Báo cáo khoa học: Characterization of the lipopolysaccharide and b-glucan of the fish pathogen Francisella victoria ppt

Characterization of the lipopolysaccharide and b-glucan
of the fish pathogen Francisella victoria
William Kay
1
, Bent O. Petersen
2
, Jens Ø. Duus
2
, Malcolm B. Perry
3
and Evgeny Vinogradov
3
1 Department of Biochemistry and Microbiology, University of Victoria, BC, Canada
2 Carlsberg Laboratory, Copenhagen, Denmark
3 Institute for Biological Sciences, National Research Council, Ottawa, ON, Canada
Members of the bacterial genus Francisella belong to
the Gram-negative Proteobacteria. Their taxonomic
position is not completely clear, as no closely related
microorganisms have been detected. Francisella
includes two species: Francisella tularensis and Franci-
sella philomiragia. There are four subspecies of F. tula-
rensis: tularensis, holarctica, mediasiatica , and novicida.
Of all subspecies, the F. tularensis subspecies tularensis
is the most infective and fatal for humans and, due to
its very low infective dose, is considered as a biological
weapon or bioterrorist agent [1]. With the introduct-
ion of rapid PCR-based methods of screening of
environmental samples, potential new variants of
Francisella were detected [2–4] lipopolysaccharide
(LPS) of Francisella has unusually low biological activ-
ity, and is considered as a potential component of

antitularemia vaccines [5–8].
Recently a virulent bacterial fish pathogen was iso-
lated from a moribund Tilapia (Oreochromis niloticus
niloticus). Tilapia sp. are warm water finfish of con-
siderable commercial importance world-wide. This
pathogen, often and incorrectly referred to as a Rick-
ettsia-like organism, was characterized and identified
as a unique Francisella sp. by 16S rRNA gene
Keywords
core; Francisella; Francisella victoria; lipid A;
lipopolysaccharide; O-chain
Correspondence
E. Vinogradov, Institute for Biological
Sciences, National Research Council,
100 Sussex Drive, K1A 0R6 Ottawa,
ON, Canada
Fax: +1 613 9529092
Tel: +1 613 9900832
E-mail:
(Received 20 January 2006, revised 26 April
2006, accepted 5 May 2006)
doi:10.1111/j.1742-4658.2006.05311.x
Lipopolysaccharide (LPS) and b-glucan from Francisella victoria, a fish
pathogen and close relative of highly virulent mammal pathogen Francisella
tularensis, have been analyzed using chemical and spectroscopy methods.
The polysaccharide part of the LPS was found to contain a nonrepetitive
sequence of 20 monosaccharides as well as alanine, 3-aminobutyric acid,
and a novel branched amino acid, thus confirming F. victoria as a unique
species. The structure identified composes the largest oligosaccharide eluci-
dated by NMR so far, and was possible to solve using high field NMR

with cold probe technology combined with the latest pulse sequences, inclu-
ding the first application of H2BC sequence to oligosaccharides. The non-
phosphorylated lipid A region of the LPS was identical to that of other
Francisellae, although one of the lipid A components has not been found
in Francisella novicida. The heptoseless core-lipid A region of the LPS con-
tained a linear pentasaccharide fragment identical to the corresponding
part of F. tularensis and F. novicida LPSs, differing in side-chain substitu-
ents. The linkage region of the O-chain also closely resembled that of other
Francisella. LPS preparation contained two characteristic glucans, previ-
ously observed as components of LPS preparations from other strains of
Francisella: amylose and the unusual b-(1–6)-glucan with (glycerol)
2
phos-
phate at the reducing end.
Abbreviations
BABA, b-aminobutyric acid; CHA or CHB, cysteine heart agar or broth; Fuc4N, 4-amino-4,6-dideoxygalactose; HPAEC, high-performance
anion-exchange chromatography; Kdo, 3-deoxy-
D-manno-octulosonic acid; LOS, lipooligosaccharide; LPS, lipopolysaccharide; LVS, live
vaccine strain; P, phosphate; QuiN, 2-amino-2,6-dideoxy-
D-glucose; Qui3N, 3-amino-3,6-dideoxy-D-glucose; Qui4N, 4-amino-4,6-dideoxy-D-
glucose.
3002 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS
sequencing and serological cross-reactivity with other
Francisella sp., and it was subsequently named Franci-
sella victoria (unpublished results). Comparative analy-
sis of the LPS from related species is important in
order to gain an understanding of the molecular basis
of their biological properties, the host specificity and
diversity of members of this important bacterial genus,
as well as the nature of its pathogenicity. The results

could also be useful for vaccine development against
fish diseases and possibly against tularemia in humans.
Here we present the results of the structural analysis of
the lipopolysaccharide of the first known fish Francisel-
la sp., F. victoria.
Results
Silver-stained SDS ⁄ PAGE of F. victoria LPS, whole-
cell proteinase digests or western blot stained with
polyclonal antisera to F. victoria revealed a predomin-
ant immuno-staining band of the large lipooligosaccha-
ride (LOS) component and a smaller less intense band
of the core-lipid A component (Fig. 1). An additional
diffuse band of higher molecular mass components
was visible at high sample load. Immunostaining using
rabbit polyclonal antisera raised to whole cells of
F. tularensis live vaccine strain (LVS), F. novicida or
with monoclonal antibodies to LPS from F. tularensis
or F. novicida showed no reaction with the dominant
LOS band of F. victoria. Cross-reactivity was observed
with the low molecular weight band of F. victoria (data
not shown). However, staining of proteinase K-diges-
ted samples of F. novicida with anti-F. victoria anti-
serum were negative. Thus the LPS fractions of whole
cells of F. victoria were immunochemically distinct
from the LPS fractions of the other Francisella sp.
Monosaccharide analysis (GC MS of alditol ace-
tates) of the whole LPS revealed the presence of
rhamnose, fucose, 3-amino-3,6-dideoxyglucosamine
(Qui3N), quinovosamine (QuiN), 4-amino-4,6-dide-
oxyhexosamine (probably a mixture of Fuc4N and

Qui4N, as determined from NMR results), mannose,
glucose, and glucosamine with dominant peak of glu-
cose 10 times larger than that of any other compo-
nent. High glucose content was due to the presence of
glucans in the LPS preparation. GC of acetylated or
trimethylsilylated (R)-2-butyl glycosides was used to
determine the absolute configuration of the monosac-
charides, which turned out to be l for Rha and Fuc,
and d for QuiN, Qui3N, Glc, Man, and GlcN. Config-
urations of Fuc4N and Qui4N have not been deter-
mined because of unclear results.
LPS was subjected to mild acid hydrolysis, which
gave water-insoluble lipid A and water-soluble prod-
ucts. Lipid A was purified by conventional silica gel
chromatography in a CHCl
3
–MeOH solvent system.
Comparison of the
1
H-NMR spectra of the unfraction-
ated lipid A and chromatographically fractionated
samples indicated that the fraction eluted with 10%
MeOH in CHCl
3
contained the major component. It
was used in further studies as ‘lipid A’. Fatty acid ana-
lysis of the purified lipid A showed the presence of
C14 : 0, C16 : 0 (minor), C16 : 0 (3-OH), and C18 : 0
(3-OH) straight-chain acids. Two-dimensional NMR
spectra of this product were identical to the spectra of

the lipid A from F. tularensis [9]. The MALDI mass
spectrum of the lipid A showed one major peak at m ⁄ z
1391.9, which corresponds to the [M + Na]
+
ion of
the structure with two GlcN, one C14 : 0, one C16 : 0
(3-OH), and two C18 : 0 (3-OH) fatty acids. Smaller
peaks at 1364.9 and 1419.9 indicated the presence of
the variants with two methylene groups shorter or lon-
ger acids, respectively. A peak corresponding to the
glycosyl cation of unit B was observed at 654.4 Da,
corresponding to the acylation of the unit B with
C14 : 0 and C18 : 0 (3-OH) acids. The same results
were observed for F. tularensis lipid A [9].
For a more detailed analysis of the distribution of
O-linked acids, the lipid A was treated with NH
4
OH
1 2 3
High molecular
mass components
LOS bands
Core-Lipid A band
Fig. 1. Western blot of F. victoria LPS products. Lane 1: whole
F. victoria cells treated with Proteinase K overnight at 60 °C. Lane
2: Non-precipitated fraction of LPS after overnight ultracentrifuga-
tion at 120 000 g. Lane 3: LPS ultracentrifuge precipitate.
W. Kay et al. Francisella victoria lipopolysaccharide
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3003
and products were analyzed by MALDI mass spectr-

ometry, as described [10]. This treatment removes all
O-linked fatty acids except those acylating OH-3
of the amide-linked 3-hydroxyacyl groups. F. victoria
lipid A contained only one hydrolyzable acyl substitu-
ent at O-3 of residue A. Indeed, the mass spectrum of
the products showed the new peak of the compound
with a mass of 1137.8, corresponding to the loss of
C16 : 0 (3-OH) acyl from O-3 of GlcN A residue.
Together with the above described data, this informa-
tion can correspond only to the acyl group distribution
shown in Fig. 2. This experiment also confirmed the
previously determined structure of F. tularensis and
F. novicida lipid A.
Water-soluble products from the mild acid hydrolysis
of F. victoria LPS were reduced with NaBH
4
and separ-
ated by size-exclusion chromatography. It gave minor
amounts of polymeric product, which was shown to be a
starch-like glucan, and three oligosaccharide fractions.
Oligosaccharides were further separated by anion-
exchange chromatography to give oligosaccharides 1
and 2 (as a mixture), 3, 4, b-glucan (see Scheme 1), and
several other products, apparently being fragments of
structure 3. Oligosaccharide 4 and b-glucan were addi-
tionally purified by high performance anion exchange
chromatography (HPAEC).
Oligosaccharides 1–4 were analyzed using two-dimen-
sional NMR spectroscopy (COSY, TOCSY, NOESY,
heteronuclear single quantum coherence (HSQC), het-

eronuclear two bond correlation (H2BC), heteronuclear
multiple quantum coherence (HMQC)-TOCSY, and
heteronuclear multiple bond correlation (HMBC)) and
MS. Spectra of the simplest oligosaccharide 4, represent-
ing the core part of the LPS, were completely assigned
in agreement with the proposed structure, consisting of
four mannose residues and one residue of Kdo-ol
(Scheme 1, Table 1). ESI MS gave a mass of 888.9 Da,
which agreed with the structure.
Assignment of the spectra of the large oligosaccha-
rides 1–3 presented a significant experimental challenge
due to an unusually high number of nonrepeating
components and consequential signal overlap even at
800 MHz (Fig. 3). Additional problems arose from the
presence of the oligosaccharides 1 and 2 as an unsepa-
rable mixture, and partial O-acetylation of the Qui3N
unit V. In order to simplify spectra, oligosaccharides
were O-deacetylated and spectra of native and deacyl-
ated products were analyzed. For the interpretation of
the NMR spectra, a new method called H2BC [11–13]
was used. It produces spectra containing three-bond
H–C–C correlations, which makes it possible to iden-
tify C-2 signals starting from H-1, and C-5 starting
from H-6, as well many other signals.
NMR analysis of the oligosaccharide 3 showed that
it contains all components of the core oligosaccharide
4. Additionally a clearly visible nonreducing end struc-
ture was present, consisting of the monosaccharide
residues Y-V-U-Z-R, a core-linked sequence L-M-[K-]-
J-I, and a number of glucose and fucose residues

between these two fragments, with an integral intensity
of their signals being 1.5–2 times higher than that of
above mentioned residues. This pointed to the possible
presence of loosely defined ‘repeating units’. Relative
and anomeric configurations of the monosaccharides
were deduced from the proton–proton coupling con-
stants and chemical shifts of proton and carbon sig-
nals. Connections between monosaccharides were
identified on the basis of NOE and HMBC correla-
tions (Table 1). A very large number of the NOE cor-
relations from the H-6 of the 6-deoxysugars was
observed, and most of them could be rationalized
within the proposed structures (Scheme 1). It should
be noted that the signals of the quinovosamine unit I
were of low intensity. A similar feature was observed
previously in the analysis of the repeating unit-core
oligosaccharides prepared from F. novicida LPS [14],
and probably can be explained by the restrained
motion of this residue in densely packed structure.
The ESI MS of the oligosaccharide 3 contained triple
and quadruple charged peaks corresponding to a mass
of 4381.7 Da, which corresponds to a composition
Hex
10
dHex
12
HexNAc
1
dHexNAc
3

Kdool
1
(average mass
of 4380.1), thus involving two copies of a fragment
P-W-X-[Q-]-T or X-[Q-]-T-[S-]-N (these structures are
identical). The spectrum also contained smaller peaks
of oligomers of lower and higher mass, differing by
hexose units (162 Da). MS-MS analysis generally con-
firmed structural assignment, but provided no new data
Fig. 2. Structure of the F. victoria lipid A.
Francisella victoria lipopolysaccharide W. Kay et al.
3004 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS
regarding the structure of the most obscure region
between Fuc R and Fuc L (data not shown).
After the determination of the structure of the prod-
uct 3, the full sequence of the oligosaccharides 1 and 2
was determined by NMR. All signals of the components
present in the oligosaccharide 3 were found at the same
positions except for the substituted Qui3NAc residue Y.
Oligosaccharide 1 contained three nonsugar compo-
nents: alanine, 3-aminobutyric acid (homoalanine or
BABA), and a novel branched amino acid designated
AA. HMBC correlations allowed us to trace the
BABA-acylated alanine, which was in turn linked to
N-4 of terminal Fuc4N residue RR. BABA had a free
amino group.
Component AA contained a methyl group and two
other protons. All proton signals were singlets and
showed only NOE correlations between each other.
Methyl group protons gave HMBC correlations to

A
B
C
D
E
F
Scheme 1. Structures of the isolated oligosaccharide fragments of the F. victoria lipopolysaccharide (LPS). (A) BABA = 3-aminobutyric acid
(homoalanine). AcOH hydrolysis products (R
1
= Ac or H): full structure = 1; structure ending at PP (PP’) = 2; structure ending at Y (Y’) = 3;
structure ending at F (F’) = 4. (B) O,N-deacylated LPS products: full structure = 5; structure ending at PP (PP’ in this case) = 6; structure
ending at Y (Y’ in this case) = 7; structure ending at F (F’ in this case) = 8.(C)b-glucan, structures of core-lipid A backbone of different Fran-
cisella LPSs: (D) F. tularensis,(E)F. novicid and, (F) F. Victoria.
W. Kay et al. Francisella victoria lipopolysaccharide
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3005
Table 1. NMR data for oligosaccharides 1–8. Components having close chemical shift values are grouped and average data are presented
for them.
Unit,
compound Nucleus H ⁄ C1 H⁄ C2 H⁄ C3 H⁄ C4 H⁄ C5 H⁄ C6a H⁄ C7⁄ 6b H ⁄ C8a⁄ 8b
NOE
from H-1
b-Fuc4N H 4.72 3.43 4.02 3.59 4.10 1.34 PP3
RR; 5 C 103.7 71.5 70.2 56.5 68.7 16.6
b-Fuc4Nac H 4.54 3.46 3.85 4.19 3.81 1.16 PP3
RR; 1 C 105.1 72.4 72.5 54.9 71.1 17.1
b-Qui4N H 4.55 3.61 3.90 3.10 3.84 1.41 KK4
PP; 5 C 103.5 73.8 79.3 56.5 70.2 18.0
b-Qui4N H 4.53 3.40 3.64 3.02 3.81 1.40 KK4
PP¢; 6 C 103.5 73.9 72.9 57.9 70.1 18.0
b-Qui4NAA H 4.47 3.58 3.79 3.74 3.74 1.30 KK4

PP; 1 C 103.8 75.0 72.2 56.4 72.8 18.2
b-Glc H 4.60 3.39 3.64 3.70 3.67 3.86 4.01 Y4
KK; 5,6 C 103.5 75.4 75.4 79.5 76.0 60.9
b-Glc H 4.48 3.25 3.61 3.54 3.55 3.78 4.03 Y4
KK; 1,2 C 103.5 74.5 75.5 80.7 76.0 61.9
b-Qui3N H 4.82 3.67 3.36 3.66 3.82 1.46 V2
Y; 5,6 C 103.7 71.4 58.1 80.3 74.8 18.4
b-Qui3NAc H 4.53 3.29 3.86 3.41 3.67 1.43 V2
Y; 1,2 C 105.1 72.7 52.2 81.6 73.9 17.9
b-Qui3N H 4.75 3.54 3.14 3.34 3.63 1.37 V2
Y¢; 7 C 104.5 72.2 58.9 72.8 75.0 17.9
b-Qui3NAc H 4.53 3.25 3.82 3.15 3.56 1.36 V2
Y¢; 3 C 105.1 72.9 57.7 74.7 74.5 18.2
b-Qui3N H 4.70 3.66 3.11 3.22 3.53 1.27 U1,2
V; 5–7 C 104.7 81.0 58.9 74.3 74.3 18.0
b-Qui3NAc H 4.67 3.61 3.97 3.17 3.56 1.28 U1,2
V*; 1–3 deac C 105.4 79.3 57.6 74.3 74.6 18.3
b-Qui3NAc4Ac H 4.72 3.73 4.16 4.63 3.78 1.19 U1,2
V; 1–3 C 105.2 78.6 55.6 75.1 72.2 17.9
a-Rha H 5.45 4.09 3.84 3.41 3.84 1.29 Z3
U; 1–3, 5–7 C 101.7 83.4 71.6 74.3 70.1 17.8
a-Rha H 4.95 4.08 3.95 3.56 4.02 1.30 R3
Z; 1–3, 5–7 C 97.8 71.4 79.0 73.0 70.1 17.9
a-Fuc H 5.01 3.92 4.01 4.03 4.53 1.19 W4,6
R; 1–3, 5–7 C 101.6 68.3 75.5 69.4 68.1 16.6
a-Glc H 5.41 3.56 3.75 3.44 3.70 3.84 3.71 W1,2
P; 1–3, 5–7 C 100.8 72.8 74.0 70.6 73.8 62.0
a-Fuc H 5.10 4.12 4.29 3.91 4.58 1.29 X4,6
W; 1–3, 5–7 C 101.3 75.3 70.3 82.3 69.3 16.3
a-Fuc H 5.15 3.85 4.11 3.91 4.42 1.46 T4,6

X; 1–3, 5–7 C 101.1 70.0 69.8 82.5 68.8 18.0
a-Glc H 5.37 3.58 3.78 3.42 3.79 3.74 3.84 T3
Q; 1–3, 5–7 C 100.8 72.8 74.2 71.0 73.9 62.1
a-Fuc H 5.00 4.15 4.15 4.06 4.60 1.31 N4,6
T; 1–3, 5–7 C 101.7 70.5 76.3 80.7 69.8 17.0
a-Glc H 5.39 3.56 3.73 3.44 3.71 3.85 3.70 N1,2
S; 1–3, 5–7 C 100.8 73.0 73.9 70.6 73.8 62.0
a-Fuc H 5.10 4.11 4.29 3.91 4.56 1.28 L4,6
N; 1–3, 5–7 C 101.3 75.5 70.3 82.3 69.3 16.4
a-Fuc H 4.98 3.82 3.93 3.91 4.44 1.35 M4,6
L; 1–3, 5–7 C 100.7 69.5 70.2 82.4 68.6 17.1
b-Glc H 4.61 3.39 3.52 3.56 3.63 4.02 3.82 J4,6
M; 5–7 C 103.5 74.8 76.0 77.4 74.9 60.1
b-Glc H 4.60 3.43 3.62 3.57 3.51 4.03 3.85 J4,6
M; 1–3 C 104.2 75.5 75.5 78.0 76.7 60.7
Francisella victoria lipopolysaccharide W. Kay et al.
3006 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS
three carbons, two of them protonated (59.4 and
66.5 p.p.m.) and one quaternary carbon at 79.0 p.p.m.
Proton signals at 4.23 and 4.73 p.p.m. correlated with
a quaternary carbon atom and carbonyl carbon atom
signals at 176.4 and 170.5 p.p.m. These data suggested
that AA was a five-carbon dicarboxylic acid with a
methyl group at C-3 and amino or hydroxy substitu-
ents at positions 2, 3, and 4.
For detailed analysis of the structure of AA, LPS
was depolymerized with anhydrous HF and Qui4NAA
was isolated using reverse-phase HPLC. NMR spectra
of this monosaccharide (not shown) confirmed its
gluco configuration. HMBC correlation was observed

between C-1 of AA at 170.9 p.p.m. and H-4 of the
Qui4N, indicating acylation of NH
2
-4 of Qui4N with
one of AA carboxyl groups. Spectra contained signals
of one N-acetyl group acylating N-4 of the AA. The
ESI mass spectrum of Qui4NAA showed the molecular
mass of 361.3 Da, 18 units less than expected for the
linear structure of the AA, which pointed to the lac-
tam formation. MS ⁄ MS experiments led to the obser-
vation of a signal at m ⁄ z 199.2, corresponding to a
cyclic AA component. The cyclic structure of the AA
was confirmed by the observation of the weak C-5:
H-2 HMBC correlation. There was no data for the
determination of the configuration of chiral atoms.
Taken together, these experimental data agreed with
the structure (Fig. 4).
As only one Qui4N (residue PP) was present in the
oligosaccharides 1 and 2, isolated Qui4NAA repre-
sented residue PP monosaccharide. Close values of
NMR shifts for AA in the oligosaccharides and in the
Table 1. (Continued).
Unit,
compound Nucleus H ⁄ C1 H⁄ C2 H⁄ C3 H⁄ C4 H⁄ C5 H⁄ C6a H⁄ C7⁄ 6b H ⁄ C8a⁄ 8b
NOE
from H-1
a-GlcN H 5.52 3.32 3.92 3.50 3.96 3.89 3.74 J3
K; 5–7 C 97.2 54.8 70.5 70.3 73.1 60.9
a-GlcNAc H 5.22 3.95 3.80 3.48 3.95 3.67 3.78 J3
K; 1–3 C 99.4 54.8 73.6 71.3 73.4 62.1

a-Fuc H 5.10 4.13 4.15 4.20 4.38 1.31 I3
J; 5–7 C 100.4 69.4 74.0 79.3 68.4 16.4
a-Fuc H 4.98 3.93 4.02 4.17 4.41 1.28 I3
J; 1–3 C 101.2 69.9 73.9 80.3 69.0 16.4
b-QuiN H 4.53 2.95 3.48 3.29 3.57 1.35 F4
I; 5–7 C 101.7 57.5 84.8 74.0 72.9 16.5
b-QuiNAc H 4.56 3.87 3.64 3.32 3.56 1.35 F4
I; 1–3 C 102.1 56.8 80.9 74.7 73.1 18.1
a-Man H 5.12 4.09 3.92 3.90 4.02 3.63 3.85 F1,2
G; 1–8 C 103.0 71.5 71.8 67.3 73.8 61.8
b-Man H 4.75 4.26 3.86 3.74 3.55 3.95 3.77 E4
F; 1–3, 5–7 C 101.3 76.2 71.6 78.6 76.5 62.0
b-Man H 4.72 4.18 3.78 3.64 3.45 3.77 3.96 E4
F¢; 4,8 C 101.3 77.9 74.9 68.3 78.6 62.4
a-Man H 5.00 4.06 3.83 3.68 3.55 3.79 3.91 E6
H; 1–8 C 102.2 71.2 72.1 68.1 75.0 62.4
a-Man H 5.22 4.15 3.95 4.02 4.04 3.82 4.05 C5,7
E; 1–8 C 102.4 71.2 70.6 77.5 71.9 67.6
Kdo H 1.83 2.17 4.01 4.03 3.72 4.00 3.81 ⁄ 3.98
D; 5–8 C 102.5 36.0 67.6 68.1 73.4 71.6 64.4
Kdo H 1.97 2.12 4.20 4.27 3.66 3.78 3.63 ⁄ 3.90
C; 5–8 C 101.2 36.0 71.3 75.9 73.8 70.9 65.0
b-GlcN H 4.73 3.11 3.61 3.55 3.63 3.58 3.58 A6
B; 5–8 C 100.2 56.4 73.5 70.5 75.3 61.8
GlcNol H 3.65 ⁄ 3.78 3.59 3.73 3.93 3.88 4.12 3.74
A; 5–8 C 63.1 56.1 72.4 69.4 71.3 71.8
BABA H 2.34 ⁄ 2.83 3.25 1.23
1 C 173.8 38.8 53.7 15.1
Ala H 4.26 1.40
1 C 177.8 51.2 17.7

AA H 4.23 4.73 1.51
1,2 C 170.5 66.5 79.0 59.4 176.4 23.2
W. Kay et al. Francisella victoria lipopolysaccharide
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3007
HF-released product indicated that its structure was
not modified during HF treatment.
A number of oligosaccharide fractions not presented
in Scheme 1 were isolated after acetic acid hydrolysis,
which had no components of the core, and contained
mostly fragments of the oligosaccharide chain from unit
L to Y or from J to Y, including side chains. Their
NMR spectra contained many minor signals, mostly of
a-glucose. As PAGE of the LPS showed the presence of
high molecular mass chains, it seems reasonable to
believe that these oligosaccharides formed the polymeric
chain beyond units Y or RR, and for some reason were
cleaved off in both acidic and alkaline conditions.
Deacylation of the LPS with 4 m KOH in the pres-
ence of NaBH
4
with subsequent fractionation by
gel-chromatography on Sephadex G50 gave four frac-
tions. As in the case of acetic acid hydrolysis, the
product eluted with the void volume turned out to be
starch-like material. A second (major) fraction
contained large oligosaccharides 5–7. A third fraction
contained b-glucan with aglycon, modified due to alka-
line conditions; it was not further analyzed. The lowest
molecular mass component, eluted near the salt peak,
contained mostly b-glucan and core oligosaccharide 8,

purified further by HPAEC. Minor fractions from
HPAEC contained the variants of structure 8 with
partly degraded lipid A glucosamine due to alkaline
conditions of deacylation. Products 5–7 were found
impossible to separate in conditions used, and they
were analyzed in the mixture.
Oligosaccharide 8 was analyzed by NMR. Complete
assignment of two-dimensional NMR spectra led to
the identification of two a-Kdo residues, one b-GlcN,
one glucosaminitol, and four mannose residues. The b-
configuration of the Man F followed from the observa-
tion of intraresidual strong NOEs between H-1 and
H-3, H-5, and also from the low field position of the
C-5 signal at 78.6 p.p.m. Characteristic NOE between
H-3 of the Kdo C and H-6 of the Kdo D residues indi-
cated the attachment of Kdo D in a-configuration to
O-4 of Kdo C. All glycosidic linkages were identified
on the basis of transglycosidic NOE and HMBC
[p.p.m.]5 4 3 2
Fig. 3. The 800 MHz
1
H-NMR spectrum of the mixture of oligosac-
charides 1 and 2 is shown.
Table 2. NMR data for b-glucan. A¢ is nonreducing end residue, A – repeating, A¢ – linked to Gro.
Unit 1 ⁄ 1¢ 23⁄ 3¢ 4566¢
b-Glc A¢ H 4.44 3.25 3.43 3.33 3.38 3.84 3.66
C 103.8 74.0 76.6 70.6 76.9 61.7
b-Glc A H 4.45 3.26 3.43 3.39 3.56 3.79 4.15
C 103.9 74.0 76.6 70.5 75.9 69.7
b-Glc A¢ H 4.42 3.25 3.34 3.38 3.56 3.77 4.14

C 103.5 74.0 76.6 70.5 75.9 69.7
Gro B H 3.83 ⁄ 3.88 3.99 3.73 ⁄ 3.87
C 67.3 70.1 71.5
Gro C H 3.80 ⁄ 3.86 3.83 3.54 ⁄ 3.61
C 67.3 71.6 63.1
Fig. 4. Structure of the isolated 4-amino-4,6-dideoxy glucose, N-ac-
ylated with the amino acid AA (unit PP-AA in the oligosaccharides).
Francisella victoria lipopolysaccharide W. Kay et al.
3008 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS
correlations. The following NOEs were observed in the
product 8: B1A6, C3D6, E1C5, E1C7, F1E4, F1E6,
G1F2, G1F3, H1E6, which corresponds to the struc-
ture presented on Scheme 1.
Analysis of the mixture of oligosaccharides 5–7 by
NMR (Fig. 5, Table 1) confirmed the sugar backbone
structure determined from the analysis of the products
1–3 with all nonsugar components of 1–3 absent in
5–7. Compound 5 had the most complete structure;
in the oligosaccharide 6 terminal Fuc4N was missing;
in 7, the nonreducing sequence b-Fuc4N-3-b-Qui4N-4-
b-Glc- was missing (Scheme 1).
The structure of b-glucan (Scheme 1) was studied by
NMR (Table 2), MS and chemical analysis. Monosac-
charide analysis showed the presence of glucose and
glycerol. NMR data indicated that short b-(1–6)-linked
glucose oligomers have an aglycon, consisting of two
glycerol residues, linked by a phosphodiester bond
(
31
P-NMR signal at 1.08 p.p.m.). Methylation analysis

of b-glucan revealed the presence of terminal and
4-substituted glucopyranose. Positive-mode MALDI
mass spectrum of the b-glucan (Fig. 6) contained a ser-
ies of peaks that could be attributed to ions
[M + Na]
+
and [M +2Na)1]
+
, with maximum con-
tent of oligomer Glc
9
, which gave disodium peak at
m ⁄ z 1749.2. The same b-glucans were found previously
in LPS preparations from F. victoria and F. tularensis
[9,15].
5.5
5.0 4.5 4.0
4.5
4.0
3.5
3.0
K2
K4
K3
S2
Q2
Q4
P4P4
P2
P5

P3
S3
Q3
U3+U1:Z3
U4
U2
K1:J3
Q1:T3
S1:N2
Q1:T4
P1:W2
E2
E1:C5
E3
X2
X4
X1:T4
X3
X1:T3
N3
W3
T1:N3
W1:X5
T3
T2
W2
G2
G1:F2
G3
W1:X4

J2
J3
J4
Z2
Z1:R3
Z1:R4
Z3
L2
R2
R3
R4
T1:W4
L3
H2
H3
H1:E6
W1:P5
J1:I3
V1:U2
F2
F1:E4
I1:F4
F3
F5
V2
Y'2
Y'5
M1:J4
N35
W35

T35
T45
J45
J35
W23
N23
X5:T3
X35
R35
R45
W45
X45
W5:X2
L1:M4
M4,5
M3
V5
I3
B2
V3,4
RR2
RR4
Y'3
Y3
I2
I4
Y4
B3,5
M2
KK2

PP2
PP'3
PP'2
PP4
PP'4
KK3
KK4
X5:Q3
W34
N34
Y1:V2
RR3
PP1:KK4
KK6
PP3
Y2
Fig. 5. Overlap of COSY (green), TOCSY (magenta) and NOESY (black) spectra of the mixture of oligosaccharides 5–7, showing correlations
from anomeric protons. Intensity is set to high level for clarity, but some important correlations became invisible.
W. Kay et al. Francisella victoria lipopolysaccharide
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3009
Discussion
The LPS of F. victoria contains two variants: a rough-
type structure consisting of the core and lipid A, and a
much larger structure with a nonrepetitive oligosaccha-
ride linked to the core. SDS ⁄ PAGE of the proteinase
K-treated whole cells or of the purified LPS appears to
reflect this composition with low molecular weight
bands, probably representing the lipid A and core
oligosaccharide regions, and the more abundant,
higher molecular weight bands, probably representing

lipooligosaccharide conjugated to the polysaccharide
component. None of these LPS components were
cross-reactive with LPS of the other Francisella sp.,
with the exception of some nonreciprocal cross-reactiv-
ity with F. novicida, perhaps due to the similarity of
their core-region oligosaccharides. These results serve
to emphasize the uniqueness of the F. victoria oligosac-
charide and to confirm it as a unique Francisella sp.
Lipid A had the same structure as determined earlier
for F. tularensis [9,16] and F. novicida [14,15] LPS with
a characteristic nonphosphorylated free reducing end.
Another variant of the lipid A, which seems to be not
substituted with core and has a phosphorylated redu-
cing end, and which has been found in F. tularensis
and F. novicida, was not detected in F. victoria. The
inner core of the LPS of F. victoria resembles the
core of F. tularensis and F. novicida in the presence of
oligosaccharide fragment b-Man-4-a-Man-5-a-Kdo
(marked in bold font together with lipid A backbone,
Scheme 1), but it has an additional side-chain Kdo
residue and different branching substituents. The poly-
saccharide part is linked to the core via b-N-acetylqui-
novosamine or its close relative b-N,N-diacetyl-
bacillosamine in all Francisella LPSs (Scheme 1).
Oligosaccharide 1 was found to consist of 33
monosaccharide residues and some nonsugar compo-
nents, which to the best of our knowledge is the lar-
gest complex carbohydrate structure elucidated to
date. The oligosaccharide had no repeating units in
the usual sense, although it contained two copies of

the same pentasaccharide fragment. Its analysis
required application of all available NMR methods,
and it is still matter of good luck that signals were
spread sufficiently to allow interpretation of the spec-
tra. Assignment relied strongly on the combined
usage of several heteronuclear
1
H–
13
C correlated
experiment including the new experiment H2BC, as
described recently [13].
LPS preparation from F. victoria contained two pol-
ymers of glucose, a starch-like polymeric material and
short b-1–6-glucan, found previously in F. tularensis
and F. novicida [9,15]. These components seem to be
characteristic for Francisella species. Overall the simi-
larity of structural elements of LPS and other compo-
nents clearly shows that newly discovered F. victoria is
indeed a new species of Francisella genus.
There are several unresolved questions concerning
Francisella LPS biological activity. Thus the role of
the ‘starch’-like material and other glucans, coex-
tracted with LPS, is not clear. Polymers such as
these are conserved microbial structures called
‘pathogen-associated molecular patterns’, which are
ligands for pattern recognition receptors expressed
Fig. 6. MALDI mass spectrum of the b-glucan, isolated from acetic acid hydrolysis products.
Francisella victoria lipopolysaccharide W. Kay et al.
3010 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS

on various immune cells as part of the innate immu-
nity recognition system [17,18]. b-1,3 ⁄ 1,6-Glucans are
cell wall components of various bacteria, fungi, and
plants, which effect the immune response of various
vertebrates species, including fish [19]. However,
these b-glucan polymers are known to have a schizo-
phrenic activity; at low concentrations they have
been shown to be immuno-stimulatory, whereas at
higher concentrations they can be immuno-inhibitory
[20,21] and seem to modulate the release of potent
cytokines induced by LPS. Possibly the starch-like
polymer and glucans coextracted here with F. victoria
LPS are immuno-modulatory ancillary polymers. A
structural understanding of specific polymers, such as
those shown here, may shed light on how Francisella
sp. so effectively evades or suppresses the host
immune response and why, after so many years,
there are still no efficacious vaccines available.
Experimental procedures
Growth of F. victoria
F. victoria was initially isolated as the predominant Gram-
negative pathogen from the kidney of a moribund, appar-
ently wild Tilapia sp., Oreochromis niloticus. F. victoria
grew slowly at <30 °C on cysteine heart agar or broth
(CHA or CHB).
Stock cultures were held at )80 °C in 25% glycerol and
cultured at 28 °C on CHA. Three colonies were inoculated
into 0.2 mL of CHB, grown at 28 °C and sequentially
scaled up to 5 mL and 100 mL of CHB, which was then
used to inoculate a 25-L CHB fermentation broth. Fermen-

tation was carried out in CHB at 28 °C in a Chemap fer-
mentor fitted with Chemap’s Fundafom foam breaker and
a bottom-driven, three-tier, flat-bladed impeller system.
Aeration was initially adjusted to 25 LÆmin
)1
. The control
systems provide proportional control of impeller speed, air-
flow, and temperature; pH and aeration were under control.
Growth was monitored at A
600
until the culture grew to
4A
600
(48 h). Data from the fermentor was logged using
a PC and Genesis
TM
process control software. pH was con-
trolled by the addition of 5% H
2
SO
4
. For harvest and con-
centration a high capacity Millipore Pellicon
TM
tangential
flow filtration system was used and cells were finally pellet-
ed by centrifugation at 10 000 g.
PAGE of F. victoria LPS
LPS samples (1 mgÆmL
)1

) were boiled in SDS sample buf-
fer for 10 min and 50 lL used for SDS ⁄ PAGE according
to the Laemmli method as modified. SDS gels were washed
for 15 min in dH
2
O and chemically stained for LPS with
Gelcode (Pierce, Rockford, IL, USA). For western blotting,
whole cells were grown in CHB at 28 °C overnight, harves-
ted, washed once with NaCl ⁄ P
i
, resuspended in
SDS ⁄ PAGE sample buffer and 50 lL samples digested with
5 lL Proteinase K (1 mgÆmL
)1
) for 2 h at 60 °C and boiled
for 10 min to stop the digestion. For western blotting, the
prospective antigens were electrophoretically transferred to
nitrocellulose membranes for 1 h at 50 mAÆgel
)1
(Bio-Rad,
Hercules, CA, USA). These transblots were blocked using
5% w ⁄ v skimmed milk ⁄ NaCl ⁄ P
i
⁄ Tween-20 and reacted for
1 h at room temperature with a 1 : 3000 dilution of rabbit
polyclonal antisera (anti-F. tularensis LVS; anti-F. novicida;
anti-F. victoria) or mouse mAb (anti-F. tularensis LVS LPS;
anti-F. novicida LPS) in NaCl ⁄ P
i
)0.5% Tween-20. The

membranes were then washed and further reacted with a
1 : 4000 dilution of second antibody, goat antirabbit IgG
conjugated to alkaline phosphatase (Caltag Laboratories,
Burlingame, CA, USA) and developed for 4 h at room tem-
perature with 5-bromo-4-chloro-3-indolyl phosphate and 4-
nitro blue tetrazolium chloride. Francisella sp. and LPS-spe-
cific antisera were kindly provided by F. Nano, University
of Victoria, BC, Canada.
For large-scale LPS preparations F. victoria cells (300 g
wet mass) were extracted by stirring with 50% aqueous
phenol (500 mL, 70 °C, 15 min). The cooled extract was
diluted by equal amounts of water, dialyzed against tap
water until phenol-free and lyophilized. The respective resi-
dues were resuspended in 50 mL 0.02 m sodium acetate,
pH 7.0, and treated sequentially with RNase, DNase and
proteinase K (37 °C, 2 h each). Enzyme-treated samples
were subjected to ultracentrifugation (120 000 g,12h,
4 °C) and the precipitated gels were dissolved in water and
lyophilized to yield 350 mg of LPS.
Mild acid hydrolysis
Aqueous phase LPS (100 mg) was hydrolyzed with 2%
acetic acid (6 mL) at 100 °C for 4 h and, following removal
of precipitated lipid A by centrifugation at 30 000 g, the
concentrated water soluble products were fractionated by
Sephadex G-50 chromatography to yield for fractions I–IV.
Fractions II–IV were further separated by anion exchange
chromatography on Hitrap Q column (5 mL, Amersham,
Piscataway, NJ, USA) in a 3 mLÆmin
)1
. gradient of water

(first 20 min) to 1 m NaCl over 1 h with UV detection at
220 nm and sugar detection by charring aliquots from each
fraction on silica gel TLC plates after dipping in 2% H
2
SO
4
in MeOH. Thus acidic oligosaccharides 1–3 were isolated
from fraction II, b-glucan from fraction III, and oligosac-
charide 4 from fraction IV. Neutral oligosaccharides eluted
with water were not numbered.
O,N-Deacylation of the LPS
LPS (80 mg) was dissolved in 4 m KOH (4 mL) containing
NaBH
4
(50 mg), kept overnight at 100 °C, and neutralized
W. Kay et al. Francisella victoria lipopolysaccharide
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3011
with 2 m HCl. Precipitated material was removed by cen-
trifugation and the solution was applied to a Sephadex G50
column. Three oligosaccharide fractions were obtained and
were further separated by HPAEC in a 3 mLÆmin
)1
. gradi-
ent of 0.1 m NaOH (A) to 1 m sodium acetate in 0.1 m
NaOH (B), 3–50% of B, to give after desalting products
5–7 (mixture), b-glucan, and oligosaccharide 8.
The
1
H- and
13

C-NMR spectra were recorded using a
Varian Inova 800 equipped with a 5 mm
1
H observe,
13
C,
15
N decouple cold probe or 600 spectrometers in D
2
O
solutions at 25 °C and referenced to the acetone standard
(
1
H, 2.225 p.p.m.,
13
C, 31.5 p.p.m.). Varian standard pulse
sequences COSY, TOCSY (mixing time 120 ms), NOESY
or ROESY (mixing time 200 ms), HSQC, gHMBC (opti-
mized for 5 Hz coupling constant) were used. Both a stand-
ard H2BC [11] and an edited H2BC [12] were obtained at
800 MHz for
1
H and 201.12 MHz for
13
C, as described
recently [13]. GC-MS and GC were performed as des-
cribed [22].
All ESI MS and ESI MS ⁄ MS experiments were per-
formed using Q-TOF (Micromass, Manchester, UK) hybrid
quadrupole ⁄ time-of-flight instrument coupled to the Crystal

model 310 CE instrument (ATI Unicam, Boston, MA,
USA) via a coaxial sheath–flow interface for sample injec-
tion. A sheath solution (70 : 30, isopropanol–methanol)
was delivered at a flow rate of 1.5 lLÆmin
)1
to a low dead
volume tee. Injections were performed on 90 cm length bare
fused silica at an applied voltage of 30 kV with an electro-
lyte solution composed of 30 mm aqueous ammonium acet-
ate pH 8.5, containing 5% methanol for positive ion
detection. In combined MS-MS analyses, collisional activa-
tion was performed using argon collision gas at an energy
(laboratory frame of reference) of 60 eV.
MALDI MS was carried out on a Perseptive Voyager
STR model (PE Biosystem, Courtaboeuf, France) time-of-
flight mass spectrometer. Gentisic acid (2,5-dihydroxyben-
zoic acid), 10 mm in water, was purchased from Sigma
Chemical Co. (St. Louis, MO, USA) and used as matrix.
Samples (0.5 lg ⁄ 0.5 lL) were deposited on the target, cov-
ered with 0.5 lL of the matrix in aqueous solution and
dried. Analyte ions were desorbed from the matrix with
pulses from a 337 nm nitrogen laser. Spectra were obtained
in the positive ion mode at 20 kV with an average of 128
pulses. The masses are average masses.
Acknowledgements
This work was performed with support from Canadian
Bacterial Diseases Network. The NMR spectra at
800 MHz were obtained at the Varian Unity Inova
spectrometer of the Danish Instrument Center for
NMR Spectroscopy of Biological Macromolecules.

The authors are grateful to F. Nano (University of
Victoria, BC, Canada) for samples of antisera
to Francisella sp. and to J. Burian and J. Barlow (Mic-
rotek Intl. Ltd, Saanichton, BC, Canada) for the fer-
mentor growth of F. victoria.
References
1 Oyston PC, Sjostedt A & Titball RW (2004) Tularae-
mia: bioterrorism defence renews interest in Francisella
tularensis. Nat Rev Microbiol 2, 967–978.
2 Fulop M, Leslie D & Titball R (1996) A rapid, highly
sensitive method for the detection of Francisella tularen-
sis in clinical samples using the polymerase chain reac-
tion. Am J Trop Med Hyg 54, 364–366.
3 Barns SM, Grow CC, Okinaka RT, Keim P & Kuske
CR (2005) Detection of diverse new francisella-like bac-
teria in environmental samples. Appl Environ Microbiol
71, 5494–5500.
4 Johansson A, Ibrahim A, Goransson I, Eriksson U,
Gurycova D, Clarridge JE III & Sjostedt A (2000)
Evaluation of PCR-based methods for discrimination
of Francisella species and subspecies and development
of a specific PCR that distinguishes the two major
subspecies of Francisella tularensis. J Clin Microbiol
38, 4180–4185.
5 Fulop MJ, Mastroeni P, Green M & Titball RW (2001)
Role of antibody to lipopolysaccharide in protection
against low- and high-virulence strains of Francisella
tularensis. Vaccine 19, 4465–4472.
6 Isherwood KE, Titball RW, Davies DH, Felgner PL &
Morrow WJ (2005) Vaccination strategies for Francisella

tularensis. Adv Drug Deliv Rev 57, 1403–1414.
7 Sandstrom G, Sjostedt A, Johansson T, Kuoppa K &
Williams JC (1992) Immunogenicity and toxicity of lipo-
polysaccharide from Francisella tularensis LVS. FEMS
Microbiol Immunol 5, 201–210.
8 Sjostedt A (2003) Virulence determinants and protective
antigens of Francisella tularensis. Curr Opin Microbiol 6,
66–71.
9 Vinogradov E, Perry MB & Conlan JW (2002) Struc-
tural analysis of Francisella tularensis lipopolysacchar-
ide. Eur J Biochem 269, 6112–6118.
10 Silipo A, Lanzetta R, Amoresano A, Parrilli M &
Molinaro A (2002) Ammonium hydroxide hydrolysis: a
valuable support in the MALDI-TOF mass spectro-
metry analysis of lipid A fatty acid distribution. J Lipid
Res 43, 2188–2195.
11 Nyberg NT, Duus JO & Sorensen OW (2005) Hetero-
nuclear two-bond correlation: suppressing heteronuclear
three-bond or higher NMR correlations while enhancing
two-bond correlations even for vanishing 2J (CH).
J Am Chem Soc 127, 6154–6155.
12 Nyberg NT, Duus JO & Sorensen OW (2005) Editing
of H2BC NMR spectra. Magn Reson Chem 43, 971–
974.
Francisella victoria lipopolysaccharide W. Kay et al.
3012 FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS
13 Petersen BO, Vinogradov E, Kay WW, Wurtz P,
Nyberg NT, Duus JO & Sorensen OW (2005) H2 BC:
a new technique for NMR analysis of complex carbohy-
drates. Carbohydr Res 341, 550–556.

14 Vinogradov E & Perry MB (2004) Characterisation of the
core part of the lipopolysaccharide O-antigen of Franci-
sella novicida (U112). Carbohydr Res 339, 1643–1648.
15 Vinogradov E, Conlan JW, Gunn JS & Perry MB
(2004) Characterization of the lipopolysaccharide
O-antigen of Francisella novicida (U112). Carbohydr Res
339, 649–654.
16 Phillips NJ, Schilling B, McLendon MK, Apicella MA
& Gibson BW (2004) Novel modification of lipid A of
Francisella tularensis. Infect Immun 72, 5340–5348.
17 Janeway CA Jr & Medzhitov R (2002) Innate immune
recognition. Annu Rev Immunol 20, 197–216.
18 Beutler B (2004) Innate immunity: an overview. Mol
Immunol 40, 845–859.
19 Iliev DB, Liarte CQ, MacKenzie S & Goetz FW (2005)
Activation of rainbow trout (Oncorhynchus mykiss )
mononuclear phagocytes by different pathogen asso-
ciated molecular pattern (PAMP) bearing agents. Mol
Immunol 42, 1215–1223.
20 Olson EJ, Standing JE, Griego-Harper N, Hoffman OA
& Limper AH (1996) Fungal beta-glucan interacts with
vitronectin and stimulates tumor necrosis factor alpha
release from macrophages. Infect Immun 64, 3548–3554.
21 Hoffman OA, Olson EJ & Limper AH (1993) Fungal
beta-glucans modulate macrophage release of tumor
necrosis factor-alpha in response to bacterial lipopoly-
saccharide. Immunol Lett 37, 19–25.
22 St Michael F, Vinogradov E, Li J & Cox AD (2005)
Structural analysis of the lipopolysaccharide from Pas-
teurella multocida genome strain Pm70 and identification

of the putative lipopolysaccharide glycosyltransferases.
Glycobiology 15, 323–333.
FEBS Journal 273 (2006) 3002–3013 ª 2006 The Authors Journal compilation ª 2006 FEBS 3013
W. Kay et al. Francisella victoria lipopolysaccharide