assigned to United States of America.

[33] Yu C, Mielewczyk SS, Breslauer KJ, Kohn J. Tyrosine-PEG-derived poly(ether carbonate)s as new biomaterials. Part II:study of inverse temperature transitions. Biomaterials 1999;20(3):265—72.

[34] Graham NB, Zulfiqar M. Interaction of poly(ethylene oxide) withsolvents: 3. Synthesis and swelling in water of crosslinked poly(ethylene glycol) urethane networks. Polymer 1989;30:2130—5.

[35] Mathiowitz E, Kline D, Langer R. Morphology of polyanhydridemicrosphere delivery systems. J Scanning Micros 1990;4(2):329—40.

[36] Sudo H, Kodama H, Amagai Y, Yammamoto S, Kasai S. In vitrodifferentiation and calcification in a new clonal osteogenic cell linederived from newborn mouse calvaria. J Cell Biol 1983;96:191—8.

[37] Hooper KA, Cox JD, Kohn J. Comparison of the effect of ethy-lene oxide and c-irradiation on selected tyrosine-derived polycar-bonates and poly(L-lactic acid). J Appl Polym Sci 1997;63(11):1499—510.

[38] Breitmaier E, Voelter W. Carbon-13 NMR spectroscopy: high-resolution methods and applications in organic chemistry andbiochemistry. New York: VCH Publishers, 1987:183—325.

[39] Franson NM, Peppas NA. Influence of copolymer compositionon non-fickian water transport through glassy copolymers. J ApplPolym Sci 1983;28:1299—310.

[40] Peppas NA, Gurny R, Doelker E, Buri P. Modelling of drugdiffusion through swellable polymeric systems. J Memb Sci1980;7(3):241—53.

264 C. Yu, J. Kohn / Biomaterials 20 (1999) 253—264