The synthesis of a series of deoxygenated 2,3-difluoro-N-acteylneuraminic acid derivatives as potential sialidase inhibitors

Article abstract of DOI:10.1016/j.carres.2013.03.026

Here we describe the successful syntheses of a series of 4-, 7-, 8- and 9-deoxygenated 2,3-difluoro-N-acetylneuraminic acid derivatives as potential mechanism-based inhibitors of sialidases. The syntheses commenced utilising an enzyme-catalysed aldolase reaction between N-acetyl mannosamine and β-fluoropyruvic acid to give 3-fluoro-N-acetyl-neuraminic acid. This common intermediate was then used in selective protection protocols and Barton-McCombie deoxygenations to generate the complete set of mono-deoxygenated 3-fluoro-N-acetylneuraminic acid derivatives. Finally, a fluorination step utilising (diethylamino)sulfur trifluoride (DAST) was used to successfully generate each of the target difluorides.

Full text of DOI:10.1016/j.carres.2013.03.026

Carbohydrate Research 374 (2013) 23–28
Contents lists available at SciVerse ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
The synthesis of a series of deoxygenated 2,3-difluoro-N-acteylneu-
raminic acid derivatives as potential sialidase inhibitors
⇑
Stefan Hader, Andrew G. Watts
Medicinal Chemistry Group, Department of Pharmacy and Pharmacology, University of Bath, Claverton Down, Bath BA2 7AY, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Here we describe the successful syntheses of a series of 4-, 7-, 8- and 9-deoxygenated 2,3-difluoro-N-
acetylneuraminic acid derivatives as potential mechanism-based inhibitors of sialidases. The syntheses
commenced utilising an enzyme-catalysed aldolase reaction between N-acetyl mannosamine and b-flu-
oropyruvic acid to give 3-fluoro-N-acetyl-neuraminic acid. This common intermediate was then used in
selective protection protocols and Barton–McCombie deoxygenations to generate the complete set of
mono-deoxygenated 3-fluoro-N-acetylneuraminic acid derivatives. Finally, a fluorination step utilising
(diethylamino)sulfur trifluoride (DAST) was used to successfully generate each of the target difluorides.
Ó 2013 Elsevier Ltd. All rights reserved.
Received 8 January 2013
Received in revised form 4 March 2013
Accepted 23 March 2013
Available online 2 April 2013
Keywords:
Sialic acid
Barton–McCombie deoxygenation
Luperox
Fluorination
Sialidases
1. Introduction
sylation and deglycosylation. A ‘good leaving group’ at the anomer-
ic position, such as fluorine, selectively rescues the rate of
Sialidases are a superfamily of sialic acid-degrading enzymes
(EC 3.2.1.18) widely found in higher eukaryotes and in a vast num-
ber of microbial pathogens.1 Owing to the importance that sialid-
glycosylation allowing the covalently linked sialosyl–enzyme
intermediate to be kinetically accessible and accumulate to a high
steady-state concentration.
ases play in
a
range of biological processes, the ability to
Given the important role of sialidases in numerous diseases it is
considered that 2,3-difluoro sialic acids represent a novel class of
sialidase inhibitors with the potential to be developed into thera-
peutics. However, the reported inactivation of Trypanosoma cruzi
trans-sialidase by 2,3-difluoro-N-acetylneuraminic acid 1 required
very high concentrations of inhibitor (5 mM), which was consid-
ered to be largely attributable to the rapid turn-over of the cova-
lent intermediate (high k2). This rapid turn-over of the covalent
intermediate has also been observed with sialidases from other
pathogenic organisms inhibited by difluorosialic acids, suggesting
that increasing the stability (half-life) of the covalent intermediate
will be necessary to improve the inhibitory properties of this class
of compound.11 Towards this aim, we considered that a detailed
understanding of the individual contribution by each hydroxyl
group of sialic acid towards transition-state stabilisation of siali-
dase catalysed hydrolysis would provide valuable information for
improving the half-lives of the covalently bound species.
A previous study by Street et al. has investigated substrate/pro-
tein interactions through specific modifications of the substrate,
creating carefully selected analogues that differ in their electronic
properties and hydrogen-bonding capability, but retaining their
ability to be accepted and/or processed by the enzyme.14,15 These
studies, performed on glycogen phosphorylase, demonstrated the
usefulness of deoxygenated substrates to gain insight into both
moderate their enzymatic function is associated with numerous
therapeutic applications ranging from the treatment of type II dia-
betes,2 to their targeting by antiviral3–5 and cancer6 therapeutics.
Sialidases are known to effect glycoside hydrolysis with a net
retention of anomeric stereochemistry (retaining glycosidases)
however debate still exists over the nature of the catalytic mecha-
nism through which this hydrolytic reaction is performed. Re-
cently, a number of studies on family GH33 (CAZy)7 sialidases
have demonstrated that these enzymes operate through a two-
step, double-displacement mechanism similar to the majority of
retaining glycosidases, but involve the participation of a tyrosine
residue as the catalytic nucleophile to form, uniquely, a covalent
aryl-glycoside intermediate.8–11 This covalent intermediate was
originally observed with Trypanosoma cruzi trans-sialidase using
the novel mechanism-based inhibitor 2,3-difluoro N-acetylneu-
raminic acid 1, developed to attenuate glycosylation (k1) and
deglycosylation (k2) rates in the catalytic cycle of sialidases
(Scheme 1).12,13 The introduction of an electronegative fluorine
atom at C-3 serves to destabilise the formation of a positive charge
during the transition-state, which reduces the rates of both glyco-
⇑
Corresponding author. Tel.: +44 (0) 1225 386788; fax: +44 (0) 1225 386114.
E-mail address: a.watts@bath.ac.uk (A.G. Watts).
0008-6215/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.carres.2013.03.026

24
S. Hader, A. G. Watts / Carbohydrate Research 374 (2013) 23–28
OH
HO
HO
F
O
HO
OH
OH
OH
HO
H2O
HF
HO
HO
OH
HO
OH
HO
CO2
CO2
O
O
O
CO2
AcHN
AcHN
AcHN
OH
F
F
F
k1
k2
1
Scheme 1. The glycosylation (k1) and deglycosylation (k2) rate constants are attenuated by fluorine-containing mechanism-based inactivators.
the electronic structure of the enzymatic ground-state Michaelis
complex, as well as the transition-state of the reaction.14,15
Herein we describe the synthesis of the complete series of
mono-deoxygenated 2,3-difluorsialic acids 2–5, designed to facili-
tate the dissection of individual hydroxyl group contributions to-
wards transition-state stabilisation during the processing of this
novel class of inhibitors by sialidases.
4 to generate the 4-deoxy-3-fluoro methyl ester 11 from the
thionocarbonate 10. Although the deoxygenation of sialic acid at
C-4 has been reported previously, the radical initiator used in the
vast majority of these reports (AIBN) is now difficult to obtain com-
mercially, prompting us to investigate a variety of radical initiators,
radical carriers, solvents and temperatures for this transformation.
Initially, bis(tributyltin) oxide was used to generate tributyltin
hydride in situ according to the procedure previously reported of
Lopez et al.18 Here, poly(methylhydrosiloxane) (PMHS) and n-
butanol in toluene are used to generate two equivalents of tributyl-
tin hydride, with 1,10 azo bis(cyanocyclohexane) (ACHN) used as
the radical initiator. As such, the 4-O-(phenoxy)thiocarbonyl deriv-
ative 10 was subjected to bis(tributyltin) oxide, n-BuOH and
azobiscyanocyclohexane in toluene/DMF overnight at 80 °C (Ta-
ble 1, entry 1).18 However, under these conditions only the starting
compound 10 was recovered. Attempts to drive the conversion for-
ward by increasing the temperature to 90 °C only resulted in
decomposition of compound 10 (Table 1, entry 2).
2. Results and discussion
2.1. Synthesis of 5-N-acetyl-2,3,4,5-tetradeoxy-2,3-difluoro-
D-
erythro-b- -manno-non-2-ulopyranosonic acid (2)
L
The synthesis of 5-N-acetyl-2,3,4,5-tetradeoxy-2,3-difluoro-
D-
erythro-b- -manno-non-2-ulopyranosonic acid 2 began with the
L
formation of 3-fluoro-N-acetyl-neuraminic acid 8, as previously
described by Watts and Withers using an aldolase mediated reac-
tion (Scheme 2).10
As a consequence of this decomposition, an alternative method
reported by Park et al.19 to effect facile Barton–McCombie deoxy-
genation was then attempted. Here, the 4-O-(phenoxy)thiocarbonyl
derivative 10 was subjected to the initiator tetrabutylammonium
peroxydisulfate and radical promoter sodium formate in DMF at
60 °C for 24 h as previously described by Park et al. (Table 1, entry
3). Following aqueous work up and purification by silica chromatog-
raphy (EtOAc/MeOH), the desired product 11 was successfully ob-
tained in 17% yield. However, in addition to the formation of 11,
both starting material 10 and the product of hydrolysis 9 were also
observed.20,21 Several attempts to improve the yield by altering the
time and temperature of reaction could only improve the conversion
to provide 11 in up to 24% yield (Table 1, entry 4). At this stage it was
consideredthat the reducedrates of conversion observed in compar-
ison to previous reports were likely resulting from the presence of
fluorine at C-3 having a significant influence on the progress of the
deoxygenation. It was considered that the use of a more effective
promoter in tributyltinhydride may overcome the retardive influ-
ence of the fluorine.
According to this procedure, 2-N-acetyl-D-mannosamine 6 and b-
fluoropyruvic acid 7 were subjected to Neu5Ac aldolase (EC 4.1.3.3)
in aqueous solution at room temperature for 5 days to give 3-fluoro-
N-acetyl-neuraminic acid 8. The crude residue was subsequently
esterified using trifluoroacetic acid (TFA) in methanol and then trea-
ted with 2,20-dimethoxypropane and catalytic amounts of p-tolu-
enesulfonic acid (TSA) in acetone according to the procedure of
Ogura et al.16 to give the 3-fluoro-8,9-O-isopropylidene-N-acetyl-
neuraminic acid methyl ester 9 in 71% overall yield from b-fluoro-
pyruvic acid 7.
For the subsequent deoxygenation of C-4, we first sought to
investigate the use of thionocarbonates and O-thiocarbamates as
these groups can be introduced under essentially neutral conditions
from O-phenyl chlorothionoformate and 1,10-thiocarbonyldiimi-
dazole respectively. Additionally, both thionocarbonates and
O-thiocarbamates have been shown to react at lower temperatures
during radical deoxygenation and have been reported previously for
the successful deoxygenation of sialic acid derivatives.17
As such, 3-fluoro-8,9-O-isopropylidene-N-acetyl-neuraminic
acid methyl ester 9 was treated with phenyl chlorothionoformate
in pyridine and dichloromethane to give the 4-O-(phenoxy)thio-
carbonyl methyl ester 10 in 76% yield. We next sought to identify
suitable conditions for the Barton–McCombie deoxygenation at C-
As such, the 4-O-(phenoxy)thiocarbonyl derivative 10 was sub-
jected to a general Barton–McCombie protocol utilising tributyltin
hydride and ACHN in toluene under reflux conditions (Table 1, en-
try 5). Following aqueous workup and purification, the desired
O
O
OH
OH
O
OH
O
HO
O
HO
HO
OH
OH
NHAc
O
OH
OH
O
O
a
CO2Me
b,c
AcHN
PhO
d
O
O
+
CO2H
CO2Me
AcHN
AcHN
HO
F
F
CO2Na
HO
HO
OH
F
F
S
6
7
8
9
10
Scheme 2. Reagents and conditions: (a) N-Acetylneuraminic aldolase, H2O, 5 d, rt; (b) MeOH, TFA, 16 h, rt; (c) 2,2-dimethoxypropane, p-toluenesulfonic acid, Et3N, acetone,
4 h, 50 °C, 71% (over 3 steps); (d) phenyl chlorothionoformate, pyridine/CH2Cl2, 4 h, ꢀ40 °C ? rt, 76%.

S. Hader, A. G. Watts / Carbohydrate Research 374 (2013) 23–28
25
Table 1
Overview of the different conditions used for the synthesis of 4-deoxy-3-fluoro-8,9-O-isopropylidene-N-acetylneuraminic acid methyl ester 11
O
O
O
O
OH
OH
O
O
OH
O
OH
O
O
O
OH
OH
OH
OH
O
+
O
+
CO2Me
CO2Me
AcHN
PhO
AcHN
PhS
O
O
CO2Me
CO2Me
AcHN
AcHN
HO
F
F
F
F
S
O
9
12
11
10
Entry
1
Initiator (equiv)
ACHNb (0.15)
Promotor (equiv)
Solvent
Time (h)
O/N
T (°C)
Product
s.m.
Yielda (%)
n.d.
(Bu3Sn)2O (0.037)
PMHS (5)
Toluene/DMF
80
n-BuOH (5.5)
(Bu3Sn)2O (0.037)
PMHS (5)
2
ACHNb (0.15)
Toluene/DMF
3.5
90
dec
n.d.
n-BuOH (5.5)
NaHCO3 (6)
NaHCO3 (6)
Bu3SnH (3.7)
Bu3SnH (3.7)
3
4
5
6
(Bu4N)2S2O8 (3)
(Bu4N)2S2O8 (3)
ACHNb (0.3)
DMF
DMF
Toluene
1,4-Dioxane
24
5
24
4
60
70
120
100
11
11
11
11
17
24
24
81
BTBPBc (0.45)
dec = Decomposition.
n.d. = Not determined.
a
Isolated yields after silica chromatography.
b
1,10 Azo bis(cyanocyclohexane).
c
50% 2,2-Bis(tert-butylperoxy)-butane mineral oil-solution.
OAc
O
AcO
O
OAc
OAc
AcO
OH
OH
OAc
OH
11
c
OAc
b
a
O
O
O
CO2Me
CO2Me
AcHN
CO2Me
AcHN
AcHN
F
F
F
14
13
OH
OH
HO
OAc
OAc
AcO
CO2H
F
CO2Me
d
O
O
F
AcHN
AcHN
F
F
2
15 α,β
Scheme 3. Reagents and conditions: (a) (i) AcOH/H2O, 2 h, 60 °C, (ii) Ac2O, pyridine, 3 d, 40 °C, 76% (2 steps); (b) hydrazine acetate, MeOH/CH2Cl2, 20 h, 4 °C, 51%; (c) DAST,
CH2Cl2, 1.5 h, ꢀ40 °C ? ꢀ10 °C, -anomer: 45%, b-anomer: 30%; (d) (i) NaOMe, MeOH, 30 min, rt, (ii) 1 M NaOH, pH 12, 1 h, rt, 52% (2 steps).
a
product 11 was isolated in 24% yield. A major side product result-
ing under these conditions was also identified as the sulfur–oxygen
exchanged carbonate 12. The rearrangement of thionocarbonates
under Barton–McCombie conditions, giving derivatives such as
the O-thiocarbonyl 12, has been described previously by Powers
and Tarbell.22 Hence, in an effort to reduce the degree of rearrange-
ment, 2,2-bis(tert-butylperoxy)-butane (BTBPB) was substituted as
the radical initiator to shorten reaction times while using 1,4-diox-
ane as solvent. Here, 3-fluoro-8,9-O-isopropylidene-4-O-(phen-
oxy)thiocarbonyl-N-acetylneuraminic acid methyl ester 10 was
then subjected to radical Barton–McCombie conditions using tri-
butyltin hydride in the presence of BTBPB in 1,4-dioxane under re-
flux to provide compound 11 in 81% yield (Table 1, entry 6).23
Having successfully deoxygenated the C-4 position, compound
11 was then subjected to 80% aqueous acetic acid according to
the procedure of Anazawa et al.24 to remove the O-8,9-isopropyli-
dene group, followed by acetylation using acetic anhydride in
pyridine to give the per-O-acetylated 4-deoxy-3-fluoro-N-acetyl-
neuraminic acid 13 in 76% yield over the two steps (Scheme 3).
Selective deprotection at the C-2 position of compound 13 was
then performed according to a known procedure using hydrazine
acetate in dichloromethane/MeOH,10,25–27 to afford the hemiketal
14 in 51% yield. Subsequent fluorination by treatment with DAST
in dichloromethane at ꢀ40 to ꢀ10 °C for 1.5 h resulted in the for-
mation of compounds 15-a,b as an anomeric mixture which
proved separable by silica chromatography.
Unfortunately, the assignment of absolute stereochemistry to
the two compounds on the basis of coupling constants, by either
1H or 19F NMR analysis, was not possible. As such, small molecule
X-ray crystallographic analysis of the less polar compound was uti-
lised to confirm that the fluorine atom at position C-2 for this com-
pound was axial, being labelled F1 in the X-ray structure shown
(Fig. 1). Upon this basis, it was determined that the desired a-ano-
mer was isolated in 45%, along with the b-anomer isolated in 30%
yield.
The desired
raminic acid methyl ester 15-
lowing Zemplén28 conditions using sodium in methanol at room
a-per-O-acetyl-4-deoxy-2,3-difluoro-N-acetyl-neu-
a
was then globally deacetylated fol-

26
S. Hader, A. G. Watts / Carbohydrate Research 374 (2013) 23–28
ester 20 in 95% yield (Scheme 5). The 7-O-thiocarbamate derivative
20 was then subjected to Barton–McCombie deoxygenation condi-
tions using LuperoxÒ 101 as a radical initiator and tributyltin hy-
dride in 1,4-dioxane under reflux conditions to give the 2,4-di-O-
benzoyl-7-deoxy-3-fluoro-8,9-O-isopropylidene-N-acetylneurami-
nic acid methyl ester 21 in 89% yield. Here, LuperoxÒ 101 was found
to have advantages over BTBPB as a radical initiator such as ease of
use and increased stability. Compound 21 was then subjected to so-
dium in methanol and the resultant crude mixture then treated with
80% aqueous acetic acid, followed by acetylation using acetic anhy-
dride in pyridine to give the per-O-acetyl-7-deoxy-3-fluoro-N-acet-
ylneuraminic acid methyl ester 22 in 83% yield over the 3 steps.
Compound 22 was then selectively deprotected at the C-2 posi-
tion using hydrazine acetate in a mixture of dichloromethane and
methanol at 4 °C to afford the hemiketal 23 in 84% yield. Treatment
of the hemiketal 23 with DAST in dichloromethane at ꢀ40 to
ꢀ10 °C according to the method described previously, then affor-
ded the products of fluorination as a mixture of
Despite exhaustive efforts, it was not found to be possible to sep-
arate the - and b-anomers using silica chromatography at this
a- and b-anomers.
a
stage, though it was considered that purification might be achieved
following deacetylation of the mixture.
Figure 1. X-ray crystallographic structure of the C-2 axial fluorine containing per-
O-acetyl-4-deoxy-2,3-difluoro-N-acetyl-neuraminic acid methyl ester 15-b.
As such, the crude anomeric mixture of difluorides was deacet-
ylated using sodium in methanol at room temperature to give the
7-deoxy-2,3-difluoro-N-acetyl-neuraminic acid methyl esters 24-
temperature, followed by saponification of the crude residue using
0.5 M aqueous sodium hydroxide and purification by silica chro-
matography to give the target compound, 5-N-acetyl-2,3,4,5-tetra-
a
,b. Subsequent purification by silica chromatography provided
the desired -anomer in 41% yield along with the b-anomer in
a
deoxy-2,3-difluoro-D-erythro-b-L-manno-non-2-ulopyranosonic acid
20% yield, over the two steps.
2, in 52% yield from 15-
a.
Finally, saponification of the desired methyl ester 24-a with
0.5 M aqueous sodium hydroxide at room temperature yielded
2.2. Synthesis of 5-N-acetyl-2,3,5,7-tetradeoxy-2,3-difluoro-
D
-
the target compound 3 in 49% yield.
erythro-b- -manno-non-2-ulopyranosonic acid (3)
L
2.3. Synthesis of 5-N-acetyl-2,3,5,8-tetradeoxy-2,3-difluoro-
D-
For the synthesis of 5-N-acetyl-2,3,5,7-tetradeoxy-2,3-difluoro-
D-
erythro-b- -manno-non-2-ulopyranosonic acid (4)
L
erythro-b- -manno-non-2-ulopyranosonic acid 3, the previously
L
generated intermediate 9 was treated with benzoyl chloride in
dichloromethane and pyridine at ꢀ40 °C for 90 min to afford 2,4-
di-O-benzoyl-3-fluoro-8,9-O-isopropylidene-N-acetylneuraminic acid
methyl ester 16 in 54% yield (Scheme 4). A major side product also
formed under these conditions was identified as the 2,4,7-tri-O-
benzoyl-3-fluoro-8,9-O-isopropylidene-N-acetyl-neuraminic acid
methyl ester 17, which was isolated in 18% yield. The introduction
of an O-thionocarbonate at OH-7 was then performed using phenyl
chlorothionoformate under conditions previously employed for the
synthesis of 10, which yielded the desired product 18 along with
the product of sulfur–oxygen exchange 19, previously observed
by Powers and Tarbell.22 as an inseparable mixture in a ratio of 1:1.
Given that the attempted introduction of a thionocarbonate at
O-7 had resulted in formation of an inseparable mixture, it was con-
sidered that the use of a thiocarbamate according to the method of
Miyazaki et al.17 could prove more successful. Hence, the 2,4-di-O-
benzoyl-3-fluoro-8,9-O-isopropylidene-sialic acid methyl ester 16
was treated with 1,10-thiocarbonyldiimidazole in anhydrous dichlo-
romethane to provide the 2,4-di-O-benzoyl-3-fluoro-8,9-O-isopro-
pylidene-7-O-thiocarbamate-N-acetylneuraminic acid methyl
For the synthesis of 5-N-acetyl-2,3,5,8-tetradeoxy-2,3-difluoro-
-erythro-b- -manno-non-2-ulopyranosonic acid 4, the previously
synthesised methyl ester derivative 9 was subjected to acetic
anhydride in pyridine for 7 days to give the 2,4,7-tri-O-acetyl-3-
fluoro-8,9-O-isopropylidene-N-acetyl-neuraminic acid methyl es-
ter 25 in 91% yield (Scheme 6).
Next, compound 25 was hydrolysed using 80% aqueous acetic
acid, followed by the addition of benzoyl chloride in pyridine at
room temperature to give selectively the O-9-benzoyl derivative
26 in 70% yield over the two steps. Compound 26 was then treated
with phenyl chlorothionoformate in pyridine at 0 °C to afford the 2,
4,7-tri-O-acetyl-9-O-benzoyl-3-fluoro-8-O-(phenoxy)thiocarbonyl-N-
acetylneuraminic acid methyl ester 27 in 84% yield. Subsequent
Barton–McCombie deoxygenation of compound 27 using LuperoxÒ
101 and tributyltin hydride in 1,4-dioxane under reflux as de-
scribed earlier, provided the 2,4,7-tri-O-acetyl-9-O-benzoyl-8-
deoxy-3-fluoro-N-acetyl-neuraminic acid methyl ester 28 in 80%
yield.
D
L
The selective anomeric deprotection of compound 28 was then
accomplished using hydrazine acetate in dichloromethane and metha-
S
O
OPh
OBz
SPh
O
O
O
O
O
O
O
O
R1
OBz
OH
R2
OH
HO
O
O
a
b
+
O
O
O
F
O
CO2Me
CO2Me
CO2Me
CO2Me
AcHN
AcHN
AcHN
AcHN
BzO
BzO
R1
F
F
F
19
16 R1 = OBz, R2 = OH
17 R1 = R2 = OBz,
18
9
Scheme 4. Reagents and conditions: (a) Benzoylchloride, CH2Cl2/pyridine, 1.5 h, ꢀ40 °C ? ꢀ20 °C, 16: 54%, 17: 18%; (b) phenyl chlorothionoformate, CH2Cl2/pyridine, O/N, rt,
18:19; 1:1.

S. Hader, A. G. Watts / Carbohydrate Research 374 (2013) 23–28
27
N
S
N
O
O
O
O
O
O
OBz
OBz
OBz
OH
O
a
c
b
O
O
O
CO2Me
CO2Me
CO2Me
AcHN
AcHN
AcHN
BzO
BzO
BzO
F
F
F
20
21
16
OH
AcHN
HO
OAc
CO2Me
F
OAc
AcO
AcO
OH
OAc
f
e
d
O
3
O
O
CO2Me
CO2Me
AcHN
AcO
AcHN
AcO
HO
F
F
F
24
23
22
Scheme 5. Reagents and condtions: (a) 1,10 Thiocarbonyldiimidazole, CH2Cl2, 20 h, 40 °C, 95%; (b) LuperoxÒ 101, Bu3SnH, 1,4-dioxane, 4 h, 100 °C, 89%; (c) (i) NaOMe, MeOH,
36 h, 40 °C, (ii) 80% AcOH/H2O, 2 h, 60 °C, (iii) Ac2O, DMAP, pyridine, O/N, rt, 83% (3 steps); (d) hydrazine acetate, MeOH/CH2Cl2, 20 h, 4 °C, 84%; (e) (i) DAST, CH2Cl2, 1.5 h,
ꢀ40 °C ? ꢀ10 °C, (ii) NaOMe, MeOH, 3 h, rt,
a-anomer: 41% (2 steps), b-anomer: 20% (2 steps); (f) 1 M NaOH, pH 12, 1 h, rt, 49%.
O
O
O
OH
O
BzO
OAc
OH
OAc
OAc
OH
a
OAc
c
b
O
O
O
CO2Me
CO2Me
CO2Me
AcHN
AcO
AcHN
AcHN
AcO
HO
F
F
F
9
25
26
OPh
S
O
BzO
BzO
AcHN
BzO
OAc
OAc
OH
OAc
d
e
f
OAc
OAc
O
O
O
O
CO2Me
CO2Me
CO2Me
AcHN
AcO
AcHN
AcO
28
AcO
F
F
F
27
29
HO
AcHN
CO2Me
OH
g
F
4
HO
F
30
Scheme 6. Reagents and conditions: (a) Ac2O, pyridine, 7 d, rt, 91%; (b) (i) 80% AcOH/H2O, 2 h, 60 °C, (ii) benzoylchloride, pyridine, O/N, rt, 70% (2 steps); (c) phenyl
chlorothionoformate, pyridine, 15 h, 0 °C ? rt, 84%; (d) LuperoxÒ 101, Bu3SnH, 1,4-dioxane, 4 h, 100 °C, 80%; (e) hydrazine acetate, MeOH/CH2Cl2, 20 h, 4 °C, 83%; (f) (i) DAST,
CH2Cl2, 1.5 h, ꢀ40 °C ? ꢀ10 °C, (ii) NaOMe, MeOH, O/N, rt, (iii) TFA, MeOH, O/N, rt,
a-anomer: 25%, b-anomer: 11% (3 steps); (g) 1 M NaOH, pH 12, 1 h, rt, 22%.
nol at 4 °C to give the hemiketal 29 in 83% yield. The subsequent treat-
ment of 29 with DAST in dichloromethane at ꢀ40 to ꢀ10 °C then affor-
jected to Barton–McCombie deoxygenation using LuperoxÒ 101 and tri-
butyltin hydride in 1,4-dioxane under reflux as described previously, to
yield the 9-deoxy-3-fluoro-8-O-methylthiocarbonyl-N-acetylneurami-
nic acid methyl ester 33 in 98% yield. The 9-deoxy derivative 33 was then
treated with sodium in methanol and the subsequent product then
acetylated using acetic anhydride in pyridine to give the 9-deoxy
methyl ester derivative 34 in 77% yield over the 2 steps.28
ded the difluorides 30-
30-
icachromatography,thenprovided the desired
a,b as an inseparable mixture. Deacetylation of
a
,b using sodium in methanol at room temperature, followed by sil-
a
-anomer in 25% yield,
as well as the b-anomer in 11% yield, over the 3 steps.
Finally, ester saponification of the desired compound 30-
a
using 0.5 M aqueous sodium hydroxide then gave the target com-
pound 4 in 22% yield from 29.
Next, compound 34 was selectively deprotected at position C-2
using hydrazine acetate in dichloromethane and methanol at 4 °Ctoaf-
ford the hemiketal derivative 35. Treatment of compound 35 with
DAST in dichloromethane at ꢀ40 to ꢀ10 °C as mentioned previously,
2.4. Synthesis of 5-N-acetyl-2,3,5,9-tetradeoxy-2,3-difluoro-
D-
afforded 36-a,b as a 1:1 mixture of the a- and b-fluorides, each isolated
erythro-b- -manno-non-2-ulopyranosonic acid (5)
L
in 38% yield at this stage following silica chromatography.
The methyl ester derivative 36- was then deacetylated using
a
For the synthesis of 5-N-acetyl-2,3,5,9-tetradeoxy-2,3-difluoro-
-erythro-b- -manno-non-2-ulopyranosonic acid 5, the previously
sodium in methanol, followed by saponification of the resulting
crude residue using 0.5 M aqueous sodium hydroxide to give the
desired 9-deoxy-2,3-difluoro-N-acetylneuraminic acid 5 in 99%
over the 2 steps.
D
L
prepared methyl ester 17 was hydrolysed using 80% aqueous acetic
acid and the crude mixture then treated with 1,10-thiocarbonyldi-
imidazole in anhydrous dichloromethane to give the desired 2,4,7-tri-
O-benzoyl-8,9-O-carbonothioate-3-fluoro-N-acetylneuraminic
acid methyl ester 31 in 81% yield over the 2 steps (Scheme 7).
Compound 31 was then subjected to freshly distilled iodomethane at
56 °C overnight as previously reported by Schreiner et al.29 to give the 9-
iodo-8-O-methylthiocarbonyl-N-acetylneuraminic acid derivative 32 in
95% yield. The 8-O-methylthiocarbonyl derivative 32 was then sub-
3. Conclusions
Herein we have described the successful syntheses of the four
novel monodeoxygenated 2,3-difluorosialic acids 2–5 as potential
inhibitors and mechanistic probes for sialidases. A number of ap-
proaches were investigated for the synthesis of the 4-deoxy deriv-

28
S. Hader, A. G. Watts / Carbohydrate Research 374 (2013) 23–28
S
S
O
I
O
O
O
O
OBz
OBz
O
OBz
OBz
OBz
OBz
c
a
b
O
O
O
CO2Me
CO2Me
AcHN
AcHN
CO2Me
AcHN
BzO
BzO
BzO
F
F
F
17
32
31
S
O
OAc
OAc
OAc
OAc
O
OH
OAc
OBz
e
OBz
d
f
O
O
O
CO2Me
CO2Me
CO2Me
AcHN
AcO
AcHN
AcHN
BzO
AcO
F
F
F
35
33
34
OAc
OAc
CO2Me
F
g
5
O
AcHN
AcO
F
36
Scheme 7. Reagents and conditions: (a) (i) 80% AcOH/H2O, 2 h, 60 °C, (ii) 1,10 thiocarbonyldiimidazole, CH2Cl2, 2 d, 40 °C, 81% (2 steps); (b) CH3I, 20 h, 56 °C, 95%; (c) LuperoxÒ
101, Bu3SnH, 1,4-dioxane, 4 h, 100 °C, 98%; (d) (i) NaOMe, MeOH, O/N, rt, (ii) Ac2O, DMAP, pyridine, O/N, rt, 77% (2 steps); (e) hydrazine acetate, MeOH/CH2Cl2, 20 h, 4 °C, 68%;
(f) DAST, CH2Cl2, 1.5 h, ꢀ40 °C ? ꢀ10 °C,
a-anomer: 38%, b-anomer: 38%; (g) (i) NaOMe, MeOH, 3 h, rt, (ii) 1 M NaOH, pH 12, 1 h, rt 99% (2 steps).
ative 2, where a Barton–McCombie protocol using the radical car-
rier tributyltin hydride under reflux conditions in 1,4-dioxane was
found to be effective and proved to be generally applicable to the
synthesis of the remaining monodeoxygenated 2,3-difluorosialic
acid derivatives 3–5. Analysis of the small molecule X-ray structure
of the difluoride 15 proved necessary for the unambiguous assign-
ment of its anomeric stereochemistry, enabling the successful syn-
thesis of 2. The Barton–McCombie conditions identified for the
synthesis of 2 were also applied to the syntheses of targets 3–5,
with the exception that Luperox 101 was employed as a free radi-
cal initiator in preference to BTBPB. The peroxide Luperox 101 is a
commonly used initiator for free radical polymerisation reactions,
though its use has not been reported previously for application un-
der Barton–McCombie deoxygenation conditions. Here, we found
Luperox 101 to be an effective radical initiator for the Barton–
McCombie deoxygenation of sialic acids, suggesting that this re-
agent may prove to be generally useful as an alternative to the tra-
ditional initiator AIBN for tinhydride mediated deoxygenations.
Another notable observation is the variability in the ratio of
anomeric fluorides generated here from treatment of the hemi-ke-
tals with DAST. Previous studies generating 2,3-difluoro sialic acid
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.carres.2013.
03.026.
References
1. Vimr, E.; Lichtensteiger, C. Trends Microbiol. 2002, 10, 254–257.
2. Treadway, J. L.; Mendys, P.; Hoover, D. J. Expert Opin. Invest. Drugs 2001, 10,
439–454.
3. Groopman, J. E. Rev. Infect Dis. 1990, 12, 908–911.
4. Zitzmann, N.; Mehta, A. S.; Carrouee, S.; Butters, T. D.; Platt, F. M.; McCauley, J.;
Blumberg, B. S.; Dwek, R. A.; Block, T. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
11878–11882.
5. Laver, W. G.; Bischofberger, N.; Webster, R. G. Sci. Am. 1999, 280, 78–87.
6. Gloster, T. M.; Davies, G. J. Org. Biomol. Chem. 2010, 8, 305–320.
7. Cantarel, B. L.; Coutinho, P. M.; Rancurel, C.; Bernard, T.; Lombard, V.; Henrissat,
B. Nucleic Acids Res. 2009, 37, D233–D238.
8. Koshland, D. E. Biol. Rev. Camb. Philos. Soc. 1953, 28, 416–436.
9. Zechel, D. L.; Withers, S. G. Acc. Chem. Res. 2000, 33, 11–18.
10. Watts, A. G.; Withers, S. G. Can. J. Chem. 2004, 82, 1581–1588.
11. Watts, A. G.; Oppezzo, P.; Withers, S. G.; Alzari, P. M.; Buschiazzo, A. J. Biol.
Chem. 2006, 281, 4149–4155.
12. Watts, A. G.; Damager, I.; Amaya, M. L.; Buschiazzo, A.; Alzari, P.; Frasch, A. C.;
Withers, S. G. J. Am. Chem. Soc. 2003, 125, 7532–7533.
13. Buchini, S.; Buschiazzo, A.; Withers, S. G. Angew. Chem., Int. Ed. 2008, 47, 2700–
2703.
14. Street, I. P.; Armstrong, C. R.; Withers, S. G. Biochemistry 1986, 25, 6021–6027.
15. Street, I. P.; Rupitz, K.; Withers, S. G. Biochemistry 1989, 28, 1581–1587.
16. Ogura, H.; Furuhata, K.; Sato, S.; Anazawa, K.; Itoh, M.; Shitori, Y. Carbohydr.
Res. 1987, 167, 77–86.
17. Miyazaki, T.; Sato, H.; Sakakibara, T.; Kajihara, Y. J. Am. Chem. Soc. 2000, 122,
5678–5694.
18. Lopez, R. M.; Hays, D. S.; Fu, G. C. J. Am. Chem. Soc. 1997, 119, 6949–6950.
19. Park, H. S.; Lee, H. Y.; Kim, Y. H. Org. Lett. 2005, 7, 3187–3190.
20. Tsuda, Y.; Sato, Y.; Kanemitsu, K.; Hosoi, S.; Shibayama, K.; Nakao, K.; Ishikawa,
Y. Chem. Pharm. Bull. 1996, 44, 1465–1475.
21. Barton, D. H. R.; Dalko, P. I.; Gero, S. D. Tetrahedron Lett. 1992, 33, 1883–1886.
22. Powers, D. H., Jr.; Tarbell, D. S. J. Am. Chem. Soc. 1956, 78, 70–71.
23. Barton, D. H. R.; McCombie, S. W. J. Chem. Soc., Perkin Trans. 1 1975, 1574–1585.
24. Anazawa, K.; Furuhata, K.; Ogura, H. Chem. Pharm. Bull. 1988, 36, 4976–4979.
25. Excoffier, G.; Gagnare, D.; Utille, J. P. Carbohydr. Res. 1975, 39, 368–373.
26. Alais, J.; David, S. Carbohydr. Res. 1992, 230, 79–87.
derivatives consistently report over 80% yield of the desired a-ano-
mer when DAST fluorinations were performed at a constant tem-
perature of ꢀ30 °C.10,11 We consider that variable ratios of
a-
anomer observed here (50%-
a for 36 to 70%-a for 24) may result
from our modified procedure where the DAST fluorinations were
performed over a temperature range from ꢀ40 to ꢀ10 °C.
The series of deoxygenated difluorosialic acids 2–5 are currently
being investigated as inhibitors against a host of sialidases of bac-
terial, trypanosomal and viral origin, with the results of these ki-
netic and crystallographic studies to be presented in the future.
Acknowledgments
S.H. was supported by an MRC capacity building studentship.
A.G.W. was supported by MRC Grant G0600514. We gratefully
acknowledge Dr. Tim Woodman (Department of Pharmacy and
Pharmacology, University of Bath) for NMR support and Dr. Mary
Mahon (Department of Chemistry, University of Bath) for perform-
ing the X-ray crystallographic analysis.
27. Iida, M.; Endo, A.; Fujita, S.; Numata, M.; Sugimoto, M.; Nunomura, S.; Ogawa,
T. J. Carbohydr. Chem. 1998, 17, 647–672.
28. Zemplen, G.; Pacsu, E. Ber. Dtsch. Chem. Ges. 1929, 62B, 1613–1614.
29. Schreiner, E.; Christian, R.; Zbiral, E. Liebigs Ann. Chem. 1990, 93–97.