Syntheses and NMR characterizations of epimeric 4-deoxy-4-fluoro carbohydrates

Article abstract of DOI:10.1016/j.carres.2011.07.019

The synthesis and NMR characterizations of 1,2,3,6-tetra-O-benzoyl-4-deoxy- 4-fluoro-β-D-galactopyranoside, the 4-deoxy-4-fluoro epimer of an intermediate in the synthesis of a drug substance, needed for use as a potential impurity standard and to confirm the stereoselectivity of a key fluorination step, are described.

Full text of DOI:10.1016/j.carres.2011.07.019

Carbohydrate Research 346 (2011) 2323–2326
Contents lists available at SciVerse ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Syntheses and NMR characterizations of epimeric 4-deoxy-4-fluoro
carbohydrates
⇑
Witold Subotkowski , Dirk Friedrich , Franz J. Weiberth
Chemical Development, sanofi-aventis US, 1041 Route 202-206, Bridgewater, NJ 08807, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 27 April 2011
Received in revised form 29 June 2011
Accepted 19 July 2011
Available online 26 July 2011
The synthesis and NMR characterizations of 1,2,3,6-tetra-O-benzoyl-4-deoxy-4-fluoro-b-D-galacto-
pyranoside, the 4-deoxy-4-fluoro epimer of an intermediate in the synthesis of a drug substance, needed
for use as a potential impurity standard and to confirm the stereoselectivity of a key fluorination step, are
described.
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Stereoselectivity
Fluorination
Carbohydrate
Epimer
1. Introduction
sulfur trifluoride (DAST), Bu4NF, Et3NꢀHF, and SO2F2. The formation
of the corresponding olefin 6 was a competing side reaction, and in
The strategic incorporation of fluorine atoms as a means to opti-
mize pharmacology, safety, and metabolic stability of small mole-
cules is a well established technique in drug design. While
replacing hydrogen with fluorine during analoging is a common
practice, replacing hydroxy with fluorine is also effective because
of similarities in van der Waals radii and electronegativity of fluo-
rine and oxygen.1 In this regard, 4-deoxy-4-fluoro carbohydrates
were of immediate interest. For example, SAR7226 (1), a molecule
containing a 4-fluoroglucopyranose glycon, progressed in our com-
pany as a promising SGLT1/SGLT2 inhibitor, potentially useful for
the treatment of diabetes.2 Scheme 1 illustrates a retrosynthetic,
convergent approach to 1 that features the coupling of two ad-
vanced intermediates, 4-fluoroglycosyl donor 5 and hydroxypyraz-
ole 2, to afford the SAR7226 backbone structure.3 Glycosyl donor 5
some cases the major pathway (Scheme 2). Of these reagent sys-
tems, BAST was the most effective reagent, and provided an
85:15 mixture of 4:6 (1.3 equiv BAST, 20 °C ? 55 °C, 2-meth-
yltetrahydrofuran as a solvent). The identification of 1-butanol as
an ideal antisolvent was key in developing an efficient and simple
isolation of 4. Thus, after quenching with water and extractive
workup in toluene, the final organic phase was treated with 1-
butanol to crystallize 4. After filtration, 4 was isolated as a white
solid in 74–78% isolated yield and with purity P98.5 A% (area %
UV absorption by HPLC assay). The olefin and other impurities
were predominately retained in the filtrate. This simple isolation
of 4 was an important aspect of an efficient overall synthesis of 5
that was demonstrated on a multi-kilogram pilot-plant scale.
It appeared that the BAST fluorination occurred with inversion
of configuration as was widely reported for DAST.4 BAST was a
preferred replacement for DAST for the fluorination because of its
process safety advantages.5 However, to more firmly establish
the stereoselectivity of the BAST fluorination step, to confirm
whether any of the C-4 epimer formed even in low levels, and to
help establish the impurity profiles of the SAR7226 synthetic route,
the corresponding 4-F epimer, 1,2,3,6-tetra-O-benzoyl-4-deoxy-4-
was derived from D-galactose in three chemical transformations;
namely, selective benzoylation to give the tetrabenzoate 3, fol-
lowed by fluorination with inversion at the 4-position, and finally
manipulation at the anomeric position to give the desired glycosyl
bromide 5. Hydroxypyrazole 2 was derived from mono-alkylation
of ethyl 4,4,4-trifluoroacetoacetate with 4-methoxybenzyl chloride
followed by treatment with benzylhydrazine.
During the development of the fluorination step via either di-
rect fluorination of 3 or by way of the corresponding triflates or
nonaflates, several reagent systems were evaluated, including
bis(2-methoxyethyl)aminosulfur trifluoride (BAST), diethylamino-
fluoro-b-D-galactopyranoside (9), was independently synthesized
and characterized as described below.
2. Results and discussion
⇑
Corresponding author. Tel.: +1 908 534 7811.
E-mail address: w.subotkowski@gmail.com (W. Subotkowski).
Member of Analytical Sciences Department at the time this work was conducted.
4-Deoxy-4-fluoro epimer 9 was prepared in three steps starting
with selective tetrabenzoylation of D-glucose to tetrabenzoyl-b-D-
0008-6215/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2011.07.019

2324
W. Subotkowski et al. / Carbohydrate Research 346 (2011) 2323–2326
N
Bn
CF3
N
H
N
CF3
HO
N
OMe
OH
F
O
O
2
OMe
OH .H2O
OBz
OBz
F
OBz
OH
OH
O
OBz
OBz
O
OBz
OBz
O
OH
OH
O
Br
OBz
1
SAR7226
HO
HO
F
OBz
OBz
OH
OBz
5
4
3
D-Galactose
Scheme 1. Retrosynthesis of SAR7226.
quenching with water, the organic layer was partially concen-
trated, and 9 was isolated in 91% yield (99.5 A% purity) after crys-
tallization upon addition of 1-butanol.
OBz
OBz
F
OBz
O
OBz
OBz
O
OBz
OBz
O
OBz
OBz
BAST
+
As expected, 8 produced less olefin 6 (2.5 A%) during fluorina-
tion than the corresponding BAST-activated galacto-intermediate
due to less favorable syn- versus anti-elimination.8 Similarly, fluo-
rination of the analogous triflate 10 derived from tetrabenzoylga-
lactose 3 with tetrabutylammonium fluoride gave olefin 6 as the
major product (94.7 A%) due to more favorable elimination with
anti-H compared to syn-H in 88 (Scheme 4). Olefin 6 was isolated
and found to be consistent (1H NMR) with material that was syn-
thesized independently.9
NMR spectra are consistent with the proposed structures and
relative stereochemistries (Table 1). For example, large (8–10 Hz)
vicinal H–H couplings throughout the ring are consistent with
chair-like conformation and all-equatorial substituents in 4. In
contrast, the small (62.5 Hz) vicinal couplings of H-4 in epimer 9
are consistent with equatorial H-4 and axial fluorine. Similarly,
the larger vicinal H-F couplings in 9 (JHF = 27 and 26 Hz) versus
those in 4 (JHF = 14 and 2.5 Hz) are also consistent with axial fluo-
rine in 9 and equatorial fluorine in 4. Finally, differences in the 1H
and 19F chemical shifts of 4 and 9 are consistent with assignments.
For example, the downfield shift of H-2 in epimer 9 can be ratio-
nalized by its spatial proximity to fluorine as a result of their
1,3-diaxial relationship; and the H-4 and fluorine chemical shifts
follow a typical pattern of being shifted downfield in an equatorial
position versus a corresponding axial position (H-4eq in 9 is shifted
downfield relative to H-4ax in 4, and Feq in 4 is shifted downfield
relative to Fax in 9).
HO
OBz
OBz
OBz
3
4
6
Scheme 2. Fluorination of 3.3
glucopyranose 7 followed by formation of triflate (8) and fluorine
substitution with tetrabutylammonium fluoride (Scheme 3).6 For
the first step, tetrabenzoylation methodology used previously for
galactose3 was adapted. However, because the equatorial 4-OH in
D
-glucopyranose is more reactive than the axial 4-OH in D-galacto-
pyranose, it was anticipated that the desired selectivity of 1,2,3,6-
tetrabenzoylation would be more difficult to achieve.7 Indeed,
benzoylation of glucose led to a complex mixture of products con-
taining only about 30 A% of 7 (HPLC) together with anomeric mix-
tures of positional isomers and over- and under-benzoylated
species. However, because the intent of the exercise was to isolate
an analytical reference standard of epimer 9, not to develop a scal-
able process that was required for 4, this poor conversion was sat-
isfactory for current purposes. Purification of 7 (100 A% HPLC) was
achieved by column chromatography. The transformation to the
triflate 8 proceeded in high yield (99.7%). The last step, fluoride dis-
placement of triflate using Bu4NF, proceeded at room tempera-
ture.6 After diluting the reaction mixture with 2-MeTHF and
OH
OBz
OBz
OBz
F
BzCl
Bz2O
O
OH
OH
O
OBz
OBz
O
OBz
OBz
O
OBz
OBz
Tf2O
Bu4NF
HO
HO
TfO
OH
OBz
OBz
OBz
9
7
D-Glucose
8
Scheme 3. Synthesis of 4-deoxy-4-fluoro epimer 9.
OBz
OBz
OBz
F
OBz
O
OBz
OBz
O
OBz
OBz
O
OBz
OBz
O
OBz
OBz
Bu4NF
Tf2O
+
HO
TfO
OBz
OBz
OBz
OBz
3
10
4
6
minor 4.1 %
major 94.7 %
Scheme 4. Fluorination by displacement of triflate 10.

W. Subotkowski et al. / Carbohydrate Research 346 (2011) 2323–2326
2325
Table 1
1H NMR assignments for 7, 4, and 9
OBz
OBz
OBz
F
O
OBz
OBz
O
OBz
OBz
O
OBz
OBz
HO
F
OBz
OBz
OBz
7
4
9
1H NMR (300 MHz)
H-1
H-2
H-3
H-4
H-5
H-6,60
6.20 d, J = 8.0
6.24 d, J = 8.3
5.78 dd, J = 9.5, 8.3
6.25 d, J = 8.3
6.11 dd, J = 10.3, 8.3
5.59 ddd, J = 10.3, 2.5; JHF = 27
5.24 br dd, J = 2.5; JHF = 50
4.39 br ddd, J = 6.5, 6.5; JHF = 26
4.65 AB-m, 2H
5.75 dd, J = 9.5, 8.0
5.64 dd, J = 9.5, 8.5
3.99 ddd, J = 10.0, 8.5, 4.0
4.05 ddd, J = 10.0, 3.5, 2.0
4.89 dd, J = 12.5, 3.5
4.63 dd, J = 12.5, 2.0
7.2–7.6 m, 12H
5.99 ddd, J = 9.5, 9.0; JHF = 14
4.94 ddd, J = 9.5, 9.0; JHF = 50
4.30 dddd, J = 9.5, 4.5, 2.5; JHF = 2.5
4.75 ddd, J = 12.5, 2.5; JHF = 1.5
4.64 ddd, J = 12.5, 4.5; JHF = 1.0
7.3–7.6 m, 12H
Ar-H
7.2–7.6 m, 12H
7.8–8.1 m, 8H
7.8–8.2 m, 8H
7.9–8.1 m, 8H
OH
3.60 d, J = 4.0
13C NMR (75 MHz)
167.5, 167.2, 165.5,
164.9, 134.0, 133.7,
130.4, 130.2, 130.0,
129.6, 129.1, 129.0,
128.8, 128.6, 93.0,
76.2, 75.7, 70.7,
166.3, 165.7, 165.4,
164.8, 134.2, 133.8,
133.6, 130.4, 130.1,
129.7, 129.1, 128.8,
128.7, 128.6, 92.8,
88.4, 85.9, 72.9,
166.3, 166.0, 165.4,
164.9, 134.1, 134.0,
133.7, 133.6, 130.5,
130.2, 130.1, 130.0
129.6, 129.1, 128.9,
93.0, 87.6, 85.9,
69.3, 63.2
70.6, 62.5
72.4, 68.5, 61.9, 53.7
19F NMR (282 MHz)
ꢂ199.0 br dd, J = 50, 14
-216.3 ddd, J = 50, 27, 26
In conclusion, 4-deoxy-4-fluoro epimer 9 was synthesized and
its stereochemistry was confirmed by NMR analyses. With pure 9
in hand as an analytical standard sample and HPLC marker, it
was then confirmed that the BAST fluorination of 3 proceeded with
inversion of configuration at C-4 to give 4 highly selectively. Anal-
yses of isolated product 4 and its concentrated filtrates did not
indicate the presence of 9 (HPLC and 19F NMR), thus helping estab-
lish the purity profile of SAR7226 and its intermediates.
The solution was cooled to 20 °C, then was diluted with toluene
(200 mL) and water (250 mL). The biphasic solution was stirred
for 1 h at 20 °C, and then the phases were separated. The organic
phase was washed with water (1 ꢁ 160 mL), aq 6% NaHCO3
(3 ꢁ 150 mL) and water (2 ꢁ 150 mL). The organic layer was con-
centrated (50 torr/45 °C), redissolved at 50 °C in toluene (100 mL)
and crystallized by the addition of n-heptane (100 mL). The solid
(62.6 g, ca. 33 A% product by HPLC) was collected by filtration, then
was redissolved in CH2Cl2 (50 mL) and chromatographed on 210 g
of silica gel (60–230 mesh) using CH2Cl2 as an eluent. Fractions
containing product were combined and concentrated to give
8.95 g of 7 as a white solid (91.8 A% product by HPLC). The solid
was redissolved in CH2Cl2 (50 mL), heated to gentle reflux, and
n-heptane (80 mL) was slowly added. The product crystallized
after ca. 5 min at 40 °C. The suspension was allowed to cool to
20 °C and then was filtered yielding 7.57 g (12.7 mmol) (100 A%
by HPLC, 10% yield from glucose) of 7 as a white solid. The struc-
tural assignment of 7, as the b-anomer, is fully supported by
NMR data provided in Table 1. The synthesis of 7 has previously
been reported.11
3. Experimental
3.1. General
HPLC analyses were performed on Agilent 1100 HPLC using Zor-
bax Eclipse XDB-C8, 5
l
m, 4.6 ꢁ 150 mm column. Mobile phase A:
H2O (0.1% TFA), B: acetonitrile (0.1% TFA). System: 0–1 min 40/60
A/B, then gradient to 20/80 for 16 min, flow rate 1 mL/min, UV
detection at 220 nm.
NMR studies were performed on a Varian 300 spectrometer in
CDCl3. Chemical shifts are given in ppm relative internal TMS (1H
and 13C) and external CFCl3
reported in Hz.
(
19F). Coupling constants (J) are
HPLC retention time: 7 (9.93 min) (purity 100% at 220 nm).
1H NMR (CDCl3), 13C NMR: see Table 1.
MS (ES+): 619.18 (M+Na)+.
The preparations of compounds 3, 4, and 5 have been previously
described.3
Anal. Calcd for C34H28FO10: C, 68.45; H, 4.73. Found: C, 68.36; H,
4.52.
3.1.1. 1,2,3,6-Tetra-O-benzoyl-b-D-glucopyranose (7)
A three-necked, 500-mL reaction flask equipped with a mechan-
ical stirrer, thermocouple, addition funnel, condenser, and nitrogen
3.1.2. 1,2,3,6-Tetra-O-benzoyl-4-(1,1,1-
trifluoromethanesulfonyl)-b-D-glucopyranose (8)
A 250-mL flask was charged at 20 °C with 7 (6.7 g, 10.4 mmol,
1.0 equiv), CH2Cl2 (65 mL) and pyridine (2.5 mL, 131.2 mmol,
3.0 equiv). The solution was stirred and cooled in an ice bath,
and triflic anhydride (2.5 mL, 14.6 mmol, 1.4 equiv) was added
via syringe while keeping the temperature below 5 °C. The bath
was removed and the reaction was allowed to warm to 20 °C. After
30 min, the reaction was judged to be complete (by HPLC). The sol-
vent was evaporated to give 8 (99.7% purity by HPLC at 220 nm) as
a white solid which was used directly in the next step.
inlet was charged with
D-glucose (22.9 g, 127 mmol, 1.0 equiv),
N-methylpyrrolidone (NMP) (200 mL), and pyridine (56.4 mL,
697 mmol, 5.5 equiv). The suspension was cooled to ca. 5 °C and
benzoyl chloride (44.9 mL, 387 mmol, 3.0 equiv) was slowly added
over 25 min keeping the temperature below 20 °C (all solids dis-
solved by the end of addition). Cooling and stirring were continued
(solids started to precipitate at ca. 13 °C). Benzoic anhydride10
(43.0 g, 190 mmol, 1.5 equiv) was added and the reaction mixture
was heated to 50 °C (clear solution at 45 °C). After heating at 50 °C
for 18 h, the reaction mixture contained about 30 A% (HPLC) of 7.
HPLC retention times: 7 (9.93 min), 8 (16.34 min).

2326
W. Subotkowski et al. / Carbohydrate Research 346 (2011) 2323–2326
3.1.3. 1,2,3,6-Tetra-O-benzoyl-4-deoxy-4-fluoro-b-D-
Acknowledgment
galactopyranose (9)
A 250-mL flask was charged with crude 8 (7.6 g, 10.4 mmol,
1.0 equiv, based on quantitative trifylation) and 2-meth-
yltetrahydrofuran (35 mL). The solution was cooled in an ice bath
and tetrabutylammonium fluoride (1.0 M solution in THF, 90 mL,
120 mmol, 8.6 equiv) was added over 5 min while keeping the
temperature below 12 °C. The cooling bath was removed and the
reaction mixture was allowed to warm to room temperature. After
1 h, 37 A% (HPLC) starting material remained. Additional tetrabu-
tylammonium fluoride solution (30 mL, 2.9 equiv) was added. After
1 h (<0.3 A% 8, 92.7 A% 9), the reaction mixture was diluted with
2-methyltetrahydrofuran (150 mL) and quenched with water
(60 mL). The organic layer was washed with water (1 ꢁ 60 mL),
6% aq NaHCO3 (2 ꢁ 60 mL) and water (2 ꢁ 60 mL). The organic
layer was partially concentrated (50 Torr/24 °C) to a volume of
ca. 70 mL and the product was crystallized by addition of 1-butanol
(100 mL). The suspension was partially concentrated to ca. half of
the volume (50 Torr/50 °C), then was cooled in an ice bath and fil-
tered. The solid was washed with 1-butanol (50 mL) and dried
(50 Torr/42 °C/N2 bleed) to yield 5.7 g of 9 (91% yield, 99.5 A% pur-
ity) as a white solid.
We are grateful to colleagues in the Analytical Department in
sanofi-aventis, US, for their support.
References
1. Hagmann, W. K. J. Med. Chem. 2008, 51, 4359–4369.
2. (a) Frick, W.; Glombik, H.; Kramer, W.; Heuer, H.; Brummerhop, H.;
Plettenburg, O. US Patent 2005/0014704A1, 2005 (WO2004052902); Chem.
Abstr. 2005, 141, 38810.; (b) Frick, W.; Glombik, H.; Kramer, W.; Heuer, H.;
Brummerhop, H.; Plettenburg, O. US Patent 2004/0259819A1, 2004
(WO2004052903); Chem. Abstr. 2005, 141, 38811.
3. Weiberth, F. J.; Gill, H. S.; Jiang, Y.; Lee, G. E.; Lienard, P.; Pemberton, C.; Powers,
M. R.; Subotkowski, W.; Tomasik, W.; Vanasse, B. J.; Yu, Y. Org. Process Res. Dev.
2010, 14, 623–631.
4. Hudlicky, M. Org. React. 1988, 35, 513–637.
5. Lal, G. S.; Pez, G. P.; Pesaresi, R. J.; Prozonic, F. M.; Cheng, H. J. Org. Chem. 1999,
64, 7048–7054.
6. For similar examples of fluoride displacement of triflate using
tetrabutylammoniun fluoride, see: (a) Marson, C. M.; Melling, R. C. Chem.
Commun. 1998, 1223–1224; (b) Szarek, W. A.; Hay, G. W.; Doboszewski, B. J.
Chem. Soc., Chem. Commun. 1985, 663–664.
7. (a) Haines, A. H. In Advances in Carbohydrate Chemistry and Biochemistry;
Tipson, R. S., Horton, D., Eds.; Academic Press: New York, 1976; Vol. 33, pp 11–
109; (b) Williams, J. M.; Richardson, A. C. Tetrahedron 1967, 23, 1369–1378.
8. Examples: (a) Nemr, A. E.; Tsuchiya, T. Carbohydr. Res. 1977, 301, 77–87; (b)
Nemr, A. E.; Tsuchiya, T. Tetrahedron Lett. 1995, 36, 7665–7668.
9. Blattner, R.; Ferrier, R. J.; Tyler, P. C. J. Chem. Soc., Perkin Trans. 1 1980, 1535–1539.
10. The use of benzoyl chloride and benzoic anhydride in tandem was adapted
from Ref. 3, where improved selectivity during benzoylation was observed.
11. (a) Brigl, G.; Grüner, H. Chem. Ber. 1932, 65, 1428–1434; (b) Lundt, P.; Pedersen,
C. Acta Chem. Scand. 1971, 25, 2320–2326.
HPLC retention times: triflate 8: 16.34 min, product 9:
14.17 min, olefin 6: 14.59 min.
1H NMR, 13C NMR, 19F NMR (CDCl3): see Table 1.
MS (ES+): 621.17 (M+Na)+.
Anal. Calcd for C34H27FO9: C, 68.22; H, 4.55; F, 3.17. Found: C,
67.98; H, 4.28; F, 3.17.