 |
INTRODUCTION |
Haemophilus influenzae frequently colonizes the human
nasopharynx. Up to 80% of the population harbor this organism as part of their normal flora (1). Although normally an innocuous inhabitant of
the upper respiratory tract, H. influenzae is an
opportunistic pathogen. The diseases caused by the organism can be
ordered in two groups based on the presence or absence of a capsule.
Encapsulated or typeable organisms, which range from capsule types
a-f, can cause systemic infections such as bacteremia, septicemia, and bacterial meningitis (1, 2). Of the various encapsulated types,
H. influenzae type b has been associated most often
with pathogenesis (2). The nonencapsulated or nontypeable
(NTHi)1 strains of H. influenzae cause more localized infections, such as chronic
bronchitis or otitis media, and rarely cause systemic infections (3,
4).
There are a number of virulence factors associated with both H. influenzae type b and NTHi that contribute to their pathogenesis, one of these being the lipooligosaccharide (LOS) (5-9). LOS is a
complex glycolipid containing three main regions: lipid A, core, and a
variable branched region (10). The core region is a conserved structure
containing a phosphorylated Kdo residue linked to three heptose
residues, whereas the variable branched region contains a heterogeneous
mix of hexoses and N-acetylhexosamines as well as other
factors, such as phosphoethanolamine (PEA), phosphorylcholine, and
NeuAc (11-16). LOS differs from its enterobacterial counterpart, lipopolysaccharide, in that the variable branched region or O-antigen is a nonrepeating unit (10). A great deal of work has been undertaken to understand the biosynthesis and role of LOS in pathogenesis (10).
H. influenzae LOS is very heterogeneous and contains a
number of phase-varying epitopes (17, 18). Phase variation is known at
least in part to occur through a process of slipped-strand mispairing
(19). Three well characterized loci involved in LOS expression and
phase variation, designated lic1, lic2, and
lic3, phase-vary through this mechanism (20-22). Phase
variation may play a role for the bacterium in the evasion of the host
immune response. LOS structures have also been found to mimic human
blood group antigens, such as the Pk antigen and
paragloboside (23). This may be another method for bacterial immune evasion.
The lsg (lipooligosaccharide
synthesis genes) locus is another region
involved in LOS biosynthesis (24). Seven genes are in the locus, six of
which have identity to various glycosyltransferases and one that acts
as a regulator (25). This locus is not controlled by the slipped strand
mispairing mechanism. Through studies expressing chimeric
Haemophilus structures in Escherichia coli
lipopolysaccharide, we know that this locus is involved in the
expression of a terminal N-acetyllactosamine structure (25).
One of the genes in this locus, lsgB, has homology (27%
identity, 46% similarity) to the sialyltransferase in Neisseria
meningitidis. In various Neisseria and
Haemophilus species, a terminal
N-acetyllactosamine structure has been shown to be an
acceptor for sialylation (23, 26, 27).
NeuAc is a constituent of the LOS in about half of the H. influenzae strains tested (23, 28). Sialylation in H. influenzae has been shown to affect its ability to evade the lytic
effects of human serum (28, 29). Two genes have been identified that are involved in LOS sialylation, siaB and lic3A
(28, 29). siaB is a CMP-NeuAc synthetase, and a mutation in
this gene eliminates all sialylation (28). The second gene,
lic3A, has been shown to function as an
2-3-sialyltransferase, responsible for sialylating terminal lactose
structures. The lic3A gene has about 40% identity to
cstII from Campylobacter jejuni (29). This gene
is one of two sialyltransferases identified in this organism (30). A
mutation in lic3A in one strain of H. influenzae
still contained sialylated glycoforms, indicating the possibility of a
second sialyltransferase in this organism (29). H. influenzae contains a homologue to a sialyltransferase from
Haemophilus ducreyi, which is designated as HI0871 in the
H. influenzae Rd genome data base (31). In a study looking
at a number of genes from H. influenzae and their possible
role in LOS biosynthesis, no function for this gene (designated orfY) was found (32). This gene, which we call
siaA, was studied for its possible role in LOS sialylation.
We report evidence that siaA is a sialyltransferase in
H. influenzae and that lsgB is required for the
biosynthesis of a third, distinct sialylated glycoform, and our
evidence strongly suggests that LsgB is the third sialyltransferase.
| |
EXPERIMENTAL PROCEDURES |
Bacterial Strains and Growth Conditions--
All bacterial
strains and plasmids used in this study are listed in Table
I. Parental strains 2019, A2, and their
derivatives were grown on brain heart infusion agar (Difco)
supplemented with 10 µg/ml 
-nicotinamide adenine dinucleotide
(Sigma) and 10 µg/ml hemin (ICN Biochemicals) at 37 °C. When
appropriate, 15 µg/ml ribostamycin (Sigma) (a kanamycin analogue), 1 µg/ml chloramphenicol, 15 µg/ml spectinomycin, and 20 µg/ml NeuAc
(Sigma) were added to the media.
DNA Isolation and Manipulation--
Chromosomal DNA was isolated
using standard protocols. Restriction enzymes were purchased from
either New England Biolabs or Promega. Polymerase chain reactions
(PCRs) were performed with either Taq DNA Polymerase (Roche
Molecular Biochemicals) or the Expand Long Template kit (Roche
Molecular Biochemicals). All plasmid constructs were maintained in
either E. coli DH5 or DH10B (Invitrogen). Gel
purification was performed with SeaPlaque GTG-agarose (BioWittaker Molecular Applications) using standard protocols.
Southern Hybridization--
DNA was digested to completion with
the appropriate restriction enzymes, fractionated in 0.7% agarose
gels, and transferred to Hybond-N nylon membranes (Amersham
Biosciences). Southern blots were hybridized with probes generated by
the random primed digoxygenin DNA labeling kit (Roche Molecular
Biochemicals). All blots were processed by following the
digoxygenin protocols. Chemiluminescent detection was performed
with Kodak XAR-5 or BMR-1 film (Eastman Kodak Co.).
DNA Sequencing--
DNA was sequenced with an Applied Biosystems
automated sequencer using fluorescent terminator dye tags at the DNA
Sequencing Facility (University of Iowa). Analysis of the sequence was
performed using various programs of the Wisconsin GCG package and the
Jellyfish software package developed by Biowire.com (available on the
World Wide Web at www.biowire.com). Similarity searches against DNA and
protein sequence data bases were performed with the FASTA, BLAST, or
BLASTX algorithms.
Cloning of siaA and Mutant Construction of A2STF--
Using the
DNA sequence from the H. ducreyi lst gene (33), an open
reading frame (ORF) in the H. influenzae Rd data base named
HI0871 was identified, which contained 48% identity and 59%
similarity over the length of the predicted protein sequence. HI0871
was renamed siaA. Primers were made to amplify
siaA and some flanking DNA from strain 2019 based on the Rd
sequence from the TIGR data base (available on the World Wide Web at
www.tigr.org/). A 4.3-kb product was amplified using the Expand Long
Template PCR kit (Roche Molecular Biochemicals) and primers HSTR8
(5'-CTG CAA AAT ACA GAT AAA GCA ACA CTG GGG-3') and HSTR9 (5'-CAG CGG CAA GAA ATA TAG GGT TAG AAA AAG C-3'), which was then TA-cloned into
pCR2.1 (Invitrogen), forming pHS89-21. This insert was sequenced (accession number AY061634). The 4.3-kb DNA fragment was then subcloned
into the EcoRI site of pBluescript KS II
(Stratagene), forming pHS89-103. The insertional mutant of
siaA was constructed by cloning a PstI-digested
kanamycin antibiotic resistance gene from pBSL86 into a unique
NsiI site in the middle of siaA, forming
pHS89-103K6. The orientation of the kanamycin gene was discerned with a
restriction digest and was found to be transcribed in the same
direction as siaA (data not shown). pHS89-103K6 was
linearized with ScaI and transformed into strains A2 and
276.4 (34). Transformants were obtained and analyzed using both
PCR with internal siaA primers HSTR1 (5'-GAT GTT ATT TTT ATT TTT GTT A-3') and HSTR2 (5'-ACT TAG GGT GTA TTT TGG TTC C-3')
and Southern blots (data not shown).
Cloning of siaB and Mutant Construction of A2SB--
Primers
SiaBU2 (5'-CGG ACT ATC ATA ACG GGC-3') and SiaBD2 (5'-CTC AGA ATT CGG
GCT TCG-3') were designed based on the H. influenzae Rd
genome. A 1.5-kb DNA fragment was amplified from NTHi 2019 and
TA-cloned into pCR2.1, and both DNA strands were sequenced. This new
plasmid, called pSiaB2, contained a 675-base pair ORF with 95%
identity to HI1279 from H. influenzae Rd. This plasmid was
digested with SspI, which cuts at a unique site after
nucleotide 276. A spectinomycin resistance cassette gene was digested
with SmaI from pABR3 and ligated into the SspI
site of pSiaB2, forming pSiaB2Spec. This new construct was digested
with NotI to linearize the DNA and then transformed into
strain A2 using the MIV method (34). Transformants were obtained and
tested using PCR and Southern hybridization to confirm the proper
insertion of the spectinomycin cassette in siaB (data not shown).
Cloning of lsgB and Mutant Construction of A2lsgB and
A2STFL3AlsgB--
A 3456-bp BamHI-BsbI fragment
of H. influenzae A2 DNA containing lsgA,
-B, -C, and -D was cloned into
pGEM3zf+. A 502-bp (BsrGI-XcmI) region of lsgB was deleted and replaced by an
erythromycin cassette. This new plasmid was called pGEMLOS2ABCDerm.
This plasmid was digested with NdeI and transformed into
strains A2 and A2STFL3A using the MIV method (34). Transformants were
obtained and tested using PCR and Southern hybridization to confirm the
proper insertion of the erythromycin cassette in lsgB (data
not shown). The mutants were designated strains A2lsgB and A2STFL3AlsgB.
TA Cloning of ira from Strain 2019--
Primers iraF (5'-AGG GGG
ATA AAA CAA AGG-3') and iraR (5'-GGC AAG TCC CTG TTC AAA-3') for PCR
were designed from the published H. influenzae Rd genome.
These primers were used to amplify an intergenic region between bases
794506 and 796038. Amplification resulted in a PCR product of ~1.6 kb
from the genome of NTHi strain 2019. The product was then cloned into
the vector pTAV1 via TA cloning. The nucleotide sequence of the cloned
fragment was elucidated and compared with sequences included in the
genome data base. The resolved consensus sequence was entered into
MacVector to identify useful restriction sites. One SphI
site was predicted that would cut the cloned region into 711- and
822-base fragments and could be used to linearize pIRA. The presence of
the unique SphI site was verified by restriction
endonuclease digestion of pIRA.
Construction of pIRACM--
The plasmid pCMR containing a
chloramphenicol resistance cassette possessing the consensus uptake
sequence for Haemophilus transformation was kindly provided
by Dr. Terrence Stull (35). To obtain the necessary SphI
sites, the chloramphenicol resistance cassette was excised from pCMR
using PstI, gel-purified, and then ligated into the vector
pBSL15 that had been previously cut with PstI. The resulting
plasmid was named pHiCM1. Finally, the chloramphenicol resistance
cassette was excised from pHiCM1 by cleavage with SphI, gel-purified, and ligated into the SphI site of pIRA,
forming pIRACM. Insertion was confirmed by PCR and by sequencing the
cloning junctions.
Complementation of A2STF--
A 2.5-kb EcoRV DNA
fragment was excised from pHS89-103, gel-purified, and blunt
end-ligated into the SfoI site of pIRACM, forming pSAIRCM.
This construct was used as a template for PCR using primers iraR and
iraF, and the resulting PCR product was transformed into A2STF
using the MIV method (34). Transformants were selected for with both
kanamycin and chloramphenicol. Verification that both the full-length
and the mutant forms of siaA were present on the chromosome
was performed with PCR using primers HSTR1 and HSTR2 and with Southern
blots (data not shown). A control strain was constructed in a similar
fashion by transforming A2STF with pIRA. The insertion into the
chromosome was confirmed with Southern blots, and the resultant strain
was named A2STFIRA (data not shown).
Cloning of lic3A and Mutant Construction of A2L3A and
A2STFL3A--
Primers 0352ELTU1 (5'-ATG TCC AAA AGC AGC CAA CCA AAT
AAA CCC-3') and 0352ELTL1 (5'-CAA CGC CGA AAT CAA CCC AAA TAG AAA
GCC-3') were designed using the H. influenzae Rd genome data
base and a 4.6-kb DNA fragment containing the HI0352 ORF
(lic3A) was amplified from strain A2 using PCR. The DNA
fragment was TA-cloned into pCR2.1 and was named p0352EX. Both DNA
strands were sequenced, and a unique SwaI site was found
after nucleotide 683 of the 981-nucleotide lic3A sequence. A
nonpolar chloramphenicol cassette (pRSM1775), which was constructed in
a manner similar to that described by Menard et al. (36)
using the chloramphenicol gene (cat) from pACYC184,2 was digested with
SmaI and cloned into the SwaI site of
lic3A. This construct was named p0352EXCM. The nonpolar
cassette contains translation stop codons in all three reading frames
upstream from the start codon of cat. Five bases after the
cat stop codon, there is a Shine-Dalgarno sequence followed
by a start codon, which is in-frame with the remainder of the
lic3A gene. The insertion sites of the chloramphenicol
cassette were sequenced to ensure proper insertion and orientation.
p0352EXCM was linearized by digestion with BamHI and
transformed using the MIV method (34) into strains A2 and A2STF,
forming strains A2L3A and A2STFL3A respectively. Proper insertion into
the chromosome of these strains was confirmed with PCR amplification
using primers to the lic3A sequence and Southern
hybridization (data not shown).
LOS Preparation and Neuraminidase Treatment--
The LOS was
prepared by a modification of the Hitchcock and Brown method (37).
Organisms were grown on solid media supplemented with 20 µg/ml of
NeuAc. The organisms from a single plate were suspended in 2 ml of
phosphate-buffered saline buffer to a final A650 of 0.9. They were washed twice with
phosphate-buffered saline, resuspended in 200 µl of lysis buffer
(0.06 M Tris base, 10 mM EDTA, 2.0% SDS, pH
6.8), and incubated in a boiling water bath for 5-10 min. The samples
were allowed to cool, and 30 µl of a proteinase K solution (2.5 mg/ml
diluted in lysis buffer; Sigma) was added to 150 µl of the boiled
sample. The samples were incubated at 37 °C for 16-24 h. The LOS
was precipitated by adding one-tenth volume of 3 M sodium
acetate and 2 volumes of 100% ethanol, put on dry ice for 10 min or in
a 
80 °C freezer for 1 h, and then centrifuged at 15,000 × g for 5 min. The samples were washed twice with 70%
ethanol and brought up in double distilled H2O to a final volume of 180 µl. For SDS-PAGE gel analysis, 1-5 µl (~0.5-2.5 µg of LOS) of a typical preparation was treated with 0.5 milliunits of neuraminidase purified from Vibrio cholerae (Roche
Molecular Biochemicals) in neuraminidase buffer (0.15 M
NaCl, 4 mM CaCl2, pH 5.5) and incubated at
37 °C for 2 h.
SDS-PAGE, Silver Staining, and Western Blotting--
SDS-PAGE
gels were prepared as described by Lesse et al. (38). The
gel was loaded with 0.5-1 µl from each LOS preparation (~0.25-0.5
µg of LOS). Silver staining was performed by the method of Tsai and
Frasch (39). The Western blot was performed by the method of Towbin
(40). The monoclonal antibody 3F11 recognizes a terminal
N-acetyllactosamine structure and has been characterized previously (41). Detection of the antibody was performed using a
peroxidase-labeled goat anti-mouse secondary antibody (Kirkegaard and
Perry Laboratories) and Super Signal West Pico Chemiluminescent Substrate (Pierce). LOS from N. gonorrhoeae strain PID2 was
used as a molecular weight standard (42).
Whole-cell 3F11 ELISA--
A whole-cell enzyme-linked
immunosorbent assay (ELISA) was performed using the monoclonal antibody
3F11 and a modified method of Abdillahi and Poolman (43). Whole
bacteria were harvested from plates containing 20 µg/ml of NeuAc
(Sigma), suspended in phosphate-buffered saline, and washed twice. The
cell suspensions were diluted with double-distilled H2O to
an A600 of 0.150. 100 µl of the cell
suspension was added to the wells of flat bottom high binding
polystyrene 96-well plates (Costar) and allowed to dry completely at
37 °C. Unbound material was removed by rinsing the plates with wash
buffer (0.12 M sodium acetate, 0.15 M NaCl, 0.05% Tween 20). Half of the wells were treated with neuraminidase from V. cholerae (Roche Molecular Biochemicals) in
neuraminidase buffer (0.15 M NaCl, 4 mM
CaCl2, pH 5.5), and the other half were incubated in
neuraminidase buffer alone for 6 h at 37 °C in a humidified
chamber. The plates were washed again, and 100 µl of the monoclonal
antibody 3F11 in antibody buffer (0.15 M NaCl, 10 mM Tris, pH 7.4, 0.3% Tween 20) was serially diluted
2-fold from a starting dilution of 1:40. The plates were incubated at 25 °C overnight. The plates were washed, and 100 µl of a 1:2000 dilution of goat anti-mouse IgM phosphatase-conjugated secondary antibody (Kirkegaard and Perry Laboratories) was added to each well and
allowed to incubate at 25 °C for 1 h. The plates were washed,
and 100 µl of developer (1 mg/ml p-nitrophenyl phosphate (Sigma), 0.96% diethanolamine, 2 mM MgCl2, pH
10.0) was added to each well. The samples were allowed to develop for
30 min, and 50 µl of 4 N NaOH was added to each well to
quench the reaction. The plates were read at 405 nm with a Bio-Tec
Instruments EL 311SX microplate reader.
Preparation of O-Deacylated Lipooligosaccharides (O-LOS) and
Neuraminidase Treatment--
To make the LOS more amenable for mass
spectrometric analysis, O-linked fatty acids were removed
from the lipid A moiety. The crude LOS (~90 µg, from a single
plate) was incubated in anhydrous hydrazine (50 µl; Sigma) at
37 °C for 25 min in a microcentrifuge tube with occasional
sonication. Samples were cooled at 10 °C for 10 min prior to and
after the addition of ice-cold acetone (300 µl; Aldrich). The
quenched reaction mixture was centrifuged (12,000 × g)
for 45 min at 4 °C. The supernatant was removed, and the pelleted
O-LOS was dissolved in MilliQ water (40 µl) and evaporated
on a speed vacuum system. To remove salts and other low molecular
weight contaminants, the O-LOS (~10-20 µg) was
suspended on a nitrocellulose membrane (type VS, 0.025 µm; Millipore
Corp.) over water for 1 h. The desalted O-LOS was removed from the
membrane, concentrated with a speed vacuum system, and analyzed by
matrix-assisted laser desorption ionization-mass spectrometry
(MALDI-MS). For removal of neuraminic acid, the O-LOS
(~10-20 µg) was digested in 10 mM ammonium acetate, pH
6.0, with immobilized neuraminidase from Clostridium
perfringens (type VI; 80 milliunits) for 20 h at 30 °C.
The enzyme was pelleted by centrifugation, and the supernatant (15 µl) was transferred to a nitrocellulose membrane for drop dialysis.
The desialylated O-LOS was also concentrated and analyzed by
MALDI-MS.
MALDI-MS of O-LOS--
Dowex 50 beads (100-200 mesh,
NH
form) were added to a mixture
containing equal volumes of dialyzed O-LOS (~2 µg/µl)
and a saturated solution of 2,5-dihydroxybenzoic acid in acetone
(Aldrich). Samples were spotted on a stainless steel MALDI target and
analyzed on a Voyager-DE time of flight mass spectrometer (Applied
Biosystems) in the negative ion mode with an accelerating voltage of 20 kV. Mass spectra were smoothed once by a 19-point Savitsky-Golay
function and calibrated internally with the deprotonated molecular ions
corresponding to LOS glycoforms A1
(m/z = 2438.1), A2
(m/z = 2561.2), B1
(m/z = 2600.3), B2
(m/z = 2723.3), C1
(m/z = 2762.4), C2
(m/z = 2885.5) (see Table
II) and the prompt fragment for lipid A
(m/z = 952.0). All masses are given as their
average values.
View this table:
[in this window]
[in a new window]
|
Table II
Summary of asialo-LOS glycoforms present in the wild type and mutant
strains of H. influenzae A2
Proposed compositions contain a minimum core structure including
Hep3, Kdo(P), and O-deacylated lipid A. PEA moieties
are denoted by subscripts. All masses listed are average values.
|
|
Exoglycosidase Treatments of O-LOS--
To sequence nonreducing
terminal saccharides, O-LOS samples were treated with
exoglycosidases. The following enzymes were used in these experiments:
-galactosidase from Mortierella vinacea (Seikagaku Corp.)
or from green coffee beans (Glyko, Inc.), -galactosidase from jack
bean meal (Glyko, Inc.), -N-acetylhexosaminidase from jack bean meal (Glyko), -N-acetylgalactosaminidase from
chicken liver (Sigma), -glucosidase from Bacillus
stearothemophilus (Sigma), -glucosidase from almonds (Sigma),
and immobilized neuraminidase from C. perfringens, type VI
(Sigma). In general, enzyme reactions were run in 25-50 mM
ammonium acetate buffer, at pH 4.5 or 6.0, depending on the enzyme (pH
4.5 for the three-enzyme mixture consisting of -galactosidase,
-galactosidase, and -N-acetylhexosaminidase). The immobilized
neuraminidase was used as described above, except that in these
experiments the incubation temperature was 37 °C. Soluble enzyme
concentrations were typically in the range of 5-10 units/ml, and total
O-LOS concentrations were estimated to be in the range of
~100-200 µM. For the 276.4STF sample, the minor acceptor glycoform of interest was estimated to represent 
1% of the
total O-LOS mixture, making its concentration in the enzyme digest reactions on the order of ~1-2 µM.
Various sequences of enzyme digests were carried out, generally
starting with the O-LOS from 1-1.5 plates' worth of
bacteria, in a reaction volume of 60-90 µl. Digests were incubated
at 37 °C for 20-24 h. Except in the case of the immobilized
neuraminidase, reactions were stopped by heating in a boiling water
bath for 3 min. After quenching, reaction mixtures were typically
delivered to prewashed Microcon YM-10 filter units (Millipore) and spun at 10,000 × g for 5-15 min, depending on the reaction
volumes. Samples were then washed with several 50-100-µl aliquots of
MilliQ water, totaling about 400 µl. Retentates were recovered by
inverting the filters and centrifuging at low speed. Filters were
washed with three portions of 20-40 µl of MilliQ water, and the
washings were combined with the retentates and evaporated to dryness on a speed vacuum system. Because of its tendency to aggregate in solution, the O-LOS was recovered in the retentate fraction
using these membrane filters. Samples were then redissolved in MilliQ water for MALDI-MS analysis as described above.
Dephosphorylation of O-LOS--
O-LOS samples were
dephosphorylated by treatment with 48% aqueous HF for 16 h at
4 °C. HF was removed under vacuum using an in-line NaOH trap.
Samples were then redissolved in a small volume of MilliQ water and
evaporated to dryness on a speed vacuum system.
| |
RESULTS |
Cloning and Mutagenesis of siaA--
The ORF HI0871 from the
H. influenzae Rd genome data base encodes a predicted
protein of 306 amino acids. This protein sequence has 48% identity and
59% similarity along the entire length of the recently identified Lst
protein from H. ducreyi (33). Lst is the sialyltransferase
responsible for the addition of NeuAc to galactose of a terminal
N-acetyllactosamine moiety of the LOS (33). Based on the Rd
sequence, primers were made to amplify the HI0871 ORF and about 1.5 kb
of flanking sequence on either side of the ORF. The primers were able
to amplify a DNA fragment of similar expected size from strains Rd,
2019, and A2. The DNA fragment from strain 2019 was sequenced in its
entirety and the gene order was found to be similar but not identical
to the Rd sequence. The DNA fragment contained the 3' partial sequence
of HI0868, the full-length HI0871 (renamed siaA), and the 5'
partial sequence of HI0872 (wbaP homologue, formerly
rfbP). There was 46% identity and 57% similarity between
the predicted proteins of Lst from H. ducreyi and SiaA from
NTHi 2019 and 58% identity and 70% similarity between the predicted
proteins of SiaA from NTHi 2019 and HI0871. There is a five-base
overlap at the 3'-end of siaA and the 5'-end of
wbaP. HI0869 contained an insertion after position 533 causing a frameshift resulting in an extension of the ORF. In addition
to this, there was a small ORF (ORF1) upstream of siaA that
was not present in the Rd sequence. Both HI0869 and ORF1 do not have
any identity to any known genes in the current data bases.
An insertional mutation was made in siaA of strains A2 and
276.4 by inserting a kanamycin cassette from pBSL86 into a
NsiI site in the middle of the gene. This construct was
called pHS89-103K6. Restriction digest analysis demonstrated that the
kanamycin cassette, which does not contain a transcriptional
termination sequence, was inserted in the same orientation as
siaA. The transformation of strains A2 and 276.4 with
pHS89-103K6 employing the MIV method (34) yielded numerous
transformants. The mutant strains from A2 and 276.4 were named A2STF
and 276.4STF, respectively. PCR analysis using probes that amplify an
intergenic region of siaA, along with Southern
hybridization, confirmed the single insertion of the kanamycin cassette
in siaA of both strains A2STF and 276.4STF.
Identification of LOS Glycoforms in H. influenzae A2--
Prior to
investigating the effects of inactivating genes responsible for LOS
biosynthesis in H. influenzae, the population of glycoforms
assembled by stain A2 was assessed with MALDI-MS in the negative ion
mode. All observed molecular ions and the proposed LOS compositions for
structures containing a minimum core structure of lipid A linked to a
phosphorylated Kdo and three heptoses are summarized in Table II.
Letters represent individual glycoforms to which NeuAc (each denoted by
an asterisk) and 1-3 PEAs (denoted by subscripts) are added. MALDI-MS
analysis of the O-LOS revealed that the A2 strain expresses
an even more complex mixture of glycoforms on its outer membrane than
previously thought (15). Most structures represented extensions of the
major species (B1 and B2) by up to four
additional hexoses (Fig. 1, Table
III). An examination of the
O-LOS proceeding the enzymatic removal of NeuAc with
neuraminidase confirmed the removal of a total of eight sialylated
glycoforms, which coincided with the appearance of the asialo
counterparts (Fig. 1, Table III). In general, the most abundant
structures terminating with NeuAc contained a single N-acetylhexosamine and a total of six or seven hexoses
(H1*, H2*, I1*, and
I2*). Tandem mass spectrometric analysis of the
oligosaccharide portions of the H and I glycoforms revealed the
structures contained a nonreducing terminal Hex-HexNAc (consistent with
a N-acetyllactosamine structure), the putative acceptor site
for NeuAc.3 Interestingly,
previous investigations of the A2 strain by electrospray mass
spectrometry revealed only two sialylated LOS glycoforms, each composed
of a single N-acetylhexosamine and a total of five or six
hexoses. This discrepancy presumably reflects the altered growth
conditions, since the bacteria were grown on solid media supplemented with NeuAc in the current study (15). Furthermore, several
novel sialylated species were observed including a structure containing
two NeuAcs (H2**) and three structures lacking
N-acetylhexosamine (B2*, D2*, and
E2*). The H2** glycoform was found to be
resistant to enzymatic digestion with a mixture of -galactosidase,
-galactosidase, and -N-acetylhexosaminidase,
suggesting that it may be sialylated on two different
branches.4
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 1.
Negative ion MALDI-MS spectra of
O-LOS from H. influenzae A2, A2STF,
and A2STFC.P4. Mass spectra are shown comparing LOS isolated from
H. influenzae strains before and after treatment with
neuraminidase. See Tables II and III for molecular weights and proposed
compositions. The asterisks indicate the addition of NeuAc,
and the number of PEA moieties is denoted by subscript
type.
|
|
View this table:
[in this window]
[in a new window]
|
Table III
Summary of sialylated LOS glycoforms in H. influenzae A2, A2STF, and
A2STFC.P4
Refer to Table II for glycoform compositions. The number of NeuAc and
PEA moieties are denoted by asterisks and subscripts, respectively. All
masses listed are average values.
|
|
Analysis of LOS from A2STF--
MALDI-MS analysis of the LOS from
the siaA mutant strain (A2STF) revealed no gross changes in
the expression of LOS whose branch structures contain only hexose
relative to those observed in the parental strain A2 (Fig. 1, Table
III). However, the deletion of siaA impaired the ability of
the mutant strain to produce the major sialylated species
(H1*, H2*, H2**, I1*,
and I2*), leaving the putative LOS substrates of SiaA
(H1, H2, I1, and I2)
unmodified. The A2STF mutant strain retained the capacity to produce
the previously identified sialylated species (B2*,
D2*, and E2*) as well as additional structures
containing both NeuAc and N-acetylhexosamine
(K2*, K3*, K3**, L2*,
and L3*) that were not observed in the parental strain A2
(Fig. 1, Table III). The emergence of the unique set of glycoforms in
strain A2STF coincided with the extension of the free SiaA acceptors
(H1, H2, I1, and I2) by
the addition of a ~617-Da moiety, suggesting the addition of HexNAc
(203 Da), PEA (123 Da), and NeuAc (291 Da) (discussed below). The
expression of these and other sialylated LOS structures upon the
mutation of siaA provided strong evidence that multiple
sialyltransferases of distinct substrate specificity reside on the
outer membrane of H. influenzae strain A2.
Complementation of A2STF--
To verify that the changes in the
LOS glycoforms observed in A2STF were the result of a mutation in
siaA, the mutation was complemented in cis. This
was accomplished using the construct pSAIRCM, which was made from a
1.5-kb intergenic region identified in the Rd data base and found to be
present in the chromosome of strain A2 by PCR analysis. pSAIRCM was
transformed into A2STF using the MIV method (34). A homologous
recombination event at the intergenic region resulted in the
incorporation of both the chloramphenicol cassette and full-length
siaA genes into that region of the chromosome. It was very
important to select for transformants with both kanamycin and
chloramphenicol to avoid the loss of the original insertional mutant in
siaA. This strain was named A2STFC.P4.
A control strain was constructed to verify that the effects seen in the
complemented mutant were not due to the insertion of the
chloramphenicol cassette into the intergenic region. Strain A2STFIRA
was constructed in an identical manner to A2STFC.P4, with the exception
that pIRACM was used instead of pSAIRCM. The insertion of the
chloramphenicol cassette into the intergenic region was confirmed with
Southern hybridization. There were no differences in either the growth
curves or the MALDI-MS spectra when strain A2STF was compared with
strain A2STFIRA. Complementation of the siaA mutation was
verified by MALDI-MS analysis of LOS assembled by the A2STFC.P4 mutant
strain (Fig. 1, Table III). Reversion to the wild type phenotype was
evident by the reappearance of both the minor and major sets of
sialylated glycoforms detected in the parental strain.
Analysis of 276.4 and 276.4STF--
The mutation in strain 276.4 has been characterized previously (26). The mutation lies in
lsgE, one of the seven genes in the lsg locus of
H. influenzae (24, 25). The inactivation of lsgE
results in the expression of three major glycoforms of ~4.2, 4.4, and
5.4 kDa, a much simpler profile when compared with the parental strain
A2 (Fig. 2A). The 5.4-kDa
glycoform is of particular interest to this study because it contains a
terminal sialyl N-acetyllactosamine structure (26). When
treated with neuraminidase, the disappearance of the 5.4-kDa glycoform
coincides with the appearance of a band of ~5.1 kDa, a 300-kDa shift
corresponding to the loss of a single NeuAc residue (Fig.
2A). The LOS was transferred to a nylon membrane and probed
with the monoclonal antibody 3F11, which recognizes the terminal
N-acetyllactosamine structure. If NeuAc or other sugars are
present extending from this epitope, the binding of 3F11 is blocked.
3F11 was able to bind LOS from strain 276.4 only after neuraminidase
treatment (Fig. 2B). The binding corresponded to the 5.1-kDa
glycoform. 3F11 binding to the 276.4 LOS before neuraminidase treatment
could not be demonstrated by Western blotting, even after the blot was
overexposed (Fig. 2C). An insertional mutant in the
siaA gene in strain 276.4 was constructed. Strain 276.4 was
transformed using the MIV method (34) with pHS89-103K6, and
transformants were obtained. Verification of the proper insertion of
the kanamycin cassette in the siaA gene was confirmed with
PCR and Southern hybridization analysis. The resulting strain was named
276.4STF. The LOS profile from 276.4STF contained four major glycoforms
of ~4.2, 4.4, 5.1, and (a new glycoform) ~5.8 kDa. Both the 5.1- and 5.8-kDa glycoforms were minor species and appear as faint bands
when visualized with SDS-PAGE (Fig. 2A). When this sample
was treated with neuraminidase, the 5.8-kDa glycoform shifted to ~5.4
kDa, indicating the loss of NeuAc. When the LOS was transferred to a
nylon membrane and probed with 3F11, the antibody bound to the 5.1-kDa
glycoform in the neuraminidase-treated 276.4STF LOS sample but with
less intensity than the neuraminidase-treated 276.4 LOS sample
(Fig. 2B). After overexposure of the blot, 3F11 was also
visualized binding to the 276.4STF LOS sample before neuraminidase
treatment (Fig. 2C). There was no antibody binding observed
to the LOS of strain 276.4STF in the 5.4-kDa size range, even after the
blot was overexposed (Fig. 2, B and C). These
data indicated that although the 5.8-kDa glycoform contained NeuAc, it
lacked a terminal N-acetyllactosamine structure. MALDI-MS
analysis of these samples confirmed these results (discussed
below).
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 2.
SDS-PAGE and Western blot. LOS samples
from 276.4 and 276.4STF were run on a SDS-PAGE gel, transferred to a
nylon membrane, and probed with 3F11. The approximate relative
molecular weights based on gel mobility of the five LOS glycoforms from
PID2 are noted. A, SDS-PAGE gel; B, Western blot
probed with 3F11; C, overdeveloped Western blot. LOS samples
are as follows: PID2 from N. gonorrhoeae (lane
1), 276.4 (lane 2), 276.4 treated with
neuraminidase (lane 3), 276.4STF (lane
4), and 276.4STF treated with neuraminidase (lane
5).
|
|
Mutant Construction and MALDI Analysis of lic3A in Strains A2 and
A2STF--
The higher molecular weight sialylated species seen in both
A2STF and 276.4STF indicated the possibility of a second
sialyltransferase in this organism. Our analysis suggested that this
second sialyltransferase would have a different specificity than a
terminal N-acetyllactosamine structure. This was based on
the fact that Western blot analysis with 3F11 indicated that this
terminal structure was lost in the 276.4STF mutant, but sialylation was
retained. Recently, a sialyltransferase was reported in a H. influenzae gene called lic3A, which is the first gene
in the lic3 locus (20, 29). This gene is phase-variable, and
its protein product sialylates both lactose and
N-acetyllactosamine in vitro but only sialylates
lactose-containing structures in vivo (29). Lic3A has around
40% identity to CstII from C. jejuni (29). This protein in
C. jejuni is a bifunctional sialyltransferase, with the
ability to transfer NeuAc to terminal galactose residues as well as the
O-8 position of terminal NeuAcs (30). To investigate the possibility of
Lic3A being the sialyltransferase responsible for sialylating the
higher molecular weight species of A2STF, a mutation in
lic3A was made in both strains A2 and A2STF. Primers were
made to the DNA sequence upstream and downstream of HI0352, the
lic3A homologue in Rd, based on the published H. influenzae Rd genomic sequence. A DNA fragment was PCR-amplified
from strain A2 that was similar in size to the expected Rd fragment.
This fragment was TA-cloned, forming plasmid p0352EX. Both strands of
the amplified DNA of p0352EX were sequenced and compared with the known
sequences in the data base. The gene order in p0352EX was identical to
that in Rd. Downstream from lic3A was the galE homologue HI0351 and the 5' portion of HI0350, a hypothetical membrane
protein. Upstream from the lic3A gene was HI0354 and the 5'
portion of HI0355. A nonpolar chloramphenicol cassette was inserted in
the middle of lic3A, forming p0352EXCM. This construct was
linearized and transformed into strains A2 and A2STF. Transformants from the A2 and A2STF transformation were named A2L3A and A2STFL3A, respectively. Proper insertion of the chloramphenicol cassette was
confirmed with Southern hybridization (data not shown).
MALDI-TOF analysis of the LOS expressed by strain A2L3A revealed the
loss of both the disialylated (H2**) and hexose-containing (B2*, D2*, and E2*) species but
confirmed the presence of the major sialylated species
(H1*, H2* I1*, and H2*)
observed in the parental strain A2 (Fig.
3, Table
IV). This indicated that in strain A2,
the Lic3A homologue has substrate specificity distinct from SiaA,
preferring to transfer NeuAc to LOS glycoforms devoid of
N-acetylhexosamine. The loss of the disialylated glycoform H2** may reflect the inability of the A2L3A mutant to
sialylate one branch of the H2 structure. The mass spectra
of the LOS isolated from strain A2STFL3A revealed that the double
mutant lost the ability to produce the SiaA products (H1*,
H2* I1*, and H2*) containing N-acetylhexosamine (Fig. 3, Table IV). Strain A2STFL3A did
produce sialylated species (K2*, K3*,
L2*, and L3*) that were not previously observed
in either strain A2L3A or the parental strain A2 but were identified in
strain A2STF. Since the deletion of both siaA and
lic3A precludes the assembly of only certain subsets of
NeuAc containing LOS glycoforms of H. influenzae strain A2,
we conclude that their gene products function as sialyltransferases and
infer the existence of a third sialyltransferase in this complex
system.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 3.
Negative ion MALDI-MS spectra of
O-LOS from H. influenzae A2, A2L3A,
and A2STFL3A. Mass spectra comparing LOS isolated from H. influenzae strains before and after treatment with neuraminidase.
See Tables II and IV for molecular weights and proposed compositions.
Asterisks indicate the addition of NeuAc, and the number of
PEA moieties is denoted by subscript type.
|
|
View this table:
[in this window]
[in a new window]
|
Table IV
Summary of sialylated LOS glycoforms in H. influenzae A2, A2L3A, and
A2STFL3A
Refer to Table II for glycoform compositions. The number of NeuAc and
PEA moieties are denoted by asterisks and subscripts, respectively. All
masses listed are average values.
|
|
Exoglycosidase Treatments of 276.4STF O-LOS--
To investigate
the structures of the extended glycoforms produced in the
siaA mutants, experiments were conducted on the
O-LOS from strain 276.4STF. The sialylated acceptor species
in 276.4STF, designated J3*, is a minor component in the
O-LOS mixture occurring at m/z ~3706
Da (Table V). Compared with the
sialylated G2* glycoform in O-LOS from strain
276.4 (Table V), this species was ~324-326 Da higher in molecular
mass. Our initial assumption was that the added moiety was a
Hex2 extension (which would have corresponded to the same
composition as the A2 wild-type I2* glycoform). However, dephosphorylation of the O-LOS sample with aqueous HF
revealed that the novel acceptor species contained three PEAs, rather
than the one or two PEAs present on all of the major components in the mixture. This observation indicated that the added structural pieces had to be HexNAc plus PEA (+326 Da), rather than
Hex2 (+324 Da).
View this table:
[in this window]
[in a new window]
|
Table V
Summary of LOS glycoforms in H. influenzae 276.4 and 276.4STF
Refer to Table II for glycoform compositions. The number of NeuAc and
PEA moieties are denoted by asterisks and subscripts, respectively. All
masses listed are average values.
|
|
To sequence the terminus of the J3* species, the
O-LOS from strain 276.4STF was subjected to a series of
exoglycosidase treatments and analyzed by MALDI-MS (data not shown). As
a first step, the sample was treated with a mixture of three enzymes
consisting of -galactosidase, -galactosidase, and
-N-acetylhexosaminidase. This treatment was done to
ensure that any nonreducing terminal saccharides of this type that were
not blocked by sialylation would be removed from the species before
sequencing of the acceptor began. In the case of strain 276.4STF, the
mixture treatment did not alter the glycoform profile of the sample
except to remove a small peak for the G2 species. However,
when strain A2STFL3A O-LOS was subjected to the mixture
treatment, various high molecular weight glycoforms were digested with
the enzymes, such that the final profile of the strain A2STFL3A
O-LOS population was virtually identical to the strain
276.4STF O-LOS profile, except for minor differences in the
PEA content of the species. The high molecular weight sialylated
glycoforms in the strain A2STFL3A sample (the K2*/K3* and L2*/L3*
glycoforms given in Table IV) were converted to the
J2*/J3* species by the loss of 1 and 2 galactoses, respectively. This indicated that the J3*
species contained the essential structure of the novel acceptor branch
found in the O-LOS from both the A2STFL3A and 276.4STF strains.
After the mixture treatment, the strain 276.4STF sample was
neuraminidase-treated and then tested with a battery of
exoglycosidases. Using this approach, none of the following enzymes
were effective at removing a terminal sugar from the asialo acceptor:
-galactosidase, -galactosidase, -glucosidase, -glucosidase,
-N-acetylhexosaminidase, and
-N-acetylgalactosaminidase.
As an alternate approach, the mixture-treated strain 276.4STF sample
was subjected to dephosphorylation with aqueous HF prior to further
enzymatic treatments. To minimize the loss of sialic acid in the
dephosphorylation step, the reaction time was kept to 16 h. Under
these conditions, microheterogeneity was created in the sample, with
asialo and sialylated species appearing as pairs containing 0 or 1 residual phosphate group. When the HF-treated strain 276.4STF sample
was subjected to -N-acetylgalactosaminidase treatment,
all of the asialo and sialylated glycoforms of the acceptor were
shifted to lower mass by the loss of 203 Da, indicating the removal of
an -linked GalNAc (-GalNAc). At this point, the sample was split
into two portions, one treated with -galactosidase and one treated
with neuraminidase followed by -galactosidase. In the sample treated
first with -galactosidase, only the asialo glycoform was shifted,
indicating the removal of -linked galactose. The sialylated
glycoform was shifted to the same species in a sequential fashion by
neuraminidase treatment (291 Da) followed by -galactosidase
treatment (162 Da).
Cloning and Mutagenesis of lsgB--
The lsgB has been
previously described in the lsg locus of H. influenzae as orf2 (accession number Q48211) and
in the TIGR H. influenzae Rd data base as ORF HI1699. It
encodes a predicted protein of 304 amino acids. This protein sequence
has 27% identity and 46% similarity along the entire length of the
recently identified 2-3-sialyltransferase (Lst) protein from
Neisseria gonorrhoeae and N. meningitidis (25,
33). Lst is the sialyltransferase responsible for the addition of NeuAc
to galactose of a terminal N-acetyllactosamine moiety of the
LOS. To clone this gene for mutagenesis and transformation, a
BbsI-PstI digest of the lsg locus in
the plasmid pGEMLOS4 released a 2161-bp fragment. The plasmid was
religated containing lsgA (lsg orf1),
lsgB (lsg orf2), and lsgC
(lsg orf3) and part of lsgD (lsg
orf4). Using this construct, a deletion mutant was made in
lsgB by deleting 506 bases with a
BsrGI-XcmI digest of this fragment and inserting
an erythromycin cassette. This cassette was placed in the forward
orientation and does not contain a transcriptional termination
sequence. The plasmid was linearized and transformed into strains A2
and A2STFL3A by the MIV method using appropriate selection. The
resulting strains were designated A2lsgB and A2STFL3AlsgB. The
chromosomal insertions were confirmed by PCR and Southern hybridization.
MALDI-TOF Analysis of A2lsgB and A2STFL3AlsgB--
The
O-deacylated LOS from the lsgB mutant (A2lsgB)
and the triple mutant (A2STFL3AlsgB) were analyzed by MALDI mass
spectrometry, both before and after neuraminidase treatment (Fig.
4). The A2STFL3AlsgB triple mutant did
not produce any of the sialylated LOS structures characteristic of the
H. influenzae A2 strain; nor did it produce the novel
sialylated species (K2*, K3*, L2*,
and L3*) found in the siaA mutant (Fig. 1) and
the siaA and lic3A double mutant (Fig. 3).
Additionally, the asialo forms of the novel acceptors (K2,
K3, L2, and L3) were not detected
in the A2STFL3AlsgB O-LOS population. The only LOS
modifications observed involved a shift to higher phosphorylation
states. In addition to an increase in the relative amount of glycoforms
containing two PEAs, a formerly minor series of glycoforms containing
two PEAs plus an additional phosphate moiety was prominent in the
A2STFL3AlsgB O-LOS population (Fig. 4). Since this new
series nearly coincided with some of the predicted masses for the
asialo N-acetylhexosamine containing glycoforms
(H1, H2, I1, and I2),
the A2STFL3AlsgB O-LOS was dephosphorylated with aqueous HF
and reanalyzed by MALDI mass spectrometry. No significant peaks for the
asialo N-acetylhexosamine acceptors were revealed in the
dephosphorylated mixture (data not shown).
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 4.
Negative ion MALDI-MS spectra of
O-LOS from H. influenzae A2, A2lsgB,
and A2STFL3AlsgB. Mass spectra are shown comparing LOS isolated
from H. influenzae strains before and after treatment with
neuraminidase. See Tables II and IV for molecular weights and proposed
compositions. The asterisks indicate the addition of NeuAc,
and the number of PEA moieties is denoted by subscript
type. Species present in strain A2STFL3AlsgB containing an
additional phosphate moiety are labeled in italic
type, with the additional subscript P.
|
|
Like the A2STFL3AlsgB triple mutant, the lsgB mutant
(A2lsgB) did not produce the asialo or sialylated forms of the novel LOS structures (K2, K3, L2, and
L3) found in the siaA mutants (Fig. 4).
Additionally, there was almost no detectable production of the
sialyl-N-acetylhexosamine-containing glycoforms
(H1*, H2*, H2**, I1*,
and I2*) characteristic of the parental strain A2 (Fig. 4).
Only very minor peaks for the H1* and I1*
glycoforms were detected by MALDI mass spectrometry. It was not
possible to detect the asialo glycoforms of these acceptors
(H1, H2, I1, and I2) in
the A2lsgB O-LOS mixture, but small peaks for the H1 and
I1 acceptors were distinguished after neuraminidase
treatment (Fig. 4). While showing reduced production of the
N-acetylhexosamine-containing glycoforms, the A2lsgB mutant
was capable of expressing the hexose-containing sialylated species
(B1*, B2*, D2*, E1*,
and E2*) found in H. influenzae A2.
Whole-cell 3F11 ELISA--
To more directly measure the level of
sialylation of N-acetyllactosamine structures on the LOS of
whole bacteria, a whole-cell ELISA assay was developed using the
monoclonal antibody 3F11 (Fig. 5).
Because of the relatively low abundance of the sialylated glycoforms,
this method gives greater sensitivity in the analysis of the
N-acetyllactosamine-terminating glycoforms. The parent strain A2, in the absence of neuraminidase, bound 3F11 poorly (Fig. 5).
When neuraminidase was added, 3F11 was able to bind with much greater
efficiency, indicating the terminal N-acetyllactosamine containing glycoforms were substituted with NeuAc. Previous work by
Hood et al. (28) showed that a mutation in siaB,
a CMP-NeuAc synthetase, eliminated sialylation. When a siaB
mutant (A2SB) was analyzed with this assay, the binding of 3F11
indicated that all of the N-acetyllactosamine was free of
NeuAc; however, the binding was ~50% of that seen in strain A2 after
neuraminidase treatment (Fig. 5). This indicated that in the absence of
sialylation, a population of the terminal
N-acetyllactosamine was modified. In support of this, the
MALDI-MS analysis of strain A2SB contained no sialylated glycoforms but
did contain higher molecular weight glycoforms not found in strain
A2.5 The siaA
mutant strain (A2STF), before neuraminidase treatment, reacted with
3F11 in a similar fashion as strain A2 before treatment with
neuraminidase. After neuraminidase treatment, 3F11 binding to A2STF was
reduced to 50% of that seen when neuraminidase-treated strain A2 was
the target antigen. This decrease in binding between the
neuraminidase-treated samples was similar to the difference in the
level of binding between strain A2SB and the neuraminidase-treated strain A2 (Fig. 5). This suggests that in strain A2STF, a population of
the terminal N-acetyllactosamine was sialylated, and an
additional population was modified, similar to that observed with
strain A2SB. The MALDI-MS data from strain A2STF support the hypothesis that other sugars are added to the terminal
N-acetyllactosamine. LOS from strain A2STF had a reduced
amount of the sialylated N-acetylhexosamine-containing glycoforms (H1*, H2*, I1*, and
I2*) when compared with the parental strain A2 (Fig. 1,
Table III). This change was not accompanied by an increase in the
acceptor glycoforms (H1, H2, I1,
and I2) but rather higher molecular weight glycoforms
(K2*, K3*, L2*, and
L3*) that were extended by a HexNAc and a PEA moiety when compared with glycoforms H1*, H2*,
I1*, and I2*. Further evidence that sugars are
being added to the terminal N-acetyllactosamine comes from
analysis of strain 276.4STF, which indicates that in a siaA
mutant, the sugars are added to the terminal
N-acetyllactosamine structure (Fig. 2).
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 5.
Measurement of the level of LOS
sialylation on terminal N-acetyllactosamine using
3F11. The monoclonal antibody 3F11 recognizes a terminal
N-acetyllactosamine structure on the LOS. The presence of
NeuAc on the N-acetyllactosamine structure inhibits the
binding of 3F11. Sialic acid was removed from the LOS with
neuraminidase. The level of sialylation was measured by comparing the
samples before and after enzymatic treatment. The samples were measured
at a 1:40 dilution of 3F11. Clear bars represent
samples before treatment with neuraminidase, and solid
bars represent samples after treatment with
neuraminidase.
|
|
Our data combined with a previous study by Hood et al. (29)
indicate that H. influenzae contains multiple
sialyltransferases. It is conceivable that in the absence of one
sialyltransferase, the other sialyltransferase could compensate by
sialylating terminal structures not normally sialylated in the parental
strain. We analyzed the lic3A mutant (A2L3A), which has been
shown previously to be a sialyltransferase capable of sialylating both
lactose and N-acetyllactosamine structures in
vitro but only lactose structures in vivo (29). Using
the ELISA assay, strain A2L3A was similar to strain A2, regardless of
neuraminidase treatment (Fig. 5). This indicated that lic3A
was not responsible for sialylating N-acetyllactosamine
structures in the parent strain A2. A double mutant in both
siaA and lic3A (A2STFL3A) was analyzed before and after neuraminidase treatment. The binding of 3F11 was identical both
before and after treatment with neuraminidase and was ~50% of that
seen with the neuraminidase-treated parental strain A2 (Fig. 5). This
result indicated that strain A2STFL3A lacked NeuAc on the terminal
N-acetyllactosamine, and in a similar fashion to both
strains A2SB and A2STF, a population of the terminal
N-acetyllactosamine was further modified. The MALDI-MS
analysis of strain A2STFL3A supports this contention (Fig. 3, Table
IV). Strain A2STFL3A was completely devoid of the sialylated
N-acetylhexosamine-containing glycoforms representing peaks
H1*, H2*, I1*, and I2*
but did contain peaks representing the acceptor glycoforms
(H1, H2, I1, and I2). This strain also contained higher molecular weight
N-acetylhexosamine containing glycoforms modified by the
addition of a HexNAc and a PEA moiety (K2*,
K3*, L2*, and L3*) when compared
with the N-acetylhexosamine-containing glycoforms designated
by peaks H1*, H2*, I1*, and
I2*. Both the ELISA assay and the MALDI-MS results of
strain A2STFL3A indicated that the sialylation observed in the
hexose-containing glycoforms in strain A2STF was the result of the
second sialyltransferase, Lic3A.
The 3F11 ELISA analysis of strain A2lsgB showed lower binding activity,
similar to A2 prior to neuraminidase (Fig. 5). After enzyme treatment,
antibody binding to strain A2lsgB increases 4-fold, which is indicative
of sialic acid release from N-acetyllactosamine. ELISA
analysis of the triple mutant showed low levels of binding before and
after neuraminidase treatment. This would indicate that all of the
N-acetyllactosamine sites are not sialylated and that the
levels of this acceptor are very low. This is confirmed by the MALDI
data, which show a disappearance of the sialylated glycoforms and an
inability to identify N-acetyllactosamine-containing glycoforms.
| |
DISCUSSION |
The biosynthesis of LOS is a complex process (5-9). Although
previous investigations utilizing electrospray mass spectrometry revealed the presence of only two sialylated glycoforms in H. influenzae strain A2 (26), MALDI-MS analysis of the LOS isolated from plate-grown organisms demonstrated eight LOS glycoforms
terminating with NeuAc, including a disialylated species never observed
before. This greater structural diversity of LOS seen in these MALDI-MS data can be attributed in part to advances made in mass spectrometry as
well as an appreciation of the effects of glycoform biosynthesis when
LOS is recovered from organisms grown on plates as opposed to broth
culture (44).
The goal of this study was to determine the roles of siaA
and lsgB in the sialylation of H. influenzae LOS.
From the evidence presented here, we conclude that both SiaA and LsgB
function as sialyltransferases in strain A2. There are several lines of
evidence that lead us to this conclusion. First, both proteins have
high amino acid identity to previously described sialyltransferases and
no other known glycosyltransferases. The homologous Lst proteins in
H. ducreyi and Neisseria have been shown to
sialylate a terminal N-acetyllactosamine structure on their
respective LOS (33). Second, mutations in siaA and
lsgB only affect the sialylation of
N-acetyllactosamine-containing glycoforms from stains A2 and 276.4. MALDI-MS analysis of LOS obtained from the siaA
mutant strain A2STF showed that the sialylated
N-acetyllactosamine-containing glycoforms that are composed
of six and seven hexoses disappear and are replaced by a new set of
sialylated glycoforms that are extended by the addition of a HexNAc and
a PEA moiety. In order to eliminate the possibility that the effects
seen by a mutation of siaA were caused by polar effects on
downstream genes, the mutant was complemented. We developed a construct
that would allow us to insert a functional siaA gene into
the chromosome at a putative intergenic region. The LOS and ELISA
profiles of the complemented mutant strain A2STFC.P4 were identical to
the parental strain A2 (Fig. 1, Table III). This is clear evidence that
the effects seen from the siaA mutants are the result of the
inactivation of the siaA gene.
The most likely explanation for the observation of novel sialylated
glycoforms is that, in the absence of sialylation by SiaA, the HexNAc
and PEA are added to the oligosaccharide branch and this new terminal
structure is modified by sialic acid by yet another sialyltransferase
(LsgB). Alternatively, it is possible that in the absence of SiaA, the
HexNAc and PEA are added to a terminal branch other than the one
containing sialic acid. The SDS-PAGE analysis of strain 276.4 and its
siaA mutant strain 276.4STF show this is not the case (Fig.
2). H. influenzae contains a triheptose core, and branched
structures can be assembled from any of the three heptoses (15, 16,
45). Strain 276.4 contains a mutation in lsgE, which results
in a very defined LOS phenotype containing a single sialylated
glycoform where the NeuAc extends from a terminal N-acetyllactosamine structure (26). The siaA
mutant strain 276.4STF contains only a very small amount of this
glycoform but also produces a higher molecular weight sialylated
glycoform extended by the addition of one HexNAc and one PEA moiety
(Table V). This extended glycoform did not bind to monoclonal antibody
3F11 after neuraminidase treatment, indicating that the 3F11
N-acetyllactosamine epitope had been modified, by the
addition of the extra moieties.
The third line of evidence comes from analysis of the strains with a
whole-cell ELISA assay using 3F11 (Fig. 5). The parental strain A2
binds 3F11 minimally before neuraminidase treatment, but after
treatment with neuraminidase the binding of 3F11 increased considerably, indicating that terminal N-acetyllactosamines
were sialylated. Strain A2STF reacted in a similar fashion to strain A2, with the exception that the level of 3F11 binding after
neuraminidase treatment was about 50% that of strain A2. This
indicated that either the 3F11-positive glycoforms were not being
produced in as great a quantity or that the 3F11 epitope was being
modified. The difference in 3F11 binding of strain A2STF LOS before and after neuraminidase treatment indicated that although there was a
decrease in the expression of the terminal
N-acetyllactosamine epitope in this strain, it was still
capable of being sialylated. A mutation in lic3A did not
affect sialylation of N-acetyllactosamine as compared with
the parent strain A2 using ELISA, but there was no sialylation detected
in the siaA and lic3A double mutant by ELISA
(Fig. 5). The level of 3F11 binding in the siaA and
lic3A double mutant was the same as that of the
neuraminidase-treated A2STF strain and a sialylation-deficient
siaB mutant strain (A2SB) (28). Hood et al. (28)
showed that a mutation in siaB, a CMP-NeuAc synthetase gene,
eliminated sialylation in NTHi strains. MALDI-MS of the LOS from this
double mutant showed the presence of sialylated hexosamine-containing
glycoforms. The failure of 3F11 to recognize these
neuraminidase-treated glycoforms most probably reflects the fact that
these new glycoforms have been altered in such a way that this antibody
cannot bind. The appearance of novel HexNAc- and PEA-containing higher
molecular weight glycoforms observed by MALDI-MS supports the
hypothesis that the previously available acceptor for sialic acid,
N-acetyllactosamine, is modified.
Finally, a fourth line of evidence comes from studies conducted on the
siaA, lic3A, and lsgB triple mutant
(strain A2STFL3AlsgB). The presence of higher molecular weight
sialylated species in the siaA and lic3A double
mutant indicated that there had to be a third sialyltransferase in this
system. The only other gene in H. influenzae with identity
to a known sialyltransferase is lsgB. The LOS of strain
A2STFL3AlsgB did not contain any sialylated glycoforms and did not
produce any sialylated N-acetyllactosamine-containing glycoforms when analyzed with the ELISA assay (Figs. 4 and 5), confirming LsgB as the third sialyltransferase.
LsgB would predictably have a different acceptor specificity from
lactose or N-acetyllactosamine, since these structures are recognized by either lic3A or siaA. The
siaA mutants have a HexNAc and a PEA moiety added to their
structures, and in the case of 276.4STF, we know that this eliminates
the N-acetyllactosamine epitope (Fig. 2). To get a
preliminary assessment of the structures of the novel acceptors in the
siaA mutants, we conducted a series of exoglycosidase
treatments on 276.4STF O-LOS. Mass spectrometric measurements indicated that that glycoform was related to the sialyl-N-acetyllactosamine-containing glycoform in strain
276.4 O-LOS by the addition of a HexNAc and a PEA moiety.
Enzymatic digestions conducted on the 276.4STF O-LOS
suggested that these structural pieces formed a new branch on the
sialylated terminus. Once the sample was dephosphorylated, it could
lose a nonreducing terminal -GalNAc by treatment with
-N-acetylgalactosaminidase, suggesting that the added PEA
moiety had been linked to the -GalNAc residue. This -GalNAc could
be released from the acceptor both before and after desialylation,
indicating that it was not the sugar being sialylated. Treatment of the
species lacking -GalNAc with -galactosidase had no effect,
whereas neuraminidase treatment followed by -galactosidase treatment
confirmed that the sialic
§,
,
TOP
ABSTRACT
INTRODUCTION
