A simple synthesis of C-8 modified 2-keto-3-deoxy-d-manno-octulosonic acid (KDO) derivatives

Article abstract of DOI:10.1055/s-0029-1219356

This paper describes a simple and efficient method with which to prepare C-8 modified 2-keto-3-deoxy-D-manno-octu-losonic acid (KDO) derivatives from C-5 modified arabinose derivatives. Georg Thieme Verlag Stuttgart.

Full text of DOI:10.1055/s-0029-1219356

LETTER
583
A Simple Synthesis of C-8 Modified 2-Keto-3-deoxy-D-manno-octulosonic Acid
(KDO) Derivatives
S
ynthesis of C-
8
e
Modified K
n
D
O Derivat
e
Institute for Glycomics, Griffith University, Gold Coast Campus, Queensland 4222, Australia
Fax +61(7)55528098; E-mail: m.kiefel@griffith.edu.au
Received 18 December 2009
This paper is dedicated to Prof. Gerry Pattenden on the occasion of his 70th birthday – thank you for the wonderful insight into chemistry.
HO
O
HO
HO
HO
Abstract: This paper describes a simple and efficient method with
which to prepare C-8 modified 2-keto-3-deoxy-D-manno-octu-
losonic acid (KDO) derivatives from C-5 modified arabinose deriv-
atives.
HO
OH
+
O
1. pH 11
CO2H
O
2. pH ~5 (– CO2)
HO2C
CO2H
HO
HO
OH
1 KDO
Key words: carbohydrates, ulosonic acids, aldol condensation,
benzylation, protecting groups
Scheme 1 Cornforth’s method for the synthesis of KDO
Whilst there are several published procedures for the syn-
thesis of KDO, many of these methods require the use of
expensive reagents or involve multiple steps. Our aim was
to find a single method that utilized relatively cheap re-
agents and would be flexible enough to allow the prepara-
tion of structural analogues of KDO. Towards this end,
the method based on Cornforth’s procedure, although not
giving high yields, appeared attractive to us. In 1986
McNicholas et al.9a described a modification of the origi-
nal procedure for the synthesis of KDO, wherein sodium
carbonate was used to ensure that an optimal pH 11 was
maintained for the aldol condensation. In this way KDO
was consistently obtained in yields of ~35%. Subsequent-
ly, it was suggested9b that the modest yields for KDO may
be due to the lack of optimised conditions for the decar-
boxylation step that follows the aldol condensation. By
using a catalytic amount of NiCl2 in the decarboxylation
step, the yield of KDO was reported to increase to an ac-
ceptable 66%.9b
The eight-carbon acidic sugar 2-keto-3-deoxy-D-manno-
octulosonic acid (KDO, 1) is a member of the family of 3-
deoxy-2-ulosonic acids. KDO is an essential component
of the outer membrane lipopolysaccharide (LPS), or lipo-
oligosaccharide (LOS), of Gram-negative bacteria, where
it forms the link between the lipid A and polysaccharide
components of the LPS.1–3 KDO appears to be unique to
Gram-negative bacteria and, since KDO biosynthetic
pathway knockout mutants are no longer viable,1b,4 it is
reasonable to assume that small molecule inhibitors of
KDO biosynthesis have the potential to act as a new class
of antibacterial agent.3 As part of a program aimed at ex-
ploring the binding specificity of enzymes involved in the
biosynthesis of KDO and its incorporation into the LPS,
we required an efficient and flexible route towards a range
of structurally modified KDO derivatives. Specifically,
we were interested in preparing C-8 modified KDO deriv-
atives since, of those bacteria that have had their LPS
structures elucidated, most have more than one KDO unit
present in their core region, with the additional KDO res-
idues usually attached to KDO through the C-4 or C-8 hy-
droxy groups.2
In using Cornforth’s method to prepare our target C-8
modified KDO derivatives, we required ready access to
C-5 modified arabinose derivatives, since carbon-5 of
arabinose becomes carbon-8 of KDO after the aldol con-
densation. Our target compounds were those in which the
normal 5-hydroxy group of arabinose was replaced by
–OR, –SR or –NHR. It was reasoned that, after transfor-
mation into the corresponding C-8 modified KDO deriva-
tives, such functional groups still have the potential to
participate in hydrogen bonding interactions with KDO-
recognising enzymes. In order to introduce these types of
groups at C-5, we envisaged preparing a differentially
protected arabinose derivative such as 2 (Scheme 2).
Compounds such as 2 have been described before,10–12
and the synthetic approaches generally rely on selectively
introducing a temporary protecting group at the 5-hy-
droxy group, protecting the three remaining hydroxy
groups, and then removal of the C-5 protecting group.
However, as noted by others,10 this proved more challeng-
The chemical synthesis of KDO (1) was first reported by
Ghalambor and Heath in 1963,5 following a method de-
veloped by Cornforth for the synthesis of sialic acid.6 The
method involved a base-catalysed aldol condensation be-
tween D-arabinose and oxalacetic acid, followed by decar-
boxylation under mildly acidic conditions (Scheme 1).
Since this first report there have been numerous papers
describing the synthesis of KDO,1a,7 often involving either
the use of D-mannose and the addition of a two-carbon
unit,8 or starting from D-arabinose and adding a three-
carbon unit.9
SYNLETT 2010, No. 4, pp 0583–0586
0
1.
0
3.
2
0
1
0
Advanced online publication: 02.02.2010
DOI: 10.1055/s-0029-1219356; Art ID: D37009ST
© Georg Thieme Verlag Stuttgart · New York

584
R. Winzar et al.
LETTER
ing than could be inferred from previously reported proce- tive unmasking of the thioacetate group in 7 was achieved
dures.
by treatment with hydrazine acetate;13 subsequent expo-
sure to dimethyl sulfate in situ gave the thiomethyl deriv-
ative 8 in 61% yield.
Our initial synthetic approach to compounds such as 2 in-
volved attempts at introducing a silyl protecting group
onto the C-5 hydroxy group of arabinose. However, we At this stage, we felt that introducing fluorine at C-5
found it difficult to achieve consistent outcomes using this would be beneficial for our studies, since fluorine is de-
approach, a finding comparable with the observations of tectable by NMR and would therefore provide us with an
others.10 Ultimately, we settled on a synthetic sequence to excellent ‘spectroscopic handle’ for studies involving in-
2 involving the use of a trityl ether at C-5 and acetate teractions between our KDO derivatives and KDO-recog-
groups on the other hydroxyls. Accordingly, treatment of nising enzymes. Additionally, fluorine is known to have
D-arabinose with trityl chloride in pyridine gave the similar electronic properties to hydroxy groups but is un-
known12 5-O-trityl ether 3 which, without purification, able to participate in hydrogen bonding interactions;14 it is
was acetylated (pyridine and acetic anhydride) to give the therefore an excellent functionality for use in studies
fully protected compound 412 (Scheme 2). Removal of the aimed at exploring ligand–protein interactions. Unfortu-
O-trityl group of 4 was achieved smoothly and efficiently nately, we found the introduction of fluorine to be quite
by treatment with 80% acetic acid at 65 °C for one hour.12 problematic. Attempted fluoride displacement of the me-
In this way the key 5-hydroxy arabinose derivative 2 was sylate group in 5 using tetrabutylammonium fluoride
obtained in 49% yield from arabinose. Importantly, we (TBAF),15 in our hands gave a complex mixture of prod-
found that when working on a large scale (>5 g of arabi- ucts from which the desired 5-fluoro-arabinose derivative
nose), the tritylation, acetylation, de-O-tritylation se- 9 was obtained in poor yield. A common reagent for con-
quence could be carried out without any purification of verting alcohols directly into fluorides is diethylamino
intermediates.
sulfur trifluoride (DAST). Unfortunately however, treat-
ment of 2 with DAST in dichloromethane,16 gave poor
(~15%) yields of the desired product. It has been reported
that changing the solvent to diglyme can improve the out-
come of this reaction10 but, in our hands, this was not the
case. We reasoned that the attempted introduction of
fluoride failed because the acetate protecting groups were
too labile for this type of transformation, since some of the
components we isolated from these reactions contained
fewer acetate groups than expected.
O
HO
O
HO
O
AcO
HO
OH
TrO
ix
OH
R
OAc
i
ii
HO
TrO
HO
AcO
3
4 R = OTr
iii
O
BnO
OBn
2 R = OH
iv
O
BnO
10
O
MsO
OAc
X
OAc
AcO
AcO
v, or vi, or vii
Dissatisfied with this approach, we investigated an alter-
native strategy involving the use of the more stable benzyl
ether protecting groups. Accordingly, the 5-O-trityl ether
3 was treated with sodium hydride and benzyl bromide in
DMF. Unfortunately, but perhaps not unexpectedly, this
resulted in a highly complex reaction mixture from which
the desired tri-O-benzyl ether 10 was obtained in modest
(<40%) and highly variable yield. Interestingly, one of the
products from this benzylation reaction appeared, by in-
spection of the 1H NMR spectrum, to contain four benzyl
ether groups in addition to the 5-O-trityl ether moiety,
which was possibly due to sodium hydride mediated re-
duction of the acyclic aldehyde form of the sugar.17 After
considerable experimentation, we opted for a longer se-
quence of reactions that proved far more reliable and effi-
cient overall. Accordingly, acetylation of arabinose
(Ac2O/pyridine) followed by treatment with SnCl4 and
benzyl alcohol18 gave the a-glycoside 11 in 86% yield
(Scheme 3). Removal of the acetate groups, tritylation of
the C-5 hydroxy group, and then benzylation of the re-
maining two hydroxy groups gave the desired arabinoside
10 in 30% yield from 11. Removal of the trityl group from
10 proceeded as expected to give the target alcohol 12 in
72% yield (Scheme 3).
AcO
AcO
5
6 X = N3
7 X = SAc
8 X = SMe
9 X = F
viii
Scheme 2 Reagents and conditions: (i) Ph3CCl, pyridine, r.t., 48 h,
58%; (ii) Ac2O, pyridine, r.t., 16 h, 96%; (iii) 80% aq AcOH, 65 °C,
1 h, 88%; (iv) MsCl, CH2Cl2, DIPEA, 0 °C to r.t., 1 h, 97%; (v) NaN3,
DMF, 60 °C, 16 h, 73%; (vi) KSAc, acetone, 40 °C, 72 h, 82%; (vii)
TBAF or DAST; (viii) H2NNH2·HOAc, DMF, (CH3O)2SO2, r.t., 4 h,
61%; (ix) NaH, BnBr, TBAI, DMF, 0 °C to r.t.
With the key alcohol 2 in hand, we turned our attention to
introducing alternative functionality at C-5, and opted for
a strategy involving displacement of a good leaving
group. Thus, exposure of 2 to methanesulfonyl chloride
gave the mesylate 5 (97%) which, upon treatment with so-
dium azide in DMF, afforded the known10,11 5-azido de-
rivative 6 (Scheme 2). We felt that an azide would provide
maximum flexibility for the preparation of a range of
KDO derivatives with nitrogen-based functionality, since
the azide group could be reduced after the aldol condensa-
tion, allowing the introduction of both amine and amide
functionality at C-8 in KDO.
For the introduction of sulfur functionality at C-5, the me-
sylate 5 was treated with potassium thioacetate in acetone
to give the 5-thioacetyl derivative 712 in 82% yield. Selec-
To see if changing the protecting groups would indeed re-
sult in an improved outcome, the alcohol 12 was exposed
to DAST. To our delight, the desired 5-fluoro derivative
Synlett 2010, No. 4, 583–586 © Thieme Stuttgart · New York

LETTER
Synthesis of C-8 Modified KDO Derivatives
585
O
O
AcO
O
AcO
O
BnO
HO
O
HO
OH
O
H
AcO
OAc
AcO
TrO
i
ii, iii, iv
R
OBn
OBn
4
HO
H
HO
CO2H
Re face
OH
AcO
AcO
BnO
HO HO
D-arabinose
HO HO
CO2H
H
11
10
KDO stereochemistry at C-4
Felkin–Ahn model
v
Scheme 4 Diagram showing the preferential Re-face attack on D-
arabinose, giving the desired KDO stereochemistry at C-4
O
BnO
O
BnO
HO
X
vi or vii
OBn
OBn
arabinose and oxalacetic acid appears to favour the forma-
tion of KDO over 4-epi-KDO, with a ratio of KDO to 4-
epi-KDO typically around 4:1. Purification of our KDO
product using Dowex 1 × 8 (200–400 mesh) anion-ex-
change chromatography following the method reported
by Kragl et al.,20 resulted in separation of KDO from 4-
epi-KDO, allowing complete assignment of their respec-
tive 1H NMR spectra (see the Supporting Information).
BnO
BnO
13 X = F
14 X = OMe
12
Scheme 3 Reagents and conditions: (i) SnCl4, BnOH, MeCN, 1 h,
86%; (ii) aq NaOH (1 M), MeOH, r.t., 16 h, 72%; (iii) Ph3CCl, pyri-
dine, r.t., 48 h, 52%; (iv) NaH, BnBr, TBAI, DMF, 0 °C to r.t., 16 h,
81%; (v) 80% aq AcOH, 65 °C, 1 h, 72%; (vi) DAST, CH2Cl2, 0 °C
to r.t., 3 h, 58%; (vii) NaH, MeI, DMF, 16 h, 81%.
13 was obtained in a modest 58% yield, which was a sig-
nificant improvement over our results with substrates
bearing acetate protecting groups. Attempts to increase
this yield by altering the reaction temperature and the
number of equivalents of DAST failed to significantly im-
prove the outcome. Having prepared the alcohol 12, we
felt we should exploit the robust nature of the benzyl
ethers by introducing a methyl ether at C-5. Thus, treat-
ment of 12 with sodium hydride and methyl iodide
smoothly gave the methyl ether 14 in 81% yield
(Scheme 3).
Although satisfied that we could prepare and isolate pure
KDO, the poor yield was a significant limitation of this
approach, especially given that many of our arabinose de-
rivatives were not trivial to prepare. In an attempt to opti-
mise this reaction, we repeated the condensation between
arabinose and oxalacetic acid using the NiCl2 modifica-
tion reported by Shirai and Ogura,9b and obtained a 60%
yield of KDO as a ~5:1 epimeric mixture at C-4. Whilst
delighted with this improved chemical yield, a limitation
with this method (as indeed with all previous reports on
the preparation of KDO using this approach) is the use of
a molar excess of arabinose. Since, in our strategy, the ar-
abinose derivative is the more difficult coupling partner to
prepare, the aldol condensation was repeated using a mo-
lar excess of oxalacetic acid. In this way, KDO could be
routinely obtained in ~65% yield based on arabinose.
Having established appropriate reaction conditions21 for
the synthesis of KDO, we turned our attention to using our
C-5 modified arabinose derivatives in the aldol condensa-
tion with oxalacetic acid. The results are summarised in
Table 1, and clearly show that this approach represents an
efficient and flexible synthesis of C-8 modified KDO de-
rivatives 15. The KDO derivatives 15 were all obtained
together with their 4-epi analogues, with the ratio of 15 to
Each of the C-5 modified arabinose derivatives was
deprotected in the standard way. Acetate protecting
groups were removed by treatment with dilute sodium hy-
droxide (1 M) in methanol, whilst benzyl ethers were re-
moved by hydrogenolysis. Interestingly, deprotection of
the 5-thioacetyl derivative 7 gave the known 5-thio-D-
arabinose,12 as a mixture of the pyranose and furanose
forms in a 3:1 ratio, respectively. Attempted deprotection
of the ester groups of the 5-thiomethyl derivative 8 result-
ed in extensive decomposition, even when dilute aqueous
lithium hydroxide and shorter reaction times were em-
ployed.
Having successfully prepared a range of C-5 modified ar-
abinose derivatives, we next explored their condensation
with oxalacetic acid to give KDO derivatives. Initially we
followed the method reported by McNicholas et al.9a us-
ing arabinose itself as a model compound, and obtained
the ammonium salt of KDO in ~30% yield after ion-
1
4-epi-15 typically better than 5:1 (as determined by H
NMR). Reduction of the azide group in 15 (X = N3) to an
amine (15, X = NH2) was accomplished by hydrogena-
tion.
Table 1 Synthesis of C-8 Modified KDO Derivatives
–
exchange chromatography on Amberlite CG-400 (HCO3 )
X
resin. As expected, the product obtained in this way was a
complex mixture of components due to both the lack of
stereoselectivity of the aldol condensation, giving a mix-
ture of epimers at C-4, as well as the fact that reducing
sugars exist as a mixture of a- and b-anomers in both pyr-
anose and furanose forms. Since this aldol condensation is
conducted in the absence of chelation control, the stereo-
chemical outcome of the enolate addition to the aldehyde
should fit the Felkin–Ahn model,19 where the incoming
nucleophile preferentially attacks from the sterically less-
hindered (Re) face of the aldehyde (Scheme 4).
O
HO
HO
HO
X
OH
+
O
1. pH 11
CO2H
O
2. pH ~5 (– CO2)
HO2C
CO2H
HO
HO
OH
15
X
Yield of 15 (%)
OH
65
84
87
67
OMe
N3
In this way we, as well as others,1a,9a,b have observed that
the stereochemical outcome for the condensation between
SH
Synlett 2010, No. 4, 583–586 © Thieme Stuttgart · New York

586
R. Winzar et al.
LETTER
(9) (a) McNicholas, P. A.; Batley, M.; Redmond, J. W.
In conclusion, we have shown that C-8 modified KDO de-
rivatives can be prepared efficiently from readily avail-
able starting materials. We are currently investigating
simple ways to improve the stereochemical outcome of
the aldol condensation between structurally modified ara-
binose derivatives and oxalacetic acid. We are also using
the C-8 modified KDO derivatives described herein as
probes for KDO-utilizing proteins. These investigations
will be described in due course.
Carbohydr. Res. 1986, 146, 219. (b) Shirai, R.; Ogura, H.
Tetrahedron Lett. 1989, 30, 2263. (c) Li, L.-S.; Wu, Y.-L.
Tetrahedron 2002, 58, 9049. (d) Gao, J.; Härter, R.;
Gordon, D. M.; Whitesides, G. M. J. Org. Chem. 1994, 59,
3714.
(10) Smellie, I. A.; Bhakta, S.; Sim, E.; Fairbanks, A. J. Org.
Biomol. Chem. 2007, 5, 2257.
(11) Legler, G.; Stutz, A. E.; Immich, H. Carbohydr. Res. 1995,
272, 17.
(12) Izumi, M.; Tsuruta, O.; Hashimoto, H. Carbohydr. Res.
1996, 280, 287.
(13) Park, W. K. C.; Meunier, S. J.; Zanini, D.; Roy, R.
Supporting Information for this article is available online at
Carbohydr. Lett. 1995, 1, 179.
http://www.thieme-connect.com/ejournals/toc/synlett.
(14) Berkowitz, D. B.; Karukurichi, K. R.; De La Salud-Bea, R.;
Nelson, D. L.; McCune, C. D. J. Fluorine Chem. 2008, 129,
731.
(15) Marcus, D. M.; Westwood, J. H. Carbohydr. Res. 1971, 17,
269.
(16) Lloyd, A. E.; Coe, P. L.; Walker, R. T.; Howarth, O. W.
J. Fluorine Chem. 1993, 60, 239.
(17) (a) See for example: Meng, Q.; Gong, B.; Hui, C.; Gao, Z.
Synth. Commun. 2009, 39, 1708. (b) For an interesting
discussion on the potential problems of using NaH in DMF,
see: Hesek, D.; Lee, M.; Noll, B. C.; Fisher, J. F.;
Mobashery, S. J. Org. Chem. 2009, 74, 2567.
(18) Pathak, A. K.; Pathak, V.; Maddry, J. A.; Suling, W. J.;
Gurcha, S. S.; Besra, G. S.; Reynolds, R. C. Bioorg. Med.
Chem. 2001, 9, 3145.
Acknowledgment
The authors would like to thank Dr Tareq Abu Izneid and Ms Lucile
Audoire for some preliminary investigations, and Prof Antony
Fairbanks (University of Canterbury, NZ) for helpful discussions.
Griffith University (GURG) and the Institute for Glycomics (GIGS)
are thanked for financial support.
References and Notes
(1) (a) Unger, F. M. Adv. Carbohydr. Chem. Biochem. 1981, 38,
323. (b) Raetz, C. R.; Whitfield, C. Annu. Rev. Biochem.
2002, 71, 635.
(2) Holst, O.; Molinaro, A. In Microbial Glycobiology,
Structures, Relevance & Applications; Moran, A. P., Ed.;
Academic Press: London, 2009, 29.
(3) Cipolla, L.; Polissi, A.; Airoldi, C.; Galliani, P.; Sperandeo,
P.; Nicotra, F. Curr. Drug Discovery Technol. 2009, 6, 19.
(4) (a) Gronow, S.; Brade, H. J. Endotoxin Res. 2001, 7, 3.
(b) Reynolds, M. C.; Raetz, C. R. H. Biochem. 2009, 48,
9627. (c) Klein, G.; Lindner, B.; Brabetz, W.; Brade, H.;
Raina, S. J. Biol. Chem. 2009, 284, 15369. (d) Raetz, C. R.
H.; Purcell, S.; Meyer, M. V.; Qureshi, N.; Takayama, K.
J. Biol. Chem. 1985, 260, 16080. (e) Mamat, U.; Meredith,
T. C.; Aggarwal, P.; Kuehl, A.; Kirchhoff, P.; Lindner, B.;
Hanuszkiewicz, A.; Sun, J.; Holst, O.; Woodard, R. W. Mol.
Microbiol. 2008, 67, 633.
(19) Evans, D. A.; Siska, S. J.; Cee, V. J. Angew. Chem. Int. Ed.
2003, 42, 1761.
(20) Kragl, U.; Godde, A.; Wandrey, C.; Lubin, N.; Auge, C.
J. Chem. Soc., Perkin Trans. 1 1994, 119.
(21) General procedure for the synthesis of KDO and C-8
modified derivatives: D-Arabinose (500 mg, 3.3 mmol) was
added to a solution of Na2CO3 (860 mg, 8.1 mmol) in H2O
(8 mL). Oxalacetic acid (525 mg, 4.0 mmol) was added
portionwise over 5 min, and the solution was adjusted to pH
11 using NaOH (10 M). After stirring for 2 h at r.t. the
solution was acidified to pH 5 using AcOH, NiCl2 (7.5 mg,
0.03 mmol) added, and the mixture was heated at 50 °C for
1 h. After cooling to r.t., the reaction was neutralised to pH
8 with ammonia and the KDO product was isolated by
–
column chromatography using CG-400 (HCO3 ) resin,
(5) Ghalambor, M. A.; Heath, E. C. Biochem. Biophys. Res.
Commun. 1963, 11, 288.
(6) Cornforth, J. W.; Firth, M. E.; Gottschalk, A. Biochem. J.
washing first with H2O and then eluting with ammonium
hydrogen carbonate (0.5 M). The eluant was concentrated
under reduced pressure, and then freeze-dried. The
lyophilised residue was purified using reversed-phase (C18)
silica gel with H2O as the mobile phase. Fractions containing
KDO can be visualised as bluish-grey spots on silica gel
TLC plates (CHCl3–MeOH–H2O, 5:5:1) by staining with
anisaldehyde–sulfuric acid dip.
1958, 68, 57.
(7) For an excellent overview on the early syntheses of KDO
and some derivatives, see: Li, L.-S.; Wu, Y.-L. Curr. Org.
Chem. 2003, 7, 447.
(8) (a) Hekking, K. F. W.; Moelands, M. A. H.; van Delft, F. L.;
Rutjes, F. P. J. T. J. Org. Chem. 2006, 71, 6444.
(b) Wardrop, D. J.; Zhang, W. Tetrahedron Lett. 2002, 43,
5389. (c) Ichiyanagi, T.; Sakamoto, N.; Ochi, K.; Yamasaki,
R. J. Carbohydr. Chem. 2009, 28, 53. (d) Kuboki, A.;
Tajimi, T.; Tokuda, Y.; Kato, D. I.; Sugai, T.; Ohira, S.
Tetrahedron Lett. 2004, 45, 4545. (e) Kim, B. G.; Schilde,
U.; Linker, T. Synthesis 2005, 1507.
Synlett 2010, No. 4, 583–586 © Thieme Stuttgart · New York