Báo cáo khoa học: Novel globoside-like oligosaccharide expression patterns in nontypeable Haemophilus influenzae lipopolysaccharide pdf

The ESI-MS spectrum of LPS-OH from strain 1200 showed the same ions as 1268 except for those corresponding to HexNAcHex4 glycoforms with-out a PCho substituent Table 1.. The ion correspo

in nontypeable Haemophilus influenzae lipopolysaccharide Susanna L Lundstro¨m1, Brigitte Twelkmeyer1, Malin K Sagemark1, Jianjun Li2,

James C Richards2, Derek W Hood3, E Richard Moxon3and Elke K H Schweda1

1 Clinical Research Centre, Karolinska Institutet and University College of South Stockholm, Huddinge, Sweden

2 Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada

3 Molecular Infectious Diseases Group and Department of Paediatrics, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, UK

Haemophilus influenzae is an important cause of

human disease worldwide and exists in encapsulated

(type a–f) and unencapsulated (nontypeable) forms

Type b capsular strains are associated with invasive

bacteraemic diseases, including meningitis, epiglottitis,

cellulitis and pneumonia, whereas acapsular or non-typeable strains of H influenzae (NTHi) are primary pathogens in otitis media and cause both acute and chronic lower respiratory tract infections [1,2] The potential of H influenzae to cause disease depends

Keywords

globoside; globotetraose; Haemophilus

influenzae; lipopolysaccharide; sialyllactose

Correspondence

E Schweda, University College of South

Stockholm, Clinical Research Centre,

Novum, S-141 86 Huddinge, Sweden

Fax: +46 85 858 3820

Tel: +46 85 858 3823

E-mail: Elke.Schweda@crc.ki.se

(Received 29 May 2007, revised 18 July

2007, accepted 25 July 2007)

doi:10.1111/j.1742-4658.2007.06011.x

We report the novel pattern of lipopolysaccharide (LPS) expressed by two disease-associated nontypeable Haemophilus influenzae strains, 1268 and

1200 The strains express the common structural motifs of H influenzae; globotetraose [b-d-GalpNAc-(1fi3)-a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-b-d-Glcp] and its truncated versions globoside [a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-b-d-Glcp] and lactose [b-d-Galp-(1fi4)-[a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-b-d-Glcp] linked to the terminal heptose (HepIII) and the corresponding structures with an a-d-Glcp as the reducing sugar linked to the middle heptose (HepII) in the same LPS mole-cule Previously these motifs had been found linked only to either the proxi-mal heptose (HepI) or HepIII of the triheptosyl inner-core moiety l-a-d- Hepp-(1fi2)-[PEtnfi6]-l-a-d-Hepp-(1fi3)-l-a-d-Hepp-(1fi5)-[PPEtnfi4]-a-Kdo-(2fi6)-lipid A This novel finding was obtained by structural studies

of LPS using NMR techniques and ESI-MS on O-deacylated LPS and core oligosaccharide material, as well as electrospray ionization-multiple-step tandem mass spectrometry on permethylated dephosphorylated oligosaccha-ride material A lpsA mutant of strain 1268 expressed LPS of reduced complexity that facilitated unambiguous structural determination Using capillary electrophoresis-ESI-MS⁄ MS we identified sialylated glycoforms that included sialyllactose as an extension from HepII, this is a further novel finding for H influenzae LPS In addition, each LPS was found to carry phosphocholine and O-linked glycine Nontypeable H influenzae strain 1200 expressed identical LPS structures to 1268 with the difference that strain 1200 LPS had acetates substituting HepIII, whereas strain 1268 LPS has glycine at the same position

Abbreviations

AnKdo-ol, reduced anhydro Kdo; CE, capillary electrophoresis; Hep, L -glycero- D -manno-heptose; Hex, hexose; HexNAc, N-acetylhexosamine; Kdo, 3-deoxy- D -manno-oct-2-ulosonic acid; lipid A-OH, O-deacylated lipid A; LPS, lipopolysaccharide; LPS-OH, O-deacylated lipopolysaccharide;

MSn, multiple-step tandem mass spectrometry; Neu5Ac, N-acetyl neuraminic acid; NTHi, nontypeable Haemophilus influenzae; OS,

oligosaccharide; PCho, phosphocholine; PEtn, phosphoethanolamine; PPEtn, pyrophosphoethanolamine; tHep, terminal heptose.

upon its surface-expressed carbohydrate antigens,

cap-sular polysaccharide [3] and lipopolysaccharide (LPS)

[4] LPS is an essential and characteristic surface

com-ponent of H influenzae This bacterium has been

found to express short-chain LPS, lacking O-specific

polysaccharide chains and is often referred to as

lipo-oligosaccharide Extensive structural studies of LPS

from H influenzae by us and others have led to the

identification of a conserved glucose-substituted

trihep-tosyl inner-core moiety

l-a-d-Hepp-(1fi2)-[PEtnfi6]-l-a-d-Hepp-(1fi3)-[b-d-Glcp-(1fi4)]-l-a-d-Hepp linked

to lipid A via 3-deoxy-d-manno-oct-2-ulosonic acid

(Kdo) 4-phosphate This inner-core unit provides the

template for attachment of oligosaccharides and

non-carbohydrate substituents [5] The outer core region of

NTHi LPS mimics host glycolipids and the expression

of terminal epitopes is subject to high-frequency phase

variation, leading to a very heterogeneous population

of LPS molecules within a single strain Phase

varia-tion is thought to provide an adaptive mechanism

which is advantageous for the survival of bacteria

con-fronted by the differing microenvironments and the

immune responses of the host Several structures

mim-icking the globoside series of mammalian glycolipids

have been identified in NTHi LPS such as globotetraose

[b-d-GalpNAc-(1fi3)-a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-b-d-Glcp-(1fi], globoside

[a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-b-d-Glcp], lactose [b-d-Galp-(1fi4)-b-d-Glcp]

and sialyllactose

[a-Neu5Ac-(2fi3)-b-d-Galp-(1fi4)-b-d-Glcp(1fi] [5] Biosynthesis of these oligosaccharide

extensions has been shown to proceed in a stepwise

fashion [6] It has also been shown that H influenzae

can express sialyllacto-N-neotetraose

[a-Neu5Ac-(2fi3)-

b-d-Galp-(1fi4)-b-d-GlcpNAc-(1fi3)-b-d-Galp-(1fi4)-b-d-Glcp-(1fi] or the related structure,

(PEtnfi6)-a-d-

GalpNAc-(1fi6)-b-d-Galp-(1fi4)-b-d-GlcpNAc-(1fi3)-b-d-Galp-(1fi4)-b-d-Glcp-(1fi, both linked to

l-gly-cero-d-manno-heptose (Hep)I Biosynthesis of these

ter-minal tetrasaccharide moieties has been found to

resemble that of the O-antigen repeating unit, with the

tetrasaccharide being added en bloc [7]

Noncarbo-hydrate substituents such as pyrophosphoethanolamine

(PPEtn), phosphoethanolamine (PEtn), phosphocholine

(PCho), acetate (Ac) and glycine (Gly) are common in

NTHi LPS [5]

Our previous studies have focused on the

conserva-tion and variability of patterns of LPS expressed in a

representative set of NTHi clinical isolates obtained

from otitis media patients [8–16] and relating this to the

role of LPS in commensal and virulence behaviour

Recently, we demonstrated that oligosaccharides

containing terminal sialic acid epitopes are essential

virulence determinants in experimental otitis media [17]

In this study, we present novel LPS structures expressed in NTHi strains 1268 and 1200 The strains were previously shown to be very closely related [18] Herein, we demonstrate that the two strains, as pre-dicted, have almost identical LPS structures, the only difference being the presence of O-acetyl groups in strain 1200 Both strains were found to express LPS glycoforms containing globoside and globoside-like epitopes extending simultaneously from HepIII and HepII, respectively These LPS glycoforms have not previously been found in H influenzae In order to unambiguously establish this, we made use of a geneti-cally defined isogenic mutant strain, NTHi 1268lpsA, which had oligosaccharide extensions from HepI and HepII only The mutant strain also allowed us to iden-tify sialyllactose units substituting HepII This is the first time that sialyllactose has been detected in that molecular environment The presence of sialylated gly-coforms likely contributes to the resistance of the strain to killing by normal human serum

Results

NTHi wild-type strains 1268 and 1200 and mutant strain 1268lpsA

NTHi strains 1268 and 1200 are clinical isolates origi-nating from the Finnish Otitis Media Study Group The strains have the same ribotype, and by multilocus sequence typing had identical nucleotide sequences in three of seven LPS alleles [18] Because of the hetero-geneous mixtures of LPS glycoforms typical of wild-type NTHi strains, a lpsA mutant strain of 1268 was made to facilitate the elucidation of its structure It has previ-ously been shown that the lpsA gene is responsible for addition of a hexose (Hex) to the distal heptose (HepIII)

of the inner-core of the Hi LPS molecule [6] By disrupt-ing the lpsA gene, we sought to construct a mutant (1268lpsA) lacking any chain elongation from HepIII, but otherwise identical to the wild-type 1268 strain The two NTHi wild-type and the 1268lpsA mutant strains were grown in liquid media, the bacteria harvested and desiccated, the LPS was then isolated by extraction using the phenol⁄ chloroform ⁄ light petroleum method

Characterization of LPS from NTHi strains 1268,

1200 and 1268lpsA

In earlier investigations it was found that the LPS of NTHi strains 1268 and 1200 contained ester-linked glycine and Neu5Ac, as shown by high-performance anion-exchange chromatography with pulsed ampero-metric detection following treatment of samples with

0.1 m NaOH and neuraminidase [19,20] Furthermore,

the lipid A backbone of the respective LPS has been

described by Helander et al [21,22]

LPS from all strains was treated with anhydrous

hydrazine under mild conditions to give the

water-sol-uble O-deacylated lipopolysaccharide (LPS-OH) which

was subjected to compositional and linkage analyses as

well as ESI-MS

Compositional sugar analysis of LPS-OH from the

wild-type strain (1268) indicated d-glucose (Glc),

d-galactose (Gal), 2-amino-2-deoxy-d-glucose (GlcN),

2-amino-2-deoxy-d-galactose (GalN) and

l-glycero-d-manno-heptose (Hep) in a ratio of 32 : 28 : 21 : 9 : 10,

as identified by GLC-MS of their corresponding alditol

acetate and 2-butyl glycoside derivatives (Table S1)

Sugar analysis of LPS-OH from strain 1200 revealed

the presence of the same sugars as in 1268 in

compara-ble amounts (Tacompara-ble S1)

LPS-OH samples were dephosphorylated with 48%

hydrogen fluoride prior to methylation analysis

Mate-rial from 1268 showed terminal Glc, terminal Gal,

4-substituted Gal, 4-substituted Glc, 3-substituted Gal,

terminal Hep, 2-substituted Hep, 3,4-substituted Hep,

terminal GalN, 2,3-substituted Hep, 4-substituted

GlcN and 6-substituted GlcN in the relative amounts

of 16 : 4 : 7 : 14 : 3 : 11 : 6 : 14 : 2 : 19 : 2 : 2

Meth-ylation analysis on dephosphorylated LPS-OH from

NTHi strain 1200 revealed the presence of the same

sugars as 1268 in comparable amounts (Table S2) The

methylation analysis data were consistent with

trian-tennary structures in NTHi 1268 and 1200, containing

the common inner-core element,

l-a-d-Hepp-(1fi2)-l

-a-d-Hepp-(1fi3)-[b-d-Glcp-(1fi4)]-l-a-d-Hepp-(1fi5)-a-Kdop of H influenzae LPS

The ESI-MS spectrum of LPS-OH from 1268 revealed

abundant molecular peaks corresponding to triply and

quadruply deprotonated ions (Table 1) The MS data

indicated the presence of heterogeneous mixtures of

glycoforms, consistent with each molecular species

con-taining the conserved PEtn substituted triheptosyl

inner-core moiety attached via a phosphorylated Kdo

linked to the O-deacylated lipid A (lipid A-OH) As a

characteristic feature, populations of glycoforms were

observed that differed by 123 Da (i.e a PEtn group),

consistent with either phosphate or PPEtn substitution

at the O-4 position of the Kdo residue [23–25] For

clar-ity, glycoforms containing five Hex with no

N-acetyl-hexosamine (HexNAc) residue are referred to as Hex5

glycoforms Glycoforms containing five Hex including a

HexNAc residue are referred to as HexNAcHex4

glyco-forms In the ESI-MS spectrum (negative mode) major

quadruply charged ions were observed at m⁄ z 609.4 and

640.2 corresponding to glycoforms with respective

com-positions PChoÆHex2ÆHep3ÆPEtnÆPÆKdoÆlipid A-OH and PChoÆHex2ÆHep3ÆPEtn2ÆPÆKdoÆlipid A-OH Ions corresponding to HexNAc containing glycoforms with respective compositions PChoÆHexNAcÆHex4ÆHep3Æ PEtnÆPÆKdoÆlipid A-OH and PChoÆHexNAcÆHex4ÆHep3Æ PEtn2ÆPÆKdoÆlipid A-OH, and PChoÆHexNAcÆHex5Æ Hep3ÆPEtnÆPÆKdoÆlipid A-OH and PChoÆHexNAcÆHex5Æ Hep3ÆPEtn2ÆPÆKdoÆlipid A-OH were detected at m⁄ z 741.3⁄ 772.1 and 781.8 ⁄ 812.8 Furthermore, glycoforms with compositions PChoÆHex5ÆHep3ÆPEtnÆPÆKdoÆlipid A-OH and PChoÆHex5ÆHep3ÆPEtn2ÆPÆKdoÆlipid A-OH were indicated at m⁄ z 731.1 and 761.7, respectively Peaks of low intensity corresponding to a minor Hex-NAcHex4 glycoform without a PCho substituent were also identified The ESI-MS spectrum of LPS-OH from strain 1200 showed the same ions as 1268 except for those corresponding to HexNAcHex4 glycoforms with-out a PCho substituent (Table 1)

ESI-MS data of LPS-OH from 1268lpsA showed less heterogeneity with no indications of Hex5 or Hex5 glycoforms Ions corresponding to the HexNAc-Hex4 glycoforms lacking PCho were moderately higher

in abundance in 1268lpsA than in 1268 (Table 1) Ions corresponding to sialylated glycoforms were not unambiguously identified in the full ESI-MS spec-tra of LPS-OH samples due to extensive overlap with those corresponding to major, nonsialylated glyco-forms, and⁄ or low abundance However, their presence was confirmed for LPS-OH of 1268lpsA in precursor ion monitoring tandem mass spectrometry experiments

by scanning for loss of m⁄ z 290 (Neu5Ac, negative ion mode) or m⁄ z 274 (Neu5Ac-H2O, positive mode) fol-lowing capillary electrophoresis (CE)-ESI-MS⁄ MS The data are shown in Fig 1 and summarized in Table S3 In the precursor negative-mode ion-scan spectrum (Fig 1A) quadruply and triply charged ions corresponding to a complex mixture of sialylated gly-coforms containing three to six hexose residues were observed The major ion at m⁄ z 909.5 corresponded to

a Hex3 glycoform with the composition Neu5AcÆHex3Æ Hep3ÆPEtnÆPÆKdoÆlipid A-OH Particularly noteworthy are HexNAc-containing glycoforms detected at m⁄ z 1086.0, 1127.5, 1180.5, 1207.0 and 1249.5 having the respective compositions, PChoÆNeu5AcÆHexNAcÆHex4Æ Hep3ÆPEtnÆPÆKdoÆlipid A-OH, PChoÆNeu5AcÆHexNAcÆ Hex4ÆHep3ÆPEtn2ÆPÆKdoÆlipid A-OH, PChoÆNeu5AcÆ HexNAcÆHex5ÆHep3ÆPEtn2ÆPÆKdoÆlipid A-OH, PChoÆ Neu5AcÆHexNAc2ÆHex5ÆHep3ÆPEtnÆPÆKdoÆlipid A-OH and PChoÆNeu5AcÆHexNAc2ÆHex5ÆHep3ÆPEtn2ÆPÆKdoÆ lipid A-OH In the precursor ion scan spectrum obtained in the positive mode (Fig 1B) ions corre-sponding to sialylated Hex3 glycoforms were not observed, whereas ions corresponding to sialylated

glycoforms containing HexNAc were readily detected

confirming their presence

Characterization of oligosaccharides derived from

NTHi strains 1268, 1200 and 1268lpsA

Mild acid hydrolysis of LPS with dilute aqueous acetic

acid afforded insoluble lipid A and core oligosaccharide

material (OS), which after purification by gel-filtration

chromatography resulted in OS samples from the vari-ous strains Strains 1268 and 1200 gave leading fractions

of higher molecular mass referred to as 1268OS and 1200OS which were investigated in detail Strain 1268lpsA gave lpsAOS

Sugar analyses (Table S1) performed on 1268OS, lpsAOS and 1200OS were consistent with the data obtained on LPS-OH for both the wild-type and mutant strains, revealing the presence of Glc, Gal,

Table 1 Negative ion ESI-MS data and proposed compositions for LPS-OH and OS of strains 1268, 1200 and 1268lpsA Average mass units were used to calculate molecular mass based on proposed compositions as follows: Hex, 162.14; HexNAc, 203.19; Hep, 192.17; Kdo, 220.18; P, 79.98; PEtn, 123.05; PCho, 165.13; AnKdo-ol, 222.20; Gly, 57.05; Ac, 42.04; lipid A-OH, 953.02 All LPS-OH and OS glycoforms contain Hep3ÆPEtnÆPÆKdoÆlipid A-OH or Hep3ÆPEtnÆAnKdo-ol, respectively nr, not rationalized.

Sample

Observed ions (m ⁄ z) Molecular mass (Da) Relative Abundance (%)

Proposed composition

Hep, GalN and GlcN The considerable decrease in

GlcN in the OS samples confirmed this sugar to be

part of lipid A but also indicated traces of glycoforms

that contain GlcN

1268OS, lpsAOS and 1200OS were dephosphorylated

with 48% hydrogen fluoride prior to methylation

anal-ysis Methylation analysis (Table S2) on the resulting

material from 1268OS showed terminal Glc, terminal

Gal, 4-substituted Gal, 4-substituted Glc, 3-substituted

Gal, terminal Hep, 2-substituted Hep, 3,4-substituted

Hep, terminal GalN, 2,3-substituted Hep and

4-substi-tuted GlcN Methylation analysis performed on

nonde-phosphorylated 1268OS revealed the same sugars but

with a decrease in terminal Glc, 4-substituted Glc and

2,3-substituted Hep, which indicated phosphorylation

on those sugars (data not shown) Methylation

analy-sis on dephosphorylated lpsAOS gave the same sugar

derivatives as 1268OS but showing a significant

increase in terminal Hep and the absence of

2-substi-tuted Hep Moreover, a decrease of 4-substi2-substi-tuted Gal,

4-substituted Glc and 3-substituted Gal was observed

in the methylation analysis of this OS sample

Methyl-ation analysis data for 1200OS was comparable with

the data obtained for 1268OS (Table S2)

ESI-MS on OS samples (Table 1) indicated all

strains to be glycylated In addition, OS samples from

NTHi 1200 showed ions corresponding to acetylated

glycoforms ESI-MS on 1268OS and 1200OS revealed

major HexNAcHex4 and HexNAcHex5 glycoforms

Lower molecular mass glycoforms were minor, in

agreement with OS samples being leading fractions

after GPC Glycoforms, of which the O-glycylated ones were of minor abundance, were evidenced as doubly negatively charged ions as follows: PChoÆ Hex2ÆHep3ÆPEtnÆAnKdo-ol and PChoÆGlyÆHex2ÆHep3Æ PEtnÆAnKdo-ol (m⁄ z 704.3 ⁄ 732.9), PChoÆHex5ÆHep3Æ PEtnÆAnKdo-ol and PChoÆGlyÆHex5ÆHep3ÆPEtnÆ AnKdo-ol (m⁄ z 947.6⁄ 976.0), PChoÆHexNAcÆHex4Æ Hep3ÆPEtnÆAnKdo-ol and PChoÆGlyÆHexNAcÆHex4Æ Hep3ÆPEtnÆAnKdo-ol (m⁄ z 968.1 ⁄ 996.3), and PChoÆ HexNAcÆHex5ÆHep3ÆPEtnÆAnKdo-ol and PChoÆGlyÆ HexNAcÆHex5ÆHep3ÆPEtnÆAnKdo-ol (m⁄ z 1049.2⁄ 1077.5) In addition, ions at m⁄ z 866.5 ⁄ 895.0 indicated the glycoforms PChoÆHex4ÆHep3ÆPEtnÆAnKdo-ol and PChoÆGlyÆHex4ÆHep3ÆPEtnÆAnKdo-ol The glycoforms indicated in lpsAOS were in agreement with those found in the equivalent LPS-OH and showed major Hex2 glycoforms

ESI-MS data of 1200OS revealed the presence of glycoforms substituted by up to two acetate groups Information on the location of Ac was provided by ESI multiple-step tandem mass spectrometry (MSn) in the positive-ion mode The product ion spectrum obtained from the molecular ion at m⁄ z 1496.4 (composition: PChoÆAc2ÆHex2ÆHep3ÆPEtnÆAnKdo-ol) (Fig 2A) con-tained, inter alia, the ion at m⁄ z 919.1 resulting from the loss of Hex-HepI-AnKdo-ol MS3performed on this ion revealed a prominent ion at m⁄ z 643.3 (composition: PChoÆHexÆHepIIÆPEtn) (Fig 2B) resulting from the loss

of a diacetylated heptose subunit indicative of HepIII being substituted with two acetates These experiments also confirmed that PCho substituted the hexose linked

Fig 1 CE-ESI-MS ⁄ MS spectra of LPS-OH derived from NTHi 1268lpsA The indicated compositions include the PÆKdoÆlipid A-OH element (A) Precursor ion spectrum (nega-tive mode) using m ⁄ z 290 as the fragment ion for identification of sialylated compo-nents in 1268lpsA (B) Precursor ion spec-trum (positive mode) using m ⁄ z 274 as the fragment ion for identification of sialylated components in 1268lpsA.

to HepII and that PEtn substituted HepII MS3

experi-ments on ions corresponding to monoacetylated

glycoforms revealed the same substitution pattern (data

not shown)

Sequence analysis on dephosphorylated and

permethylated oligosaccharide samples using

ESI-MSn

Sequence and branching details of the various

glyco-forms in 1268OS, lpsAOS and 1200OS were obtained

using ESI-MSn in the positive mode on dephosphoryl-ated and permethyldephosphoryl-ated material [13,26] Because of the increased MS response obtained by permethylation

in combination with added sodium acetate, several gly-coforms were observed in the MS spectra that were not detected in underivatized samples Thus, the

ESI-MS mass spectrum of 1268OS (positive mode) (Fig 3A) showed sodiated singly charged adduct ions ([M+Na]+) corresponding to the glycoforms Hex2Æ Hep3ÆAnKdo-ol, Hex3ÆHep3ÆAnKdo-ol, Hex4ÆHep3Æ AnKdo-ol and Hex5ÆHep3ÆAnKdo-ol (m⁄ z 1467.9, 1672.4, 1875.8 and 2080.1), HexNAcÆHex4ÆHep3Æ AnKdo-ol and HexNAcÆHex5ÆHep3ÆAnKdo-ol (m⁄ z 2120.8 and 2325.3) and HexNAc2ÆHex4ÆHep3ÆAnKdo-ol (m⁄ z 2366.7) The HexNAc2Hex4 glycoform was not detected in the underivatized samples due to low abundance

In order to obtain sequence and branching informa-tion, these molecular ions were further fragmented in

MS2 and MS3 experiments For most glycoforms the presence of several isomeric compounds was revealed

by identifying product ions in MS2 spectra (Table S4)

MS3 experiments were used when necessary to confirm structures

Two isomeric Hex2 glycoforms were identified in 1268OS by fragmenting the molecular ion m⁄ z 1467.9 The resulting spectrum revealed ions at m⁄ z 1206.1 (major) and 1002.0 (minor) corresponding to loss of terminal (t)Hep and tHex-Hep The ion at m⁄ z 754.3 corresponded to the fragment tHex-HepI-AnKdo-ol Thus in the major Hex2 isomer terminal hexoses substituted both HepI and HepII In the minor Hex2 glycoform both HepI and HepIII were substituted with terminal hexose residues Performing MS2 on the ion m⁄ z 1672.4 and subsequent MS3 on the resulting

Fig 3 ESI-MS n analysis of permethylated

OS of strain 1268 (A) Full-scan spectrum

(positive mode) on permethylated

dephos-phorylated 1268OS (B) Product ion

spectrum of [M+Na] + m ⁄ z 2120.8

corre-sponding to the HexNAcHex4 glycoform.

Proposed key fragments are indicated in the

structure (C) MS 3 of the ion at m ⁄ z 1859.0

from MS2of m ⁄ z 2120.8 Proposed key

fragments are indicated in the structure.

Fig 2 ESI-MSn analysis of OS derived from NTHi strain 1200.

(A) Product ion spectrum of [M+H] + m ⁄ z 1496.4 corresponding to

the PChoÆAc2ÆHex2ÆHep3ÆPEtnÆAnKdo-ol glycoform The proposed

fragmentation is shown beside the spectrum (B) MS 3 on fragment

ion m ⁄ z 919.1, corresponding to the loss of Hex-HepI-AnKdo-ol.

The proposed fragmentation is shown beside the spectrum.

product ions determined two major and two minor

Hex3 isomeric glycoforms The ion corresponding to

the loss of tHepIII at m⁄ z 1409.9 was further

frag-mented to give the ion at m⁄ z 753.5 due to the loss of

a tHex-Hex-HepII unit, thus evidencing a structure

with one hexose residue substituted to HepI and a

disaccharide moiety substituting HepII Furthermore,

in the same MS3 spectrum a minor ion was detected at

m⁄ z 957.3 corresponding to the loss of tHex-HepII

This ion confirmed the structure of a glycoform in

which a disaccharide moiety substitutes HepI and one

hexose substitutes HepII The ion at m⁄ z 1206.0

corre-sponded to the loss of tHex-HepIII from the molecular

ion It was further fragmented to give the ion at m⁄ z

753.4 which originated from the loss of a tHex-HepII

unit, thus indicating a structure with one hexose

resi-due substituting each heptose resiresi-due Finally, a minor

structure with one hexose residue linked to HepI and

two hexoses linked to HepIII could be determined

when the product ion, m⁄ z 1001.5, corresponding to

the loss of tHex-Hex-HepIII, was further fragmented

to give the ion at m⁄ z 753.2 due to the loss of HepII

Four isomeric Hex4 glycoforms were identified by

per-forming MS2on the parent ion at m⁄ z 1875.8 and

sub-sequent MS3 on the resulting product ions at m⁄ z

1614.0, 1206.1 and 1001.7 due to the losses of a

termi-nal heptose, a termitermi-nal Hex-Hex-Hep residue and a

terminal Hex-Hex-Hex-Hep residue, respectively

When the ion at m⁄ z 1614.0 was further fragmented in

a MS3 experiment it gave product ions at m⁄ z 883.5

and 753.4 due to losses of the epitopes

tHex-HepI-AnKdo-ol and tHex-Hex-Hex-HepII, respectively

Fur-thermore, a product ion at m⁄ z 1161.2 indicated the

loss of tHex-HepII Thus two structures with terminal

HepIII were identified: the first with one hexose linked

to HepI and a trisaccharide group linked to HepII and

the second containing elongation of a trisaccharide

group substituting HepI and one hexose on HepII

When the ion at m⁄ z 1206.1 was further fragmented in

MS3 experiments a product ion at m⁄ z 753.4 was

observed defining the loss of tHex-HepII, which

indi-cated a major glycoform containing a disaccharide unit

on HepIII and one hexose residue on each of HepI

and HepII The product ion at m⁄ z 1001.7 was further

fragmented to give the ion at m⁄ z 753.3 due to the loss

of HepII, revealing a minor glycoform containing a

tri-saccharide unit on HepIII and with one hexose on

HepI One isomeric Hex5 glycoform was observed by

fragmenting the molecular ion at m⁄ z 2080.1 The

iso-mer was defined by the ions at m⁄ z 1349.9 and 1206.2

corresponding to the loss of tHex-HepI-AnKdo-ol and

tHex-Hex-Hex-HepIII which indicated HepI to be

substituted by one hexose and HepIII to be elongated

by three hexoses The structure was confirmed in MS3 experiments on m⁄ z 1206.2 where the product ion at

m⁄ z 754.4 indicated the loss of tHex-HepII

The molecular ion at m⁄ z 2120.8 corresponded to a glycoform with four hexoses and one hexosamine One single isomer (Fig 3B,C) was identified by fragmenting the molecular ion In the resulting spectrum, fragment ions were observed at m⁄ z 1862.4, 1859.0 and 1390.7 resulting from the loss of tHexNAc, tHep and tHex-HepI-AnKdo-ol, respectively A MS3 experiment on

m⁄ z 1859.0, showing the loss of tHex-HepI-AnKdo-ol (m⁄ z 1129.4) confirmed that this glycoform contained

a tHexNAc-Hex-Hex-Hex elongation from HepII and

a single hexose substituting HepI The molecular ion

at m⁄ z 2325.3 corresponded to a HexNAcHex5 glyco-form When this ion was further fragmented it gave ions at m⁄ z 2065.5, 1594.8 and 1205.9 resulting from the loss of tHexNAc, HepI-AnKdo-ol and tHex-NAc-Hex-Hex-Hex-Hep, respectively The ion at m⁄ z 1205.9 was further fragmented to give the ion at m⁄ z 754.1 due to the loss of a tHex-HepII unit, thus evi-dencing a structure with one hexose residue substituted

to each of HepI and HepII, and a tetrasaccharide moi-ety with terminal hexosamine substituting HepIII The molecular ion at m⁄ z 2366.7 corresponded to

a glycoform with the composition HexNAc2ÆHex4Æ Hep3ÆAnKdo-ol The single isomer of this glycoform was defined in the MS2 spectrum by the ions at m⁄ z 2108.0, 2105.0, 1903.5 and 1658.3 corresponding to the loss of tHexNAc, tHep, Hex and tHexNAc-Hex-HexNAc MS3 performed on the ion at m⁄ z 1658.3 indicated the loss of tHep (m⁄ z 1396.1) and the fragment ion of -Hex-Hex-HepI-AnKdo-ol (m⁄ z 944.1) Thus, this glycoform contained a tHexNAc-Hex-HexNAc-Hex-Hex- unit elongating from HepI and with one hexose substituting HepII

ESI-MSn data obtained from lpsAOS clearly indi-cated the absence of glycoforms expressing chain extension from HepIII The major isoforms observed were otherwise equivalent to those found in the wild-type strain, except for an extra Hex1 glycoform (m⁄ z 1264.1) containing one hexose substituent on HepI determined from the fragment ion at m⁄ z 753.2 indi-cating the loss of tHepIII-HepII (Table S4)

Strain 1200 contained virtually the same glycoforms

as observed in strain 1268 except for those having elongations from HepI However, traces of three other higher molecular mass forms; HexNAcÆHex6ÆHep3Æ AnKdo-ol, HexNAcÆHex7ÆHep3ÆAnKdo-ol and Hex-NAc2ÆHex7ÆHep3ÆAnKdo-ol at m⁄ z 2528.9, 2732.9 and 2976.9, respectively, were observed and investigated

MS2 of m⁄ z 2528.9 gave fragment ions at m ⁄ z 2269.2, 1410.0 and 1799.5 corresponding to losses of

tHexNAc, tHexNAc-Hex-Hex-Hex-HepIII and

tHex-HepI-AnKdo-ol, indicating HepI to be substituted by

one hexose, HepII by two hexoses and HepIII by

a tHexNAc-Hex-Hex-Hex unit The same spectrum

showed a second glycoform defined by the ion at m⁄ z

1859.7 indicating that HepIII was elongated with two

hexoses This was confirmed by MS3 on the ion at

m⁄ z 1799.5 giving the fragment ion at m ⁄ z 1128.9

corresponding to the loss of tHex-Hex-HepIII, thus

indicating HepII to be substituted by a

tHexNAc-Hex-Hex-Hex- unit MS2 of m⁄ z 2732.9 gave fragment ions

at m⁄ z 2474.4, 1613.7 and 2002.9 corresponding to the

losses of tHexNAc, tHexNAc-Hex-Hex-Hex-HepIII

and tHex-HepI-AnKdo-ol, respectively, revealing HepI

to be substituted by one hexose, HepII by three

hexoses and HepIII by the tHexNAc-Hex-Hex-Hex

unit MS2 of m⁄ z 2976.9 gave the fragment ions m ⁄ z

2718.4 and 1858.5, indicating the loss of tHexNAc and

tHexNAc-Hex-Hex-Hex-HepIII The ion at m⁄ z 1858.5

was further fragmented which gave the fragment ions

at m⁄ z 1599.7 and 1129.0 corresponding to the losses

of tHexNAc and tHex-HepI-AnKdo-ol This indicated

the HexNAc2Hex7 isomer to be substituted by one

hexose at HepI and tHexNAc-Hex-Hex-Hex units

substituting both HepII and HepIII

Characterization of lpsAOS, 1268OS and 1200OS

by NMR

Major structures were elucidated by detailed 1H, 13C

and31P NMR analyses 1H and 13C NMR resonances

were assigned using gradient chemical shift correlation

techniques (COSY, TOCSY and HMQC experiments)

The chemical shift data corresponding to 1268OS,

lpsAOS and 1200OS are given in Table 2 Prior to

NMR analyses the samples were treated with 1 m NH3

to remove O-acyl groups Subspectra corresponding to

the individual glycosyl residues were identified on the

basis of spin-connectivity pathways delineated in the

1H chemical shift correlation maps, the chemical shift

values, and the vicinal coupling constants The

mono-saccharide sequences of the major glycoforms were

confirmed from transglycosidic NOE connectivities

between anomeric and aglyconic protons on adjacent

residues (Table S5) The chemical shift data are

consis-tent with each sugar residue being present in the

pyr-anosyl ring form Further evidence for this conclusion

was obtained from NOE data which also served to

confirm the anomeric configurations of the linkages

and the monosaccharide sequence NOESY spectra of

1268OS, lpsAOS and 1200OS revealed inter-residue

NOE connectivities between the anomeric protons of

HepIII to HepII H-1⁄ H-2, HepII to HepI H-3, HepI

to Kdo H-5⁄ H-7 and GlcIV to HepI H-4 ⁄ H-6, which confirmed the sequence of the conserved triheptosyl inner core unit Several signals for methylene protons

of AnKdo-ol were observed in the COSY and TOCSY spectra in the region d 1.87–2.18 This is due to the fact that several anhydro-forms of Kdo are formed during the hydrolysis by elimination of phosphate or pyrophosphoethanolamine from the C-4 position [27]

1H–31P correlation experiments indicated PEtn (dP0.01) to be linked to O-6 of HepII

Structure of the Hex2, Hex4 and HexNAcHex4 glycoforms inlpsAOS

Sequence analysis of lpsAOS by ESI-MSn revealed a predominant Hex2 glycoform having a triheptosyl inner-core from which chain elongation by hexoses only appeared from HepI and HepII (Table S4) In addition, glycoforms having further extensions from HepII by HexNAc-Hex-Hex-Hex or truncated versions thereof were detected In the 1H NMR spectrum of lpsAOS, anomeric resonances corresponding to the triheptosyl moiety (HepI–HepIII) were identified at

d 5.05–5.16, 5.83 and 5.03, respectively Subspectra cor-responding to the hexose residues were identified in the 2D COSY and TOCSY (Fig 4A) spectra at d 5.28 (Glc residue V), 4.97 (Gal residue VII), 4.92 (Gal resi-due VII), 4.66 (GalNAc resiresi-due VIII), 4.57⁄ 4.64 (Gal residue VI) and 4.54 (Glc residue IV), respectively The chemical shift data were consistent with VII (dH-14.97) and VIII being terminal residues The terminal and 4-substituted forms of residue V could be distinguished

by different H-2 and H-4 shifts (dH-23.54⁄ 3.59 and

dH-43.50⁄ 3.80), which was also confirmed in COSY and 1H)13C HMQC experiments (dC-469.8⁄ 76.3) The high H-6A ⁄ B shifts of V (d 4.11⁄ 4.18) indicated this position to be substituted with a PCho subunit, which was confirmed in 1H–31P correlation experiments showing a 31P resonance at d)0.05 correlating to H-6A⁄ B of V and the methylene protons of PCho at

d 4.35 The spin systems at d 4.57 and 4.64 could both

be assigned to residue VI indicating the anomeric proton of this residue to be sensitive to changes in molecular environment due to the microheterogeneity

of the sample Because the oligosaccharide contains Hex4 glycoforms with and without PCho, we assume that the proton at d 4.57 corresponds to glycoforms substituted by PCho and the one at d 4.64 to those that do not Inter-residue NOE between the proton pairs of V H-1⁄ II H-3 confirmed these residues to be linked to position O-3 in HepII Inter-residue NOE were observed between the VII H-1⁄ VI H-4 confirming

a a-d-Galp-(1fi4)-b-d-Galp unit within the extension from HepII Inter-residue NOE from VI H-1 and

Table 2 1 H and 13 C chemical shifts for 1268OS, lpsAOS and 1200OS Prior to NMR analyses the samples were O-deacylated Spectra were recorded in D2O at 25 C Chemical shift values compared between the three strains could vary by up to ± 0.01 p.p.m Signals originating from the Hex5 and HexNAcHex5 glycoforms were not observed in the lpsA mutant Signals corresponding to PCho methyl protons and car-bons occurred at d 3.23 and 54.7, respectively Pairs of deoxy protons of reduced AnKdo-ol were identified in COSY and TOCSY spectra at d 1.87–2.18 Signals corresponding to GalNAc methyl 1 H and 13 C occurred at d 2.05 and 23.02, respectively.

a Observed from intense NOE signals b –, not determined c Observed in TOCSY of strain 1268 only d Observed as intra-residue NOE from H-1 of d 4.64 only e An extra terminal b-hexose was observed in strain 1200 and 1268 (weak) at dH-1,C-14.46, 102.7; dH-2,C-23.54,72.9;

d H-3,C-3 3.69,72.7; d H-4 3.54 and d H-5 3.74, respectively Also, intra-residue NOE signals from the anomeric proton to H-3 and H-5 were observed No inter-residue NOE connections could be detected.

V H-3 were not observed, probably due to low

abun-dance of the corresponding glycoforms However,

NMR data combined with data from methylation and

tandem MS analysis corroborate the sequence of the

extending glycose unit from HepII as [b-d-GalNAcp-(1fi3)-a-d-Galp-(1fi4)-b-d-Galp-(1fi4)-a-d-Glcp-(1fi] The HexNAcHex4 glycoform in lpsAOS is shown in Fig 5 Also indicated are the truncated Hex4 and Hex2 glycoforms

Structure of the Hex5 and HexNAcHex5 glycoforms in 1268OS and 1200OS

Sequence analysis of 1268OS by ESI-MSn revealed in addition to HexNAcHex4 glycoforms, abundant Hex-NAcHex5 glycoforms having the structure observed for lpsAOS and also those with chain elongation from HepIII (Table S4) In the 1H NMR spectrum of 1268OS (Fig S1A), anomeric resonances correspond-ing to the triheptosyl moiety (HepI–HepIII) were iden-tified at d 5.05–5.16, 5.71 and 5.13, respectively Spin systems corresponding to the hexose residues were identified in the COSY and TOCSY spectra The occurrence of inter-residue NOESY connectivities between protons on contiguous residues in 1268OS confirmed an identical structural element as shown in Fig 5 In addition, in the COSY and TOCSY spectra

of 1268OS anomeric signals at d 4.43 and 4.52 could

be attributed to fi4)-b-d-Glcp (IX) and fi4)-b-d-Galp (X) residues, respectively Additional spin systems cor-responding to terminal GalNAc and Gal residues indi-cated by methylation analysis were not observed It was reasonable to assume that these overlapped with the resonances of the corresponding sugars extending from HepII Thus resonances at d 4.92 and 4.66 were assigned to correspond to residues XI and XII, respec-tively Inter-residue NOE between X H-1⁄ IX H-4 and IX H-1⁄ III H-1,2 (Fig 6A) gave evidence for the fi4)-b-d-Galp-(1fi4)-b-d-Glcp-(1fi2)-l-a-d-HepIIIp-(1fi unit Because inter-residue NOE between XII H-1⁄

XI H-3 and XI H-1⁄ X H-4 was observed we propose that a globotetraose unit is the full extension from HepIII in 1268OS The HexNAcHex5 glycoform in 1268OS and 1200OS is shown in Fig 7 as well as the truncated Hex5 glycoform

Fig 4 Selected region of phase sensitive TOCSY spectra (mixing

time 180 ms) of (A) lpsAOS and (B) 1200OS Cross-peaks of

impor-tance are labelled See Table 2 for an explanation of the roman

numerals (A) Signals corresponding to structures with full

exten-sion from HepII (Fig 5) are indicated (B) Signals corresponding to

structures with full extension from HepIII (Fig 7) are indicated.

IV

β-D-Glcp-(1→4)-L-α-D-HepIp-(1→5)-AnKdo-ol

PCho 3

Hex4 Hex2 ↓ ↑

6 1

β-D-GalNAcp-(1→3)-α-D-Galp-(1→4)-β-D-Galp-(1→4)- α-D-Glcp-(1→3)-L-α-D-HepIIp6 ←PEtn

2

VIII VII VI V ↑

1

L-α-D-HepIIIp

Fig 5 Structure proposed for the HexNAcHex4 glycoform in lpsAOS Also indicated are the truncated Hex4 and Hex2 glycoforms.