Substituent and solvent effects of TMS triflate mediated C1′ epimerization of β-thymidine to α-thymidine

Article abstract of DOI:10.1002/1099-0690(20021)2002:1<87::AID-EJOC87>3.0.CO;2-B

This paper deals with kinetic studies of TMSOTf-mediated C1′ epimerization of β-thymidine to α-thymidine. The effect of neighboring group participation by various 5′-hydroxy protecting groups, such as toluoyl, Et₂CHC(O), Et₂NC(O), and Et₂NC(S), on the β→α conversion is described in detail. The time dependence of the ratio of the α and β anomers in the C1′ epimerization was estimated by ¹H NMR. The α/β equilibrium constants K and the rate constants k_α and k_β were calculated on the basis of the experimental data. As the result, it was concluded that, in acetonitrile, the α/β equilibrium constants K are thermodynamically affected by steric hindrance from the 5′-hydroxy protecting group. On the other hand, the rate constants k_α and k_β are mainly influenced by the stability of the oxonium ion intermediate. In particular, formation of an intramolecularly cyclized iminium ion intermediate from the oxonium ion intermediate, due to the neighboring group participation by the diethylthiocarbamoyl group, tended to decrease the overall reaction rate. Finally, the α/β C1′ epimerization could be carried out with a high α anomer selectivity of 89% through the use of the Et₂CHC(O) group. Thus, 5′-O-pixyl-α-thymidine could be synthesized from β-thymidine as a key intermediate for the synthesis of α-DNA in a considerably improved overall yield of 40%.

Full text of DOI:10.1002/1099-0690(20021)2002:1<87::AID-EJOC87>3.0.CO;2-B

FULL PAPER
Substituent and Solvent Effects of TMS Triflate Mediated C1Ј Epimerization
of β-Thymidine to α-Thymidine
Yuichi Sato,[a] Gohsuke Tateno,[a] Kohji Seio,[a] and Mitsuo Sekine*[a]
Keywords: Epimerization / Kinetics / Protecting groups / Steric hindrance / Trimethylsilyl triflate
This paper deals with kinetic studies of TMSOTf-mediated
C1Ј epimerization of β-thymidine to α-thymidine. The effect
of neighboring group participation by various 5Ј-hydroxy
protecting groups, such as toluoyl, Et2CHC(O), Et2NC(O),
and Et2NC(S), on the βǞα conversion is described in detail.
hand, the rate constants kα and kβ are mainly influenced by
the stability of the oxonium ion intermediate. In particular,
formation of an intramolecularly cyclized iminium ion inter-
mediate from the oxonium ion intermediate, due to the
neighboring group participation by the diethylthiocarbamoyl
The time dependence of the ratio of the α and β anomers in group, tended to decrease the overall reaction rate. Finally,
the C1Ј epimerization was estimated by 1H NMR. The α/β
the α/β C1Ј epimerization could be carried out with a high α
equilibrium constants K and the rate constants kα and kβ were anomer selectivity of 89% through the use of the Et2CHC(O)
calculated on the basis of the experimental data. As the re- group. Thus, 5Ј-O-pixyl-α-thymidine could be synthesized
sult, it was concluded that, in acetonitrile, the α/β equilibrium
constants K are thermodynamically affected by steric hind-
rance from the 5Ј-hydroxy protecting group. On the other
from β-thymidine as a key intermediate for the synthesis of
α-DNA in a considerably improved overall yield of 40%.
Introduction
mixture of the α and β anomers, which were recovered in
67% and 27% yields, respectively.[17] Later, Yamaguchi and
Saneyoshi applied this C1Ј epimerization to the reverse con-
version of β-thymidine 1(β) to α-thymidine 1(α), and ob-
tained α-thymidine 1(α) from 3Ј,5Ј-di-O-toluoylthymidine
2a(β)[18] in an overall yield of 25%, as shown in Scheme 1.
They used an elevated temperature of 70 °C to accelerate
the epimerization. Our own study found that this epimeriz-
ation resulted in a 75:25 mixture of the α and β anomers
2a(α) and 2a(β), and serious competitive decomposition was
observed. Thus, compound 2a(β) could not be obtained in
good yield.
α-Oligodeoxynucleotides[1Ϫ5] and derivatives[6] of nucle-
otide components with α configurations at the anomeric
centers have proven to be promising new sources in the anti-
sense strategy.[7] Imbach and others have reported that these
substances are resistant to nucleases[2] in cells and can se-
lectively hybridize with mRNA,[5,6] ssDNA,[2] and
dsDNA.[4] However, the α-nucleoside monomers required
as the starting materials for the synthesis of α-oligodeoxy-
nucleotides are not available in nature, so must be prepared
by chemical approaches. Unfortunately, α-deoxynucleosides
have long been viewed only as undesired by-products in the
synthesis of β-deoxynucleosides by glycosylation of deoxyri-
boside derivatives with silylated or metallated bases.[8] Des-
pite this background, several methods for the synthesis of
α-deoxynucleosides have been reported.[8Ϫ16] However,
these methods involve multi-step reactions and do not at-
tain ideal selectivity. Compared with these de novo syn-
theses of α-deoxynucleosides, simple C1Ј epimerization of
commercially available β-deoxynucleosides would provide a
more straightforward route to α-deoxynucleosides. In 1981,
Vorbrüggen reported that treatment of α-(5-ethyldeoxy-
uridine) with TMSOTf in CH3CN at 25 °C for 46 h gave a
Therefore, in order to identify the essential factors for
enhancement of the α/β ratio of 2a(α) and 2a(β) in the
TMSOTf-mediated epimerization, and also to control the
competitive decomposition, we studied the C1Ј epimeriz-
ation of various 5Ј-O-acylthymidine derivatives in more de-
tail. In particular, our interest was focused on the possibility
of neighboring group participation by the 5Ј-O-protecting
group in this type of epimerization.[19] It was also important
to see whether the α/β equilibrium in the
VorbrüggenϪSaneyoshi method could be shifted more pref-
erentially to the α isomer and whether the reaction rate
could be accelerated. For this purpose, four kinds of acyl
groups Ϫ toluoyl (Tol), (2-ethyl)butyryl [Et2CHC(O)Ϫ],
N,N-diethylcarbamoyl [Et2NC(O)Ϫ], and N,N-diethylthio-
carbamoyl [Et2NC(S)Ϫ] were chosen. The Tol group was
chosen as a reference, since Saneyoshi used this protecting
group previously.[18] The latter two were expected to have
strong proximal effects on the stabilization of oxonium in-
[a]
Department of Life Science, Tokyo Institute of Technology,
Nagatsuta, Midoriku,
Yokohama 227-8501, Japan
Fax: (internat.) ϩ 81-45/924-5772
E-mail: msekine@bio.titech.ac.jp
Eur. J. Org. Chem. 2002, 87Ϫ93
WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
1434Ϫ193X/02/0101Ϫ0087 $ 17.50ϩ.50/0
87

Y. Sato, G. Tateno, K. Seio, M. Sekine
FULL PAPER
Scheme 1
termediates generated by the TMSOTf-mediated activation dine 9(α) as a key synthetic intermediate for the synthesis
of the glycosidic bond. This inherent proximal effect has of α-DNA.
previously been exploited for the α-selective glycosylation
of 1-O-acetyl-3-O-benzyl-5-O-(N,N-diethylthiocarbonyl)-2-
Results and Discussion
deoxyribofuranose with 2,4-O-bis(trimethylsilyl)thymine in
the presence of Lewis acids.[20] The observed α selectivity
was explained in terms of the formation of a cyclic iminium
salt that could be attacked by the silylated thymine from
the α face.
In this paper we report that the β/α epimerization of thy-
midine 1(β) was greatly affected by the nature of the 5Ј-O-
acyl group and the solvent, and also describe its ultimate
application to a convenient synthesis of 5Ј-O-pixyl-α-thymi-
3Ј,5Ј-Di-O-toluoylthymidine[21] 2a(β) was prepared by
acylation of 1(β) with toluoyl chloride. Compound 2b(β),
with an Et2CHC(O) group, was synthesized in 73% yield by
a two-step procedure from thymidine 1(β), by way of the
5Ј-O-acylated species 4. The other 3Ј,5Ј-di-O-acylated thy-
midine derivatives 2c and 2d were synthesized from 3Ј-O-
(4,4Ј-dimethoxytrityl)thymidine[22] (5) and 3Ј-O-(tert-butyl-
dimethylsilyl)thymidine[23] (6), respectively, as shown in
Scheme 2.
In the β/α epimerization, thymidine derivatives 2a؊d
were silylated by treatment with hexamethyldisilazane
(HMDS) in toluene under reflux for 1 h. After addition of
1 equiv. of TMSOTf to the silylated thymidine derivatives
3a؊d(β), the mixture was stirred at 30 °C for 30Ϫ96 h
(Scheme 3). It was found that when an elevated temperature
of 70 °C was employed, as reported by Saneyoshi,[18] the
reaction rate was accelerated but the yield of 2a(α) was re-
duced because of unfavorable decomposition, giving rise to
a considerable quantity of charcoal-like materials.[24] The
current reactions were therefore carried out at 30 °C with 1
equiv. of TMSOTf, which produced a marked improvement
in the recovery of the α/β mixture. To check solvent effects,
the reaction was also examined in CH2Cl2 in place of
CH3CN. At appropriate times, the ratio of 2a؊d(α)/
2a؊d(β) was estimated by 1H NMR. These results are
shown in Figure 1.
To our surprise, a marked solvent effect between CH3CN
and CH2Cl2 was observed. In the former, the βǞα epi-
merization reaction rates ranked in the following order:
Scheme 2
88
Eur. J. Org. Chem. 2002, 87Ϫ93

C1Ј Epimerization of β-Thymidine to α-Thymidine
FULL PAPER
2c(β) Ͼ 2b(β) Ͼ 2a(β) Ͼ 2d (β). On the other hand, dramat-
ically different results were obtained in the latter solvent.
Namely, no epimerization occurred at all in the cases of
2a؊c(β), and only 2d(β) showed a very slow epimerization
(α/β ratio ϭ 9:91 after 96 h).
The mixtures of 2a؊d(α) and 2a؊d(β) obtained after
20 h with CH3CN were purified by silica gel column chro-
matography without further separation of anomer pairs.
The isolated (or recovered) yields of 2a؊d(α) and 2a؊d(β)
and their ratios at this time are summarized in Table 1. The
mixtures of 2a؊d(α) and 2a؊d(β) could be recovered in
82Ϫ91% yields.
The final α/β ratios for 2a(α) and 2a(β) in the reaction
carried out at 30 °C reached 81:19, which was better than
that (75:25) obtained at 70 °C. It was also found that, com-
pared with 2a(β) with the toluoyl group, the α anomer se-
lectivity in the α/β C1Ј epimerization after 20 h was in-
creased to 89:11 and 88:12 by introduction of the
Et2CHC(O) and Et2NC(O) groups, respectively, at the 5Ј-
hydroxy group of thymidine, as shown in Table 1.
In the case of 2a؊c(β), all epimerizations in CH3CN re-
ached equilibrated mixtures within 20 h. The epimerization
of 2c(β) was the fastest among them. Compound 2d(β), with
the Et2NC(S) group, underwent very slow epimerization
that reached a plateau after 50 h, but α-selective epimeriz-
ation of 91% was finally accomplished. The equilibrium
constants K of these α/β epimerizations were calculated
from the data plotted in Figure 1. (These results are also
shown in Table 1.) Compound 2d(β), with an Et2NC(S)
group, gave the best equilibrium constant K (9.95) but took
longer to reach equilibrium.
Scheme 3
This result implies that the ultimate α/β ratio in these
epimerizations depends not on electronic factors, but on
steric factors relating to the 5Ј-O-protecting groups, since
compounds 2b؊d(β), with structurally similar acyl groups,
showed very similar α/β ratios (88:12 to 91:9). It is likely
that the Et2NC(O), Et2NC(S), and Et2CHC(O) groups in-
teract sterically more strongly than the toluoyl group with
the silylated thymine base. Accordingly, the β-thymidine de-
rivatives 2b؊d(β) with these protecting groups are more
susceptible to α attack of the silylated thymine and are more
thermodynamically unfavorable than compound 2a(β) with
the toluoyl group, as shown in Figure 2.
Figure 1. Time dependence of C1Ј epimerization of 5Ј-O-acyl-3Ј-
O-toluoylthymidine derivatives 2a؊d(β) in acetonitrile and dichlor-
1
omethane, based on H NMR analysis; in acetonitrile: open circle
2a(β), open diamond 2b(β), open triangle 2c(β), open square 2d(β);
in dichloromethane: closed circle 2a(β), closed diamond 2b(β),
closed triangle 2c(β), closed square 2d(β)
It is suggested from Taft steric parameters that the
Et2CHC(O) group is larger than the benzoyl group.[25]
Table 1. TMSOTf-mediated C1Ј epimerization of 2a؊d(β) in acetonitrile, kinetic data, and Rf values of the products
Compd.
R
Recovery of 2(α) and 2(β) α/β[a]
after epimerization
(%)
Rf (TLC)
Equilibrium Rate constant
Half-time
[h]
constant
[ϫ 106 sϪ1
]
α isomer β isomer
K
kα
kβ
2a(β)
2b(β)
2c(β)
2d(β)
toluoyl
Et2CHC(O)Ϫ 91
Et2NC(O)Ϫ
Et2NC(S)Ϫ
89
81:19
89:11
88:12
64:36
0.53[b]
0.59[c]
0.35[b]
0.54[b]
0.59[b]
0.64[c]
0.42[b]
0.59[b]
4.38
8.08
7.12
9.95
36.0
56.7
82.7
15.8
7.7
7.0
10.2
1.6
6.0
4.5
2.5
12.5
82
89
77 (50 h)
91:9 (50 h)
[a]
[b]
[c]
The ratio measured after 20 h. CHCl3/EtOAc (1:1, v/v). CHCl3/EtOAc/toluene (1:1:1, v/v/v), developed 3 times.
Eur. J. Org. Chem. 2002, 87Ϫ93
89

Y. Sato, G. Tateno, K. Seio, M. Sekine
FULL PAPER
Figure 2. Interaction of the 5Ј-acyl moiety with the thymine base
Therefore, it is implied that the use of sterically hindered
protecting groups here resulted in better α/β ratios.
The 5Ј-O-(N,N-diethylthiocarbamoyl)thymidine derivat-
ive 2d(β) gave the best result as far as the α anomer selectiv-
ity was concerned, but the C1Ј epimerization rate in
CH3CN was the slowest. To clarify the reason for the differ-
ences in reaction rate between 2d(β) and the other three
compounds 2a؊c(β), we determined the rate constants kα
and kβ, respectively, of the βǞα epimerization and of its
αǞβ reverse reaction. To this end, the equation ln(KЈ/
{KЈ Ϫ xα}) ϭ (kα ϩ kβ)t was used [xα ϭ α/(α ϩ β), KЈ ϭ K/
(K ϩ 1)] (see Figure 3).
Scheme 4
In the second reaction pathway (B), the C1Ј epimeriz-
ation is retarded by formation of the cyclic iminium ion
intermediate 8. Such a cyclic intermediate would be ex-
pected to have a considerable degree of stability[20] and as-
sumed to reduce the effective concentration of the reactive
intermediate 7. It is predicted that attack by the most nucle-
ophilic Et2NC(S) group on the oxonium C1Ј carbon atom
is more likely than that by the less nucleophilic Tol,
Et2NC(O), and Et2CHC(O) groups. In general, neighboring
group participation by acyl groups is less effective in polar
solvents,[26] because the iminium ion intermediates can be
stabilized by solvation. Therefore, the compound 2d(β),
with the Et2NC(S) group as a strong nucleophile, under-
went the slowest epimerization in CH3CN, as shown in Fig-
ure 1.
When the C1Ј epimerization of 2a؊c(β) was carried out
in less polar dichloromethane, no reaction occurred. Only
the C1Ј epimerization of 2d(β) in dichloromethane pro-
ceeded, extremely slowly, as shown in Figure 1. Consider-
able degradation of the starting materials accompanied the
reaction. These results can be explained as follows. In the
less polar solvent CH2Cl2, the oxonium ion intermediate 7
Figure 3. Kinetics of C1Ј epimerization of 2a؊d in acetonitrile
As shown in Table 1, it was found that kα and kβ for 2d(β) cannot be stabilized because of weaker solvation, so that
were very small compared with those of the other three the first cleavage of the glycosidic bond does not occur
compounds 2a؊c(β). If there is a possibility that the sulfur favorably. It is likely that the oxonium ion 7 can, in the
atom of the Et2NC(S) group may attack the C1Ј carbon case of 2d(β), be trapped by the thiocarbonyl group and
atom, it is reasonable to assume that two kinds of reaction be converted into a thermodynamically more stable cyclic
pathways (A) and (B) may simultaneously exist in the α/β intermediate 8. Therefore, the α anomer product was
equilibrium, as shown in Scheme 4. The rate constants kα3 formed through attack of 8 with the silylated thymine only
and kβ3 between the oxonium intermediate 7 and the imin- in this case.
ium ion intermediate 8 require discussion. Our results sug-
From a practical point of view, α-thymidine would best
gest that the constant kβ3 is much smaller than kα3 and that be obtained from 2a(β) or 2b(β), as multi-step reactions
this additional slow step affects the overall reaction rate so were required for preparation of compounds 2c(β) and
that the rate constants kα and kβ are decreased.
2d(β) (see Scheme 2). However, it turned out that it was not
In the first reaction pathway (A), the C1Ј epimerization easy to separate the pure α anomers from the mixtures of
between the β anomer 2d(β) and the α anomer 2d(α) arises 2a(α)/2a(β) and 2b(α)/2b(β) obtained by epimerization, ow-
from formation of the oxonium ion intermediate 7. In par- ing to the close Rf values of the compounds on TLC, as
ticular, the epimerization and the reverse reaction at 30 °C shown in Table 1. Therefore, we decided to remove the 3Ј-
in polar solvents such as acetonitrile proceed through the and 5Ј-protecting groups from the most highly α-enriched
solvated oxonium ion intermediate.
(89:11) mixture of 2b(α)/2b(β) and protect the 5Ј-hydroxy
90
Eur. J. Org. Chem. 2002, 87Ϫ93

C1Ј Epimerization of β-Thymidine to α-Thymidine
FULL PAPER
tack of the silylated thymine from the β face. Further stud-
ies in this direction are now underway.
Experimental Section
1
General Remarks: H and 13C NMR spectra were recorded at 270
and 68 MHz, respectively. The chemical shifts were measured from
tetramethylsilane for 1H NMR spectra, CDCl3 (δ ϭ 77) for 13C
NMR. TLC was performed with Kieselgel 60-F-254 (0.25 mm).
Column chromatography was performed with silica gel C-200 or
C-300 purchased from Wako Co. Ltd., and an aquarium minipump
could conveniently be used to attain sufficient pressure for rapid
chromatographic separation. Pyridine was distilled twice from p-
toluenesulfonyl chloride and from calcium hydride and then stored
˚
over molecular sieves (4 A). HMDS and TMSOTf were purchased
from Shinetsu Chemical Co. Ltd. Elemental analyses were per-
formed by the Microanalytical Laboratory, Tokyo Institute of
Technology at Nagatsuta.
3Ј,5Ј-Di-O-toluoylthymidine [2a(β)]:[18,21] Toluoyl chloride (680 mg,
4.4 mmol) was added to a solution of thymidine (485 mg, 2 mmol)
in dry pyridine (10 mL). The solution was stirred at room temper-
ature for 1 h. The mixture was partitioned between CHCl3 (50 mL)
and 5% NaHCO3 (50 mL). The organic layer was collected, washed
with 5% NaHCO3 (50 mL), dried with Na2SO4, and filtered, and
the solvents were evaporated under reduced pressure. The residue
was chromatographed on a column of silica gel (30 g) with CHCl3/
MeOH (100:1, v/v) to give 2a(β) as a foam (919 mg, 96%). 1H
NMR (270 MHz, CDCl3): δ ϭ 1.62 (s, 3 H), 2.31Ϫ2.69 (m, 2 H),
2.42, (s, 3 H), 2.43 (s, 3 H), 4.52 (m, 1 H), 4.63Ϫ4.80 (m, 2 H),
5.63 (m, 1 H), 6.46 (dd, J ϭ 8.9, 5.6 Hz, 1 H), 7.25Ϫ7.96 (m, 9 H),
8.86 (br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 12.2, 21.8, 21.8,
38.1, 64.2, 74.9, 82.8, 84.9, 111.6, 126.2, 126.5, 129.2, 129.4, 129.4,
129.7, 134.3, 144.4, 150.1, 163.2, 165.8, 165.9. C26H26N2O7 (478.5):
C 65.26, H 5.48, N 5.85; found C 65.43, H 5.36, N 6.11.
Scheme 5
group of the resulting 1(α)/1(β) mixture in situ with pixyl
chloride. As a result, 5Ј-O-pixylated α-thymidine 9(α) was
successfully separated from its β anomer 9(β)[27] and could
finally be crystallized from methanol (Scheme 5). Thus,
compound 9(α) was obtained in an overall yield of 40%
from β-thymidine 1(β), while the β anomer could not be
1
detected by H NMR, which suggested it was more than
99.5% pure. This compound can be directly used for the
synthesis of α-oligodeoxynucleotides. The 5Ј-O-pixylated α-
thymidine 9(α) could, in our hands, also be obtained in an
overall yield of 15Ϫ20% from thymidine by the original
method.[18] Therefore, we had been able to improve the isol-
ated yield of 9(α) significantly. Development of this new
route to the key synthetic intermediate 9(α) would facilitate
the synthesis of a variety of modified α-oligonucleotides.
5Ј-O-(2-Ethylbutyryl)-3Ј-O-toluoylthymidine [2b(β)]: 2-Ethylbutyryl
chloride (444 mg, 3.3 mmol) was added to a solution of thymidine
(727 mg, 3 mmol) in dry pyridine (30 mL). The solution was stirred
at room temperature for 2 h and toluoyl chloride (510 mg,
3.3 mmol) was then added. The mixture was stirred at room tem-
perature for 1 h, and the reaction was quenched by addition of
water (2 mL). After having been kept for 10 min, the mixture was
partitioned between CHCl3 (100 mL) and NaHCO3 (5%, 50 mL).
The organic layer was collected, washed with 5% NaHCO3
(50 mL), dried with Na2SO4, and filtered, and the solvents were
evaporated under reduced pressure. The residue was chromato-
graphed on a column of silica gel (15 g) with hexane/CHCl3 (20:80,
v/v) to give 2b(β) as a foam (1.01 g, 73%). 1H NMR (270 MHz,
CDCl3): δ ϭ 0.92 (t, J ϭ 7.4 Hz, 6 H), 1.55Ϫ1.67 (m, 4 H), 1.97
(s, 3 H), 2.19Ϫ2.70 (m, 6 H), 4.30Ϫ4.63 (m, 2 H), 4.43 (m, 1 H),
5.43 (m, 1 H), 6.43 (dd, J ϭ 8.7, 5.4 Hz, 1 H), 7.25Ϫ7.95 (m, 5 H),
9.53 (br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 11.9, 11.9, 12.6,
21.8, 24.9, 25.0, 37.8, 48.9, 63.6, 74.5, 82.3, 84.8, 111.5, 126.1,
129.1, 129.7, 134.4, 144.4, 150.3, 163.6, 165.8, 175.4. C24H30N2O7
(458.5): C 62.87, H 6.59, N 6.11; found C 62.26, H 6.64, N 6.19.
Conclusion
In conclusion, we have shown that α/β C1Ј epimerization
of thymidine derivatives at 30 °C is possible, and that the α
anomer selectivity can be improved to ca. 90% through the
use of Et2CHC(O)Ϫ, Et2NC(O)Ϫ, and Et2NC(S)Ϫ as 5Ј-
O-protecting groups. Use of the thiocarbamoyl group,
which was effective for irreversible selective glycosyl-
ation,[20] resulted in slow epimerization in acetonitrile but
only effected epimerization in dichloromethane because of
its strong neighboring group participation ability. It was
also shown that the kinetic rates in the C1Ј epimerization
are considerably affected by solvent effects. The α/β ratio
was highly dependent on the thermodynamic stability of the
α- and β-thymidine derivatives. The main factor determin-
ing the α/β ratio is not neighboring group participation, but
the steric hindrance due to the 5Ј-protecting group.
5Ј-O-(N,N-Diethylcarbamoyl)-3Ј-O-toluoylthymidine [2c(β)]: So-
dium hydride (1.5 g, 37.5 mmol) was added to a solution of 3Ј-O-
(4,4Ј-dimethoxytrityl)thymidine (8.17 g, 15 mmol) in dry THF
(30 mL). The solution was stirred at room temperature for 30 min
and N,N-diethylcarbamoyl chloride (4.07 g, 30 mmol) was then ad-
These results suggest that it may be possible to enhance
the ratio of α and β anomers through a choice of more
hindered 5Ј-O-protecting groups, which should inhibit at- ded. After having been stirred at room temperature for 1 d, the
Eur. J. Org. Chem. 2002, 87Ϫ93 91

Y. Sato, G. Tateno, K. Seio, M. Sekine
FULL PAPER
mixture was quenched by addition of phosphate buffer (0.2 ,
1 h. The mixture was concentrated in vacuo. After the residue had
pH ϭ 6.0, 200 mL). The mixture was extracted with CHCl3 been dissolved in anhydrous CH3CN or CH2Cl2 (20 mL), TMSOTf
(100 mL ϫ 2), and the organic layer was collected. Trifluoroacetic
acid (6 mL) was added to the solution, and the mixture was stirred
at room temperature for 1 h. After neutralization by addition of
(360 µL, 2 mmol) was added. The mixture was stirred at 30 °C for
30Ϫ96 h. At appropriate times, an aliquot was taken from the mix-
1
ture and analyzed by H NMR. Each α anomer formed was isol-
saturated NaHCO3, the mixture was extracted with CHCl3 ated by silica gel column chromatography and characterized by 1H
(100 mL). The organic layer was collected, dried with Na2SO4, and
filtered, and the solvents were evaporated under reduced pressure.
The residue was chromatographed on a column of silica gel (50 g)
with hexane/EtOAc (30:70, v/v). The product was dissolved in dry
pyridine (75 mL). Toluoyl chloride (2.55 g, 16.5 mmol) was added
to the mixture. After having been stirred at room temperature for
1 h, the mixture was partitioned between CHCl3 (100 mL) and 5%
NaHCO3 (100 mL). The organic layer was collected, dried with
Na2SO4, and filtered, and the solvents were evaporated under re-
duced pressure. The residue was chromatographed on a column of
silica gel (50 g) with hexane/EtOAc (50:50, v/v) to give 2c(β) as a
foam (3.65 g, 53%). 1H NMR (270 MHz, CDCl3): δ ϭ 1.14 (t, J ϭ
7.1 Hz, 6 H), 1.94 (s, 3 H), 2.12Ϫ2.71 (m, 2 H), 2.43 (s, 3 H), 3.31
(br., 4 H), 4.43Ϫ4.48 (m, 3 H), 5.49 (m, 1 H), 6.39 (dd, J ϭ 8.7,
5.4 Hz, 1 H), 7.25Ϫ7.95 (m, 5 H), 9.15 (br., 1 H). 13C NMR
(68 MHz, CDCl3): δ ϭ 12.6, 21.8, 37.9, 64.5, 74.6, 82.9, 85.1, 111.3,
126.2, 129.1, 129.7, 134.5, 144.3, 150.2, 155.0, 163.5, 165.8.
C23H29N3O7 (459.5): C 60.12, H 6.36, N 9.14; found C 59.97, H
6.31, N 9.11.
and 13C NMR as follows.
3Ј,5Ј-Di-O-toluoyl-α-thymidine [2a(α)]:[18,21] 1H NMR (270 MHz,
CDCl3): δ ϭ 1.87 (s, 3 H), 2.41Ϫ3.01 (m, 8 H), 4.46Ϫ4.59 (m, 2
H), 4.89 (m, 1 H), 5.61 (m, 1 H), 6.38 (dd, J ϭ 7.3, 1.6 Hz, 1 H),
7.21Ϫ7.95 (m, 9 H), 9.03 (br., 1 H). 13C NMR (68 MHz, CDCl3):
δ ϭ 12.7, 21.8, 38.9, 64.1, 74.8, 85.3, 87.2, 110.2,126.0, 126.4, 129.3,
129.3, 129.4, 129.6, 135.2, 144.2, 144.7, 150.1, 163.7, 165.4, 166.0.
C26H26N2O7 (478.5): C 65.26, H 5.48, N 5.85; found C 64.73, H
5.41, N 5.95.
5Ј-O-(2-Ethylbutyryl)-3Ј-O-toluoyl-α-thymidine [2b(α)]: 1H NMR
(270 MHz, CDCl3): δ ϭ 0.92 (t, J ϭ 7.4 Hz, 6 H), 1.52Ϫ1.72 (m,
4 H), 1.96 (s, 3 H), 2.24Ϫ2.97 (m, 6 H), 4.30 (m, 2 H), 4.78 (m, 1
H), 5.50 (m, 1 H), 6.33 (dd, J ϭ 6.9, 1.6 Hz, 1 H), 7.21Ϫ7.81 (m,
5 H), 9.32 (br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 11.9, 12.7,
21.8, 25.0, 25.0, 38.9, 48.8, 63.7, 74.8, 85.0, 87.2, 110.2, 125.9,
129.3, 129.4, 135.1, 144.7, 150.2, 163.8, 165.4, 175.4.
C24H30N2O7·0.5H2O (467.5): C 61.66, H 6.68, N 5.99; found C
61.70, H 6.27, N 6.22.
5Ј-O-(N,N-Diethylcarbamoyl)-3Ј-O-toluoyl-α-thymidine [2c(α)]: 1H
NMR (270 MHz, CDCl3): δ ϭ 1.16 (t, J ϭ 7.1 Hz, 6 H), 1.88 (s, 3
H), 2.41Ϫ2.89 (m, 5 H), 3.31 (br., 4 H), 4.19Ϫ4.35 (m, 2 H), 4.77
(m, 1 H), 5.56 (m, 1 H), 6.33 (dd, J ϭ 6.9, 1.6 Hz, 1 H), 7.21Ϫ7.80
(m, 5 H), 9.29 (br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 12.7,
21.8, 38.7, 64.5, 74.6, 85.3, 87.1, 110.1, 126.1, 129.2, 129.4, 135.2,
144.6, 150.1, 155.0, 163.8, 165.3. C23H29N3O7 (459.5): C 60.12, H
6.36, N 9.14; found C 59.94, H 6.34, N 9.18.
5Ј-O-(N,N-Diethylthiocarbamoyl)-3Ј-O-toluoylthymidine [2d(β)]: So-
dium hydride (288 mg, 7.2 mmol) was added to a solution of 3Ј-
O-(tert-butyldimethylsilyl)thymidine (1.07 g, 3 mmol) in dry THF
(10 mL). The solution was stirred at room temperature for 30 min
and N,N-diethylthiocarbamoyl chloride (958 mg, 6 mmol) was then
added. After having been stirred at room temperature for 1 d, the
mixture was quenched by addition of phosphate buffer (pH ϭ 6.0,
0.2 , 100 mL). The mixture was extracted with CHCl3 (100 mL ϫ
2), and the organic layer was collected, dried with Na2SO4, and
filtered, and the solvents were evaporated under reduced pressure.
The residue was dissolved in dry THF (6 mL). Tetrabutylammo-
nium fluoride hydrate (1.57 g, 6 mmol) was added to the solution.
After having been stirred at room temperature for 4 h, the mixture
was partitioned between CHCl3 (100 mL) and 5% NaHCO3
(100 mL). The organic layer was collected, dried with Na2SO4, and
filtered, and the solvents were evaporated under reduced pressure.
The residue was chromatographed on a column of silica gel (30 g)
with hexane/EtOAc (50:50, v/v). The product was subsequently dis-
solved in dry pyridine (3.7 mL). Toluoyl chloride (324 mg,
2.1 mmol) was added to the mixture. After having been stirred at
room temperature for 1 h, the mixture was partitioned between
CHCl3 (100 mL) and 5% NaHCO3 (100 mL). The organic layer
was collected, dried with Na2SO4, and filtered, and the solvents
were evaporated under reduced pressure. The residue was chroma-
tographed on a column of silica gel (20 g) with CHCl3/MeOH
(100:1, v/v) to give 2d(β) as a foam (798 mg, 56%). 1H NMR
(270 MHz, CDCl3): δ ϭ 1.15Ϫ1.23 (m, 6 H), 1.96 (s, 3 H),
2.20Ϫ2.71 (m, 2 H), 2.43 (s, 3 H), 3.46Ϫ3.87 (m, 4 H), 4.51 (m, 1
H), 4.83Ϫ4.86 (m, 2 H), 5.49 (m, 1 H), 6.40 (dd, J ϭ 8.6, 5.6 Hz,
1 H), 7.25Ϫ7.95 (m, 5 H), 9.33 (br., 1 H). 13C NMR (68 MHz,
CDCl3): δ ϭ 11.9, 12.6, 13.4, 21.8, 37.6, 43.3, 48.1, 69.8, 74.2, 82.3,
85.0, 111.5, 126.1, 129.1, 129.7, 134.5, 144.3, 150.2, 163.5, 165.7,
186.4. C23H29N3O6S (475.6): C 58.09, H 6.15, N 8.84, S 6.74; found
C 57.61, H 6.08, N 8.82, S 6.69.
5Ј-O-(N,N-Diethylthiocarbamoyl)-3Ј-O-toluoyl-α-thymidine [2d(α)]:
1H NMR (270 MHz, CDCl3): δ ϭ 1.20Ϫ1.29 (m, 6 H), 1.88 (s, 3
H), 2.41Ϫ2.93 (m, 5 H), 3.51Ϫ3.88 (m, 4 H), 4.53Ϫ4.73 (m, 2 H),
4.85 (m, 1 H), 5.58 (m, 1 H), 6.33 (dd, J ϭ 7.1, 1.8 Hz, 1 H),
7.21Ϫ7.80 (m, 5 H), 9.08 (br., 1 H). 13C NMR (68 MHz, CDCl3):
δ ϭ 11.9, 12.7, 13.5, 21.8, 38.5, 43.6, 48.2, 69.3, 74.4, 85.0, 86.9,
110.2, 126.0, 129.2, 129.4, 135.2, 144.6, 150.1, 163.7, 165.2, 186.3.
C23H29N3O6S (475.6): C 58.09, H 6.15, N 8.84, S 6.74; found C
57.75, H 6.48, N 8.51, S 6.50.
5Ј-O-Pixyl-α-thymidine [9(α)]: Sodium methoxide (177 mg,
3.28 mmol) was added to a solution of an 89:11 mixture of 5Ј-O-
(2-ethylbutyryl)-3Ј-O-toluoyl-α-thymidine [2b(α)] and 5Ј-O-(2-
ethylbutyryl)-3Ј-O-toluoylthymidine [2b(β)] (301 mg of the α/β mix-
ture, 656 µmol) in methanol (20 mL). After having been stirred at
room temperature for 1 d, the mixture was quenched by addition
of hydrochloric acid in water (1 , 3.3 mL). The mixture was parti-
tioned between CHCl3 (20 mL) and water (20 mL). The water layer
was collected and washed with CHCl3 (20 mL), and the solvents
were evaporated under reduced pressure. The residue was rendered
anhydrous by repeated coevaporation and finally dissolved in dry
pyridine (3.3 mL). 9-Chloro-9-phenylxanthene (288 mg, 984 µmol)
was added to the solution. After stirring had been continued for
an additional 3 h, the resulting mixture was partitioned between
CHCl3 (100 mL) and 5% NaHCO3 (50 mL). The organic layer was
collected, dried with Na2SO4, and filtered, and the solvents were
evaporated under reduced pressure to give an 86:14 mixture of 9(α)
and 9(β) (268 mg, 82%). This mixture was chromatographed four
times on a column of silica gel (15 g) with CHCl3/MeOH (100:1,
Synthesis of α-Nucleosides by β/α Epimerization: A mixture of an
appropriate thymidine derivative (2 mmol) and hexamethyldisilaz-
ane (HMDS) (8.4 mL, 40 mmol) was refluxed in toluene (8 mL) for v/v) containing 1% pyridine, to give 9(α) as a foam [196 mg, 73%
92 Eur. J. Org. Chem. 2002, 87Ϫ93

C1Ј Epimerization of β-Thymidine to α-Thymidine
FULL PAPER
O. Zelphati, J.-L. Imbach,
[6b]
ids Res. 1992, 20, 1193Ϫ1200.
from the mixture of 9(α)/9(β), α-purity Ͼ 99.5%], m.p. 133 °C (crys-
tallization from methanol). 1H NMR (270 MHz, CDCl3): δ ϭ 1.88
(s, 3 H), 2.18Ϫ2.80 (m, 2 H), 2.91Ϫ3.06 (m, 2 H), 4.36Ϫ4.42 (m,
2 H), 6.11 (dd, J ϭ 7.7, 2.1 Hz, 1 H), 7.01Ϫ7.45 (m, 14 H), 8.69
(br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 12.5, 40.9, 63.9, 72.4,
76.0, 77.2, 88.3, 88.7, 109.0, 116.3, 122.3, 122.5, 123.6, 126.2, 126.7,
127.8, 128.9, 129.2, 129.3, 137.5, 148.3, 150.7, 151.0, 151.1, 164.4.
C29H26N2O6 (498.5): C 69.87, H 5.26, N 5.62; found C 69.43, H
5.08, N 5.64.
N. Signoret, G. Zon, B. Rayner, Nucleic Acids Res. 1994, 22,
[6c]
4307Ϫ4314.
S. Peyrottes, J.-J. Vasseur, J.-L. Imbach, B.
Rayner, Tetrahedron Lett. 1996, 37, 5869Ϫ5872. [6d] A. Laurent,
F. Debart, J.-C. Bologna, J.-J. Vasseur, B. Rayner, Nucleosides
Nucleotides 1998, 17, 1645Ϫ1649. [6e] A. Laurent, M. Naval, F.
Debart, J.-J. Vasseur, B. Rayner, Nucleic Acids Res. 1999, 27,
[6f]
4151Ϫ4159.
A. Laurent, M. Naval, F. Debart, J.-J. Vasseur,
B. Rayner, Nucleosides Nucleotides 1999, 18, 1629Ϫ1630. [6g] K.
Pongracz, S. M. Gryaznov, Nucleic Acids Res. 1998, 26,
[6h]
1099Ϫ1106.
F. Debart, A. Meyer, J.-J. Vasseur, B. Rayner,
5Ј-O-Pixylthymidine [9(β)]:[27] 1H NMR (270 MHz, CDCl3): δ ϭ
1.67 (s, 3 H), 2.25Ϫ2.49 (m, 2 H), 3.12Ϫ3.28 (m, 2 H), 4.00 (m, 1
H), 4.43 (m, 1 H), 6.37 (dd, J ϭ 7.6, 5.9 Hz, 1 H), 7.01Ϫ7.62 (m,
14 H), 8.82 (br., 1 H). 13C NMR (68 MHz, CDCl3): δ ϭ 12.2, 41.2,
63.5, 72.7, 85.0, 86.0, 111.1, 116.5, 116.5, 122.0, 122.4, 123.6, 123.7,
126.2, 127.0, 128.0, 128.8, 129.2, 129.4, 129.6, 135.4, 147.9, 150.2,
151.1, 151.4, 163.5. C29H26N2O6·0.5H2O (507.5): C 68.63, H 5.36,
N 5.52; found C 68.60, H 5.13, N 5.54.
Nucleic Acids Res. 1998, 26, 4551Ϫ4556.
E. Uhlmann, A. Peyman, Chem. Rev. 1990, 90, 543Ϫ584; and
references cited therein
L. B. Townsend (Ed.), in: Chemistry of Nucleosides and Nucle-
otides, Plenum Press, New York, 1994, vol. 3.
H. Aoyama, Bull. Chem. Soc. Jpn. 1987, 60, 2073Ϫ2077.
H. Sugimura, K. Sujino, K. Osumi, Nucleic Acids Symp. Ser.
1992, 27, 111Ϫ112.
S. Janardhanam, K. P. Nambiar, Tetrahedron Lett. 1994, 35,
3657Ϫ3660.
K. Shinozuka, N. Yamada, A. Nakamura, H. Ozaki, H. Sawai,
Bioorg. Med. Chem. Lett. 1996, 6, 1843Ϫ1848.
E. J. Michael, C. Claire, I. K. Saeed, Nucleosides Nucleotides
1998, 17, 2383Ϫ2387.
H. Sawai, A. Nakamura, H. Hayashi, K. Shinozuka, Nucleos-
ides Nucleotides 1994, 13, 1647Ϫ1654.
K. Shoda, T. Wada, M. Sekine, Nucleosides Nucleotides 1998,
17, 2199Ϫ2210.
M. Sekine, ‘‘N-glycosylation’’ in: Glycoscience Ϫ Chemistry and
Chemical Biology, chapter 3.6, Medio, Berlin, 2001, pp.
673Ϫ690.
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
Acknowledgments
This work was supported by a Grant from ‘‘Research for the Fu-
ture’’ Program of the Japan Society for the Promotion of Science
(JSPS-RFTF97I00301) and a Grant-in-Aid for Scientific Research
from the Ministry of Education, Culture, Sports, Science, and Tech-
nology, Japan.
[1]
F. Morvan, B. Rayner, J.-L. Imbach, D.-K. Chang, J. W. Lown,
Nucleic Acids Res. 1986, 14, 5019Ϫ5035.
[17]
[18]
[19]
[20]
[2] [2a]
H. Vorbrüggen, K. Krolikiewicz, B. Bennua, Chem. Ber. 1981,
F. Morvan, B. Rayner, J.-L. Imbach, S. Thenet, J.-R. Ber-
114, 1234Ϫ1255.
trand, J. Paoletti, C. Malvy, C. Paoletti, Nucleic Acids Res.
[2b]
T. Yamaguchi, M. Saneyoshi, Chem. Pharm. Bull. 1984, 32,
1441Ϫ1450.
W. Wierenga, H. I. Skulnick, Carbohydrate Reserch 1981, 90,
1987, 15, 3421Ϫ3437.
C. Cazenova, M. Chevrier, N. T.
´ `
Thuong, C. Helene, Nucleic Acids Res. 1987, 15, 10507Ϫ10521.
[2c]
F. Morvan, B. Rayner, J.-L. Imbach, M. Lee, J. A. Hartley,
41Ϫ52.
D.-K. Chang, J. W. Lown, Nucleic Acids Res. 1987, 15,
T. Mukaiyama, T. Ishikawa, H. Uchino, Chem. Lett. 1997,
389Ϫ390.
M. Hoffer, Chem. Ber. 1960, 93, 2777Ϫ2781.
M. D. Matteucci, M. H. Caruthers, Tetrahedron Lett. 1980,
21, 3243Ϫ3246.
K. K. Ogilvie, Can. J. Chem. 1973, 51, 3799Ϫ3807.
M. P. Kotick, C. Szantay, T. J. Bardos, J. Org. Chem. 1969,
34, 3806Ϫ3813.
7027Ϫ7044.
[3]
C. Gagnor, B. Rayner, J.-P. Leonetti, J.-L. Imbach, B. Lebleu,
Nucleic Acids Res. 1989, 17, 5107Ϫ5114.
[21]
[22]
[4] [4a]
J. S. Sun, C. Giovannangeli, J. C. Francois, R. Kurfurst, T.
M. Garestier, U. Asseline, T. S. Behmoaras, N. T. Thuong, C.
[23]
[24]
[4b]
´ `
Helene, Proc. Natl. Acad. Sci. USA 1991, 88, 6023Ϫ6027.
F. Morvan, H. Porumb, G. Degols, I. Lefebvre, A. Pompon, B.
S. Sproat, B. Rayner, C. Malvy, B. Lebleu, J.-L. Imbach, J.
[25]
[26]
[27]
[4c]
N. S. Isaacs, in: Physical Organic Chemistry, 2nd. ed., Longman
Scientific & Technical,New York, 1987, pp. 319Ϫ368.
H. G. Howell, P. R. Brodfuehrer, S. P. Brundidge, D. A. Be-
nigni, C. J. Sapino, J. Org. Chem. 1988, 53, 85Ϫ88.
J. B. Chattopadhyaya, C. B. Reese, J. Chem. Soc., Chem. Com-
mun. 1978, 15, 639Ϫ640.
Med. Chem. 1993, 36, 280Ϫ287.
J. Liquier, R. Letellier, C.
Dagneaux, M. Ouali, F. Morvan, B. Raynier, J.-L. Imbach, E.
Taillandier, Biochemistry 1993, 32, 10591Ϫ10598.
[5]
[5a] E. Bloch, M. Lavignon, J.-R. Bertrand, F. Pognan, F. Mor-
gan, C. Malvy, B. Rayner, J.-L. Imbach, C. Paoletti, Gene 1988,
[5b]
72, 349Ϫ360.
C. Gagnor, B. Rayner, J.-P. Leonetti, J.-L.
Imbach, B. Lebleu, Nucleic Acids Res. 1989, 17, 5107Ϫ5114.
Received August 6, 2001
[6] [6a]
F. Debart, B. Rayner, G. Degols, J.-L. Imbach, Nucleic Ac-
[O01388]
Eur. J. Org. Chem. 2002, 87Ϫ93
93