Enzymatic parallel kinetic resolution of mixtures of d / l 2′-deoxy and ribonucleosides: An approach for the isolation of β- L -nucleosides

Article abstract of DOI:10.1021/jo101368z

We have developed a lipase-catalyzed parallel kinetic resolution of mixtures of β-d/l-nucleosides. The opposite selectivity during acylation exhibited by Pseudomonas cepacia lipase (PSL-C) with β-d- and β-l-nucleosides furnished acylated compounds that have different R _f values. As a consequence, isolation of both products was achieved by simple column chromatography. Computer modeling of the transition-state analogues during acylation of β-d- and β-l-2′-deoxycytidine with PSL-C was carried out to explain the high selectivity. PSL-C favored the 3′-O-levulination of the β-d enantiomer, whereas the 5′-OH group was acylated in 2′-deoxy-β-l-cytidine. In both cases, the cytosine base was placed in the alternate hydrophobic pocket of PSLs substrate-binding site, where it can form extra hydrogen bonds (in addition to the five essential catalytically relevant hydrogen bonds) that stabilize these intermediates catalyzing the selective acylation of β-d/l-nucleosides.

Full text of DOI:10.1021/jo101368z

pubs.acs.org/joc
Enzymatic Parallel Kinetic Resolution of Mixtures of D/L 20-Deoxy and
Ribonucleosides: An Approach for the Isolation of β-L-Nucleosides†
‡
Saul Martınez-Montero, Susana Fernandez, Yogesh S. Sanghvi, Vicente Gotor,* and
‡
§
,‡
ꢀ
ꢀ
Miguel Ferrero*,‡
‡
´
ꢀ
ꢀ
´
Departamento de Quımica Organica e Inorganica and Instituto Universitario de Biotecnologıa de Asturias,
Universidad de Oviedo, 33006-Oviedo (Asturias), Spain, and §Rasayan Inc., 2802 Crystal Ridge Road,
Encinitas, California 92024-6615
MFerrero@uniovi.es; VGS@uniovi.es
Received July 12, 2010
We have developed a lipase-catalyzed parallel kinetic resolution of mixtures of β-D/L-nucleosides.
The opposite selectivity during acylation exhibited by Pseudomonas cepacia lipase (PSL-C) with β-D-
and β-L-nucleosides furnished acylated compounds that have different Rf values. As a consequence,
isolation of both products was achieved by simple column chromatography. Computer modeling of
the transition-state analogues during acylation of β-D- and β-L-20-deoxycytidine with PSL-C was
carried out to explain the high selectivity. PSL-C favored the 30-O-levulination of the β-D enantiomer,
whereas the 50-OH group was acylated in 20-deoxy-β-L-cytidine. In both cases, the cytosine base was
placed in the alternate hydrophobic pocket of PSL’s substrate-binding site, where it can form extra
hydrogen bonds (in addition to the five essential catalytically relevant hydrogen bonds) that stabilize
these intermediates catalyzing the selective acylation of β-D/L-nucleosides.
Introduction
Recently, telbivudine (LdT; 4), the L-isomer of thymidine,
was approved for the treatment of HBV infection. Clinical
trials have shown LdT to be significantly more effective than
lamivudine or adefovir and less likely to cause resistance.3
Another anti-HBV L-nucleoside is valtorcitabine4 (val-LdC;
5), the 30-O-valinyl ester of L-deoxycytidine, which has been
developed as a fixed dose combination with LdT. In clinical
trials, the combination of the two agents had demonstrated
greater antiviral activity than either drug alone. One promising
candidate for the treatment of chronic HBV is β-L-20,30-
dideoxy-30-thia-5-fluorocytidine (emtricitabine, FTC; 2), the
5-fluorinated analogue of 3TC, available in combination with
other antiretroviral agents as an anti-HIV therapeutic agent.
This drug is currently being evaluated as a potential treatment
for chronic HBV.5
In recent years, unnatural β-L-nucleoside analogues have
received much attention as a new class of antiviral and anticancer
agents.1 Since the discovery of lamivudine2 (L-20,30-dideoxy-30-
thiacytidine, 3TC; 1, Chart 1) for the treatment of human
immunodeficiency virus (HIV) and hepatitis B virus (HBV)
infections, various L-nucleoside derivatives have been synthe-
sized and evaluated for their potential therapeutic applications.
†
ꢀ
ꢀ
ꢀ
In memoriam to Professor Jose Manuel Concellon.
(1) See a mini-review: Mathe, C.; Gosselin, G. Antiviral Res. 2006, 71,
276–281.
(2) (a) Jarvis, B.; Faulds, D. Drugs 1999, 58, 101–141. (b) Cameron, J. M.;
Collins, P.; Daniel, M.; Storer, R. Drugs Future 1993, 18, 319–323. (c) Chang,
C.-N.; Doong, S.-L.; Zhou, J. H.; Beach, J. W.; Jeong, L. S.; Chu, C. K.; Tsai,
C.-H.; Cheng, Y.-C. J. Biol. Chem. 1992, 267, 13938–13942. (d) Doong, S. L.;
Tsai, C.-H.; Schinazi, R. F.; Liotta, D. C.; Cheng, Y.-C. Proc. Natl. Acad.
Sci. U. S. A. 1991, 88, 8495–8499.
(3) Charuworn, P.; Keeffe, E. B. Clin. Med. Ther. 2009, 1, 157–166.
(4) (a) Gosselin, G.; Pierra, C.; Benzaria, S.; Dukhan, D.; Imbach, J.-L.;
Loi, A. G.; La Colla, P.; Cretton-Scott, E.; Bridges, E. G.; Standring, D. N.;
Sommadossi, J. P. β-L-20-Deoxythymidine (L-dT) and β-L-20-Deoxycytidine
(LdC): How Simple Structures can be Potent, Selective and Specific Anti-
HBV Drugs. In Frontiers in Nucleosides and Nucleic Acids; Schinazi, R. F.,
Liotta, D. C., Eds.; IHL Press: Tucker, GA, 2004; pp 309-317. (b) Licensed
by Novartis. It has been discontinued in 2008 for reasons that were not
disclosed: Ryder, N. S. Expert Opin. Invest. Drugs 2010, 19, 1–21.
(5) (a) Furman, P. A.; Davis, M.; Liotta, D. C.; Paff, M.; Frick, L. W.;
Nelson, D. J.; Dornsife, R. E.; Wurster, J. A.; Wilson, L. J.; Fyfe, J. A.;
Tuttle, J. V.; Miller, W. H.; Condreay, L.; Averett, D. R.; Schinazi, R. F.;
Painter, G. R. Antimicrob. Agents Chemother. 1992, 36, 2686–2692. (b)
Schinazi, R. F.; McMillan, A.; Cannon, D.; Mathis, R.; Lloyd, R. M.; Peck,
A.; Sommadossi, J.-P.; Clair, M. St.; Wilson, J.; Furman, P. A.; Painter,
G. R.; Choi, W. B.; Liotta, D. C. Antimicrob. Agents Chemother. 1992, 36,
2423–2431. (c) Jenh, A. M.; Thio, C. L.; Pham, P. A. Pharmacotherapy 2009,
29, 1212–1227. (d) Saag, M. S. Clin. Infect. Dis. 2006, 42, 126–131.
DOI: 10.1021/jo101368z
r
Published on Web 09/09/2010
J. Org. Chem. 2010, 75, 6605–6613 6605
2010 American Chemical Society

Martınez-Montero et al.
JOCArticle
CHART 1. β-L-Nucleosides Used As Antiviral and Antitumor
Agents
analogue to show potent preclinical antitumor activity.10
Use of L-nucleoside 3 is mainly being studied in patients with
advanced pancreatic cancer or blast phase chronic myelo-
genous leukemia.
Interestingly, novel β-L-adenosine analogues have been
synthesized and evaluated as cardioprotective agents. Thus,
β-L-30-amino-30-deoxy-N6-dimethyladenosine was found to
enhance functional recovery from ischemia.11 On the other
hand, β-L-adenosine and β-L-thymidine (4) were selectively
transported into malaria-infected cells through the induced
transporter.12
General favorable features of β-L-nucleosides are lower
cellular toxicity and more resistance to degrading enzymes
than their D-counterparts.13 Several routes have been re-
ported for the synthesis of β-L-nucleoside derivatives. How-
ever, formation of a mixture of β-D- and β-L-nucleosides is a
quite common occurrence during the preparation of these
compounds. Due to the increased therapeutic applications of
L-nucleosides, there is a need for a separation method that
permits easy isolation of β-D-nucleosides from β-L-nucleo-
sides. Enzymes are useful catalysts for the separation of
racemic mixtures of nucleosides.14 Enzymatic resolutions
were performed for isolating the chiral forms with high
enantiomeric excess. Relevant examples are the pig liver
esterase-mediated kinetic resolution of β-D- and β-L-config-
ured dideoxynucleosides (10, Chart 2),15 the enantioselective
synthesis of bicyclo[3.1.0]hexane carbocyclic nucleosides
(11) via a lipase-catalyzed asymmetric acetylation,16 or the
preparative scale resolution of spiro[2.3]hexane carbocyclic
nucleosides (12) by Pseudomonas cepacia lipase.17 Recently,
preparation of isoxazolidinyl nucleoside (13) enantiomers
by double lipase-catalyzed resolution has been described.18
Taking into account the opposite preference shown for
Pseudomonas cepacia lipase during the acylation or hydro-
lysis reactions of β-D- and β-L-20-deoxynucleosides, our
preliminary work described the resolution of β-D/L-thy-
midine.19a Herein, we report the validation and extension
of this procedure to the resolution of other pyrimidine and
purine 20-deoxynucleoside derivatives. Also a study of the
enzymatic acylation of ribonucleosides and its application in
the resolution of racemic mixtures has been carried out. In
addition, the results presented here have been rationalized
using a computational approach.
A β-L-nucleoside analogue that has received much attention
as an HBV agent is clevudine [L-FMAU, 1-(20-fluoro-5-methyl-
β-L-arabinofuranosyl)uracil], but its phase III clinical trial was
terminated due to some myopathy cases in patients.6
Additional β-L-nucleosides are under clinical investigation
as potential antiviral agents. These include the unsaturated
nucleoside analogues β-L-20,30-didehydro-20,30-dideoxy-5-
fluorocytidine (elvucitabine, β-L-d4FC; 6), β-L-20,30-didehy-
dro-20,30-dideoxy-30-fluorocytidine (pentacept, β-L-30-Fd4C; 7),
β-L-20,30-didehydro-20,30-dideoxy-20-fluorocytidine (β-L-20-
Fd4C; 8),7 and the L-enantiomer of ribavirin, 1-β-L-ribo-
furanosyl-1,2,4-triazole-3-carboxamide (levovirin; 9), which
was investigated as a potential anti-hepatitis C virus (HCV)
agent.8 However, the limited absorption of levovirin induced
the development of the 50-valinate monoester of levovirin,
namely, R1518, a prodrug that results in improved bioavail-
ability and significantly higher plasma concentrations of
levovirin than oral administration of levovirin.9
β-L-Nucleoside analogues are also promising anticancer
agents. Troxacitabine (3) is the first unnatural β-L-nucleoside
Results and Discussion
The potential therapeutic application of val-LdC for the treat-
ment of HBV disease motivated us to develop a successful
(6) (a) Discovered: Chu, C. K.; Ma, T. W.; Shanmuganathan, K.; Wang,
C.-G.; Xiang, Y.-J.; Pai, S. B.; Yao, G.-Q.; Sommadossi, J.-P.; Cheng, Y.-C.
Antimicrob. Agents Chemother. 1995, 39, 979–981. (b) Terminated: Drugs.
com, Princeton, NJ, April 20, 2009.
(7) Komiotis, D.; Manta, S.; Tsoukala, E.; Tzioumaki, N. Anti-Infective
Agents Med. Chem. 2008, 7, 219–244.
(11) Kasiganesan, H.; Wright, G. L.; Chiacchio, M. A.; Gumina, G.
Bioorg. Med. Chem. 2009, 17, 5347–5352.
(12) Gero, A. M.; Dunn, C. G.; Broen, D. M.; Pulenthiran, K.; Gorovits,
E. L.; Bakos, T.; Weis, A. L. Curr. Pharm. Des. 2003, 9, 867–877.
(13) (a) Gumina, G.; Chong, Y.; Choo, H.; Song, G.-Y.; Chu, C. K. Curr.
Top. Med. Chem. 2002, 2, 1065–1086. (b) Gumina, G.; Song, G.-Y.; Chu,
C. K. FEMS Microbiol. Lett. 2001, 202, 9–15. (c) Jurovcik, M.; Holy, A.;
Sorm, F. FEBS Lett. 1971, 18, 274–276.
(8) Walker, M. P.; Appleby, T. C.; Zhong, W.; Lau, J. Y. N.; Hong, Z.
Antiviral Chem. Chemother. 2003, 14, 1–21.
(9) Huang, Y.; Ostrowitzki, S.; Hill, G.; Navarro, M.; Berger, N.; Kopeck,
P.; Mau, C. I.; Alfredson, T.; Lal, R. J. Clin. Pharmacol. 2005, 45, 578–588.
(10) (a) Damaraju, V. L.; Bouffard, D. Y.; Wong, C. K. W.; Clarke, M. L.;
Mackey, J. R.; Leblond, L.; Cass, C. E.; Grey, M.; Gourdeau, H. BMC Cancer
2007, 7, 121. (b) Swords, R.; Giles, F. Hematology 2007, 12, 219–227. (c)
Adema, A. D.; Radi, M.; Daft, J.; Narayanasamy, J.; Hoebe, E. K.; Alexander,
L. E.; Chu, C. K.; Peters, G. J. Nucleosides Nucleotides Nucleic Acids 2007, 26,
1073–1077. (d) Gumina, G.; Chong, Y.; Chu, C. K. L-Nucleosides as Chemo-
therapy Agents. In Cancer Drug Discovery and Development: Deoxynucleo-
side Analogs in Cancer Therapy; Peters, G. J., Ed.; Humana Press: Totowa,
NJ, 2006; pp 173-198. (e) Lapointe, R.; Letourneau, R.; Steward, W.;
Hawkins, R. E.; Batist, G.; Vincent, M.; Whittom, R.; Eatock, M.; Jolivet,
J.; Moore, M. Ann. Oncol. 2005, 16, 289–293.
(14) (a) Ferrero, M.; Gotor, V. Chem. Rev. 2000, 100, 4319–4347. (b)
Ferrero, M.; Gotor, V. Monatsh. Chem. 2000, 131, 585–616.
€
(15) Albert, M.; De Souza, D.; Feiertag, P.; Honig, H. Org. Lett. 2002, 4,
3251–3254.
(16) Yoshimura, Y.; Moon, H. R.; Choi, Y.; Marquez, V. E. J. Org.
Chem. 2002, 67, 5938–5945.
(17) Bondada, L.; Gurnina, G.; Nair, R.; Ning, X. H.; Schinazi, R. F.;
Chu, C. K. Org. Lett. 2004, 6, 2531–2534.
(18) Carnovale, C.; Iannazzo, D.; Nicolosi, G.; Piperno, A.; Sanfilippo,
C. Tetrahedron: Asymmetry 2009, 20, 425–429.
6606 J. Org. Chem. Vol. 75, No. 19, 2010

Martınez-Montero et al.
JOCArticle
CHART 2. Enzymatic Resolutions for the Synthesis of
β-L-Nucleosides
SCHEME 2
SCHEME 1
30-hydroxyl group of β-D-nucleosides, whereas the same lipase
PSL-C exhibited an opposite selectivity for the β-L-isomer,
catalyzing exclusively the acylation of the 50-hydroxyl group.
The mixture of the two acylated products 16a and 17a showed
different Rf values on the TLC and are easily separated by silica
column chromatography (entry 1, Table 1). Enantiomeric excess
were determined with chiral HPLC analysis by comparing the
product from the enzymatic reaction and an artificial racemic
sample. The data clearly supported the enantiomeric purity of
the nucleosides (ee >99% by HPLC).
Next, we decided to evaluate the same process for the
separation of a mixture of N6-benzoyl-20-deoxy-β-D/L-ade-
nosine (14b/15b). After 24 h, the reaction was incomplete and
a large amount of insoluble starting material was observed in
the reaction mixture. In order to improve the solubility of the
nucleosides 14b/15b and reach total conversion, more dilute
conditions (0.05 M instead of 0.1 M) were used. However,
unreacted nucleoside still remained in the solution. Next, a
higher reaction temperature was employed to drive the
reaction to completion. We speculate that the higher tem-
perature offered improved flexibility for binding to the active
site on PSL-C. Entry 2 in Table 1 shows successful conver-
sion under these conditions, furnishing 16b and 17b in high
yields. We were pleased to note that the PSL-C exhibited the
same regiospecificity with the deoxycytidine and deoxyade-
nosine derivatives. Again, this lipase catalyzes the acylation
at the secondary hydroxyl group of β-D-dABz with remark-
able selectivity. Both compounds 16b and 17b were obtained
with >99% ee determined by chiral HPLC.
TABLE 1. Enzymatic Resolution of D/L-20-Deoxynucleoside Racemic
Mixtures with PSL-Ca
16
17
conc
yield
ee
yield
(%)b
ee
entry substrate (M) T (°C) t (h) (%)b
(%)c
(%)c
1
2
14a/15a 0.1
14b/15b 0.05
30
45
24
20
93
80
>99
>99
82
87
>99
>99
aRatio of substrate:PSL-C, 1:3 (w/w); 3 equiv of acetonoxime
c
levulinate. bCalculated after column chromatography. Calculated by
chiral HPLC analysis.
method for the separation of a β-D/L-dC mixture. For that, we
selected Pseudomonas cepacia lipase (PSL-C) as the catalyst and
acetonoxime levulinate as the acylating agent since these condi-
tions have proven to be excellent during the acylation of β-D-
dCBz and β-L-dCBz.19 Treatment of N-benzoyl-β-D/L-dC (14a/
15a)20 with acetonoxime levulinate in the presence of PSL-C at
30 °C in THF afforded a mixture of two acylated products
The resolution of N2-isobutyryl-20-deoxy-β-D/L-guano-
sine resulted in incomplete acylation reaction catalyzed by
PSL-C.19a The acylation on β-D-dGIbu (14c) afforded a
mixture of mono- and di-O-levulinyl derivatives (18-20),
in addition to unreacted starting nucleoside (Scheme 2). In
order to reach 100% conversion, the process was carried out
with a large excess of acylating agent (4 equiv). However, the
reaction was incomplete even after 6 days probably due to
the low solubility of the starting nucleoside in THF (see
Table 2). The use of cosolvents such as pyridine or DMF
inhibited the activity of the enzyme. In recent years, ionic
liquids as a choice of solvent for biocatalysis have shown
promising results.21 Therefore, 1-butyl-3-methylimidazolium
1
(Scheme 1). After workup, H NMR spectra confirmed the
formation of β-D-30-O-levulinyl ester 16a and β-L-50-O-levulinyl
derivative 17a as the sole products. The formation of these
compounds is attributed to the opposite acylation preference
exhibited by PSL-C in β-D- and β-L-20-deoxynucleosides. Thus,
PSL-C catalyzes with total regioselectivity the acylation at the
ꢀ
(19) (a) Garcıa, J.; Fernandez, S.; Ferrero, M.; Sanghvi, Y. S.; Gotor, V.
Org. Lett. 2004, 6, 3759–3762. (b) Garcıa, J.; Fernandez, S.; Ferrero, M.;
ꢀ
Sanghvi, Y. S.; Gotor, V. Tetrahedron: Asymmetry 2003, 3533–3540.
(20) The equimolar mixture of the two nucleosides was prepared by
mixing the equal weight of β-D-dCBz (14a) and β-L-dCBz (15a). Mixtures of
14b and 15b, and 23a and 25a were prepared in a similar manner.
J. Org. Chem. Vol. 75, No. 19, 2010 6607

Martınez-Montero et al.
JOCArticle
TABLE 2. Solubility of Nucleosides in THF
SCHEME 4
solubility
(mg/mL)
entry
nucleoside
1
2
3
4
5
6
7
T
4.4
2.1
5.4
0.4
3.0
0.5
,0.1
dCBz
dABz
dGIbu
U
ABz
CBz
SCHEME 3
β-D-uridine (23a) and N6-benzoyl-β-D-adenosine (23b) with
acetonoxime levulinate in the presence of CAL-B afforded
the 50-O-levulinyl esters 24a and 24b with excellent regios-
electivity and high yields in short reaction times (Scheme 4,
entries 1 and 2 of Table 3). Total selectivity was also observed
when the same process was carried out with N4-benzoyl-β-D-
cytidine (23c). However, the reaction was slower and did
not go to completion, despite the large excess of acylating
agent (9 equiv vs 3 equiv). This is probably due to the poor
solubility of the starting nucleoside in the reaction medium
(see Table 2). More dilute conditions and a substrate:CAL-B
ratio of 1:2 (w/w) drove the reaction to completion and
furnished 24c in 93% yield (entry 4, Table 3). The acylation
of 23a with PSL-C took place with poor selectivity, furnish-
ing a mixture of 20-O-Lev- and 30-O-Lev-protected nucleo-
sides as the major products (entry 5, Table 3). In the case of
β-D-ABz, the unreacted starting material was observed after
24 h (entry 6, Table 3). The low solubility of β-D-ABz in THF
and slow reactivity of ribonucleosides in the presence of
PSL-C may have been the cause of failure to acylate. More
dilute conditions increased the conversion rate to 22% but low
selectivity persisted (entry 7, Table 3). This lipase also dis-
played low selectivity during acylation of β-D-CBz (entry 8,
Table 3).
Acylation reaction catalyzed by CAL-B on β-L-uridine
(25a) and N6-benzoyl-β-L-adenosine (25b) afforded exclu-
sively 50-O-levulinyl-β-L-uridine (26a) and N6-benzoyl-50-O-
levulinyl-β-L-adenosine (26b), respectively (Scheme 5, entries
1 and 3 of Table 4). Next, we performed the acylation
catalyzed by PSL-C. The latter exhibited the same behavior as
CAL-B in L-ribonucleosides, catalyzing selectively 50-O-acyla-
tion of β-L-uridine (entry 2, Table 4) and N6-benzoyl-β-L-
adenosine (entry 4, Table 4). When the same process was
carried out on β-L-CBz, PSL-C showed selective acylation of
the 50-hydroxyl group. However, the reaction stalled after 75%
conversion (entry 6, Table 4). Prolonged reaction times did not
increase the conversion, but the selectivity was lower. In the
case of CAL-B, the 50-O-Lev derivative was obtained as the
major compound in addition to diacylated compounds (entry
5, Table 4).
tetrafluoroborate [(BMIM)BF4], an ionic liquid, was tried
for acylation of 14c. Surprisingly, the solubility of 14c in
(BMIM)BF4 was insufficient to drive the reaction to completion.
However, the acylation of β-L-dGIbu under very diluted con-
ditions (0.025 M) and large excess of acylating agent
(9 equiv) led to 100% conversion but furnished mixtures of
acylated compounds.
Unsuccessful results were also observed in the hydrolysis
reaction of corresponding 30,50-di-O-levulinyl D and L deri-
vatives (Scheme 3). Hydrolysis reaction catalyzed by PSL-C
on 30,50-di-O-Lev-β-D-dGIbu (20) afforded 50-O-Lev-β-D-
dGIbu (18), whereas CAL-B furnished the 30-O-Lev com-
pound 19.22 CAL-B exhibited the same selectivity with the
L-isomer, and selective 50-hydrolysis was accomplished by
the reaction of 30,50-di-O-Lev-β-L-dGIbu (21) with phosphate
buffer. However, PSL-C exhibited low selectivity, furnishing
a mixture of acylated derivatives. Although several strategies
were tried, we could not find an adequate enzyme for the
desired separation of β-D/L-dGIbu mixtures.
On the basis of our experience, β-D/L-uridine was subjected
to enzymatic resolution. The lack of selectivity observed with
PSL-C on β-D-ribonucleosides makes the separation of a
mixture of β-D/L-isomers difficult. Among the approaches
tested, the route specified in Scheme 6 provided the most
convenient strategy. Treatment ofβ-D/L-uridine (23a/25a) with
acetonoxime levulinate in the presence of PSL-C at 30 °C in
THF furnished multiple inseparable monoacylated products
We also became interested in the resolution of D/L-mix-
tures of ribonucleosides due to their applications in the
synthesis of therapeutic aptamers.23 Therefore, we decided
to investigate the enzymatic acylation of β-D- and β-L-ribo-
nucleosides with CAL-B and PSL-C. Enzymatic acylation of
(21) van Rantwijk, F.; Sheldon, R. A. Chem. Rev. 2007, 107, 2757–2785.
ꢀ
(22) Garcıa, J.; Fernandez, S.; Ferrero, M.; Sanghvi, Y. S.; Gotor, V.
J. Org. Chem. 2002, 67, 4513–4519.
1
with the same Rf. H NMR analysis showed the presence of
20-, 30-, and 50-O-levulinyl derivatives. We attributed the 50-O-
levulinylation of the L-isomer to the expected selectivity
(23) Eulberg, D.; Jarosch, F.; Vonhoff, S.; Klussmann, S. Spiegelmers for
Therapeutic Applications. In The Aptamer Handbook; Klussmann, S., Ed.;
Wiley-VCH: Weinheim, 2006; Chapter 18, pp 417-439.
6608 J. Org. Chem. Vol. 75, No. 19, 2010

Martınez-Montero et al.
JOCArticle
TABLE 3. Enzymatic Acylation of β-D-Ribonucleosidesa
entry
substrate
enzyme
conc (M)
T (°C)
t (h)
23 (%)b
24 (%)b,c
1
2
3
4
5
23a
23b
23c
23c
23a
CAL-B
CAL-B
CAL-Bd
CAL-Bd,e
PSL-C
0.1
30
45
50
55
30
2
2
55
24
13
97 (80)
>97 (83)
70 (56)
>97 (93)
mixture of acylated
compounds
0.05
0.05
0.025
0.1
30
6
7
23b
23b
PSL-C
PSL-C
0.05
0.025
45
45
24
27
>97
78
mixture of acylated
compounds
mixture of acylated
compounds
8
23c
PSL-Cd
0.025
55
24
aRatio of 23:CAL-B, 1:1 (w/w); ratio of 23:PSL-C, 1:3 (w/w); 3 equiv of acetonoxime levulinate. bPercentage of compounds calculated by 1H NMR.
cIsolated yields are given in parentheses. d9 equiv of acetonoxime levulinate. eRatio of 23:CAL-B, 1:2 (w/w).
SCHEME 5
SCHEME 6
TABLE 4. Enzymatic Acylation of β-L-Ribonucleosidesa
entry substrate enzyme conc (M) T (°C) t (h) 25 (%)b 26 (%)b,c
1
2
3
4
5
6
25a
25a
25b
25b
25c
25c
CAL-B
PSL-C
CAL-B
PSL-C
0.1
0.1
30
30 10.5
45
45 26
55 41
55 46
5.5
97 (81)
>97 (87)
>97 (80)
>97(88)
63 (58)
0.05
0.025
3
CAL-Bd,e 0.025
18
25
PSL-Cd,f
0.025
75 (70)
aRatio of 25:CAL-B, 1:1 (w/w); ratio of 25:PSL-C, 1:3 (w/w); 3 equiv
of acetonoxime levulinate. bPercentage of compounds calculated by 1H
NMR. cIsolated yields are given in parentheses. d9 equiv of acetonoxime
levulinate. eRatio of 25:CAL-B, 1:2 (w/w). fRatio of 25:PSL-C, 1:6 (w/w).
observed with PSL-C, whereas the secondary hydroxyl groups
of the D-isomer were acylated in a nonselective manner with
PSL-C. In order to accomplish the separation of these pro-
ducts, we elected to carry out a second acylation step catalyzed
by CAL-B. Since this lipase is regioselective toward the
50-hydroxyl group of the D-nucleosides, 20- and 30-O-acyl
derivatives were transformed into 20,50- and 30,50-di-O-levuli-
nyl nucleosides, while the 50-O-levulinyl compound remained
unreacted. The diacylated products and 50-O-Lev-β-L-U are
easily separated by column chromatography. Subsequently,
the diacyl derivatives were treated with aqueous ammonium
in MeOH to give the corresponding nucleoside β-D-U (23a),
which was obtained with a satisfactory 92% ee, determined
by transformation into 50-O-Lev-β-D-U. The chiral HPLC
analysis of 50-O-Lev-β-L-U revealed 83% ee during acylation
with PSL-C.
Thus, we obtained the X-ray structure of PSL24 from the
Protein Data Bank.25 The X-ray crystal structure of the open
form of PSL (3LIP) shows an active site with three subsites
that bind to the substrate (Figure 1A).24 Viewing the catalytic
triad Asp264-His286-Ser87 from left to right, the three
subsites are (a) the large hydrophobic pocket, where the acyl
chain binds; (b) the medium-sized pocket, where the nucleo-
phile is placed; and (c) the alternate hydrophobic pocket to
the right of the medium pocket, which can also bind parts
of the nucleophile. This alternate hydrophobic pocket lies
below the catalytic triad in a narrow region.26 Most other
lipases, including CAL-B, lack this alternate hydrophobic
pocket.27 During the PSL-catalyzed acylation of N4-ben-
zoyl-20-deoxycytidine, the nucleoside acts as the nucleophile
and thus binds in the medium-sized pocket. Since N4-ben-
zoyl-20-deoxycytidine is much larger than this pocket, it
might extend into other pockets or into the solvent.
Molecular Modeling. The regio- and stereoselectivity ob-
tained during enzymatic resolution of racemic mixtures of 20-
deoxy-β-D/L-nucleosides inspired us to undertake a modeling
study of these processes to explain the observed results.
(25) http://www.rcsb.org/pdb/explore/explore.do?structureId=3LIP.
ꢀ
(26) Lavandera, I.; Fernandez, S.; Magdalena, J.; Ferrero, M.; Grewal, H.;
Savile, C. K.; Kazlauskas, R. J.; Gotor, V. ChemBioChem 2006, 7, 693–698.
(24) Schrag, J. D.; Li, Y.; Cygler, M.; Lang, D.; Burgdorf, T.; Hecht,
H. J.; Schmid, R.; Schomburg, D.; Rydel, T. J.; Oliver, J. D.; Strickland,
L. C.; Dunaway, C. M.; Larson, S. B.; Day, J.; McPherson, A. Structure
1997, 5, 187–202.
ꢀ
(27) Lavandera, I.; Fernandez, S.; Magdalena, J.; Ferrero, M.; Kazlauskas,
R. J.; Gotor, V. ChemBioChem 2005, 6, 1381–1390.
J. Org. Chem. Vol. 75, No. 19, 2010 6609

Martınez-Montero et al.
JOCArticle
SCHEME 7. Tetrahedral Intermediate for 30- or 50-Levulination
of β-D- or β-L-Nucleosidesa
aThe five catalytically essential hydrogen bonds are as follows: one from
the carboxylate of Asp264 to Nδ of His286 (bond d1), two from Nε of
His286 to the oxygens of Ser87 (bond d2) and the oxygen of the
corresponding nucleoside (bond d3), and two from the oxy anion to
the main chain amides of Gln88 (bond d4) and Leu17 (bond d5). For
computer modeling the different groups are introduced in a stepwise
fashion as described in the text, forming both 30- and 50-regioisomers of
each β-D- or β-L-enantiomer.
of the enzyme, but might omit subtle details about the
transition state.
We started with a simplified intermediate that mimicked
the tetrahedral intermediate for levulination of ethanol.
Geometry optimization yielded a structure containing
all five catalytically essential hydrogen bonds (d1 to d5 in
Scheme 7). To model the nucleoside substrates, we replaced
the ethoxy moiety with the 20-deoxyribosyl group linked to
the tetrahedral intermediate at either the 50- or 30-position
with each of the nucleoside enantiomers (D or L) and then
added the cytosine ring followed by the benzoyl protecting
group. Since the structures contained only two (30-acylation)
or three (50-acylation) rotatable bonds, a systematic search
identified the catalytically productive conformations (see
Supporting Information). The shape of the nucleophile-
binding region below the catalytic triad (medium pocket
and alternate hydrophobic pocket) restricted the nucleoside
orientations, especially the cytosine-ring orientation, so that
only a few conformations (a) contained all five catalytically
essential hydrogen bonds (d1-d5)28 shown in Scheme 7, (b)
avoided steric hindrances between the nucleoside inter-
mediate and the lipase, and (c) avoided internal steric clashes
within the nucleoside intermediate. Acylation at the 50-posi-
tion allowed rotation of the three bonds in the nucleoside
to adjust its position, while acylation at the 30-position
allowed only two bonds to adjust. For this reason, modeling
the 50-levulination yielded a larger number of productive
conformations than the 30-levulination.
FIGURE 1. Best models of intermediates for PSL-catalyzed levu-
lination of 20-deoxy-β-D-cytidine. Residues surrounding the large
hydrophobic pocket (Val266, Val267, Leu167, Phe119, and Pro113)
are colored green, and residues surrounding the alternate hydro-
phobic pocket (Ile290, Leu287, Thr18, and Tyr29) are colored dark
blue. The oxyanion hole residues Leu17 and Gln88 are colored red
and yellow, respectively. (A) Conformation of the intermediate of
levulination corresponding to the 30-hydroxyl with the β-D-enantiomer.
(B) When placed in this alternate pocket, three potential hydrogen-
bond stabilizations appear (see text). (C) Conformation that mimics
levulination of the 50-hydroxyl with the β-D-enantiomer. To allow a
better view of the active site, panel A displays Ala247 and Leu248 in a
line representation, and panel C displays Ala247, Leu248, Gln292,
Leu293, and Leu294 in a line representation.
When geometry optimization was carried out at 30-hydroxyl
of N4-benzoyl-20-deoxy-β-D-cytidine (Figure 1A), the tetrahe-
dral intermediate for the levulination placed the cytosine base
and its benzoyl protecting group in the alternate hydrophobic
pocket and the 20-deoxyribose moiety into the medium-sized
pocket. Both moieties avoid steric hindrances in their respec-
tive pockets. Importantly, in addition to the five catalytically
To qualitatively explain the unusual regioselectivity of
PSL-catalyzed acylations on each enantiomer of the racemic
mixtures of thymidine,19a N4-benzoyl-20-deoxycytidine, and
N6-benzoyl-20-deoxyadenosine, we focused on N4-benzoyl-
20-deoxycytidine as a representative example of this study.
Thus, we modeled the key intermediates for levulination
of N4-benzoylated 20-deoxycytidine in both the 30- and
50-positions of each enantiomer, namely, β-D and β-L. This
approach focused on how the substrate fits in the active site
(28) Hydrogen bonds must have a donor atom to acceptor atom distance
˚
of less than 3.30 A and a donor atom-hydrogen-acceptor atom angle of
120° or greater.
6610 J. Org. Chem. Vol. 75, No. 19, 2010

Martınez-Montero et al.
JOCArticle
key hydrogen bonds (d1 = 3.04; d2 =3.03; d3=2.84; d4=3.17;
˚
d5=3.23 A), this intermediate formed three extra hydrogen
bonds: one between the C2 carbonyl group of cytosine and the
5 -OH of the sugar (bond a, 2.95 A, 171°, Figure 1B),28 another
0
˚
˚
between the later and Gly16 (bond b, 2.84 A, 146°), and the
˚
third additional hydrogen bond (bond c, 2.71 A, 171°) between
the oxygen carbonyl group of the levulinyl substituent and a
water molecule placed inside the large hydrophobic pocket.
We propose that these stabilizing hydrogen bonds cause PSL
to favor acylation at the 30-hydroxyl group in the β-D-enantio-
mer of N4-benzoyl-20-deoxycytidine. However, when the tetra-
hedral intermediate for the levulination of 50-hydroxyl of
20-deoxy-β-D-cytidine was built and minimized, no single
productive conformation was obtained. In all structures some
of the key H-bond interactions were lost. In the best fit
(Figure 1C; d1 =2.83; d2 =3.03; d3 =2.71; d4 =3.62; d5 =
˚
3.84 A), the enzyme situates the cytosine ring in the alternate
hydrophobic pocket but in such a way that the protecting
benzoyl group cannot fit inside the active site of the enzyme. In
another option the benzene ring is placed outside the active
˚
side, forming an improbable bond of 9.06 A between the
carbonylic carbon and the aromatic ring. The binding of the
base and position of the protecting group appear to be the key
factors in supporting the enzyme regioselectivity.
The geometry optimization of both regioisomers of
N4-benzoyl-20-deoxy-β-L-cytidine, which come from levulina-
tion of the 30- or 50-hydroxyl group, yielded structures that are
shown in Figure 2. The best model of the tetrahedral inter-
mediate for the 30-acylation (Figure 2A) shows the sugar in the
medium-sized pocket and the cytosine and its protecting group
pointing out to the solvent. Three of the five key hydrogen
bonds were present while avoiding inter- and intramolecular
steric interactions (d1 = 3.10; d2 = 4.07; d3 = 3.15; d4 = 3.57;
˚
d5 = 3.10 A). On the contrary, in the tetrahedral intermediate
for the levulination of 50-hydroxyl of N4-benzoyl-20-deoxy-β-L-
cytidine (Figure 2B), the sugar was placed in the medium-sized
pocket and the cytosine base, including its benzoyl protecting
group, was inside the alternate hydrophobic pocket.
It is noteworthy that in addition to the five catalytically
key H-bonds (d1 = 2.86; d2 = 2.86; d3 = 2.82; d4 = 3.30;
˚
d5 = 3.15 A), this intermediate formed four hydrogen bonds:
FIGURE 2. Best models of intermediates for PSL-catalyzed levu-
lination of N4-benzoyl-20-deoxy-β-L-cytidine. The color code is the
same as that used in Figure 1. (A) Conformation of the intermediate
of levulination corresponding to the 30-hydroxyl with the β-L-
enantiomer. (B) Conformation that mimics the levulination of the
50-hydroxyl with the β-L-enantiomer. (C) When placed in this alter-
nate pocket, the intermediate is stabilized since four extra hydrogen
bonds are formed (see text). To allow a better view of the active site,
panels A and B display Ala247 and Leu248 in a line representation.
between the 30-hydroxyl group of the sugar and the 50-O that
is directly bound to the quaternary carbon of the tetrahedral
intermediate (bond a, 3.29 A, 124°, Figure 1C), from the
water molecule placed inside the alternate hydrophobic
pocket to the N-3 (bond b, 3.11 A, 140°) and N-4 (bond c,
3.17 A, 134°) of the cytosine base, respectively, and from that
water molecule to the backbone hydroxyl group of Thr18
˚
˚
˚
˚
(bond d, 3.10 A, 155°). As in the previous case for levulina-
tion of N4-benzoyl-20-deoxy-β-D-cytidine, the binding base
and protecting group rings in 20-deoxy-β-L-cytidine appear
to be the key factors supporting the enzyme regioselectivity.
The overall result shows that when a racemic mixture of
20-deoxy-β-D/L-cytidine reacts with PSL in the presence
of acetonoxime levulinate, only the 30-hydroxyl group of
20-deoxy-β-D-cytidine and the 50-hydroxyl group of 20-deoxy-
β-L-cytidine are selectively acylated. Thus, molecular modeling
corroborates the experimental results.
L-nucleoside racemic mixtures via an acylation reaction. This
lipase showed different selectivity with both isomers and
furnished a mixture of easily separable compounds. Pseudo-
monas cepacia lipase also exhibited total selectivity toward
the acylation of the 50-hydroxyl group of β-L-uridine,
whereas a mixture of acylated derivatives was obtained
during bioacylation of β-D-uridine. Thus, a double sequen-
tial parallel kinetic resolution catalyzed by Pseudomonas
cepacia lipase and Candida antarctica lipase B has been
developed to separate the racemic mixture β-D/L-uridine.
The methodology described here has proven to be very useful
for the isolation of β-L-nucleosides, which are increasingly
used especially as antiviral and anticancer therapeutic agents.
In summary, we have described a parallel kinetic resolu-
tion approach to obtain chiral nonracemic β-D- and β-L-
nucleosides. Pseudomonas cepacia lipase was found to be
a suitable biocatalyst for the resolution of 20-deoxy-β-D/
J. Org. Chem. Vol. 75, No. 19, 2010 6611

Martınez-Montero et al.
JOCArticle
Results obtained from a molecular modeling study on the
PSL-C-catalyzed levulination of mixtures of 20-deoxy-β-D/
L-nucleosides are in agreement with the observed experi-
mental data. Thus, PSL-C showed that only the 30-hydroxyl
group of 20-deoxy-β-D-cytidine and 50-hydroxyl group of
20-deoxy-β-L-cytidine selectively are acylated into the corre-
sponding levulinate derivatives, allowing convenient and
practical separation of the original racemic mixture. This
study further reinforces the utilization of biocatalysis as an
important tool for organic chemists working on the manu-
facturing of nucleoside-based therapeutic products.29
of 22 are in full accordance with those described for its
enantiomer.22
Enzymatic Resolution of an Artificial 1:1 Mixture of β-D/L-
Uridine (23a and 25a). A suspension of a mixture of β-D/L-
uridine (60 mg, 0.25 mmol), acetonoxime levulinate (131 mg,
0.76 mmol), and PSL-C (180 mg) in anhydrous THF (2.5 mL)
under N2 atmosphere was stirred for 13 h at 250 rpm and 30 °C.
Once the reaction was finished, the enzyme was filtered off and
washed with CH2Cl2 and MeOH, and the solvents were distilled
under vacuum. Flash chromatography of the residue (2-4%
MeOH/CH2Cl2) afforded 20-, 30-, and 50-levulinyl derivatives
(same Rf). Subsequently, an Erlenmeyer flask containing this
fraction, acetonoxime levulinate (113 mg, 0.66 mmol), CAL-B
(75 mg), and anhydrous THF (2.2 mL) was stirred at 250 rpm
and 30 °C for 17 h. The enzyme was filtered off, washed and
solvents, and evaporated. The crude reaction was subjected
to flash chromatography (2-4% MeOH/CH2Cl2). The first
fraction contained diacyl derivatives, while 50-O-levulinyl-β-D-
uridine (26a) was obtained from the second fraction in 85%
yield and 83% ee. Hydrolysis of the diacyl derivatives in a 1:1
mixture of aqueous ammonia and MeOH (3.2 mL) led after 2 h
of reaction to β-D-uridine (23a) in 62% yield and 92% ee.
50-O-Levulinyl-β-D-uridine (24a) or 50-O-Levulinyl-β-L-uridine
(26a). White solid; Rf (10% MeOH/CH2Cl2) 0.32; mp 60-62 °C;
IR (KBr) ν 3172, 2920, 2850, 1650, 1460 cm-1; 24a [R]D20=þ14 (c 1,
Experimental Section
General Comments. Enzyme Activities. Candida antarctica
lipase B (CAL-B, Novozym 435, 7300 PLU/g) was a gift from
Novo Nordisk Co., and Pseudomonas cepacia lipase (PSL-C,
1387 U/g) was purchased from Amano Phamaceuticals. The ee
values of 16a and 17a were determined by HPLC using a
Chiralcel OD column (250 ꢀ 46 mm), 0.8 mL/min, 20 °C, and
hexane/isopropyl alcohol (1:1) as eluent. The ee values of 16b
and 17b were determined using a Chiralcel OJ-H column (250 ꢀ
46 mm), 0.75 mL/min, 35 °C, and hexane/isopropyl alcohol
(65:35) as eluent. The ee values of 24a and 26a were determined
by HPLC using a Chiralcel OD column (250 ꢀ 46 mm), 0.8 mL/min,
30 °C, and hexane/isopropyl alcohol (45:55) as eluent. Opti-
cal rotations were recorded on a polarimeter, and values are
1
MeOH); 26a [R]2D0 =-13 (c 1, MeOH); H NMR (DMSO-d6,
300.13 MHz) δ 2.11 (s, 3H, CH3-Lev), 2.50 (t, 2H, CH2-Lev,
3JHH 6.2 Hz), 2.70 (t, 2H, CH2-Lev, 3JHH3 6.0 Hz), 3.91-4.08 (m,
3
0
0
0
0
3H, H2 þH3 þH4 ), 4.14 (dd, 1H, H5 , JHH 4.9 Hz, JHH 11.9
reported as follows: [R]T (c: g/100 mL, solvent). Derivatives
Hz), 4.26 (dd, 1H, H5 , JHH 2.8 Hz, 3JHH 11.9 Hz), 5.28 (d, 1H,
3
λ
0
16a,19b 16b,19b 17a,19a 17b,19a 18,22 19,22 and 2022 have been
previously described.
OH, 3JHH 4.5 Hz), 5.47 (d, 1H, OH, 3JHH 5.3 Hz), 5.67 (d, 1H,
H5, 3JHH 8.1 Hz), 5.74 (d, 1H, H1 , JHH 4.9 Hz), 7.65 (d, 1H, H6,
3
0
General Procedure for the Enzymatic Acylation of Nucleo-
sides. In a standard procedure, THF was added to an Erlen-
meyer flask that contained nucleoside derivative (0.2 mmol),
acetonoxime levulinate, and lipase under nitrogen (ratio of
acylating agent, ratio of enzyme, concentration, temperature,
and time are indicated in Tables 1, 3, and 4). The reactions were
stirred at 250 rpm and were monitored by TLC (10% MeOH/
CH2Cl2). The enzyme was filtered off and washed with CH2Cl2
and MeOH, and the solvents were distilled under vacuum. Flash
chromatography (gradient eluents 2-5% MeOH/CH2Cl2) af-
forded derivatives 16, 17, 24, and 26 (yields and ee are indicated
in Tables 1,3, and 4).
3JHH 8.1 Hz); 13C NMR (DMSO-d6, 75.5 MHz) δ 27.5 (CH2-
0
Lev), 29.5 (CH3-Lev), 37.4 (CH2-Lev), 63.7 (C5 ), 69.7, 72.8
0
0
0
0
(C2 þC3 ), 81.2 (C4 ), 82.7 (C1 ), 102.0 (C5), 140.7(C6), 150.6 (C2),
162.9 (C4), 172.2 (CdO), 206.9 (CdO); MS (APCIþ, m/z) 343
[(M þ H)þ, 90%], 366 [(M þ Na)þ, 35]; HMRS (ESIþ) calcd for
C14H19N2O8 [M þ H]þ 343.1141, found 343.1139. Anal. Calcd
for C14H18N2O8: C, 49.12; H, 5.30; N, 8.18. Found: C, 49.1; H,
5.4; N, 8.0.
N6-Benzoyl-50-O-levulinyl-β-D-adenosine (24b) or N6-Benzoyl-
50-O-levulinyl-β-L-adenosine (26b). White solid; Rf (10% MeOH/
CH2Cl2) 0.38; mp 73-75 °C; IR (KBr) ν 3060, 2930, 1737, 1711
cm-1; 24b [R]D20=-15 (c 2, MeOH); 26b [R]2D0=þ17 (c 2, MeOH);
1H NMR (DMSO-d6, 600.15 MHz) δ 2.09 (s, 3H, CH3-Lev),
Synthesis of N2-Isobutyryl-30,50-di-O-levulinyl-β-L-20-deoxy-
guanosine (21). To a stirred mixture of β-L-dGIbu (0.40 mmol)
and Et3N (0.16 mL, 1.0 mmol) in 1,4-dioxane (2.7 mL) under
nitrogen were added levulinic acid (0.21 mL, 2.08 mmol), DCC
(428 mg, 2.08 mmol), and DMAP (3.9 mg, 0.03 mmol). The
reaction was stirred at rt during 2 h. The insoluble material was
collected by filtration, and the filtrate was evaporated under
vacuum. The residue was subjected to flash chromatography
with 2% MeOH/CH2Cl2 to afford 21 in 93% yield. The spectro-
scopic data of 21 are in full accordance with those described for
its enantiomer.22
2.49 (t, 2H, CH2-Lev, 3JH3H 6.0 Hz), 2.71 (t, 2H, CH2-Lev, 3JHH
3
6.4 Hz), 4.14 (c, 1H, H4 , JHH 4.6 Hz), 4.22 (dd, 1H, H5 , JHH
0
0
5.9 Hz, 3JHH 11.9 Hz), 4.28 (c, 1H, H3 , JHH 4.9 Hz), 4.36 (dd,
3
0
1H, H5 , JHH 3.6 Hz, 3JHH 11.9 Hz), 4.75 (c, 1H, H2 , JHH 5.2
3
3
0
0
Hz), 5.44 (d, 1H, OH, 3JHH 5.4 Hz), 5.67 (d, 1H, OH, 3JHH 5.8
Hz), 6.14 (d, 1H, H1 , JHH 5.2 Hz), 7.56 (t, 2H, Hm, 3JHH 7.7 Hz),
3
0
7.66 (t, 1H, Hp, 3JHH 7.4 Hz), 8.05 (d, 2H, Ho, 3JHH 7.5 Hz), 8.57
(s, 1H, H8), 8.71 (s, 1H, H2); 13C NMR (DMSO-d6, 90.14 MHz)
28.0 (CH2-Lev), 30.0 (CH3-Lev), 37.8 (CH2-Lev), 64.4 (C50),
0
0
0
0
70.7, 73.4 (C2 þC3 ), 82.3 (C4 ), 88.3 (C1 ), 125.3 (C5), 126.4 (Co),
128.9 (Cm), 132.9 (CP), 133.8 (Ci), 143.7 (C8), 150.9-152.7
(C2þC6þC4), 166.1 (Ph-CdO), 172.7 (CdO), 207.3 (CdO);
MS (APCIþ, m/z) 469 [(M þ H)þ 75%]; HMRS (ESIþ) calcd for
C22H24N5O7 [M þ H]þ 470.1670, found 470.1677. Anal. Calcd
for C22H23N5O7: C, 56.29; H, 4.94; N, 14.92. Found: C, 56.2; H,
5.0; N, 14.8.
Enzymatic Hydrolysis of N2-Isobutyryl-30,50-di-O-levulinyl-β-
L-20-deoxyguanosine (21). Synthesis of 22. To a solution of 21
(0.1 mmol) in 1,4-dioxane (0.35 mL) was added 0.15 M phos-
phate buffer pH 7 (1.65 mL) and the corresponding lipase [ratio
of 21:CAL-B was 1:1 (w/w), ratio of 21:PSL-C was 1:3 (w/w)].
The mixture was allowed to react at 250 rpm and at 40 °C for the
time indicated in Scheme 3. The enzyme was filtered off and
washed with CH2Cl2 and MeOH. The solvents were distilled
under vacuum, and the residue was subjected to flash chromato-
graphy with 2% MeOH/CH2Cl2 to afford 22 (acylation
reaction with CAL-B) in 80% yield. The spectroscopic data
N4-Benzoyl-50-O-levulinyl-β-D-cytidine (24c) or N4-Benzoyl-
50-O-levulinyl-β-L-cytidine (26c). White solid; Rf (10% MeOH/
CH2Cl2) 0.47; mp 193-195 °C; IR (KBr) ν 3493, 3323, 3162,
1728, 1703, 1651 cm-1; 24c [R]2D0 = þ35 (c 1, DMSO); 26c
[R]2D0=-36 (c 1, DMSO); 1H NMR (DMSO-d6, 300.13 MHz) δ
2.12 (s, 3H, CH3-Lev), 2.54 (t, 2H, CH2-Lev, 3JHH 6.3 Hz), 2.75 (t,
3
2H, CH2-Lev, JHH 6.2.3 Hz), 3.97 (m, 1H, H4 ), 4.09 (m, 2H,
0
(29) Pollard, D.; Woodley, J. M. Trends Biotechnol. 2007, 25, 66–73.
6612 J. Org. Chem. Vol. 75, No. 19, 2010

Martınez-Montero et al.
JOCArticle
H2 þH3 ), 4.25 (dd, 1H, H5 , JHH 5.1 Hz, 3JHH 12.2 Hz), 4.35 (dd,
Molecular Modeling. All minimizations were performed with
an iMac by using the molecular modeling package Molecular
Operating Environment (MOE) 2009.10 (Chemical Computing
Group, Inc.).30 The Amber99 force field31 and the correspond-
ing dictionary charges were used as implemented in MOE. Due
to the lack of parameters for THF compatible with the force
field used, we opted for a continuum solvation method, an
approach that has already afforded good results in similar
studies.32 In particular, a distance-dependent relative dielectric
constant of 2 was selected as a good compromise between the
dielectric constant of THF33 and that of a globular protein
matrix.34 The convergence criterion of the energy minimization
was set to a rms gradient value of 0.000 01 kcal mol-1. High-
quality pictures of representative structures in Figures 1 and 2
were created by using PyMOL 0.99.35
3
0
0
0
1H, H5 , JHH 2.3 Hz, 3JHH 12.1 Hz), 5.82 (d, 1H, H1 , JHH 2.7
Hz), 7.38 (d, 1H, H5, 3JHH 7.5 Hz), 7.51 (t, 2H, Hm, 3JHH 7.5 Hz),
7.63 (t, 1H, Hp, 3JHH 7.3 Hz), 8.01 (d, 2H, Ho, 3JHH 7.4 Hz), 8.16
(d, 1H, H6, 3JHH 7.5 Hz); 13C NMR (DMSO-d6, 75.5 MHz) 28.0
3
3
0
0
0
(CH2-Lev), 30.0 (CH3-Lev), 37.9 (CH2-Lev), 63.9(C5 ), 69.7 (C4 ),
0
0
0
0
74.3 (C2 ), 81.3 (C3 ), 91.4 (C1 ), 96.9 (C5), 128.9 (CmþCo), 133.2,
133.6 (CpþCi), 145.5 (C6), 154.9 (C2), 163.5 (C4), 168.0
(Ph-CdO), 172.7 (CdO), 207.4 (CdO); MS (ESIþ, m/z) 468.1
[(M þ Na)þ, 100%] and 446.1 [(M þ H]þ, 35]; HMRS (ESIþ)
calcd for C21H24N3O8 [M þ H]þ 446.1563, found 446.1564. Anal.
Calcd for C21H23N3O8: C, 56.50; H, 5.42; N, 9.41. Found: C, 56.3;
H, 5.2; N, 9.3.
(30) Molecular Operating Environment (MOE) 2009.10; Chemical Comput-
ꢀ
ꢀ
ing Group, Inc.: Montreal, Quebec, Canada; http://www.chemcomp.com.
(31) Wang, J.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000, 21,
1049–1074.
Acknowledgment. Financial support of this work by the
ꢀ
ꢀ
(32) (a) Garcıa-Urdiales, E.; Rıos-Lombardıa, N.; Mangas-Sanchez, J.;
Gotor-Fernandez, V.; Gotor, V. ChemBioChem 2009, 10, 1830–1838. (b)
Spanish Ministerio de Educacion y Ciencia (MEC) (Project
MEC-CTQ-2007-61126) isgratefully acknowledged. S.M.-M.
thanks MEC for a predoctoral FPU fellowship.
ꢀ
ꢀ
Leonard, V.; Fransson, L.; Lamare, S.; Hult, K.; Graber, M. ChemBioChem
2007, 8, 662–667. (c) Cammenberg, M.; Hult, K.; Park, S. ChemBioChem
2006, 7, 1745–1749. (d) Raza, S.; Fransson, L.; Hult, K. Protein Sci. 2001, 10,
329–338.
(33) The dielectric constant of THF at 20° C is 7.58. In Speight, J. G.
Perry’s Standard Tables and Formulas for Chemical Engineers; McGraw-Hill:
New York, 2003.
(34) Laberge, M. Biochim. Biophys. Acta 1998, 1386, 305–330.
(35) DeLano, W. L. The PyMOL Molecular Graphics System; DeLano
Scientific LLC: San Carlos, CA; http://www.pymol.org.
Supporting Information Available: H, 13C, and DEPT NMR
1
spectral data and 2D NMR experiments, in addition to
HPLC analysis. Molecular modeling information. This mate-
rial is available free of charge via the Internet at http://pubs.
acs.org.
J. Org. Chem. Vol. 75, No. 19, 2010 6613