Practical synthesis of (2′ R)-2′-deoxy-2′-C-methyluridine by highly diastereoselective homogeneous hydrogenation

Article abstract of DOI:10.1021/jo101822j

Diastereoselective hydrogenation of 2′-deoxy-2′-exo- methyleneuridine was carried out under homogeneous conditions using a low loading of a chiral Rh catalyst. This, coupled with improvements in the synthesis of the substrate, allowed the smooth pilot plant preparation of the title compound on >10 kg scale.

Full text of DOI:10.1021/jo101822j

pubs.acs.org/joc
Practical Synthesis of (20R)-20-Deoxy-20-
C-methyluridine by Highly Diastereoselective
Homogeneous Hydrogenation
Sebastien Lemaire,*,† Ioannis Houpis,†
ꢀ
Rainer Wechselberger,† Jaak Langens,†
Wim A. A. Vermeulen,† Nico Smets,† Ulrike Nettekoven,‡
Youchu Wang,§ Tingting Xiao,§ Haisheng Qu,§
Renmao Liu,§ Tim H.M. Jonckers, Pierre Raboisson,
Koen Vandyck, Karl M. Nilsson,^ and Vittorio Farina†
FIGURE 1. Key compounds described in this study.
compound,3 and its derivatives have been widely incorpo-
rated into modified nucleic acids,4 no practical synthesis of
this interesting building block has been reported, and even its
physicochemical properties are not fully described. Uridine
(2), in view of its low cost and ready availability, represents
an ideal starting point for a practical synthesis of 1. Although
1 is formally the result of a simple substitution of the 20-OH
group by a methyl group with inversion of configuration,
we deemed that replacement of a suitable leaving group at
C-20 with a methyl organometallic species would be thwarted
by the well-known ready participation of the uracil C-2
carbonyl,5 and we therefore settled on the development of
a face-selective hydrogenation of the known 20-exomethy-
lene derivative 3. Such hydrogenation has been reported to
proceed with modest selectivity using heterogeneous Pd
catalysts (e.g., β/R = 3:1 with Pd on CaCO3).3
†Johnson and Johnson Pharmaceutical Development,
Turnhoutseweg 30, B-2340 Beerse, Belgium, ‡Solvias AG,
Mattenstrasse 22, CH-4002 Basel, Switzerland,
§WuXi PharmaTech Co., Ltd., 288 Fute Zhong Road,
Waigaoqiao Free Trade Zone, Shanghai 200131, PR China,
Tibotec BVBA, Medicinal Chemistry, Turnhoutseweg 30,
B-2340 Beerse, Belgium, and ^Medivir AB, PO Box 1086,
SE-141 22 Huddinge, Sweden
slemaire@its.jnj.com
Received September 17, 2010
The selected route is shown in Scheme 1 and is based on the
well-established protecting group strategy of Robins.6 The
TIPDS protecting group is ideal for selective protection of
the 30 and 50 hydroxyl groups in nucleosides and leaves the 20
hydroxyl largely unprotected. The original conditions, which
employ pyridine as solvent, were inconvenient due to the
high boiling point of pyridine, its toxicity, and its tendency to
inhibit the next (oxidation) step (vide infra), even when
present in small amounts. The silylation reaction was there-
fore carried out in dichloromethane using 4 equiv of imid-
azole as the base. This protection reaction is not quantitative,
and LCMS data indicate the presence of eight impurities,
each with intensity >0.2%. While the detailed analysis of the
reaction crude is outside the scope of this note, three general
observations are important: (1) A large excess of silylating
agent must be avoided, because the 20-position is readily
silylated as well. It was found convenient to employ ca. 1.1
equiv. (2) The medium has to be anhydrous. It was found
most practical to simply azeotropically dry the solution of
uridine and imidazole before addition of the silylating agent.
(3) The concentration cannot be increased beyond the one
Diastereoselective hydrogenation of 20-deoxy-20-exo-methy-
leneuridine was carried out under homogeneous conditions
using a low loading of a chiral Rh catalyst. This, coupled
with improvements in the synthesis of the substrate, allowed
the smooth pilot plant preparation of the title compound
on >10 kg scale.
Modified nucleosides continue to hold a central position
in the chemotherapy of cancer and viral diseases,1 and in
particular, 20-modified nucleosides play an important role in
research aimed at understanding the detailed molecular basis
of RNA function.2
In connection with a program in the antiviral area, we
needed to prepare multikilo amounts of (20R)-20-deoxy-20-C-
methyluridine (1) (Figure 1). Although the latter is a known
(4) (a) Li, N.-S.; Piccirilli, J. A. J. Org. Chem. 2003, 68, 6799. (b) Cicero,
D. O.; Gallo, M.; Neuner, P. J.; Iribarren, A. M. Tetrahedron 2001, 57, 7613.
(c) Matsuda, A.; Takenuki, K.; Sasaki, T.; Ueda, T. J. Med. Chem. 1991, 34,
234. (d) Ioannidis, P.; Classon, B.; Samuelsson, B.; Kvarnstrom, I. Nucleo-
sides Nucleotides 1992, 11, 1205. (d) Sugimura, H.; Osumi, K.; Yamazaki, T.;
Yamaya, T. Tetrahedron Lett. 1991, 32, 1809. (e) De Mesmaeker, A.;
Lebreton, J.; Hoffmann, P.; Freier, S. M. Synlett 1993, 677. (f) Awano, H.;
Shuto, S.; Miyashita, T.; Ashida, N.; Machida, H.; Kira, T.; Shigeta, S.;
Matsuda, A. Arch. Pharm. Pharm. Med. Chem. 1996, 329, 66.
(5) Marton-Mersz, M.; Kuszmann, J.; Pelczer, I.; Parkanyi, L.; Koritsansky;
Kalman, A. Tetrahedron 1983, 39, 275.
(1) (a) Simons, C.; Wu, Q.; Htar, T. T. Curr. Top. Med. Chem. 2005, 5,
1191. (b) Parker, W. B.; Secrist, J. A.; Waud, W. R. Curr. Opin. Invest. Drugs
2004, 5, 592.
(2) (a) Gesteland, R. F.; Cech, T. R.; Atkins, J. F. The RNA World, 2nd ed.;
Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 1999. (b)
Saenger, W. Principles of Nucleic Acid Structure; Springer Verlag: Berlin, 1983.
(3) Cicero, D. O.; Neuner, P. J. S.; Franzese, O.; D’Onofrio, C.; Iribarren,
A. M. Bioorg. Med. Chem. Lett. 1994, 4, 861.
(6) Robins, M. J.; Wilson, J. S.; Hansske, F. J. Am. Chem. Soc. 1983, 105,
4059.
DOI: 10.1021/jo101822j
r
Published on Web 12/06/2010
J. Org. Chem. 2011, 76, 297–300 297
2010 American Chemical Society

Lemaire et al.
JOCNote
SCHEME 1. Practical Synthesis of 1 from Uridine (2)
FIGURE 2. Relevant impurities and intermediates.
inactive hydroxylamine and the active oxammonium species,
which PhI(OAc)2 is unable to produce.10 (3) The product
displayed a hitherto unreported tendency to form a stable
gem-diol (7, Figure 2), both in solution and in the isolated
product if drying was incomplete. This was shown by 1H and
13C NMR data in the presence of traces of water and by
LCMS. In particular, in the 13C NMR spectrum the reso-
nance for the C-20 carbonyl group at δ 204.7 (CDCl3)
decreased upon saturation with D2O, with the concomitant
formation of a new resonance at δ 99.1 (20-gem-diol C).11 The
tendency to form the hydrate was strong even under our
reversed-phase HPLC monitoring conditions, and therefore,
a special normal-phase method had to be developed (silica
gel, EtOH/heptane). However, the distillation prior to crys-
tallization removed the water of hydration, and the product
was obtained as a sharp-melting, crystalline white powder.
The Wittig reaction has been reported to occur in 51-92%
yield using a base like the dimsyl anion in DMSO or sodium
tert-butoxide in ether/benzene.12 On a large scale, a major
problem is, of course, the removal of the triphenylphosphine
oxide (TPPO) formed in the reaction on a stoichiometric
basis. We also found that traces of TPPO inhibited our
hydrogenation step (vide infra) and a specification for TPPO
content in 3 had to be set at e1 mol %. Our preferred
conditions for the reaction made use of tert-amylate as base
(a 25% w/w solution in toluene) and toluene as the sole
solvent. We found that the [2 þ 2] cycloaddition took place
immediately at rt (as shown by disappearance of starting
material in the HPLC) to yield an intermediate that was
readily detected by HPLC and gave the characteristic NMR
pattern of an oxaphosphetane (8, 13P: δ -66.6). This re-
quired prolonged warming to yield 6 (5 h at 35 ꢀC). This
phenomenon is unusual but precedented in related systems
described by Robins.12b The yield in solution, determined by
HPLC assay, was 85-90%. Removal of TPPO entailed a
solvent switch to heptane, from which the bulk of TPPO
crystallizes, and filtration of the supernatant through a silica
gel pad. The ensuing heptane solution contained 3 in
79-85% yield on a 0.1-10 kg scale. The TPPO content at
this stage was 5-8 mol %, which we knew could be sub-
stantially reduced (to below 1%) in the next step. It was
possible to isolate the product, if desired, by crystallization
from MeOH/water, but on plant scale it was found much
more convenient to take this heptane solution directly into
the deprotection step. Desilylation with fluoride presented
workup problems because of the aqueous solubility of 3. It
was found most practical to carry out the deprotection using
Reaction conditions: (i) TPDSCl2, imid, CH2Cl2, rt; (ii) TEMPO
(0.2 equiv), PhI(OAc)2, CH2Cl2, 15 ꢀC; (iii) Ph3PCH3Br, t-BuOK, THF,
50 ꢀC; (iv) HCl, MeOH, rt; (v) [Rh(COD)Cl]2, MeOH, ligand 10, 50
bar H2.
reported here, under batch conditions, without producing
substantial amounts of a variety of double condensation pro-
ducts (i.e., compounds containing 1 equiv of TIPDS and 2 equiv
of uridine), and the reaction temperature should be kept at or
below 20 ꢀC. With these modifications, the reaction gave up to
88% yield (by in situ HPLC assay), a considerable improvement
over the 65-70% obtained with the original method. The
organic layer was simply washed with water and used for the
next step. A sample was chromatographed to yield pure 4,which
was a foam and retained much solvent: an analytical standard of
high purity was therefore difficult to produce. The assay yield of
the organic solution, under these imperfect conditions, was
consistently 85-88.5% on a scale from 100 g up to10 kg.
Although the oxidation of 4 to 5 is known, the oxidizing
agents used are either based on extremely toxic Cr(VI)7 or on
hazardous Dess-Martin periodinane.8 Therefore, we devel-
oped a more scalable procedure, which makes use of catalytic
TEMPO and stoichiometric oxidant PhI(OAc)2.9 This reac-
tion was conveniently carried out using the dichloromethane
solution from step 1, after a water wash. At the end of the
oxidation, simple solvent switch to heptane and cooling
produced crystalline 5 in 90-91% yield (76% isolated over
two steps) on 100 g scale.10 The material was >99% pure by
HPLC (area %). A few important experimental considera-
tions apply to this reaction as well: (1) The reaction was run
at 15 ꢀC with 0.2 equiv of TEMPO. Lower temperatures or
lower catalyst loads led to exceedingly long reaction times,
whereas higher temperature led to premature catalyst loss.
Under the optimized conditions the reaction took 16-24 h.
(2) The reaction displayed sigmoidal kinetics. Reasoning
that the 2 equiv of AcOH produced in the reaction may
catalyze the reaction, AcOH was introduced in stoichio-
metric amounts from the beginning, dramatically reducing
the initiation phase. This result is due to the catalytic effect of
acids in promoting the disproportionation of TEMPO to
(7) Hansske, F.; Madej, D.; Robins, M. J. Tetrahedron 1984, 40, 125.
(8) Samano, V.; Robins, M. J. Synthesis 1991, 283.
(9) De Mico, A.; Margarita, R.; Parlanti, L.; Vescovi, A.; Piancatelli, G.
J. Org. Chem. 1997, 62, 6974.
(11) Upon saturation of the solvent with D2O, 7 constituted 40% of the
7 þ 5 mixture in CDCl3, 51% in CD3CN, and 68% in THF-d8.
(12) (a) Lin, T.-S.; Luo, M.-Z.; Liu, M.-C.; Clarke-Katzenburg, R. H.;
Cheng, Y.-C.; Prusoff, W. H.; Mancini, W. R.; Birnbaum, G. I.; Gabe, E. J.;
Giziewicz, J. J. Med. Chem. 1991, 34, 2607. (b) Samano, V.; Robins, M. J.
Synthesis 1991, 283.
(10) The yield for the two steps averaged 74% on a 10 kg scale. The
reaction was inhibited by base, including imidazole, and promoted by water
and AcOH. In spite of the thorough attempts to identify all variables in this
reaction, on scale the reaction displayed a variable reaction time, and up to
48 h and addition of more reagent were employed in one case.
298 J. Org. Chem. Vol. 76, No. 1, 2011

Lemaire et al.
JOCNote
TABLE 1. Selected Screen Results (eq 1)
The optimized hydrogenation was carried out with an s/c
ratio of 800 in degassed MeOH at 40 bar and 40 ꢀC over
12-16 h. On a 6-7 kg scale, upon increasing the concentra-
tion to 25% w/v, the de dropped slightly from 97 to 98% in
the lab to 93-95% in a 50 L stainless steel autoclave. After
removal of the catalyst by pad filtration, isolation of the final
product could be carried out, with some losses, by recrys-
tallization from THF-acetone (71%) with 96.4% purity and
94.02% de. In conclusion, we have described the first prac-
tical diastereoselective and chromatography-free synthesis
of 1, and we have scaled the protocol to produce tens of
kilograms of this important intermediate. Our protocol
should facilitate the application of this modified uridine
analogue to biological studies and drug design.
entry
metal used (4 mol %)
ligand (4 mol %)
% de (HPLC)
1
2
3
4
5
6
[Rh(nbd)2]BF4
[RuCl2(p-cymene)]2
[Rh (COD)]BF4
[Rh(nbd)2]BF4
[Rh(nbd)2]BF4
[Rh(nbd)2]BF4
10
11
dipf, 12
ent-10
dppf
13
98.1
96.6
91.2
78.7
64.0
84.8
Experimental Section
1-((6aR,8R,9aR)-2,2,4,4-Tetraisopropyl-9-oxotetrahydro-6H-furo-
[3,2-f ][1,3,5,2,4]trioxadisilocin-8-yl)pyrimidine-2,4(1H,3H)-dione (5).
In a 2 L four-necked flask equipped with temperature probe,
addition funnel, overhead stirrer, and distillation head was charged
uridine (100.1 g, 0.410 mol), followed by imidazole (111.4 g, 1.110
mol) and anhydrous dichloromethane (1.0 L); the solution was
stirred and distilled at atmospheric pressure to a volume of 750 mL
(KF water analysis indicated <0.05%). TIPDS dichloride (141.9 g,
0.450 mol) was added dropwise as a solution in CH2Cl2 (200 mL)
over 40 min while being cooled to 10-15 ꢀC (internal temperature),
and then the mixture was stirred at this temperature for 3 h. Water
was added (1 L), and the two-phase mixture was separated. The
organic layer was diluted to 1.4 L, assayed by HPLC (content: 169 g
of 4), and used directly for the oxidation reaction as follows: in a 2 L
four-necked flask equipped with temperature probe, addition
funnel, overhead stirrer, and nitrogen inlet was introduced the
solution of 4, which was then cooled to 15 ꢀC and then treated
with bis(acetoxy)iodobenzene (145 g, 0.451 mol), acetic acid (20.8 g,
0.347 mol), and with medium stirring, a solution of TEMPO (10.8 g,
0.069 mol, 0.2 equiv) in DCM (40 mL) over 10 min. The mixture
was stirred at 13-18 ꢀC until the starting material was consumed,
typically 18-24 h. After the reaction was complete, aqueous sodium
thiosulfate was added (1 N, 850 mL) with stirring for 10 min. The
phases were separated, and the organic phase was washed with
water (1 L). The organic phase was concentrated to 550 mL, and
then heptane (700 mL) was added, and the mixture was distilled at
reduced pressure (internal T < 50 ꢀC) until no more dichloro-
methane was left; the residue was cooled in ice and stirred for 3 h.
The copious precipitate was filtered and washed with cold heptane
(2 ꢀ 200 mL). The cake was dried in vacuo at 45 ꢀC for 14 h to yield
5 as a white powder (152 g, 76.5% from uridine). Mp: 187.0 ꢀC dec.
1H NMR (CDCl3, 400 MHz): δ 8.70 (br s, 1H), 7.16 (d, J = 8.0 Hz,
1H), 5.76 (dd, J = 8.0 Hz, J0 = 1.5 Hz, 1H), 5.05 (d, J = 9.2 Hz,
1H), 5.00 (s, 1H), 4.20-4.10 (m, 2H), 3.94 (m, 1H), 1.20-0.90 (m,
28H). HRMS: calcd for C21H36N2O7Si2 [M þ H]þ 485.2134, found
485.2131. HRMS for 7: calcd for C21H38N2O8Si2 [M þ H]þ
503.2239, found 503.2242.
FIGURE 3. Representative ligands screened.
2 M aqueous HCl in MeOH at 35-40 ꢀC for 7 h. Solvent
switch in vacuo to acetonitrile and cooling yielded white
crystals of 3 in 84-87% isolated yield over two steps. The
material was >99.5% pure by HPLC assay. It is important
to carry out the crystallization slowly and to determine the
content of the cleaved disilanol in the product. A specifica-
tion of 0.5 mol % had to be set on this impurity because it
poisoned the homogeneous catalyst used in the final step.13
With a practical four-step, two-isolation synthesis of 3 in
hand, we were ready to attempt the unprecedented face-
selective hydrogenation of 3 (eq 1).14 A 96-plate screen was
carried out using three metal salts: Rh(I), Ru(II), and Ir(I) in
different alcoholic solvents using a battery of 43 phosphorus
ligands, both achiral and chiral. The s/c ratio was set at 25
with T at 40 ꢀC and the hydrogen pressure at 40 bar. Whereas
Ir(I) catalysts were essentially inert under these conditions,
excellent turnover was obtained with several Rh(I) and Ru-
(II) salts (Table 1 and Figure 3).
The most diastereoselective achiral ligand was dipf (12), in
conjunction with Rh(I) sources. However, chiral ligand 10
gave the best de in this preliminary screen, and this was a
“matched” effect, as evidenced by the lower de obtained for
ent-10 (Table 1, entry 4).15 Experiments 1-3 were repeated as
individual runs at a more acceptable s/c ratio of 100, and the
Ru(II) catalysts proved more sluggish than the Rh(I). For all
these reasons, it was decided to optimize and scale the
reaction in entry 1.
Intermediate 4 can be isolated by silica gel chromatography
(methanol-dichloromethane 2:98) as a foam and characterized.
1HNMR(CDCl3, 400 MHz) δ10.04 (br s, 1H), 7.81 (d, J=8.1Hz,
1H), 5.74, (s, 1H), 5.70 (d, J= 8.1 Hz, 1H), 4.30-4.17 (m, 4H), 3.99
(dd, J=13.1 Hz, J0 =2.1 Hz, 1H), 3.68 (br s, 1H), 1.20-0.95 (m,
28H). HRMS: calcd for C21H39N2O7Si2 [M þ H]þ 487.2296, found
487.2322.
1-((2R,4S,5R)-4-Hydroxy-5-(hydroxymethyl)-3-methylenete-
trahydrofuran-2-yl)pyrimidine-2,4(1H,3H )-dione (3). In a 3 L
three-necked round-bottom flask equipped with temperature
probe, overhead stirrer, and addition funnel was suspended
methyltriphenylphosphonium bromide (147 g, 0.412 mol) in
anhydrous toluene (1.0 L). To this suspension was added
(13) 1H NMR is quite sufficient to determine the level of silanol-related
contaminants, given the high intensity of the isopropyl peaks.
(14) In preliminary tests, hydrogenation of 6 was sluggish and less
diastereoselective.
(15) A full description of the screening results can be found in the
Supporting Information.
J. Org. Chem. Vol. 76, No. 1, 2011 299

Lemaire et al.
JOCNote
1
potassium tert-amylate (25% w/w in toluene, 209 g, 0.412 mol)
over 20 min at rt (slight exotherm). A solution of 5 (100 g, 0.206
mol) in anhydrous THF (250 mL) was added over 25 min at rt
(slight exotherm). The mixture was then heated to 35 ꢀC for 5 h
and cooled in ice, and acetic acid (8.7, 0.144 mol) was added in
one lot as a quench. Water (620 mL) was added and the brown
organic layer separated. The water wash was repeated, and then
the organic layer was distilled under reduced pressure to 300 mL
(T<35-40 ꢀC). Upon cooling, the solution was seeded with
triphenylphosphine oxide, and the mixture was stirred for 2-3 h
with ice cooling until a suspension formed. Then heptane (800 mL)
was added dropwise over 2 h, and the suspension was stirred for a
further 3 h under ice cooling. Filtration was followed by washing
the TPPO with heptane (100 mL). The filtrate was treated with ethyl
acetate (400 mL), and the solution was filtered through a silica pad
(200 g), washing the pad with a mixture of ethyl acetate (800 mL)
and heptane (1.7 L). The filtrate was concentrated in vacuo to
300 mL (T < 45 ꢀC). Methanol (350 mL) was added, and the
solution was again evaporated to 300 mL. After the procedure was
repeated once more, the methanol solution was assayed by HPLC
and found to contain 84.7 g (85%) of 6. The product could be
isolated by crystallization from MeOH/water (2/1 v/v). Mp: 130 ꢀC.
1H NMR (CDCl3, 400 MHz): δ 8.94 (br s, 1H), 7.45 (d, J = 8.0 Hz,
1H), 6.52 (d, J = 1.5 Hz, 1H), 5.71 (dd, J = 8.0 Hz, J0 = 1.8 Hz,
1H), 5.55 (d, J = 1.5 Hz, 1H), 5.46 (d, J = 1.5 Hz, 1H), 4.82 (m,
1H), 4.15 (dd, J = 13.2 Hz, J0 = 2.2 Hz, 1H), 4.05 (dd, J = 13.2,
J0 = 2.7 Hz, 1H), 3.70 (m, 1H), 1.25-0.90 (m, 28H). HRMS: calcd
for C22H38N2O6Si2 [M þ H]þ 483.2341, found 483.2346.
Yield: 42.2 g (85%). Mp: 173.5 ꢀC dec. H NMR: (DMSO-d6,
400 MHz) δ 11.38 (exch s, 1H), 7.53 (d, J = 8.0 Hz, 1H), 6.45 (d,
J = 1.8 Hz, 1H), 5.68-5.61 (m, 2H), 5.39 (d, J = 2.2 Hz, 1H),
5.26 (d, J = 2.2 Hz, 1 H), 4.96 (exch s, 1H), 4.47 (exch s, 1H),
3.67-3.57 (m, 3H). 13C NMR (DMSO-d6) δ 161.7, 149.4, 148.5,
140.2, 110.6, 101.0, 83.4, 82.2, 68.5, 59.2. HRMS: calcd for
C10H12N2O5 [M þ K]þ 279.0378, found 279.0374.
1-((2R,3S,4S,5R)-4-Hydroxy-5-(hydroxymethyl)-3-methyl tetrahy-
drofuran-2-yl)pyrimidine-2,4(1H,3H )-dione (1). A degassed solu-
tion of 3 (6.00 g, 25.0 mmol) in MeOH (16 mL) was placed in a
50 mL autoclave, which was purged with argon. [Rh(nbd)2]BF4
(11.7 mg, 0.0312 mmol) and ligand 10 (32.3 mg, 0.0343 mmol) were
added as a solution in dry, degassed MeOH (16 mL) from a
Schlenk tube. The autoclave was purged again with three cycles of
argon and then three times with 10 bar of hydrogen. The mixture
was stirred and heated at 40 ꢀC under 40 bar hydrogen for 16-18 h.
The reaction was allowed to cool down and then filtered through a
small pad of silica. The filtrate was evaporated to dryness and
recrystallized from hot acetone/THF (ca. 20:1, total 20-25 mL),
then dried at 40 ꢀC overnight. Yield: 4.3 g (71%). Mp: 138-141 ꢀC.
1H NMR (DMSO-d6, 400 MHz): δ 11.30 (exch. s, 1H), 7.97 (d,
J= 8.1 Hz, 1H), 6.08 (d, J= 7.7 Hz, 1H), 5.60 (d, J=8.1Hz, 1H),
5.37 (exch br d, 1H), 5.14 (exch br t, 1 H), 3.73-3.59 (m, 4H), 2.40
(m, 1H), 0.81 (d, J = 6.9 Hz, 3H). 13C NMR (DMSO-d6): δ 161.7,
149.2, 139.7, 99.6, 84.0, 83.4, 70.9, 57.5, 43.2, 9.8. HRMS: calcd for
C10H15N2O5 [M þ H]þ 243.0975, found 243.0973.
The material still contains 2-3 mol % of the epimer 9, which
can be separated by silica gel chromatography (ethyl acetate/
heptane) and characterized: 1H NMR (DMSO-d6, 400 MHz): δ
11.32 (exch. s, 1H), 7.85 (d, J = 8.1 Hz, 1H), 5.81 (d, J = 9.1 Hz,
1H), 5.68 (d, J = 8.1 Hz, 1H), 5.22 (exch br d, 1H), 5.05 (exch br
t, 1 H), 4.05 (m, 1H), 3.85 (m, 1H), 3.56 (m, 2H), 2.22 (m, 1H),
0.92 (d, J = 6.5 Hz, 3H). 13C NMR (DMSO-d6) δ 161.6, 149.5,
139.1, 100.7, 86.9, 85.7, 71.2, 60.4, 40.8, 7.2. HRMS: calcd for
C10H15N2O5 [M þ H]þ 243.0975, found 243.0969.
However, it was advantageous to carry out the deprotection
directly on the crude product: the MeOH solution obtained
above, in a three-necked flask equipped with overhead stirrer,
temperature probe, and condenser, was treated with aqueous
HCl (2M, 98.2 g, 1.09 equiv) and heated to 40 ꢀC for 7 h. After
being cooled to 10 ꢀC, the solution was evaporated at reduced
pressure (T < 15 ꢀC) to a volume of 300 mL, acetonitrile (400 mL)
was added, and the solution was concentrated to 270 mL in vacuo
(T < 15 ꢀC) and then cooled and stirred for 3 h at 0 ꢀC, while the
product slowly crystallized. The precipitate was filtered off and
then slurried in methyl tert-butyl ether (300 mL) at 40 ꢀC for 2 h
(to remove residual disilanol). The mixture was allowed to cool
and then filtered. The solid was dried in vacuo at 45 ꢀC over 14 h.
Supporting Information Available: Full description of the
96-plate catalyst screen, compound characterization data, and
copies of spectra supporting the structural assignments. This
material is available free of charge via the Internet at http://
pubs.acs.org.
300 J. Org. Chem. Vol. 76, No. 1, 2011