Synthesis of and evaluation of lipid A modification by 4-substituted 4-deoxy arabinose analogs as potential inhibitors of bacterial polymyxin resistance

Article abstract of DOI:10.1016/j.bmcl.2007.12.061

Three sets of novel 4-deoxy-l-arabinose analogs were synthesized and evaluated as potential inhibitors of the bacterial resistance mechanism in which lipid A, on the outer membrane, is modified with 4-amino-4-deoxy-l-arabinose (l-Ara4N). One compound diminished the transfer of l-Ara4N onto lipid A. These results suggest that small molecules might be designed that would effect the same reversal of bacterial resistance observed in genetic knockouts.

Full text of DOI:10.1016/j.bmcl.2007.12.061

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1507–1510
Synthesis of and evaluation of lipid A modification
by 4-substituted 4-deoxy arabinose analogs as potential inhibitors
of bacterial polymyxin resistance
T. Kline,a,* M. S. Trent,d C. M. Stead,d M. S. Lee,e M. C. Sousa,e H. B. Felise,a
H. V. Nguyena and S. I. Millera,b,c
aDepartment of Genome Sciences, University of Washington, Seattle, WA 98195, USA
bDepartment of Microbiology, University of Washington, Seattle, WA 98195, USA
cDepartment of Medicine, University of Washington, Seattle, WA 98195, USA
dDepartment of Biochemistry and Molecular Biology, Medical College of Georgia, Augusta, GA 30912, USA
eDepartment of Chemistry and Biochemistry, University of Colorado Boulder, CO 80309, USA
Received 2 October 2007; revised 21 December 2007; accepted 21 December 2007
Available online 27 December 2007
Abstract—Three sets of novel 4-deoxy-L-arabinose analogs were synthesized and evaluated as potential inhibitors of the bacterial
resistance mechanism in which lipid A, on the outer membrane, is modified with 4-amino-4-deoxy-L-arabinose (L-Ara4N). One com-
pound diminished the transfer of L-Ara4N onto lipid A. These results suggest that small molecules might be designed that would
effect the same reversal of bacterial resistance observed in genetic knockouts.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Polymyxin is a currently prescribed antibiotic whose use
is predominately topical as a result of its systemic toxic-
ity. With the advent of devices that include as ‘topical’
the aerosol route of administration, polymyxin is now
widely used for chronic airway infections, notably in
cystic fibrosis patients for whom this is a significant
cause of morbidity and mortality.1 Polymyxin, like all
cationic antimicrobial peptides (CAPs), binds to nega-
tive charges on phosphates of the lipopolysaccharide
component lipid A that makes up the outer leaflet of
the Gram-negative bacterial outer membrane. Lipid A
binding by CAPs results in critical disorganization of
the outer membrane.2 Bacteria can resist the action of
polymyxin by modification of lipid A phosphates, which
decreases surface negative charge and presumably re-
duces polymyxin binding to the altered electrostatic
topography of the remodeled membrane.3–5
Five enzymes catalyze seven reactions that produce an
undecaprenylated aminosugar at the cytoplasmic sur-
face of the inner membrane. This glycolipid is then
translocated to the periplasmic side of the inner mem-
brane where ArnT/PmrK catalyzes the displacement
of the prenyl group by the 40- and/or 10-phosphate(s)
of lipid A.6
If the mechanisms of lipid A modification could be dis-
abled, CAP-resistant bacteria might be returned to a
susceptible state. An agent capable of this restoration
could also promote the activity of host antimicrobial
peptides in controlling infection, and indeed, mutants
defective for the addition of aminoarabinose are attenu-
ated for virulence in mouse models of infection. A
re-sensitizing agent could also expand the repertoire of
utility for polymyxin and enable the durable therapeutic
utility of the newer CAPs under investigation.7 While
the ultimate goal of the project is to develop small
molecule inhibitors of the L-Ara4N-lipid A that could
exogenously reproduce the effects of genetic knockouts,
we have initially focused on the synthesis of compounds
with inhibitory activity in vitro. We now report on the
synthesis and evaluation of a series of 4-modified arab-
inose analogs.
The pathway that results in aminoarabinose incorpora-
tion into lipid A has been elucidated in some detail.
Keywords: Lipid A remodeling; Aminoarabinose; 4-Deoxy-4-substi-
tuted arabinose analogs; Polymyxin resistance.
*
Corresponding author. Tel.: +1 206 543 9587; fax: +1 206 543
5383; e-mail: klinet@u.washington.edu
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.12.061

1508
T. Kline et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1507–1510
Figure 1.
The 4-amino function of aminoarabinose is formylated
for several of the cytoplasmic biosynthetic steps. N-
formylation drives the otherwise-unfavorable transami-
nation reaction (ArnB/PmrH) forward,8 and the formyl
group is subsequently removed by ArnD.9 We con-
structed analogs (Fig. 1) that might mimic the N-formy-
lation transition state or serve as surrogates for the
formylated amine. Three isosteres for the N–C(OH)–N
framework were selected, phosphonamidate, phosphor-
amidate, and azide. For the azide and phosphoramidate,
several substituents at the anomeric carbon were ex-
plored: O-phosphoryl, O-acetyl, O-benzyl, and, the most
substrate-mimetic, O-uridyldiphosphoryl. Although the
initial biological characterization was to be in cell-free
enzyme preparations, we designed in some structural
features that anticipated our longer term plans for com-
pounds having good stability in the human host and the
ability to penetrate bacterial outer membranes. Since
five of the L-Ara4N biosynthetic enzymes, both the sol-
uble and membrane bound, lie beyond the inner mem-
brane, while the lipid A modifying enzyme, ArnT/
PmrK, is on the inner membrane, access to these en-
zymes was expected to be a significant determinant of
our compounds’ ultimate capabilities against intact bac-
teria. With this in mind, the 2- and 3-hydroxyls were
acetylated, and the phosphoramidates/phosphonami-
dates maintained as esters. We anticipated that these
esters would encounter nonselective esterases and be
hydrolyzed in vivo. The selection of the compounds ben-
efited from an economy in design, since many candidates
were also intermediates en route to the more complex
structures. The 4-azido-2,3-isopropylidene-1-methyl
acetal10 served as the single common precursor.
The 4-azido analogs 1, 2, and 3 were synthesized as
shown in Scheme 1. Conversion of the isopropylidene
N3
N3
N3
N3
O
O
O
O
OCH3
b
d
O
c
a
O
O
O
O
O
O
O
O
O
O
O
O
OH
PH
O
O
O
O
O
O
e
60:40
β:α
9
8
1
+H3N
N3
N3
O
O
O
O
O
2
O-
- +NEt3
O
O
O
O
O
O
O
P
P
O
NH
O
OH
O-
O
O
O
O
O
P
N
O
O
3
2
HO
OH
Scheme 1. Reagents and conditions: (a) i—1 M HCl/acetone, then NaHCO3, ii—1 M HCl, 90 ꢁC, iii—Li2CO3, iv—Ac2O, NaOAc, reflux; (b)
(CH3)2NH, CH3CN, ꢀ20 ꢁC; (c) i—2-chloro-4[H]-benzodioxaphosphorin-4-one, NEt3, THF–dioxane, 0 ꢁC, ii—H2O; (d) i—Dowex 50W · 8, H+
form, THF–dioxane, ii—t-BuOOH/decane, cat. I2, NEt3; (e) i—tri-n-octylamine/pyridine, ii—UMP morpholidate, DCCM salt, tetrazole, pyridine.
CH3O
P
CH3O
CH3O
P
O
OCH3
O
O
OCH3
O
O
OCH3
O
P
N3
N3
HN
HN
+H
N
HN
*
c
b
d
a
O
O
O
O-
O
O
O
O
O
O
O
O
O
2
O
O
O
O
O
O
O
O
O
P O
OH
O
O
O
O
O
O- +HNEt3
P
O
O
P
O
P
OH
O
O
NH
O
O
O
O
O
P
-O
N
O
O
6
8
10
4
5
HO
OH
Scheme 2. Reagents: (a) i—n-BuLi, THF, ii—tetrabenzylpyrophosphate, THF; (b) (CH3)3P, pyridine, CH2Cl2; (c) i—H2, Pd–C, CH3OH, ii—NEt3;
(d) UMP morpholidate, tetrazole, pyridine.

T. Kline et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1507–1510
1509
O
OC2H5
O
O
O
OC2H5
O
P
P
N3
P
HN
O
HN
O
HN
O
OC2H5
O
+
H2N
O
O-
O
O
O
O
O
2
O
O
O
b
a
O
c
P O
O
O
O
O
O
O- +HNEt3
O
O
O
O
P
O
O
P
O
O
P
NH
O
O
O
O
O
P
OH
-O
N
O
O
10
12
11
7
HO
OH
Scheme 3. Reagents: (a) i—ethyl benzyl H-phosphinate, BTSTFA, pyridine, ii—1:3 CH3OH–CHCl3, D; (b) i—H2, Pd–C, CH3OH, ii—NEt3; (c) i—
tri-n-octylamine, pyridine, ii—UMP morpholidate, tetrazole, pyridine.
methyl acetal directly to the triacetate was accomplished
in a three-step sequence using Takeo’s procedure.11
Selective 1-O-deacetylation was achieved by treatment
of the triacetate with a saturated solution of dimethyl-
amine in acetonitrile at ꢀ20 ꢁC.12 Installation of the
phosphate via the two-step method13 gave the b-L-arab-
inose phosphate 2 in modest yield but good stereochem-
ical purity.14 Reaction of the 1-phosphate, as its tri-n-
octylamine salt, with uridine monophosphoryl morphol-
idate15 in the presence of tetrazole, completed the syn-
thesis of 3. For the synthesis of 4, 5, and 6 the azide
was converted via a modified Staudinger reaction to
the phosphoramidate (Scheme 2).16 Without the syn-
thetic incompatibility presented by the azide function,
dibenzylphosphate could be directly introduced (8–10)
and readily removed (4–5).
tive bacterium Salmonella typhimurium strain JSG43523
were assayed for the addition of aminoarabinose to
32P labeled precursors of lipid A.6 Salmonella JSG435
overproduces the enzymatic machinery required for
the modification of lipid A with aminoarabinose, includ-
ing the membrane-bound glycosyl transferase PmrK/
ArnT.6 As determined by densiometry24 of the 32P
TLC (Fig. 2), compound 1 effected a threefold diminu-
tion of aminoarabinose addition to the phosphate
groups of the lipid A precursor lipid IVA, demonstrating
inhibition of PmrK/ArnT.
Acting against the transferase PmrK/ArnT, azido com-
pound 1 represents the first reported small molecule to
The efficiency of the Staudinger reaction that retains the
N–P bond for the preparation of phosphoramidates
encouraged us to develop a similar strategy for the
synthesis of phosphonamidate 7. Intermediate 10 was
reacted with ethyl benzyl H-phosphinate17,18 in the pres-
ence of 6 equivalents of BTSTFA, followed by thermal
rearrangement to give phosphonamidate 11 (Scheme
3).16,19 Hydrogenolysis of the benzyl esters gave the free
phosphate 12. Solubility mandated prior formation of
the bis-n-octylammonium salt for the reaction with
UMP morpholodate, giving 7 as the corresponding
bis-n-octylammonium salt.
Compounds 1–7 were evaluated in vitro in a single glo-
bal assay for the decarboxylase and transformylase
activity encoded in ArnA/PmrI as well as the transami-
nase activity of ArnB/PmrH as previously described.20,21
These enzymes were unaffected at compound concentra-
tions up to 100 lM. Crystal structures of ArnA and
ArnB20–22 suggest that the 2- and 3-acetates, as well as
excessive functionality at the 4-position, may be imped-
ing rather than promoting activity against these two
aminoarabinose biosynthetic enzymes.
Figure 2. Assay of PmrK/ArnT aminoarabinose transferase activity.
Membranes from S. typhimurium strain JSG435 were assayed for the
transfer of L-Ara4N from endogenous sources to the tetra-acylated
lipid A precursor 40-32P-lipid IVA. The protein concentration was
1 mg/ml and the concentration of the inhibitor when present was
1 mM. Assays were carried out as previously described6 for 0.5 h (A)
or for 6 h (B) at 30 C using 5 lM 40-32P-lipid IVA (20,000 cpm/nmol).
The reaction products were separated by thin layer chromatography in
the solvent chloroform, pyridine, 88% formic acid, water (50:50:16:5,
v/v) and visualized by Phosphorimaging. Lipid IIA arises from the
addition of L-Ara4N to the starting substrate lipid IVA. Lipids IIB and
IVB arise from the PagP-dependent addition of palmitate to lipid IIA
and lipid IVA, respectively. Reaction products modified with L-Ara4N
are indicated by an asterisk.
Considering that the 4-azido group might be a neutral
surrogate for the 4-amino group, compounds 1, 2, and
3 were selected for evaluation in an assay against
PmrK/ArnT.6 Compound 7 was included as a structural
(and presumed negative) control of the most substrate-
like of the azide trio, compound 3. To monitor PmrK/
ArnT activity, whole membranes from the Gram-nega-

1510
T. Kline et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1507–1510
9. Williams, G. J.; Breazeale, S. D.; Raetz, C. R.; Naismith,
J. H. J. Biol. Chem. 2005, 280, 23000.
10. Naleway, J. J.; Raetz, C. R.; Anderson, L. Carbohydr. Res.
1988, 179, 199.
11. Takeo, K.; Murata, Y.; Kitamura, S. Carbohydr. Res.
1992, 224, 311.
12. Upreti, M. R. Dipali; Vishwakarma, Ram A. Tetrahedron
2000, 56, 6577.
interfere with lipid A modification. With the synthetic
methods developed, the structural biology of the en-
zymes well characterized, and with compound 1 as a po-
tential lead, additional analogs may now be prepared
with increased activity against the lipid A modifying en-
zymes and against resistant bacteria.25
13. Takaya, K.; Nagahori, N.; Kurogochi, M.; Furuike, T.;
Miura, N.; Monde, K.; Lee, Y. C.; Nishimura, S. J. Med.
Chem. 2005, 48, 6054.
Acknowledgments
14. For the synthesis of compound 2 and analytical data,
please see Supplementary information.
15. Wittmann, V.; Wong, C. H. J. Org. Chem. 1997, 62, 2144.
16. For the synthesis of compound 4 and analytical data,
please see Supplementary information.
17. Abrunhosa-Thomas, I.; Ribiere, Patrice; Adcock, Alicia
C. Synthesis 2006, 325.
18. Deprele, S. M.; Jean-Luc, A. J. Organiomet. Chem. 2002,
634–644, 154.
The authors thank Stona Jackson and Kathleen Barry
for technical assistance in the synthetic chemistry. This
work was supported by an NIAID award for the North-
west Regional Center of Excellence for Biodefense and
Emerging Infectious Diseases U54 AI057141. The gener-
ous support of the National Institutes of Health NIH
(RO1-AI064184 to M.S.T) and (AI060841-01 to
M.C.S.) is also acknowledged.
19. For the synthesis of compound 11 and analytical data,
please see Supplementary information. Compound 11.
20. Gatzeva-Topalova, P. Z.; May, A. P.; Sousa, M. C.
Biochemistry 2005, 44, 5328.
Supplementary data
21. Gatzeva-Topalova, P. Z.; May, A. P.; Sousa, M. C.
Structure 2005, 13, 929.
22. Gatzeva-Topalova, P. Z.; May, A. P.; Sousa, M. C.
Biochemistry 2004, 43, 13370.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2007.12.061.
23. Gunn, J. S.; Ryan, S. S.; Van Velkinburgh, J. C.; Ernst, R.
K.; Miller, S. I. Infect. Immun. 2000, 68, 6139.
24. TLC plates were exposed overnight to a PhosphorImager
Screen and product formation detected using a Bio-Rad
Molecular Imager PhosphorImager. Enzyme activity was
calculated by determining the percentage of the substrate
converted to product by densitometry using Quantity One
Software.
25. A preliminary assay against whole bacteria showed that
none of the compounds, at highest concentrations rang-
ing from 90 mM (compound 1) to 500 mM (compound
7), were able to restore susceptibility to the CAP
polymyxin in Salmonella strain JSG435 that constitutively
synthesizes a lipid A modified with aminoarabinose. The
lack of effect against intact bacteria could result from
insufficient activity against the target, poor permeability/
retention, metabolic inactivation, or a combination of
these factors.
References and notes
1. Beringer, P. Curr. Opin. Pulm. Med. 2001, 7, 434.
2. Brogden, K. A. Nat. Rev. Microbiol. 2005, 3, 238.
3. Peschel, A. Trends Microbiol. 2002, 10, 179.
4. McPhee, J. B.; Lewenza, S.; Hancock, R. E. Mol.
Microbiol. 2003, 50, 205.
5. Bader, M. W.; Navarre, W. W.; Shiau, W.; Nikaido, H.;
Frye, J. G.; McClelland, M.; Fang, F. C.; Miller, S. I. Mol.
Microbiol. 2003, 50, 219.
6. Trent, M. S.; Ribeiro, A. A.; Lin, S.; Cotter, R. J.; Raetz,
C. R. J. Biol. Chem. 2001, 276, 43122.
7. Baron, C.; Coombes, B. Infect. Disord. Drug Targets 2007,
7, 19.
8. Breazeale, S. D.; Ribeiro, A. A.; McClerren, A. L.; Raetz,
C. R. J. Biol. Chem. 2005, 280, 14154.