Novel enzymatic synthesis of levulinyl protected nucleosides useful for solution phase synthesis of oligonucleotides

Article abstract of DOI:10.1016/j.tetasy.2003.07.009

An efficient synthesis of 3′- and 5′-O-levulinyl-2′- deoxy- and 2′-O-alkylribonucleosides has been developed from appropriate nucleosides by enzyme-catalyzed regioselective acylation in organic solvents. Several lipases were screened in combination with a

Full text of DOI:10.1016/j.tetasy.2003.07.009

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 14 (2003) 3533–3540
Novel enzymatic synthesis of levulinyl protected nucleosides
useful for solution phase synthesis of oligonucleotides
Javier Garc´ıa,a Susana Ferna´ndez,a Miguel Ferrero,a Yogesh S. Sanghvib,* and Vicente Gotora,*
aDepartamento de Qu´ımica Orga´nica e Inorga´nica, Facultad de Qu´ımica, Universidad de Oviedo, 33071 Oviedo, Spain
bDevelopment Chemistry, Isis Pharmaceuticals, 2292 Faraday Avenue, Carlsbad, CA 92008, USA
Received 2 June 2003; accepted 9 July 2003
Abstract—An efficient synthesis of 3%- and 5%-O-levulinyl-2%-deoxy- and 2%-O-alkylribonucleosides has been developed from
appropriate nucleosides by enzyme-catalyzed regioselective acylation in organic solvents. Several lipases were screened in
combination with acetonoxime levulinate as an acylating agent. Immobilized Pseudomonas cepacia lipase (PSL-C) was selected for
acylation of the 3%-hydroxyl group in nucleosides, furnishing 3%-O-levulinylated products 1 and 3 in excellent yields. Similarly,
Candida antarctica lipase B (CAL-B) provided 5%-O-levulinyl nucleosides 2 and 4 in high yields. Base-protected cytidine and
adenosine analogs were found to be good substrates for lipase-mediated acylations. To demonstrate the industrial utility of this
method, 3%-O-levulinyl thymidine and N2-Ibu-5%-O-levulinyl-2%deoxyguanosine were synthesized on a 25 g scale. Additionally,
PSL-C was reused to make the processes further economical.
© 2003 Elsevier Ltd. All rights reserved.
1. Introduction
been carried out through several tedious chemical pro-
tection/deprotection steps.6
Chemical synthesis of therapeutic oligonucleotides1 has
undergone a renaissance during the last decade due to
their potential use in the treatment of a wide range of
human diseases.2 Although classical solid-support
methodology for the synthesis of oligonucleotides is
well established, the solution-phase synthesis is perhaps
the next best alternative should any of these drugs
become commercially successful and are required in ton
quantity. Therefore, establishment of the solution-phase
synthesis of oligonucleotides is an important task.3 In
order to make large quantities of oligonucleotides via
solution-phase synthesis, one would require appropriate
nucleosidic building-blocks on an industrial scale.4 The
key building blocks for solution-phase oligonucleotide
synthesis are appropriately protected nucleosidic
monomers. Among the limited protecting groups avail-
able, the levulinyl group is frequently chosen to protect
the 3%- and/or 5%-hydroxyl of the nucleosides. This
group is stable during oligonucleotide coupling reac-
tions and can be selectively cleaved without affecting
other protecting groups in the same molecule.5 Until
recently, the preparation of these building blocks has
The potential of enzymes in organic synthesis is well
recognized, in particular when substrates possess sev-
eral functional groups of similar reactivity.7 The chiral
nature of enzymes allows selective acylation and/or
deacylation processes on a variety of molecules avoid-
ing additional chemical protection/deprotection steps.
In our ongoing research related to the synthesis of
selectively modified 3%-O-levulinyl and 5%-O-levulinyl
nucleosides from 2%-deoxy and 2%-O-alkyl nucleosides,
we have recently developed a general chemoenzymatic
synthesis of monolevulinyl protected nucleosides using
commercial lipases (Chart 1).8 The original approach
was a two-step method based on chemical acylation of
both hydroxyl groups followed by the regioselective
enzymatic hydrolysis of one of the levulinyl groups.
Candida antarctica lipase B (CAL-B) was found to
selectively hydrolyze the 5%-O-levulinate esters furnish-
ing 3%-O-levulinyl-2%-deoxynucleosides 1 in >80% iso-
lated yields. Immobilized Pseudomonas cepacia lipase
(PSL-C) and C. antarctica lipase A (CAL-A) exhibited
the opposite selectivity toward the hydrolysis at 3%-posi-
tion affording 5%-O-levulinyl derivatives 2 in >70% iso-
* Corresponding authors. Tel.: +34 98 510 3448; fax: +34 98 510 3448
(V.G.); tel.: +1 760 931 9200; fax: +1 760 931 9639 (Y.S.S.); e-mail:
ysanghvi@isisph.com; vgs@sauron.quimica.uniovi.es
lated yields.
A similar hydrolysis procedure was
successfully extended to the synthesis of 3%- and 5%-O-
0957-4166/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2003.07.009

3534
J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
levulinyl protected 2%-O-alkylribonucleosides 3 and 4.
Unfortunately, attempts to extend this methodology to
nucleosides with N-benzoylated bases using CAL-B or
unprotected bases using PSL-C were not successful.
30°C in THF indicated (TLC) that after 7 h the starting
material was completely consumed (entry 1, Table 1).
After work-up, 1H NMR data of the product confirmed
the formation of 3%-O-levulinylthymidine 1a as the sole
product due to regioselective acylation of the 3%-
hydroxyl group. Protected 1a was isolated in 85% yield
as a crystalline product without chromatography. Simi-
lar conditions were utilized for N-benzoylcytidine 6b to
obtain total regioselectivity at the 3%-position, however,
the reaction was much slower (34 h, entry 2, Table 1)
compared to thymidine. This could be attributed to the
limited solubility of the starting material 6b in THF.
Chart 1.
Scheme 2.
Herein we describe a new and concise approach for the
synthesis of levulinyl protected building blocks through
regioselective enzymatic acylation of base-protected
nucleosides overcoming the limitations of our original
method. The improved method is a one step enzymatic
protocol that utilizes the complementary behavior
shown by enzymes in acylation and hydrolysis
processes.
Similarly, PSL-C regioselectively acylated the 3%-
hydroxyl group of nucleoside 6c, furnishing N-benzoyl-
3%-O-levulinyl-2%-deoxyadenosine 1c in 89% isolated
yield (entry 3, Table 1). It is noteworthy that
nucleosides 6b and 6c bearing N-benzoyl protected
bases were completely and selectively acylated to form
N-benzoyl-3%-O-levulinyl-2%-deoxycytidine 1b and N-
benzoyl-3%-O-levulinyl-2%-deoxyadenosine 1c, respec-
tively. The ease of preparation of 1b and 1c indicates
the compatibility of the lipase toward acylation. This
attribute overcomes the previous limitations of our
method where we could not make 1b and 1c via enzy-
matic hydrolysis of 3%,5%-dilevulinyl nucleosides.8 The
acylation reaction catalyzed by PSL-C on N2-isobu-
tyryl-2%-deoxyguanosine 6d afforded a mixture of 3%-O-
levulinyl 1d, 5%-O-levulinyl 2d, and 3%,5%-di-O-levulinyl
derivatives, in addition to unreacted nucleoside 6d
(entry 4, Table 1). Increasing the reaction temperature
to 60°C did not improve the selectivity (entry 5, Table
1). Next, we evaluated the influence of co-solvents on
the catalytic activity of the enzyme using different
ratios of organic solvents. In this case, 10% of pyridine
or acetonitrile were employed as co-solvents, 1,4-diox-
ane, acetonitrile and DMF were used as solvents with-
out much success. Nevertheless, we are able to make 6d
via our original enzymatic hydrolysis protocol in good
yields should there be a need for this nucleoside.8
2. Results and discussion
We had previously reported that oxime esters are excel-
lent substrates towards regioselective enzymatic
acylations of various natural products, including
nucleosides.9 Therefore, acetonoxime levulinate 5 was
selected as an acylating agent for the present study.
Synthesis of 5 was accomplished in quantitative yield in
a single step starting from inexpensive industrial quality
levulinic acid and acetonoxime (Scheme 1).10
Scheme 1.
2.1. Acylation studies of 2%-deoxynucleosides
Selective 5%-O-acylation of thymidine was accomplished
with CAL-B at 30°C in a facile manner (entry 6, Table
1) with a trace amount of di-levulinyl product. In order
to increase the selectivity, the reaction temperature was
reduced to 10°C. The low temperature reaction did not
The overall enzymatic approach for the protection of
2%-deoxynucleosides is shown in Scheme 2 and the
experimental data is summarized in Table 1. Treatment
of 5 with thymidine 6a in the presence of PSL-C at

J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
Table 1. Regioselective enzymatic acylation of 2%-deoxynucleosides 6
3535
Entry
Substrate
Enzyme
T (°C)
t (h)
1 (%)a,b
2 (%)a,b
6 (%)a
diLevc (%)a
1
2
3
4
5
6
7
8
6a
6b
6c
6d
6d
6a
6a
6b
6c
6d
6a
6a
6b
6c
6a
6b
6c
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
CAL-Be
CAL-Be
CAL-Be
CAL-Be
CAL-Be
PSL-Cf
PSL-Cf,g
PSL-Cf
PSL-Cf
PSL-Cf,h
PSL-Cf,h
PSL-Cf,h
30
30
30
30
60
30
10
30
30
30
40
40
40
40
30
30
30
7
34
11
24
14
3
>97 (85)
98 (77)
96 (89)
44
35
5
3
6
2
4
23
12
92 (74)
91
11
22
22
31
3
8
6
18
11
16
4.5
8
92 (76)
94 (71)
>97 (79)
2
1
9
5
10
11
12
13
14
15
16
17
>97 (83)
>97 (80)
>97 (86)
94 (82)
>97 (92)
>97 (81)
>97 (90)
Traces
5
10
14
14
14
6
a Based on 1H NMR signal integration.
b Percentages of isolated yields are given in parenthesis.
c diLev=3%,5%-di-O-Levulinyl derivative.
d Ratio 6:PSL-C is 1:3 (w/w); 0.1 M concentration.
e Ratio 6:CAL-B is 1:1 (w/w); 0.1 M concentration.
f Ratio 6:PSL-C is 1:2 (w/w); 0.2 M concentration.
g Recycled PSL-C from entry 11.
h Scale-up to 5 g of 6.
completion. Similarly, nucleosides 6b and 6c were sub-
jected to enzyme catalysis at 40°C in 0.2 M concentra-
tion, allowing isolated preparation of 1b (86%) and 1c
(82%), respectively (entries 13 and 14, Table 1). Finally,
the optimized processes with PSL-C at 30°C and 0.2 M
concentration were scaled up to 5 g of nucleoside,
furnishing 1a (92%), 1b (81%), and 1c (90%), respec-
tively, in 14 h (entries 15, 16, and 17, Table 1). It is
noteworthy that 3%-O-Lev- and 5%-O-Lev-deoxynu-
cleosides 1 and 2 were isolated by precipitation from
the crude reaction mixture to avoid any expensive
chromatographic steps. Only in those exceptional cases
where the mixture of acylated byproducts were
obtained and precipitation attempts failed, flash chro-
matography was employed to isolate pure products.
reduce the formation of the byproduct (entry 7, Table
1). Good selectivity towards the 5%-hydroxyl group was
accomplished when nucleosides 6b–d reacted with
CAL-B at 30°C in THF, furnishing 5%-O-levulinyl
derivatives 2b–d in high yields (entries 8, 9, and 10,
Table 1). Although there were minor byproducts
formed during the acylation with CAL-B, we were able
to isolate 6b–d in high purity and acceptable yields
(71–79%) after chromatography. In order to obtain
total regioselectivity, Chirazyme L-2, c-f, C3, lyo (lipase
comp. B from C. antarctica on a different carrier than
Novozym SP435) were tried, without much avail.
Scale-up studies. All of the above PSL-C mediated
acylation experiments were carried out with a high
enzyme ratio (6:PSL-C, 1:3, w/w) to insure the comple-
tion of the reaction. For cost-efficient scale-up it was
necessary to reduce the enzyme ratio (1:2) and increase
the reaction concentration to 0.2 M, more appropriate
for industrial processes. Additionally, the reaction tem-
perature was increased to 40°C with the aim being
ability to reduce the reaction times and plant occu-
pancy. Treatment of 5 with 6a (0.2 M) in the presence
of PSL-C at 40°C provided 3%-O-levulinylnucleoside 1a
in 83% isolated yield with complete regioselectivity in
4.5 h (entry 11, Table 1). The industrial productivity of
the biocatalytic reactions could be further increased
when immobilized enzymes are utilized for chemical
transformations. Since PSL-C is an immobilized lipase,
we further explored the possibility of reusing the
enzyme. In entry 12 (Table 1), the data confirmed that
reusing PSL-C was possible and it worked as well as in
our previous experiment (entry 11, Table 1) except an
extended time was necessary to drive the reaction to
2.2. Acylation of 2%-O-alkyl nucleosides
We also became interested in the preparation of
levulinyl protected 2%-O-alkyl nucleosides 3 and 4
(Scheme 3) as building blocks for solution-phase syn-
thesis of second-generation oligonucleotides. The incor-
poration of 2%-O-alkyl residue in therapeutic
oligonucleotides provided enhanced stability, higher
binding affinity, and improved potency relative to their
2%-deoxy predecessors in vivo.11
The experimental data in Table 2 shows that PSL-C
was selective towards the acylation of the 3%-position of
2%-O-MOE nucleosides which bears T, 5-Me-CBz, and
ABz as bases (entries 1, 3, and 5, Table 2). However, no
reaction occurred when both 5-Me-C and A unpro-
tected bases were used (entries 2 and 4, Table 2). All the
reactions took place at 30°C, except for 2%-O-MOE-ABz

3536
J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
observed at 30 or 60°C even during extended period of
reaction time (entries 8 and 9, Table 2).
Scale-up studies. The utility of PSL-C for large-scale
reactions was demonstrated with nucleosides 7c and 7e
as pyrimidine and purine examples, respectively. Thus,
3%-O-Lev-2%-O-MOE-CBz 3c and 3%-O-Lev-2%-O-MOE-
ABz 3e were conveniently synthesized using similar
enzymatic acylation method (1:2 substrate: enzyme
ratio at 0.2 M) in 87 and 88% isolated yields, respec-
tively (entries 10 and 11, Table 2).
As observed for 2%-deoxynucleosides, CAL-B exhibited
opposite selectivity from PSL-C, catalyzing the
acylation at the 5%-position of 2%-O-alkylribonucleosides
7 with absolute regioselectivity (entries 12–18, Table 2).
The acylation process was complete at 40°C within
2.5–8 h to afford 5%-O-Lev-2%-O-alkylribonucleosides
4a–g. The increased efficiency of 5%-O-acylations on
2%-O-alkylnucleosides compared to the 2%-deoxynu-
cleoside with CAL-B may be attributed to the ability of
the enzyme to recognize the sugar conformation (North
versus South). From these results it appears that CAL-
B prefers acylation of sugar with North conformation
(e.g. 2%-O-alkylnucleosides) relative to sugar with South
conformation (e.g. 2%-deoxynucleoside).12
Scheme 3.
(7e), where 40°C was needed to reach conversion close
to 100%. Interestingly, when the reaction was carried
out at 30°C it stalled at 50% conversion and did not
proceed further even after longer reaction times. Similar
stalling behavior was observed with 2%-O-Me-A 7f.
Although PSL-C catalyzed the acylation of 7f, the
selectivity was moderate due to the absence of the
N-benzoyl group, a plausible enzyme recognition site
for facile reactions (entries 6 and 7, Table 2). In the
case of 2%-O-MOE-GIbu 7g, PSL-C showed total selec-
tivity toward the 3%-position, but low conversions were
Flash chromatography was used to isolate some of the
levulinyl derivatives (Table 2) wherever precipitation of
the crude was not practical due to solubility related
issues. When enzymatic reactions gave mainly one
acylated product, dry column vacuum chromatography
described by Pedersen and Rosenbohm13 was an excel-
lent alternative for large-scale isolation. This technique
Table 2. Regioselective enzymatic acylation of 2%-O-alkylribonucleosides 7
Entry
Substrate
Enzyme
T (°C)
t (h)
3 (%)a,b
80 (70)
>97 (82)
4 (%)a,b
7 (%)a
diLevc (%)a
1
2
3
4
5
6
7
8
7a
7b
7c
7d
7e
7f
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Cd
PSL-Ce,f
PSL-Cd,g
CAL-Bh
CAL-Bh
CAL-Bh
CAL-Bh
CAL-Bh
CAL-Bh
CAL-Bh
30
30
30
30
40
30
60
30
60
30
40
40
40
40
40
40
40
40
23
23
2.5
23
18
37
38
172
172
2.5
23
7
2.5
5.5
8
2.5
2.5
3.5
16
4
>99
>99
4
22
10
68
64
>95 (82)
64
57
32
36
6
11
8
22
7f
7g
7g
7c
7e
7a
7b
7c
7d
7e
7f
9
10
11
12
13
14
15
16
17
18
>97 (87)
96 (88)
4
>97 (70)
>97 (61)
>97 (86)
>97 (86)
>97 (84)
>97 (82)
>97 (88)
7g
a Based on 1H NMR signal integration.
b Percentage of isolated yields are given in parenthesis.
c diLev=3%,5%-di-O-levulinyl derivative.
d Ratio 7:PSL-C is 1:3 (w/w).
e Scale-up to 4 g of 7.
f Ratio 7:PSL-C is 1:2 (w/w).
g Scale-up to 3 g of 7.
h Ratio 7:CAL-B is 1:1 (w/w).

J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
3537
improved the yield of 5%-O-Lev-2%-O-MOE ca. 20% and
also reduced the amount of solvent and silica gel used,
making it cost-efficient for scale-up.
conditions, HPLC analysis showed that the reaction did
not proceed after 30 h (65% conversion) probably due
to the poor solubility of the starting material. In order
to increase the conversion rate and to try and maintain
the high concentration useful for the industrial scale a
higher temperature (40°C) was selected to increase the
solubility of the starting material. Increased tempera-
ture gave a small improvement in conversion rate (67%)
in 40 h. The best results were obtained (100% conver-
sion at 40°C) when a large excess of acylating agent 5
(1:3) was utilized. Next, the reaction was carried out
with 25 g of 6d. As indicated (entry 1, Table 4), CAL-B
catalyzed the acylation process with total selectivity
furnishing 2d in 82% isolated yield. Again with the
exception of longer reaction time, similar behavior was
observed in a successive run with recycled enzyme from
the previous reaction. Since direct precipitation was not
satisfactory, isolation of 5%-O-Lev-dGIbu 2d was accom-
plished by dry column vacuum chromatography from
the crude reaction mixture.
2.3. Large-scale acylation studies
The industrial utility of these synthetic processes was
proved via scale-up of the enzymatic acylation of
thymidine. The reaction was carried out with 25 g (0.1
mol) of thymidine with 1:2 ratio of acylating agent
(6a:5) and 1:2 (w/w) ratio of lipase (6a:PSL-C) (run 1,
Table 3). The reaction was completed in 12.5 h and
work-up of the reaction provided crude 3%-O-Lev-T 1a,
which was isolated in pure state (87%) via precipitation.
An extra 9% product was recovered after flash chro-
matography of the mother liquors from the precipita-
tion step. In the mother liquor, trace amounts of
dilevulinate derivative were found. Four consecutive
reactions were carried out with the recycled PSL-C
making the acylation process more economical (runs
2–4, Table 3). Importantly, PSL-C maintained total
selectivity toward acylation of the 3%-position with the
exception being the reaction time, which got longer as
the enzyme was recycled in subsequent runs.
Table 4. Large scale acylation of N2-isobutyryl-2%-
deoxyguanosine with CAL-Ba
Run
t (h)
2d (%)b
6d (%)b,c
Yield (%)d
Table 3. Large scale acylation of thymidine with PSL-Ca
1
24
216
87.5
77
12.5
23
82
86
2e
Run
t (h)
1a (%)b
6a (%)b
Yield (%)c
a The reaction was carried out at 40°C, 220 rpm, and under nitrogen
in 0.2 M concentration with 25 g of dGIbu in a ratio 1:3 of acylating
agent (dGIbu:oxime ester) and 1:1 (w/w) of lipase (dGIbu:CAL-B).
b Based on HPLC signal integration.
1
12.5
15.5
17
98.3
98.2
>99
1.7
1.8
87 (9)
90 (7)
88 (9)
88 (9)
2d
3d
4d
24.5
98.7
1.3
c 6d is $91% pure by HPLC (impurity: N-Ibu-guanine). In the
HPLC the impurity and 6d have very close retention times and
when the ratio between them are similar it is not possible to
integrate both signals separately, therefore the value is the sum of
both. In addition, the recycled enzyme contains the impurity from
previous run, consequently, processes were allowed to react until no
evolution was observed.
a The reaction was carried out at 30°C, 220 rpm, and under nitrogen
in 0.2 M concentration with 25 g of T in a ratio 1:2 of acylating
agent (T:oxime ester) and 1:2 (w/w) of lipase (T:PSL-C).
b Based on HPLC signal integration.
c Percentages of isolated yields by precipitation of 3%-O-Lev-T. Iso-
lated yields by flash chromatography of the ether mother liquors are
given in parenthesis.
d Percentages of isolated yields.
d Recycled enzyme from the previous run.
e Recycled enzyme from the previous run.
Next, we focused our attention to the levulinate oxime
ester 5 that was used in 1:2 excess with respect to the
starting material. In order to recapture the excess of 5
after completion of the reaction, the desired product 1a
was collected via precipitation and the filtrate subjected
to a dry column vacuum chromatography to furnish
unreacted 5.13 The enzymatic acylation of thymidine
with recaptured oxime ester 5 took place in a similar
manner as if it was fresh acylation agent, with the
exception of the extended reaction time (22 h versus
12.5 h). The fact that both the acylating agent and
lipase can be recycled after each acylation reaction
makes the new process atom efficient14 and very attrac-
tive for industrial production.
Next,
large-scale
acylation
of
N6-benzoyl-2%-
deoxyadenosine 6c was studied with recycled enzyme.
Treatment of 6c (5 g) with 3 equivalents of oxime ester
and CAL-B (entry 1, Table 5) furnished >90% (HPLC)
of 5%-O-Lev nucleoside 2c after 24 h, with a trace of
starting material and minor amounts of other acyl
derivatives. Standard work-up provided pure 2c in 87%
yield. Crystallization of 2c from MeOH provided ana-
lytically pure sample. Two consecutive runs were car-
ried out with the recycled enzyme from run 1 (entries 2
and 3, Table 5). A conversion of ꢀ80% was reached
after 112 h during the third experiment, where N-ben-
zoyl-5%-O-levulinyl-2%-deoxyadenosine was isolated in
61% yield after work-up.
To further demonstrate the scalability of enzymatic
acylation with CAL-B, N-isobutyryl-2%-deoxyguanosine
6d was chosen. The initial experiment was carried out
with 5 g of 6d at 0.2 M concentration instead of 0.1 M
(entry 10 of Table 1), in a ratio of 1:2 acylating agent
(6d:5) and 1:1 (w/w) of lipase (6d:CAL-B). Under these
3. Summary
3%- and 5%-O-Levulinyl protected derivatives of 2%-deoxy
and 2%-O-alkyl nucleosides have been prepared by
regioselective enzymatic acylation using a variety of

3538
J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
Table 5. Large scale acylation of 2%-deoxyadenosine with CAL-Ba
Run
t (h)
1c (%)b
2c (%)b
6c (%)b
DiLevc (%)b
Yield (%)d
1
24
64
112
Traces
Traces
Traces
>90
>89
>75
7
7
19
Traces
Traces
Traces
87 (78)
85 (79)
61 (53)
2e
3e
a The reaction was carried out at 30°C, 250 rpm, and under nitrogen in 0.2 M concentration with 5 g of dABz in a ratio 1:3 of acylating agent
(dABz:oxime ester) and 1:1 (w/w) of lipase (dABz:CAL-B).
b Based on HPLC and 1H NMR signal integration.
c diLev=3%,5%-di-O-Lev-dABz
.
d Percentages of isolated yields by precipitation of 5%-O-Lev-dABz. Isolated yields by recrystallization of the products are given in parenthesis.
e Recycled enzyme from the previous run.
lipases and acetonoxime levulinate as acylating agent.
N-Benzoyl-3%-O-levulinyl cytidine and adenosine
derivatives were obtained in good yield overcoming the
limitations of our original hydrolysis protocol. The new
and improved method is shorter than the one described
before, allowing greater atom efficiency14 and lowering
the cost of enzyme and reagents via recycling. To the
best of our knowledge this is the first example of
enzyme directed regioselective levulinylation reaction of
base protected nucleosides. The industrial suitability of
this process was proven via large-scale acylation of
several natural and unnatural nucleosides. Additionally,
we have demonstrated that the biosynthesis of pro-
tected nucleosides has the potential to allow synthesis
under environmentally friendly conditions while achiev-
ing high specificity, yield and generated significantly less
waste products than alternate chemical processes.15
4.1. Procedure for O-levulinyl acetoxime from levulinic
acid and acetonoxime
To a solution of acetonoxime (4.60 g, 63 mmol) in Et2O
(70 mL) under nitrogen was added levulinic acid (6.44
mL, 63 mmol) and dicyclohexylcarbodiimide (12.98 g,
63 mmol). The suspension was stirred at rt during 0.5 h.
After separation of the solid by filtration, the filtrate
was evaporated under vacuum to afford 10.77 g (quan-
titative yield) of acetonoxime levulinate10 as pale yellow
viscous oil, which solidified on cooling. Rf (EtOAc):
0.37; IR (NaCl): w 1756 and 1719 cm−1; 1H NMR
(CDCl3, 300 MHz): l 1.99 (s, 3H, Me), 2.01 (s, 3H,
3
Me), 2.19 (s, 3H, MeCO), 2.66 (t, 2H, CH2-Lev, JHH
3
6.6 Hz), and 2.82 (t, 2H, CH2-Lev, JHH 6.8 Hz); 13C
NMR (CDCl3, 75.5 MHz): l 16.9 (Me), 21.8 (Me), 26.5
(CH2-Lev), 29.8 (Me), 37.7 (CH2-Lev), 163.9 (C=N),
170.4 (C=O), and 206.3 (C=O). MS (ESI+, m/z): 172
[(M+H)+, 30%] and 194 [(M+Na)+, 100]. Anal. calcd
(%) for C8H13NO3: C, 56.13; H, 7.65; N, 8.18. Found:
C, 56.3; H, 7.4; N, 8.3.
4. Experimental16
Candida antarctica lipase B (CAL-B, Novozym 435,
10000 PLU/g), Chirazyme L-2, c-f, C3, lyo (lipase from
C. antarctica B, 400 U/g), and immobilized Pseu-
domonas cepacia lipase (PSL-C, 904 U/g) were pur-
chased from Novo Nordisk Co., Roche Diagnostics,
and Amano Pharmaceuticals, respectively. Solvents
were distilled over an adequate desiccant under nitro-
gen. High performance liquid chromatography analyses
(HPLC) were carried out at 254 nm using a Kromasil
100, C18, 5m column (150×46 mm) and two mobile
phase (A: MeCN; B: 1.54 g ammonium acetate in 1-L
water containing 1% MeCN). Gradient condition:
4.2. General procedure for the regioselective enzymatic
acylation of 2%-deoxynucleosides
A suspension of 6 (0.2 mmol), the oxime ester (0.6
mmol), and the lipase in dry THF (2 mL) under
nitrogen was stirred at 250 rpm for the time and at the
temperature indicated in Table 1. The reactions were
monitored by TLC (10% MeOH/CH2Cl2). The enzyme
was filtered off and washed with CH2Cl2 and THF. The
solvents were distilled under vacuum, and the residue
was taken up in NaHCO3 (aq) and extracted with
CH2Cl2. The combined organic layers were dried over
Na2SO4 and evaporated. The residue was precipitated
in diethyl ether to afford after filtration the
monolevulinyl nucleosides 1 or 2 as solids. No further
purification was necessary except in entries 8 and 9
t (min)
A
B
0.0
10.0
15.0
20.0
25.0
30.0
40.0
1
4
4
10
20
50
70
99
96
96
90
80
50
30
i
(Flash chromatography, gradient eluent 5–20% PrOH/
CH2Cl2) to separate the traces of other acyl derivatives.
4.2.1. N4-Benzoyl-3%-O-levulinyl-2%-deoxycytidine 1b. Rf
(10% MeOH/CH2Cl2): 0.30; mp: 189–191°C; [h]2D0=
+20.5 (c 1.0, MeOH); IR (KBr): w 3289, 2929, 1727, and
1
Compounds 1a, 2a–d, 3a, 4a, 4c, 4e, and 4g have been
1709 cm−1; H NMR (CDCl3, 300 MHz): l 2.20 (s, 3H,
previously obtained by enzymatic hydrolysis.8
Me-Lev), 2.44 (m, 1H, H2%), 2.59 (m, 2H, CH2-Lev),

J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
3539
2.73 (m, 3H, H2%+CH2-Lev), 3.97 (m, 2H, H5%), 4.20 (m,
1H, H4%), 5.37 (m, 1H, H3%), 6.26 (apparent t, 1H, H1%,
3JHH 6.8 Hz), 7.58 (m, 4H, H5+Hm+Hp), 7.90 (apparent
(C1%), 111.8 (C5), 128.1, 129.9 (Co+Cm), 132.5 (Cp),
136.8 (Ci), 139.2 (C6), 148.0 (C2), 159.5 (C4), 172.2
(C=O), 179.4 (PhC=O), and 206.5 (C=O); MS (ESI+,
m/z): 518 [(M+H)+, 20%], 540 [(M+Na)+, 100], and 556
[(M+K)+, 15]. Anal. calcd (%) for C25H31N3O9: C,
58.02; H, 6.04; N, 8.12. Found: C, 58.1; H, 6.2; N, 8.2.
3
3
d, 2H, Ho, JHH 7.3 Hz), and 8.30 (d, 1H, H6, JHH 7.4
Hz); 13C NMR (DMSO-d6, 75.5 MHz): l 27.8 (CH2-
Lev), 29.6 (Me-Lev), 37.5 (CH2-Lev), 38.2 (C2%), 61.2
(C5%), 74.9, 85.6 (C3%+C4%), 86.4 (C1%), 96.3 (C5), 128.5
(Co+Cm), 132.8 (Ci), 133.1 (Cp), 145.0 (C6), 154.5
(PhC=O), 163.3 (C4), 167.4 (C2), 172.1 (C=O), and
207.0 (C=O); MS (ESI+, m/z): 430 [(M+H)+, 30%], 452
[(M+Na)+, 52], and 468 [(M+K)+, 95]. Anal. calcd (%)
for C21H23N3O7: C, 58.74; H, 5.40; N, 9.79. Found: C,
58.7; H, 5.5; N, 9.8.
4.3.2. N6-Benzoyl-3%-O-levulinyl-2%-O-(2-methoxyethyl)-
adenosine 3e. Rf (10% MeOH/CH2Cl2): 0.62; mp: 44–
46°C; [h]2D0=−30.6 (c 0.4, MeOH); IR (KBr): w 3019,
2849, 1734, 1716, and 1698 cm−1; 1H NMR (CDCl3, 300
MHz): l 2.21 (s, 3H, Me-Lev), 2.71 (m, 2H, CH2-Lev),
2.80 (m, 2H, CH2-Lev), 3.04 (s, 3H, OMe), 3.27 (m, 2H,
CH2-MOE), 3.45 (m, 1H, CH-MOE), 3.59 (m, 1H,
CH-MOE), 3.90 (m, 2H, H5%), 4.37 (m, 1H, H4%) 4.93
4.2.2. N6-Benzoyl-3%-O-levulinyl-2%-deoxyadenosine 1c.
Rf (10% MeOH/CH2Cl2): 0.44; mp: 125–127°C; [h]2D0=
−26.7 (c 0.9, CHCl3); IR (KBr): w 2360, 2341, 1733,
3
3
(dd, 1H, H2%, JHH 7.8, JHH 5.1 Hz), 5.64 (apparent d,
3
3
1H, H3%, JHH 4.9 Hz), 5.94 (d, 1H, H1%, JHH 7.9 Hz),
7.56 (m, 3H, Hm+Hp), 8.02 (apparent d, 2H, Ho, JHH
1
1715, and 1687 cm−1; H NMR (CDCl3, 200 MHz): l
3
2.22 (s, 3H, Me-Lev), 2.61 (m, 3H, CH2-Lev+H2%), 2.82
(m, 2H, CH2-Lev), 3.18 (m, 1H, H2%), 3.97 (m, 2H, H5%),
7.2 Hz), 8.11 (s, 1H, H2 or H8), and 8.77 (s, 1H, H8 or
H2); 13C NMR (CDCl3, 75.5 MHz): l 27.7 (CH2-Lev),
29.8 (Me-Lev), 37.7 (CH2-Lev), 58.6 (Me-MOE), 62.8
(C5%), 70.4 (CH2-MOE), 71.7 (CH2-MOE), 72.3 (C3%),
79.9, 86.1 (C2%+C4%), 89.6 (C1%), 124.4 (C5), 127.8, 128.8
(Co+Cm), 132.9 (Cp), 133.2 (Ci) 143.2 (C2 or C8), 150.2
(C4), 150.4 (C6), 151.9 (C8 or C2), 164.5 (PhC=O),
171.9 (C=O), and 206.3 (C=O); MS (ESI+, m/z): 528
[(M+H)+, 100%] 550 [(M+Na)+, 50], and 566 [(M+K)+,
35]. Anal. calcd (%) for C25H29N5O8: C, 56.92; H, 5.54;
N, 13.28. Found: C, 56.8; H, 5.4; N, 13.2.
3
4.31 (s, 1H, H4%), 5.57 (apparent d, 1H, H3%, JHH 5.4
3
3
Hz), 6.38 (dd, 1H, H1%, JHH 9.8, JHH 5.2 Hz), 7.57 (m,
3H, Hm+Hp), 8.05 (m, 2H, Ho), 8.17 (s, 1H, H2 or H8),
and 8.79 (s, 1H, H8 or H2). 13C NMR (CDCl3, 75.5
MHz): l 27.7 (CH2-Lev), 29.6 (Me-Lev), 37.6 (CH2-
Lev+C2%), 62.8 (C5%), 76.1, 86.8 (C3%+C4%), 87.0 (C1%),
124.3 (C5), 127.8, 128.6 (Co+Cm), 132.7 (Cp), 133.2 (Ci),
142.3 (C2 or C8), 150.1 (C4), 150.6 (C6), 151.9 (C8 or
C2), 164.8 (PhC=O), 172.1 (C=O), and 206.4 (C=O);
MS (ESI+, m/z): 454 [(M+H)+, 20%], 476 [(M+Na)+,
33], and 492 [(M+K)+, 5]. Anal. calcd (%) for
C22H23N5O6: C, 58.27; H, 5.11; N, 15.44. Found: C,
58.2; H, 5.2; N, 15.5.
4.3.3. 5%-O-Levulinyl-2%-O-(2-methoxyethyl)-5-methylcy-
tidine 4b. Rf (10% MeOH/CH2Cl2): 0.32; mp: 135–
137°C; [h]2D0=+32.9 (c 0.4, MeOH); IR (KBr): w 3396,
1
3333, 3264, 3225, 2919, 1734, and 1662 cm−1; H NMR
4.3. General procedure for the regioselective enzymatic
acylation of 2%-alkylribonucleosides
(CDCl3, 200 MHz): l 1.89 (s, 3H, Me), 2.15 (s, 3H,
Me-Lev), 2.58 (m, 2H, CH2-Lev), 2.76 (m, 2H, CH2-
Lev), 3.32 (s, 3H, OMe), 3.52 (m, 2H, CH2-MOE), 3.81
(m, 1H, CH-MOE), 4.05 (m, 4H, H2%+H3%+H4%+CH-
MOE), 4.38 (m, 2H, H5%), 5.80 (m, 1H, H1%), and 7.43 (s,
1H, H6); 13C NMR (CDCl3, 75.5 MHz): l 13.1 (Me),
27.6 (CH2-Lev), 29.6 (Me-Lev), 37.6 (CH2-Lev), 58.6
(Me-MOE), 62.9 (C5%), 68.7 (C3%), 69.9 (CH2-MOE),
71.4 (CH2-MOE), 80.8, 82.2 (C2%+C4%), 89.5 (C1%), 101.6
(C5), 137.6 (C6), 155.6 (C2), 165.7 (C4), 172.2 (C=O),
and 206.4 (C=O); MS (ESI+, m/z): 436 [(M+Na)+,
10%] and 452 [(M+K)+, 20]. Anal. calcd (%) for
C18H27N3O8: C, 52.29; H, 6.58; N, 10.16. Found: C,
52.3; H, 6.5; N, 10.2.
The reactions were carried out at 0.2 M concentration
following the same procedure as previously described
for 2%-deoxynucleosides except the purification step.
The monolevulinyl derivatives of 2%-alkylribonu-
cleosides were soluble in Et2O and separation from the
remaining oxime ester by precipitation was not possi-
ble. Thus, after enzyme filtration the residue was
purified by flash chromatography (gradient eluent 2–
i
10% PrOH/CH2Cl2, except 5–15% MeOH/CH2Cl2 for
4g).
4.3.1. N4-Benzoyl-3%-O-levulinyl-2%-O-(2-methoxyethyl)-
5-methylcytidine 3c. Rf (10% MeOH/CH2Cl2): 0.56; mp:
102–104°C; [h]2D0=+55.7 (c 0.4, MeOH); IR (KBr): w
4.3.4. 5%-O-Levulinyl-2%-O-(2-methoxyethyl)adenosine 4d.
Rf (10% MeOH/CH2Cl2): 0.37; [h]2D0=−40.0 (c 2.7,
MeOH); IR (NaCl): w 3336, 3198, 2928, 1733, 1715, and
1
3426, 3067, 2922, 1738, 1705, and 1645 cm−1; H NMR
3
(CDCl3, 200 MHz): l 2.09 (d, 3H, Me, JHH 1.0 Hz),
1
2.20 (s, 3H, Me-Lev), 2.66 (m, 2H, CH2-Lev), 2.78 (m,
2H, CH2-Lev), 3.28 (s, 3H, OMe), 3.46 (m, 2H, CH2-
MOE), 3.71–4.02 (m, 4H, CH2-MOE+H5%), 4.26 (m,
1651 cm−1; H NMR (CDCl3, 300 MHz): l 2.12 (s, 3H,
Me-Lev), 2.55 (m, 2H, CH2-Lev), 2.72 (m, 2H, CH2-
Lev), 3.30 (s, 3H, OMe), 3.47 (m, 2H, CH2-MOE), 3.74
(m, 1H, CH-MOE), 3.87 (m, 1H, CH-MOE), 4.26 (m,
1H, H4%), 4.38 (m, 3H, H3%+H5%), 4.62 (m, 1H, H2%), 6.06
3
1H, H4%), 4.51 (apparent t, 1H, H2%, JHH 5.4 Hz), 5.32
3
(apparent t, 1H, H3%, JHH 4.9 Hz) 5.70 (d, 1H, H1%,
3JHH 5.4 Hz), 7.45 (m, 3H, Hm+Hp), 7.64 (s, 1H, H6),
and 8.30 (m, 2H, Ho); 13C NMR (CDCl3, 50.3 MHz): l
13.5 (Me), 27.7 (CH2-Lev), 29.8 (Me-Lev), 37.7 (CH2-
Lev), 58.9 (Me-MOE), 61.6 (C5%), 70.4 (CH2-MOE),
70.7 (C3%), 71.8 (CH2-MOE), 79.4, 83.3 (C2%+C4%), 91.7
(d, 1H, H1%, JHH 4.3 Hz), 6.59 (br s, 2H, NH2), 8.05 (s,
3
1H, H2 or H8), and 8.25 (s, 1H, H8 or H2); 13C NMR
(CDCl3, 75.5 MHz): l 27.6 (CH2-Lev), 29.6 (Me-Lev),
37.7 (CH2-Lev), 58.7 (Me-MOE), 63.4 (C5%), 69.6 (C3%),
70.1 (CH2-MOE), 71.5 (CH2-MOE), 81.7, 82.0 (C2%+

3540
J. Garc´ıa et al. / Tetrahedron: Asymmetry 14 (2003) 3533–3540
C4%), 87.2 (C1%), 119.7 (C5), 139.1 (C2 or C8), 149.3 (C4),
152.7 (C8 or C2), 155.6 (C6), 172.3 (C=O), and 206.5
(C=O); MS (ESI+, m/z): 424 [(M+H)+, 70%], 446
[(M+Na)+, 90], and 462 [(M+K)+, 50]. Anal. calcd (%)
for C18H25N5O7: C, 51.06; H, 5.95; N, 16.54. Found: C,
51.0; H, 5.8; N, 16.6.
168–169; (b) For comprehensive literature, see: Current
Protocols in Nucleic Acid Chemistry; Beaucage, S. L.;
Bergstrom, D. E.; Glick, G. D.; Jones, R. A., Eds.; John
Wiley: New York, 1999; Chapter 2.
6. (a) Kumar, G.; Poonian, M. S. J. Org. Chem. 1984, 49,
4905–4912; (b) Iwaki, S.; Ohtsuka, E. Nucleic Acids Res.
1988, 16, 9443–9456; (c) Iwai, S.; Toshiro, S.; Ohtsuka, E.
Tetrahedron 1990, 46, 6673–6688; (d) Krotz, A. H.;
Klopchin, P.; Cole, D. L.; Ravikumar, V. T. Nucleosides
Nucleotides 1997, 16, 1637–1640; (e) Eleuteri, A.; Cheru-
vallath, Z. S.; Capaldi, D. C.; Cole, D. L.; Ravikumar, V.
T. Nucleosides Nucleotides 1999, 18, 1803–1807; (f) Reese,
C. B.; Song, Q. Nucleic Acids Res. 1999, 27, 963–971.
7. For recent reviews on enzymatic transformations in
nucleosides, see: (a) Ferrero, M.; Gotor, V. Chem. Rev.
2000, 100, 4319–4347; (b) Ferrero, M.; Gotor, V.
Monatsh. Chem. 2000, 131, 585–616.
8. (a) Garc´ıa, J.; Ferna´ndez, S.; Ferrero, M.; Sanghvi, Y. S.;
Gotor, V. J. Org. Chem. 2002, 67, 4513–4519; (b) Garc´ıa,
J.; Ferna´ndez, S.; Ferrero. M.; Sanghvi, Y. S.; Gotor, V.
Nucleosides Nucleotides Nucleic Acids 2003, 22, in press.
9. Ferrero, M.; Gotor, V. In Enzymes in Action. Green
Solutions for Chemical Problems; Zwanenburg, B.; Miko-
lajczyk, M.; Kielbasinski, P., Eds.; Kluwer Academic
Plublishers: Dordrecht, The Netherlands, 2000; Chapter
18, pp. 347–363.
4.3.5. 5%-O-Levulinyl-2%-O-methyladenosine 4f. Rf (10%
MeOH/CH2Cl2): 0.30; mp: 124–126°C; [h]2D0=−15.2 (c
0.5, MeOH); IR (KBr): w 3146, 2968, 2927, 1731, 1713,
1
and 1651 cm−1; H NMR (CDCl3, 300 MHz): l 2.17 (s,
3H, Me-Lev), 2.60 (m, 2H, CH2-Lev), 2.77 (m, 2H,
CH2-Lev), 3.57 (s, 3H, OMe), 4.24 (m, 1H, H4%), 4.44
3
(m, 4H, H2%+H3%+H5%), 6.12 (d, 1H, H1%, JHH 3.1 Hz),
6.18 (br s, 2H, NH2), 8.10 (s, 1H, H2 or H8), and 8.32
(s, 1H, H8 or H2); 13C NMR (CDCl3, 75.5 MHz): l
27.6 (CH2-Lev), 29.6 (Me-Lev), 37.7 (CH2-Lev), 58.7
(OMe), 63.2 (C5%), 69.4 (C3%), 81.9, 83.0 (C2%+C4%), 86.7
(C1%), 119.7 (C5), 138.8 (C2 or C8), 149.2 (C4), 152.9 (C8
or C2), 155.6 (C6), 172.4 (C=O), and 206.7 (C=O); MS
(ESI+, m/z): 380 [(M+H)+, 10%], 402 [(M+Na)+, 100],
and 418 [(M+K)+, 10]. Anal. calcd (%) for C15H21N5O6:
C, 49.04; H, 5.76; N, 19.06. Found: C, 49.1; H, 5.6; N,
19.3.
Acknowledgements
10. Acetonoxime levulinate 5 is now commercially available
from Sai Dru Syn Laboratories, Hyderabad, India,
www.saiintgroup.com.
11. (a) Martin, P. Helv. Chim. Acta 1995, 78, 486–504; (b)
Altmann, K.-H.; Martin, P.; Dean, N. M.; Monia, B. P.
Nucleosides Nucleotides 1997, 16, 917–926; (c) Grotli, M.;
Douglas, M.; Eritja, R.; Sproat, B. S. Tetrahedron 1998,
54, 5899–5914.
Financial support from Principado de Asturias (Spain;
Project GE-EXP01-03) and MCYT (Spain; Project
PPQ-2001-2683) is gratefully acknowledged. S. F. also
thanks MCYT (Spain) for a personal grant (Ramo´n y
Cajal Program). J.G. thanks ISIS Pharmaceuticals for a
predoctoral fellowship.
12. We are currently investigating the effect of enzymatic
acylations with nucleosides locked in North and South
conformation; For more details on nucleoside conforma-
tion, see: Thibaudeau, C.; Chattopadhyaya, J. Stereoelec-
tronic Effects in Nucleosides and Nucleotides and Their
Structural Implications; Uppsala University Press: Upp-
sala, Sweden, 1999.
References
1. Sanghvi, Y. S. In Comprehensive Natural Products Chem-
istry; Kool, E. T., Ed.; Elsevier: Amsterdam, 1999; Vol.
7, Chapter 7, pp. 285–311 and references cited therein.
2. For a recent monograph on this topic, see: Antisense
Drug Technology Principles, Strategies and Applications;
Crooke, S. T., Ed.; Marcel Dekker: New York, 2001.
3. Reese, C. B.; Yan, H. J. Chem. Soc., Perkin Trans. 1
2002, 2619–2633.
4. (a) Sanghvi, Y. S. Org. Process R&D 2000, 4, 168–169;
(b) Sanghvi, Y. S.; Ravikumar, V. T.; Scozzari, A. N.;
Cole, D. L. Pure Appl. Chem. 2001, 73, 175–180.
5. (a) Protective Groups in Organic Synthesis; Green, T. W.;
Wuts, P. G. M., Eds.; John Wiley: New York, 1999; pp.
13. Pedersen, D. S.; Rosenbohm, C. Synthesis 2001, 2431–
2434.
14. Trost, B. M. Acc. Chem. Res. 2002, 35, 695–705.
15. Green Chemical Syntheses and Processes: ACS Sympo-
sium Series 767; Anastas, P. T.; Heine, L. G.;
Williamson, T. C., Eds.; American Chemical Society:
Washington, DC, 2000.
16. The purity of isolated compounds was estimated by
HPLC analysis, and in all cases it was >98%.