Ž .Journal of Molecular Catalysis B: Enzymatic 6 1999 407–410

Lipase-catalysed alkoxycarbonylation of thymidine: influence ofthe carbonate on the regioselectivity of the process

Luis F. Garcıa-Alles, Vicente Gotor )´Departamento de Quımica Organica e Inorganica, Facultad de Quımica, UniÕersidad de OÕiedo, 33071 OÕiedo, Spain´ ´ ´ ´

Accepted 16 October 1998

Abstract

Serine-hydrolases are known to catalyse alkoxycarbonylation processes. In the field of nucleoside chemistry they can beused to perform regioselective transformations at the carbohydrate frame. In this paper we present data regarding theinfluence of the structure of the carbonate on the regioselectivity of the alkoxycarbonylation of thymidine catalysed byCandida antarctica lipase. As expected, assuming the formation of an acyl-enzyme intermediate, the structure of the leavinggroup portion of the carbonate has no effect, whereas the remaining alcoholic portion of the carbonate has a marked effect.q 1999 Elsevier Science B.V. All rights reserved.

Keywords: Enzymes; Alkoxycarbonylation; Thymidine; Organic solvents; Regioselectivity; Mechanism

1. Introduction

Enzymes are widely accepted synthetic tools,which can be used to promote selective transfor-

w xmations of substrates 1 . Among them, serine-hydrolases are probably the most commonlyemployed biocatalysts. Serine-hydrolases catal-

w xyse a wide variety of acylation reactions 2w xincluding alkoxycarbonylations 3–9 . The im-

portance of these transformations, especially inw xprotective group chemistry 10 , and the selec-

tivity inherent to enzyme-catalysed reactionsrender these processes extraordinarily interest-ing. On the basis of these arguments, some

) Corresponding author. Tel.: q34-985-103-448; Fax: q34-985-103-448; E-mail: vgs@sauron.quimica.uniovi.es

stereo-, regio- and chemoselective syntheticroutes have been recently described in our groupw x4–9 . In concrete, this methodology has beensuccessfully applied to develop alternative pro-cedures toward 5X- and 3X-carbonates of nucleo-

w xsides 4,5,8 . The regioselectivity in these caseswas governed by the simple choice of the bio-catalyst employed.

Despite the aforementioned interest, to thebest of our knowledge, a systematic analysis ofthe regioselectivity of these transformations as afunction of the alkoxycarbonylating agent em-ployed has not yet been published. For thatreason we have measured the influence of suchfactor, in the alkoxycarbonylation of thymidinecatalysed by the lipase from Candida antarc-tica. The experimental data obtained are shownin the present work.

1381-1177r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.Ž .PII: S1381-1177 99 00004-1

( )L.F. Garcıa-Alles, V. GotorrJournal of Molecular Catalysis B: Enzymatic 6 1999 407–410´408

2. Experimental

2.1. Materials

Lipase from C. antarctica SP 435L was gent-ly gifted by Novo Nordisk. It was stored undernitrogen at 58C. Thymidine and other chemicalswere purchased from Sigma and Aldrich. THFwas dried by reflux over and distillation fromsodium. Pyridine was freshly distilled fromKOH. Vinyl carbonates 2a–f were prepared bytreating the corresponding alcohol or oxime withvinylchloroformate in pyridine at 08C. In thesame way, acetone oxime carbonates 2g–l wereobtained by treating acetone oxime with thecorresponding chloroformate. Products wereisolated by conventional procedures, as previ-

w xously described 4,5 .

2.2. Determination of the initial rate of forma-tion of the 5X- and 3X-carbonates of thymidine

A 10 ml solution of dry THF containingŽ . Ž .thymidine 12.4 mM , carbonate 2a–l 40 mM

Žand a small amount of the internal standard 1.mgrml of pertrimethylsilylated uridine was

prepared in an erlenmeyer flask. The solutionwas stopped with a rubber septum, introduced

Ž .in an orbital shaker 250 rpm , and allowed tostand for at least 30 min at 308C. The immo-

Ž .bilised CAL enzyme 50 mg was then added tothe solution, and at regular intervals of timeŽ .usually 15–20 min 200-ml samples were taken

Ž .from the flask 6–10 samples per reaction . Forthe reactions with carbonates 2i–l, samples weretaken after several hours, since they proceededvery slowly. Finally, the solvent present in thesesamples was removed under vacuum. For eachcarbonate, experiments were done by triplicate.

2.3. Silylation and chromatographic analysis

The silylation procedure employed had beenw xalready described by Sweely et al. 11 : 100 ml

of pyridine, 8 ml of 1,1,1,3,3,3-hexamethyldi-silazane and 4 ml of trimethylsilane chloridewere added to each reaction sample. The mix-ture was stirred and the reaction allowed toreach completion after 5–10 min. At this time,1 ml was withdrawn from each sample mixturewith a Hamilton syringe, and injected into aHewlett-Packard 5890 Series II gas chromato-graph equipped with an HP1 crosslinked methylsilicone gum capillary column. For the detectionof the products of reaction with carbonates 2a–i,isocratic conditions were employed: an oventemperature of 2608C and a head-column pres-sure of 120 kPa. However, for the rest of reac-tions this pressure was increased to 140 kPa,and the oven temperature was programmed tobe 2808C during 5.5 min, then a temperatureramp of 208Crmin, and finally 7 min at 2908C.

Ž . X XRetention times min for the 5 r3 -carbonateswere as follows: in the reaction with 2a–fŽ . Ž . Ž .8.9r9.4 , 2g 8.6r9.2 , 2h 10.2r10.8 , 2iŽ . Ž . Ž .10.9r11.8 , 2j 8.8r9.2 , 2k 11.4r11.9 , 2iŽ .12.8r14.2 . Products were detected with the

Ž .aid of a flame ionisation detector 3508C , theirsignal peaks integrated, and their concentrationsdetermined by comparison with the internalstandard. In this way, the concentration of thecarbonates 3, 4 could be plotted against thereaction time, and reaction rates were calcu-lated. Reactions never exceeded 10% of totalconversion.

3. Results and discussion

One of the most interesting processes devel-oped in our group has been the reaction ofalkoxycarbonylation of nucleosides catalysed by

w xenzymes 4,5,8 . Several factors are alreadyknown to affect the regioselectivity of this kindof processes, being the most important one theselection of the biocatalyst. Hence, the reactingposition can be easily turned from the 5X to the3X hydroxyl group of the nucleoside by simplychanging the enzyme. However, it still remained

( )L.F. Garcıa-Alles, V. GotorrJournal of Molecular Catalysis B: Enzymatic 6 1999 407–410´ 409

Fig. 1. CAL-catalysed vinyloxycarbonylation of thymidine using carbonates which present different leaving groups.

to be investigated another factor that might haveinfluence on the regioselectivity pattern: thealkoxycarbonylating agent employed.

The alkoxycarbonylation of thymidine catal-ysed by the lipase from C. antarctica lipaseŽ .CAL was the reaction selected to performsuch study. Therefore, several carbonates 2a–lwere synthesised and their effect on the regiose-lectivity of the reaction was measured. Theresults have been organised in two groups:

Ž .1 Carbonates carrying different leavingŽ .group X , but the same alkoxycarbonylatingŽ .portion vinyl carbonates 2a–f . All of these

carbonates give place to the same couple ofŽ .products 3 and 4, Fig. 1 . Initial rates for the

formation of these two possible vinyl regioiso-Žmers were measured, and their ratios regiose-

. Ž .lectivity calculated Table 1 . As we can see,initial rate values were dependent on the leavinggroup present on the starting carbonate. Thisindicates that ‘acylation’ is rate-limiting underthese reaction conditions. However, regioselec-tivity values were independent of the X leavinggroup. This conclusion was not unexpected sinceit is probably an indirect evidence of the occur-rence of a common intermediate in the reaction

Žmechanism the vinyloxycarbonyl-enzyme inter-.mediate when all of these carbonates are em-w xployed 12 .

It must be pointed out that in the case of theoxime carbonates 2e,f the regioselectivity val-ues are slightly higher than for the rest ofcarbonates. Only with these two carbonates theratio of 5Xr3X product concentration increases as

Table 1Ž . Ž .Regioselectivity of CAL-catalysed alkoxycarbonylation of thymidine, as a function of both the leaving group X and the alkoxy group R

present in the starting carbonatea

X b c dŽ .Carbonate X V8 5 Regioselectivity Carbonate R Regioselectivitye e2a Ph– – – 2g Me– 2.5"5%

i2b p-NO Ph– 3 1.5"9% 2h Propenyl– 2.6"8%2

2c o-NO Ph– 1 1.7"10% 2i Allyl– 1.0"7%2

2d CCl CH – 1 1.5"9% 2j CCl CH – 1.7"6%3 2 3 2Ž .2e CH C5N– 8 2.1"14% 2k Ph– 20"15%3 2

2f Cy5N– 7 1.9"11% 2l PhCH – 4.7"6%2

a Mean values from experiments performed by triplicate in THF at 308C.b X Ž . y7 Ž .Initial rate for the formation of the 5 -regioisomer 3 , relative values: 1 approximately equals 6.6=10 Mrmin for a CALconcentration of 1 mgrml.c Ž X . Ž X .RegioselectivitysV8 5 rV8 3 .d w X x w X xRegioselectivitys 5 r 3 .eFormation of products was not detected at this temperature.

( )L.F. Garcıa-Alles, V. GotorrJournal of Molecular Catalysis B: Enzymatic 6 1999 407–410´410

Fig. 2. CAL-catalysed alkoxycarbonylation of thymidine employing different acetone oxime carbonates.

the reaction takes place. 1 This effect is respon-sible for the higher regioselectivity valuesobtained with 2e,f with regard to the rest ofvinylcarbonates. In fact, the value for the 5Xr3X

concentration relationship extrapolated to time 0Ž .would be closer around 1.7 to the regioselec-

tivity values obtained with the other carbonates.Ž .2 Different carbonates carrying the same

Ž .leaving group 2g–l . In this case, different pairof products are formed depending on the car-

Ž .bonate employed Fig. 2 results have beencollected in Table 1. 2 As we can see, regiose-lectivity is now spread in a wide range ofvalues, going from null regioselection in thecase of the reaction with the allyl carbonate 2i,to the highly regioselective reaction with thephenyl carbonate 2k. 3 In the last case, the5X-carbonate of thymidine is formed 20-foldfaster than the 3X-carbonate. 4 Obviously, thevariation in regioselectivity values comes nowfrom the generation of, and product evolutionfrom different covalent intermediates.

1 One suggestion to explain this observation would be that oncethe oxime leaving group is released from the carbonate, it behavesas a molecular lubricant, modifying the structural properties of thecatalyst, and consequently its regioselectivity.

2 Initial rates could not be determined with most of carbonates2g–l, since reactions proceeded too slowly. For that reason,regioselectivity values were taken to be the mean ratio of 5Xr3X

product concentrations at several reaction times.3 All thymidine carbonates 3, 4 were isolated and characterized

by 1H and 13C-NMR spectroscopy. Most of them have beenw xalready described 4,5,8 .

4 The high regioselectivity of this reaction had been alreadyexploited in a chemoenzymatic route toward 3X-amino-3X-de-

w xoxyxylonucleosides 8 .

In conclusion, data has been presented in thepresent work which reflect that the nature of thecarbonate employed can have a profound influ-ence on the regioselectivity of the biocatalyticalkoxycarbonylation reaction. However, the in-fluence of the leaving group present in thereacting carbonate has been shown to be practi-cally null.

Acknowledgements

Financial support of this work by CICYTŽ .project BIO-95-0687 , is gratefully acknowl-edged. L.F.G-A. also thanks to the II PRIŽ .Asturias for a doctoral scholarship.

References

w x1 C.-H. Wong, G.M. Whitesides, Enzymes in Synthetic Or-ganic Chemistry, Pergamon, Oxford, 1994.

w x Ž .2 Y.F. Wang, C.-H. Wong, J. Org. Chem. 53 1988 3127.w x Ž .3 D. Pioch, P. Lozano, J. Graille, Biotechnol. Lett. 13 1991

633.w x Ž .4 F. Morıs, V. Gotor, J. Org. Chem. 57 1992 2490.´w x Ž .5 F. Morıs, V. Gotor, Tetrahedron 48 1992 9869.´w x6 S. Fernandez, R. Brieva, F. Rebolledo, V. Gotor, J. Chem.´

Ž .Soc., Perkin Trans. 1 1992 2885.w x Ž .7 M. Pozo, V. Gotor, Tetrahedron: Asymmetry 6 1995 2797.w x8 L.F. Garcıa-Alles, J. Magdalena, V. Gotor, J. Org. Chem. 61´

Ž .1996 6980.w x Ž .9 M. Ferrero, S. Fernandez, V. Gotor, J. Org. Chem. 62 1997´

4358.w x10 T.V. Greene, P.G.M. Wuts, Protective Groups in Organic

Chemistry, Wiley, Chichester, 1991.w x11 C.C. Sweely, R. Bentley, M. Makita, W.W. Wells, J. Am.

Ž .Chem. Soc. 85 1963 2497.w x12 A.R. Fersht, Enzyme Structure and Mechanism, Freeman,

New York, 1885, pp. 200–211.