Ester Synthesis FromTrimethylammonium Alcohols in DryOrganic Media Catalyzed by ImmobilizedCandida antarctica Lipase B

Pedro Lozano, Mirta Daz,* Teresa de Diego, Jose L. Iborra

Departamento de Bioquımica y Biologıa Molecular B e Inmunologıa,Facultad de Quımica, Universidad de Murcia, P.O. Box 4021, E-30100Murcia, Espana; telephone: 34-968-36-7398; fax: 34-968-36-4148;e-mail: jliborra@um.es

Received 16 May 2002; accepted 8 October 2002

DOI: 10.1002/bit.10580

Abstract: Twenty-one different organic solvents were as-sayed as possible reaction media for the synthesis ofbutyryl esters from trimethylammonium alcohols in dryconditions catalyzed by immobilized Candida antarcticalipase B. The reactions were carried out following atransesterification kinetic approach, using choline and L-carnitine as primary and secondary trimethylammoniumalcohols, respectively, and vinyl butyrate as acyl donor.The synthetic activity of the enzyme was strictly depen-dent on the water content, the position of the hydroxylgroup in the trimethylammonium molecule, and the LogP parameter of the assayed solvent. Anhydrous condi-tions and a high excess of vinyl butyrate over L-carnitinewere necessary to synthesize butyryl-L-carnitine. Thesynthetic reaction rates of butyryl choline were practi-cally 100-fold those of butyryl-L-carnitine with all the as-sayed solvents. In both cases, the synthetic activity of theenzyme was dependent on the hydrophobicity of the sol-vent, with the optimal reaction media showing a Log Pparameter of between −0.5 and 0.5. In all cases, 2-meth-yl-2-propanol and 2-methyl-2-butanol were shown to bethe best solvents for both their high synthetic activity andnegligible loss of enzyme activity after 6 days. © 2003Wiley Periodicals, Inc. Biotechnol Bioeng 82: 352–358, 2003.Keywords: lipase; trimethylammonium ester; butyryl-L-carnitine; butyryl choline; ester synthesis

INTRODUCTION

The technological usefulness of enzymes can be greatlyenhanced if they are used in organic media rather than intheir natural aqueous reaction media. In this way, enzymesin nonconventional media may exhibit new catalytic prop-erties, such as synthetic activity (by hydrolytic enzymes),stereo- and regioselectivity for substrates, and enhanced sta-bility (Klibanov, 2001). Among enzymes, lipases have beenused successfully as a catalyst for the synthesis of esters,

both on a small scale and on an industrial scale (Balco et al.,1996; Jaeger and Reetz, 1998). Recently, research has beenfocusing on amphiphilic molecules, for example, fatty acidsugar esters (Arcos et al., 1998) and dihydroxyacetone fattyacid esters (Virto et al., 2000) for use as nonionic surfac-tants, esters of �-hydroxyacids as humectants (Torres andOtero, 1999; Torres et al., 1999), and phospholipids (Virtoet al., 1999; Virto and Adlercreutz, 2000) for their importantfood and pharmaceutical applications. Lipase-catalyzedsynthesis of these compounds requires nearly anhydrousconditions and the use of nonpolar solvents as reaction me-dia to avoid the stripping of essential water molecules fromthe enzyme structure. However, polar acyl-acceptors exhibitlimited solubility in common “enzyme-friendly” organicsolvents (e.g., hexane). Several strategies have been used toovercome this problem, including the complexation of polarsubstrates (Castillo et al., 1994), the adsorption of polarsubstrates on inert silica-gel (Castillo et al., 1997), as well asthe use of hydrophilic solvents and/or free-solvent systemscontaining partially soluble polar substrates (Ljunger et al.,1994). The use of polar solvents or free-solvent systemscombined with an efficient removal of water from the mediahas proved to be an adequate strategy for enzymatic estersynthesis (Torres et al., 1999; Virto et al., 1999). Both directesterification and transesterification approaches have beenassayed in the lipase-catalyzed synthesis of amphiphilic es-ters. Among the numerous acyl-donors employed duringtransesterification, vinyl esters are the most popular, be-cause the vinyl alcohol formed during the process tautomer-izes to low-boiling-point acetaldehyde, thus shifting theequilibrium toward ester formation (Virto and Adlercreutz,2000).

Choline and L-carnitine [R-(-)-3-hydroxy-4-N,N,N-tri-methylaminobutyrate] are ubiquitous trimethylammoniumalcohols, which play different roles in living cells. In addi-tion to the classical functions of these alcohols, many phar-macological applications have been described for their

Correspondence to: J. L. Iborra*Permanent address: Departamento de Quımica. Facultad de Ciencias

Exactas, Universidad Nacional de Salta, Salta, Republica ArgentinaContract grant sponsor: CICYTContract grant number: BIO99-0492-C02-01

© 2003 Wiley Periodicals, Inc.

derivatives. For example, long-chain esters of choline andL-carnitine exhibit activity against bacteria, yeasts, andfungi (Ahlstroem et al., 1995; Calvani et al., 1998), whereasshort-chain esters of L-carnitine are used for cardiovasculartreatment (Arsenian, 1997), dermatoses (Cavazza andCavazza, 1994), and Alzheimer’s disease (Calvani andCarta, 1996). In all cases, both choline and L-carnitine estersare obtained by chemical synthetic processes.

This study describes, for the first time, the enzymaticsynthesis of butyryl esters of both choline and L-carnitine indry conditions, using 21 different organic solvents as or-ganic reaction media. Novozyme 435 (Candida antarticalipase B immobilized on acrylic macroporous resin) wasused as catalyst, and synthetic reactions were carried out bytransesterification, using vinyl butyrate as acyl-donor.

MATERIALS AND METHODS

Chemicals and Enzyme

Candida actarctica lipase B immobilized on a macroporousresin (Novozyme 435) was a gift from Novo-Nordisk A/S(Bagsværd, Denmark). L-carnitine, choline, butyryl choline,and �-naphthylamine were obtained from Sigma Co. (St.Louis, MO). Ethyl chloroformate, vinyl butyrate, triethyl-amine, sodium dodecylsulfate (SDS), acetanilide, molecularsieve UOP Type 4 (pore diameter 4 Å), and butyric acidwere from Fluka. All of the remaining reagents and solventswere of analytical grade.

Drying of Chemicals and Water Content Analysis

Water was removed from the solvents and vinyl esters byadding molecular sieves (0.1 g/mL), shaking the resultingmixture for 24 h at room temperature, and finally storingthem in the presence of the adsorbent. Both choline andL-carnitine were dried as follows: 0.5 g of trimethylammo-nium alcohol were added to 25 mL of dry acetone, and theresulting mixture was stirred for 1 h at room temperature.Then, the remaining solid was recovered by filtration anddried under nitrogen. After the drying process, the remain-ing water content, as determined by the Karl-Fisher (MKS210 Kyoto Electronics) method, of all the components of thereaction media ranged from 0.02% to 0.06% (w/w) for sol-vents, whereas choline and L-carnitine showed 0.5% and0.95% (w/w) water contents, respectively. Novozyme wasused without any drying process, showing a 3.5% (w/w)water content.

Synthetic Reactions

One hundred micromoles of choline or L-carnitine and 100mg of molecular sieves were placed in a screw-capped vialwith a Teflon seal. Then, 1 mL of an appropriate mixture ofvinyl ester and solvent was added, and the resulting reaction

medium was incubated with gentle shaking overnight at40°C to reach solubility equilibrium. The reaction wasstarted by adding 20 mg of immobilized enzyme and run at40°C with shaking. At regular time intervals, 50-mL ali-quots were extracted, centrifuged for 3 min at 13,000g, andthe resulting solutions were analyzed by high-performanceliquid chromatography (HPLC). At the end of the assay, theoverall solid fraction was recovered and divided in twoparts. From the first part, the immobilized enzyme particleswere recovered and then used to quantify the residual es-terase activity. The second part was suspended in dry etha-nol, then mixed for 5 min, and finally centrifuged for 3 minat 13,000g. A 50-mL aliquot was analyzed to quantify thepossible precipitated product by HPLC. Experiments werecarried out in duplicate.

HPLC Analysis

Substrate and product concentrations were determined byHPLC (Shimadzu LC10) using a LiChrocart RP18 column(12.5-cm length and 4-mm internal diameter, 5-mm particlesize, and 10-nm pore size). Samples (50 mL) containingcholine derivatives were eluted in a linear gradient (PhaseA: acetonitrile 30% with 10 mM SDS and 15 mM phospho-ric acid; Phase B: acetonitrile 60% with 10 mM SDS and15 mM phosphoric acid) at 1.5 mL min−1, using 1 mMacetanilide in ethanol, as internal standard. Elution profileswere monitored at 210 nm, and showed a detection limit of0.25 mM.

Samples containing L-carnitine and L-butyrylcarnitinewere chemically modified to a chromophore for UV detec-tion, using a modification of the method described byKagawa et al. (1999). Two hundred microliters of appropri-ately diluted sample in ethanol were added to a screw-capped vial with a Teflon seal, containing 0.2 mL of 50 mMethyl chloroformate in chloroform, 0.2 mL of 50 mM tri-ethylamine in chloroform, and 0.2 mL of 20 mM �-naph-thylamine in ethanol. The mixture was incubated at 40°Cfor 30 min, reaching the full chromophoric modification.Then, 20 �L of 0.1 M HCl was added to avoid hydrolysis ofL-butrylcarnitine to L-carnitine by TEA. The chomophoricproduct mixture was stable for 2 days at 4°C. Finally, 50 �Lof internal standard solution (100 mM acetanilide in etha-nol) was added. The final injection sample (50 mL), con-taining substrate and product derivatives at the appropriatedilution in ethanol, was eluted in a linear gradient (Phase A:acetonitrile 10% in 20 mM TEA/phosphoric acid buffer so-lution, pH 2.5; Phase B: acetonitrile 80% in 20 mM TEA/phosphoric acid buffer solution, pH 2.5) at 1 mL min−1, andmonitored at 280 nm. Standard calibration was carried outusing either L-carnitine, L-butyrylcarnitine, or butyric acidstandard solutions (from 0.01 to 1 mM) in ethanol. Theresulting straight-line equations were respectively as fol-lows: y � 0.657x + 1.12 � 10−4 (r2 � 0.998), y � 0.754x+ 2.51 � 10−4 (r2 � 0.997), and y � 0.546x + 3.56 � 10−4

(r2 � 0.995), where y is the ratio between the integrated

LOZANO ET AL.: SYNTHESIS OF ESTERS FROM TRIMETHYLAMMONIUM ALCOHOLS 353

areas of the compound and the internal standard, and x is theratio between the concentrations of the compound and theinternal standard. The detection limit exhibited by thismethod was 0.02 mM. The L-butyrylcarnitine used as stan-dard for HPLC was synthesized according to the method ofBohmer and Bremer (1968). One unit of synthetic activitywas defined as the amount of enzyme that produced 1 �molof product per minute.

Assay of Esterase Activity

The esterase activity of Novozyme (fresh and recoveredfrom organic media) was assayed tritrimetrically, using avideo-titrator VIT-90 equipped with an autoburette (ABU91) and a sample station (SAM 90, Radiometer, Copenha-gen), under the following conditions: a solution containing0.1 M KCl, 1% (w/v) sodium deoxycholate, and 6% (v/v)ethyl hexanoate was sonicated over 3 min and its pH ad-justed to 7.0 with 10 mM NaOH. Ten milliliters of thissolution was incubated at 40°C and the reaction was startedby the addition of 20 mg of Novozyme. The pH was main-tained constant at 7.0 by continuous addition of 10 mMNaOH as titrant. One unit of activity was defined as theamount of enzyme that hydrolyzes 1 �mol of ethyl hexano-ate per minute.

RESULTS AND DISCUSSION

Effect of Solubility ofTrimethylammonium Alcohols

In lipase-catalyzed reactions, a covalently linked acyl-enzyme intermediate is formed (see Fig. 1A). Nucleophilicattack of this intermediate by water resulted in ester hydro-lysis, although the presence of another nucleophile (e.g.,R�-OH) may involve the formation of the transesterificationproduct. This synthetic pathway can be regarded as a ki-netically controlled process, where the rapid accumulationof the acyl-enzyme intermediate and preferential nucleo-philic attack, determined at kT[R�-OH] >> kH[H2O], are es-sential. The first condition is enhanced by the use of acti-vated acid acyl-donors as vinyl esters, because the vinylalcohol released in the degradation of the vinyl ester tau-tomerizes to acetaldehyde, which cannot act as a substratefor the enzyme (Degueil-Castaing et al., 1987; Virto andAdlercreutz, 2000). The second condition may arise fromusing both reaction media with very low water content anda high nucleophile (R�-OH)/acyl donor concentration ratio.However, the use of quaternary ammonium alcohols (e.g.,choline or L-carnitine; Fig. 1B) as acyl-acceptors wouldgreatly limit the extension of the synthetic pathway due totheir poor solubility in many of the organic solvents used asreaction media (see Table I). Thus, an increase in the Log Pparameter of the solvent reduces the solubility of both sub-stances. The solubilization of choline and L-carnitine mol-ecules involves bond–dipole interactions with solvent mol-

ecules due to the charged nitrogen atom they contain, bothcompounds being nonsoluble in hydrophobic solvents. Inthis way, Calvani et al. (1998) described a linear correlationbetween the capacity factors (k�) in a C18 column and thecalculated Log P (CLOGP) of 34 different quaternary am-monium L-carnitine esters, which contained long-chain al-kyl substituents. The large influence of the charged nitrogenon the polar properties of these compounds is demonstratedby the low differences in the experimental k� parameter,suggesting that all the substances exhibit similar amphipath-icity, with no dependence on the nature of the alkyl sub-stituent.

In spite of the low solubility of choline and L-carnitine indry organic solvents, the synthesis of butyryl ester deriva-tives from these quaternary ammonium compounds cata-lyzed by Novozyme 435 was assayed in 21 different organicsolvents. The synthetic yields obtained with each solvent at40°C are also shown in Table I. However, it should first bepointed out that when the reactions proceeded in the ab-sence of molecular sieves no synthetic products were ob-tained, so that inclusion of this dehydrating agent in thereaction media was absolutely necessary. Also, syntheticproducts were not detected in the insoluble fraction of thereaction media, suggesting that a solubilization of trimeth-ylammonium substrates occurred during the process for sol-vents with limited solubilization ability. As can be seen, thesynthetic yield obtained was strongly dependent on the na-ture of the assayed solvent for both butyryl trimethylammo-nium esters. 2-Methyl-2-butanol was shown to be the bestsolvent for the lipase-catalyzed synthesis of butyryl choline(19.0 U/g initial synthetic activity, 91% synthetic productyield at 24 h), whereas, in the case of butyryl-L-carnitine,the best results were obtained with acetonitrile (0.17 U/ginitial synthetic activity, 66.4% synthetic product yield at 6days).

In addition, neither synthetic product was detected in re-actions carried out in any of the water-immiscible solvents.A comparison of the data included in Table I shows how thesolubility of these trimethylammonium alcohols was not theonly factor, because the best solvents for dissolving the

Figure 1. (A) Synthesis of esters from vinyl ester and alcohols catalyzedby lipase. (B) Structure of choline and L-carnitine.

354 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 3, MAY 5, 2003

substrates were not the best reaction media. For example,formamide was able to dissolve 2.5 M choline, but the initialsynthetic activity of the enzyme in this solvent (0.9 U/g)was 21-fold lower than that in 2-methyl-2-butanol. Thesefindings could be explained by both the solvent–water andsolvent–enzyme interactions that took place. In all cases(Table II), the water content of the reaction media was verylow, and the ability of the solvents to sequester essentialwater molecules from the microenvironment of the immo-bilized enzyme in these anhydrous media may lead to en-zyme deactivation by water-stripping (Klibanov, 2001; Lev-itsky et al., 2000). Table II shows the initial activity of theimmobilized enzyme for butyryl-L-carnitine synthesis in thebest solvents at 40°C, as well as the residual esterase activ-ity of the recovered immobilized enzyme from these mediaafter 6 days of operation. As can be seen, an increase insolvent polarity reduced the initial synthetic activity andincreased activity loss. The synthetic activity of lipase pro-duced good yields in those polar solvents, which showed anability to dissolve a critical amount of choline and L-carnitine (i.e., 2-methyl-2-butanol), with negligible deacti-vation effects.

Kinetic Characteristics of Synthesis ofTrimethylammonium Esters

The solvent 2-methyl-2-butanol was chosen to analyze thekinetic characteristics of the process. Figure 2 shows the

time-course of both the hydrolytic (butyric acid) and syn-thetic (butyryl choline or butyryl-L-carnitine) products ob-tained in 2-methyl-2-butanol by the Novozyme 435 action.For both trimethylammonium alcohols, the hydrolytic ac-tivity of the enzyme was clearly higher than the syntheticactivity, even when all the components of the reaction me-dia were previously dried, and a large quantity of molecularsieves (100 mg/mL) was included. Butyryl choline was syn-

Table I. Solubility of choline and L-carnitine, determined by HPLC from saturated solution, in different dry organic solvents after 24 h of magnetic stirringat 40°C.

Solvent Log P

Solubility Yield

Choline(M)

L-carnitine(mM)

BuChol at 24 hours(%)

BuCar at 6 days(%)

Tetra(ethylene-glycol) dimethyl ether −2.47 NA 0.2 NA 8.0Formamide −1.61 2.5 3.0 29.7 3.7Di(ethylene-glycol) dimethyl ether −1.30 NA 2.0 NA 20.4Dioxane −1.10 NA ND NA 0.0N,N-dimethylformamide −1.01 0.5 2.0 18.0 5.2N-methylformamide −0.97 1.7 1.0 32.0 19.7Acetonitrile −0.33 17 × 10−3 2.6 70.0 66.4Di(ethylene-glycol) diethyl ether −0.27 NA 1.0 NA 5.4Acetone −0.23 NA 1.2 NA 55.8Propionitrile 0.16 NA 0.7 NA 48.52-Methyl-2-propanol 0.35 25 × 10−3 10.5 78.0 45.2Tetrahydrofurane 0.49 NA ND NA 0.0Diethylether 0.85 NA ND NA 0.02-Methyl-2-butanol 0.89 12 × 10−3 10.5 91.0 55.93-Methyl-2-pentanol 1.71 NA 2.1 NA 7.0Benzene 2.00 ND ND 0.0 0.0Toluene 2.50 ND ND 0.0 0.0Tetrachloromethane 3.00 ND ND 0.0 0.0n-Hexane 3.50 ND ND 0.0 0.0n-Heptane 4.00 ND ND 0.0 0.0n-Octane 4.50 ND ND 0.0 0.0

Synthetic yield of butyryl choline (BuChol) and butyryl-L-carnitine (BuCar), with respect to the initial choline and L-carnitine amount (100 micromol),produced by immobilized CALB (Novozyme 435) in different dry organic solvents at 40°C. Initial vinyl butyrate concentration was 20% (v/v). ND, notdetected; NA, not assayed.

Table II. Initial water content and initial synthetic activity of immobi-lized CALB for butyryl-L-carnitine synthesis in different organic solventsat 40°C.

Solvent

Watercontent(mM)

Initialsynthetic

activity (U/g)

Residualesterase

activity (%)

Di(ethylene-glycol)dimethyl ether 47.2 0.06 48.5

N-methylformamide 45.5 0.06 49.9Acetonitrile 29.4 0.16 59.8Acetone 66.5 0.11 66.72-Methyl-2-propanol 79.4 0.12 98.72-Methyl-2-butanol 119.3 0.14 99.5

All synthetic reaction media contained 1.56 M (20% v/v) dry vinylbutyrate in dry solvent, 100 �mol in total of dry L-carnitine, 100 mg ofmolecular sieves, and 20 mg of Novozyme 435. Residual esterase activity(see “Materials and Methods”) was determined using the recovered immo-bilized enzyme from the different reaction media after 6 days.

aInitial esterase activity 22.1 U/g.

LOZANO ET AL.: SYNTHESIS OF ESTERS FROM TRIMETHYLAMMONIUM ALCOHOLS 355

thesized much faster than butyryl-L-carnitine, with a 60%butyryl choline yield being obtained after 2 h (Fig. 2A),whereas reaching the same yield of butyryl-L-carnitine took6 days (Fig. 2B).

These results were the consequence of the low nucleo-philic power of L-carnitine than of choline, due to the sec-ondary position of the hydroxyl group in the former and theprimary position in the latter (Fig. 1B). Taking into accountthe low solubility of these trimethylammonium alcohols, thehydrolytic activity seems to be a kinetic way of controllingfree water molecules for two reasons. First, in both cases,this activity was stopped when only 6% of the initial vinylbutyrate concentration (1.56 M) had been consumed. As thehydrolytic activity consumed free water molecules, it canalso be seen how the final butyric acid concentration wasabout 80% of the initial water content (see Table II),whereas the remaining water content of the medium maycorrespond to non-free water molecules. Second, becauseL-carnitine has a lower nucleophilic power than choline, theenzymatic synthesis of butyryl-L-carnitine proceeds at alower water concentration than in the case of butyryl cho-line (see Fig. 1A). Hence, no butyryl-L-carnitine was syn-thesized during the first day of the reaction, whereas thehydrolytic product was obtained very rapidly. After thehydrolytic activity had reduced the water concentration toa critical level (approx. 50 mM), butyryl-L-carnitine syn-hesis was started by the enzyme. Furthermore, it is neces-sary to point out that the concentration of both trimethyl-ammonium substrates was constant during the first-order step of the synthetic reaction, indicating that thesolid fraction of the substrate was a nucleophile reservoir.These results clearly demonstrate the kinetic control ofthe water content of the synthetic activity exhibited by theenzyme.

Ljunger et al. (1994) reported the C. antarctica lipase-catalyzed synthesis of octanoylglucose from insoluble glu-

cose and octanoic acid in acetonitrile. In this case, whereinthe glucose was not fully soluble in the solvent, the enzymewas also able to catalyze the synthetic products at high acidconcentration.

Figure 3 shows the influence of the vinyl butyrate (acyl-donor) concentration on the ability of Novozyme 435 tocatalyze the synthesis of butyryl choline and butyryl-L-carnitine in 2-methyl-2-butanol in the presence of the sametrimethylammonium alcohol (100 mmol) and molecularsieve (100 mg/mL) content. As can be seen, in both cases,the synthetic activity was clearly dependent on the concen-tration of vinyl butyrate, with the critical concentration be-ing 0.3 M. This coincides with the kinetic control of thewater concentration in the reaction media, wherein a de-crease in water content as a result of the hydrolytic activityof the enzyme allows for the expression of synthetic activ-ity. In the case of choline ester (Fig. 3A), the syntheticactivity was enhanced exponentially to its maximum level(16 U/g), which was observed at an acyl-donor concentra-tion of 1 M. However, in the case of butyryl-L-carnitinesynthesis (Fig. 3B), the maximum activity (0.15 U/g) wasreached at an acyl-donor concentration 1.56 M. At abovethis concentration, there was a gradual reduction in syn-thetic activity until no activity was seen at 8 M (100% [v/v])vinyl butyrate. In this case, vinyl butyrate plays a doublerole, as substrate and solvent, and its low polarity (LogP � 1.71) was not able to dissolve L-carnitine. In addition,it can be seen how the ability of the enzyme to synthesizebutyryl choline was 100-fold higher than in the case ofbutyryl-L-carnitine. As mentioned earlier, secondary alco-hols are poorer nucleophiles than primary alcohols. In ad-dition, from the lipase-catalyzed esterification of the sec-ondary hydroxyl group of lactic acid by decanoic acid, Fromet al. (1997) observed that the steric hindrance produced bythe presence of the carboxyl group of the nucleophile isimportant in determining the decrease in enzymatic syn-thetic activity.

Figure 2. Time-courses of reaction products (�, butyric acid; �, butyrylcholine; �, butyryl-L-carnitine) for immobilized CALB-catalyzed estersynthesis from choline (A) and L-carnitine (B) in 2-methyl-2-butanol at40°C. Both reaction media contained 1.56 M (20% [v/v]) dry vinyl buty-rate, 100 mmol (total) of dry choline or L-carnitine, 100 mg of molecularsieves, and 20 mg of Novozyme 435.

Figure 3. Effect of vinyl butyrate concentration on the ability of Novo-zyme 435 to catalyze the synthesis of butyryl choline (A, �) and butyryl-L-carnitine (B, �) in 2-methyl-2-butanol at 40°C. See Figure 2 for otherreaction conditions.

356 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 3, MAY 5, 2003

Influence of Solvent Hydrophobicity onTrimethylammonium Ester Synthesis

To ascertain the influence of the polar characteristics of thesolvent used as reaction medium, the ability of Novozyme435 to synthesize both trimethylammonium esters was as-sayed with 21 different solvents (listed in Table II) in dryconditions and the optimum vinyl butyrate concentration asdetermined earlier. Figure 4 shows the evolution of both theinitial butyryl choline and butyryl-L-carnitine synthetic ac-tivity as a function of Log P. As can be seen, the immobi-lized enzyme showed an increase in both synthetic activitieswhen Log P decreased, reaching a maximum value of −0.5to 0.5. Taking into account the constant composition of allreaction media with regard to the polarity of solvents, theseresults could be explained by both the enzyme–solvent in-teractions and the ability of solvents to reach a critical dis-solved trimethylammonium alcohol concentration. Even ifthe most hydrophilic solvents can easily dissolve these sub-strates, the increase in solvent polarity could enhance theloss of essential water molecules from the enzyme micro-environment, increasing the enzyme solvent–interaction,which may result in enzyme deactivation (Klivanov, 2001;Levitsky et al., 2000). Most hydrophobic solvents can pre-serve the essential water-shell around the enzyme, permit-ting the expression of its activity, but are not able to dissolvethe nucleophile substrate and so the synthetic reaction doesnot occur. However, the polar solvents with a limited abilityto dissolve substrates and intermediate water-mimickingproperties (e.g., tertiary alcohols) exhibited the best resultswith regard to synthetic activity and stability.

CONCLUSIONS

Immobilized Candida antarctica lipase B was able to cata-lyze ester synthesis from vinyl ester and trimethylammo-nium alcohols (e.g., choline and L-carnitine) by a transes-terification mechanism. The extremely high polarity of

these alcohols requires extremely anhydrous conditions forall the components of the reaction media and the presence ofmolecular sieves to limit the hydrolytic activity of the en-zyme. Indeed, the synthetic activity of the enzyme was con-trolled by the kinetic consumption of free water moleculesby the hydrolytic activity. Substrates need not be fullysoluble in the bulk reaction media, and only a criticalamount of soluble trimethylammonium alcohol is neces-sary. Neither non-water-miscible solvents (e.g., hexane) norwater mimics (e.g., formamide) were suitable for the lipase-catalyzed synthesis of trimethylammonium esters. Novo-zyme 435 in a midrange polar solvent (Log P of −0.5 to 0.5)exhibited the best synthetic activity, which, in the case ofcholine ester, was 100-fold higher than for L-carnitine ester.These results clearly show how the synthetic action oflipases could be enhanced by an adequate microenviron-ment.

Mirta Daz was a postdoctoral fellow from the Consejo Nacionalde Investigaciones Cientıficas y Tecnicas (CONICET), Repub-lica Argentina.

References

Ahlstroem B, Chelminska-Bertilsson M, Thompson RA, Edebo L. 1995.Long-chain alkanoylcholines: A new category of soft antimicrobialagents that is enzymatically degradable. Antimicrob Agents Chemoter39:50–55.

Arcos JA, Bernabe M, Otero C. 1998. Quantitative enzymatic productionof 1,6-diacyl fructofuranoses. Enzyme Microb Technol 22:27–35.

Arsenian MA. 1997. Carnitine and its derivatives in cardiovascular disease.Progr Cardiovasc Dis 40:265–286.

Balcao VM, Pavia AL, Malcata FX. 1996. Bioreactors with immobilizedlipases: State of the art. Enzyme Microbiol Technol 18:392–416.

Bohmer T, Bremer J. 1968. Propionylcarnitine physiological variations invivo. Biochim Biophys Acta 152:559–567.

Calvani M, Carta A. 1996. Use of acetyl-L-carnitine for treatment of pre-senile Alzheimer’s disease. Application EP 96-830532.

Calvani M, Critelli L, Gallo G, Giorgi F, Gramiccioli G, Santaniello M,Scafetta N, Tinti MO, De Angelis F. 1998. L-carnitine esters as “soft,”broad spectrum antimicrobial amphiphiles. J Med Chem 41:2227–2233.

Castillo E, Dossat V, Marty A, Condoret JS, Combes D. 1997. The role ofsilica gel in lipase-catalyzed esterification reactions of high polar sub-strates. J Am Oil Chem Soc 74:77–86.

Castillo E, Marti A, Combes D. 1994. Polar substrates for enzymic reac-tions in supercritical CO2: How to overcome the solubility limitation.Biotechnol Lett 16:169–174.

Cavazza C, Cavazza P. 1994. Use of esters of L-carnitine with hydroxyacids for producing pharmaceutical compositions for treating derma-toses. Application EP 94-108921.

Degueil-Castaing M, de Jeso B, Drouillard S, Maillard B. 1987. Enzymaticreactions in organic synthesis: 2-ester interchange of vinyl esters. Tet-rahed Lett 28:953–954.

From M, Adlercreutz P, Mattiasson B. 1997. Lipase catalysed esterificationof lactic acid. Biotechnol Lett 19:315–317.

Jaeger K-E, Reetz MT. 1998. Microbial lipases form versatile tools forbiotechnology. TIBTECH 16:396–403.

Kagawa M, Machida Y, Nishi H. 1999. Enantiomeric purity determinationof acetyl-L-carnitine by reversed-phase high-performance liquid chro-matography using chiral derivatization. J Chromatogr A 857:127–135.

Figure 4. Influence of hydrophobicity of the reaction media (measuredby Log P) on the initial enzyme activity for butyryl choline (BuCho) (�)and butyryl-L-carnitine (BuCar) (�) synthesis catalyzed by Novozyme 435at 40°C. See Table I for the list of solvents and Figure 2 for other reactionconditions.

LOZANO ET AL.: SYNTHESIS OF ESTERS FROM TRIMETHYLAMMONIUM ALCOHOLS 357

Klibanov AM. 2001. Improving enzymes by using them in organic sol-vents. Nature 409:241–246.

Levitsky VY, Lozano P, Iborra JL. 2000. Designing enzymatic kyotorphinsynthesis in organic media with low water content. Enz Microb Tech-nol 26:608–613.

Ljunger G, Adlercreutz P, Mattiasson B. 1994. Lipase catalyzed acylationof glucose. Biotechnol Lett 16:1167–1172.

Torres C, Bernabe M, Otero C. 1999. Part II. Two enzymatic proceduresfor the selective synthesis of malic acid monoesters. Enz Microb Tech-nol 25:753–761.

Torres C. Otero C. 1999. Part I. Enzymatic synthesis of lactate and gly-colate esters of fatty alcohols. Enz Microb Technol 25:745–752.

Virto C, Adlercreutz P. 2000. Lysophosphatidylcholine synthesis withCandida antarctica lipase B (Novozyme 435). Enz Microb Technol26:630–635.

Virto C, Svensson I, Adlercreutz P. 1999. Enzymatic synthesis of lyso-phosphatidic acid an phosphatidic acid. Enz Microb Technol 24:651–658.

Virto C, Svensson I, Adlercreutz P. 2000. Candida antarctica lipase B-catalysed synthesis of dihydroxiacetone fatty acid esters. Biocatal Bio-trans 18:13–29.

358 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 82, NO. 3, MAY 5, 2003