Synthesis of lipid A derivatives and their interactions with polymyxin B and polymyxin B nonapeptide

Article abstract of DOI:10.1021/ja0284456

Lipid A is the causative agent of Gram-negative sepsis, a leading cause of mortality among hospitalized patients. Compounds that bind lipid A can limit its detrimental effects. Polymyxin B, a cationic peptide antibiotic, is one of the simplest molecules c

Full text of DOI:10.1021/ja0284456

Published on Web 02/08/2003
Synthesis of Lipid A Derivatives and Their Interactions with
Polymyxin B and Polymyxin B Nonapeptide
Ning Yin, Ryan L. Marshall, Sannali Matheson, and Paul B. Savage*
Contribution from the Department of Chemistry and Biochemistry, Brigham Young UniVersity,
ProVo, Utah 84602
Received September 6, 2002 ; E-mail: paul_savage@byu.edu
Abstract: Lipid A is the causative agent of Gram-negative sepsis, a leading cause of mortality among
hospitalized patients. Compounds that bind lipid A can limit its detrimental effects. Polymyxin B, a cationic
peptide antibiotic, is one of the simplest molecules capable of selectively binding lipid A and may serve as
a model for further development of lipid A binding agents. However, association of polymyxin B with lipid
A is not fully understood, primarily due to the low solubility of lipid A in water and inhomogeneity of lipid A
preparations. To better understand lipid A-polymyxin B interaction, pure lipid A derivatives were prepared
with incrementally varied lipid chain lengths. These compounds proved to be more soluble in water than
lipid A, with higher aggregation concentrations. Isothermal titration calorimetric studies of these lipid A
derivatives with polymyxin B and polymyxin B nonapeptide indicate that binding stoichiometries (peptide
to lipid A derivative) are less than 1 and that affinities of these binding partners correlate with the aggregation
states of the lipid A derivatives. These studies also suggest that cooperative ionic interactions dominate
association of polymyxin B and polymyxin B nonapeptide with lipid A.
Introduction
The outer membranes of Gram-negative bacteria are made
up primarily of the lipid A portion of lipopolysaccharide (LPS);1
lipid A linked to a series of oligosaccharides constitutes LPS
(also known as Gram-negative bacterial endotoxin). LPS
released as a consequence of bacterial growth or death stimulates
monocytes and macrophages to produce and release a series of
proinflammatory cytokines that in turn can trigger a response
generally known as sepsis.2 The lipid A portion of LPS is the
causative agent of this response.3 In extreme cases cytokine
release can lead to organ failure and death, and despite
considerable efforts to limit the effects of lipid A, sepsis
continues to be a leading cause of death among hospitalized
patients.4
The basic structure of lipid A is conserved among most strains
of Gram-negative bacteria.1 The molecule consists of a â-D-
glucosaminyl-(1-6)-D-glucosamine disaccharide phosphorylated
at the 1 and 4′ positions. Lipid A is acylated by up to seven
ester- or amide-linked fatty acid residues, and the number and
length of these chains can vary depending upon the bacterial
source. For example, lipid A from Escherichia coli carries up
to seven fatty acids that are 14 carbons in length (Figure 1).2
Figure 1. Structure of lipid A from E. coli.
A number of compounds have been identified that associate
with lipid A and prevent its stimulation of cytokine production.
These compounds include anti-LPS antibodies,5 endogenous
LPS-binding proteins (e.g., bactericidal/permeability increasing
protein6 and lipopolysaccharide binding protein7), novel cationic
lipids,8 and the polymyxin family of antibiotics.9 One member
of the family of polymyxins, polymyxin B (PMB) (Figure 2),
(1) Nikaido, H.; Vaara, M. Microbiol. ReV. 1985, 49, 1.
(2) Reitschel, E. T.; Brade, H.; Holst, O.; Brade, L.; Mu¨ller-Loennies, S.;
Mamat, U.; Za¨hringer, U.; Beckmann, F.; Seydel, U.; Brandenburg, K.;
Ulmer, A. J.; Mattern, T.; Heine, H.; Schletter, J.; Loppnow, H.; Scho¨nbeck,
U.; Flad, H.-D.; Hauschildt, S.; Schade, U. F.; Di Padova, F.; Kusumoto,
S.; Schumann, R. R. In Pathology of Septic Shock; Reitschel, E. T., Wagner,
H., Eds; Springer: New York, 1996; p 39.
(5) For reviews see (a) Pollack, M.; Ohl, C. A. Curr. Top. Microbiol. Immunol.
1996, 216, 275. (b) Poxton, I. R. J. Immunol. Methods 1995, 186, 1.
(6) Taylor, A. H.; Heavner, G.; Nedelman, M.; Sherris, D.; Brant, E.; Knight,
D.; Ghrayeb, J. J. Biol. Chem. 1995, 270, 17934.
(7) Little, R. G.; Kelner, D. N.; Lim, E.; Burke, D. J.; Conlon, P. J. J. Biol.
Chem. 1994, 269, 1865.
(3) Galanos, C.; Luderitz, O.; Rietshel, E. T.; Westphal, O.; Brade, H.; Brade,
L.; Freudenberg, M.; Schade, U.; Imoto, M.; Yoshimura, H.; Kusumoto,
S.; Shiba, T. Eur. J. Biochem. 1985, 148, 1.
(8) (a) David, S. A.; Pe´rez, L.; Infante, M. R. Bioorg. Med. Chem. Lett. 2002,
12, 357. (b) David, S. A.; Silverstein, R.; Amura, C. R.; Kielian, T.;
Morrison, D. C. Antimicrob. Agents Chemother. 1999, 43, 912.
(4) M. M. W. R. 1990, 39, 31.
9
2426
J. AM. CHEM. SOC. 2003, 125, 2426-2435
10.1021/ja0284456 CCC: $25.00 © 2003 American Chemical Society

Interactions of Lipid A Derivatives with Polymyxin B
A R T I C L E S
amine salts of lipid A by use of triethylamine or tris-
(hydroxymethyl)aminomethane (Tris).12,13 However, ionic in-
teractions between the phosphate groups on lipid A and the
amine groups on PMB are believed to contribute to association,
and competitive salt formation is likely to influence association.
A potential solution to the problems associated with measur-
ing interactions of lipid A with other molecules would be to
synthesize homogeneous, well-characterized lipid A derivatives
that are soluble in aqueous solution. Specific association of many
compounds with lipid A occurs via interactions with the
disaccharide headgroup;14a,19 consequently, modifications to the
disaccharide would be expected to interfere with binding by
other molecules. However, modification of the length and
number of the lipid chains within lipid A would provide a means
of generating lipid A derivatives that are more soluble than the
parent compound while preserving interactions of binding
partners with the phosphorylated disaccharide.
Figure 2. Structures of polymyxin B2 (PMB) and polymyxin B nonapeptide
(PMBN).
has been especially well studied in this context. For example,
in animal studies PMB can prevent the onset of sepsis caused
by a bacterial challenge,9 and PMB immobilized on a solid
support can be used to remove LPS from blood.10 However,
the use of the polymyxins, including PMB, is severely limited
because of their toxicity.11
Because PMB is among the smallest compounds known to
selectively associate with lipid A, the interactions of these two
molecules may provide the simplest model of how lipid A can
be bound by other molecules. Consequently, PMB-LPS and
PMB-lipid A interactions have been the subject of a number
of studies.12-15 These studies have relied exclusively on lipid
A and LPS derived directly from bacterial sources. However,
there are drawbacks in using lipid A or LPS from bacterial
sources. While the basic structure of lipid A is generally
conserved among strains of Gram-negative bacteria, significant
inhomogeneity typically exists in LPS preparations, even those
derived from a single species of bacteria.16 This inhomogeneity
may complicate studies of association of LPS with other
molecules. Possibly more significant are problems associated
with cleaving lipid A from the oligosaccharides that make up
the remainder of LPS. This cleavage is typically performed
under acidic conditions that can cause hydrolysis of the labile
anomeric phosphate, yielding monophosphoryl lipid A.17,15b
A major obstacle that must also be addressed in studies of
lipid A binding with other molecules is the limited solubility
of lipid A in aqueous solutions. LPS and lipid A are sparingly
soluble in water.18 In an effort to increase the solubility of lipid
A for binding studies, a number of researchers have prepared
Studies of the interactions of lipid A and LPS with PMB
have been performed by a variety of methods including
isothermal titration calorimetry (ITC),12 surface plasmon reso-
nance,13 and NMR spectroscopy.14 Results from calorimetric
studies are particularly informative because they provide as-
sociation constants and thermodynamic parameters of associa-
tion, which can indicate the relative roles of noncovalent forces
in association. However, experimental results from reported ITC
studies are surprisingly inconsistent regarding issues critical to
the understanding of PMB interactions with lipid A and LPS:
binding stoichiometry and the change in enthalpy associated
with complex formation. For example, Surolia and co-workers12a
reported 1:1 binding stoichiometries for interactions of PMB
with lipid A and LPS from a variety of bacterial sources. In
contrast, two other groups12b,c have reported binding stoichi-
ometries of much less than 1 (LPS to PMB). These two groups
reported primarily exothermic association of PMB with LPS,
while Surolia and co-workers reported that binding of PMB to
lipid A and LPS is an endothermic process and concluded that
hydrophobic interactions dominate these binding events. Sur-
prisingly, in the latter study12a diphosphoryl lipid A and
monophosphoryl lipid A (both derived from cleavage of
oligosaccharides from LPS) gave essentially identical binding
stoichiometries and enthalpies of association with PMB. Pro-
vided that ionic interactions play even a small role in association,
one would expect that loss of a phosphate-amine interaction
would influence binding.
(9) For examples see (a) Rifkind, D. J. Bacteriol. 1967, 93, 1463. (b) From,
A. L.; Fong, J. S. C.; Good, R. A. Infect. Immun. 1979, 23, 660.
(10) Tani, T.; Cahng, T. M. S.; Kodama, M.; Tsuchiya, M. Biomater., Artif.
Cells, Immobilization Biotechnol. 1992, 20, 457.
(11) For examples see (a) Yow, E. M.; Moyer, J. M.; Smith, C. P. Arch. Inter.
Med. 1953, 92, 248. (c) Garrod, L. P.; Lambert, H. P.; O’Grady, F.
Antibiotic and Chemotherapy; Churchill Livingstone: London, 1981.
(12) Studies using ITC: (a) Srimal, S.; Surolia, N.; Balasubramanian, S.; Surolia,
A. Biochem. J. 1996, 315, 679. (b) Brandenburg, K.; Moriyon, I.; Arraiza,
M. D.; Lewark-Yvetot, D.; Koch, M. H. J.; Seydel, U. Thermochem. Acta
2002, 382, 189. (c) Koch, P.-J.; Frank, J.; Schu¨ler, J.; Kahle, C.; Bradeczek,
H. J. Colloid Interface Sci. 1999, 213, 557-564.
(13) Study using surface plasmon resonance: (a) Thomas, C.; Surolia, N.;
Surolia, A. J. Biol. Chem. 1999, 274, 29624. (b) Thomas, C.; Surolia, A.
FEBS Lett. 1999, 445, 420.
(14) Studies using NMR spectroscopy and molecular modeling: (a) Bruch, M.
D.; Cajal, Y.; Koh, J. T.; Jain, M. K. J. Am. Chem. Soc. 1999, 121, 11993.
(b) Pristovsek, P.; Kidric, J. J. Med. Chem. 1999, 42, 4604. (c) Bhatta-
charjya, S.; David, S.; Mathan, V. I.; Balaram, P. Biopolymers, 1997, 41,
251.
(15) Examples of studies using UV and/or fluorescence spectroscopy: (a)
Thomas, C. J.; Gangadhar, B. P.; Surolia, N.; Surolia, A. J. Am. Chem.
Soc. 1998, 120, 12428. (b) David, S. A.; Balasubramanian, K. A.; Mathan,
V. I.; Balaram, P. Biochim. Biophys. Acta 1992, 1165, 147. (c) Moore, R.
A.; Bates, N. C.; Hancock, R. E. W. Antimicrob. Agents Chemother. 1986,
29, 496.
If hydrophobic interactions play a significant role in PMB-
LPS or PMB-lipid A association, the lipophilic acyl chain on
PMB would be expected to play a role in binding. However,
studies of the association of PMB and a truncated form of PMB,
polymyxin B nonapeptide (PMBN)20 (Figure 2), with either
bacterial surfaces or LPS suggest that PMB and PMBN have
comparable affinities for LPS.12b,21 Consequently, the role and
source of hydrophobic interactions in this system are not clear.
In this report we describe the synthesis of well-characterized
lipid A derivatives with modified lipid chains, studies of their
aggregation properties, and ITC studies of their interactions with
(19) PMB associates with intact bacterial membranes where lipid chains are
sequestered (see ref 21). See also Kirikae, T.; Schade, F. U.; Za¨hringer,
U.; Kirikae, F.; Brade, H.; Kusumoto, S.; Kusame, T.; Rietschel, E. T.
FEMS Immunol. Med. Microbiol. 1993, 8, 13.
(20) Vaara, M.; Vaara, T. Nature 1983, 303, 526.
(21) Vaara, M.; Viljanen, P. Antimicrob. Agents Chemother. 1985, 27, 548.
(16) Lehamann, V.; Rupprecht, E. Eur. J. Biochem. 1977, 81, 443. (b) Chang,
C.-M.; Nowotny, A. Immunochemistry 1975, 12, 19.
(17) O’Connor, J. V.; Barker, R. Carbohydr. Res. 1979, 254, 5906.
(18) Din, Z. Z.; Mukerjee, P.; Kastowsky, M.; Takayama, K. Biochemistry 1993,
32, 4579.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2427

A R T I C L E S
Yin et al.
Scheme 1. Preparation of Glycosyl Acceptor 10a
a Reagents [yields in brackets]: (a) Ac2O, pyridine, CH2Cl2, 70 °C; (b) AcOH, H2NCH2CH2NH2, THF [72% for two steps]; (c) TBSCl, imidazole,
acetonitrile; (d) NH3, MeOH; (e) 2,2-dimethoxypropane, PPTS, CH2Cl2 [63% for three steps]; (f) hexanoic anhydride, pyridine, CH2Cl2; (g) PPTS, MeOH
[65% for two steps]; (h) TBSCl, imidazole, acetonitrile [91%]; (i) benzylchloroformate, DMAP, CH2Cl2 [71%]; (j) HF, H2O, acetonitrile [59%].
Scheme 2. Preparation of Glycosyl Donor 13a
a Reagents [yields in brackets]: (a) benzylchloroformate, DMAP, CH2Cl2
[69%]; (b) (BnO)2PN(iPr)2, 5-phenyltetrazole; H2O; MMPP [70%]; (c) HF,
MeCN [74%].
Figure 3. Structures of lipid A derivatives 1-4.
we found it necessary to modify reported syntheses of lipid A
to satisfy our needs with an emphasis on the ability to prepare
large quantities of glycosyl donors and acceptors. A representa-
tive synthesis of 2 is shown in Schemes 1-3.
PMB and PMBN in solution buffered with either phosphate or
Tris. These studies reveal the influences on association of lipid
Formation of the glycosyl acceptor (10) was begun with
amide 524 (Scheme 1). Peracylation and removal of the anomeric
acetate25 gave 6. The anomeric alcohol was protected as a silyl
ether, the remaining acetate groups were removed, and an
acetonide was formed, protecting the hydroxyl groups at C4
and C6. Acylation at C3 and hydrolysis of the acetonide gave
8. Selective protection of the primary alcohol at C6 as a silyl
ether was followed by incorporation of a benzyloxycarbonyl
protecting group at C4. Selective deprotection of the TBS ether
at C6 gave glycosyl acceptor 10.23a
chains, both on the lipid A derivatives and on PMB, and the
roles of buffers on the thermodynamic profiles of the interac-
tions. In addition, studies performed with a monosaccharide
model of lipid A demonstrate the importance of the disaccharide
structure of lipid A in its interactions with PMB.
Results
Modifications of lipid A by decreasing lipid chain length were
intended to leave the disaccharide headgroup intact while
increasing the solubility of the derivatives. As a means of
determining the effects of these changes on association with
PMB and PMBN, lipid chain lengths were incrementally
shortened, yielding compounds 1-3 (Figure 3). In addition, a
control compound, 4, was prepared that consisted of one of the
modified glucosamine groups of lipid A (Figure 3). The total
synthesis of the putative structure of lipid A from E. coli was
reported in 1985 by Shiba and co-workers,22 and since that report
syntheses of lipid A and variants have been described by a
number of groups.23 For our studies, relatively large quantities
(hundreds of milligrams) of pure lipid A derivatives 1-3 were
required, and considering the length of the synthesis (>20 steps)
The synthesis of glycosyl donor 13 is shown in Scheme 2.
The pathway used to generate 8 from 5 was followed to prepare
11 from Troc-glucosamine. Selective protection of the primary
alcohol at C6 as the benzyloxycarbonate, followed by incor-
poration of a protected phosphate at C4 via phosphoramidate
chemistry, gave 12. Deprotection of the anomeric silyl ether
provided 13. Treatment of 8 (Scheme 1) with the reagents and
under the conditions given in steps a-c in Scheme 2, followed
by removal of the benzyl and benzyloxy protecting groups via
treatment with palladium on carbon and hydrogen, gave 4.
Coupling of donor 13 and acceptor 10 gave disaccharide 14
(Scheme 3). The Troc group was replaced22 by the appropriate
amide to yield 15. Hydrolysis of the remaining TBS ether was
followed by incorporation of the protected phosphate at the
anomeric position, which gave only the R-anomer of 16.
(22) Imoto, M.; Yoshimura, H.; Sakaguchi, N.; Kusumoto, S.; Shiba, T.
Tetrahedron Lett. 1985, 26, 1545.
(23) For examples, see (a) Christ, W. J.; McGuinness, P. D.; Asano, O.; Wang,
Y.; Mullarkey, M. A.; Perez, M.; Hawkins, L. D.; Blythe, T. A.; Dubuc,
G. R.; Robidoux, A. L. J. Am. Chem. Soc. 1994, 116, 3637. (b) Ikeda, K.;
Takahashi, T.; Kondo, H.; Achiwa, K. Chem. Pharm. Bull. 1987, 35, 1311.
(c) Kusumoto, S. In Bacterial Endotoxic Lipopolysaccharides; Morrison,
D. C., Ryan, J. L., Eds.; CRC Press: Boca Raton, FL, 1992; p 81.
(24) Vafona, M. G.; Molodstov, N. V. Carbohydr. Res. 1976, 47, 188.
(25) Zhang, J.; Kovac, P. J. Carbohydr. Chem. 1999, 18, 4461.
9
2428 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Interactions of Lipid A Derivatives with Polymyxin B
A R T I C L E S
Scheme 3. Coupling of Donor 13 and 10 Leading to the Preparation of 2a
a Reagents [yields in brackets]: (a) Cl3CCN, K2CO3; (b) 10, TMSOTf, 4 Å sieves, ClCH2CH2Cl [53% for two steps]; (c) Zn/Cd, AcOH, sonication; (d)
hexanoic anhydride, pyridine [50% yield for two steps]; (e) HF, MeCN [68%]; (f) (BnO)2PN(iPr)2, 5-phenyltetrazole; H2O; MMPP [76%]; (g) H2, Pd/C
(10%), THF [81%].
Figure 4. Prodan fluorescence as a function of lipid concentration: (]) 2
in phosphate buffer (50 mM); (4) 2 in Tris buffer (50 mM); ([) 3 in
phosphate buffer (50 mM); (2) 3 in Tris buffer (50 mM).
Figure 5. Titration of 3 (1 mM) with PMB (6 mM) in Tris buffer (50
mM, pH 7.2) at 25 °C. Complex formation is exothermic as characterized
by a negative ∆H value and by comparison to complex formation of Ba2+
with 18-crown-6.
Deprotection of the benzyl and benzyl carbonate groups gave
lipid A derivative 2.
Aqueous solubilities and critical aggregation concentrations
the binding partners were used to establish a set of standard
(CACs) of 1-3 were determined in phosphate and Tris buffers
conditions for the titrations; in general, PMB and PMBN at 15
to determine conditions under which these compounds could
mM were titrated into lipid A derivative solutions at 2 mM at
be studied. The concentrations we expected to use for ITC
25 °C with buffer concentrations at 50 mM. The only exception
experiments were in the millimolar range, and each of the
was the titration of 3 with PMB and PMBN. Because of greater
derivatives proved to be soluble in this range in both buffers
heats of association with these binding partners, concentrations
(50 mM). CAC values were determined by use of the fluorescent
of 6 mM (peptides) and 1 mM (3) were used. Attempts to
probe Prodan. This probe changes its fluorescence in response
observe PMB interactions with 3 near the CAC of 3 (20 µM)
to its environment, and its fluorescence is modulated upon
yielded heats of association that were too small to be useful in
moving from solution into hydrophobic aggregates.26 The
characterizing binding. The standard conditions used gave
reproducible results while only slowly exhausting our supply
of binding partners. Each experiment was performed a minimum
fluorescence responses of Prodan to varied concentrations of 2
and 3 are shown in Figure 4. As would be expected, aggregation
was not observed with 1, and 2 did not appear to display critical
of three times, and similar results were obtained for each
aggregation behavior although some indication of aggregation
replicate. An ITC isotherm from a typical titration (PMB with
is suggested near 1 mM. Also, a change from one buffer to the
other did not appear to influence aggregation of 2. In contrast,
3 in Tris buffer) is given in Figure 5.
Compound 4 was prepared to establish the requirement of
the disaccharide structure of lipid A for interaction with PMB.
As compared to heats of dilution of 4 and PMB alone, titration
of a solution of 4 at 2 mM with PMB in either phosphate or
Tris buffer produced no measurable heat of association. This
result contrasts the heats of association of PMB with 1-3
measured under the same conditions (vide infra) and is evidence
that the disaccharide portion of lipid A is required for binding.
3 displayed behavior consistent with critical aggregation with
a CAC near 20 µM, and the buffer appeared to influence
aggregation; phosphate buffer apparently facilitated aggregation.
Titration experiments of PMB and PMBN with lipid A
derivatives 1-3 were conducted under a variety of conditions.
Considerations of heats of association and the availability of
(26) Weber, G.; Farris, F. J. Biochemistry 1979, 18, 3075.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2429

A R T I C L E S
Yin et al.
Table 1. Results from ITC Titrations of Lipid A Derivatives with
PMB and PMBN
isotherm that initially described an exothermic process, which
changed to an endothermic interaction (Figure 6). Titrations of
PMB and PMBN with 3 in Tris buffer gave simpler isotherms
(Figure 5) that could be interpreted to yield association constants
and binding stoichiometries (Table 1). PMB displayed a higher
affinity for 3 over 2, and the interaction was highly exothermic.27
Similarly, the association of PMBN with 3 was also exothermic,
while its affinity for 3 was similar to that for 2. As with 2,
binding of the peptides with 3 gave a stoichiometry with PMBN
that was greater than that with PMB.
lipid A
peptide derivative stoichiometry
K (M-1
)
∆H (kJ/mol) buffer
a
a
PMB
PMB
PMB
PMBN
PMB
PMBN
PMB
1
1
2
2
3
3
4
NRa
NR
NR
b
c
0.62 ( 0.09 (3.34 ( 0.11) × 102 -35 ( 7
0.60 ( 0.008 (1.20 ( 0.15) × 104 10.7 ( 0.20 b
0.82 ( 0.03 (2.85 ( 0.39) × 104 -8.7 ( 0.11 b
0.54 ( 0.007 (7.16 ( 1.61) × 104 -54 ( 1.2
0.62 ( 0.02 (2.71 ( 0.97) × 104 -57.1 ( 3.3
c
c
b, c
NR
NR
NR
a NR, no significant heat response observed compared to the dilution.
b Potassium phosphate buffer. c Tris buffer.
Discussion
Aggregation properties of 1-3 were similar to those antici-
pated: disaccharide 1, without a significant lipid component,
did not aggregate, and as lipid chains increased in length,
propensities to aggregate increased. The three lipid A derivatives
provide an opportunity to observe binding in the absence of
aggregation, with some degree of aggregation, and with the
compounds highly aggregated, respectively. As discussed, the
solubility of lipid A can be increased by forming its ammonium
salts; Aurell and Winstrom28 determined the CAC values in Tris
buffer of lipid A to be 5 µM, and LPS from a variety of sources
had CAC values of 11-38 µg/ mL. Similarly, the propensity
of 3 to aggregate is decreased in Tris buffer (Figure 4).
Observation of association of PMB with 1 provides a view
of association with a minimal lipid A headgroup structure and
without aggregation of the lipid A binding partner. Measured
association is exothermic, which is consistent with a dominant
role for ionic interactions. Considering the differences in
responses of PMB to 1 in phosphate and Tris buffers, it is clear
that these buffers can influence binding. Ionic association
between PMB and lipid A involves interaction of amine groups
with phosphates. For these ionic interactions to contribute to
association, they must outcompete the ionic buffers, which are
present in much higher concentrations. Cooperative effects29 of
multiple amine groups on PMB associating with diphosphoryl
lipid A, as proposed by Bruch et al.,14a may allow effective
competition with buffer. This requirement for the diphosphoryl
structure of lipid A is corroborated by the fact that comparable
binding was not observed with monophosphate 4.
The affinities of PMB for 1-3 roughly paralleled their
aggregation states under the conditions used in the study. That
is, PMB displayed the highest affinity for 3, which displays
the greatest propensity to aggregate. Aggregation may play a
key role in complex formation. The binding stoichiometries of
PMB and PMBN with the lipid A derivatives were less than 1,
and aggregation would facilitate simultaneous interaction of
PMB and PMBN with more than one disaccharide. The CAC
values of lipid A and LPS are much lower (<50 nM)18 than
concentrations used in other studies of PMB-LPS or PMB-
lipid A interactions; consequently, the aggregated states of the
binding partners in these studies likely influenced observed
association.
Figure 6. Comparison of heats of complex formation vs molar ratios of
lipid A derivatives with PMB in different aqueous buffers. Titrations of
PMB (15 mM) into indicated lipid A derivative (2 mM) in indicated buffer
(50 mM, pH 7.2) at 25 °C: (2) 2 in phosphate buffer; (4) 2 in Tris buffer;
(9) 3 in phosphate buffer.
Titration of PMB into a solution of 1, the minimal structure
containing the disaccharide headgroup of lipid A, gave little or
no heat of association in phosphate buffer. In Tris buffer only
relatively weak association was observed with PMB and 1
(Table 1). Association was exothermic, and the best fit of the
data gave a stoichiometry less than 1.
In phosphate buffer the association of 2 with PMB was
endothermic, in contrast to the association of 1 with PMB (Table
1). In addition, the affinity of PMB for 2 was nearly a 100-fold
greater than that for 1. PMBN, lacking the hydrophobic chain,
associated with 2 in an exothermic manner. Surprisingly, PMBN
demonstrated a slightly higher affinity for 2 than PMB. Also,
the binding stoichiometry of PMBN was higher than that with
PMB. Titration of 2 with PMB in Tris buffer gave a very
different result from that observed in phosphate buffer. Figure
6 provides the relationship between the amount of PMB added
and the microjoules of heat of association. In a typical titration
(e.g., PMB with 2 in phosphate), heat is absorbed or given off
as each aliquot of the titrated binding partner is added, and after
saturation heat is neither absorbed nor given off (neglecting heats
of dilution). However, the titration of 2 with PMB in Tris buffer
gave a binding isotherm that was much more complicated than
the experiment in phosphate buffer. The titration moved from
endothermic to exothermic behavior.
The measured stoichiometries of PMB interactions with 1-3
may reflect, in part, the relative number of ionic groups in the
binding partners. PMB contains five amine groups, and if
(27) The exothermic nature of the interaction was reaffirmed by comparison to
the ∆H of association of Ba2+ with 18-crown-6.
(28) Aurell, C. A.; Winstrom, A. O. Biochem. Biophys. Res. Commun. 1998,
253, 119.
Titration of 3 with PMB in phosphate also gave a complex
(29) Jenks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 4046.
9
2430 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Interactions of Lipid A Derivatives with Polymyxin B
A R T I C L E S
association occurs with individual amines binding each phos-
phate, one would expect a stoichiometry of 0.4. Therefore,
because stoichiometries are greater than 0.4, two amine groups
on PMB may in concert associate with one phosphate as
proposed by Bruch and co-workers.14a Our results, taken with
others,12b,c provide strong evidence that binding stoichiometry
is less than unity. Truncation of PMB to yield PMBN results
in loss of one amine group, and as would be expected
considering relative numbers of charged groups, stoichiometries
with PMBN are greater than those with PMB (Table 1).
Comparison of the binding of 2 and 3 by PMB and PMBN
provides an indication of the role(s) that the acyl chain of PMB
plays in lipid A binding. The affinities of PMBN and PMB for
the outer membranes of Gram-negative bacteria are compa-
rable,21 and others have noted the similarity in their interactions
with LPS.12b PMB and PMBN gave similar binding constants
with 2 and 3. The lack of a significant role of the acyl chain of
PMB in lipid A binding suggests that hydrophobic contributions
to binding are due to amino acid side chains. PMB and PMBN
contain only two hydrophobic residues, so contributions of
hydrophobic interactions to binding may be small as compared
to those from polar and ionic interactions. Therefore, it is not
surprising that association is exothermic. For its bactericidal
activity, PMB must traverse the outer membranes of Gram-
negative bacteria, a process termed self-promoted transport,30
and the acyl chain of PMB is required for this activity. Lacking
this acyl chain, PMBN is not bactericidal.20
The buffers used in the titrations greatly influenced the
observed results. The fact that no heat of association was
observed in the titration of 1 with PMB in phosphate buffer
does not necessarily signify that association did not occur;
however, the titration of 1 with PMB clearly indicated associa-
tion in Tris buffer. Differences in these two titrations may be
due to competition of the buffers for interactions with the
binding partners. If the interaction of 1 and PMB is indeed
weaker in phosphate buffer, it is likely due to a greater ability
of phosphate to compete for interaction with the binding partners
than Tris.
The influence of buffer on titrations of 2 and 3 with PMB
was also significant. Brandenburg et al.,12b in ITC titrations of
LPS with PMB, observed results very similar to those described
in Figure 6. These results were explained as a superimposition
of multiple phenomena: association of LPS with PMB and
changes in the aggregation state of LPS. Likewise, the results
given for titrations with 2 in phosphate buffer and 3 in Tris
buffer in Figure 6 appear to come from simultaneous changes
in the binding and aggregation states of the lipid A derivatives
and PMB. It is also possible that during other titrations changes
in aggregation state may occur that do not appreciably affect
interpretation of binding. Changing from one buffer to the other
allowed measurement of association without other apparent
changes in the system. Nevertheless, because of the demon-
strated effects of buffer on association of PMB with lipid A
derivatives, direct comparisons of titrations of 2 (in phosphate
buffer) and titrations of 3 (in Tris buffer) may not be informa-
tive.
be attributed to the difference in structures of PMB and PMBN
and would suggest a role for the acyl chain of PMB in
association with 2. However, as compared to PMB, PMBN
reproducibly displayed higher affinity for 2, indicating that
associative interactions of the acyl chain with 2 were offset by
weakening of other interactions. Due presumably to solubility
problems, titrations of lipid A and LPS by PMB or PMBN have
not been performed in the absence of amine buffers. Conse-
quently, it is not clear if this change in heats of association
(endothermic with PMB and exothermic with PMBN) occurs
in the interactions of these peptides with lipid A and LPS.
Titrations of 3 with PMB and PMBN in Tris buffer gave very
similar results: comparable association constants and heats of
association.
Conclusions
The interactions of lipid A and LPS with other molecules
are inherently difficult to study because of the inhomogeneity
of these glycolipids derived from bacterial sources and their
insolubility/aggregation propensity in aqueous solution. The
ability to synthesize a series of pure compounds with incre-
mentally shortened lipid chain lengths has provided an improved
means of studying interactions of lipid A derivatives with other
molecules. The two observations that (1) titrations of lipid A
derivatives with PMB and PMBN in an amine buffer were
exothermic and (2) PMB and PMBN displayed similar affinities
for 2 and 3 strongly suggest that ionic interactions play a
dominant role in association. However, both PMB and PMBN
contain hydrophobic amino acids, and these may play a role in
association with lipid A. In addition, these studies provide
conclusive evidence that the binding stoichiometry of PMB and
PMBN with the disaccharide portion of lipid A is less than unity
and that the stoichiometry is determined by the number of ionic
groups on the binding partners. This conclusion does not
preclude specific association of PMB with lipid A, as proposed
by Bruch et al.,14a but it does suggest that amine groups in a
PMB-lipid A complex are available for further interactions.
Because PMB and PMBN can bind more than one lipid A
molecule, the strength of these interactions appears to be
influenced by the aggregation state of lipid A. Monomeric lipid
A (represented by 1) was bound less strongly than partially
aggregated (2) and fully aggregated (3) forms. The propensities
of lipid A-PMB complexes to aggregate may also contribute
to this observation. These studies also reveal specific structural
characteristics of PMB and lipid A that are necessary for
association. The acyl chain of PMB does not play a central role
in binding, and as we have shown in a distinct system,31 this
lipid chain serves to facilitate traversal of the outer membranes
of Gram-negative bacteria. The diphosphate structure of lipid
A derivatives (as compared to 4) provides for cooperative
interactions of the appropriately spaced amine groups on PMB
that can compete with buffers for ionic interactions. It is
anticipated that information from these studies will clarify
unresolved issues regarding association of PMB and PMBN with
lipid A (e.g., exothermic vs endothermic association and
stoichiometries of binding). Furthermore, these results should
assist in the further development of lipid A binding agents,
Association of PMB with 2 in phosphate buffer proved to be
an endothermic process, while the same titration with PMBN
was exothermic. This difference in enthalpies of association must
(31) Li, C.; Budge, L. P.; Driscoll, C. D.; Willardson, B. M.; Allman, G. W.;
(30) Hancock, R. E. W. Annu. ReV. Microbiol. 1984, 38, 237.
Savage, P. B. J. Am. Chem. Soc. 1999, 121, 931.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2431

A R T I C L E S
Yin et al.
chromatography was performed (CH2Cl2/MeOH 20:1) to yield a clear
oil (5.13 g, 10.5 mmol, 65% for two steps). H NMR (CDCl3) δ 6.33
which in turn may be useful in preventing the detrimental
biological effects of lipid A.
1
(d, J ) 9.3 Hz, 1H), 5.12 (dd, J ) 10.7, 8.8 Hz, 1H), 4.79 (d, J ) 7.8
Hz, 1H), 3.91-3.82 (m, 3H), 3.74 (t, J ) 9.3 Hz, 1H), 3.47 (dt, J )
9.9, 4.4 Hz, 1H), 2.40-2.29 (m, 2H), 2.15-2.04 (m, 2H), 1.61-1.50
(m, 4H), 1.31-1.22 (m, 8H), 0.90-0.84 (m, 15H), 0.08 (s, 3H), 0.04
(s, 3H). 13C NMR (CDCl3) δ 175.09, 173.58, 96.485, 75.52, 75.22,
69.69, 62.60, 55.94, 37.00, 34.32, 31.64, 31.38, 25.51, 24.74, 22.52,
22.48, 18.00, 14.03, -3.89, -5.04. HRFAB-MS (thioglycerol + Na+
matrix) m/e ([M + Na]+) 512.3014 (100%); calcd 512.3019
Experimental Section
General Methods. 1H NMR, 13C NMR, and 31P NMR spectra were
recorded on Varian Unity 500 MHz or Varian Unity 300 MHz
instruments. Mass spectrometric data were obtained on a Jeol SX 102
A spectrometer. Tetrahydrofuran (THF) and CH2Cl2 were dried over
Na0 or CaH2. Chemicals were obtained from Fluka, Aldrich, Acros,
and Sigma and were used as received unless otherwise noted.
1,6-Di-(tert-butyldimethylsilyl) 4-O-(Benzyloxycarbonyl)-2-deoxy-
3-O-hexanoyl-2-(hexanoylamino)-â-D-glucopyranoside (9). Com-
pound 8 (5.13 g, 10.5 mmol) was dissolved in CH3CN (50 mL). Tert-
Butyldimethylsilyl chloride (1.90 g, 12.6 mmol) and imidazole (0.86
g, 12.6 mmol) were added to the solution and allowed to stir for 4 h.
The solvent was removed in vacuo and the residue was dissolved in
CH2Cl2 (100 mL). The mixture was washed with 5% HCl (100 mL)
and brine (100 mL), after which the organic phase was dried over
MgSO4, filtered, and concentrated. SiO2 chromatography (ethyl acetate/
3,4,6-Tri-O-acetyl-2-deoxy-2-(hexanoylamino)-D-glucopyranose
(6). To a round-bottom flask were added 5 (10.0 g, 36.0 mmol), acetic
anhydride (33.9 mL, 360 mmol), and pyridine (29.1 mL, 360 mmol)
in CH2Cl2 (300 mL). The mixture was refluxed for 6 h. The solution
was then washed with 5% HCl solution (2 × 200 mL) and brine (200
mL). The organic phase was dried (MgSO4) and concentrated. The
remaining clear oil was dissolved in THF (200 mL) and added to a
flask containing a mixture of ethylenediamine (2.8 mL, 41.4 mmol)
and acetic acid (2.5 mL, 43.2 mmol). The solution was stirred for 2 h,
and the THF was removed in vacuo. The residual oil was dissolved in
CH2Cl2 (300 mL), and the organic layer was washed with water (2 ×
200 mL) and brine (200 mL), dried (MgSO4), and concentrated. SiO2
chromatography (CH2Cl2/MeOH 20:1) yielded a clear oil (10.4 g, 25.8
1
hexanes, 1:3) afforded a clear oil (5.77 g, 9.56 mmol, 91%) H NMR
(CDCl3) δ 5.50 (d, J ) 9.3 Hz, 1H), 5.00 (dd, J ) 10.8, 9.3 Hz, 1H),
4.66 (d, J ) 7.8 Hz, 1H), 3.95 (ddd, J ) 11.2, 9.8, 8.3 Hz, 1H), 3.89,
3.84 (ABq of ABX, JAB ) 10.3 Hz, JAX ) 4.6 Hz, JBX ) 5.9 Hz, 2H),
3.74 (dt, J ) 9.3, 2.4 Hz, 1H), 3.42 (ddd, J ) 9.3, 5.9, 4.9 Hz, 1H),
3.27 (d, J ) 2.4 Hz, 1H), 2.40-2.29 (m, 2H), 2.14-2.04 (m, 2H),
1.63-1.54 (m, 4H), 1.33-1.24 (m, 8H), 0.90-0.85 (m, 24H), 0.10 (s,
3H), 0.09 (s, 6H), 0.07 (s, 3H). 13C NMR (CDCl3) δ 175.08, 172.87,
96.90, 75.37, 74.53, 71.98, 65.02, 55.75, 37.16, 34.50, 31.71, 31.45,
26.06, 25.80, 25.51, 24.84, 22.60, 22.52, 18.43, 18.10, 14.09, -4.15,
-4.83, -5.30, -5.50. HRFAB-MS (thioglycerol + H+ matrix) m/e
([M + H]+) 604.4071 (74.4%), calcd 604.4065. N,N-Dimethyl-4-
aminopyridine (DMAP, 2.33 g, 19.1 mmol) and benzyl chloroformate
(2.7 mL, 19.1 mmol) were added to a solution of 1,6-di-tert-
butyldimethylsilyl 2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glu-
copyranoside (5.77 g, 9.56 mmol) in CH2Cl2 (50 mL). The mixture
was stirred for 16 h at room temperature. After removal of the solvent,
the residue was chromatographed on SiO2 (ethyl acetate/hexanes 1:6)
1
mmol, 72% for two steps). H NMR (CDCl3) δ 5.95 (d, J ) 9.4 Hz,
1H), 5.31 (t, J ) 9.7 Hz, 1H), 5.25 (d, J ) 3.5 Hz, 1H), 5.14 (t, J )
9.7 Hz, 1H), 4.30 (dt, J ) 9.6, 3.4 Hz, 1H), 4.24-4.20 (m, 2H), 4.25-
4.11 (m, 1H), 2.17-2.14 (m, 2H), 2.10 (s, 3H), 2.04 (s, 3H), 2.02 (s,
3H), 1.59-1.56 (m, 2H), 1.32-1.24 (m, 4H), 0.89 (t, J ) 7.2 Hz, 3H).
13C NMR (CDCl3) δ 173.92, 171.25, 169.59, 91.50, 71.03, 68.41, 67.28,
62.18, 52.15, 36.55, 31.26, 25.25, 22.31, 20.76, 20.74, 20.63, 13.90.
HRFAB-MS (glycerol + H+ matrix) m/e ([M + H]+) 404.1919 (97.2%),
calcd 404.1921.
tert-Butyldimethylsilyl 2-Deoxy-4,6-O-isopropylidene-2-(hexanoyl-
amino)-â-D-glucopyranoside (7). Compound 6 (10.4 g, 25.8 mmol)
was dissolved in CH3CN (100 mL). tert-Butyldimethylsilyl chloride
(5.83 g, 38.7 mmol) and imidazole (2.64 g, 38.7 mmol) were added,
and the solution was allowed to stir for 4 h. The solvent was removed
in vacuo, and residue was dissolved in CH2Cl2 (100 mL). The mixture
was washed with 5% HCl (50 mL) and brine (50 mL). The organic
phase was dried with MgSO4, the solvent was removed in vacuo, and
a clear oil was obtained. The crude product was dissolved in methanol
saturated with ammonia (100 mL), and the mixture was stirred for 8 h
at 0 °C. After evaporation of the solvent, the residue was dissolved in
CH2Cl2 (100 mL). Pyridinium p-toluenesulfonate (PPTS, 1.30 g, 5.16
mmol) and 2,2-dimethoxypropane (15.9 mL, 129 mmol) were added,
and the mixture was refluxed for 4 h. The solvent was removed in
vacuo, and SiO2 chromatography was performed (CH2Cl2/MeOH 40:
1
to afford the product as a colorless oil (4.97 g, 6.75 mmol, 71%). H
NMR (CDCl3) δ 7.36-7.27 (m, 5H), 5.50 (d, J ) 9.3 Hz, 1H), 5.16-
5.09 (m, 3H), 4.94 (t, J ) 9.8 Hz, 1H), 4.69 (d, J ) 7.8 Hz, 1H), 4.00
(ddd, J ) 10.3, 9.3, 7.8 Hz, 1H), 3.70 (d, J ) 3.6 Hz, 2H), 3.52 (dt,
J ) 9.8, 3.5 Hz, 1H), 2.20-2.02 (m, 4H), 1.76-1.46 (m, 4H), 1.29-
1.19 (m, 8H), 0.99-0.87 (m, 24H), 0.11 (s, 3H), 0.08 (s, 3H), 0.02 (s,
3H), 0.00 (s, 3H). 13C NMR (CDCl3) δ 174.28, 172.73, 154.15, 135.25,
128.81, 128.46, 96.83, 74.67, 72.96, 70.08, 62.34, 56.17, 37.07, 34.19,
31.65, 31.35, 26.02, 25.78, 25.45, 24.69, 22.56, 22.44, 18.46, 18.05,
14.06, 14.05, -3.86, -4.99, -5.32, -5.38. HRFAB-MS (thioglycerol
+ Na+ matrix) m/e ([M + Na]+) 760.4265 (100%), calcd 760.4252.
1
1) to yield a clear oil (6.97 g, 16.2 mmol, 63% for three steps). H
NMR (CDCl3) δ 6.27 (d, J ) 7.3 Hz, 1H), 4.79 (d, J ) 7.8 Hz, 1H),
3.83-3.71 (m, 3H), 3.54 (t, J ) 9.3 Hz, 1H), 3.48 (q, J ) 8.3 Hz,
1H), 3.25-3.19 (m, 1H), 2.16 (dt, J ) 7.7, 3.0 Hz, 2H), 1.29-1.56
(m, 2H), 1.46 (s, 3H), 1.37 (s, 3H), 1.29-1.25 (m, 4H), 0.86-0.83
(m, 12H), 0.06 (s, 3H), 0.05 (s, 3H). 13C NMR (CDCl3) δ 174.86, 99.83,
96.35, 74.54,71.82, 67.43, 62.19, 60.33, 36.82, 31.55, 29.17, 25.79,
25.74, 25.38, 22.50, 19.17, 17.96, 14.00, -3.94, -5.04. HRFAB-MS
(thioglycerol + Na+ matrix) m/e ([M + Na]+) 454.2591 (100%), calcd
454.2601
tert-Butyldimethylsilyl 2-Deoxy-3-O-hexanoyl-2-(hexanoylamino)-
â-D-glucopyranoside (8). Hexanoic anhydride (5.6 mL, 24.3 mmol)
and pyridine (3.9 mL, 48.8 mmol) were added to a solution of 7 (6.97
g, 16.2 mmol) in CH2Cl2 (50 mL). After being stirred for 8 h, the
mixture was diluted with CH2Cl2 (100 mL). The solution was washed
with 5% HCl (100 mL), H2O (100 mL), and brine (100 mL), dried
(MgSO4), and concentrated. The residue was dissolved in CH3OH (50
mL), and PPTS (0.81 g, 3.24 mmol) was added. The mixture was
refluxed for 3 h, and the solvent was removed in vacuo. SiO2
tert-Butyldimethylsilyl 4-O-(Benzyloxycarbonyl)-2-deoxy-3-O-
hexanoyl-2-(hexanoylamino)-â-D-glucopyranoside (10). In a plastic
bottle, 9 (4.97 g, 6.75 mmol) was dissolved in CH3CN (20 mL). A
49% aqueous solution of HF was added dropwise (∼3 mL). The mixture
was stirred for 30 min, and saturated aqueous NaHCO3 (25 mL) was
added to quench the reaction. The mixture was then washed with brine
(60 mL) and the aqueous washes were extracted with CH2Cl2 (100
mL). The organic layer was dried (MgSO4) and concentrated. SiO2
chromatography (ethyl acetate/hexanes 1:3) was performed to afford
the product as a clear oil (2.48 g, 3.98 mmol, 59%). 1H NMR (CDCl3)
δ 7.31-7.25 (m, 5H), 6.43 (d, J ) 9.8 Hz, 1H), 5.29 (t, J ) 9.8 Hz,
1H), 5.10-5.05 (m, 2H), 4.88 (t, J ) 9.8 Hz, 1H), 4.74 (d, J ) 7.8
Hz, 1H), 4.03 (dd, J ) 9.8, 8.3 Hz, 1H), 3.62-3.59 (m, 2H), 3.53-
3.50 (m, 1H), 2.20-1.99 (m, 4H), 1.55-1.41 (m, 4H), 1.27-1.14 (m,
8H), 0.85-0.76 (m, 15H), 0.03 (s, 3H), -0.01 (s, 3H). 13C NMR
(CDCl3) δ 174.19, 172.82, 154.52, 134.80, 128.76, 128.63, 128.36,
96.33, 73.43, 73.31, 72.67, 70.20, 61.64, 55.65, 36.79, 33.93, 31.50,
31.06, 25.53, 25.50, 24.52, 22.39, 22.26, 17.77, 13.88, 13.84, -4.10,
9
2432 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Interactions of Lipid A Derivatives with Polymyxin B
A R T I C L E S
-5.35. HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+)
646.3385 (100%), calcd 646.3387.
69.88, 69.83 66.07, 58.26, 34.02, 31.32, 25.60, 24.54, 22.36, 17.92,
13.99, -4.10, -5.25. 31P NMR (CDCl3) δ -1.70. HRFAB-MS
(thioglycerol + Na+ matrix) m/e ([M + Na]+) 982.2288 (80.1%), calcd
982.2300.
tert-Butyldimethylsilyl 2-Deoxy-3-O-hexanoyl-2-[[(2,2,2-trichlo-
roethoxy)carbonyl]amino]-â-D-glucopyranoside (11). tert-Butyldi-
methylsilyl 2-deoxy-2-[[(2,2,2-trichloroethoxy)carbonyl]amino]-â-D-
glucopyranoside (6.50 g, 13.9 mmol) was dissolved in CH2Cl2 (100
mL). PPTS (0.70 g, 2.78 mmol) and 2,2-dimethoxypropane (8.6 mL,
69.5 mmol) were added, and the mixture was refluxed for 4 h. The
solvent was removed in vacuo, and the residue was dissolved in ethyl
acetate (100 mL). The solution was washed with water (100 mL) and
brine (100 mL), dried (MgSO4), and concentrated. The remaining clear
oil was dissolved in CH2Cl2 (100 mL). Hexanoic anhydride (3.2 mL,
13.9 mmol) and pyridine (4.5 mL, 55.6 mmol) were added to the
solution. The mixture was stirred for 4 h at room temperature and then
washed with 5% HCl (100 mL) and brine (100 mL), dried (MgSO4),
and concentrated. The residue was dissolved in CH3OH (100 mL), and
PPTS (0.70 g, 2.78 mmol) was added to the solution. The mixture was
refluxed for 4 h, and the solvent was removed in vacuo. SiO2
chromatography (CH2Cl2/MeOH 20:1) was performed to afford the
6-O-(Benzyloxycarbonyl)-4-O-[bis(benzyloxy)phosphoryl]-2-deoxy-
3-O-hexanoyl-2-[[(2,2,2-trichloroethoxy)carbonyl]amino]-D-gluco-
pyranose (13). In a plastic bottle, 12 (3.32 g, 3.46 mmol) was dissolved
in CH3CN (20 mL). A 49% aqueous solution of HF was added dropwise
(∼6 mL). The mixture was stirred for 8 h. Saturated aqueous NaHCO3
(15 mL) was added to quench the reaction. The mixture was then
washed with brine (60 mL), and the aqueous washes were extracted
with CH2Cl2 (100 mL). The organic layer was dried (MgSO4) and
concentrated. SiO2 chromatography (ethyl acetate/hexanes 1:3) gave
13 (2.16 g, 2.56 mmol, 74%) as a clear oil. 1H NMR (CDCl3) δ 7.39-
7.22 (m, 15H), 5.44 (d, J ) 9.8 Hz, 1H), 5.41 (dd, J ) 10.7, 9.3 Hz,
1H), 5.25 (br s, 1H), 5.12 (s, 2H), 5.02-4.89 (m, 4H), 4.71, 4.62 (ABq,
J ) 12.0 Hz, 2H), 4.47 (q, J ) 9.3 Hz, 1H), 4.49, 4.29 (ABq of ABX,
JAB ) 12.0 Hz, JAX ) 2.0 Hz, JBX ) 4.6 Hz, 2H), 4.20 (ddd, J ) 6.4,
4.4, 1.7 Hz, 1H), 3.99-3.91 (m, 2H), 2.15 (t, J ) 7.7 Hz, 2H), 1.52-
1.35 (m, 2H), 1.28-1.07 (m, 4H), 0.83 (t, J ) 7.1 Hz, 3H). 13C NMR
(CDCl3) δ 174.24, 155.04, 154.34, 135.61 (d, J ) 6.6 Hz), 135.52 (d,
J ) 7.4 Hz), 135.26, 128.85, 128.82, 128.79, 128.78, 128.77, 128.63,
128.20. 128.19, 95.46, 91.68, 74.83, 73.56 (d, J ) 6.2 Hz), 70.56 (d,
J ) 2.5 Hz), 70.08, 70.05, 70.02, 70.00, 69.98, 68.37 (d, J ) 5.4 Hz),
65.90, 54.37, 34.14, 31.36, 24.41, 22.48, 14.08. 31P NMR (CDCl3) δ
-2.16. HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+)
868.1440 (76.4%); calcd 868.1435.
1
product as a clear oil (4.01 g, 7.10 mmol, 51% for three steps). H
NMR (CDCl3) δ 5.57 (d, J ) 7.4 Hz, 1H), 5.11 (t, J ) 10.0 Hz, 1H),
4.77 (d, J ) 8.3 Hz, 1H), 4.70, 4.63 (ABq, J ) 11.7 Hz, 2H), 3.88,
3.83 (ABq of ABX, JAB ) 11.7 Hz, JAX ) 4.4 Hz, JBX ) 2.9 Hz, 2H),
3.73 (t, J ) 9.5 Hz, 1H), 3.60 (q, J ) 9.8 Hz, 1H), 3.46-3.43 (m,
1H), 2.38-2.31 (m, 2H), 1.64-1.58 (m, 2H), 1.32-1.26 (m, 4H), 0.93-
0.84 (m, 12H), 0.11 (s, 3H), 0.09 (s, 3H). 13C NMR (CDCl3) δ 175.27,
154.60, 96.60, 75.51, 74.98, 74.87, 69.82, 62.57, 58.22, 34.50, 31.47,
31.39, 25.75, 24.83, 24.68, 22.54, 22.49, 18.11, 14.13, -3.92, -5.05.
HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+) 588.1331
(100%), calcd 588.1330.
tert-Butyldimethylsilyl 4-O-(Benzyloxycarbonyl)-6-O-[6-O-(benz-
yloxycarbonyl)-4-O-[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-hexanoyl-
2-[[(2,2,2-trichloroethoxy)carbonyl]amino]-â-D-glucopyranosyl]-2-
deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glucopyranoside (14).
Compound 13 (1.0 g, 1.18 mmol) and freshly powdered K2CO3 (1.63
g, 11.8 mmol) were stirred in trichloroacetonitrile (20 mL) for 2 h.
The solvent was removed in vacuo, and the activated donor was passed
through a SiO2 plug with 1:1 ethyl acetate/hexanes. The solvents were
removed in vacuo. To a flask containing the activated glycosyl donor
were added the glycosyl acceptor 10 (0.74 g, 1.18 mmol), freshly
crushed 4 Å molecular sieves, and 1,2-dichloroethane (20 mL).
Trimethylsilyl trifluoromethanesulfonate (TMSOTf, 42.8 µL, 0.24
mmol) was added dropswise. The mixture was stirred for 8 h at room
temperature. The solvent was removed in vacuo, and the remaining
slurry was chromatographed (SiO2, CH2Cl2/MeOH 40:1). The product
tert-Butyldimethylsilyl 6-O-(Benzyloxycarbonyl)-4-O-[bis(benz-
yloxy)phosphoryl]-2-deoxy-3-O-hexanoyl-2-[[(2,2,2-trichloroethoxy)-
carbonyl]amino]-â-D-glucopyranoside (12). As described in the
preparation of 9, compound 11 (4.01 g, 7.10 mmol) was allowed to
react with benzyl chloroformate (1.2 mL, 8.52 mmol) and N,N-dimethyl-
4-aminopyridine (DMAP, 0.87 g, 7.10 mmol) to afford tert-butyldi-
methylsilyl 6-O-benzyloxycarbonyl-2-deoxy-3-O-hexanoyl-2-[[(2,2,2-
trichloroethoxy)carbonyl]amino]-â-D-glucopyranoside as a colorless oil
1
(3.43 g, 4.91 mmol, 69%). H NMR (CDCl3) δ 7.25-7.21 (m, 5H),
5.53 (d, J ) 9.3 Hz, 1H), 5.04-4.99 (m 3H), 4.65 (d, J ) 8.3 Hz,
1H), 4.58-4.51 (m, 2H), 4.37-4.26 (m, 2H), 3.50-3.43 (m, 3H), 2.24-
2.21 (m, 2H), 1.51-1.45 (m, 2H), 1.19-1.15 (m, 4H), 0.80-0.74 (m,
12H), -0.02 (s, 3H), -0.04 (s, 3H). 13C NMR (CDCl3) δ 175.13,
155.45, 154.43, 135.16, 128.14, 128.70, 128.37, 96.37, 74.70, 73.77,
69.60, 66.91, 63.52, 58.08, 34.36, 31.25, 25.70, 25.67, 24.69, 22.36,
18.03, 14.05, -4.10, -5.25. HRFAB-MS (thioglycerol + Na+ matrix)
m/e ([M + Na]+) 722.1708 (100%), calcd 722.1698. To a round-bottom
flask were added tert-butyldimethylsilyl 6-O-benzyloxycarbonyl-2-
deoxy-3-O-hexanoyl-2-[[(2,2,2-trichloroethoxy)carbonyl]amino]-â-D-
glucopyranoside (3.43 g, 4.91 mmol), 5-phenyl-1H-tetrazole (0.72 g,
4.91 mmol), and dibenzyl diisopropylphosphoramidite (1.6 mL, 4.91
mmol) in CH2Cl2 (60 mL). The mixture was stirred for 2 h. Water (4
mL) was added with magnesium monoperoxyphthalate hexahydrate
(MMPP, 2.43 g, 4.91 mmol), and stirring was continued for 3 h. The
solution was washed with saturated aqueous NaHCO3 (50 mL) and
brine (50 mL). The organic phase was dried (MgSO4) and concentrated.
The remaining clear oil was purified by SiO2 chromatography (ethyl
acetate/hexanes 1:4). The product was isolated as a clear oil (3.32 g,
1
was isolated as a colorless oil (0.90 g, 0.62 mmol, 53%). H NMR
(CDCl3) δ 7.39-7.25 (m, 20H), 5.55 (d, J ) 9.3 Hz, 1H), 5.42 (d, J
) 8.8 Hz, 1H), 5.25-5.16 (m, 3H), 5.15 (s, 2H), 5.16, 5.09 (ABq, J )
12.0 Hz, 2H), 5.02-4.90 (m, 4H), 4.86 (t, J ) 9.5 Hz, 1H), 4.75-
4.72 (m, 2H), 4.50-4.48 (m, 1H), 4.33 (q, J ) 9.3 Hz, 1H), 4.26-
4.22 (m, 1H), 4.17 (d, J ) 8.3 Hz, 1H), 3.92, 3.41 (ABq of ABX, JAB
) 11.7 Hz, JAX ) 2.5 Hz, JBX ) 4.4 Hz, 2H), 3.88-3.86 (m, 1H),
3.60 (ddd, J ) 6.6, 4.2, 2.4 Hz, 1H), 3.56-3.51 (m, 1H), 3.46-3.44
(m, 1H), 2.18-2.00 (m, 6H), 1.57-1.44 (m, 6H), 1.31-1.18 (m, 12H),
0.88-0.83 (m, 18H), 0.10 (s, 3H), 0.07 (s, 3H). 13C NMR (CDCl3) δ
174.01, 173.71, 172.87, 163.73, 154.93, 154.41, 135.65 (d, J ) 6.6
Hz), 135.58 (d, J ) 7.0 Hz), 135.30, 129.11, 129.00, 128.97, 128.86,
128.83, 128.80, 128.79, 128.76, 128.70, 128.55, 128.52, 128.21, 128.17,
101.00, 96.71, 96.62, 74.75, 74.01, 73.68 (d, J ) 6.2 Hz), 73.00, 72.94,
72.68 (d, J ) 5.0 Hz), 72.60, 72.35, 72.25, 72.15, 40.46, 70.34, 70.02,
70.01, 69.99, 69.97, 68.04, 65.98, 61.69, 60.57, 56.13, 36.96, 34.17,
34.14, 24.04, 31.74, 31.58, 31.35, 31.29, 31.27, 25.77, 25.74, 25.70,
25.37, 24.70, 24.64, 24.37, 22.81, 22.51, 22.46, 22.40, 22.38, 21.20,
18.00, 14.35, 14.04, 14.01, -3.80, -4.92. 31P NMR (CDCl3) δ -1.69.
HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+) 1473.4817
(64.8%); calcd 1473.4819.
1
3.46 mmol, 70%). H NMR (CDCl3) δ 7.25-7.12 (m, 15H), 5.56 (d,
J ) 9.8 Hz, 1H), 5.25 (t, J ) 10.0 Hz, 1H), 5.01 (s, 2H), 4,91-4.77
(m, 4H), 4.69 (d, J ) 7.8 Hz, 1H), 4.58-4.52 (m, 2H), 4.40-4.20 (m,
3H), 3.62-3.48 (m, 2H), 2.17-2.02 (m, 2H), 1.38-1.28 (m, 2H), 1.18-
1.02 (m, 4H), 0.78-0.72 (m, 12H), -0.02 (s, 3H), -0.05 (s, 3H). 13
C
NMR (CDCl3) δ 173.82, 154.84, 154.18, 135.85, 135.52, 135.42,
128.73, 128.67, 128.64, 128.61, 128.54, 128.32, 128.99, 96.168, 74.70,
73.94 (d, J ) 5.5 Hz), 72.55 (d, J ) 5.5 Hz), 71.90, 69.96, 69.91,
tert-Butyldimethylsilyl 4-O-(Benzyloxycarbonyl)-6-O-[6-O-(benz-
yloxycarbonyl)-4-O-[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-hexa-
noyl-2-(hexanoylamino)-â-D-glucopyranosyl]-2-deoxy-3-O-hexanoyl-2-
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2433

A R T I C L E S
Yin et al.
(hexanoylamino)-â-D-glucopyranoside (15). Freshly filed zinc and
cadmium (excess) were added to a solution of 14 (0.90 g, 0.62 mmol)
in acetic acid. The mixture was sonicated for 2 h. Insoluble materials
were removed by filtration, and the filtrate was concentrated in vacuo.
The residue was dissolved in ethyl acetate (100 mL), washed with
saturated NaHCO3 aqueous solution (100 mL) and brine (100 mL),
dried (MgSO4), and concentrated. The residue was dissolved in CH2-
Cl2 (100 mL), and hexanoic anhydride (290 µL, 1.24 mmol) and
pyridine (250 µL, 3.10 mmol) were added to the solution. The mixture
was stirred for 2 h, washed with 5% HCl (100 mL) and brine (100
mL), dried (MgSO4), and concentrated. SiO2 chromatography (ethyl
acetate/hexanes 1:4) was performed to afford the product as a clear oil
6H), 1.94-1.81 (m, 2H), 1.61-1.55 (m, 2H), 1.54-1.43 (m, 6H), 1.30-
1.12 (m, 16H), 0.89-0.82 (m, 12H). 13C NMR (CDCl3) δ 174.06,
173.77, 173.52, 173.05, 154.89, 154.36, 135.65 (d, J ) 6.4 Hz), 135.59
(d, J ) 6.9 Hz), 135.33 (d, J ) 6.0 Hz), 135.22 (d, J ) 6.4 Hz), 135.23,
135.03, 129.32, 129.22, 128.14, 129.09, 128.84, 128.70, 128.66, 128.33,
128.23, 100.76, 96.01 (d, J ) 6.9 Hz), 73.65 (d, J ) 6.0 Hz), 73.19 (d,
J ) 2.3 Hz), 72.69 (d, J ) 5.0 Hz), 71.38, 70.50, 70.23 (d, J ) 5.5
Hz), 70.12 (d, J ) 5.5 Hz), 69.89, 69.88, 69.86, 69.83, 69.82, 69.62,
66.35, 66.04, 53.49, 52.00 (d, J ) 8.2 Hz), 36.65, 36.37, 34.03, 33.99,
31.70, 31.38, 31.35, 31.18, 25.48, 25.15, 24.56, 24.31, 22.49, 22.47,
22.41, 22.34, 14.09, 14.03, 14.00, 13.97. 31P NMR (CDCl3) δ -1.75,
-3.11. HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+)
1543.6241 (100%); calcd 1543.6246.
1
(0.43 g, 0.31 mmol, 50%). H NMR (CDCl3) δ 7.40-7.26 (m, 20H),
5.84 (d, J ) 8.3 Hz, 1H), 5.25-5.02 (m, 6H), 4.99-4.92 (m, 4H),
4.86 (t, J ) 9.8 Hz, 1H), 4.76 (d, J ) 8.3 Hz, 1H), 4.49 (d, J ) 11.7
Hz, 1H), 4.32 (q, J ) 9.3 Hz, 1H), 4.23 (dd, J ) 11.7, 5.4 Hz, 1H),
4.12 (d, J ) 8.3 Hz, 1H), 3.93-3.86 (m, 2H), 3.72 (q, J ) 8.3 Hz,
1H), 3.58 (dt, J ) 9.8, 3.4 Hz, 1H), 3.44-3.42 (m, 1H), 3.35 (dd, J )
11.7, 3.9 Hz, 1H), 2.33 (t, J ) 7.3 Hz, 2H), 2.21-2.02 (m, 6H), 1.66-
1.34 (m, 8H), 1.39-1.12 (m, 16H), 0.91-0.06(m, 21H), 0.09 (s, 3H),
0.06 (s, 3H), -3.86, -5.02. 13C NMR (CDCl3) δ 178.63, 173.83,
173.78, 173.01, 154.85, 154.45, 135.54 (d, J ) 6.8 Hz), 135.48 (d, J
) 7.3 Hz), 135.20, 135.17, 129.18, 128.73, 128.69, 128.58, 128.51,
128.46, 128.13, 128.08, 100.79, 96.42, 73.70 (d, J ) 6.0 Hz), 73.13,
72.57, 72.52, 72.32, 72.03, 70.32, 69.96, 69.95, 69.90, 67.79, 65.99,
56.13, 56.87, 36.87, 36.70, 34.18, 34.10, 33.93, 31.61, 31.52, 31.37,
31.34, 31.22, 25.90, 25.37, 25.35, 24.62, 24.60, 24.28, 22.47, 22.45.
22.43, 22.36,17.93, 14.07, 14.02, 13.98, -3.86, -5.02. 31P NMR
(CDCl3) δ -1.76. HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M
+ Na]+) 1397.6492 (56.3%); calcd 1397.6506.
4-O-(Benzyloxycarbonyl)-6-O-[6-O-(benzyloxycarbonyl)-4-O-[bis-
(benzyloxy)phosphoryl]-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-
â-D-glucopyranosyl]-1-O-[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-
hexanoyl-2-(hexanoylamino)-r-D-glucopyranose (16). As described
for 13, compound 15 (0.43 g, 0.31 mmol) was treated with a 49%
aqueous solution of HF to give 4-O-(benzyloxycarbonyl)-6-O-[6-O-
(benzyloxycarbonyl)-4-O-[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-
hexanoyl-2-(hexanoylamino)-â-D-glucopyranosyl]-2-deoxy-3-O-hexanoyl-
2-(hexanoylamino)-D-glucopyranose (0.26 g, 0.21 mmol, 68%) as a
colorless oil.1H NMR (CDCl3) δ 7.39-7.23 (m, 20H), 5.82 (d, J )
7.8 Hz, 1H), 5.75 (d, J ) 9.3 Hz, 1H), 5.37 (dd, J ) 10.3, 8.8 Hz,
1H), 5.30 (dd, J ) 10.7, 9.8 Hz, 1H), 5.20 (d, J ) 3.4 Hz, 1H), 5.16-
5.10 (m, 5H), 4.98-4.90 (m, 5H), 4.81 (d, J ) 8.3 Hz, 1H), 4.75 (t, J
) 9.8 Hz, 1H), 4.49 (dd, J ) 12.2, 2.5 Hz, 1H), 4.42-4.38 (m, 1H),
4.27-4.18 (m, 3H), 4.13-4.09 (m, 1H), 3.80-3.77 (m, 1H), 3.58-
3.46 (m, 3H), 2.21-2.05 (m, 8H), 1.60-1.43 (m, 8H), 1.34-1.12 (m,
16H), 0.92-0.82 (m, 12H). 13C NMR (CDCl3) δ 174.33, 173.97,
173.68, 173.12, 154.95, 154.58, 135.56 (d, J ) 6.9 Hz), 135.49 (d, J
) 7.6 Hz), 135.24, 135.18, 100.83, 91.36, 73.75 (d, J ) 6.1 Hz), 72.80,
72.60, 72.56, 72.15, 70.95, 70.26, 69.98, 69.94, 69.92, 69.89, 69.87,
54.74, 52.24, 36.76, 36.75, 34.16, 33.98, 31.56, 31.51, 31.36, 31.27,
25.39, 25.35, 24.60, 24.38, 22.49, 22.40, 14.10, 14.05, 14.02. 31P NMR
(CDCl3) δ -1.92. HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M
+ Na]+) 1283.5652 (100%); calcd 1283.5644. As in the preparation
of 12, 4-O-(benzyloxycarbonyl)-6-O-[6-O-(benzyloxycarbonyl)-4-O-
[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-
â-D-glucopyranosyl]-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-D-glu-
copyranose (260 mg, 0.21 mmol) was treated with 5-phenyl-1H-
tetrazole (59 mg, 0.41 mmol), dibenzyl diisopropylphosphoramidite
(140 µL, 0.41 mmol), and then H2O and MMPP (202 mg, 0.41 mmol)
to afford 16 (243 mg, 0.16 mmol, 76%) as a clear oil. 1H NMR (CDCl3)
δ 7.40-7.25 (m, 30H), 6.32 (d, J ) 8.8 Hz, 1H), 5.63 (dd, J ) 5.4,
3.9 Hz, 1H), 5.59 (d, J ) 8.3 Hz, 1H), 5.17-4.93 (m, 14H), 4.84 (t, J
) 9.8 Hz, 1H), 4.51 (dd, J ) 11.72, 2.0 Hz, 1H), 4.32 (q, J ) 9.3 Hz,
1H), 3.96-3.88 (m, 2H), 3.77 (d, J ) 11.2 Hz, 1H), 3.38 (ddd, J )
7.3, 5.4, 1.5 Hz, 1H), 3.32 (dd, J ) 12.7, 4.4 Hz, 1H), 2.20-2.09 (m,
2-Deoxy-6-O-[2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glu-
copyranosyl]-3-O-hexanoyl-2-(hexanoylamino)-r-D-glucopyranose
1,4′-Bisphosphate (2). Compound 16 (240 mg, 0.16 mmol) was
dissolved in THF (20 mL) and transferred to a hydrogenation vessel.
Palladium on carbon (10%, 100 mg) was added to the vessel. The vessel
was subjected to H2 (200 psi) at room temperature for 8 h. The catalyst
was removed via filtration through a Celite pad, and solvent was
removed in vacuo. The residue was lyophilized from a water solution
1
to give 5 (117 mg, 0.13 mmol, 81%) as a white powder. H NMR
(CD3OD) δ 5.48-5.45 (m, 1H), 5.25 (t, J ) 9.3 Hz, 1H), 5.17 (t, J )
10.8 Hz, 1H), 4.74 (d, J ) 8.3 Hz, 1H), 4.24-4.23 (m, 1H), 4.15 (d,
J ) 10.7 Hz, 1H), 4.08-4.01 (m, 2H), 3.90-3.79 (m, 4H), 3.59 (t, J
) 9.8 Hz, 1H), 3.52-3.51 (m, 1H), 2.41-2.25 (m, 4H), 2.23-2.11
(m, 4H), 1.63-1.53 (m, 8H), 1.36-1.24 (m, 16H), 0.93-0.89 (m, 12H).
13C NMR (CD3OD) δ 176.85, 176.54, 175.34, 175.94, 102.64, 96.48
(d, J ) 4.1 Hz), 76.81, 74.59, 74.55, 74.11, 69.42, 69.16, 62.04, 55.54,
53.31 (d, J ) 7.6 Hz), 37.55, 37.11, 35.30, 35.11, 32.79, 32.71, 32.63,
32.59, 26.84, 26.72, 25.86, 25.53, 23.63, 23.60, 23.57, 14.50, 14.42.
31P NMR (CD3OD) δ 3.88, 2.02. HRFAB-MS (thioglycerol + Na+
matrix) m/e ([M + Na]+) 915.3608 (100%); calcd 915.3633. Anal.
Calcd for C36H66N2O19P2: C, 48.43; H, 7.45. Found: C, 48.59; H, 7.46.
3-O-Acetyl-6-O-[3-O-acetyl-2-acetamino-2-deoxy-â-D-glucopyra-
nosyl]-2-acetamino-2-deoxy-r-D-glucopyranose 1,4′-Bisphosphate
1
(1). H NMR (CD3OD) δ 5.50-5.49 (m, 1H), 5.22 (t, J ) 9.8 Hz,
1H), 5.15 (dd, J ) 10.7, 9.3 Hz, 1H), 4.77 (d, J ) 8.3 Hz, 1H), 4.24-
4.20 (m, 1H), 4.14 (dt, J ) 10.7, 2.9 Hz, 1H), 4.08 (d, J ) 11.7 Hz,
1H), 4.04-4.01 (m, 1H), 3.90-3.78 (m, 4H), 3.58 (t, J ) 9.3 Hz, 1H),
3.51-3.50 (m, 1H), 2.05 (s, 3H), 2.03 (s, 3H), 1.96 (s, 3H), 1.93 (s,
3H). 13C NMR (CD3OD) δ 174.19, 173.80, 172.75, 172.39, 102.45,
96.23 (d, J ) 4.8 Hz), 76.75, 74.96, 74.55, 74.44, 74.03, 69.51, 69.25,
62.01, 55.80, 53.50 (d, J ) 7.6 Hz), 23.16, 22.78, 21.25, 21.13. 31P
NMR (CD3OD) δ 3.85, 2.15. HRFAB-MS (thioglycerol + Na+ matrix)
m/e ([M + 4Na - 3H]+) 757.0571 (100%); calcd 757.0587.
3-O-Decanoyl-2-(decanoylamino)-6-O-[3-O-decanoyl-2-(deca-
noylamino)-2-deoxy-â-D-glucopyranosyl]-2-deoxy-r-D-gluco-
pyranose 1,4′-Bisphosphate (3). 1H NMR (CD3OD) δ 5.48-5.46 (m,
1H), 5.25 (t, J ) 9.8 Hz, 1H), 5.17 (t, J ) 9.8 Hz, 1H), 4.77 (d, J )
8.3 Hz, 1H), 4.22 (q, J ) 9.3 Hz, 1H), 4.14 (dt, J ) 10.7, 2.9 Hz, 1H),
4.07 (d, J ) 11.7 Hz, 1H), 4.02-3.99 (m, 1H), 3.92-3.79 (m, 4H),
3.60 (t, J ) 9.8 Hz, 1H), 3.51-3.50 (m, 1H), 2.33-2.31 (m, 4H), 2.21-
2.16 (m, 4H), 1.59-1.55 (m, 8H), 1.33-1.25 (m, 48H), 0.91-0.88
(m, 12H). 13C NMR (CD3OD) δ 176.80, 176.49, 175.28, 174.97, 102.77,
96.28 (d, J ) 5.3 Hz), 76.89, 74.63, 74.35, 73.94, 69.40, 69.19, 62.02,
55.58, 53.40 (d, J ) 8.4 Hz), 37.68, 37.23, 35.42, 35.20, 33.32, 33.27,
33.25, 30.94, 30.88, 30.84, 30.80, 30.78, 30.75, 30.69, 30.67, 30.64,
30.54, 27.24, 27.14, 26.24, 26.14, 23.97, 23.93, 14.66, 14.63. 31P NMR
(CD3OD) δ 3.88, 2.06. HRFAB-MS (thioglycerol + Na+ matrix) m/e
([M + 4Na - 3H]+) 1183.5773 (100%); calcd 1183.5776.
2-Deoxy-3-O-hexanoyl-2-(hexanoylamino)-2-r-D-glucopyranose
4-Phosphate (4). As described for 12, compound 8 (500 mg, 1.02
mmol) was treated with benzyl chloroformate (180 µL, 1.22 mmol)
and DMAP (124 mg, 1.02 mmol) to afford tert-butyldimethylsilyl 6-O-
(benzyloxycarbonyl)-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-
9
2434 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Interactions of Lipid A Derivatives with Polymyxin B
A R T I C L E S
glucopyranoside as a colorless oil (516 mg, 0.83 mmol, 82%). 1H NMR
(CDCl3) δ 7.37-7.30 (m, 5H), 6.34 (d, J ) 9.3 Hz, 1H), 5.23 (dd, J
) 10.7, 8.8 Hz, 1H), 5.15 (s, 2H), 4.81 (t, J ) 7.8 Hz, 1H), 4.47, 4.40
(ABq of ABX, JAB ) 11.7 Hz, JAX ) 2.0 Hz, JBX ) 5.9 Hz, 2H), 3.87-
3.83 (m, 1H), 3.68-3.62 (m, 6H), 3.58-3.55 (m, 1H), 2.45-2.31 (m,
2H), 2.13-2.07 (m, 2H), 1.64-1.54 (m, 4H), 1.33-1.21 (m, 8H), 0.91-
0.85 (m, 15H), 0.08 (s, 3H), 0.06 (s, 3H). 13C NMR (CDCl3) δ 175.17,
173.34, 155.54, 135.33, 128.86, 128.79, 128.73, 128.43, 96.39, 75.49,
74.01, 69.93, 69.63, 63.75, 56.28, 37.11, 34.39, 31.72, 31.41, 25.83,
25.80, 25.74, 25.67, 24.81, 22.62, 22.52, 18.11, 14.13, -3.93, -5.01.
HRFAB-MS (thioglycerol + Na+ matrix) m/e ([M + Na]+) 646.3395
(100%); calcd 646.3387. tert-Butyldimethylsilyl 6-O-(benzyloxycar-
bonyl)-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glucopyrano-
side (516 mg, 0.83 mmol) was treated with 5-phenyl-1H-tetrazole (158
mg, 1.08 mmol) and dibenzyl diisopropylphosphoramidite (360 µL,
1.08 mmol) and then H2O and MMPP (532 mg, 1.08 mmol) to afford
tert-butyldimethylsilyl 6-O-(benzyloxycarbonyl)-4-O-[bis(benzyloxy)-
phosphoryl]-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glucopyra-
[bis(benzyloxy)phosphoryl]-2-deoxy-3-O-hexanoyl-2-(hexanoylamino)-
â-D-glucopyranoside (282 mg, 0.37 mmol) was hydrogenated in the
presence of palladium on carbon (10%, 50 mg), and the resulting oil
was lyophilized to afford 4 (143 mg, 0.31 mmol, 84%) as a white
powder. 1H NMR (CDCl3) δ 5.36 (dd, J ) 10.7, 9.3 Hz, 1H), 5.06 (d,
J ) 3.4 Hz, 1H), 4.25 (q, J ) 9.3 Hz, 1H), 4.14 (dd, J ) 10.7, 3.4 Hz,
1H), 3.99-3.97 (m, 1H), 3.86-3.75 (m, 2H), 2.40-2.26 (m, 2H), 2.19-
2.13 (m, 2H), 1.60-1.52 (m, 4H), 1.35-1.24 (m, 8H), 0.92-0.89 (m,
6H). 13C NMR δ 176.47, 175.35, 92.79, 74.37, 72.85, 71.98 (d, J )
4.6 Hz), 62.12, 53.78, 37.10, 35.23, 32.71, 32.64, 26.91, 25.58, 23.64,
23.62, 14.43. 31P NMR (CDCl3) δ 4.04. HRFAB-MS (thioglycerol +
Na+ matrix) m/e ([M + Na]+) 478.1816 (74.4%); calcd 478.1818.
Prodan Fluorescence Aggregation Studies. Prodan was purchased
from Molecular Probes and was used as received. Fluorescence data
were obtained by use of a FluoroMax-3 fluorometer (Jobin Yvon Inc.).
Incrementally varied concentrations of 2 and 3 were prepared by dilution
of a 2 mM stock solution with Millipore water. An aliquot of 200 µL
of a 5 µM solution of Prodan in Millipore water was added to 1.8 mL
of each solution while it was gently stirred with a small magnetic stir
bar in a cuvette for 60 s, followed by sonication for an additional 30
s before measurements were made. Prodan fluorescence spectra were
measured over a range of 335-600 nm (λex ) 351 nm, slit width ) 5
nm).
1
noside (535 mg, 0.60 mmol 72%) as a clear oil. H NMR (CDCl3) δ
7.35-7.23 (m, 15H), 5.87 (d, J ) 9.3 Hz, 1H), 5.31 (t, J ) 9.3 Hz,
1H), 5.12 (s, 2H), 5.00-4.89 (m, 4H), 4.79 (d, J ) 8.3 Hz, 1H), 4.49
(dd, J ) 11.7, 2.0 Hz, 1H), 4.38 (q, J ) 9.3 Hz, 1H), 4.30 (dd, J )
12.2, 6.3 Hz, 1H), 3.91 (q, J ) 9.3 Hz, 1H), 3.71 (ddd, J ) 8.8, 5.86,
2.0 Hz, 1H), 2.25-2.14 (m, 2H), 2.13-2.03 (m, 2H), 1.58-1.39 (m,
4H), 1.31-1.12 (m, 8H), 0.90-0.81 (m, 15H), 0.08 (s, 3H), 0.06 (s,
3H). 13C NMR (CDCl3) δ 174.12, 172.79, 154.87, 135.81 (d, J ) 6.9
Hz), 135.51 (d, J ) 6.4 Hz), 137.46 (d, J ) 6.9 Hz), 135.23, 128.69,
128.67, 128.64, 128.62, 128.54, 128.07, 128.01, 96.31, 74.01 (d, J )
6.4 Hz), 72.58 (d, J ) 2.7 Hz), 72.40 (d, J ) 5.5 Hz), 69.86, 69.82,
69.80, 69.76, 69.73, 69.36 (d, J ) 5.5 Hz), 66.13, 56.03, 36.83, 33.99,
31.54, 31.26, 26.62, 25.34, 24.39, 22.46, 22.38, 17.89, 13.96, -4.10,
-5.21. 31P NMR δ -1.77. HRFAB-MS (thioglycerol + Na+ matrix)
m/e ([M + Na]+) 906.3997 (100%); calcd 906.3990. The silyl ether
(533 mg, 0.60 mmol) was treated with a 49% aqueous solution of HF
to give 6-O-(benzyloxycarbonyl)-4-O-[bis(benzyloxy)phosphoryl]-2-
deoxy-3-O-hexanoyl-2-(hexanoylamino)-â-D-glucopyranoside (282 mg,
Isothermal Titration Calorimetry. Lipid A derivatives (1-4) and
PMB or PMBN (Sigma) were dissolved in 50 mM potassium phosphate
or Tris buffer. Solutions were degassed before titrations. Titrations were
performed in a CSC4200 isothermal titration calorimeter (Calorimetry
Sciences Corp.). The instrument was calibrated by measuring the area
under a test electrical heat pulse and comparing it with heat input. The
effective volume of sample cell was measured by titration of 100 mM
barium chloride into 10 mM 18-crown-6 (volume ) 1.381 mL). Lipid
A derivatives were loaded into the sample cell, and PMB or PMBN
was placed in a 250 µL syringe at concentrations 6-7.5-fold greater
than that of lipid A derivatives. The system was allowed to equilibrate
and a stable baseline was recorded before initiation of an automated
titration. A typical titration sequence involved 20 injections at 1000 s
intervals of 10 µL aliquots of PMB or PMBN into the sample cell.
Throughout the titration, the cell was stirred continuously at 297 rev/
min. The data generated in a titration experiment were analyzed by
Bindworks (Applied Thermodynamics).
1
0.37 mmol, 62%) as a colorless oil. H NMR (CDCl3) δ 7.36-7.22
(m, 15H), 5.97 (d, J ) 9.3 Hz, 1H), 5.40 (t, J ) 9.3 Hz, 1H), 5.17-
5.11 (m, 3H), 4.99-4.88 (m, 4H), 4.49-4.43 (m, 2H), 4.29 (dd, J )
11.7, 4.4 Hz, 1H), 4.23-4.18 (m, 2H), 2.18-2.06 (m, 4H), 1.56-1.38
(m, 4H), 1.29-1.10 (m, 8H), 0.88-0.82 (m, 6H). 13C NMR δ 174.46,
173.54, 155.01, 135.55 (d, J ) 7.6 Hz), 135.49 (d, J ) 7.7 Hz), 135.20,
128.77, 128.72, 128.70, 128.52, 128.11, 128.10, 91.47, 73.71 (d, J )
6.1 Hz), 70.87 (d, J ) 2.3 Hz), 69.95, 69.93, 69.89, 68.00 (d, J ) 5.4
Hz), 65.99, 52.22, 36.69, 34.07, 31.51, 31.32, 25.38, 24.36, 22.48, 22.44,
14.04, 14.02. 31P NMR (CDCl3) δ -2.15. HRFAB-MS (thioglycerol
+ Na+ matrix) m/e ([M + Na]+) 792.3137 (100%); calcd 792.3125.
As described in the preparation of 2, 6-O-(benzyloxycarbonyl)-4-O-
Acknowledgment. Generous financial support from the NSF
(CAREER) and the NIH (GM 54616) is gratefully acknowl-
edged. We thank Bryce Harrison, Glade Roper, Benjamin
Gordon, and Adam Peters for participation in synthesis of lipid
A derivatives.
JA0284456
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2435