Practical synthesis of a potent hepatitis C virus RNA replication inhibitor

Article abstract of DOI:10.1021/jo0491096

A practical, efficient synthesis of 1, a hepatitis C virus RNA replication inhibitor, is described. Starting with the inexpensive diacetone glucose, the 12-step synthesis features a novel stereoselective rearrangement to prepare the key crystalline furanose diol intermediate. This is followed by a highly selective glycosidation to couple the C-2 branched furanose epoxide with deazapurine.

Full text of DOI:10.1021/jo0491096

P r a ctica l Syn th esis of a P oten t Hep a titis C Vir u s RNA
Rep lica tion In h ibitor
Matthew M. Bio,* Feng Xu,* Marjorie Waters, J . Michael Williams, Kimberly A. Savary,
Cameron J . Cowden, Chunhua Yang, Elizabeth Buck, Zhiguo J . Song, David M. Tschaen,
R. P. Volante, Robert A. Reamer, and Edward J . J . Grabowski
Process Research, Merck Research Laboratories, P.O. Box 2000, Rahway, New J ersey 07065
matthew_bio@merck.com
Received May 26, 2004
A practical, efficient synthesis of 1, a hepatitis C virus RNA replication inhibitor, is described.
Starting with the inexpensive diacetone glucose, the 12-step synthesis features a novel stereose-
lective rearrangement to prepare the key crystalline furanose diol intermediate. This is followed
by a highly selective glycosidation to couple the C-2 branched furanose epoxide with deazapurine.
SCHEME 1
In tr od u ction
Hepatitis C virus (HCV) currently infects 3.9 million
Americans leading to 8 000-10 000 deaths annually as
a result of chronic liver disease. Furthermore, infection
by HCV is currently the leading indication for liver trans-
plants. Co-infection of HIV patients with HCV averages
about 70% with the result that HCV-related liver disease
is a major reason for hospitalization and death among
HIV-infected persons. The current HCV infection treat-
ment protocols include interferon-R either alone or in
combination with Ribavirin. While these regimens have
been somewhat effective, not all patients respond to these
drug therapies1 and greater than 10% experience mild
to moderate side effects.2 Given the current state of the
art and the fact that there is no vaccine available for the
prevention of HCV,3 it is clear that new protocols for the
treatment of HCV infection are needed.
HCV is an enveloped virus with a single-stranded,
positive-sense RNA genome. The HCV RNA encodes a
handful of life-cycle essential enzymes including an RNA
polymerase. As part of the process to replicate new viral
particles, RNA polymerase acts to assemble copied por-
tions of the viral RNA sequence into a single strand prior
to encapsulation. In light of the key role HCV polymerase
plays in the HCV life cycle, inhibition of this enzyme is
a potential therapeutic strategy for treating HCV-infected
patients. During a collaborative effort between Merck and
ISIS Pharmaceuticals, the unnatural nucleoside 1 was
identified as a potent inhibitor of HCV replication.4 To
provide the quantities of nucleoside 1 necessary for
further biological testing, an efficient and practical
synthesis of 1 was required.
Our initial foray into this chemistry began with the
synthesis of 1 via the route disclosed by the Merck-ISIS
collaboration.5 The synthesis follows the logical, conver-
gent retrosynthesis outlined in Scheme 1. Disconnection
of 2 at the anomeric center reveals the commercially
available, but prohibitively expensive, 4-chloro-7H-pyr-
rolo[2,3-d]pyrimidine 46 and 3,5-bis-O-(2,4-dichloroben-
zyl)-2-C-methyl-R-D-ribosylbromide 3. While this route
proved adequate for the small-scale preparation of 1,
providing material for further studies via this route
would present some difficulties. As such, a more practical
and efficient synthesis of 1 from readily available starting
materials was needed. In particular, we sought a syn-
thesis that would provide key intermediates as crystalline
solids to facilitate purification. We required an efficient
and stereoselective glycosidation reaction. The ultimate
route we developed achieved each of these goals.
(1) Rosen, H. R.; Gretch, D. R. Mol. Med. Today 1999, 5, 393.
(2) Centers for Disease Control and Preventions. Recommendations
for prevention and control of hepatitis C virus (HCV) and HCV-related
chronic disease. In Morbidity Mortality Weekly Rep. 1998, 47, 1.
(3) Lemon, S. M.; Thomas, D. L. New Engl. J . Med. 1997, 336, 196.
(4) Carroll, S. S.; Lafemina, R. L.; Hall, D. L.; Himmelberger, A. L.;
Kuo, L. C.; Maccoss, M.; Olsen, D. B.; Rutkowski, C. A.; Tomassini, J .
E.; An, H.; Bhat, B.; Bhat, N.; Cook, P. D.; Eldrup, A. B.; Guinosso, C.
J .; Prhavc, M.; Prakash, T. P. Patent WO 02/057425 A2, 2002.
(5) Carroll, S. S.; Maccoss, M.; Olsen, D. B.; Bhat, B.; Bhat, N.; Cook,
P. D.; Eldrup, A. B.; Prhavc, M.; Song, Q. Patent WO 02/057287 A2,
2002.
(6) Chloro-7H-pyrrolo[2,3-d]pyrimidine, also known as 6-chloro-7-
deazapurine is available from Chem Genes, 299 Homer Av., Ashland,
MA 01721. For a synthesis of 6-chloro-7-deazapurine see: Davoll, J .
J . Chem. Soc. 1960, 131.
10.1021/jo0491096 CCC: $27.50 © 2004 American Chemical Society
Published on Web 08/13/2004
J . Org. Chem. 2004, 69, 6257-6266
6257

Bio et al.
SCHEME 2a
We wish to report a novel synthesis of 1 making use
of a highly selective ester migration and a highly stereo-
selective glycosidation via a nucleophilic epoxide ring
opening. Furthermore, the synthesis of 1 is made con-
vergent by use of a protected form of 4-amino-7H-pyrrolo-
[2,3-d]pyrimidine in the glycosidation coupling. Finally,
a much-improved synthesis of the required 4-amino-7H-
pyrrolo[2,3-d]pyrimidine was developed.
Resu lts a n d Discu ssion
Glycosid a tion via Ep oxid e In ter m ed ia tes. We
initiated our studies by repeating the previously de-
scribed synthesis of 1, paying particular attention to the
key glycosidation reaction. Treating intermediate 5 with
38% HBr-AcOH in dichloromethane resulted in forma-
tion of the corresponding ribosyl bromide 3 as a 5:1
mixture of R/â-anomers, as determined by 1H NMR.
Following concentration of the reaction solution to re-
move excess HBr-AcOH below 20 °C, the mixture of
ribosyl bromides was treated with 1.5 equiv of the sodium
salt of 4-chloro-7H-pyrrolo[2,3-d]pyrimidine (generated
in situ). Despite the modest 43% yield of 2 observed in
this reaction, we were pleased to confirm that none of
the R-anomer of 2 was observed. If the glycosidation were
proceeding via simple SN2 displacement of bromide, we
would have seen a 5:1 ratio of â/R anomers. In situ 1H
NMR experiments revealed that the reaction took place
not through bromide displacement but through an inter-
mediate epoxide (6).7 It is known that nucleosides can
be prepared by treatment of 1,2-anhydro sugars with
heterocyclic bases.7b,8 SN2 opening of the R-epoxide of a
1,2-anhydro sugar results in highly selective formation
of the â-nucleoside. Application of this method, however,
in the preparation of nucleosides has been limited.8 By
carefully monitoring the reaction by HPLC and 1H NMR,
it was found that the moderate glycosidation yield is
mainly due to significant cleavage of the 2,4-dichloroben-
zyl groups in the presence of HBr-HOAc. On the basis
of this understanding of the glycosidation reaction, we
developed a route to the R-epoxide, eliminating the need
for HBr, which was incompatible with the benzyl ether
protecting groups. Hydrolysis of the 1-R-O-methyl ribose
5 gave diol 7, which upon treatment with methane-
sulfonic anhydride, triethylamine, and DMAP gave ep-
oxide 6.9 Treatment of the resulting epoxide solution with
4-chloro-7H-pyrrolo[2,3-d]pyrimidine and sodium hex-
amethyldisilazane gave a 65% yield of the desired 2.
a
Reagents: (a) HBr/AcOH; (b) 6-chloro-7-deazapurine, NaH.
SCHEME 3a
a
Reagents: (a) TFA/H2O; (b) Ms2O, Et3N, cat. DMAP.
Despite the improved glycosidation yield, we still desired
crystalline intermediates for ease of purification in large-
scale preparation.
Vor br u1 ggen Rou te. Contemporaneous with our study
of the route based on the previously described synthesis,
we were also pursuing an alternate route to 1 based on
the widely used Vorbru¨ggen glycosidation conditions.10
From our review of the literature concerning the synthe-
sis of 2-C-methyl ribonucleosides, we were cognizant of
recent work by both the Wolfe11 and Piccirilli12 groups.
Each group had reported the synthesis, under nearly
identical conditions, of a series of closely related ana-
logues of 1 in which only the nucleoside base differed.11,12
The stereochemical control at the anomeric center ob-
served in the Wolfe and Piccirilli syntheses was proposed
to arise from neighboring group participation of the R-2-
O-benzoate under Vorbru¨ggen reaction conditions. The
Vorbru¨ggen reaction calls for the treatment of a peracyl-
ated carbohydrate with a strong Lewis acid such as
TMSOTf, resulting in the formation of an oxonium ion
intermediate (8 f 9). The resulting oxonium ion 9 then
reacts with a silylated heterocyclic base to give â-nucleo-
side 10. The selectivity in this reaction results from
coordination of the neighboring 2-O-benzoate on the
(7) Formation of an epoxide in the reaction of C-2 hydroxyl-furanosyl
halides with base was proposed previously. For examples, see: (a)
Reist, E. J .; Hart, P. A.; Goodman, L.; Baker, B. R. J . Am. Chem. Soc.
1959, 81, 5176. (b) J arman, M.; Ross, W. C. J . J . Chem. Soc. (C) 1969,
199.
(8) For examples, see: (a) Ning, J .; Xing, Y.; Kong, F. Carbohydr.
Res. 2003, 338, 55. (b) J ung, M. E.; Toyota, A. Terahedron Lett. 2000,
41, 3577-3581. (c) Ning, J .; Kong, F. J . Carbohydr. Chem. 1997, 16,
311. (d) Ning, J .; Kong, F. Carbohydr. Res. 1997, 300, 355. (e)
McDonald, F. E.; Gleason, M. M. J . Am. Chem. Soc. 1996, 118, 6648.
(f) Baumgartner, H.; Marschner, C.; Pucher, R.; Griengl, H. Tetrahe-
dron Lett. 1991, 32, 611. (g) Chow, K.; Danishefsky, S. J . Org. Chem.
1990, 55, 4211. (h) Biggadike, K.; Borthwick, A. D.; Exall, A. M.; Kirk,
B. E.; Roberts, S. M.; Youds, P. J . Chem. Soc., Chem. Commun. 1987,
1083. (i) Konno, K.; Hayano, K.; Shirahama, H.; Saito, H.; Matsumoto,
T. Tetrahedron 1982, 38 3281. (j) Holy´, A.; Sˇorm, F. Collect. Czech.
Chem. Commun. 1969, 34, 3383.
(9) The initially formed mixture of anomeric mesylates gave only
the R-epoxide, likely via a solvolysis mechanism. When the reaction is
run at room temperature epoxide formation is rapid, but at lower
temperatures both R- and â-mesylate anomers can be observed as
precursors to the epoxide.
(10) For an extensive review of this reaction see: Vorbru¨ggen, H.;
Ruh-Pohlenx, C. Org. React. 2000, 55, 1.
(11) Harry-O’kuru, R. E.; Smith, J . M.; Wolfe, M. S. J . Org. Chem.
1997, 62, 1754.
(12) Tang, X.-Q.; Liao, X.; Piccirilli, J . A. J . Org. Chem. 1999, 64,
747.
6258 J . Org. Chem., Vol. 69, No. 19, 2004

Synthesis of a Hepatitis C Virus Inhibitor
SCHEME 4a
SCHEME 6
as â-8 were crystalline. To combine the favorable features
in both approaches, we subsequently sought to design a
route that would rely on ester protection to give crystal-
line intermediates leading to a â-selective glycosidation
via the epoxide.
a
Reagents: (a) TMSOTf; (b) persilylated base.
SCHEME 5
Th e Dia ceton e-D-glu cose Rou te to th e Ep oxid e.
Seeking a route to a diol analogous to 7 but bearing ester-
protected 3,5-hydroxyls, we were aware of reports of the
2-C-methyl-D-ribose structure mapped back to com-
mercially available diacetone-D-glucose (Scheme 6).16-18
We were further encouraged by the report of a few
crystalline intermediates en route from diacetone-D-
glucose to 13. As per the literature reports, we found the
10-step synthesis gave 13 in about 40% overall yield from
(15) Literature concerning the mechanism of Vorbru¨ggen glycosi-
dation suggests a reason for the nonreactive nature of 6-chloro-7-
deazapurine.
R-face, effectively directing attack of the silylated nu-
cleophile to the â-face. Given the close analogy between
the Wolfe and Piccirilli targets and our own, we naturally
chose to investigate the synthesis of 1 using Vorbru¨ggen
conditions. We began with the reported three-step syn-
thesis of 8 from commercially available 1,3,5-tri-O-
benzoyl-R-D-ribofuranose, repeated with only minor modi-
fication.13 Attempts to use â-8 as the glycosyl donor in
reaction with 4-chloro-7H-pyrrolo[2,3-d]pyrimidine under
a variety of Vorbru¨ggen conditions failed to give the
desired product.14 To confirm that the failure was specific
to the 4-chloro-7H-pyrrolo[2,3-d]pyrimidine, the reaction
was performed with 6-chloropurine 11 and gave a near
quantitative yield of the expected â-nucleoside 12.15 While
it appeared that we were both silylating the 4-chloro-
1
7H-pyrrolo[2,3-d]pyrimidine (as evidenced by in situ H
NMR) and forming the oxonium ion intermediate, no
conditions were found to effect the desired glycosidation.
Though the Vorbru¨ggen chemistry proved incompatible
with the pyrrolo[2,3-d]pyrimidine base, we were pleased
to find that the benzoate-protected carbohydrates such
Vorbru¨ggen has proposed that the attack of the silylated base on the
intermediate oxonium takes place through the nucleophile â to the
silane rather than in an ipso sense. In the case of the deazapurine it
is unlikely that silyl migration to the 7-C occurs, thus only reaction
through one of the remotely conjugated, silylated pyrimidine nitrogen
atoms may be invoked. Then again, there are a number of literature
examples of Vorbru¨ggen glycosidation successfully employed for nu-
cleophiles that do not fit well with this mechanistic rationalization.
For an extensive discussion of the mechanism of reaction of silylated
purines with similar oxonium ions, see ref 10. Also insightful are the
following: Dempcy, R. O.; Skibo, E. B. J . Org. Chem. 1991, 56, 776.
Krˇen, V.; P´ıskala, A.; Sedmera, P.; Havl´ıcˇek, V.; Prˇikrylova´, V.;
Witvrouw, M.; De Clercq, E. Nucleosides Nucleotides, 1997, 16, 97.
Other examples may be found in the tables given in ref 10.
(16) Brimacombe, J . S.; Rollins, A. J .; Thompson, S. W. Carbohydr.
Res. 1973, 31, 108.
(13) See ref 11. The oxidation of 1,3,5-tri-O-benzoyl-R-D-ribofuranose
(I f II) was modified from the original Dess-Martin periodate
oxidation to a RuCl3 catalyzed bleach oxidation.
(17) Funabashi, M.; Yamazaki, S.; Yoshimura, J . Carbohydr. Res.
1975, 44, 275.
(18) Beigelman, L. N.; Ermolinsky, B. S.; Gurskaya, G. V.; Tsapkina,
E. N.; Karpeisky, M. Y.; Mikhailov, S. N. Carbohydr. Res. 1987, 166,
219.
(14) The â-anomer of 8 could be isolated from the anomeric mixture
by selective crystallization. The formation of the oxonium ion inter-
mediate 9 from the â-anomer upon treatment with TMSOTf was found
to be significantly more rapid than that with the R-anomer.
J . Org. Chem, Vol. 69, No. 19, 2004 6259

Bio et al.
SCHEME 7
SCHEME 8
SCHEME 9a
diacetone-D-glucose. Unfortunately, 13 was isolated as
an oil. With only minor modification of the published
route, however, we could access the crystalline intermedi-
ate 14.
We found that both 13 and 14 could serve as glycosi-
dation precursors as treatment of either with HBr-AcOH
gave the same ribosyl bromides 16.19 We noted that at
the end of reaction the bromides were present as an 11:1
mixture of anomers favoring the R-16. Upon concentra-
tion with toluene to remove excess HBr-AcOH, however,
the ratio degraded to 4:1 R/â. Glycosidation of these
bromides with potassium tert-butoxide and 4-chloro-7H-
pyrrolo[2,3-d]pyrimidine gave only the desired â-anomer
1
18. H NMR showed once again that the glycosidation
a
Reagents: (a) 1 mol % of TEMPO, NaOCl, NaBr, NaOAc; (b)
MeMgCl; (c) KOH, Triton X-405, BnCl; (d) 5% aq H2SO4.
conditions were converting both bromides to an epoxide
intermediate, in this case 17, which then reacted to give
the desired â-glycosidation product 18 in about 62%
overall yield from triol 14. The use of HBr-AcOH to
access epoxide 17 was operationally less than optimal,
requiring azeotropic removal of excess reagent. Addition-
ally, the 3-acetate appeared to be somewhat labile under
the glycosidation conditions. Loss of the acetate led to
formation of what appear to be dimeric and oligomeric
products. From what we had learned to this point, we
realized that the ideal substrate for the glycosidation
would be a 1,2-diol allowing mild and efficient epoxide
formation via mesylation. Additionally, the ideal sub-
strate would possess robust protecting groups on the 3
and 5 hydroxyls. It occurred to us that it might be
possible to access such a substrate using a sequence
analogous to the one used to prepare 14 if 15 were
esterified at the start to give the diester rather than the
monoester. Esterification of both hydroxyls of 15 followed
by acetonide hydrolysis, oxidative cleavage, and selective
formyl hydrolysis would give 19. Success would then
require defining conditions promoting acyl migration and
affording furanose 20 rather than pyranose 21.20 Subse-
quent hydrogenolysis of the benzyl ether would afford a
1,2-diol substrate for use in the glycosidation. To deter-
mine the potential for success in this strategy, aldehyde
27 was prepared from the diester 25 by a procedure
similar to the one developed for preparation of 14.
Treatment of 27 with Et3N in MeOH at 60 °C gave the
desired furanose in 90% selectivity (20:21 ) 9:1). This
was quite an encouraging result for the first trial. We
were pleased to find that the diol 28, obtained upon
hydrogenolysis of 20, is a crystalline solid. Subjecting 28
to reaction with methanesulfonic anhydride and triethyl-
amine gave the desired epoxide 29, which reacted with
the 4-chloro-7H-pyrrolo[2,3-d]pyrimidine to give the â-
glycosidation product. In light of these results, we set
about to optimize the synthesis of 1 via the crystalline
diol 28. A number of steps in the literature synthesis from
which we were working proved inefficient or impractical
for the preparation of 1. Specifically, the oxidation of
diacetone-D-glucose to give ketone 22 called for the use
of excess pyridinium dichromate (PDC) and 300 wt %
powdered 4 Å sieves.21 The use of molecular sieves to
absorb water was required; otherwise, a mixture of
(20) Acyl migration is well-known in carbohydrate chemistry where
the migration takes place on a rigid framework and has been used to
prepare compounds that would be difficult to attain by direct esteri-
fication. (a) Haines, A. H. Relative Reactivities of Hydroxyl Groups in
Carbohydrates. In Advances in Carbohydrate Chemistry and Biochem-
istry; Tipson, R. S., Horton, D., Eds.; Academic Press: New York, 1976;
Vol. 33, pp 11-109. (b) Horrobin, T.; Tran, C. H.; Crout, D. J . Chem.
Soc., Perkin Trans. 1 1998, 1069. (c) Goueth, P. Y.; Gogalis, P.; Bikenga,
R.; Gode, P.; Postel, D.; Ronco, G.; Villa, P. J . Carbohydr. Chem. 1994,
13, 249. (d) Chaplin, D.; Crout, D. H.; Bornemann, S.; Hutchinson, D.
W.; Khan, R. J . Chem. Soc., Perkin Trans. 1 1992, 235. There are fewer
examples where acyl migration has been employed in an acyclic system.
(e) Suh, Y.-G.; J ung, J .-K.; Seo, S.-Y.; Min, K.-H.; Shin, D.-Y.; Lee. Y.-
S.; Kim, S.-H.; Park, H.-J . J . Org. Chem. 2002, 67, 4127. (f) Vares, L.;
Rein, T. Org. Lett. 2000, 2, 2611. (g) VanMiddlesworth, F.; Lopez, M.;
Zweerink, M.; Edison, A. M.; Wilson, K. J . Org. Chem. 1992, 57, 4753.
(h) Maycock, C. D.; Barros, M. T.; Santos, A. G.; Godinho, L. S.
Tetrahedron Lett. 1992, 33, 4633.
(19) Treatment of 14 with HBr-AcOH resulted in complete bromi-
nation at the anomeric position but only ∼85% of the bromide was
acylated at the 3-OH despite the use of excess HBr-AcOH. While no
further work was done to elucidate the mechanism of 3-OH acylation,
our observations suggest that intramolecular transfer of a 1-O-acyl to
the 3-OH is involved.
6260 J . Org. Chem., Vol. 69, No. 19, 2004

Synthesis of a Hepatitis C Virus Inhibitor
SCHEME 10a
TABLE 1. Con d ition s for F or m a tion of 20
area %
entry
base
Et3N
Et3N
Et3N
solvent
conditions conv, %
21a
1
2
3
4
5
6
7
8
9
MeOH/water
MeOH/water
MeOH
THF/water
THF
rt, 48 h
99
97
98
54
4
99
98
99
80
45
97
8
11
10
12
37
4
4
12
7
60 °C, 5 h
60 °C, 10 h
60 °C, 48 h
60 °C, 48 h
0 °C, 24 h
Et3N
Et3N
iPr2NH
iPr2NH
iPr2NH
DMAP
MeOH
MeOH/iPrOAc 0 °C, 48 h
MeOH
MeOH
rt, 27 h
rt, 48 h
10
11
Imidazole MeOH
NaHCO3 MeOH
45 °C, 72 h
45 °C, 48 h
4
12
a
HPLC area % 21 of total products (20 + 21).
HBF4 gave 26 at a reasonable reaction rate, but the
hydrolysis stalled at 90% conversion. By removing ac-
etone liberated during hydrolysis by vacuum distillation,
we were able to achieve a satisfactory 96% conversion.
Treatment of diol 26 with periodic acid resulted in a
mixture of 27 and 19. To optimize the formation of 20,
we needed to identify conditions that would minimize the
formation of pyranose 21. After screening solvent, tem-
perature, and base, we discovered that using 2 equiv of
diisopropylamine in methanol-isopropyl acetate led to
20 with >23:1 selectivity over 21.22 Hydrogenolysis of the
2-benzyl ether gave the desired diol 28 as a crystalline
solid from isopropyl acetate-hexanes in 65% overall yield
from 15.
With 28 in hand we found that epoxide formation could
be effected with methanesulfonyl chloride in place of the
more expensive methanesulfonic anhydride. The epoxide
29 turned out to be remarkably stable to hydrolysis,
tolerating aqueous workup to give a 93% yield from 28.23
With an efficient synthesis of 29 now fully developed we
turned our attention to the critical glycosidation reaction.
Parallel with our carbohydrate studies we were work-
ing to identify a nucleoside base which was compatible
with the nucleophilic epoxide ring opening. We began our
glycosidation studies with the commercially available, but
prohibitively expensive, 4-chloro-7H-pyrrolo[2,3-d]pyri-
midine. The chloro substituent was to be converted to
the required amino function by ammonolysis after coup-
ling.24 The ammonolysis was expected to require a high
temperature in alcoholic ammonia with use of a pressure-
a
Reagents: (a) toluoyl chloride, pyridine; (b) HBF4; (c) H5IO6;
i
i
(d) Pr2NEt, MeOH, PrOAc; (e) H2, Pd/C; (f) MsCl, Et3N.
ketone and hydrate was isolated. These conditions are
environmentally undesirable due to the large volume of
chromium-contaminated waste produced. Furthermore,
to isolate the ketone free from chromium required a
filtration through either silica gel or Florosil, which also
created chromium-laden waste. We found the alcohol
oxidation could be accomplished using catalytic TEMPO
with aqueous bleach as oxidant. This protocol gave a
mixture of ketone and hydrate. Dehydration of a toluene
solution by azeotropic distillation gave the ketone 22 as
a solution in toluene. The resulting toluene solution was
treated with methylmagnesium chloride at 0 °C to give
23 as a crystalline solid from toluene-octane in 85%
overall yield from diacetone-D-glucose. Use of the dehy-
drated ketone is critical as the hydrated form does not
react with methylmagnesium chloride to give the desired
product but is returned as the hydrate upon quench-
ing.
While published methods for benzylation of the 3-hy-
droxyl to give 24 call for the use of NaH in DMSO, we
wished to avoid the use of NaH due to safety concerns.
We found this benzylation could be accomplished by
heating 23 in toluene with solid KOH, benzyl chloride,
and Triton X-405 as the solid-liquid phase transfer
catalyst. The crude benzyl ether product was subse-
quently treated with dilute acid to effect selective cleav-
age of the 5,6-acetonide. Under these optimized condi-
tions, diol 15 is isolated by crystallization from toluene-
heptane in a 84% yield from 23. Reaction of 15 with
toluoyl chloride in acetonitrile using pyridine as base
afforded the diester 25. After screening a variety of acids
for hydrolysis of the acetonide, we found that aqueous
(22) See entries 6 and 7 from Table 1. We chose the mixed solvent
system, MeOH-iPrOAc, despite the longer reaction times so that the
i
solution of 27 and 19 in PrOAc from the extractions following oxidative
cleavage could be used directly without isolation of the product mixture
i
or complete removal of PrOAc as would be required to conduct the
reaction in methanol alone.
(23) Solutions of the epoxide proved unchanged after several weeks
at room temperature. This should be contrasted with the stability of
epoxides bearing 3,5 benzyl ethers (6), which readily hydrolyzed on
exposure to water. We attribute this stability to the electron-withdraw-
ing nature of the 3,5 esters which destabilize the formation of an
oxonium ion. Such long-range effects have been noted and quantified
in the carbohydrate literature. See: Glaudemans, C. P. J .; Fletcher,
H. G., J r. J . Am. Chem. Soc. 1965, 87, 4636. Green, L. G.; Ley, S. V.
Carbohydr. Chem. Biol. 2000, 1, 427.
(21) A number of oxidations have been applied to this problem
including some catalytic methods (see: Weber, J . F.; Talhouk, J . W.;
Nachman, R. J .; You, T.-P.; Halaska, R. C.; Williams, T. M.; Mosher,
H. S. J . Org. Chem. 1986, 51, 2702); most reports, however, rely on
the high-yielding chromium-based oxidations such as: Shing, T. K.
M.; Wong, C.-H.; Yip, T. Tetrahedron: Asymmetry 1996, 7, 1323.
J . Org. Chem, Vol. 69, No. 19, 2004 6261

Bio et al.
SCHEME 11a
rated reactor. We hoped to avoid these forcing conditions
and achieve a more convergent synthesis of 1 by going
into the glycosidation with the amino group in place. To
explore the glycosidation with various protected ana-
logues of 4-amino-7H-pyrrolo[2,3-d]pyrimidines, we re-
quired an efficient synthesis of the parent nucleoside
base.
Syn th esis 4-P h th a lim id o-7H-p yr r olo[2,3-d ]p yr i-
m idin e. A review of the literature concerning the 4-amino-
7H-pyrrolo[2,3-d]pyrimidine leads one back to two similar
reports of its synthesis, one by Davoll25 and the other the
subject of a patent by the Wellcome Foundation Lim-
ited.26
The Davoll and Wellcome syntheses form the pyrimi-
dine ring by condensation of a malonate derivative with
thiourea and then remove the sulfur atom with Raney
nickel. In both cases the best route to the 4-amino-7H-
pyrrolo[2,3-d]pyrimidine was through ammonolysis of the
4-chloro-7H-pyrrolo[2,3-d]pyrimidine, a process that called
for extended high temperature and pressure exposure of
the chloro compound to alcoholic ammonia. This reaction
led to byproducts resulting from incorporation of the
alcohol at the 4-position. The chloro compound was
accessed through the 4-hydroxyl-7H-pyrrolo[2,3-d]pyri-
midine by reaction of it with POCl3, a process that
generates a great deal of acid waste. For these reasons
neither process seemed efficient. Herein we report an
improved synthesis that, while based on the Wellcome
and Davoll syntheses, includes important modifications
leading to a practical, efficient synthesis of 4-amino-7H-
pyrrolo[2,3-d]pyrimidine. The reported procedures call for
the reaction of bromoacetaldehyde diethyl acetal with
malononitrile to be carried out in the presence of sodium
ethoxide.27 Optimization of base, solvents, and stoichi-
ometry revealed that this alkylation could be performed
in DMF with potassium carbonate, milder and more
readily handled than sodium ethoxide. Under these
conditions a 60% yield of 30 is realized while limiting
over-alkylation to 10-15%. Following aqueous workup,
the crude solution of 30 and 31 is treated with thiourea
in the presence of potassium tert-butoxide to form the
pyrimidine ring in 85%. The potassium salt 32 was
conveniently isolated by crystallization from the reaction
mixture, rejecting multiple side products and resulting
in >95% pure product.28 Use of the mixture of 30 and 31
in the reaction with thiourea obviated the need for a high-
a
Reagents: (a) K2CO3, malononitrile; (b) thiourea, t-BuOK; (c)
5 N HCl; (d) H2O2; (e) EtOH, H2SO4.
temperature vacuum distillation required to isolate 30
free from 31. Isolation of the potassium salt instead of
the neutral compound was found to be more convenient
in that it eliminated the need for careful pH adjustments
and gave higher purity product. Exposure of 32 to
aqueous HCl results in near-quantitative conversion to
33.
A major limitation of both the Davoll synthesis and
that reported by Wellcome is the need to use Raney nickel
reduction for cleavage of the sulfur-carbon bond, a
procedure that generates much heavy metal waste.
Furthermore the reduction procedure requires a hot
filtration at the end of the reduction due to the low
solubility of the intermediate. We hoped to avoid these
complications in the sulfur removal by making use of the
observation that aromatic sulfinic acids readily lose
sulfur dioxide upon heating with sulfuric acid.29
Sulfinic acid 34 was prepared by reaction of 33 in 1 N
potassium hydroxide with 2 equiv of hydrogen peroxide.
The oxidation was carefully controlled so that the hy-
drogen peroxide was charged at the optimal rate to
minimize the formation of the sulfonic acid as a result of
over-oxidation. We found that slow addition of hydrogen
peroxide (3 h addition time) led to more over-oxidation
than when shorter addition times were used.30 On a large
scale, the addition was done in portions as fast as possible
while still controlling the temperature. The sulfinic acid
34 was then heated in 50 wt % sulfuric acid resulting in
loss of SO2 to give 35 as the hydrogen sulfate salt, which
(24) This conversion was reported previously in the synthesis of 1.5
Ammonolysis of a glycosylated 6-chloro-7-deazapurine was successfully
performed to synthesize the natural product tubercidin: Anderson, J .
D.; Botems, R. J .; Geary, S.; Cottam, H. B.; Larson, S. B.; Matsumoto,
S. S.; Smee, D. F.; Robins, R. K. Nucleosides Nucleotides 1989, 8, 1201.
(25) Davoll, J . J . Chem. Soc. 1960, 131.
(26) The Wellcome Foundation Limited, UK patent 812366, 1955.
(27) An obvious opportunity for improving pyrimidine synthesis
would be to do the ring forming condensation not with thiourea but
with formamidine. Despite attempts at this condensation under a
variety of conditions we were never able to realize greater than 15%
conversion to the desired I.
(29) This desulfurization reaction has been used in the synthesis of
other nitrogen heterocycles. Evans, R. M.; J ones, P. G.; Palmer, P. J .;
Stephens, F. F. J . Chem. Soc. 1956, 4106.
(30) Careful control of stoichiometry is important in the oxidation,
and use of greater than 2 equiv of H2O2 results in over oxidation. It is
not immediately clear why slower addition rates favor over oxidation.
For an investigation into the mechanism of aryl thiol oxidation with
H2O2 see: Evans, B. J .; Doi, J . T.; Musker, W. K. J . Org. Chem. 1990,
55, 2337.
(28) HPLC analysis gave 95.6 area % at 210 nm.
6262 J . Org. Chem., Vol. 69, No. 19, 2004

Synthesis of a Hepatitis C Virus Inhibitor
SCHEME 12a
imide, or both, despite our efforts to keep the reactions
anhydrous.34 The formation of this mixture turned out
to be of little consequence, however, when it was found
that heating the end of the reaction mixture with
n-butylamine converted all the products to 1.35 The target
nucleoside 1 was isolated following global deprotection
by direct crystallized from solution on addition of anti-
solvent. In this manner diol 28 could be converted to
nucleoside target 1 in 77% overall yield. The final product
was conveniently isolated by crystallization from 1-pro-
panol-water in >99% purity. The results presented here
suggest that 1,2-anhydrofuranose carbohydrates may
find fruitful employment in the synthesis of nucleosides
containing heterocycles which are not amenable to the
Vorbru¨ggen conditions.
a
i
Reagents: (a) PhCOCl; (b) phthalic anhydride, Pr2NEt.
SCHEME 13a
Con clu sion
We have developed a practical, efficient, and conver-
gent synthesis of nucleoside 1 via the crystalline 3,5-di-
O-toluoyl-2-C-methyl-D-ribofuranose starting from the
inexpensive and readily available diacetone-D-glucose. A
catalytic TEMPO-bleach oxidation of diacetone-D-glucose
to give 1,2:5,6-di-O-isopropylidene-R-D-ribo-hexofuranos-
3-ulose was developed as an environmentally sound
alternative to the typical chromium oxidation. An ester
migration was exploited to selectively give the 3,5-diester.
The â-selective glycosidation of 3,5-di-O-toluoyl-2-C-
methyl-D-ribofuranose has been developed via the nu-
cleophilic opening of an 2-â-C-methyl-1,2-R-anhydrori-
bose formed under mild, stereospecific conditions. Addition-
ally, we have developed a practical synthesis of 4-amino-
7H-pyrrolo[2,3-d]pyrimidine. These methods have been
applied to the multi-kilogram synthesis of the potential
HCV polymerase inhibitor 1, realized in 12 steps and 35%
overall yield.
a
Reagents: (a) NaH, 6-phthalimido-7-deazapurine; (b) n-butyl-
amine.
crystallized directly from the reaction mixture in 84%
yield from 33. The sulfate salt 35 served as a convenient
intermediate for derivatization of the primary amino
group. For example, reaction of 35 with benzoyl chloride
in the presence of excess base gave the benzamide 36 in
high yield. The desired 6-phthalimido-7-deazapurine 37
was prepared by treating 35 with phthalic anhydride and
Hu¨nig’s base in dimethylacetamide at 80-85 °C to give
37 as a crystalline solid isolated in >85% yield.
Glycosid a tion a n d Dep r otection . In seeking an
appropriate protecting group for the amine functionality
we first looked to amides, which might allow for depro-
tection simultaneous with deprotection of the esters at
the 3 and 5 positions. While benzoyl protection is most
obvious, its use gave rise to side products resulting from
reaction through the pyrimidine portion of the base, a
result we attributed to the acidic amide NH.31Employing
phthalimide protection32 eliminated the acidic amide NH
and, consequently, the side products from pyrimidine
reaction. With this new substrate we found that only a
catalytic amount of NaH was required to achieve com-
plete conversion of the epoxide.33 While the phthalimide
nucleoside base gave excellent yields for the C-N bond
formation, the yield of 38 suffered due to the formation
of products resulting from hydrolysis of the 3-ester, the
Exp er im en ta l Section
Gen er a l. All reactions were conducted under N2 atmo-
sphere with standard air-free manipulation techniques. Sol-
vents and common reagents were purchased from a commer-
cial source and used without further purification. Concentration
in vacuo refers to removal of the solvent with a rotary
evaporator at reduced pressure. Chemical shifts are reported
in ppm downfield from tetramethylsilane (TMS). Data are
reported as follows: (s ) singlet, d ) doublet, t ) triplet, q )
quartet, m ) multiplet; integration; coupling constant(s) in
Hz; assignment). All new compounds are fully characterized
below.
1,2:5,6-Di-O-isop r op ylid en e-r-D-r ibo-h exofu r a n os-3-u l-
ose (22). Water (10 kg) and ethyl acetate (30 L) were charged
to a 100-L vessel followed by addition of diacetone-D-glucose
(6.67 kg, 25.6 mol), sodium bromide (2.11 kg, 20 mol), sodium
acetate (3.15 kg, 38.4 mol), and then TEMPO (2,2,6,6-tetra-
methyl-1-piperidinyloxy) (40 g, 0.256 mol). Additional water
(3.3 kg) and ethyl acetate (3 L) were charged and the resulting
solution was cooled to 14 °C. Bleach (ca. 12.5% solution, 16.9
kg, 28.4 mol) was then added over 2 h maintaining the
(31) Specifically, we observed products such as I and II.
(34) The mechanism of 3 deprotection or phthalimide opening was
never clarified. These products, specifically the deprotection at 3, were
observed regardless of rigorous exclusion of water from these reactions.
The hydrolysis of 3 esters may be the result of a general base assistance
by 2-hydroxyl (alkoxides) or acceleration due to DMAP-like attack by
the pyrimidine heterocycle.
(35) Although phthalimide can be deprotected by hydrolysis in aq
NaOH or methanolysis in NaOMe/MeOH, the following workup and
isolation of 1 are not as straightforward as the use of n-BuNH2.
(32) Kume, A.; Sekine, M.; Hata, T. Tetrahedron Lett. 1982, 23,
4365-4368.
(33) As little as 20 mol % of NaH could be used to effect complete
conversion of the epoxide. We routinely employed 50 mol % to ensure
complete conversion.
J . Org. Chem, Vol. 69, No. 19, 2004 6263

Bio et al.
temperature between 10 and 20 °C with constant cooling. After
a 30-min age, 1H NMR analysis indicated 1% residual starting
material. The reaction was quenched with 2 M Na2SO3
aqueous solution (100 mL, 0.2 mol) after which a starch iodide
test proved negative. The phases were separated and the
aqueous phase was extracted with ethyl acetate (33 L). The
combined organic extracts were washed with 12% aqueous
NaCl solution (6.6 L). The major component in solution at this
point is the ketone-hydrate. The resulting organic phase was
concentrated under reduced pressure (pot temperature ca. 15
°C) to low volume. Toluene (35 L) was then added and the
distillation was continued adjusting the vacuum until the pot
temperature was 70 °C at which point azeotropic water
removal was observed. When water evolution had ceased, the
19 °C. The reaction is initially cloudy but becomes homoge-
neous during the second hour and remains so. When HPLC
indicated >98% conversion of 24 (approximately 4 h), 18 L of
toluene was added causing phase separation. The aqueous
phase was then cut and the organic layer washed with 20 L of
∼6% NaHCO3 solution. Following the phase cut the organic
phase was washed with 10 L of water. The organic layer was
then concentrated at reduced pressure with the internal
temperature of 50 °C to about 30 L. An additional 30 L of
toluene was added and the solution was again concentrated
to 30 L, allowing the temperature of the pot to rise to 80 °C.
n-Heptane (43 L) was then slowly added while the internal
temperature was maintained at about 80-75 °C. The solution
was then cooled to 70 °C at which point product began
crystallizing from solution (if seed does not form the reaction
can be self-seeded by removing an aliquot, cooling to room
temperature, and returning the crystals which formed to the
bulk solution). The seedbed was allowed to develop for 1-2 h
at 70-75 °C. The slurry was then cooled at 15 deg/h to room
temperature. Prior to isolation of the product the internal
temperature was lowered to 4 °C and the slurry was aged for
∼1 h at this temperature prior to filtration. The wet cake was
washed with 2 × 8 L of n-heptane then dried on the filter
under nitrogen overnight to give a 83% yield of 15 from 23
(9.02 kg, 94 wt %, 26.1 mol).37 HPLC assay conditions: C-8
column (250 mm × 4.6 mm i.d., 5 µm particle size), 50% MeCN:
50% 0.1% v 60% HClO4/H2O mobile phase, ramp to 70% MeCN
over 20 min, 1.5 mL/min at 35 °C and detecting at 210 nm
wavelength.
2-C-Me t h yl-3,5-d i-O-(4-m e t h ylb e n zoyl)-D -r ib ofu r a -
n ose (28). To a solution of 3-O-benzyl-1,2-O-isopropylidene-
3-C-methyl-R-D-allofuranose (15) (5.0 kg, 15.4 mol) and pyri-
dine (3.7 kg, 46.2 mol) in 35 L of acetonitrile was added
4-methylbenzoyl chloride (p-toluoyl chloride) (5.2 kg, 33.9 mol),
and the reaction was heated at 50-55 °C for 12 h. A solution
of 6.0 L (46.2 mol) of 48 wt % of HBF4 in 9 L of water was
added at 50-55 °C. After 2 h, 10 L of acetonitrile was distilled
off under reduced pressure and 10 L of acetonitrile was added.
At 97% conversion, 10 L of acetonitrile was distilled off, and
the reaction solution was cooled to 0-5 °C. A solution of
periodic acid (4.2 kg, 18.5 mol) in 10 L of water was added.
After the reaction was aged for 30 min, 35 L of isopropyl
acetate and 10 L of water were added. The organic phase was
washed with 25 L of water followed by 20 L of aqueous
NaHCO3, 15 L of 5% sodium thiosulfate in water, and 15 L of
water. The isopropyl acetate solution was concentrated to 10-
15 L, and 40 L of methanol was added. The solution was cooled
to 0 °C and diisopropylamine (0.78 kg, 7.7 mol) was added.
After 2 d at 0 °C, aqueous HCl (1 N, 7.7 L) was added at 0-5
°C followed by 30 L of isopropyl acetate and 40 L of water.
The organic phase was washed with aqueous 1 N HCl,
NaHCO3, and brine. The organic phase was dried through
azeotropic distillation and treated with activated carbon. The
carbon was removed by filtration and the resulting solution
was diluted to 75 L with isopropyl acetate and hydrogenated
(45 psi, 50 °C, 1.5 kg of 10% Pd/C) for 24 h. The filtrate was
concentrated to 15 L, and 60 L of heptane was added at 50
°C. The crystalline product was isolated by filtration washing
with a 10 L of 20% isopropyl acetate-heptane. Drying afforded
4.03 kg of the desired diol 28. 1H NMR (CDCl3, 400 MHz): The
ratio of R: â isomers in CDCl3 was about 5 to 1. For the major
isomer: δ 7.95-7.90 (m, 4H), 7.26 (d, J ) 8.0 Hz, 2 H), 7.17
(d, J ) 8.0 Hz, 2 H), 5.53 (d, J ) 7.2 Hz, 1 H), 5.22 (d, J ) 2.8
Hz, 1 H), 4.65-4.49 (m, 3 H), 3.08 (d, J ) 3.2 Hz, 1 H), 2.44
(s, 3 H), 2.38 (s, 3 H), 2.26 (s, 1 H), 1.44 (s, 3 H). For the minor
isomer: δ 7.95-7.90 (m, 4H), 7.27 (d, J ) 8.0 Hz, 2 H), 7.22
(d, J ) 8.0 Hz, 2 H), 5.16 (d, J ) 5.6 Hz, 1 H), 5.12 (d, J ) 5.6
Hz, 1 H), 4.66-4.49 (m, 3 H), 3.54 (d, J ) 5.6 Hz, 1 H), 2.91
(s, 1 H), 2.43 (s, 3 H), 2.40 (s, 3 H), 1.44 (s, 3 H). 13C NMR
1
reaction was sampled and H NMR analysis indicated complete
break of hydrate (ca 0.5 equiv of residual ethyl acetate relative
to product was observed). The solution volume was adjusted
to 45 L by addition of toluene, and the resulting solution of 22
was used as is in the next step.36
1H NMR Meth od for Oxid a tion /Hyd r a te Br ea k . For
determining the azeotropic dehydration end-point the NMR
1
chemical shift of the anomeric H proved diagnostic. For the
hydrate break, a 200-µL sample of the toluene reaction mixture
was added to 1 mL of CDCl3 and analyzed immediately
1
[species, δ anomeric H]: diacetone-D-glucose, 5.95 ppm (J )
4.0 Hz); ketone (22), 6.14 ppm (J ) 4.8 Hz); and hydrate, 5.85
ppm (J ) 4.0 Hz).
1,2:5,6-Di-O-isop r op ylid en e-3-C-m et h yl-r-D-a llofu r a -
n ose (23). Methylmagnesium chloride (10.2 L of a 3 M THF
solution) was added over 80 min to the crude solution of ketone
22 in toluene (ca. 45 L) while the internal temperature was
maintained in the range -10 to 0 °C. After a 30-min age, the
reaction was quenched by addition to a solution of ammonium
chloride (23.2 L of 4 M aqueous solution). The layers were
separated and the aqueous phase was extracted with ethyl
acetate (23 L). The combined organic phase (ca. 75 L) was
washed with 12% NaCl aqueous solution (23 L) then filtered
through a 10-µm in-line filter and concentrated under reduced
pressure to a low volume (ca. 10 L). During the concentration,
the product started to crystallize out. Octane (45 L) was added
and the slurry was warmed to 78 °C dissolving the solids. The
solution was allowed to cool and crystallization occurred at
50 °C. The mixture was cooled to -10 °C and filtered to afford
7.02 kg of tertiary alcohol 23 in 85% yield from diacetone-D-
glucose.21
3-O-Ben zyl-3-C-m eth yl-1,2-O-isop r op ylid en e-r-D-a llo-
fu r a n ose (15). A 72-L vessel fitted with a mechanical stirrer
was charged with toluene (36 L), then solid alcohol 23 (4.32
kg, 15.75 mol), followed by benzyl chloride (2.92 kg, 23.1 mol)
and Triton X-405 (70% solution in water, 0.45 kg). KOH
powder (3.7 kg, 65.9 mol) was added in a single portion
resulting in a mild exotherm (19 °C up to ∼30 °C). The reaction
vessel was fitted with a Dean-Stark trap and heated to reflux
for 20 h. Over the course of 20 h approximately 250-300 mL
of water collected in the trap. Upon reaching >97% conversion
(as determined by GC), the reaction was cooled to room
temperature and washed with 10 L of water. The layers were
separated and the organic was washed with 5 L of 1 M HCl
and finally with 5 L of 12% aqueous NaCl solution. The
resulting organic solution was then concentrated at reduced
pressure to give 3-O-benzyl-1,2:5,6-di-O-isopropylidene-3-C-
methyl-R-D-allofuranose (24) as a >70 wt % solution in toluene.
This reaction was repeated at the same scale and combined
to give an assay total of 10.81 kg of 24, 29.66 mol, as a 70 wt
% solution in toluene. The toluene solution was then diluted
with acetonitrile (47 L) in a 100-L vessel with stirring. To this
solution was added 14.5 L of a 5 vol % of sulfuric acid aqueous
solution and the reaction temperature was maintained at ca.
(36) 1H NMR data for the ketone has been reported. Soler, T.;
Bachki, A.; Falvello, L. R.; Foubelo, F.; Yus, M. Tetrahedron: Asym-
metry 2000, 11, 493. 1H NMR data for the hydrate have been
reported: see ref 21.
(37) The 1H NMR data obtained for this compound agreed with
published data. Dequin, B.; Bertounesque, E.; Guadel, G.; Florent, J .-
C.; Monneret, C. Tetrahedron 1992, 48, 4885-4892.
6264 J . Org. Chem., Vol. 69, No. 19, 2004

Synthesis of a Hepatitis C Virus Inhibitor
(CDCl3, 100 MHz): δ 166.6, 166.3, 165.9, 165.7, 144.6, 144.3,
143.8, 143.7, 129.9, 129.7, 129.3, 129.2, 129.1, 129.0, 127.0,
126.9, 126.4, 126.2, 102.9, 100.8, 79.8, 79.2, 78.7, 76.9, 76.5,
76.4, 65.5, 64.0, 23.5, 21.7, 21.6, 20.0. Anal. Calcd for
151.64, 157.46. Anal. Calcd for C12H16N4O4: C, 51.42; H, 5.75;
N, 19.99. Found: C, 51.44; H, 5.42; N, 20.00.
2-(2,2-Dieth oxyeth yl)m a lon on itr ile (30). To a 100-L
flask equipped with mechanical stirrer, thermocouple, and
nitrogen/vacuum inlet was added 33 L of DMF and potassium
carbonate (12.6 kg, granular, 91.1 mol). Malononitrile (6.02
kg, 91.1 mol) was melted at 40 °C overnight and added to the
K2CO3 slurry and the mixture was washed in with 3 L of DMF.
A purple slurry resulted with mild exotherm to 41 °C.
Bromoacetaldehyde diethyl acetal (9.00 kg, 45.7 mol) was
added. The slurry was degassed three times (placed under
vacuum and back-filled with nitrogen) and warmed to 50 °C
then aged 17.5 h. GC analysis showed >99A% conversion. The
batch was cooled to ambient temperature and mixed with 80
L of H2O and 40 L of toluene in a 200-L extractor. After layer
cut, the top organic layer was washed with 2 × 70 L of 5%
brine. GC assay of the organic layer indicated 5.41 kg (65%
yield) of 30. The dialkylated product (31) was present at 11
GC area %. The crude solution was used directly for the next
step. An analytical sample was obtained as an oil after column
chromatography on silica gel.
GC sample preparation from the reaction: ∼1 mL of
reaction mixture was quenched into 10 mL of water and 10
mL of toluene, the toluene layer was diluted 3 times for GC
assay on an 14% cyanopropyl phenyl methyl capillary column,
15 m × 530 µm, 1.00 µm film thickness; initial temperature
50 °C; ramp 15 deg/min to 250 °C then hold for 4 min; split
ratio 5:1; retention times, toluene 2.0 min, DMF 3.0 min,
bromoacetaldehyde diethyl acetal 4.2 min, malononitrile 4.9
min, 30 at 7.9 min, and 31 at 10.9 min. For conversion
calculations toluene, DMF, and malononitrile peak integra-
tions were deleted. For yield assay, a weighed 1-mL sample is
diluted to 50 mL with MTBE for quantitative assay.
C
22H24O7: C, 65.99; H, 6.04. Found: C, 65.59; H, 6.03.
1,2-An h yd r o-3,5-d i-O-(4-m eth ylben zoyl)-2-C-m eth yl-â-
D-r ibofu r a n ose (29). To a 72-L vessel was charged dry
dichloromethane (32 L), triethylamine (3.0 L), and diol 28 (3.44
kg). The mixture was warmed to 30 °C, then methanesulfonyl
chloride (0.79 L) was added over 40 min. After 1 h, the batch
was partitioned between pH 7 buffer (20 L) and methyl tert-
butyl ether (44 L). The organic phase was washed with 1 M
aqueous NaCl (38 L) then switched to toluene by vacuum
distillation. The resulting solution of epoxide 29 (2.75 assay
kg, 93%) was used directly in the subsequent glycosidation
reaction. The HPLC method was same as that for 28.
4-P h th alim ido-7-[3′,5′-di-O-(4-m eth ylben zoyl)-2′-C-m eth -
yl-â-D-r ibofu r a n osyl]-7H-p yr r olo[2,3-d ]p yr im id in e (38).
Tetrahydrofuran (5.4 L), sodium hydride (146 g of 60%
dispersion in oil, unwashed), and N,N-dimethylacetamide (4
L) were charged to a 72-L flask and the suspension was cooled
to 0 °C. 4-Phthalimido-7H-pyrrolo[2,3-d]pyrimidine (37) (2.08
kg) was added to the reaction while maintaining the temper-
ature below 25 °C (Caution! Gas evolution). N,N-Dimethylac-
etamide (1.4 L) was added followed by a solution of the epoxide
29 (2.75 assay kg) in toluene (5.5 L), and the reaction mixture
was heated at 50 °C for 9 h. After the solution was cooled to
ambient temperature, ethyl acetate (19 L) and aqueous KH2-
PO4 (0.97 kg in 19 L of water) were added. The organic phase
was washed with water. The resulting organic solution was
solvent switched to toluene by vacuum distillation (<40 °C)
to give a final volume of 10 L. The HPLC method was the same
as that for 28.
4-Am in o-7-(2′-C-m eth yl-â-D-r ibofu r a n osyl)-7H-p yr r olo-
[2,3-d ]p yr im id in e (1). To the toluene solution of 38 at
ambient temperature was added methanol (25 L) and n-
butylamine (3.13 kg, 42.87 mol). The reaction mixture was
aged at 64 °C for 24 to 30 h. The mixture was concentrated to
about 9 L, and 9 L of methanol was added. The solution was
concentrated to 9 L, and the resulting slurry was aged at 60
°C for 60 min. Toluene (13 L) was added over 1.5 h. The slurry
was aged at 60 °C for another 2 h and then allowed to cool to
ambient temperature. The solid was filtered and the wet cake
was washed with 9 L of 20% methanol in toluene then 9 L of
10% methanol in toluene. The solid was suction dried at
ambient temperature under nitrogen to give 1.79 kg of 1 as
the toluene solvate; the yield was 77% for the three steps from
diol 28. The nonsolvate form is available by two methods.
(a ) Recr ysta lliza tion of 1: The toluene solvate form of
nucleoside 1 (30 g) was dissolved in 240 mL of 25% water in
1-propanol at 50 °C and the resulting solution was seeded. The
water concentration was reduced to 2% by azeotropic distil-
lation at reduced pressure while maintaining the volume
constant with the addition of 1-propanol. The slurry was cooled
to about 20 °C and the crystallized product was isolated by
filtration to give 25 g of solid after suction drying at ambient
temperature.
P ota ssiu m 4,6-Dia m in o-5-(2,2-d ieth oxyleth yl)p yr im i-
d in e-2-th iola te (32). The crude organic solution of 30 was
charged to a 100-L flask and concentrated to ∼8 L, and the
concentrate was flushed with 15 L of toluene. The concentrate
was flushed with 18 L of anhydrous ethanol. GC showed <5
vol % of toluene. The concentrate was diluted to 34 L with
anhydrous ethanol. Thiourea (2.60 kg, 34.2 mol) was added
in one portion and the slurry cooled with an ice bath.
Potassium tert-butoxide solid (3.84 kg, 34.2 mol) was added
over 15 min with 5.5 L of ethanol rinse and the slurry
exothermed to 43 °C. The flask was fitted with a reflux
condenser and reaction was warmed to 78 °C. Upon reaching
temperature the reaction first became homogeneous then, after
a 1-2 h age at 78 °C, a slurry formed. The reaction was aged
an additional 16.5 h at 78 °C at which point the slurry was
sampled. The sample was diluted with H2O prior to HPLC
assay, which showed 7.9 area % thiourea remaining (typically
5-10 area %). The reaction mixture was diluted with 7.6 L
more of ethanol and cooled to 21 °C over 90 min. After aging
1 h 15 min at 21 °C, the batch was filtered and the cake
washed with 38 L of ethanol in three portions. The wet cake
was dried under nitrogen flow with vacuum on the filter pot
to give 8.70 kg of off-white solid 32. HPLC analysis showed
95.6 area % and acid titration showed 93 wt % purity. Loss
on drying was 6.2 wt %. The corrected yield was 85%. HPLC
assay conditions: 3 mm × 3 cm C-8 column; temperature 35
°C; flow rate 2 mL/min; eluent 0.1% H3PO4/acetonitrile, hold
1 min at 100% H3PO4 then ramped to 75/25 acetonitrile/
H3PO4 over 6 min; UV 210 nm; retention times, thiourea 0.23
min, 33 0.7 min, 32 2.6 min, and disulfide 3.0 min.
(b) Hot Slu r r y of 1 w ith 1-p r op a n ol: The toluene solvate
form of nucleoside 1 (30 g) was dissolved in 240 mL of 2% water
in 1-propanol and the slurry was heated at 60-65 °C for 4 h.
After the solution was cooled to about 20 °C, the crystalline
solid was isolated by filtration to give 25 g of the nucleoside
as the unsolvated crystal form.
Analytical data for the nonsolvated 1:5 [R]365 -156.8 (MeOH,
4-Am in o-7H-p yr r olo[2,3-d ]p yr im id in e-2-th iol (33). To
a 100-L flask equipped with a mechanical stirrer, thermo-
couple, and N2 inlet were charged 40 L of H2O and 11.75 L of
5 N HCl. The potassium salt 32 (7.50 kg, 23.5 mol) was added
over 25 min with 7 L of H2O rinse and the solution was
warmed to 50 °C over 50 min. A slurry was formed on
warming. The resulting slurry was sampled after 10 min at
50 °C and the sample was diluted with water prior to HPLC
assay. Assay at this point showed complete consumption of
1
c 1.0 w/v %); H NMR (DMSO-d6, 600 MHz) δ 0.64 (s, 3 H),
3.64 (ddd, J ) 3.4, 5.0, and 12.1 Hz, 1 H), 3.80 (ddd, J ) 2.0,
5.0, and 12.1 Hz, 1 H), 3.83 (ddd, J ) 2.0, 3.4, and 8.9 Hz, 1
H), 3.93 (dd, J ) 7.2 and 8.9 Hz, 1 H), 5.02 (s, 1 H), 5.06 (d, J
) 7.2 Hz, 1 H), 5.09 (t, J ) 5.0 Hz, 1 H), 6.12 (s, 1 H), 6.55 (d,
J ) 3.5 Hz, 1 H), 6.98 (br s, 2 H), 7.45 (d, J ) 3.5 Hz, 1 H),
8.04 (s, 1 H); 13C NMR (DMSO-d6, 150 MHz) δ 19.59, 59.76,
72.05, 78.64, 82.07, 90.53, 99.47, 102.42, 121.30, 149.69,
J . Org. Chem, Vol. 69, No. 19, 2004 6265

Bio et al.
starting material. The reaction slurry was transferred to a
cooled vessel charged with a mixture of 12 L of H2O and 3.5 L
of 10 N NaOH. Mild exotherm following reagent addition
caused the reaction to warm to 42 °C. The resulting slurry
was cooled to 22 °C and then adjusted to pH 7, aged 3 h
(typically aged 1-3 h, aging longer than 3 h led to increasing
amounts of disulfide impurity), and filtered. The filter cake
was washed with 30 L of water followed by 36 L of acetonitrile
and dried under nitrogen to give 3.95 kg of off-white solid 33.
HPLC indicated 98.6 area % purity. Acid titration and loss on
drying indicated 90.5 wt % (100% yield). HPLC conditions were
the same as for 32. Retention times: thiourea 0.23 min, 33
0.7 min, 32 2.6 min, disulfide, 3.0 min.
MHz): δ 151.4, 148.1, 142.4, 125.3, 102.6, 101.6. Anal. Calcd
for C6H8N4O4S: C, 31.03; H, 3.47; N, 24.13. Found: C, 30.76;
H, 3.17; N, 23.92.
N-(7H-P yr r olo[2,3-d]pyr im idin -4-yl)ben zam ide (36). Ac-
etonitrile (100 mL), brine (50 mL), and 10 N NaOH (20 mL)
were combined in a 500-mL round-bottom flask. The flask was
degassed by two vacuum/nitrogen fill cycles. To the vigorously
stirred biphasic mixture was added 4-amino-7H-pyrrolo[2,3-
d]pyrimidine hydrogen sulfate (35) (26.58 g, 87.4 wt %, 100
mmol) at room temperature. The slurry was cooled to 15 °C
and benzoyl chloride (46 mL, 396 mmol) was added dropwise
with slight exotherm to 20 °C. With the mixture at 10-15 °C,
10 N NaOH (40 mL) was added dropwise over 25 min,
resulting in a clearly biphasic mixture. The reaction was aged
for 3 h with gradual warming to room temperature. The
mixture was cooled to 5-10 °C and 10 N NaOH (40 mL) was
added dropwise over 1 h. The reaction was warmed to room
temperature and the reaction monitored by HPLC until
hydrolysis was complete in 2 h. During this time, the product
began to precipitate at the interface. Ammonium acetate (7.67
g, 99 mmol) was added to bring the pH of the aqueous layer
to 10.2. The solid was isolated by filtration and washed with
water (3 × 100 mL), acetonitrile (2 × 100 mL), and ethanol
(50 mL). The cake was dried on the filter under nitrogen to
give a 92% yield of benzamide 36 (21.92 g, 96.6 wt %). 1H NMR
(DMSO-d6, 400 MHz): δ 12.05 (br s, 1H), 10.97 (br s, 1H), 8.53
(s, 1H), 8.04 (d, J ) 7.4, 2H), 7.58 (t, J ) 7.3 Hz, 1H), 7.50 (t,
J ) 7.4 Hz, 2H), 7.40 (d, J ) 3.3 Hz, 1H), 6.59 (d, J ) 3.3 Hz,
1H). 13C NMR (DMSO-d6, 100 MHz): δ 166.2, 153.7, 152.1,
151.3, 150.6, 134.2, 132.7, 128.9, 124.9, 109.3, 102.9.
4-Am in o-7H-p yr r olo[2,3-d ]p yr im id in e-2-su lfin ic Acid
(34). In a 100-L cylindrical vessel with heating/cooling coils,
33 (21.8 mol) was dissolved in 43.8 L of 1 N KOH and cooled
to 5 °C. Hydrogen peroxide, 30% solution (4.72 kg, 43.6 mol),
was added over 1.5 h, maintaining the reaction temperature
at 5-15 °C. (Note: The addition was done in shots at ∼150-
200 mL/min and the batch temperature was <13 °C during
the addition.) The resulting solution was aged at 10 °C for 1
h then warmed to 15 °C. HPLC assay indicated 98% conver-
sion. Acetic acid (6.8 L) was added over 20 min, resulting in
the formation of a thick slurry. The reaction was then heated
to 35-40 °C over 30 min, aged for 10 min, then cooled to 10
°C and aged for 20 min. The temperature cycle was done to
increase the particle size to give a filterable solid. The
crystalline solid was isolated by filtration and the resulting
filter cake was washed with water (3 × 8 L) then acetone (8
L). The cake was dried on the filter under nitrogen to give
4.24 kg (99.1% yield, 92.5 area%) of sulfinic acid 34 as a pale
yellow solid. HPLC conditions: C-8 column, 4.6 mm × 7.5 cm.
Mobile phase: (A) 0.1% H3PO4; (B) acetonitrile. Gradient:
100:0 A/B hold 4 min to 50:50 A/B over 6 min. Flow: 1.5 mL/
min. UV: 210 nm. Typical retention times: 4-amino-7H-
pyrrolo[2,3-d]pyrimine-2-sulfonic acid 2.1 min, 33 2.7 min, 34
2-(7H -P yr r olo[2,3-d ]p yr im id in 4-yl)isoin d olin e-1,3-d i-
on e (37). In a 72-L flask, 35 (5.79 kg, 24.9 mol) and phthalic
anhydride (7.46 kg, 50.3 mol) were added to 25 L of N,N-
dimethylacetamide. The mixture was degassed to nitrogen 3
times then heated to 70-75 °C. N,N-Diisopropylethylamine
(9.8 L, 55.2 mol) was added at a rate such that the batch
temperature was maintained at <85 °C. Following addition,
the resulting solution was aged at 80-85 °C for 5 h, then
allowed to cool to room temperature. The conversion as
determine by HPLC was 97%. The reaction was then further
cooled to 12 °C with an ice bath. Water (12.5 L) was added to
the resulting slurry over 1.5 h, maintaining the batch tem-
perature at <25 °C. The slurry was aged for 1 h after addition
at 12-15 °C. The solid product was isolated by filtration and
the resulting filter cake was washed with 16 L of 1:1 N,N-
dimethylacetamide/water, then 3 × 16 L of water. After some
drying, the cake was washed with 12 L of ethyl acetate. The
cake was dried on the filter under nitrogen to remove the bulk
of the water, then dried at 60 °C in the vacuum oven to a water
content of <0.2 wt % as determined by Karl-Fisher titration.
The product 37 was isolated as an off-white solid in 85.6% yield
(5.6 kg, 99.7 area % purity). The HPLC conditions were the
same as those for 34. Typical retention times: hydrogen sulfate
salt 2.2 min, phthalic acid 6.7 min, amino deazapurine
1
3.6 min. H NMR (DMSO-d6, KOD in D2O, 400 MHz): δ 7.04
(s, 1H), 6.21 (s, 1H). 13C NMR (DMSO-d6, KOD in D2O, 100
MHz): δ 168.8, 158.4, 156.5, 135.8, 104.9, 96.4. Anal. Calcd
for C6H6N4O2S: C, 36.36; H, 3.05; N, 28.27. Found: C, 36.13;
H, 2.93; N, 27.95.
4-Am in o-7H-p yr r olo[2,3-d ]p yr im id in e Hyd r ogen Su l-
fa te (35). To a 100-L flask with nitrogen sweep to a caustic
scrubber, the sulfinic acid 34 was added portion-wise to 56.7
kg of 50 wt % sulfuric acid at 20-30 °C over 1.5 h. The reaction
was aged at 20-30 °C for 1 h. Gas evolution occurred during
the addition and age. The reaction was then heated to 70-75
°C and aged 3 h, until dissolution was complete. (Note: The
hot age converts the residual 4-amino-7H-pyrrolo[2,3-d]py-
rimine-2-sulfonic acid impurity to 4-amino-7H-pyrrolo[2,3-d]-
pyrimidin-2-ol, which is more easily rejected in the crystalli-
zation. The sulfonic acid was undetected by HPLC after the
hot age.) The reaction was then cooled to 15-20 °C. Ethanol
was added over 45 min maintaining the temperature at <25
°C. The slurry was cooled to 10 °C and aged 20 min. The solid
was isolated by filtration. The cake was washed with 20 L of
cold 1:1 EtOH/50% sulfuric acid then with ethanol (4 × 8 L).
The solid was dried on the filter under nitrogen giving 7.5 kg
(84% yield, 98.3 wt %) of 35. The HPLC conditions were the
same as those for 34. Typical retention times: 4-amino-7H-
pyrrolo[2,3-d]pyrimidin-2-ol 1.8 min, 35 2.2 min, 34 3.6 min.
1H NMR (DMSO-d6, 400 MHz): δ 12.61 (s, 1H), 9.14 (br s, 1
H), 8.52 (br s, 1H), 8.27 (s, 1H), 7.36 (dd, J ) 3.3, 2.5 Hz, 1H),
6.87 (dd, J ) 3.3, 1.8 Hz, 1H). 13C NMR (DMSO-d6, 100
1
phthalimide 7.5 min. H NMR (DMSO-d6, 400 MHz): δ 12.14
(s, 1H), 8.79 (s, 1H), 8.01 (dd, J ) 5.5, 3.1 Hz, 2H), 7.92 (dd, J
) 5.5, 3.1 Hz, 2H), 7.63 (dd, J ) 3.5, 2.4 Hz, 1H), 6.56 (dd, J
) 3.5, 1.8 Hz, 1H). 13C NMR (DMSO-d6, 100 MHz): δ 166.3,
154.2, 151.0, 145.2, 135.7, 132.0, 128.7, 124.4, 115.2, 100.1.
Anal. Calcd for C14H8N4O2: C, 63.64; H, 3.05; N, 21.20.
Found: C, 63.33; H, 2.74; N 21.26.
J O0491096
6266 J . Org. Chem., Vol. 69, No. 19, 2004