Experimentally revised repertoire of putative contingency loci in Neisseria meningitidis strain MC58: evidence for a novel mechanism of phase variation

Authors


  • Present addresses: The Wellcome Trust Sanger Institute, Hinxton, Cambridge, CB10 1SA; Department of Biology and Biochemistry, University of Bath, Bath, BA2 7AY.

E-mail pmartin@molbiol.ox.ac.uk; Tel. (+44) 1865 222347; Fax (+44) 1865 222626.

Summary

Analysis of the genome sequence of Neisseria meningitidis strain MC58 revealed 65 genes associated with simple sequence repeats. Experimental evidence of phase variation exists for only 14 of these 65 putatively phase variable genes. We investigated the phase variable potential of the remaining 51 genes. The repeat tract associated with 20 of these 51 genes was sequenced in 26 genetically distinct strains. This analysis provided circumstantial evidence for or against the phase variability of the candidate genes, based on the sequence and the length of the repeated motif. These predictions of phase variability were substantiated for three of these candidate genes using colony immunoblotting or β-galactosidase as a reporter. This investigation identified a novel phase variable gene (NMB1994 or nadA) associated with a repeat tract (TAAA) not previously reported to be associated with phase variable genes in N. meningitidis. Analysis of the nadA transcript revealed that the repeat tract was located upstream of the putative −35 element of the nadA promoter. Semiquantitative RT-PCR showed that variation in the number of repeats was associated with changes in the level of expression of nadA, findings consistent with a model whereby the variable number of (TAAA) repeats modulates the promoter strength.

Introduction

The survival of obligate human commensal and pathogenic bacteria depends upon colonization of specific tissues, transmission between hosts and evasion of host clearance mechanisms. This multiplicity of behaviours depends, at least in part, on surface antigens and the generation of phenotypic variants (van Belkum et al., 1998; Oliver et al., 2000). One mechanism mediating the variability of surface structures is phase variation, i.e. the reversible switching of phenotype as a consequence of changes at the level of the genotype. It can be mediated for example by DNA methylation, DNA rearrangement or slipped strand mispairing (reviewed in Henderson et al., 1999).

Neisseria meningitidis, a Gram-negative encapsulated bacterium predominantly found in the human nasopharynx, is a common constituent of the normal microbial flora. However, some strains are invasive and can enter first the bloodstream and then the cerebrospinal fluid, causing septicaemia and meningitis. In N. meningitidis, phase variation is associated with reversible changes within repeated simple sequence DNA motifs (microsatellites) located in coding or promoter regions of genes involved in the biosynthesis of surface antigens. These simple sequence repeats were defined as identical reiterated motifs less than 10 bases in length (Saunders et al., 2000). The length of the homopolymeric or heteropolymeric tracts, i.e. the number of tandemly repeated motifs, can be modified during replication through slipped strand mispairing and can consequently influence translation or transcription (Moxon et al., 1994; van der Ende et al., 1995; Hammerschmidt et al., 1996).

Analysis of the complete genome sequence of N. meningitidis serogroup B strain MC58 revealed 65 genes associated with repeat tracts, i.e. the highest number of genes associated with microsatellites described so far for any bacterium (Tettelin et al., 2000; Saunders et al., 2000). However, there is experimental evidence of phase variation for only 14 genes encoding opacity proteins (de Vries et al., 1996), lipopolysaccharide biosynthesis proteins (Jennings et al., 1999), proteins involved in the biosynthesis and modification of pili (Jonsson et al., 1991; Jennings et al., 1998), haemoglobin receptors (Lewis et al., 1999; Richardson and Stojiljkovic, 1999), PorA outer membrane protein (van der Ende et al., 2000), Opc outer membrane protein (Sarkari et al., 1994), capsular polysaccharide (Hammerschmidt et al., 1996) and ferric receptor (Biegel Carson et al., 2000).

The aim of this study was to obtain additional evidence for or against the phase variability of the remaining 51 genes. Based upon the fact that the repeat tracts associated with known phase variable genes display interstrain length polymorphism, repeat length polymorphism was investigated for a subset of 20 putative phase variable genes representing all repeat sequence types. This analysis provided circumstantial evidence supporting or not supporting phase variability based on the sequence and the length of the repeated motif. More direct evidence for or against the phase variability of three candidate genes was obtained through colony immunoblotting and specific antisera raised to the relevant proteins (Pizza et al., 2000) or using the β-galactosidase gene as a reporter. This analysis revealed a novel phase variable gene (NMB1994 or nadA) associated with a tetranucleotide repeat tract (TAAA) not previously reported to be associated with phase variation in N. meningitidis strain MC58. This repeat tract was precisely located in the promoter of the gene. We showed that the expression of nadA was regulated at the transcriptional level through a variation of the number of reiterated motifs present in the repeat tract. These data provided insights into the possible molecular mechanism resulting in the phase variable expression of nadA.

Results

Evidence for or against phase variability of putative phase variable genes by analysing sequence polymorphism of repeat tracts

Interstrain polymorphisms in the length of repeat tracts were investigated to provide circumstantial evidence of the phase variability of genes associated with simple sequence repeats. We determined, for each type of repeats, the threshold or the minimum number of repeated motifs leading to sequence polymorphism and therefore potentially to phase variation of the associated genes. Twenty candidate genes were selected to be representative of the diversity of simple sequence repeats encountered in the N. meningitidis strain MC58 genome (Table 1). Homopolymeric tracts of cytosines or guanines shorter than five bases, homopolymeric tracts of adenines or thymines shorter than eight bases, dinucleotide repeats and tetranucleotide repeats reiterated less than four and three times, respectively, were not encountered in strain MC58 genome. The only pentanucleotide repeat investigated is the longest pentanucleotide repeat tract present in strain MC58 genome, apart from the ones already described (Saunders et al., 2000). Five known phase variable genes (Table 1) were also integrated into the study.

Table 1.  Repeat-associated genes analysed in this study.
GeneaNMB
numberb
RepeatcFramedPresence in
strains (%)
Length
polymorphisme
Potential phase
variation
  • a. Name of the potentially phase variable gene based upon annotation of the Neisseria meningitidis strain MC58 genome (http://www.tigr.org). HYPO: hypothetical protein; FUN: protein of unknown function.

  • b

    . Open reading frame (ORF) designation in the annotation of the N. meningitidis strain MC58 genome.

  • c

    . Sequence of the repeat associated with the putative phase variable gene. The number of repeated units indicated is the number present in the strain MC58 genome sequence.

  • d

    . Location of the repeat. Pro: in the promoter region of the gene; In/Out: the number of repeats makes the ORF of the gene in strain MC58; In: in frame, allowing the synthesis of a mature protein; Out: out of frame, preventing the synthesis of a mature protein.

  • e

    . Lowest and highest numbers of repeated motifs observed between strains. – indicates no change in repeat length.

  • f

    . Previously characterized phase variable gene.

Homopolymeric tracts G/C
cvaA1783(C) 5Out 504–5Yes
cam2104(G) 6Out1005–6Yes
siaDf0067(C) 7In100No
HYPO2008(C) 7Pro 65No
lbpA1540(G) 8In 92No
FUN0415(G) 9Out 885–10Yes
sprA1998(C) 9Out 587–10Yes
hmbRf1668(G) 9Out 778–11Yes
spr1969(C) 10In 855–11Yes
fetAf1988(C) 11Pro 969–12Yes
pglAf0218(G) 11In1008–11Yes
lgtGf2032(C) 12Out 738–13Yes
Homopolymeric tracts A/T
mesJ1140(A) 8In100No
hsdM1223(A) 9In 54No
HYPO0065(T) 10Pro1008–10Yes
HYPO0368(A) 11In 968–11Yes
Dinucleotide repeats
D150182(AC) 4In100No
tra1277(TG) 4In100No
FUN0432(TA) 4In 58No
fixP1723(AT) 5In100No
Tetra- Penta- nucleotide repeats
aipI1985(GGCA) 3In 50No
FUN1893(TTCC) 4Out1003–12Yes
FUN1525(AAGC) 5Out1001–7Yes
nadA1994(TAAA) 9Pro 585–13Yes
funZ0961(CAAAT) 3In 38No

A DNA fragment was amplified by polymerase chain reaction (PCR) then sequenced for each of the 25 repeat tracts from a subset of 19 different N. meningitidis strains, selected to be representative of the genetic diversity of N. meningitidis from a reference collection of 107 strains, whose phylogenetic relationships have been determined by multilocus sequence typing (Maiden et al., 1998). Three N. gonorrhoeae and four N. lactamica strains were also included in the analysis (Table 2). Southern analysis was carried out to assess the presence or absence of the gene in the strains for which no PCR product was obtained.

Table 2.  Strains used in this study.
StrainSpeciesSerogroup
  1. Neisseria meningitidis strains were from a collection of strains representative of all major serogroups (Maiden et al., 1998).

MC58
NG6/88
BZ198
BZ169
BZ133 B
BZ83
NGE31
NGF26
SWZ107
H44/76Neisseria 
A22 meningitidisW135
F6124
D8 A
92001
106
90/18311  
D1 C
L93/4286
SB25
MS11N. gonorrhoeae 
150002
179008
L12
L13N. lactamica 
L18
L19

The study carried out with the 25 genes showed that the proportion of strains containing a given gene varied from 38% to 100% (Table 1). The sequence analysis revealed that 14 of the 25 genes displayed length polymorphism of the repeat tract (Table 1).

The presence or the absence of length polymorphisms was shown to depend on the sequence and length of the repetitive DNA identified in strain MC58 (Table 1). All but three homopolymeric tracts of guanine or cytosine residues studied exhibited polymorphisms. Three repeat tracts, seven cytosine residues in siaD (NMB0067) and NMB2008 and eight guanine residues in lbpA (NMB1540), did not display variation in their length. Homopolymeric tracts constituted by adenine or thymine residues greater than 10 bases in length showed length polymorphism, whereas shorter tracts of eight or nine bases were invariant in length. The four dinucleotide repeat tracts investigated, each comprising four or five reiterated units, were shown not to be polymorphic. Four or more repeated motifs of a tetranucleotide were indicative of length polymorphism in the strains studied. The pentanucleotide repeat analysed did not display any polymorphism (Table 1).

The demonstration of polymorphisms provided support for, whereas absence of polymorphism brought evidence against phase variability of the candidate genes (Table 1). The hypothesis that sequence polymorphism could have been detected if additional strains had been investigated can not be excluded.

The repeat types not previously reported to be associated with characterized phase variable genes in N. meningitidis and that showed length polymorphism in the present study are the homopolymeric tracts of adenine or thymine and tetranucleotides. Examples of these repeats are the polyA tract in NMB0368 and the (TAAA) tract in NMB1994 (Table 1).

Figure 1A shows the alignment of the nucleotide sequences determined for NMB0368, which is absent from N. meningitidis serogroup W135 strain A22, from N. gonorrhoeae strains MS11 and 150002 and from N. lactamica strain L19. The number of adenine residues in this homopolymeric tract varied from 8 to 11, with an additional substitution (AG) in N. meningitidis strains 90/18311, D1 and L93/4286. In addition to the variation of the repeat length, point mutations were detected outside of the homopolymeric tract (Fig. 1A). The deduced amino acid sequences (Fig. 1B) revealed that the variation of the number of repeats was mostly by units of three bases and although this made the number of lysine (K) residues vary it would be consistent with the synthesis of the full-length protein. However, the number of repeats (nine adenine residues) would be consistent with the translation of a truncated protein in strains SWZ107, F6124 and 92001. A similar homopolymeric tract located in the promoter region of NMB0065 was also shown to be polymorphic (Table 1).

Figure 1.

Comparison of the nucleotide (A) and deduced amino acid (B) sequences of the region including the homopolymeric tract in NMB0368 and comparison of the nucleotide C) sequences of the region including the tetranucleotide repeat in NMB1994 (nadA gene). Strain numbers are indicated to the left of the sequence. The numbers to the right of the sequences refer to the length of the sequenced regions. The numbers above the sequences refer to relative nucleotide or amino acid positions. The multiple sequence alignments were performed using the program pileup in the GCG software package. Gaps were introduced to improve the alignment and are indicated by a dash in the sequence. The nucleotides and amino acids not shared by all sequences and the repeat tracts are shaded. * indicates a stop codon.

Alignment of the nucleotide sequences determined for NMB1994 (or nadA gene), which is absent from five N. meningitidis strains, the three N. gonorrhoeae strains and three out of the four N. lactamica strains studied, revealed that the number of repeated motifs varied from 5 to 13 between strains (Fig. 1C). The tetranucleotide repeat tracts associated with NMB1893 and NMB1525 are located in the open reading frame (ORF) of the genes and also displayed length polymorphisms (Table 1).

Substantiation of the predictions of phase variability of genes using antibodies directed against the relevant gene product

The phase variability of candidate genes was experimentally investigated by colony immunoblotting using antisera containing antibodies raised to the relevant gene product to confront the predictions made from the DNA sequence analysis of the repeats.

Antibodies directed against D15 [NMB0182, dinucleotide (AC) repeat in the coding region] and NadA [NMB1994, tetranucleotide (TAAA) repeat in the promoter region] of N. meningitidis were raised in mice (Pizza et al., 2000). The cross-reactivity of the antisera with the 26 Neisserial strains used in this study was investigated by dot immunoblotting. This revealed that antisera raised against the D15 and NadA proteins were satisfactory to investigate the phase variability of the corresponding genes using colony immunoblotting.

When 19 000 and 20 000 colonies derived from each of two colonies of strain MC58 were screened with the D15 antiserum, all showed the same level of reactivity (data not shown). D15 locus was PCR amplified and sequenced from seven colonies, chosen arbitrarily. This revealed that those seven colonies presented (AC)4 in the repeat tract, supporting the absence of phase variable expression of this gene.

The level of reactivity observed with the antiserum containing antibodies raised against NadA varied according to the colony (Fig. 2A). We showed that, in strain MC58, different degrees of reactivity were associated with different numbers of motifs in the repeat tract that lies upstream of the coding region of this gene. Colonies not recognized by the antibodies had (TAAA)9 in the repeat tract, those showing moderate levels of reactivity (TAAA)12 and those displaying very high levels of reactivity with the antiserum (TAAA)8 or (TAAA)10 (Fig. 2B). When single colonies exhibiting a very high level of reactivity with the NadA antiserum and displaying (TAAA)10 and (TAAA)8 in the repeat tract were dispersed and then plated, colonies not reacting with the antiserum and exhibiting (TAAA)9 were obtained (Fig. 2B). When plated, these latter phenotypes gave rise to colonies strongly reacting with the antiserum and presenting (TAAA)10 or (TAAA)8 at a frequency of 4.4 × 10−4 (Fig. 2B). These data provided direct evidence that the expression of nadA (NMB1994) was phase variable.

Figure 2.

Phase variable expression of NadA in Neisseria meningitidis strain MC58.
A. Colony immunoblot using the NadA antiserum. The filter predominantly contains colonies not reacting with the antiserum. Colonies, one of which is indicated by an arrow, strongly reacting with the antiserum are also observed.
B. Phenotypes and the associated genotypes of variants of N. meningitidis strain MC58 displaying different levels of reactivity with the antiserum, observed in three consecutive passages (I, II and III). –, ++, +++, ++++ indicate no, moderate, high and very high levels of reactivity respectively. The number of (TAAA) units present in the repeat tract of nadA in each colony investigated is shown. Arrows represent successive plating of colonies of different phenotypes.

The experimental data obtained for D15 and nadA were consistent with the predictions of phase variability made for these two genes from the repeat length polymorphism analysis.

Substantiation of the prediction of phase variability of the spr gene using the lacZ gene as a reporter

Translational fusion utilizing the lacZ gene as a reporter was used to investigate the phase variability of a candidate gene for which antibodies are not available. We selected spr (NMB1969) which presents a repeat tract (C10) displaying interstrain length polymorphism (Table 1) and encodes a putative serine protease potentially relevant to pathogenesis. A promoterless lacZ gene was inserted in frame downstream of the cloned spr gene. This construct was transferred into the chromosome of N. meningitidis strain H44/76.

By adding BHI Top Agar containing Xgal to the colonies of the H44/76 spr/lacZ fusion strain on plates subsequent to overnight growth or by plating colonies directly on BHI medium containing Xgal, we derived blue variants expressing the enzyme and presenting 10 cytosine residues in the homopolymeric tract from a white founder colony not expressing the β-galactosidase and presenting 11 cytosines. This switching was shown to be reversible as white colonies (not expressing the enzyme and presenting 9 or 11 cytosine residues in the repeat tract) were produced in the progeny plated from a blue founder one (expressing the enzyme and displaying 10 cytosines).

These data demonstrated that the expression of the spr gene was phase variable, in agreement with the prediction made from the interstrain polymorphism analysis.

Mapping of the transcriptional start site of nadA

Our analysis of putative phase variable genes in N. meningitidis strain MC58 revealed a novel phase variable gene, nadA, associated with a simple sequence repeat tract (TAAA) not previously reported to be associated with phase variable genes in N. meningitidis. To help elucidate the mechanism of phase variation of the nadA gene and because the repeat tract is located upstream of the coding region of the gene, we investigated the exact location of the repeat tract in the promoter region of the nadA gene.

The transcription start site of the gene was mapped, using primer extension and 5′3′ rapid amplification of cDNA ends (5′3′ RACE), in two variants of strain MC58 expressing low [(TAAA)9] and high [(TAAA)10] levels of NadA protein, based on the level of reactivity with the NadA antiserum (Fig. 2).

Primer extension revealed one major mRNA end-point corresponding to a single transcriptional start site (Fig. 3A) from the variant highly expressing NadA. Transcription was initiated at the residue G-303 located 78 nucleotides upstream of the translation initiation codon and separated from the putative −10 box (TATACT) by seven nucleotides (Fig. 3B). In agreement with the data showing a difference in the amount of NadA protein synthesized, the intensity of the band obtained with the variant presenting (TAAA)10 was strong whereas the band derived from the strain exhibiting (TAAA)9 was not detectable on the gel (data not shown). 5′3′ RACE was used as an alternative method to determine the 5′ end of the nadA transcript. Sequence of the cloned PCR product synthesized from nadA mRNA revealed that, for both strain backgrounds, the residue immediately following the poly-T sequence, representing the poly-A tail that was added to the cDNA and indicating the 5′ end of the nadA mRNA, was the same residue (G-303) as that identified by primer extension (Fig. 3B).

Figure 3.

Transcription start site of nadA mRNA determined by 5′3′ RACE and by primer extension.
A. Identification of the start site for transcription using primer extension. Dideoxysequencing of the upstream and 5′ terminal portion of the nadA gene cloned into the pT7Blue cloning vector using [32P]-end labelled oligonucleotide primer nadA3 is shown with the lane corresponding to each base labelled C, T, A and G accordingly. The reverse transcriptase primer extension using the same oligonucleotide with total RNA from a variant expressing high level of NadA protein is shown (lane 1). The arrow marks the transcription start site which corresponds to the underlined nucleotide. The multiple DNA fragments shorter than the major band were interpreted as mRNA endpoints resulting from the premature termination of the reverse transcription, due to particular RNA structure.
B. Sequence of the 5′ flanking region and 5′ terminal portion of the nadA gene. The repeat tract is indicated in bold. The transcriptional initiation nucleotide is indicated by shading and labelled +1. The −10 and −35 elements are boxed, the ribosome binding site (AAGG) and the translation initiation codon (ATG) are underlined. The same start point was obtained with variants of strain MC58 synthesizing low ((TAAA)9) and high ((TAAA)10) levels of NadA protein.

Thus, the (TAAA) repeat tract was located upstream of the putative −35 element of the core promoter of the nadA gene.

The phase variable expression of nadA is transcriptionally regulated

We then investigated whether variation of the number of repeated motifs resulted in variation of the level of transcription of the gene and if this correlated with the level of expression of the protein detected with the antiserum.

The amount of NadA protein synthesized by variants of strain MC58 presenting (TAAA)8 (TAAA)9 (TAAA)10 (TAAA)11 (TAAA)12 and (TAAA)13 was quantified by performing a ELISA using the NadA antiserum (Fig. 4A). This revealed that variants presenting (TAAA)9 and (TAAA)12 synthesized a small amount of NadA protein and that the variants presenting (TAAA)8 (TAAA)10 (TAAA)11 and (TAAA)13 produced NadA at a high level. Colony immunoblot analysis (Fig. 4B) with these six variants using the same antiserum showed that the variants having (TAAA)9 and (TAAA)12 reacted weakly and that the variants harbouring (TAAA)8 (TAAA)10 (TAAA)11 and (TAAA)13 reacted strongly with the antiserum, in agreement with the ELISA results. We then compared the nadA gene expression in these variants using semiquantitative RT-PCR (Fig. 4C). The variants having (TAAA)9 and (TAAA)12 were shown to synthesize low levels of nadA transcript and variants having (TAAA)8 (TAAA)10 (TAAA)11 and (TAAA)13 expressed the gene at high levels (Fig. 4C). Thus, we showed that the level of nadA gene expression (Fig. 4C) was consistent with the level of NadA protein synthesized (Fig. 4A) through its reactivity with the antiserum (Fig. 4B) and that this expression differed according to the number of reiterated motifs present in the repeat tract.

Figure 4.

Levels of NadA protein and nadA mRNA synthesized in relation to the number of repeated (TAAA) present in the promoter region of the gene.
A. Whole cell ELISA of variants of strain MC58 presenting (TAAA)8 (TAAA)9 (TAAA)10 (TAAA)11 (TAAA)12 and (TAAA)13, using NadA antiserum.
B. Colony immunoblot using the NadA antiserum on the same six variants.
C. RT-PCR analysis of the expression of nadA gene. Oligonucleotides used were 1994.3/1994.5 (nadA) and gdh-P1/gdh-P2 (gdh, used as an internal control to semiquantify the level of nadA gene expression). Total RNA was extracted from variants of N. meningitidis strain MC58 having (TAAA)8 (lanes 1 and 7) (TAAA)9 (lanes 2 and 8) (TAAA)10 (lanes 3 and 9) (TAAA)11 (lanes 4 and 10) (TAAA)12 (lanes 5 and 11) and (TAAA)13 (lanes 6 and 12) in the repeat tract. Lanes 7–12 include reverse transcriptase (RT), lanes 1–6 are control lanes without the enzyme.
D. Statistical analysis of the level of transcription of the nadA gene in variants harbouring different numbers of repeated (TAAA) motifs. The black bars and the grey fractions, respectively, represent the means and standard errors of at least four independent measurements by semiquantitative RT-PCR. The program imagequant was used to calculate nadA levels of transcription normalized with gdh levels of transcription. Statistical analysis was performed using GraphPad software. AU, arbitrary units. The high level of NadA expression associated with (TAAA)11 (A and B) was not reproduced when additional variants presenting (TAAA)11 were analysed to allow the statistical analysis.

To determine whether a periodic pattern could be detected between the number of repeated motifs and the level of nadA gene expression, the semiquantitative RT-PCR experiment was reproduced with additional variants presenting the same numbers of repeats. A statistical analysis of the pooled data revealed that a given number of repeated motifs led to a consistent level of transcription (Fig. 4D). The levels of transcription obtained with variants presenting (TAAA)8 (TAAA)10 and (TAAA)13 were shown not to be different (P < 0.001). In the same manner, no significant difference in nadA transcription was observed between variants displaying (TAAA)9 and (TAAA)12 (P < 0.001). Finally, the nadA expression in variants exhibiting (TAAA)11 was shown to be significantly different from the level of transcript synthesized by the other variants [P < 0.001 when compared with (TAAA)9 (TAAA)10 and (TAAA)13, P < 0.01 when compared with (TAAA)8, P < 0.05 when compared with (TAAA)12]. Thus, an increase or decrease of the number of repeats led to a modification of the level of nadA transcription. However, no periodic pattern correlating repeat numbers with expression was observed.

Discussion

The repertoire of putative phase variable genes in N. meningitidis strain MC58 was originally determined by the identification of simple DNA repeats using criteria based upon the length of the repeat, its position within the gene and knowledge of tracts in established phase variable genes in Neisseria (Saunders et al., 2000). The genes were classified as ‘strong’ or ‘moderate’ candidates by a comparison of repeat length between the three pathogenic Neisseria spp. genome sequences available at that time: N. meningitidis strain MC58, N. meningitidis strain Z2491 and N. gonorrhoeae strain FA1090, as interstrain polymorphism is a trait associated with the repeat tracts of proven phase variable genes. However, for 26 out of the 51 putative phase variable genes in strain MC58 there was either no repeat at the equivalent location or no homologue of the gene in the sequence of N. meningitidis strain Z2491 and/or N. gonorrhoeae strain FA1090 (Saunders et al., 2000). Consequently, interstrain length polymorphism was not discernible for these genes. The aim of the present study was therefore to experimentally re-assess the repertoire of potentially phase variable genes in N. meningitidis.

Twenty candidate genes were selected to be representative of the diversity of repeat tracts encountered in strain MC58. The potential of these 20 candidate genes to be considered as phase variable was evaluated on the basis of length polymorphisms of their repeat tracts across a range of three N. gonorrhoeae, four N. lactamica and 19 N. meningitidis strains. These latter strains were selected to be representative of the genetic diversity of N. meningitidis from a reference set of 107 isolates of N. meningitidis originated from invasive disease and healthy carriers and whose genetic relationships were characterized by multilocus sequence typing (Maiden et al., 1998).

We have shown that, in strain MC58, genes associated with homopolymeric tracts of guanine or cytosine residues longer than five bases were likely to be phase variable. Except for siaD, the previously characterized phase variable genes included in the study exhibited interstrain polymorphism. The homopolymeric tract of seven cytosine residues contained in the siaD gene, which encodes a sialyl transferase involved in capsule biosynthesis, did not display any length variation in this study in spite of the fact that its expression was shown to be phase variable through modifications in the length of this homopolymeric tract (Hammerschmidt et al., 1996). This result points out a limitation of relying solely upon length polymorphism analysis for confirming the phase variability of genes.

We also showed that the presence of a tract of 10 or more adenine residues in N. meningitidis strain MC58 was a reliable indicator of variation of the homopolymeric tract. A repeat tract of seven adenine residues in the coding region of the porA gene has been described as an exception as it was shown to vary in size to alter expression (van der Ende et al., 2000).

The dinucleotide repeats investigated were not polymorphic. This indicates that genes associated with these repeats are probably not phase variable. In N. meningitidis, the dinucleotide repeats investigated are shorter (four or five reiterated motifs) than that described in Helicobacter pylori (Yamaoka et al., 2000) or Haemophilus influenzae (van Ham et al., 1993) and are therefore perhaps below a threshold length required for length polymorphism.

Finally, genes presenting a repeat tract containing more than four copies of a tetranucleotide in N. meningitidis strain MC58 were shown to be potentially regulated by phase variation. Tetranucleotide repeat tracts are known to regulate expression of genes in H. influenzae (Hood et al., 1996) but have not previously been reported to be associated with phase variable expression of genes in N. meningitidis. Although length polymorphism was observed in the tetranucleotide repeat tracts located in the genes NMB1893 and NMB1525, suggesting that these genes could be regulated by phase variation, in all strains investigated the number of repeats makes the ORF out of frame, thus preventing the synthesis of a mature protein. These genes, which present low numbers of tetranucleotide repeats in strain MC58 (four and five respectively), may be examples of genes which were once phase variable, but repeat numbers have contracted to those which are now fixed, at least under in vitro growth conditions. These could now be pseudogenes as they seem not to be expressed in any strain, whatever the number of repeats.

Using mouse sera containing antibodies directed against putatively phase variable proteins, D15, presenting (AC)4 in the ORF, was found not to be phase variable, whereas in contrast, NadA, associated with (TAAA)9, was demonstrated to be regulated by this mechanism in strain MC58.

We have been able to provide additional evidence for the phase variation of a candidate gene, spr, through translational fusion of the lacZ gene in frame with the repeats and the use of β-galactosidase as a reporter of gene expression. A similar approach involving a fusion with alkaline phosphatase was previously described (Belland et al., 1989).

Although based on a limited number of cases, the experimental results obtained with the antisera and the translational fusion were each consistent with the predictions we made about the likelihood of phase variation of genes based on the DNA sequence analysis.

In summary, further study of the 51 putative phase variable genes identified in N. meningitidis strain MC58 predicted that 33 genes (65%) could be considered phase variable and 18 non-variable. These 18 genes correspond to the ‘moderate’ candidates described by Saunders et al. (2000). Of the 19 additional candidate genes described in a more recent comparative whole-genome analysis (Snyder et al., 2001), 10 are associated with four reiterated units of a dinucleotide repeat, and according to our study, these are not likely to be phase variable.

In a recent study of NadA, an adhesin proposed as a novel vaccine candidate, interstrain length polymorphism of the repeat tract associated with the gene was observed (Comanducci et al., 2002). In addition to confirming this finding, the present study demonstrated that the expression of nadA was phase variable. We have also determined the transcription start site of the gene which consequently defined a putative core promoter (−10 and −35 elements). The location of this promoter is not consistent with those recently proposed based upon DNA sequence analysis (Comanducci et al., 2002). In Neisseria, a variation of the number of repeats influencing the transcription of the gene, by modifying the strength of the promoter, has previously been suggested for homopolymeric tracts of cytosine or guanine residues (Sarkari et al., 1994; Sawaya et al., 1999; Biegel Carson et al., 2000). In the case of nadA, several hypotheses could account for the absence of a simple relationship between the number of repeats and the expression of the gene. A modification of the number of repeats might result in a change of DNA conformation, including its curvature or supercoiling. These factors per se are known to affect the interaction of Escherichia coli RNA polymerase with DNA and can consequently have dramatic effects on the level of gene expression (Plaskon and Wartell, 1987 ; Bracco et al., 1989; Nickerson and Achberger, 1995) . Furthermore, a transcriptional regulator could be involved, binding either to the repeat tract or a sequence upstream of it. Alteration of the number of (TAAA) repeats, leading to a localized change of spacing of the flanking DNA sequence, could alter the binding of this putative regulator, or alter its interaction with the RNA polymerase, and consequently modify the nadA gene expression. A third hypothesis is that the repeat tract could act as an UP-like element. In E. coli, a promoter element known as the upstream (UP) element is the third recognition element in the promoter (Ross et al., 1993). This very A+T rich region located just upstream of the putative −35 element was shown to bind the α subunit of the RNA polymerase. The binding of the α subunit of the RNA polymerase to the UP element was shown to strengthen the RNA polymerase binding on the promoter (Lloyd et al., 2001). UP elements must be in the correct position and orientation within the DNA helix relative to the other elements of the RNA polymerase binding sites (−10 and −35 boxes) to permit high levels of transcription (Ross et al., 1993). The (TAAA) repeat tract associated with the nadA gene could act as an UP element and variation of the number of repeats could modulate the strength of binding between the α subunit of RNA polymerase and the promoter and subsequently increase or decrease the level of transcription of the gene. The details of the mechanism by which the (TAAA) repeat tract alters transcription remain to be elucidated.

Experimental procedures

Bacterial strains and culture

The Neisseria strains used in this study are listed in Table 2. Neisseria meningitidis strains were from a collection of 107 strains representative of all major serogroups (A, B, C, W135, X, Y and Z) (Maiden et al., 1998). All strains were grown overnight at 37°C on brain–heart infusion (BHI) medium (Oxoid) with 1.5% (w/v) added agar supplemented with Levinthal's base (Alexander, 1965), in an atmosphere of 5% CO2. For selection of Neisseria transformants, kanamycin (100 µg ml−1) was included in the growth medium.

Escherchia coli DH5α was grown on Luria–Bertani (LB) broth or plates at 37°C and was used to propagate plasmids (Sambrook et al., 1989). When necessary, E. coli was grown in the presence of kanamycin (50 µg ml−1) or ampicillin (100 µg ml−1).

Detection of phase variation by colony immunoblotting

Single colonies of N. meningitidis were picked from a plate grown overnight, resuspended in 70 µl 1% proteose peptone (w/v), 10% glycerol (w/v), and then 10 µl 100-, 200- and 1000-fold dilutions were plated and grown overnight. Colonies were transferred to a nitrocellulose filter (45 µm pore size, Schleicher and Schuell) and allowed to air dry. Non-specific binding sites were blocked for 1 h with 5% (w/v) milk, 0.1% (w/v) azide in PBS-T (phosphate-buffered saline/Tween-20, 0.05% v/v). Filters were washed three times with PBS-T and then were incubated for 2 h 30 min with an appropriate dilution of polyclonal antiserum in PBS-T-bovine serum albumin (BSA, 2% w/v). Anti-D15 and anti-NadA sera were gifts from R. Rappuoli, M. Pizza and M. Comanducci. The dilutions of antibodies used were the following: anti-D15 1/10 000 (v/v) and anti-NadA 1/10 000 (v/v). Filters were then washed three times in PBS-T. To detect the primary antibody, filters were incubated for 1 h with anti-mouse IgG-alkaline phosphatase antibody (Sigma and Cedarlane Laboratories, Ltd). Filters were washed three times in PBS-T and bound antibody was detected with 5-bromo-4-chloro-3-indolyl-phosphate-nitroblue tetrazolium (Perkin Elmer Life Sciences, Inc.). The colour reaction was stopped after 30 min by several washes with water and the blots were air dried.

Dot immunoblotting

Three microliters of bacterial suspensions (cold-inactivated lysates of Neisseria organisms, OD260 = 2.5 or 5) were applied to a nitrocellulose filter (45 µm, pore size, Schleicher and Schuell) and allowed to air dry. The same procedure as described for colonies immunoblots was followed.

Detection of β-galactosidase activity

After overnight growth of N. meningitidis strain H44/76 containing the fusion between the lacZ and spr genes, β-galactosidase activity was detected using two methods. First, 7 ml of BHI Top Agar (agar, 7 g l−1) containing Xgal (5-bromo-4-chloro-3-indolyl-β-d galactopyranoside; 300 µg ml−1) were poured over the plates and the colour allowed to develop for 2 h at 37°C. Alternatively, the strains were grown overnight on BHI plates containing Xgal (80 µg ml−1).

Enzyme linked immunosorbent assay (ELISA)

Whole cell ELISA was performed essentially as described by Plested et al. (2000). Plates were coated with 100 µl of a 5 × 107 cells ml−1 suspension of ethanol-fixed variants of N. meningitidis strain MC58 presenting (TAAA)8 (TAAA)9 (TAAA)10 (TAAA)11 (TAAA)12 and (TAAA)13 in the repeat tract located in the promoter region of NMB1994 (nadA). After evaporation of the liquid overnight at 37°C, the ELISA was carried out using 1/5000 1/10 000, 1/20 000 and 1/40 000 dilutions of the NadA antiserum.

Recombinant DNA techniques and nucleotide sequence analysis

PCRs were performed using Taq polymerase (Gibco BRL). Reactions consisted of 30 cycles of 1 min of denaturation at 94°C, followed by 1 min of annealing at 60°C followed by 1 min of extension at 72°C. The repeat tract and about 200 bp of flanking DNA each side of the repeat were amplified using primers purchased from Genosys and described in Table 3. For the hmbR gene, primers hmbRF1, hmbRF2 and hmbRR1 were used for PCR amplification and sequencing of all N. meningitidis serogroup B strains except strain BZ133. Primers hmbRF2, hmbRF4, hmbRR1 and hmbRR4 were used for PCR amplification and sequencing of all N. meningitidis serogroup A strains. Primers hmbRF1, hmbRF3, hmbRR4 were used for PCR amplification and sequencing of all the N. gonorrhoeae strains. For the fetA gene, primers fetAF2 and fetAR2 were used for PCR amplification and sequencing of all but the N. gonorrhoeae strains and N. meningitidis serogroup B strain BZ198. Primers fetAF3 and fetAR3 were used for PCR amplification and sequencing of all the N. gonorrhoeae strains. Primer 1525F1 and 1525R1 were used for PCR amplification and sequencing of NMB1525 from N. meningitidis serogroup A and serogroup B strains, N. gonorrhoeae and N. lactamica strains. Primer 1525F1 and 1525R2 were used for PCR amplification and sequencing of N. meningitidis serogroup C and serogroup W135 strains. The DNA sequence of both strands of these amplicons was obtained using the same primers in a sequencing reaction with fluorescent dye-labelled dideoxynucleotide terminators using Ampli Taq DNA Polymerase FS (Perkin Elmer), according to the instructions supplied by Applied Biosystems Inc. The sequences were analysed on an automatic sequencer (model 377, Applied Biosystems Inc.). Nucleotide sequence data were analysed using the GCG program (Wisconsin package, version 10.2-Unix, Genetics Corporation Group).

Table 3.  Primers used to amplify by PCR and sequence the DNA repeat tracts.
PrimerSequence (5′-3′)PrimerSequence (5′-3′)
  1. The relevant locus in each case is identified by the gene name or NMB ORF number given as the first part of each primer designation.

cvaAFCTATCCGCCAATGATGCAGTG ClgtGFGATTATGGCGCATTCCCA
cvaARCTGTCCGACTTCGACGTTCGlgtGRGGTCGATGGGAACAGGAA
camFGGAGGTAAACGCTTTGCGGTCGmesJFGGAAAAACAACTGCACGAAGCGC
camRCCGCCTGATGTTGTTTCAGCGGmesJRGGCACGGGGACTCAAGCCG
siaDFGCTTGCAATATTCGCAAAGGTGChsdMFGCACGGTATCGATCCGATAAACTGG
siaDRTTGTTAGCTGATTGATGAACTAACTTAGGChsdMRCGATGTCGTCAAAAGTCGTACCG
NMB2008FGTTGTTGGAGTTGGGGGCGGNMB0065FCTTAACTCGGGCTATGC
NMB2008RATCTTGT TTTCCCTGTGCCNMB0065RCCTGATCACCGAATACTACACC
lbpAFATGAATAAGAAACACGGTTTTCCGCNMB0368FGGAATGTCCAAAAATGAACGC
lbpARCCTGTTCTTTGTTCAACGTTTCCGNMB0368RATCAACACCGTATGATTTAGGG
NMB0415FCAACCGCCCTATATTTGCCGD15FGCCTTCTTCAACGACGTTACC
NMB0415RGCGCTCTCACCGATAACCACAACD15RAAAGTATTGCGACTGCGCC
sprAFCACTGACGCTTGACGGCGCtraFGCTTTGCCCGATGACCGGG
sprARGGCTTGGTGGCTGTGGCTGTGGCTGGtraRCTGCTGACCGAGAGTATCAGCAGG
hmbRF1GGCGGTTAACAACAAAGGCTCGNMB0432FCGTTTGTGGAGATTGAAGAACG
hmbRR1GGTGGTTTTCACCGGATGCTGGNMB0432RGGTGGCAATTGCCGACAC
hmbRF2TATAACCTGACCGCTTCTTCCTfixPFCAACAGGCGGAATAAGAAAAAC
hmbRF3TGCCAACCTCTTTTACGAATGGACfixPRACCAAATAACCGATACCGAAC
hmbRF4CGCAATGCCAACCTCTTTTACaipIFTTTGTCACAGATGCAGAACCT
hmbRR4GCTACTGAACACGTCGTTCCaipIRACTACCTAAGTTGACTGTGCC
sprFGGCGTTAGCTGTTGCAACAACNMB1893FAGCCTGCCAAGAAAATCATCGC
sprRCTACCCCGCGTCCTGTATAGCCNMB1893RGGATGTGGTATCGACATTGCCG
fetAF2TTTACATTTCCGACAAAACCTGTCAAC1525F1CACATTTTCAAACTGGCTTGAC
fetAR2GCGAGAGCAGGCTGAGACGG1525R1ATCGCGGAGTTTTTTTAAGCGG
fetAF3GTTCCGTACTATTTGTACTGTCTG1525R2ATCTCTAAGTTTTTTTAAGCGA
fetAR3GGTGCTTTCGTGTTTTTGTTGnadAFTCGACGTCCTCGATTACGAAGGC
pglAFCCCGATGAAAAAGAACCCGTAnadARTGGCTGTGGTCAGTACTTTGGATGG
pglARGCCTCCTGAGTGCTTCGGGfunZFTCAGCTAAAGAAAGCAAGACAC
funZRGCGTTAGCAAGACCTTTTACAC

Southern blots and hybridisations were performed under stringent conditions using radiolabelled probes, essentially as described by Sambrook et al. (1989).

Construction of translational fusion between the lacZ gene and the spr gene in N. meningitidis

Plasmid pGsprZK contains the fusion between the lacZ gene and the spr gene (NMB1969) and was constructed as follows, using plasmid pGΔZ-WT, which was previously described for a similar experiment with the mod gene in H. influenzae (de Bolle et al., 2000). pGΔZ-WT is mainly composed of a HindIII–BglII DNA fragment containing the terminal portion of the mod gene fused with the promoterless lacZ gene, which is linked to a BamHI–EagI fragment carrying a fraction of the res gene (de Bolle et al., 2000). The genomic region containing the spr repeat tract was amplified from strain MC58 using primers sprhind (5′- CCCAAGCTTCCCAAACCGTCATTCCCGCC -3′) and sprbgl (5′- GAAGATCTTTCTTGTATGCGTCATTTGGG -3′), thus originating HindIII and BglII sites (underlined) at opposite ends of the spr5′ PCR product. The primer sprbgl was designed in order for the lacZ gene and the repeat tract to be in frame once cloned in plasmid pGΔZ-WT. The PCR product was subcloned into the pCR2.1 cloning vector (Invitrogen), released by digestion with HindIII and BglII, gel purified with QIAEXII Gel Extraction Kit (Qiagen) and cloned upstream of the lacZ gene in plasmid pGΔZ-WT digested with the same restriction enzymes. Thus, the mod fragment was replaced by the spr5′ fragment. The res fragment was then excised from the resulting plasmid by EagI/BamHI digestion and replaced by the spr3′ PCR product obtained with spreag (5′-CCCGGCCGCCGCATGTTCGTC CGCACCG-3′) and sprbam (5′-CGGGATCCGGGCG GCATCCGGCGGCAGC-3′) primers, subcloned and then released with EagI and BamHI enzymes. The final step was to excise the Tn903 kanR gene from the pUC4K vector (Pharmacia) by digestion with BamHI, and to clone it into the plasmid linearized by BamHI.

To transfer the fusion construct to the chromosome of N. meningitidis by homologous recombination, strain H44/76 was transformed with plasmid pGsprZK linearized with ScaI or uncut. Transformants were streaked on selective medium and their chromosomal DNA was isolated and analysed by PCR amplification to confirm the presence of the fusion. The DNA repeat tract of phenotypic variants (lacZ+ and lacZ) was amplified with primers lacZB1 (5′-TCCCAGTCACGACGT TGT-3′) and sprhind and these products were sequenced as described above.

Semi quantitative reverse transcriptase-polymerase chain reaction (RT-PCR)

After 12 h growth on solid medium, total RNA was prepared using the SV total RNA isolation system (Promega) from variants of N. meningitidis strain MC58 presenting (TAAA)8 (TAAA)9 (TAAA)10 (TAAA)11 (TAAA)12 and (TAAA)13 in the repeat tract located in the promoter region of NMB1994 (nadA). Next, 550 ng of total RNA was reverse transcribed to produce cDNA using moloney murine leukemia virus (MMLV) reverse transcriptase (RT) and random primers (Promega). Co-amplification of the nadA transcript with an internal control allowed comparison between the different RNA samples. PCR amplification of the nadA gene was performed with oligonucleotides 1994.3 (5′-GTAGCCGCCGACTGCAGCCG-3′) and 1994.5 (5′-GCCACTTTCTGTAGCGGCGC-3′). Primers specific for the constitutively expressed housekeeping gene gdh (glucose-6-phosphate dehydrogenase), gdh-P1 (5′-ATCAATACCGATGTGGCGCGT-3′) and gdh-P2 (5′-GGTTTTCATCTGCGTATAGAG-3′) were also included in each PCR amplification to provide an internal control. As a control for chromosomal DNA contamination, RNA was used directly for PCR amplification. The relative intensities of the nadA-specific and gdh-specific products were measured by densitometry (ImageQuant, Amersham Biosciences). The experiment was reproduced with several variants presenting the same numbers of repeats and PCR amplifications were performed twice with each cDNA. All the data were pooled and a statistical analysis was performed using the prism program (GraphPad Software Inc.), one-way anova, Tukey's multiple comparison test.

Mapping of the transcriptional start site of nadA

In order to determine the 5′ end of the nadA transcript, primer extension and 5′3′ RACE (Roche Molecular Biochemicals) methods were used. In the primer extension method, oligonucleotide nadA3 (5′-CGTCGTCGCTTGTGGCTGCC-3′) was end labelled with [γ32-P]-dATP using polynucleotide kinase (Boehringer Mannheim) (Sambrook et al., 1989). Next, 25 µg total RNA was mixed with 200 ng of end-labelled oligonucleotide in the presence of AMV (avian myeloblastosis virus) RT. In parallel, a sequencing reaction was performed according to the protocol supplied by the manufacturer with the Sequenase 2.0 kit (USB) using the same primer nadA3 and the upstream and 5′-terminal portion of the nadA gene cloned into the pT7Blue cloning vector to allow the identification of the end of the mRNA. 5′3′ RACE was performed as briefly follows. cDNA was synthesized from 550 ng of total RNA using primer nadA2 (5′-AATGTCGTAGATG GTCTCTCC-3′). A 3′ dA tail was then added to the cDNA. The dA-tailed cDNA was amplified by PCR at 55°C using the oligo-dT anchor primer and nadA2. Nested PCR amplification, using oligo-dT anchor primer and nadA3 and carried out at 65°C, was required to generate a product visible on agarose gel. The PCR product was subsequently ligated into the pCR2.1 vector (Invitrogen) and sequenced using the M13 universal and nadA3 primers.

Acknowledgements

This work was supported by a grant programme from the Wellcome Trust (E.R.M).

This project was initiated with the input of Dr Nigel. J. Saunders, whom the authors wish to acknowledge. We would like to acknowledge the gift of antisera provided by Rino Rappuoli, Mariagrazia Pizza and Maurizio Comanducci, Dr Christopher D. Bayliss for giving us plasmid pGΔZ-WT, Dr Wendy Sweetman for her advice concerning the RT-PCR, P. A. Coull for help with ELISA, Dr Ruth Ripley for her advice on the statistical analysis of the semiquantitative RT-PCR.

Ancillary