The role of lex2 in lipopolysaccharide biosynthesis in Haemophilus influenzae strains RM7004 and RM153

Ruth Griffin1,{dagger}, Andrew D. Cox2, Katherine Makepeace1, James C. Richards2, E. Richard Moxon1 and Derek W. Hood1

1 Molecular Infectious Diseases Group, University of Oxford Department of Paediatrics, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK
2 Institute for Biological Sciences, National Research Council of Canada, Ottawa, Ontario, Canada K1A 0R6

Correspondence
Derek W. Hood
dhood{at}molbiol.ox.ac.uk


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The locus lex2, comprising lex2A and lex2B, contributes to the phase-variable expression of lipopolysaccharide (LPS) of Haemophilus influenzae and was found to be present in 74 % of strains investigated. lex2A contains 5'-GCAA repeats which vary in number from 4 to 46 copies between strains. The locus was cloned from the serotype b strains RM7004 and RM153 and showed >99 % nucleotide sequence identity between these strains and the published lex2 sequence. Disruption of the lex2B gene in strain RM7004 resulted in truncation of some LPS glycoforms, shown by gel fractionation, with only one glycoform reacting with a digalactoside-specific monoclonal antibody, 4C4, compared with four LPS glycoforms in the more elongated LPS of the parent strain. Mass spectrometry and NMR analyses of LPS from the lex2B mutant revealed loss of the terminal digalactoside as well as the second {beta}-glucose extending from the first heptose of the inner core. The authors conclude that Lex2B is the {beta}-(1-4)-glucosyltransferase that adds the second {beta}-glucose to the first {beta}-glucose as part of the oligosaccharide extension from the first heptose of the LPS of strain RM7004. Investigation of the expression of the lex2 locus indicated that the genes are co-transcribed and that both reading frames are required for addition of this second {beta}-glucose in a phase-variable manner.


Abbreviations: NOESY, nuclear Overhauser effect spectroscopy; T-SDS-PAGE, Tricine-SDS-PAGE

The GenBank/EMBL/DDBJ accession number for the sequence reported in this paper is AF503507.

{dagger}Present address: National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Haemophilus influenzae is a bacterium that resides in the human respiratory tract as a commensal. However, this organism has a propensity to be invasive, and capsulate, particularly serotype b, strains can disseminate into the bloodstream causing septicaemia and invade the central nervous system to cause meningitis. A major virulence determinant of H. influenzae is lipopolysaccharide (LPS) (reviewed by Hood & Moxon, 1999). Unlike the LPS of enteric bacteria, the LPS of H. influenzae lacks an O antigen. From the relatively conserved inner core, consisting of a tri-heptose backbone, linked via a single 2-keto-3-deoxyoctulosonic acid (Kdo) to lipid A, oligosaccharides comprising mainly hexose sugars are extended and vary in composition between strains (Fig. 1). The number of sugars and phosphate-containing substituent groups composing these oligosaccharides can vary at high frequency within individual strains. This reversible loss and gain of epitopes is known as phase variation. Phase variation of LPS in H. influenzae is mediated by ‘on–off’ switching of the translation of LPS-specific loci through tetranucleotide repeat tracts within the 5' end of the reading frame (Weiser et al., 1989a, b; High et al., 1993; Hood et al., 1996; Jarosik & Hansen, 1994). The number of repeat units within a tract can change due to mis-pairing of complementary strands at the repeat region during DNA replication, resulting in loss or gain of one or more repeat units and a switch in the reading frame to be translated (Levinson & Gutman, 1987). In having multiple phase-variable loci, H. influenzae can simultaneously generate a range of LPS glycoforms allowing the organism to adapt to differing microenvironments of the host and help it escape from host immune responses.



View larger version (14K):
[in this window]
[in a new window]
 
Fig. 1. Schematic representation of the structure of the fully extended LPS glycoforms of H. influenzae strains RM153 (Masoud et al., 1997) and RM7004 (Masoud et al., 2003). The LPS of strain RM7004 is identical in structure to that of RM153 except for an additional extension from the first heptose. Represented in the LPS structure are: Hep, LD-heptose; Glc, glucose; Gal, galactose; P, phosphate; ChoP, phosphorylcholine; PEtn, phosphoethanolamine. A dashed line indicates the substituents that are variably present.

 
Tetranucleotide-repeat-containing loci in H. influenzae are involved in the phase-variable expression of a digalactoside (lic2A and lgtC) (High et al., 1993; Hood et al., 1996), of phosphorylcholine (ChoP) (lic1) (Weiser et al., 1997) and of sialic acid (lic3A) (Hood et al., 2001) in the LPS. A fifth locus, lex2, comprising two reading frames lex2A and lex2B, which is absent from the published Rd genome sequence (Fleischmann et al., 1995), was identified in a serotype b strain, DL42 (Jarosik & Hansen, 1994). A tract of multiple, tandem 5'-GCAA repeats is located within the 5' end of lex2A and some features of the sequence suggest that lex2A and lex2B are co-transcribed (Jarosik & Hansen, 1994).

Disruption of the lex2B gene in strain DL42 led to loss of reactivity of this strain to an LPS-specific monoclonal antibody (mAb) 5G8. A correlation between repeat numbers within lex2A and mAb 5G8 reactivity was found (Jarosik & Hansen, 1994) but the specific function of the lex2 locus has not been elucidated since both the structure of the LPS of this strain and the composition of the epitope reactive with mAb 5G8 are unknown. In the present study we investigate the role of the lex2 locus in H. influenzae RM7004 and RM153, strains for which the details of the LPS structure are known (Masoud et al., 1997, 2003), and elucidate the function of Lex2B.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacterial strains, plasmids and culture conditions.
Haemophilus influenzae strains used in this study are listed in Table 1; they were grown at 37 °C in brain–heart infusion (BHI) broth supplemented with 10 µg haemin ml-1 and 2 µg NAD ml-1 or on BHI solidified with agar (1 %, w/v) supplemented with 10 % Levinthal's reagent (Alexander, 1965). Escherichia coli strain DH5{alpha} was cultured in Luria Broth (LB) or on agar plates at 37 °C. When necessary, IPTG (40 µg ml-1) and X-Gal (40 µg ml-1) (Gibco-BRL) were added to the medium. Antibiotics (Sigma) were used in selective media at the following concentrations: ampicillin at 100 µg ml-1 for E. coli; kanamycin at 50 µg ml-1 for E. coli and at 10 µg ml-1 for H. influenzae.


View this table:
[in this window]
[in a new window]
 
Table 1. H. influenzae strains, with capsular serotype indicated, used in this study

The strains were obtained from the collection of E. R. Moxon and were included in the taxonomic study by Musser et al. (1990). Indicated are the number of tetranucleotide repeat units in lex2A; – indicates that this locus was absent. Repeat numbers that are predicted as permissive for translation of full-length Lex2A are underlined. ND, Not determined.

 
DNA procedures.
Restriction enzymes and DNA-modifying enzymes were purchased from Boehringer Mannheim and used according to the manufacturer's instructions. Plasmid DNA was prepared from E. coli strains by the alkaline-lysis method (Sambrook et al., 1989). Chromosomal DNA was prepared from H. influenzae by the method described by Preston et al. (1996). Plasmid DNA was introduced into E. coli by transformation after CaCl2 treatment (Sambrook et al., 1989) and into H. influenzae by transformation using the MIV procedure (Herriott et al., 1970). Purification of DNA from agarose gels was performed using the Qiaex II agarose gel extraction kit (Qiagen).

Southern blot and colony blot analyses were performed as described by Sambrook et al. (1989) using Hybond-N nylon membranes (Amersham). Following blotting, the membranes were incubated at 65 °C for 2–3 h in hybridization buffer as described by Sambrook et al. (1989) but lacking formamide. Hybridizations, with added labelled DNA probe, were performed overnight at 65 °C and filters were subsequently washed twice in 2x SSC/0·1 % SDS for 15 min at 65 °C then once in 0·2x SSC/0·1 % SDS for 15 min at 65 °C.

PCR amplification was carried out in buffer containing 50 mM KCl, 10 mM Tris/HCl pH 8, 0·01 % (w/v) gelatin and 2·5 mM MgCl2. Thirty cycles of PCR were performed, each cycle consisting of 1 min periods of denaturation at 94 °C, annealing at 47 °C and extension at 72 °C.

Cloning and DNA sequence analysis of the lex2 locus.
The 801 bp PCR product amplified from chromosomal DNA of H. influenzae strain RM7004 using primers lex5' and lex3', designed from the published lex2 sequence of strain DL42 (GenBank U05670) (Table 2), was cloned into pCR2.1 (Invitrogen), generating plasmid pCR2.1lex2. The 432 bp SspI–RsaI fragment corresponding to an internal fragment of lex2B was purified and labelled with [{alpha}-32P]dCTP (Amersham Pharmacia), by random-primed labelling (Boehringer Mannheim). This hybridization probe was used for Southern analysis of chromosomal DNA of strains RM7004 and RM153, yielding signals on the blot corresponding to band sizes of 1·9 kb and 5·5 kb for EcoRI- and PstI-digested DNA respectively. The corresponding fragments were cloned by recovering DNAs of the appropriate sizes from an agarose gel and ligating into EcoRI- or PstI-digested, dephosphorylated pBluescript SK- (Stratagene) then transforming into E. coli. The transformants obtained were screened by colony hybridization, using the 432 bp lex2B probe. DNA was prepared from positively hybridizing colonies and verified by restriction digestion. Clones containing the 1·9 kb EcoRI insert were designated pB7004Elex2 and pB153Elex2 and clones containing the 5·5 kb PstI fragment from strain RM7004 were designated pB7004Plex2.


View this table:
[in this window]
[in a new window]
 
Table 2. Oligonucleotide primers used in PCR amplification and DNA sequencing

 
Plasmid primers (T7 forward primer and M13 reverse primer) and primers designed from the published lex2 sequence of strain DL42 (Table 2) were used for sequencing of the plasmid clones using a Big Dye DNA sequencing kit (PE Applied Biosystems). The reactions were subjected to 25 cycles consisting of heating at 94 °C (30 s), 50 °C (15 s) then 60 °C (4 min). The DNA obtained from the sequencing reactions was ethanol precipitated and the pellet dried in a vacuum centrifuge for 2 min before being resuspended then analysed with an ABI377 autosequencer.

For direct sequencing of 5'-GCAA repeat tracts in lex2A, DNA fragments generated by PCR amplification with a biotinylated form of primer lex2for and primer lex3' (Table 2) were denatured in 150 mM NaOH and single-stranded DNA was purified by the use of streptavidin-coated Dynabeads (Dynal) as described by High et al. (1993). The DNA obtained was sequenced by the dideoxy chain termination method using a modified T7 polymerase (Sequenase II kit: United States Biochemical Corp.) and primer lex2seq (Table 2).

DNA sequences were collated and analysed using Chromas (Conar McCarthy, Australia) and the GCG GAP program (Devereux et al., 1984). Sequence homologies were determined using the GenBank DNA and protein sequence databases through the National Centre for Biotechnology Information BLAST network server (Altschul et al., 1997).

Generation of a lex2B chromosomal mutant in H. influenzae.
A kanamycin-resistance (kanR) cassette, released by digestion of pUC4K (Pharmacia) with HincII, was inserted at a unique site cut by StyI then end-filled with Klenow enzyme and dNTPs (Sambrook et al., 1989), located 209 bp downstream of the initiation codon of lex2B in plasmid pCR2.1lex2. This knockout construct was transformed into H. influenzae strains RM7004 and RM153; correct transformants were confirmed by PCR amplification and Southern analysis.

Construction of lex2–lacZ fusion reporter strains.
To engineer lacZ fusion constructs the following oligonucleotides were used with restriction sites (underlined) incorporated (HindIII site in PR1, KpnI site in PR2 and PR3, XhoI site in PR4 and PR6, EagI site in PR5 and PR7): PR1 (5'-TATCCCAAGCTTGATTCAGTTTGGTTTGCAGGA-3'), PR2 (5'-TCGGGGTACCTGATGTACTATTTAAGTCACTCT-3'), PR3 (5'-TAGTCGGGGTACCAACATAATCTCTCCATAGTT-3'), PR4 (5'-AATGCACTCGAGGAGATTATGTTTATTACACCT-3'), PR5 (5'-GCGTGCCGGCCGTTTCTCATAAATAGTTGATTC-3'), PR6 (5'-AATGCACTCGAGATAGTAGACAAAGCAGTTG-3') and PR7 (5'-GCGTGCCGGCCGAGCAAAATCTCAAGGTAGAG-3'). A previously described plasmid vector, pGZ-MCS (De Bolle et al., 2000), was used as a source of the lacZ gene for reporter gene constructs. PR1 and PR2 were used to amplify a region of chromosomal DNA from strain RM7004 from 547 bp upstream of the initiation codon of lex2A (within purL) to the nucleotide immediately adjacent to and downstream of the final repeat of lex2A. The 741 bp PCR product was digested with HindIII/KpnI then cloned into HindIII/KpnI-digested pGZ-MCS to make clone pGZPr1.2. The 3' end of the KpnI site in pGZ-MCS forms the second codon of the lacZ gene. PR4 and PR5 were used to amplify by PCR a region of chromosomal DNA from strain RM7004 from 6 bp downstream of the stop codon of lex2A to 93 bp upstream of the stop codon of lex2B. The 678 bp PCR product digested with EagI/XhoI was ligated into EagI/XhoI-digested pGZPr1.2 generating clone pGZPr1.2.4.5 The kanR gene derived from pUC4K by digestion with BamHI was cloned into the BamHI site of pGZPr1.2.4.5, located immediately downstream of lacZ and upstream of the XhoI site. The clones obtained, designated pGZPr1.2.4.5k, were verified by DNA sequencing.

A second construct was prepared in a similar manner to pGZPr1.2.4.5k with lacZ fused in-frame with the ATG initiation codon of lex2B. PR1 and PR3 were used to amplify by PCR a region of chromosomal DNA from strain RM7004 from 547 bp upstream of the initiation codon of lex2A to 2 nucleotides downstream of the initiation codon of lex2B. The 872 bp HindIII/KpnI-digested PCR product was cloned into HindIII/KpnI-digested pGZ-MCS, generating clone pGZPr1.3. PR6 and PR7 were used to amplify by PCR a region of chromosomal DNA from strain RM7004 from 16 bp downstream of the stop codon of lex2B to 62 bp downstream of the initiation codon of HI0755. The 671 bp XhoI/EagI-digested PCR product was cloned into XhoI/EagI-digested pGZPr1.3. The kanR gene was inserted into the BamHI site in an identical manner to that described above to give clone pGZPr1.3.6.7k.

Constructs pGZPr1.2.4.5k and pGZPr1.3.6.7k were transformed into strain RM7004 to give strains RM7004lex2AlacZ and RM7004lex2BlacZ respectively.

Construction of non-polar lex2A mutant.
A lex2A construct with the repeats removed, leaving a single BglII restriction enzyme site and retaining the majority of the reading frame in-frame with the start codon, has been engineered as described elsewhere (De Bolle et al., 2000). To interrupt the lex2A reading frame, the BglII site was digested, the ends were filled by Klenow enzyme and dNTPs (Sambrook et al., 1989), then the plasmid was religated. This procedure adds 4 bp to the reading frame, making a frameshift mutation. The construct, pRM7004{Delta}GCAAlex2AEF, was transformed into strain RM7004 to produce strain RM7004{Delta}GCAAlex2A- containing a non-polar mutation of lex2A.

RNA isolation and transcript analysis.
Total RNA was isolated from strain RM7004 using the Promega SV Total RNA Isolation kit as detailed by the supplier. An extra DNase treatment was included as the last step of the isolation procedure. cDNA was prepared from 0·5 µg RNA using MLV reverse transcriptase (Gibco-BRL) and random primers (Promega) under conditions recommended by the supplier. PCR was performed under the conditions described above and using 1 µl reverse-transcribed (RT) RNA as template. Controls were carried out using primers frdBup (5'-CTTATCGTTGGTCTTGCCGT-3') and frdBdwn (5'-TTGGCACTTTCCACTTTTCC-3') to amplify frdB, a constitutively expressed metabolic gene.

LPS analysis.
For colony immunoblotting, a single colony of the appropriate H. influenzae strain was diluted in PBS and dilutions spread on agar plates to obtain single colonies. Following overnight growth, colonies were transferred to nitrocellulose membranes then incubated with hybridoma culture supernatants containing LPS-specific mAbs 4C4 or 5G8 (provided by E. J. Hansen, University of Texas), as described by Roche et al. (1994).

For gel fractionation and Western immunoblotting, whole-cell lysates were prepared from overnight cultures as described by Serino & Virji (2000). The lysates were fractionated by Tricine-sodium dodecyl sulphate polyacrylamide gel electrophoresis (T-SDS-PAGE) (Lesse et al., 1990) and LPS profiles were detected by staining with silver (Quicksilver; Amersham) (Roche et al., 1994). LPS was transferred to nitrocellulose membranes for Western analysis (Towbin et al., 1979) then incubated with mAb 4C4 as described by Weiser et al. (1997).

Determination of LPS structure.
Water-insoluble LPS was extracted using the hot-aqueous phenol method of Westphal & Jann (1965) and O-deacylated LPS was then prepared for analysis by ESI-MS on a VG Quattro mass spectrometer (Micromass) as described by Masoud et al. (1997). NMR analysis was carried out as described by Lysenko et al. (2000).

Investigation of phase variation of strains RM7004lex2AlacZ and RM7004lex2BlacZ.
Three single blue colonies of the appropriate strain from a fresh plate were independently serially diluted in PBS to obtain between 500 and 1000 c.f.u. per plate. After growth on plates including X-Gal, one single blue and one single white colony derived from each of the three founder colonies were isolated then serially diluted in PBS to yield approximately 1500 c.f.u. per plate. The proportion of blue and white variants on each plate was established.


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The lex2 locus and its sequence in strains of H. influenzae
Amplification by PCR using locus-specific primers and Southern analysis showed the lex2 locus to be present in 20 out of 27 (74 %) diverse capsulate strains of H. influenzae investigated. These strains of clinical origin collectively represent all six serotypes (a through f) of this organism (Musser et al., 1990) (Table 1). The number of 5'-GCAA repeats within the lex2A gene was investigated by further PCR amplification and DNA sequencing and was found to vary between strains from four to forty-six copies (Table 1). No correlation of repeat number with capsular type was evident.

The chromosomal regions encompassing the lex2 locus in H. influenzae strains RM7004 and RM153 were cloned. The 1·9 kb EcoRI fragments in plasmids pB7004Elex2 and pB153Elex2 were sequenced for each strain (Fig. 2). The lex2A and lex2B ORFs of strain RM7004 (GenBank AF503507) were 282 bp and 745 bp in length respectively, separated by 13 bp of intergenic sequence. A single putative promoter sequence was found upstream of lex2A and sequences exhibiting partial hyphenated dyad symmetry, which may play a role in transcriptional termination, were identified downstream of the putative translational termination codon of lex2B, as reported previously (Jarosik & Hansen, 1994). These features suggest that lex2A and lex2B may form an operon. The nucleotide sequences of lex2 from our serotype b strains and strain DL42 were more than 99 % identical. However, 12 and 20 5'-GCAA repeats were identified in lex2A, 65 bp downstream of the initiation codon, of strains RM7004 and RM153 respectively, instead of the 18 repeats found in strain DL42. Thus, strain RM7004 contained a permissive, and strain RM153 a non-permissive, number of repeats for correct translation of lex2A. Comparisons of the deduced amino acid sequences of lex2A and lex2B with those of other proteins in the public databases revealed sequence homology for Lex2B only. Lex2B showed similarity to a number of glycosyltransferase enzymes from other organisms, the highest homology being 60 % similarity to Lob1 from Haemophilus somnus (Inzana et al., 1997).



View larger version (17K):
[in this window]
[in a new window]
 
Fig. 2. (A) The equivalent regions of the chromosome of strain Rd (top) and that deduced in the lex2-containing EcoRI clone and one end of the overlapping PstI clone derived from strain RM7004 (bottom), drawn to scale. E and P represent EcoRI and PstI sites respectively. Boxes represent genes; for strain Rd the appropriate gene (HI) number is given. Each initiation codon (represented by {lceil} and {rciel}) and each stop codon (represented by |) is shown. The percentage of nucleic acid identity of homologous regions is indicated. The nucleotide positions at which homologous recombination has occurred are shown by arrows. (B) Schematic representation of plasmid constructs pGZPr1.2.4.5k and pGZPr1.3.6.7k, used to make strains RM7004lex2AlacZ and RM7004lex2BlacZ respectively. The lex2-containing region of the chromosome is shown above each construct. Boxes represent genes and the repeat tract of lex2A is shown as a striped box. Initiation codons are indicated by {lceil} or and stop codons are represented by |. Arrows showing their orientation indicate the primers used for PCR amplification. The fusion between the coding sequence of lacZ and both lex2A and lex2B occurs at the KpnI site. The kanR gene is in the opposite transcriptional direction relative to lacZ. The constructs include sequence upstream of lex2A and downstream of lex2B to facilitate recombination.

 
The sequence of the 1·9 kb EcoRI fragments from pB7004Elex2 and pB153Elex2 included 821 nucleotides upstream of lex2A but only 68 nucleotides of flanking sequence downstream of lex2B. A further 688 nucleotides downstream of lex2B were determined after sequencing one end of the insert of the PstI clone of RM7004 (pB7004Plex2). The DNA sequences flanking lex2 were searched against the Rd genome sequence. Upstream of lex2A the clones contained sequence that matched that of the purL gene (HI0752), encoding an enzyme involved in purine biosynthesis, and the first part of the intergenic sequence immediately downstream of purL. The non-coding sequence immediately downstream of lex2B had no significant homology; however, the coding sequence downstream of that matched the gene HI0755 in strain Rd. In place of the lex2 locus in strain Rd were the genes HI0753 and HI0754. Both of these genes and HI0755 encode hypothetical proteins. Fig. 2 shows the relevant regions of the chromosome of strains Rd and RM7004.

The lex2 locus contained 71 mol% A+T (including the 5'-GCAA repeat region), in contrast to genes HI0753 and HI0754 which, like the mean composition of the Rd genome sequence, contain 61 mol% A+T. This would be consistent with a hypothesis that the lex2 locus has been laterally transferred from another species and has integrated into the genome by homologous recombination with the resultant loss of HI0753 and HI0754.

Transcription and translation of lex2A and lex2B
The position of lex2A and lex2B after DNA sequence analysis predicted that the two genes would be co-transcribed. Using primer pairs lex5'/lex3' and oligo1/lex3', weak but reproducible bands (0·8 kb and 0·4 kb respectively) were obtained by RT-PCR of RNA isolated from strain RM7004 (data not shown), confirming that the two genes are transcribed together. No bands were obtained in PCR controls using total RNA as a template.

To investigate the possible interdependence of translation between the closely sited reading frames within the lex2 locus, translational fusions of both lex2A and lex2B to a {beta}-galactosidase (lacZ) reporter gene were constructed. In plasmid pGZPr1.2.4.5k, the lacZ gene was fused in-frame with the twelfth and final repeat of the 5'-GCAA tract of the strain RM7004 lex2A sequence. The lacZ ORF could only be translated from the lex2A initiation codon (Fig. 2B). Plasmid pGZPr1.3.6.7k was prepared with the lacZ gene fused in-frame with, and adjacent to, the initiation codon of lex2B (Fig. 2B). The plasmid constructs, pGZPr1.2.4.5k and pGZPr1.3.6.7k, were then used to transform strain RM7004 to generate strains RM7004lex2AlacZ and RM7004lex2BlacZ respectively. Several independent blue or white colonies of transformed strains RM7004lex2AlacZ and RM7004lex2BlacZ after growth on media containing X-Gal were each dispersed then replated. The former strain showed clear evidence of phase variation. The mean proportion of white variants generated from a blue founder colony was calculated to be 1 in 571 whilst the mean proportion of blue variants generated from a white founder colony was 1 in 1263. Upon plating, strain RM7004lex2BlacZ gave rise to only blue colonies. No white phase variants were detected in 10 000 colonies screened, consistent with lex2B being translated independently of lex2A.

Analysis of lex2B function
lex2B, encoding a predicted glycosyltransferase, was inactivated in strains RM7004 and RM153 to give mutants RM7004lex2B and RM153lex2B respectively. Upon immunoblotting, colonies of both mutants and the corresponding wild-type strains demonstrated no reactivity with mAb 5G8. This is contrary to the variable reactivity in colony phenotype demonstrated by other strains when reacted with mAb 5G8, in previous studies (Jarosik & Hansen, 1994). Differences in the pattern of reactivity between strains were observed with mAb 4C4; this antibody binds a digalactoside moiety on H. influenzae LPS (Virji et al., 1990). The majority of colonies of wild-type strain RM7004 (>98 %) reacted (R) with mAb 4C4 and a negative (off; O) phenotype was represented at a level of less than 1 % (Fig. 3A). Occasional stronger-reacting (S) colonies were also observed at a frequency of less than 0·5 %. Colonies of strain RM7004lex2B, derived from several independent transformants, showed only the R and O phenotypes upon reaction with mAb 4C4, with the majority of colonies being of the R phenotype (Fig. 3B). A majority of colonies of both strains RM153 and RM153lex2B were of the O phenotype when reacted with mAb 4C4 and 0·2 % were of an R phenotype (Fig. 3C, D). Thus, insertional inactivation of lex2B in strain RM7004 abolished the ability of this strain to generate the LPS phenotype, allowing the highest levels of mAb 4C4 binding, presumably as a result of reduced digalactoside incorporation into the LPS.



View larger version (92K):
[in this window]
[in a new window]
 
Fig. 3. Analysis of the LPS of strains RM7004 (A), RM7004lex2B (B), RM153 (C) and RM153lex2B (D) by colony immunoblotting with mAb 4C4. Reactive (R), strongly reactive (S) and off (O) phenotypes of stained colonies are indicated by arrows.

 
LPS from strain RM7004lex2B was analysed by SDS-PAGE and Western immunoblotting and compared to that of the parent strain. The lex2B mutant exhibited loss of the slowest migrating bands upon electrophoresis, indicating that some LPS glycoforms were truncated (Fig. 4A). Such truncation would be predicted to represent the loss of two or three sugars from the LPS molecule (Masoud et al., 2003). If one band on a gel is considered as one glycoform, Western immunoblots using mAb 4C4 showed binding of this antibody to one glycoform only of the lex2B mutant compared with reactivity to four LPS glycoforms of the parent strain (Fig. 4B). The presence of only one mAb 4C4-reactive LPS glycoform for strain RM7004lex2B is consistent with the finding that only one mAb 4C4-reactive phenotype was detected upon colony immunoblotting. This might suggest that this strain is capable of synthesizing LPS glycoforms containing only one digalactoside epitope and not the higher molecular mass glycoforms containing two digalactosides that have been shown for the parent strain (Fig. 1).



View larger version (74K):
[in this window]
[in a new window]
 
Fig. 4. Comparison of the LPS from strain RM7004 and derived mutants after fractionation by T-SDS-PAGE and staining with silver (A) and Western blotting with mAb 4C4 (B). Strains: lane 1, RM7004; lane 2, RM7004lex2B; lane 3, RM7004lex2BlacZ; lane 4, RM7004lex2AlacZ; lane 5, RM7004{Delta}GCAAlex2A-; lane 6, RM153. Arrows indicate the proposed equivalent bands between the gel and blot.

 
To characterize alterations in the LPS of strain RM7004lex2B, mass spectrometry (ESI-MS) analysis was performed on O-deacylated LPS of the mutant and compared to data available for the parent strain (Masoud et al., 2003). The largest glycoforms elaborated by the mutant strain consisted of only six hexose sugars (6-Hex) in the outer core (Table 3), compared to the largest glycoform of the parent strain, which comprised nine hexose residues (9-Hex) (Masoud et al., 2003). The variable presence of an additional phosphoethanolamine (PEtn) moiety was also observed for the mutant O-deacylated LPS due to the commonly encountered incomplete substitution of the Kdo-P moiety with a PEtn residue.


View this table:
[in this window]
[in a new window]
 
Table 3. Negative-ion ESI-MS data and proposed compositions of O-deacylated LPS from H. influenzae RM7004lex2B

Mean mass units were used for calculation of molecular mass based on proposed composition as follows: Lipid A, 952·00; Hex, 162·15; HexNAc, 203·19; Hep, 192·17; Kdo-P, 300·16; PEtn, 123·05. ChoP, 165·05; P, 79·98.

 
The O-deacylated LPS of strain RM7004lex2B was further analysed by 1H-NMR spectroscopy, which produced a well-resolved spectrum confirming the 6-Hex structure as the largest observed glycoform like that of the LPS of strain RM153 (Masoud et al., 1997) (Fig. 1). Confirmation of the role of Lex2B was deduced from analysis of the chemical shifts for the glucose residue (GlcI) attached at the first heptose moiety (HepI) (Table 4). For higher LPS glycoforms (7- to 9-Hex) of the RM7004 parent strain the proton resonances for GlcI demonstrated chemical shifts indicative of substitution of this residue at the 4-position (Masoud et al., 2003) (Fig. 1). However, for the RM7004lex2B LPS the chemical shifts for the proton resonances indicated that GlcI was a terminal residue (Table 4). A NOESY spectrum confirmed the identity of GlcI in both parent and lex2B mutant LPS due to assignment of diagnostic inter-residue NOE connectivities to the H-4 and H-6 proton resonances of HepI in each case (Table 4). ESI-MS and NMR analyses therefore lead us to conclude that lex2B encodes a {beta}-glucosyltransferase that adds the second {beta}-glucose in a 1,4 linkage to the first {beta}-glucose extending from the proximal heptose of the LPS of strain RM7004.


View this table:
[in this window]
[in a new window]
 
Table 4. 1H NMR assignment of glucose residue (GlcI) of H. influenzae RM7004 and RM7004lex2B O-deacylated LPS

The spectrum was recorded at 25 °C relative to HOD at 4·77 p.p.m.

 
Confirming the results of phenotypic analysis of the LPS of strain RM153lex2B, no alteration in the structure of the LPS of this strain was found when compared to wild-type upon both ESI-MS and NMR analyses.

Analysis of the role of the lex2A gene
Strain RM7004lex2AlacZ contains a mutated lex2A gene through deletion of a majority of the reading frame and insertion of the lacZ gene but retains an intact lex2B gene. SDS-PAGE and colony immunoblot analyses showed that the LPS from this strain was equivalent to that of strain RM7004lex2B (Fig. 4A). Upon reacting with mAb 4C4, no S variants of RM7004lex2AlacZ were detected amongst 10 000 colonies screened. Due to the close proximity of the two genes, there remained a possibility that there was a polar effect from the disrupted lex2A gene on transcription of lex2B. Localized disruption of the coding sequence of lex2A was introduced into strain RM7004 using plasmid pRM7004{Delta}GCAAlex2AEF. This construct was generated by replacing the repeat tract of lex2A with a BglII site then digesting with BglII and filling the recessed ends to introduce a frameshift once religated. This renders lex2A out of frame whilst minimizing any direct polar effect on lex2B transcription. Removal of the 5'-GCAA repeats from lex2A does not alter lex2 function (data not shown). Upon SDS-PAGE (Fig. 4) and colony immunoblotting (data not shown), LPS from this strain, RM7004{Delta}GCAAlex2A-, behaved in an equivalent manner to that of strain RM7004lex2AlacZ described above. These data were consistent with our proposal that lex2A is somehow necessary for the phenotypic expression of Lex2B activity.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The LPS of H. influenzae is marked by extensive intra-strain heterogeneity generated by frequent on–off translational switching in the expression of LPS biosynthetic genes containing tracts of repeated DNA. We set out to characterize the function of the lex2 locus, comprising lex2B, which forms a putative operon with a short upstream gene, lex2A, which contains tetranucleotide repeat units. The lex2 locus is absent from the published genome sequence of H. influenzae strain Rd (Fleischmann et al., 1995). By cloning and sequencing the chromosomal region encompassing lex2 from strains RM7004 and RM153 we were able to locate the relative position of lex2 in the chromosome of H. influenzae. We postulate that lex2 has recombined into the genome after lateral transfer from another species. The closest homologue of Lex2B amongst a number of glycosyltransferases is Lob1 of H. somnus. The encoding gene in this organism is extended when compared to lex2B and contains 5'-CAAT repeats upstream of the region of homology and in an equivalent position to the lex2A reading frame (Inzana et al., 1997).

Mutation of lex2B in strain RM7004 resulted in the synthesis of truncated LPS that was probably altered in its digalactoside content, as shown by the reduction in the level of reactivity of glycoforms to mAb 4C4. ESI-MS analysis of the LPS of strain RM7004lex2B confirmed the loss of three hexose sugars. The preferred order of LPS oligosaccharide biosynthesis in strain RM7004 is substitution of each heptose by a single hexose sugar followed by addition of further hexoses from the second heptose to complete a Gal-{alpha}-1-4-Gal-{beta}-1-4-Glc-{beta}-1-4-Glc-{alpha} extension (Masoud et al., 2003; J. C. Richards, unpublished). This occurs before adding Gal-{alpha}-1-4-Gal-{beta}-1-4-Glc-{beta}-1 to the {beta}-glucose linked to the first heptose to complete the fully extended LPS (Masoud et al., 2003). NMR analysis of LPS from the lex2B mutant confirmed that the terminal globoside trisaccharide was absent from the HepI oligosaccharide extension, and that the glucose linked to the first heptose was now a terminal residue. Thus, we have convincing evidence for the role of Lex2B as the glycosyltransferase that adds the second {beta}-glucose to the oligosaccharide extending from the first heptose of the LPS of strain RM7004.

lex2 is the fifth of the published H. influenzae phase-variable LPS-associated loci to have its function elucidated; a majority of these genes encode glycosyltransferases. In the original study by Jarosik & Hansen (1994), the function of the lex2 locus was correlated with phase-variable reactivity of bacteria with mAb 5G8. In the present study, we did not find a useful correlation between our strains and binding of this antibody. However, mAb 4C4, previously reported to bind a digalactoside of H. influenzae LPS (Virji et al., 1990), did show strain-dependent phase-variable expression of LPS epitopes dependent upon lex2 activity. The generation of more than one mAb 4C4-reactive phenotype for strain RM7004 noted in this study and reported by Roche et al. (1994) until now has not been understood. Our findings lead us to propose that the common reactive phenotype (R) results from the binding of mAb 4C4 to predominantly one digalactoside, presumed to be that extending from the second heptose, and that the S phenotype is generated by the binding of mAb 4C4 to cells predominantly expressing both digalactosides extending from the first and second heptose sugars. In a large population of cells only a minority of glycoforms of strain RM7004 include two digalactosides (Masoud et al., 2003), presumably due to phase variation and other microheterogeneity associated with H. influenzae LPS synthesis. In contrast to strain RM7004, only a minority of variant colonies of strain RM153 react with mAb 4C4. Since lex2A is maintained predominantly out of frame in strain RM153, extension beyond the glucose attached to HepI is prohibited in this strain. Indeed, we have shown that strain RM153 variant colonies that gain reactivity to mAb 4C4 and contain a lex2A gene with an in-frame number of repeats (21 copies of 5'-GCAA) exhibit an LPS profile upon SDS-PAGE that resembles that of strain RM7004 (data not shown).

Previous studies have indicated that digalactoside expression on LPS enhances the ability of H. influenzae to escape the bactericidal activity of serum in in vitro assays, and facilitates intravascular survival in the infant rat model of infection (Kimura & Hansen, 1986; Kimura et al., 1987; Cope et al., 1991; Maskell et al., 1992; Hood et al., 1996). We anticipate that organisms expressing lex2, facilitating the extension of a second digalactoside on the LPS, may be selected for during systemic infection. The digalactoside can mimic antigens found on the surface of a number of host cell types. Thus, the conserved presence of lex2 in serotype b strains may contribute to the virulence of these strains, those most frequently causing systemic and meningeal infection (Turk & May, 1967). Organisms contained in the cerebrospinal fluid of newly diagnosed cases of meningitis showed a majority (>99 %) of organisms binding mAb 4C4 compared to only <0·1 % when organisms were cultured in vitro (Weiser et al., 1989b). The importance to the organism of the digalactoside in the globoside extension from HepI might be indicated by the emphasis placed on the switching of its expression. Its assembly evidently requires the expression of three independent phase-variable loci, lex2, lic2A and lgtC, each adding sequential sugar residues.

Lex2B may also compete with alternative steps in LPS biosynthesis. Recently, a novel sialylated lacto-N-neotetraose structure has been reported in H. influenzae strains that contain the lex2 locus (Cox et al., 2002). The terminal four sugars of this structure [{alpha}-Neu5Ac-(2-3)-{beta}-D-gal-(1-4)-{beta}-D-glcnac-(1-3)-{beta}-D-Gal] are an alternative addition to the same {beta}-glucose attached to HepI.

What is unclear about the lex2 locus is the relationship between the glycosyltransferase, Lex2B, and its phase-variable expression apparently mediated by changes in the repeat numbers within lex2A. It had been previously postulated that the function of lex2B depends upon the successful translation of lex2A (Jarosik & Hansen, 1994). A functional relationship between the two genes has been confirmed by evidence that they are co-transcribed. It has been shown in Helicobacter pylori that expression of adjacent co-transcribed genes can co-phase vary through transcription/translational coupling (de Vries et al., 2002). However in our study, the separate in-frame fusions of the lacZ gene to lex2A and lex2B in strain RM7004 indicated that these two genes are translated independently of each other. The LPS from strain RM7004lex2AlacZ appeared different from wild-type and equivalent to that examined in detail from strain RM7004lex2B. This alteration in LPS structure is unlikely to be due only to a polar effect of the lex2A mutation on lex2B as introduction of a non-polar mutation in lex2A resulted in the same LPS phenotype as observed for strain RM7004lex2B. It remains a possibility that translation of lex2B is influenced by translation of lex2A through reinitiation of translation at the initiation codon of lex2B by ribosomes that have completed translation of lex2A, or that message stability is altered through failure to translate the first reading frame. However, the uniform colouration of blue colonies of strain RM7004lex2BlacZ grown on medium containing X-Gal indicated a consistent and constitutive translation of this gene. No clue to Lex2A function can be discerned through database searches; its contribution to the function of the lex2 locus requires further investigation.

In summary, lex2B is required for the addition of the second {beta}-glucose in the oligosaccharide extension from the first heptose of the LPS of H. influenzae strain RM7004. lex2 has a role in the phase-variable expression of H. influenzae LPS and its activity is key to the extension of a digalactoside that may be associated with enhanced survival of the organism during systemic infection.


   ACKNOWLEDGEMENTS
 
The authors thank Gaynor Randle for her assistance in RT-PCR experiments and preparing RNA.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Alexander, H. E. (1965). The Haemophilus Group. London: Pitman Medical Publishing.

Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25, 3389–3402.[Abstract/Free Full Text]

Cope, L. D., Yogev, R., Mertsola, J., Latimer, J. L., Hanson, M. S., McCracken, G. H., Jr & Hansen, E. J. (1991). Molecular cloning of a gene involved in lipooligosaccharide biosynthesis and virulence expression by Haemophilus influenzae type b. Mol Microbiol 5, 1113–1124.[Medline]

Cox, A. D., Hood, D. W., Martin, A., Makepeace, K. M., Deadman, M. E., Li, J., Brisson, J.-R., Moxon, E. R. & Richards, J. C. (2002). Identification and structural characterisation of a sialylated lacto-N-neotetraose structure in the lipopolysaccharide of Haemophilus influenzae. Eur J Biochem 269, 4009–4019.[Abstract/Free Full Text]

De Bolle, X., Bayliss, C. D., Field, D., van de Ven, T., Saunders, N. J., Hood, D. W. & Moxon, E. R. (2000). The length of a tetranucleotide repeat tract in Haemophilus influenzae determines the phase variation rate of a gene with homology to type III DNA methyltransferases. Mol Microbiol 35, 211–222.[CrossRef][Medline]

Devereux, J., Haeberli, P. & Smithies, O. (1984). A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res 12, 387–395.[Abstract]

de Vries, N., Duinsbergen, D., Kuipers, E. J., Pot, R. G. J., Weisenekker, P., Penn, C. W., van Vliet, A. H. M., Vandenbroucke-Grauls, C. M. J. E. & Kusters, J. G. (2002). Transcriptional phase variation of a type III restriction-modification system in Helicobacter pylori. J Bacteriol 184, 6615–6623.[Abstract/Free Full Text]

Fleischmann, R. D., Adams, M. D., White, O. & 37 other authors (1995). Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269, 496–512.[Medline]

Herriott, R. M., Meyer, E. Y., Vogt, M. & Modan, M. (1970). Defined nongrowth media for stage II development of competence in Haemophilus influenzae. J Bacteriol 101, 517–524.[Medline]

High, N. J., Deadman, M. E. & Moxon, E. R. (1993). The role of a repetitive DNA motif (5'-CAAT-3') in the variable expression of the Haemophilus influenzae lipopolysaccharide epitope alpha Gal(1–4)beta Gal. Mol Microbiol 9, 1275–1282.[Medline]

Hood, D. W. & Moxon, E. R. (1999). Lipopolysaccharide phase variation in Haemophilus and Neisseria. In Endotoxin in Health and Disease. Edited by H. Brade, S. Opal, S. Vogel & D. Morrison. New York: Marcel Dekker.

Hood, D. W., Deadman, M. E., Jennings, M. P., Bisercic, M., Fleischmann, R. D., Venter, J. C. & Moxon, E. R. (1996). DNA repeats identify novel virulence genes in Haemophilus influenzae. Proc Natl Acad Sci U S A 93, 11121–11125.[Abstract/Free Full Text]

Hood, D. W., Cox, A. D., Gilbert, M. & 11 other authors (2001). Identification of a lipopolysaccharide alpha-2,3-sialyltransferase from Haemophilus influenzae. Mol Microbiol 39, 341–350.[CrossRef][Medline]

Inzana, T. J., Hensley, J., McQuiston, J., Lesse, A. J., Campagnari, A. A., Boyle, S. M. & Apicella, M. A. (1997). Phase variation and conservation of lipooligosaccharide epitopes in Haemophilus somnus. Infect Immun 65, 4675–4681.[Abstract]

Jarosik, G. P. & Hansen, E. J. (1994). Identification of a new locus involved in expression of Haemophilus influenzae type b lipooligosaccharide. Infect Immun 62, 4861–4867.[Abstract]

Kimura, A. & Hansen, E. J. (1986). Antigenic and phenotypic variations of Haemophilus influenzae type b lipopolysaccharide and their relationship to virulence. Infect Immun 51, 69–79.[Medline]

Kimura, A., Patrick, C. C., Miller, E. E., Cope, L. D., McCracken, G. H., Jr & Hansen, E. J. (1987). Haemophilus influenzae type b lipooligosaccharide: stability of expression and association with virulence. Infect Immun 55, 1979–1986.[Medline]

Lesse, A. J., Campagnari, A. A., Bittner, W. E. & Apicella, M. A. (1990). Increased resolution of lipopolysaccharides and lipooligosaccharides utilising tricine-sodium dodecyl sulphate-polyacrylamide gel electrophoresis. J Immunol Methods 126, 109–117.[CrossRef][Medline]

Levinson, G. & Gutman, G. A. (1987). Slipped-strand mispairing: a major mechanism for DNA sequence evolution. Mol Biol Evol 4, 203–221.[Abstract]

Lysenko, E., Richards, J. C., Cox, A. D., Stewart, A., Martin, A., Kapoor, M. & Weiser, J. N. (2000). The position of phosphorylcholine on the lipopolysaccharide of Haemophilus influenzae affects binding and sensitivity to C-reactive protein mediated killing. Mol Microbiol 35, 234–245.[CrossRef][Medline]

Maskell, D. J., Szabo, M. J., Deadman, M. E. & Moxon, E. R. (1992). The gal locus from Haemophilus influenzae: cloning, sequencing and the use of gal mutants to study lipopolysaccharide. Mol Microbiol 6, 3051–3063.[Medline]

Masoud, H., Moxon, E. R., Martin, A., Krajcarski, D. & Richards, J. C. (1997). Structure of the variable and conserved lipopolysaccharide oligosaccharide epitopes expressed by Haemophilus influenzae serotype b strain Eagan. Biochemistry 36, 2091–2103.[CrossRef][Medline]

Masoud, H., Martin, A., Thibault, P., Moxon, E. R. & Richards, J. C. (2003). Structure of extended lipopolysaccharide glycoforms containing two globotriose units in the Haemophilus influenzae serotype b strain, RM7004. Biochemistry 42, 4463–4475.[CrossRef][Medline]

Musser, J. M., Kroll, J. S., Granoff, D. M. & 25 other authors (1990). Global genetic structure and molecular epidemiology of encapsulated Haemophilus influenzae. Rev Infect Dis 12, 75–111.[Medline]

Preston, A., Maskell, D., Johnson, A. & Moxon, E. R. (1996). Altered lipopolysaccharide characteristic of the I69 phenotype in Haemophilus influenzae results from mutations in a novel gene, isn. J Bacteriol 178, 396–402.[Abstract]

Roche, R. J., High, N. J. & Moxon, E. R. (1994). Phase variation of Haemophilus influenzae lipopolysaccharide: characterisation of lipopolysaccharide from individual colonies. FEMS Microbiol Lett 120, 279–284.[CrossRef][Medline]

Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.

Serino, L. & Virji, M. (2000). Phosphorylcholine decoration of lipopolysaccharide differentiates commensal Neisseriae from pathogenic strains: identification of licA-type genes in commensal Neisseriae. Mol Microbiol 35, 1550–1559.[CrossRef][Medline]

Towbin, H., Staehelin, T. & Gordon, J. (1979). Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Biotechnology 24, 145–149.

Turk, D. C. & May, J. R. (1967). Haemophilus influenzae, its Clinical Importance. London: The English Universities Press.

Virji, M., Weiser, J. N., Lindberg, A. A. & Moxon, E. R. (1990). Antigenic similarities in lipopolysaccharides of Haemophilus and Neisseria and expression of a digalactoside structure also present on human cells. Microb Pathog 9, 441–450.[Medline]

Weiser, J. N., Shchepetov, M. & Chong, S. T. (1997). Decoration of lipopolysaccharide with phosphorylcholine: a phase-variable characteristic of Haemophilus influenzae. Infect Immun 65, 943–950.[Abstract]

Weiser, J. N., Lindberg, A. A., Manning, E. J., Hansen, E. J. & Moxon, E. R. (1989a). Identification of a chromosomal locus for expression of lipopolysaccharide epitopes in Haemophilus influenzae. Infect Immun 57, 3045–3052.[Medline]

Weiser, J. N., Love, J. M. & Moxon, E. R. (1989b). The molecular mechanism of phase variation of H. influenzae lipopolysaccharide. Cell 59, 657–665.[Medline]

Westphal, O. & Jann, K. (1965). Bacterial lipopolysaccharides. Extraction with phenol-water and further applications of the procedure. Methods Carbohydr Chem 5, 83–91.

Received 2 April 2003; revised 18 July 2003; accepted 29 July 2003.



This Article
Abstract
Full Text (PDF)
Alert me when this article is cited
Alert me if a correction is posted
Citation Map
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Download to citation manager
Google Scholar
Articles by Griffin, R.
Articles by Hood, D. W.
Articles citing this Article
PubMed
PubMed Citation
Articles by Griffin, R.
Articles by Hood, D. W.
Agricola
Articles by Griffin, R.
Articles by Hood, D. W.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
INT J SYST EVOL MICROBIOL MICROBIOLOGY J GEN VIROL
J MED MICROBIOL ALL SGM JOURNALS
Copyright © 2003 Society for General Microbiology.