Structural Characterization of a Lipase-CatalyzedCopolymerization of E-Caprolactone and D,L-Lactide

Jessica Wahlberg,†,‡ Per Valdemar Persson,† Tove Olsson,† Erik Hedenstrom,‡ andTommy Iversen*,†

STFI, Swedish Pulp and Paper Research Institute, Box 5604, SE-114 86 Stockholm, Sweden, andChemistry, Department of Natural and Environmental Sciences, Mid Sweden University,

SE-851 70 Sundsvall, Sweden

Received March 12, 2003; Revised Manuscript Received May 6, 2003

The copolymerization ofε-caprolactone (ε-CL) and D,L-lactide catalyzed byCandida antarcticalipase Bwas studied. Copolymerizations with differentε-CL-to-lactide ratios were carried out, and the product wasmonitored and characterized by MALDI-TOF MS, GPC, and1H NMR. The polymerization ofε-CL, whichis normally promoted byC. antarcticalipase B, is initially slowed by the presence of lactide. During thisstage, lactide is consumed more rapidly thanε-CL, and the incorporation occurs dimer-wise with regard tothe lactic acid (LA) units. As the reaction proceeds, the relative amount of CL units in the copolymerincreases. The nonrandom copolymer structure disappears with time, probably due to a lipase-catalyzedtransesterification reaction. In the copolymerizations with a low content of lactide, macrocycles of poly(ε-caprolactone) and copolymers having up to two LA units in the ring were detected.

Introduction

The aliphatic polyesters, polylactide (PLA), and poly(ε-caprolactone) (PCL), have received much attention becauseof their biocompatibility and biodegradability. In recent years,blends of these polymers have attracted great interest as asimple way of achieving improved polymer properties.1-5

However, polymer blends often suffer from poor mechanicalproperties because of limited miscibility and insufficientadhesion between the phases. To optimize the productproperties, it is of interest to increase the phase adhesionbetween the polymers by using compatibilizers such as PCL/PLA block copolymers.4

Lipase-catalyzed lactone polymerizations are reported tobe both enantio- and regioselective reactions,6,7 and theycould, in principle, be an attractive route to PCL/PLA blockcopolymers. However, the lipase-catalyzed copolymerizationof lactones has been reported to result in the formation ofrandom copolymers,8-10 and it has been suggested that therandom structure is a consequence of a frequently occurringlipase-catalyzed transesterification between the originallyformed polyesters.8,11

Candida antarcticalipase B is efficient in catalyzing thering-opening polymerization ofε-caprolactone (ε-CL),12,13

but it has been reported that it does not catalyze thepolymerization ofD,L-lactide.14,15 In a recent study of thecopolymerization ofD,L-lactide with eitherε-CL or æ-pen-tadecalactone, it was suggested thatD,L-lactide acts as areversible inhibitor ofC. antarcticalipase B.16 In the presentstudy, we have used this ability ofD,L-lactide to slow

copolymerizations catalyzed byC. antarctica lipase B toinvestigate how the molecular structure of PCL/PLA co-polymers is altered as the reaction proceeds. The lipase-catalyzed copolymerization is illustrated in Scheme 1. Thelactide can be regarded as the condensation product of twounits of lactic acid (LA). The reaction between one moleculeeach ofε-CL and lactide thus yields a CL unit connected totwo LA units.

Materials and Methods

Materials. Immobilized Candida antarctica lipase B(Novozym 435, 7000 PLU/g), a gift from Novo Nordisk A/S,was dried over calcium hydride before use.ε-CL and D,L-lactide were purchased from Aldrich Chemical Co.ε-CL wasdried over activated molecular sieves (4 Å) before use,whereas the lactide was used as received.

High-Lactide-Content Copolymerization. A total of 2.2mmol of ε-CL was mixed with 1.1 mmol of lactide(equivalent to CL:LA 1:1) in a dried vial containing lipase(15 mg) and a few molecular sieves. The capped vial wasshaken at 70°C and 100 rpm. Samples of the product mixturewere withdrawn at different reaction times and immediatelyanalyzed by GPC and MALDI-TOF MS.

Low-Lactide-Content Copolymerization. ε-CL (2.2 mmol)and lactide were mixed according to the high-lactide-contentprocedure to give molar ratios of 10:1 and 20:1 (CL:LA 5:1and 10:1). Polymerization of neatε-CL was also performedas a reference. A lower temperature (60°C) and less enzyme(5 mg) were used in these experiments in order to reducethe reaction speed.

Matrix-Assisted Laser Desorption/Ionization Mass Spec-trometry (MALDI-TOF MS). Samples (about 0.5 mg) fromthe various reactions were withdrawn and dissolved in

* To whom correspondence should be addressed. Telephone:+46-8-6767000. Fax:+46-8-108340. E-mail: tommy.iversen@stfi.se.

† STFI, Swedish Pulp and Paper Research Institute.‡ Mid Sweden University.

1068 Biomacromolecules 2003,4, 1068-1071

10.1021/bm0340725 CCC: $25.00 © 2003 American Chemical SocietyPublished on Web 06/06/2003

dioxane (25µL). A volume of 10µL of this solution wasmixed with an equal volume of matrix (50 mg/mL 2,5-dihydroxybenzoic acid in methanol/water 1/1). An aliquot(0.5 µL) was applied to the sample probe, the solvent wasevaporated by vacuum, and the probe was inserted into thespectrometer (Hewlett-Packard G20205A LD-TOF). Spectrawere taken in the positive ion mode. Products havingm/zratios below 300 were disregarded in the evaluation of thespectra due to overlap with matrix components.

Gel Permeation Chromatography (GPC).Samples (5mg) of the crude reaction mixtures were dissolved in 1 mLof tetrahydrofuran and filtered through a 0.45µm PTFEmembrane before injection. The GPC system was a WatersHPLC pump 510 connected to a Waters 410 differentialrefractometer and a Rheodyne 7125 injector (20µL loop).Three columns from Waters, Ultrastyragel 50, 100, and 500Å, connected in series were used for the separation. Tet-rahydrofuran was used as eluent at a flow rate of 1 mL/min.The GPC system was calibrated using polystyrene standardsfrom Machery Nagel, 266-350 000 Da.

1H NMR. Proton spectra were obtained with a BrukerAvance DPX 250 spectrometer at 25°C using CDCl3 assolvent and TMS as internal reference.

Results and Discussion

MALDI-TOF MS Characterization. MALDI-TOF spec-tra of synthetic polymers can provide fundamental informa-tion about the mass of the repeating units and the presentend groups. In some cases with narrow distributions, theaverage molecular weight can also be determined. In spectraof copolymers, the number of masses observed increasesdramatically; resolution is hence usually limited to oligomersof relatively low molecular weight. Provided there is asufficient mass difference between the monomers and a lowoligomer-specific discrimination, the end-group pattern, themonomer composition, and, sometimes, also the presenceof preferential monomer growth in co-oligomers can bedetermined directly from the MALDI-TOF spectra.17,18

Evidence may thus be obtained about changes in monomercomposition of polymers occurring as a function of reactiontime, if it is assumed that the interpretable oligomer structuresreflect the corresponding high-molecular weight polymerstructures.

High-Lactide-Content Copolymerization. In Figure 1,the MALDI-TOF mass spectra of the 1:1 CL:LA reactionafter 1, 3, and 8 weeks are shown. The spectra are complex,especially the spectrum taken after 8 weeks reaction time;the presence of mixtures of copolymers and a correspondingrapid loss of resolution result in signal overlap. However,the interpretation of the distinct oligomer distributions in the

spectra, shown as enlargements in Figure 1, can addressseveral issues concerning the structural changes occurringas the reaction proceeds.

The reactivities of the two monomers differed as seen inthe three-dimensional diagrams in Figure 2. The diagramsshow the oligomeric products formed after 1, 3, and 8 weeksin the 1:1 CL:LA reaction, calculated from discernible peaksin the mass spectra. After one week (a), the oligomeric

Scheme 1. Lipase-Catalyzed Copolymerization of ε-CL (1) andLactide (2)

Figure 1. MALDI-TOF MS of the high-lactide-content reaction (2.2mmol ε-CL, 1.1 mmol lactide, 70 °C) catalyzed by C. antarctica lipaseB (15 mg Novozym 435) after 1, 3, and 8 weeks (I, II, and III,respectively). The marked peaks show oligomers containing 0, 1, 2,3, and 4 CL units (a-e, respectively).

Figure 2. Three-dimensional representation of the high-lactide-content reaction (2.2 mmol ε-CL, 1.1 mmol lactide, 70 °C) catalyzedby immobilized C. antarctica lipase B (15 mg) after 1 week (a), 3weeks (b), and 8 weeks (c) calculated from MALDI-TOF MS data.The diagrams to the right are obtained by a 90° clockwise rotation ofthe left diagrams.

Structural Characterization of a Copolyester Biomacromolecules, Vol. 4, No. 4, 2003 1069

fraction consisted of copolymers dominated by LA units, andno PCL homopolymer was observed in the product mixture.This is surprising becauseC. antarcticalipase B in organicsolvents, as mentioned in the Introduction, has been shownto be highly active towardε-CL and inactive toward lactide.With time, an increasing proportion of CL units was foundin the product (b and c).

The lipase-catalyzed ring-opening copolymerization isinitially a nonrandom reaction according to the MALDI-TOFmass spectra covering the oligomeric distribution. After oneweek (CL:LA 1:1, Figure 3), the oligomers contain pre-dominantly an even number of LA units (distribution withm/z difference of 144, corresponding to two LA units),indicating that lactide is polymerized as the dimer aspreviously stated.14 At this stage, only small amounts ofoligomers with an odd number of LA units were detected.However, this fraction increased with time, as seen in theenlarged oligomer spectrum taken after 8 weeks. Thedominating and interpretable peaks of this spectrum werethose having equal amounts of CL and LA units. The build-up of distributions withm/z difference of 72, i.e., one LAunit, supports this suggestion. The initial nonrandom productdistribution was thus lost, probably as a result of extensivetransesterification.8,11

After 8 weeks, all of the monomers had completely reactedto polymeric products according to GPC (Figure 4). The1HNMR spectrum of the copolymer mixture after 8 weeksshowed no monomer signals, which also indicated completepolymerization.19,20

At longer reaction times, there is a risk that nonenzymaticreactions contribute to the product formation. However, inreference reactions without lipase, no polymer products weredetected by MALDI-TOF, and thermally deactivated No-vozym 435 is reported not to catalyze the ring-openingpolymerization ofε-CL.21

To our knowledge, this is the first time thatD,L-lactide isreported to be active as a substrate in aC. antarcticalipaseB catalyzed copolymerization. Earlier studies have concludedthat D,L-lactide forms polymer products in reasonable yieldwith Pseudomonas cepacialipase, although no reaction wasdetected usingC. antarcticalipase B.14 Other studies havesuggested that lactide acts as an inhibitor in the polymeri-zation of ε-CL.16 However, no thorough structural charac-terization of the products was conducted in these studies to

verify that lactide really does inhibit the polymerization ofε-CL without being incorporated into the resulting polymer.

Low-Lactide-Content Copolymerization. To facilitate acloser investigation of the behavior of this reaction, someparameters were changed. Lowering the temperature to 60°C and using less enzyme (5 mg) slowed the polymerization,and a larger amount ofε-CL reduced the viscosity of thereaction mixture. Experiments were performed with twodifferent concentrations of lactide (CL:LA 10:1 and 5:1), andthese were monitored by GPC and MALDI-TOF MS. Thepolymerization of neatε-CL was used as a reference sample.

Figure 3. MALDI-TOF MS of the CL:LA 1:1 reaction (2.2 mmol ε-CL,1.1 mmol lactide, 70 °C) catalyzed by immobilized C. antarctica lipaseB (15 mg) after (a) 1 week and (b) after 8 weeks, showing thepositions of oligomers in the spectrum containing 1 (9), 3 (b), and 5(2) LA units.

Figure 4. GPC of (a) the product mixture containing ε-CL (I), lactide(II), and copolymer (III), and (b) the 1:1 CL:LA reaction (2.2 mmolε-CL, 1.1 mmol lactide, 15 mg C. antarctica lipase B, 70 °C) aftereight weeks showing only polymer product.

Figure 5. Consumption of ε-CL ([) and lactide (9) in (a) the neatε-CL reaction, (b) CL:LA 10:1 (2.2 mmol ε-CL, 0.11 mmol lactide, 5mg C. antarctica lipase B, 60 °C), and (c) CL:LA 5:1 (2.2 mmol ε-CL,0.22 mmol lactide, 5 mg C. antarctica lipase B, 60 °C) according toGPC.

1070 Biomacromolecules, Vol. 4, No. 4, 2003 Wahlberg et al.

The lactide-containing reactions exhibited lower reactionrates; after 2 days, the consumption ofε-CL in the referencereaction was 94% but only 51% and 31% in the copolym-erization reactions, the former value being the 10:1 CL:LAreaction. Furthermore,ε-CL was completely consumed after7 days in the neat reaction, whereas residualε-CL wasdetected after 4 weeks when lactide was present. In Figure5, the consumption of the separate monomers is shown as afunction of time for each specific case, calculated from GPC.Both 5:1 and 10:1 CL:LA reactions exhibit a rapid consump-tion of lactide in the first 2 h. During this time, few CL unitsseem to be incorporated into the product. Whenε-CL ispolymerized in the absence of lactide, the conversion is fastand the monomer is completely consumed.

After this initial phase, both monomers were convertedinto polymeric products at approximately the same rate.Further, in the presence of lactide,ε-CL was after 4 weeksstill not completely consumed, whereas no residual monomerwas detected after 7 days of the reference reaction. Eventhough lactide slows down the over-all polymerizationreaction, it is the most reactive monomer during the initialpolymerization. The presence of lactide thus influences thereactivity of ε-CL, a normally favorable substrate for thelipase used in this study. Interestingly, the polymerizationrate of ε-CL does not increase when the lactide has beencompletely consumed. This cannot be fully ascribed to athermal deactivation of the lipase, because the molecularweight of the polymers continues to increase, as seen inFigure 6.

Macrocycle Formation. Examination of the results of the

low-lactide-content experiments (Figure 7) revealed thatmacrocyclic products are initially formed.13,22-24 The m/zratios of these distributions agreed with cyclic PCL or cycliccopolyesters containing one or two LA groups, respectively.The cyclic products gradually disappeared as the reactionproceeded.

Conclusion

The copolymerization ofε-CL and D,L-lactide catalyzedby C. antarctica lipase B initially leads to a nonrandomincorporation of lactide in the polymer formed. Lactide isincorporated as the dimer in the initial stage, whenε-CL ispractically excluded from the reaction. After this initial stage,ε-CL and lactide are polymerized at approximately equalrates. Intermolecular transesterification destroys the nonran-dom product pattern, eventually leading to a random mixture,which is seen after prolonged reaction times. In experimentswith a low proportion of lactide, macrocycles of pure PCLand copolymers with up to two LA units in the ring wereformed. These cycles are gradually transformed into a linearproduct.

References and Notes

(1) Hiljanen-Vainio, M.; Karjalainen, T.; Seppala, J.J. Appl. Polym. Sci.1996, 59, 1281.

(2) Yang, J.-M.; Chen, H.-L.; You, J.-W.; Hwang, J. C.Polym. J.1997,29, 657.

(3) Wang, L.; Ma, W.; Gross, R. A.; McCarthy, S. P.Polym. Degrad.Stab.1998, 59, 161.

(4) Dell’Erba, R.; Groeninckx, G.; Maglio, G.; Malinconico, M.;Migliozzi, A. Polymer2001, 42, 7831.

(5) Na, Y.-H.; He, Y.; Shuai, X.; Kikkawa, Y.; Doi, Y.; Inoue, Y.Biomacromolecules2002, 3, 1179.

(6) Cordova, A.; Iversen, T.Macromolecules1998, 31, 1040.(7) Bisht, K. S.; Deng, F.; Gross, R. A.; Kaplan, D. L.; Swift, G.J. Am.

Chem. Soc.1998, 120, 1363.(8) Uyama, H.; Takeya, K.; Kobayashi, S.Proc. Acad. Jpn.1993, 69B,

203.(9) Kumar, A.; Kalra, B.; Dekhterman, A.; Gross, R. A.Macromolecules

2000, 33, 6303.(10) Takamoto, T.; Kerep, P.; Uyama, H.; Kobayashi, S.Macromol. Biosci.

2001, 1, 223.(11) Kumar, A.; Gross, R. A.J. Am. Chem. Soc.2000, 122, 11767.(12) Uyama, H.; Suda, S.; Kikuchi, H.; Kobayashi, S.Chem. Lett.1997,

1109.(13) Cordova, A.; Iversen, T.; Hult, K.; Martinelle, M.Polymer1998,

39, 6519.(14) Matsumura, S.; Mabuchi, K.; Toshima, K.Macromol. Rapid Com-

mun.1997, 18, 477.(15) Matsumura, S.; Mabuchi, K.; Toshima, K.Macromol. Symp.1998,

130, 285.(16) Kalra, B.; Kunioka, M.; Kumar, A.; Gross, R. A.Polym. Prepr.2002,

43, 591.(17) Nielen, M. W. F.Mass Spectrom. ReV. 1999, 18, 309.(18) Viala, S.; Tauer, K.; Antonietti, M.; Kru¨ger, R.-P.; Bremser, W.

Polymer2002, 43, 7231.(19) Thakur, K. A. M.; Kean, R. T.; Hall, E. S.; Kolstad, J. J.; Lindgren,

T. A.; Doscotch, M. A.; Siepmann, J. I.; Munson, E. J.Macromol-ecules1997, 30, 2422.

(20) Cordova, A.; Iversen, T.; Hult, K.Polymer1999, 40, 6709.(21) Deng, F.; Gross, R. A.Int. J. Biol. Macromol.1999, 25, 153.(22) Knani, D.; Gutman, A. L.; Kohn, D. H.J. Polym. Sci. Part A: Polym.

Chem.1993, 31, 1221.(23) Berkane, C.; Mezoul, G.; Lalot, T.; Brigodiot, M.; Marechal, E.

Macromolecules1997, 30, 7729.(24) Matsumura, S.; Suzuki, Y.; Tsukada, K.; Toshima, K.; Doi, Y.;

Kasuya, K.Macromolecules1998, 31, 6444.

BM0340725

Figure 6. Molecular weights (Mn) as a function of reaction time ofneat ε-CL ([), CL:LA 10:1 (9), and 5:1 (2), taken from GPC data.

Figure 7. Cyclic distributions in the MALDI-TOF MS of the 5:1 CL:LA reaction taken after 48 h. Cycles containing zero (2), one (b),and two (9) LA units are detected in the lower region of the spectrum.

Structural Characterization of a Copolyester Biomacromolecules, Vol. 4, No. 4, 2003 1071