Asymmetric synthesis of an N-acylpyrrolidine for inhibition of HCV polymerase

Article abstract of DOI:10.1021/jo800062c

(Chemical Equation Presented) A practical asymmetric synthesis of a highly substituted N-acylpyrrolidine on multi-kilogram scale is described. The key step in the construction of the three stereocenters is a [3+2] cycloaddition of methyl acrylate and an i

Full text of DOI:10.1021/jo800062c

Asymmetric Synthesis of an N-Acylpyrrolidine for Inhibition of
HCV Polymerase
Armel A. Agbodjan,† Bob E. Cooley,† Royston C. B. Copley,‡ John A. Corfield,§
Roy C. Flanagan,† Bobby N. Glover,† Rossella Guidetti,§ David Haigh,§ Peter D. Howes,§
Mary M. Jackson,† Richard T. Matsuoka,† Katrina J. Medhurst,§ Alan Millar,†
Matthew J. Sharp,† Martin J. Slater,*,§ Jennifer F. Toczko,† and Shiping Xie*,†
Chemical DeVelopment, GlaxoSmithKline, Research Triangle Park, North Carolina 27709, Analytical
Chemistry, GlaxoSmithKline, New Frontiers Science Park (North), Third AVenue, Harlow, Essex CM19
5AW, United Kingdom, and Infectious Diseases CEDD and Medicinal DiscoVery Research,
GlaxoSmithKline, GSK Medicines Research Centre, Gunnels Wood Road,
SteVenage SG1 2NY, United Kingdom
shiping.x.xie@gsk.com
ReceiVed January 10, 2008
A practical asymmetric synthesis of a highly substituted N-acylpyrrolidine on multi-kilogram scale is
described. The key step in the construction of the three stereocenters is a [3+2] cycloaddition of methyl
acrylate and an imino ester prepared from L-leucine t-butyl ester hydrochloride and 2-thiazolecarboxal-
dehyde. The cycloaddition features novel asymmetric catalysis via a complex of silver acetate and a
cinchona alkaloid, particularly hydroquinine, with complete diastereomeric control and up to 87%
enantiomeric control. The alkaloid serves as a ligand as well as a base for the formation of the azomethine
ylide or 1,3-dipole. Experiments have shown that the hydroxyl group of hydroquinine is a critical element
for the enantioselectivities observed. The cycloaddition methodology is also applicable to methylvinyl
ketone, providing access to either R- or â-epimers of 4-acetylpyrrolidine depending on the reaction
conditions utilized. The synthesis also highlights an efficient N-acylation, selective O- versus N-methylation,
and a unique ester reduction with NaBH4-MeOH catalyzed by NaB(OAc)3H that not only achieves
excellent chemoselectivity but also avoids formation of the undesired but thermodynamically favored
epimer. The highly functionalized target is synthesized in seven linear steps from L-leucine t-butyl ester
hydrochloride with all three isolated intermediates being highly crystalline.
Introduction
be infected with HCV.2 Following acute HCV infection, most
patients develop a slowly progressive chronic disease that can
eventually lead to cirrhosis and liver failure. The current standard
treatment for HCV infection is a combination therapy employing
some form of interferon-R, particularly the pegylated form, plus
ribavirin. Unfortunately, only about 50-80% of the patients
achieve sustained virological response as measured by a
reduction in serum HCV RNA levels and normalization of liver
Hepatitis C virus (HCV) infection has been identified as the
most common risk factor for developing hepatocellular carci-
noma and the main cause of adult liver transplants in developed
nations.1 Worldwide, over 170 million people are estimated to
† Chemical Development.
‡ Analytical Chemistry.
§ Infectious Diseases CEDD and Medicinal Discovery Research.
(1) Willems, M.; Metselaar, H. J.; Tilanus, H. W.; Schalm, S. W.; de
Man, R. A. Transplant Int. 2002, 15, 61.
(2) (a) Wasley, A.; Alter, M. J. Semin. LiVer Dis. 2000, 20, 16. (b) Alter,
M. J. Hepatology 1997, 26, 62S.
10.1021/jo800062c CCC: $40.75 © 2008 American Chemical Society
Published on Web 03/22/2008
3094
J. Org. Chem. 2008, 73, 3094-3102

N-Acylpyrrolidine for Inhibition of HCV Polymerase
FIGURE 1. Strategy for synthesis of 1.
enzymes. The response rate is about 50% for infections with
HCV of genotype 1, the prevalent type of HCV infection in
the United States.3 In addition, the current treatment is also
accompanied with significant systemic side effects. Clearly, new
therapies are needed to address the issues of safety, broad
antiviral response, and viral resistance mutations.
Several pharmaceutical companies, in search of a new
treatment for HCV infections, are pursuing an approach that
inhibits the HCV polymerase which synthesizes the new viral
RNA strands.4 Compound 1, a highly substituted N-acylpyrro-
lidine with three stereocenters, is a non-nucleoside small
molecule which acts as a potent inhibitor of RNA-dependent
RNA polymerase (NS5B) in enzymatic assays and inhibits viral
RNA replication in cell-based replicon assays.5
(2), which is prepared from commercially available 2-t-butyl-
3-methylphenol by methylation followed by oxidation with
potassium permanganate.6 The overall efficiency of the synthesis
depends on the efficiency in preparation of B with the three
stereocenters and the level of selectivity in the ester reduction.
A common approach to substituted pyrrolidine like B is the
[3+2] cycloaddition of a N-metalated azomethine ylide 1,3-
dipole such as C with an electron deficient dipolarophile such
as acrylic ester D.7
We herein report a practical and efficient synthesis of target
1, which includes diastereo- and enantioselective synthesis of
the acylpyrrolidine moiety through a [3+2] cycloaddition
catalyzed by a novel silver-cinchona complex, a unique and
highly selective reduction of a carboxylic ester group with
NaBH4-MeOH catalyzed by NaB(OAc)3H, as well as other
efficient downstream functionalizations of the pyrrolidine core.
Results and Discussion
Preparation of imino ester 3, the precursor to the N-metalated
azomethine ylide, is shown in Scheme 1. Condensation of
SCHEME 1
Our synthetic strategy is shown in Figure 1. The C2
carboxylic acid group in the target molecule (1) is protected as
a bulky tert-butyl ester (intermediate A) throughout the syn-
thesis. The C4 methoxymethyl group of 1 is derived from
selective reduction of an ester precursor (CO2R) in the presence
of two other acyl groups in A. Intermediate A is obtained from
the acylation of a pyrrolidine moiety (B) with a benzoic acid
L-leucine t-butyl ester hydrochloride with 2-thiazolecarboxal-
dehyde in the presence of triethylamine afforded imino ester 3
in 81% yield. The stereocenter in 3 is irrelevant because of the
subsequent formation of the planar 1,3-dipole. Thus L-leucine
was used for its low cost and wide availability. No dehydrating
reagent was needed under this set of reaction conditions. While
there have been many reports of racemic synthesis of substituted
pyrrolidines from an imino ester like 3,7 the enantioselective
synthetic strategy using a substoichiometric amount of a catalyst
is relatively new. The most notable ligands for the enantiose-
lective catalysis via cycloadditions of N-metalated azomethine
ylides include chiral bisferrocenyl amide phosphines (FAP) from
Zhang’s group,8 chiral bisoxazolines from Jørgensen’s group,9
and QUINAP P,N-ligands from Schreiber’s group.10 These
(3) (a) Fried, M. W. Hepatology 2002, 36, 237. (b) Cornberg, M.;
Wedemeyer, H.; Manns, M. P. Curr. Gastroenterol. Rep. 2002, 4, 23. (c)
Hu¨gle, T.; Cerny, A. ReV. Med. Virol. 2003, 13, 361.
(4) For recent reviews on the progress and development of small-molecule
HCV antivirals, see: (a) Nie, Z-J.; Wagman, A. S. Curr. Opin. Drug
DiscoVery DeV. 2004, 7, 446. (b) Beaulieu, P. L.; Tsantrizos, Y. S. Curr.
Opin. InVest. Drugs 2004, 5, 838. (c) Wu, J. Z.; Yao, N.; Walker, M.; Hong,
Z. Mini-ReV. Med. Chem. 2005, 5, 1103.
(5) (a) Slater, M. J.; Amphlett, E. M.; Andrews, D. M.; Bravi, G.; Burton,
G.; Cheasty, A. G.; Corfield, J. A.; Ellis, M. R.; Fenwick, R. H.; Fernandes,
S.; Guidetti, R.; Haigh, D.; Hartley, C. D.; Howes, P. D.; Jackson, D. L.;
Jarvest, R. L.; Lovegrove, V. L. H.; Medhurst, K. J.; Parry, N. R.; Price,
H.; Shah, P.; Singh, O. M. P.; Stocker, R.; Thommes, P.; Wilkinson, C.;
Wonacott, A. J. Med. Chem. 2007, 50, 897. (b) Burton, G.; Ku, T. W.;
Carr, T. J.; Kiesow, T.; Sarisky, R. T.; Lin-Goerke, J.; Baker, A.; Earnshaw,
D. L.; Hofmann, G. A.; Keenan, R. M.; Dhanak, D. Bioorg. Med. Chem.
Lett. 2005, 15, 1553. (c) Burton, G.; Ku, T. W.; Carr, T. J.; Kiesow, T.;
Sarisky, R. T.; Lin-Goerke, J.; Hofmann, G. A.; Slater, M. J.; Haigh, D.;
Dhanak, D.; Johnson, V. K.; Parry, N. R.; Thommes, P. Bioorg. Med. Chem.
Lett. 2007, 17, 1930.
(6) 4-(1,1-Dimethylethyl)-3-(methyloxy)benzoic acid is available from
CarboGen, Bubendorf, Switzerland, through custom synthesis.
(7) For reviews, see: (a) Grigg, R.; Sridharan, V. AdV. Cycloaddit. 1993,
3, 161. (b) Gothelf, K. V.; Jørgensen, K. A. Chem. ReV. 1998, 98, 863. (c)
Gothelf, K. V. In Cycloaddition Reactions in Organic Synthesis; Kabayashi,
S., Jørgensen, K. A., Eds.; Wiley-VCH: Weinheim, Germany, 2002; pp
211-245.
J. Org. Chem, Vol. 73, No. 8, 2008 3095

Agbodjan et al.
SCHEME 2
ligands, when used with silver acetate or zinc triflate, provide
excellent selectivities for certain imine substrates usually derived
from glycine. Tertiary amines such as N,N-diisopropylethy-
lamine or triethylamine are used as a base along with these cited
catalysts. However, these ligands are not readily available and
can be costly when used as a catalyst in industrial-scale
production. In addition, leucine-derived imines are much less
reactive toward a dipolarophile than glycine-derived imines.
A common feature of all of the reported ligands is their
bidentate chemical structure. This feature brought to our
attention the cinchona alkaloids which have an amino group
and an adjacent hydroxyl group. Examples of some cinchona
alkaloids include cinchonidine, quinine, and hydroquinine, as
shown in Figure 2. In addition to the chelation potential of the
need to add an achiral tertiary amine such as N,N-diisopropy-
lethylamine and triethylamine as used in the reported catalytic
systems.8-10 The availability of the pseudo enantiomers and the
abundant supply at low cost of those alkaloids were also
attractive for the industrial use that we were pursuing.11 We
were also interested in knowing what would happen if the
hydroxyl group of the cinchona alkaloids was capped as an ester
such 4-chlorobenzoate (Figure 2). Other alkaloids such as (-)-
sparteine and (-)-strychnine were also included in the screening
for comparison.
For the proof of concept studies, three reactions with imine
3 and methyl acrrylate were set up at ambient temperature, each
containing 10 mol % cinchonidine (Scheme 2). When no metal
salt was added (reaction 1), cycloaddition occurred but was only
about 50% complete after several days. Essentially no enan-
tioinduction was seen even though cinchonidine served as the
catalytic base. Reaction times were decreased to 2 h with the
addition of 1.5 equiv of lithium bromide, but the presence of
lithium bromide completely eliminated all enantioselectivity.
For the third reaction, addition of 10 mol % silver acetate
significantly increased the reaction rate. The reaction was
complete within 1 h after addition of silver acetate. Furthermore,
the reaction was cleaner, and the enantiomeric ratio of the
cycloadduct was 7:3 as measured by chiral HPLC analysis.12
In all reactions, only the C4 R-epimer was detected by HPLC-
MS analysis, which indicated excellent diastereoselectivity. The
C4 R-epimer corresponds to the addition of methyl acrylate to
the 1,3-dipole derived from the imino ester via an endo transition
state. For the reaction with silver acetate and cinchonidine, both
(11) Jørgensen and co-workers recently published their work on cy-
cloadditions of glycine-derived imino esters with acrylates catalyzed by
hydrocinchonine and silver fluoride. See: Alemparte, C.; Blay, G.;
Jørgensen, K. A. Org. Lett. 2005, 7, 4569. Our work involving glycine as
well as the less reactive leucine-derived imines was independently initiated
in the fall of 2002. The results from our study including many more
examples varying amino acids and aldehydes were presented in the
GlaxoSmithKline International Chemistry Symposium at Tonbridge, U.K.,
in 2004, and a part of it was published in an internal journal (Xie, S.;
Agbodjan, A. A.; Cooley, B. E.; Flanagan, R. C.; Glover, B. N.; Jackson,
M. M.; Matsuoka, R. T.; Sharp, M. J.; Toczko, J. F. Chemicus 2004, 20,
3).
(12) Conditions for chiral analysis of 4 and 6a were as follows.
Column: Chiracel OD-H, 4.6 × 250 mm, 5 micron; mobile phase: 90:10
heptane/i-PrOH; flow rate: 1 mL/min; detection: 230 nm; temperature:
20 oC; retention time: 7.0 min for 4 and 5.9 min for ent-4, 5.4 min for 6a,
and 6.8 min for ent-6a.
FIGURE 2. Structures of some alkaloids screened for the cycloaddi-
tions.
hydroxyl and amino groups, the tertiary amino group could
function as a base for the formation of ylides or four-electron
1,3-dipoles. Such a built in basic moiety would eliminate the
(8) Longmire, J. M; Wang, B.; Zhang, X. J. Am. Chem. Soc. 2002, 124,
13400.
(9) Gothelf, A. S.; Gothelf, K. V.; Hazell, R. G.; Jørgensen, K. A. Angew.
Chem., Int. Ed. 2002, 41, 4236.
(10) Chen, C.; Li, X.; Schreiber, S. L. J. Am. Chem. Soc. 2003, 125,
10174.
3096 J. Org. Chem., Vol. 73, No. 8, 2008

N-Acylpyrrolidine for Inhibition of HCV Polymerase
SCHEME 3
SCHEME 4
the rate acceleration and the enantioinduction clearly indicated
the presence of chiral catalysis. The absolute stereochemistry
of 4 was subsequently confirmed by X-ray analysis on the
crystal of a derivative of 4, namely, compound 10 (Scheme 4).
The Ag-cinchonidine catalysis worked similarly for the
cycloaddition of methylvinyl ketone to give predominantly the
C4 R-epimer 5 (R/â ratio 98:2); this result suggests the
involvement of the same endo addition as observed in the
acrylate addition. We found that the C4 R-epimer readily
converted to the â-epimer 6a when stirred with 10 mol % 1,8-
diazabicyclo[5.4.0]undec-7-ene (DBU) in ethyl acetate at 40 °C
(â/R ratio >95:5). The enantiomeric ratio of 6a was 73:27 by
chiral HPLC analysis.12 The hydrochloride salt 6b was prepared
for analysis. The 5/6a ratio was determined by HPLC-MS
analysis. The structures of compounds 5 and 6b were assigned
based on extensive NMR studies including gDQCOSY (gradient
temperature. Compound 6 was of interest to our drug discovery
program; however, access to 6 had proven extremely difficult
under other conditions including racemic synthesis with lithium
bromide. In contrast, the C4 epimerization of cycloadduct 4
required much stronger conditions such as heating in refluxing
THF with powdered potassium carbonate. From the data
obtained, we propose that the endo cycloadduct is kinetically
favored for both the acrylate and the vinyl ketone dipolarophiles
and that the â-epimer, which corresponds to exo addition, is
the thermodynamic product. Thus, the cycloaddition, when
combined with the base-catalyzed epimerization, in essence,
provides access to both endo and exo products depending on
the conditions applied.
With the concept of a metal alkaloid catalysis proven, we
next proceeded to optimize the cycloaddition reaction. We first
screened a number of metal salts including Mg2+, Co3+, Mn2+
,
double quantum filtered correlation spectroscopy), H {13C}
Zn2+, and Li+ with various counterions. The use of these metal
salts resulted in much lower enantioselectivity and,13 in the
reactions employing CoBr3, CoCl3, and MnCl2, less clean
reactions due to formation of impurities including the undesired
C4 â-epimer. In the end, Ag+ was, by far, the best metal salt to
use with respect to both diastereo- and enantioselectivities. As
a result, our efforts were focused on silver salts for further
optimization of the cycloaddition. The screening of silver salts
was carried out with 10 mol % cinchonidine and 5 mol % silver
1
gHSQC (gradient heteronuclear single quantum coherence), and
GOESY (gradient enhanced nuclear Overhauser effect spec-
troscopy). These studies clearly indicated that the substituents
on positions 4 and 5 of the pyrrolidine ring are geometrically
cis for 5 and trans for 6b. Further details of this analysis are
provided in the Supporting Information. In a scale-up using 10
g of imine 3, the epimerization from 5 to 6 was cleanly carried
out in just 6 h with a stoichiometric amount of DBU at ambient
J. Org. Chem, Vol. 73, No. 8, 2008 3097

Agbodjan et al.
TABLE 1. Cycloaddition of Imine 3 and Methyl Acrylate: Silver
Salt Screeninga
TABLE 2. Cycloaddition of Imine 3 and Methyl Acrylate: Solvent
Screeninga
entry
AgX, X )
solvent
enantio ratio of 4
enantio
ratio
enantio
ratio
entry
solvent
toluene
chlorobenzene
benzeneb
anisole
i-PrOAc
THF
p-xylene
entry
solvent
1
2
3
4
5
6
7
8
9
CF3CO2
CF3CO2
OTf
THF
toluene
THF
toluene
THF
toluene
THF
toluene
THF
toluene
60:40
73:27
63:37
72:28
68:32
73:27
56:44
72:28
65:35
73:27
1
2
3
4
5
6
7
75:25
72:28
72:28
70:30
69:31
69:31
68:32
8
9
1,4-dioxane
MTBE
CH2Cl2
mesitylene
trifluorotoluene
isooctane
MeCN
66:34
65:35
63:37
62:38
60:40
59:41
50:50
OTf
10
11
12
13
14
OAc
OAc
PF6
PF6
NO3
a The general procedure for screening was followed, and the enantio ratio
was analyzed by chiral HPLC.12,14 b The reaction in benzene was run at
7 °C.
10
NO3
a The general procedure for screening was followed, and the enantio ratio
was analyzed by chiral HPLC.12,14
TABLE 3. Cycloaddition of Imine 3 and Methyl Acrylate:
Alkaloid Screeninga
salt in two solvents, with the results shown in Table 1. Of the
five silver salts screened, it was found that there was not much
difference among the various silver salts in regards to the
enantioselectivity when the same solvent was used. All salts
gave enantio ratios of about 6:4 in THF and 7:3 in toluene.
Silver acetate was chosen for alkaloid optimization because it
gave a faster and slightly cleaner reaction probably because of
its better solubility in THF and toluene.
The choice of solvent had a dramatic impact on the enanti-
oselectivity in the cycloaddition. Results from screening of 14
solvents with cinchonidine as the ligand are shown in Table 2.
In general, solvents containing a phenyl group such as toluene,
chlorobenzene, benzene, and anisole (entries 1-4) gave better
selectivity. On the other hand, too much substitution on the
phenyl ring, as in p-xylene and mesitylene (entries 7 and 11),
lowered the selectivity. Ethers such as THF, methyl t-butyl ether
(MTBE), and 1,4-dioxane (entries 6, 8 and 9) were not good
solvents, even though THF was the solvent initially used in the
study. Interestingly, the use of acetonitrile as the solvent (entry
14) gave no enantioselectivity at all.
The alkaloid screening was conducted with 10 mol %
alkaloids and 5 mol % silver acetate, with the results shown in
Table 3. Pseudo enantiomeric alkaloids gave opposite enanti-
oselectivities (entries 1 and 2), as did quinine and quinidine
(entries 3 and 4). The use of hydroquinine improved the
selectivity to above 80% (entry 5). Interestingly, addition of 50
wt % 4 Å molecular sieves (MS) increased the enantioselectivity
to 87% (entry 6). We postulate that molecular sieves facilitate
the exchange of imine 3 and cycloadduct 4 from the silver-
ligand complex and thus increase the catalytic efficiency and
reduce, relatively, the various background reactions. When the
conditions for entry 6 were applied to t-butyl acrylate, the
enantiomeric ratio increased to 92:8. However, the use of C4
t-butyl ester would theoretically make the subsequent selective
ester reduction in the presence of the N-acyl group and the C2
t-butyl ester (see Scheme 4) more difficult. For practicality, we
focused our efforts on methyl acrylate, despite the 5% lower
selectivity. Minimum enantiomeric induction was observed
enantio
entry
alkaloid
cinchonidine
cinchonine
quinine
quinidine
hydroquinine
hydroquinine
solvent
THF
THF
THF
ratio of 4
1
2
3
4
5
6
69:31
33:67
62:38
31:69
81:19
87:13
THF
toluene
toluene with
4 Å MS
THF
THF
toluene
7
8
9
(-)-strychnine
(-)-sparteine
hydroquinine 4-
chlorobenzoate
hydroquinine 1,4-
phthalazinediyl diether
45:55
46:54
57:43
10
toluene
68:32
a The general procedure for screening was followed, and the enantio ratio
was analyzed by chiral HPLC.12,14
(entries 7 and 8) when using (-)-strychnine or (-)-sparteine as
the ligand. This is presumably due to lack of a hydroxyl group
and hence an inability to form a bidentate complex with the
silver cation. As expected, when the hydroxyl group of
hydroquinine was capped as an ester (entry 9) or as an ether
(entry 10), the selectivity was significantly lower by as much
as 24% compared to ratios obtained with hydroquinine (entry
5). We believed that the hydroxyl group of the cinchona
alkaloids is an important component of the ligand structure for
the silver-cinchona catalyst. It is possible that the hydroxyl
group and the amino group of the alkaloid chelate to the silver
ion for a more stable complex of the catalyst.
The process as shown in Scheme 3 was scaled up in 500
gallon reactors. The HCl salt of L-leucine t-butyl ester reacted
with the 2-thiazolecarboxaldehyde in the presence of triethy-
lamine to give imine 3. Although 3 is crystalline, this compound
is a hygroscopic and low melting solid. For the sake of
practicality, the crude 3 from the aqueous workup and partial
concentration in vacuo to remove triethylamine was used for
cycloaddition directly without any purification. The catalyzed
[3+2] cycloaddition was carried out with 6 mol % hydroquinine
(13) Zinc triflate, in combination with (S,S)-2,2′-isopropylidene-bis(4-
t-butyl-2-oxazoline) (t-Bu-BOX, 10 mol %) and triethylamine (10 mol %),
as described by Jørgensen and coworkers,9 did not yield any enantioselec-
tivity when applied to the cycloaddition of imine 3 and methyl acrylate in
THF.
3098 J. Org. Chem., Vol. 73, No. 8, 2008

N-Acylpyrrolidine for Inhibition of HCV Polymerase
and 3 mol % silver acetate in the presence of 50 wt % 4 Å
molecular sieves at -10 °C. The enantioselectivity was 85:15,
slightly lower than the typical 87:13 ratio that we were able to
consistently achieve with isolated crystalline imine 3. Cycload-
duct 4 as a crude toluene solution was partially concentrated
and diluted with isopropanol. For the purpose of isolation, (R)-
1,1′-binapthyl-2,2′-dihydrogen phosphate was used to form
crystalline salt 7 in 57% overall yield from L-leucine t-butyl
ester hydrochloride. Use of the chiral acid not only secured
excellent recovery of the desired cycloadduct 4, but it also
enhanced the enantiomeric ratio of the product to 99.9:0.1. On
the smaller scale with one-liter fixed equipment, an overall yield
as high as 65% was achieved for the three steps. Compared
with the racemic synthesis that we scaled up to 50 L to support
the early research program, the asymmetric route as shown in
Scheme 3 more than doubled the overall yield, and the process
was extremely robust and consistently reproducible on a large
scale.
Completion of the synthesis of the target N-acylpyrrolidine
(1) is shown in Scheme 4. The slurry of salt 7 in MTBE was
treated with 1 equiv of triethylamine to provide free base 4 as
a solution in MTBE. (R)-Binaphthyl hydrogenphosphate was
fully recovered as the crystalline triethylamine salt via filtration.
For the formation of acid chloride 8 from carboxylic precursor
2 and SOCl2 in CH2Cl2, the use of a catalytic amount (0.25
equiv) of pyridine to ensure that conversion to the acid chloride
was reproducible proved to be vital. The formation of the highly
reactive and corrosive acid chloride was monitored by online
ReactIR. This technique offered advantages in safety and
accuracy compared with the traditional analysis of an amide
derivative by HPLC. The carbonyl stretch at 1691 cm-1 was
used to monitor the carboxylic acid 2, and the carbonyl stretches
at 1743 and 1768 cm-1 were used to monitor the formation of
the acid chloride 8. Acylation of free base 4 with acid chloride
8 provided crystalline N-acylpyrrolidine 9 in up to 88% yield.
Reduction of the methyl ester to alcohol 10 was a significant
challenge. Both the amide bond and the t-butyl ester were
susceptible to reduction by a variety of reducing agents. In
addition to the chemoselectivity, the C4 position of 9 is prone
to epimerization to the thermodynamically more stable â-epimer
prior to the methyl ester reduction. As a result, common reagents
for reduction of an ester to an alcohol such as i-Bu2AlH, LiBH4,
Super-hydride, and NaBH4 in combination with various alcohols
led to a large amount of overreduction and C4 epimerization
on the multigram scale. The use of LiAlH4 was successful on
a small scale when the reaction temperature was carefully
controlled within a narrow and low range.5a Unfortunately, this
LiAlH4-based reduction, on scale up to just under 100 grams,
led to the formation of over 50% of 4-(1,1-dimethylethyl)-3-
(methyloxy)benzaldehyde, a byproduct from the partial reduction
of the amide bond of ester 9 or alcohol 10. HPLC-MS was used
to monitor and analyze the ratios of 9 and 10 versus their
respective epimers.15 After extensive reagent screening and
process optimization, we came up with a unique set of reduction
FIGURE 3. Atomic displacement ellipsoid plot (50% probability) for
10 from the crystal structure. A molecule of acetonitrile, also present
in the asymmetric unit and with partial occupancy, has been omitted
for clarity.
conditions involving NaBH4-MeOH (1:2) in the presence of a
catalytic amount of NaB(OAc)3H (2.5 mol % relative to
NaBH4). These conditions were an extension of previously
reported work by our colleagues on the reduction of a t-
butoxycarbonyl (BOC)-protected amide with NaBH4-MeOH
in the presence of acetic acid.16 We postulated that conversion
of NaBH4 to NaB(OMe)4-nHn (n ) 1-3) was catalyzed by
NaB(OAc)3H. To our surprise, the commercially available NaB-
(OMe)3H led to up to 50% C4 epimerization in the reduction
of 9 to 10. We believed that the in-situ-formed NaB(OMe)4-nHn
was more reactive and less basic, contributing to both the fast
and essentially epimerization-free reduction.17 The reduction was
carried out conveniently by charging the three solids (9, NaBH4,
and NaB(OAc)3H) to the reactor followed by the sequential
addition of THF and methanol with stirring. The selective
reduction was complete within 5 h at 25 °C, affording 10 as a
crystalline solid in 89% yield (as high as 97% if a second crop
was included). A molecule of alcohol 10 from a single-crystal
X-ray diffraction study is shown in Figure 3. The structure
confirmed the structural assignment for the [3+2] cycloaddition
as well as the retention of the integrity of the stereocenter from
the selective reduction.18
The O-methylation of 10 proceeded smoothly with io-
domethane or dimethyl sulfate and a strong base such as 1 equiv
of sodium hydride or 2 equiv of sodium t-butoxide, as shown
(15) (a) Conditions for HPLC analysis of 9 and its â-epimer were as
follows. Column: Phenomenex, Develosil Diol 100A, 4.6 × 250 mm, 5
micron; mobile phase: 85:15 heptane/EtOAc; flow rate: 1.5 mL/min;
detection: 250 nm; temperature: 25 °C; retention time: 5.95 min for 9
and 5.05 min for its â-epimer. (b) Conditions for HPLC analysis of 10 and
its â-epimer were as follows. Column: Phenomenex, Develosil Diol 100A,
4.6 × 250 mm, 5 micron; mobile phase: 60:40 heptane/EtOAc; flow rate:
1.0 mL/min; detection: 250 nm; temperature: 25 °C; retention time: 7.59
min for 10 and 5.26 min for its â-epimer.
(16) Daluge, S. M.; Martin, M. T.; Sickles, B. R.; Livingston, D. A.
Nucleosides, Nucleotides Nucleic Acids 2000, 19, 297.
(17) The process research work leading to the discovery of this unique
set of reduction conditions and the solution to process safety issues identified
during reaction calorimetry assessment will be the subject of a future
publication.
(14) General procedure for screening: To a solution of 282 mg (1.00
mmol) of imino ester 3 in 10 mL of toluene or another solvent was added
29.4 mg (0.10 mmol) of cinchonidine or equivalent mmolar quantities of
another alkaloid, followed by addition of 8.3 mg (0.05 mmol) of silver
acetate or equivalent mmolar quantities of another silver salt. After being
stirred with ice cooling for 15 min, the mixture was treated with 0.10 mL
(1.10 mmol) of methyl acrylate. The mixture was stirred for 30 min with
ice cooling and was allowed to gradually warm to ambient temperature.
Upon completion of the reaction (1-12 h), the reaction was quenched with
water. The organic layer was sampled for analysis by chiral HPLC.12
(18) The numbering method for the X-ray structure in Figure 3 is different
from the one used in the rest of the structure drawings and text.
J. Org. Chem, Vol. 73, No. 8, 2008 3099

Agbodjan et al.
2-(1,1-Dimethylethyl) 4-Methyl (2S,4S,5R)-2-(2-Methylpro-
pyl)-5-(1,3-thiazol-2-yl)-2,4-pyrrolidinedicarboxylate (4): A gen-
eral procedure for cycloaddition of 3 and methyl acrylate to form
4 as used in the screening is described in the note.14 For the
synthesis of target 1, free base 4 was not isolated. It was
conveniently converted to salt 7 for purification and isolation. See
the procedure for compound 7 for details.
in Scheme 4. When a weak base such as potassium hydroxide
or cesium carbonate was used, up to 50% quarternization of
the thiazole nitrogen occurred. The formation of the more
nucleophilic alkoxide in the presence of the strong base was
critical to avoid the undesired N-methylation. The methylation
was over 99% complete within 1 h after the reaction was
warmed from -30 °C to room temperature. For an unknown
reason, we could not achieve 100% methylation with 2 equiv
of methyl iodide, and the unreacted residual alcohol 10 could
not be removed by recrystallization. As a result, the initial
isolated methyl product 11 was subject to the methylation
process with additional 0.2 equiv of methyl iodide to convert
the residual amount (<1%) of 10 to 11. Finally, the hydrolysis
of the crude t-butyl ester was facile with concentrated aqueous
hydrochloric acid in MTBE. The target 1 was isolated as a
crystalline solid in 66% overall yield for the two steps from
10.
An analytical sample of 4 was obtained by neutralization of
compound 7, which was the salt of 4 and (R)-1,1′-binapthyl-2,2′-
dihydrogen phosphate, with triethylamine in MTBE. The procedure
for this free base preparation is described in the procedure for 9.
Crystallization from the concentrated MTBE solution provided 4
as a crystalline solid; enantiomeric ratio by chiral HPLC12 > 99.9:
0.1 favoring target enantiomer 4: [R]25 -29.9 (c 1.67, CHCl3);
D
1H NMR (300 MHz, CDCl3) δ 7.66 (d, J ) 3.0 Hz, 1H), 7.22 (d,
J ) 3.0 Hz, 1H), 4.87 (d, J ) 6.0 Hz, 1H), 3.44 (s, 3H), 3.48 (m,
1H), 3.30 (br s, 1H), 2.70 (m, 1H), 2.06 (m, 1H), 1.75 (m, 2H),
1.60 (m, 1H), 1.47 (s, 9H), 0.95 (d, J ) 6.0 Hz, 3H), 0.89 (d, J )
6.0 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 174.7, 172.3, 171.3,
142.7, 119.2, 100.2, 81.6, 69.0, 61.9, 51.9, 48.9, 39.0, 28.2, 25.4,
24.5, 23.7. Anal. Calcd for C18H28N2O4S: C, 58.67; H, 7.66; N,
7.60, S, 8.70. Found: C 58.47; H, 7.62; N, 7.57, S, 8.68.
Conclusions
An efficient synthesis of the highly substituted N-acylpyrro-
lidine target (1) was achieved in seven linear steps from
L-leucine t-butyl ester hydrochloride. The synthesis features a
novel asymmetric synthesis of pyrrolidines through a [3+2]
cycloaddition of an azomethine ylide with methyl acrylate
catalyzed by a complex of silver acetate and a cinchona alkaloid.
The cycloaddition is completely diastereoselective and 87%
enantioselective with readily available hydroquinine. The al-
kaloid serves as a base for the formation of the azomethine ylide
or 1,3-dipole as well as a ligand for the catalyst. The cycload-
dition methodology also proved applicable to methylvinyl
ketone, providing access to either R- or â-epimer of the
4-acetylpyrrolidine, depending on the reaction conditions uti-
lized. A unique NaBH4-MeOH reduction condition with
catalysis by NaB(OAc)3H was developed to secure a mild and
selective reduction of an ester to an alcohol. With all three
isolated intermediates being highly crystalline solids, the
synthesis is practical and has been scaled up to 500 gallon
reactors to produce multi-kilogram quantities of 1.
1,1-Dimethylethyl (4S,5R)-4-Acetyl-2-(2-methylpropyl)-5-(1,3-
thiazol-2-yl)-l-prolinate (5) and 1,1-Dimethylethyl (4R,5R)-4-
Acetyl-2-(2-methylpropyl)-5-(1,3-thiazol-2-yl)-l-prolinate (6): A
1 L round-bottom flask was charged with 10.0 g (35.4 mmol) of
imine 3, 300 mL of THF, 1.05 g (3.54 mmol) of (-)-cinchonidine,
and 600 mg (3.54 mmol) of AgOAc. The reaction mixture was
cooled to -20 °C, and 3.6 mL (42.5 mmol) of methylvinyl ketone
was added. The reaction mixture was allowed to warm to 0 °C.
After 4 h, 100 mL of MTBE and 100 mL of H2O were added, and
the layers were separated. The organic layer was washed with 100
mL of H2O and 100 mL of brine successively, dried over MgSO4,
and concentrated to give 12.5 g of crude cis product 1,1-
dimethylethyl (4S,5R)-4-acetyl-2-(2-methylpropyl)-5-(1,3-thiazol-
2-yl)-L-prolinate (5) as a thick oil. The ratio of 5 to its C4 â-epimer
was determined to be 98:2 by HPLC analysis on the crude product
(HPLC conditions: Column: Luna C18(2) 50 × 2 mm, 3 µm;
mobile phase A: H2O (0.05% TFA), mobile phase B: MeCN
(0.05% TFA); gradient: 0-95% B over 8 min; detection: 220 nm;
temperature: 40 °C; retention time: 4.07 min for 5 and 4.68 min
for the â-epimer). The crude oil was dissolved in 250 mL of THF,
and 6.50 mL (42.5 mmol) of DBU was added. After being stirred
at room temperature for 6 h, the HPLC analysis showed conversion
to the C4 â-epimer (6a). The mixture was treated with 100 mL of
20% aqueous NH4Cl and 150 mL of EtOA. The organic layer was
washed with an additional 100 mL of 20% NH4Cl, dried over
MgSO4, and concentrated in vacuo to give 12.4 g of trans product
Experimental Section
1,1-Dimethylethyl N-(1,3-Thiazol-2-ylmethylidene)-l-leucinate
(3): To a suspension of 50.0 g (223 mmol) of L-leucine t-butyl
ester hydrochloride in 500 mL of toluene was successively added
19.6 mL (223 mmol) of 2-thiazolecarboxaladehyde and 31.8 mL
(228 mmol) of triethylamine. The reaction was heated at 50 °C for
3 h. The reaction was cooled to ambient temperature and diluted
with 40 mL of MTBE and 200 mL of water. The organic layer
was successively washed with 150 mL of water and 200 mL of
saturated brine, dried with anhydrous sodium sulfate, and evaporated
in vacuo to near dryness. The thick oil was diluted with 50 mL of
MTBE and treated with about 100 mL of heptane. After being
cooled at 0 °C for 4 h, the solids were filtered, washed with 50 mL
of heptane, and dried at 50 °C overnight to give 32.1 g (51%) of
imine 3 as a light-yellow crystalline solid. The combined filtrate
and washing were concentrated to about 50 mL and stood at ambient
temperature for 2 days. The resultant solids were filtered, washed
6a as a yellow oil (er 73:27 by chiral HPLC analysis).12
A
hydrochloride salt (6b) was obtained by stirring a mixture of 353
mg (1.00 mmol) of 6a and 0.5 mL (2.00 mmol) of 4 M HCl in
1,4-dioxane in 5 mL of MTBE. Filtration and washing with cold
MTBE provided 6b as a light-yellow crystalline solid. The structures
of compounds 5 and 6b were assigned based on extensive NMR
studies including gDQCOSY, 1H {13C} gHSQC, and GOESY
experiment. Details are included in the Supporting Information.
These studies clearly indicated that the substituents on positions 4
and 5 of the pyrrolidine ring were cis for 5 and trans for 6b. 5: 1H
NMR (300 MHz, CDCl3) δ 7.64 (d, J ) 3.0 Hz, 1H), 7.22 (d, J )
3.0 Hz, 1H), 4.98 (d, J ) 9.0 Hz, 1H), 3.54 (m, 1H), 3.04 (br s,
1H), 2.70 (m, 1H), 2.01 (s, 3H), 1.98 (m, 1H), 1.76 (m, 2H), 1.60
(m, 1H), 1.47 (s, 9H), 0.95 (d, J ) 6.0 Hz, 3H), 0.88 (d, J ) 6.0
Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 207.3, 175.0, 171.7, 142.6,
119.5, 81.6, 68.7, 62.0, 56.0, 49.6, 38.3, 30.8, 28.2, 25.4, 24.6, 23.6.
HRMS calcd for C18H29N2O3S (MH+): 353.1894. Found: 353.1905.
6a: 1H NMR (300 MHz, CDCl3) δ 7.69 (d, J ) 3.0 Hz, 1H), 7.25
(d, J ) 3.0 Hz, 1H), 4.57 (m, 1H), 3.10 (m, 1H), 2.60 (m, 1H),
2.50 (m, 1H), 2.23 (s, 3H), 1.98 (m, 1H), 1.79 (m, 2H), 1.71 (m,
1H), 1.39 (s, 9H), 0.98 (d, J ) 6.0 Hz, 3H), 0.91 (d, J ) 6.0 Hz,
with 10 mL of heptane, and dried to give 18.9 g (30%) of the second
1
crop of 3: [R]25 -62.2 (c 2.21, CHCl3); H NMR (300 MHz,
D
CDCl3) δ 8.42 (s, 1H), 7.90 (d, J ) 3.0 Hz, 1H), 7.41 (d, J ) 3.0
Hz, 1H), 4.03 (t, J ) 6.0 Hz, 1H), 1.80 (m, 2H), 1.43 (s, 9H), 0.86
(d, J ) 6.0 Hz, 6H); 13C NMR (75 MHz, CDCl3) δ 170.9, 166.9,
156.7, 144.3, 122.2, 81.7, 71.5, 42.0, 28.2, 24.8, 23.3, 21.7. Anal.
Calcd for C14H22N2O2S: C, 59.54; H, 7.85; N, 9.92, S, 11.35.
Found: C 59.56; H, 7.88; N, 9.91, S, 11.36.
3100 J. Org. Chem., Vol. 73, No. 8, 2008

N-Acylpyrrolidine for Inhibition of HCV Polymerase
3H); 13C NMR (75 MHz, CDCl3) δ 207.5, 173.3, 172.8, 142.6,
119.1, 82.6, 72.8, 61.9, 58.9, 47.6, 37.5, 31.5, 28.0, 25.8, 25.1, 24.0.
HRMS calcd for C18H29N2O3S (MH+): 353.1894. Found: 353.1893.
Salt of (2S,4S,5R)-2-(2-Methylpropyl)-5-(1,3-thiazol-2-yl)-2,4-
pyrrolidinedicarboxylate and (R)-1,1′-Binapthyl-2,2′-dihydrogen
Phosphate (7): To a 300 gallon reactor were successively added
45.0 kg (201 mol) of t-butyl L-leucine hydrochloride, 450 L of
toluene, and 22.8 kg (201 mol) of 1,3-thiazole-2-carbaldehyde at
ambient temperature. The white slurry was stirred for 2 min under
nitrogen; then, 20.7 kg (205 mol) of Et3N was added. The mixture
was heated at 50 °C for 2 h. The reaction was cooled to 20 °C and
quenched with 180 L of water. Layers were separated, and the
toluene layer was washed with 130 L of water. The toluene layer
was concentrated in vacuo at 35-55 °C to about 180 L. After being
diluted with 270 L of toluene, the solution was further distilled to
about 270 L. The toluene solution of the crude 1,1-dimethylethyl
N-(1,3-thiazol-2-ylmethylidene)-L-leucinate (3) was used for the
cycloaddition without further purification.
The toluene solution of 3 as prepared above was diluted with
540 L of toluene, followed by sequential addition of 3.9 kg (11.9
mmol) of hydroquinine and 22.5 kg of 4 Å powdered molecular
sieves at ambient temperature. The thin slurry was stirred for 5
min followed by addition of 1.02 kg (6.11 mol) of AgOAc. The
mixture was cooled to -10 °C, and then, 19.1 kg (222 mol) of
methyl acrylate was added. The reaction was stirred for 6 h. The
reaction was warmed to ambient temperature and filtered through
an in-line filter. The filtrate was successively washed with 2 ×
180 L of water and 180 L of 15% brine. The organic layer was
concentrated in vacuo at 35-65 °C to about 360 L. The mixture
was diluted with 360 L of i-PrOH and further concentrated to 360
L. This dilution-concentration protocol was repeated once. The
crude cycloadduct 2-(1,1-dimethylethyl) 4-methyl (2S,4S,5R)-2-(2-
methylpropyl)-5-(1,3-thiazol-2-yl)-2,4-pyrrolidinedicarboxylate (4),
prepared as such, was used for the next step without further
purification.
and 9.80 L (130 mol) of SOCl2. After being stirred at 20 °C for 2
h, the resultant acid chloride 8 was cooled to 0 °C. The previous
filtrate in the other reactor containing 4 was slowly added to the
cold acid chloride (8) solution, the temperature was raised to 35
°C, and it was stirred overnight.
The reaction mixture was concentrated under vacuum, with mild
heating, to approximately 370 L and cooled to 20 °C. To the mixture
was added 148 L of 1 N HCl, followed by 740 L of heptane, and
the slurry was cooled to 0 °C. After 2 h at 0 °C, the solids were
filtered off, washed with 370 L of water, followed by 370 L of
heptane, and dried to a constant weight to provide 50.8 kg (88%)
of 9 as a white crystalline solid: [R]25D +73.0 (c 1.20, MeOH); 1H
NMR (300 MHz, DMSO-d6) δ 7.53 (d, J ) 3.0 Hz, 1H), 7.49 (d,
J ) 3.0 Hz, 1H), 7.19 (d, J ) 6.0 Hz, 1H), 6.85 (d, J ) 6.0 Hz,
1H), 6.62 (s, 1H), 5.82 (d, J ) 9.0 Hz, 1H), 3.90 (m, 1H), 3.67 (s,
3H), 3.27 (s, 3H), 2.83 (t, J ) 12 Hz, 1H), 2.50 (s, 1H), 2.36 (m,
1H), 2.20 (m, 2H), 1.93 (m, 2H), 1.36 (s, 9H), 1.31 (s, 9H), 1.01
(d, J ) 6.0 Hz, 3H), 0.99 (d, J ) 6.0 Hz, 3H); 13C NMR (75 MHz,
MeOH-d4) δ 170.5, 169.5, 168.1, 156.9, 139.6, 137.8, 134.0, 124.9,
118.9, 115.4, 107.5, 80.4, 69.7, 62.4, 59.5, 52.6, 33.0, 27.3, 25.9,
23.6, 23.2, 22.5. Anal. Calcd for C30H42N2O6S: C, 64.49; H, 7.58;
N, 5.01, S, 5.74. Found: C 64.48; H, 7.54; N, 5.06, S, 5.69.
1,1-Dimethylethyl (4S,5R)-1-{[4-(1,1-Dimethylethyl)-3-(me-
thyloxy)phenyl]carbonyl}-4-(hydroxymethyl)-2-(2-methylpropyl)-
5-(1,3-thiazol-2-yl)-l-prolinate (10): To a 50 L reactor were added
6.48 kg (11.6 mol) of ester 9, 880 g (23.2 mol) of NaBH4, 120 g
(0.58 mol) of NaB(OAc)3H, and 16.2 L of THF at ambient
temperature under nitrogen. The mixture was cooled to -10 °C.
To the white slurry was added 1.87 L (46.4 mol) of methanol over
about 1 h. The reaction was gradually warmed to 25 °C over 1.5 h
and stirred for 5 h. The reaction was cooled to 5 °C and treated
with 740 mL (23.2 mol) of MeOH over 30 min and 1.95 kg of
concentrated HCl (19.8 mol) over 30 min. The mixture was diluted
with 19.4 L of water and 6.5 L of MTBE. About 115 g more of
concentrated HCl was added to adjust the pH of the aqueous layer
to about 3.0 to ensure full hydrolysis of any residual reducing
reagent. Layers were separated, and the aqueous layer was extracted
with 19.5 L of MTBE. The combined organic layers were washed
with 13 L of 15% brine and concentrated in vacuo to about 15 L.
To the solution was added 19.5 L of MeCN, and the mixture was
further concentrated to about 16 L, at which point the crystallization
started. After being cooled to -5 °C and stirred for 2h, the white
mixture was filtered, washed with 2.5 L of MTBE, and dried at 55
°C in vacuo to give 5.49 kg (89%) of alcohol 10 as a white
crystalline solid.
The filtrate from the above isolation was combined with filtrates
from two other reactions involving 3.1 and 6.2 kg of ester 9. The
combined filtrates were concentrated to 13 L, diluted with 10 L of
MeCN, and further concentrated in vacuo to 7.5 L. After being
cooled at -8 °C for 2 h, the mixture was filtered, washed with 0.8
L of MTBE, and dried to give 1.57 kg (8%) of a second crop of
product. The overall yield including the second crop would be 97%
of 10: [R]25D +81.6 (c 1.10, MeOH); 1H NMR (300 MHz, MeOH-
d4) δ 7.36 (s, 2H), 7.05 (d, J ) 9.0 Hz, 1H), 6.62 (d, J ) 3.0 Hz,
1H), 6.34 (s, 1H), 5.46 (d, J ) 9.0 Hz, 1H), 3.63 (m, 1H), 3.52 (s,
3H), 3.27 (m, 1H), 3.09 (m, 1H), 2.74 (m, 1H), 2.27 (m, 1H), 2.18
(m, 2H), 1.93 (m, 2H), 1.50 (s, 9H), 1.20 (s, 9H), 1.01 (d, J ) 6.0
Hz, 6H); 13C NMR (75 MHz, MeOH-d4) δ 170.5, 169.5, 168.1,
156.9, 139.6, 137.8, 134.0, 124.9, 118.9, 115.4, 107.5, 80.4, 69.7,
62.4, 59.5, 52.6, 33.0, 27.3, 25.9, 23.6, 23.2, 22.5. Anal. Calcd for
C29H42N2O5S·0.5H2O: C, 64.53; H, 8.02; N, 5.18; S, 5.94. Found:
C 64.84; H, 8.00; N, 5.23; S, 5.91. HRMS m/z calcd for
C29H42N2O5SNa (M+Na+): 553.2707. Found: 553.2688.
A 500 gallon reactor was charged with 59.4 kg (171 mol) of
(R)-1,1′-binapthyl-2,2′-dihydrogen phosphate and 765 L of i-PrOH,
and the slurry was heated to 80 °C for 30 min. The solution of
cycloadduct 4 in i-PrOH prepared as above in the 300 gallon reactor
was then added to the 500 gallon reactor, maintaining a temperature
of 70-80 °C during the addition. After being stirred for 1 h, the
reaction mixture was cooled to 20 °C at a rate of 1 °C/min. The
solids were filtered, and the cake was washed with 225 L of i-PrOH.
The solids were dried at 50-60 °C under house vacuum to afford
81.6 kg (overall yield of 57% for three steps from t-butyl L-leucine
hydrochloride) of the salt of (2S,4S,5R)-2-(2-methylpropyl)-5-(1,3-
thiazol-2-yl)-2,4-pyrrolidinedicarboxylate and (R)-1,1′-binapthyl-
2,2′-dihydrogen phosphate (7); enantiomeric ratio by chiral HPLC
analysis of the free base 412 > 99.9:0.1 favoring target enantiomer
1
4. Analytical data of 7: [R]25 -36.9 (c 1.33, MeOH); H NMR
D
(300 MHz, DMSO-d6) δ 8.10 (d, J ) 6.0 Hz, 2H), 8.06 (d, J ) 6.0
Hz, 2H), 7.76 (m, 2H), 7.73 (m, 2H), 7.48 (m, 4H), 7.33 (m, 2H),
7.23 (m, 2H), 5.22 (br s, 1H), 3.75 (m, 1H), 3.38 (s, 3H), 2.90 (m,
1H), 2.23 (m, 1H), 1.80 (m, 2H), 1.61 (m, 1H), 1.41 (s, 9H), 0.90
(d, J ) 6.0 Hz, 3H), 0.83 (d, J ) 6.0 Hz, 3H). Anal. Calcd for
C38H41N2O8SP: C, 63.68; H, 5.77; N, 3.91, S, 4.47. Found: C
63.45; H, 5.74; N, 3.76, S, 4.39.
2-(1,1-Dimethylethyl) 4-Methyl (2S,4S,5R)-1-{[4-(1,1-Dimeth-
ylethyl)-3-(methyloxy)phenyl]carbonyl}-2-(2-methylpropyl)-5-(1,3-
thiazol-2-yl)-2,4-pyrrolidinedicarboxylate (9): A 300 gallon reac-
tor was charged with 74 kg (103 mol) of 7, followed by 370 L of
MTBE and 15.8 L (11.0 mol) of triethylamine. The slurry was
stirred at room temperature for 1 h, and the solids were filtered off
and washed with 222 L of MTBE. The reactor was rinsed with
MTBE, the filtrate containing free base 4 was recharged to the
reactor via an in-line filter (0.4-1 micron) to remove residual solids,
and 20.9 L (260 mol) of pyridine was added. Meanwhile, 25.8 kg
(120 mol) of carboxylic acid 2 was charged to a 500 gallon reactor,
followed by 222 L of dichloromethane, 2.50 L (30 mol) of pyridine,
(4S,5R)-1-{[4-(1,1-Dimethylethyl)-3-(methyloxy)phenyl]car-
bonyl}-4-[(methyloxy)methyl]-2-(2-methylpropyl)-5-(1,3-thiazol-
2-yl)-l-proline (1): A 300 gallon reactor was charged with 139 L
of 1,2-dimethoxyethane and 34.7 kg (65.4 mol) of 10. The mixture
was cooled to -30 °C. A solution of 12.6 kg (131 mol) of sodium
t-butoxide in 139 L of 1,2-dimethoxyethane was charged to the
J. Org. Chem, Vol. 73, No. 8, 2008 3101

Agbodjan et al.
cold reaction mixture over 30 min and rinsed with 29 L of 1,2-
dimethoxyethane. While maintaining the reaction temperature below
-25 °C, 18.5 kg (131 mol) of methyl iodide was added. The mixture
was held at -30 °C for 30 min and then warmed to 25 °C over 1
h. The alkylation at this point was >99% complete. The following
recycling process was used to convert all of the starting material
10. The reaction was quenched with 139 L of 0.1 N HCl and cooled
to 0 °C. The intermediate methylated t-butyl ester 11 was collected
by filtration and washed with 52 L of water. The material was
charged to a clean 300 gallon reactor along with 347 L of 1,2-
dimethoxyethane. The solution was concentrated in vacuo to 220
L to azeotropically dry the solution. To the mixture was added 90
L of 1,2-dimethoxyethane, and the solution was cooled to -30 °C.
A solution of 1.26 kg (13.1 mol) of sodium t-butoxide in 14 L of
1,2-dimethoxyethane was charged to the cold reaction mixture and
rinsed with 3.5 L of 1,2-dimethoxyethane. While maintaining the
reaction temperature below -25 °C, 1.90 kg (13.1 mol) of methyl
iodide was added. The mixture was warmed to 25 °C; then, 69 L
of MTBE and 69 L of 0.1 N HCl were added. The aqueous phase
was removed, and the organic solution was concentrated in vacuo
to 170 L. The reactor was charged with 3.30 kg of concentrated
HCl. The mixture was heated to 60 °C and stirred for 7 h. After
being cooled to 25 °C, the mixture was diluted with 139 L of MTBE
and 104 L of water. The aqueous layer was removed, and the
organic solution was concentrated in vacuo to 140 L. After being
treated with 34 L of MTBE and 35 L of 2,2,4-trimethylpentane,
the mixture was aged for 30 min at 25 °C. An additional 35 L of
2,2,4-trimethylpentane was added, and the mixture was stirred for
30 min. A final 35 L of 2,2,4-trimethylpentane was charged, and
the mixture was stirred for an additional 30 min. The suspension
was cooled to 0 °C and held for 2 h. The product was collected by
filtration and washed with 70 L of a 1:1 mixture of MTBE and
2,2,4-trimethylpentane. The product was dried in vacuo to give a
66% yield of 1: [R]25D +50.0 (c 1.43, MeOH); 1H NMR (300 MHz,
MeOH-d4) δ 7.87 (d, J ) 6.0 Hz, 1H), 7.61 (d, J ) 3.0 Hz, 1H),
7.22 (d, J ) 9.0 Hz, 1H), 6.71 (d, J ) 6.0 Hz, 1H), 6.36 (d, J )
3 Hz, 1H), 5.67 (d, J ) 9.0 Hz, 1H), 3.64 (s, 3H), 3.19 (s, 1H),
3.11 (m, 1H), 3.10 (s, 3H), 2.68 (t, J ) 9 Hz, 1H), 1.95-2.35 (m,
5H), 1.32 (s, 9H), 1.14 (d, J ) 6.0 Hz, 3H), 1.12 (d, J ) 6.0 Hz,
3H); 13C NMR (75 MHz, MeOH-d4) δ 171.2, 158.6, 140.2, 135.0,
126.5, 121.8, 116.9, 108.7, 100.2, 70.8, 63.9, 57.9, 54.3, 47.0, 43.4,
42.7, 34.6, 28.8, 24.9, 24.6, 23.7. Anal. Calcd for C26H36N2O5S:
C, 63.91; H, 7.43; N, 5.73, S, 6.56. Found: C 63.82; H, 7.37; N,
5.72, S, 6.58.
Acknowledgment. We thank Dr. Thomas O’Connell (Uni-
versity of North Carolina at Chapel Hill) for the NMR studies,
Dr. David Black (GlaxoSmithKline) for the HRMS analysis,
and Professor William Clegg and Dr. Neil Brooks (University
of Newcastle) for the GSK-funded X-ray crystallographic
analysis.
Supporting Information Available: NMR spectra for com-
pounds 1, 3, 4, 5, 6a, 7, 9, and 10, gDQCOSY and gHSQC analysis
on compounds 5 and 6b, and X-ray crystallographic data for
compound 10. This material is available free of charge via the
Internet at http://pubs.acs.org.
JO800062C
3102 J. Org. Chem., Vol. 73, No. 8, 2008