DNA repair profiles of disease-associated isolates of Neisseria meningitidis

Introduction

Neisseria meningitidis, or the meningococcus (Mc), colonizes the mucosal surfaces of the human upper respiratory tract. Although it usually resides as part of the commensal flora, Mc may traverse from this nonpathogenic carrier state and disseminate into the blood stream. Despite advances in preventive and therapeutic medicine, meningococcal disease (meningitis and/or septicaemia) is the cause of significant morbidity and mortality in industrialized and developing countries worldwide (Caugant et al., 1986). Mc has humans as its exclusive host and must be uniquely adapted to this environmental niche. This habitat offers challenges such as nutrient deprivation, competition from other microorganisms, high oxygen concentrations, and the immune responses of the host. Mc fitness for survival will be directly linked to its capability of generating change and adaptation to its environment, and strains conferring a selective advantage will excel (Bjedov et al., 2003).

Mutator strains exhibit increased spontaneous mutation rates compared with those commonly found in the corresponding wild-type species (Giraud et al., 2001). Hypermutable strains in bacterial populations might have potential advantages or disadvantages, depending on the nature, rate and magnitude of environmental change (Giraud et al., 2001; Notley-McRobb et al., 2002; Bjedov et al., 2003), and have been described in various human pathogens (LeClerc et al., 1996; Oliver et al., 2000). Mutators have generally been associated with defects in the postreplicative mismatch repair (MMR) pathway responsible for removing insertion/deletion loops (Lahue et al., 1989), suggesting that frameshift mutations and mispairing of nucleotide runs are major pathways for generating variability in pathogenic species, including Mc (LeClerc et al., 1996; Oliver et al., 2000; Richardson & Stojiljkovic, 2001; Richardson et al., 2002). Conflicting evidence exists on the association of Dam methylase variants causing hypermutable neisserial strains with enhanced phase-variable capsule switching (Bucci et al., 1999; Jolley et al., 2004). Moreover, only 40% of the Mc strains investigated by Stojiljkovic and coworkers could be complemented with wild-type alleles of MMR components MutS or MutL (Richardson et al., 2002). This means that more than half of Mc mutators are genetically uncharacterized, suggesting that DNA repair pathways yet to be identified contribute to the pool of Mc mutator phenotypes – a notion supported by findings in other pathogens such as Helicobacter pylori and Streptococcus pneumoniae (Bjorkholm et al., 2001; Morosini et al., 2003).

A number of DNA repair pathways and adaptive responses other than MMR have been related to mutagenicity in laboratory strains of Escherichia coli (Horst et al., 1999). Endogenous DNA damages, including one of the most frequent oxidative lesions, 7,8-dihydro-8oxo-2′-deoxyguanosine (8oxoG), are primarily processed through the base excision repair (BER) pathway (Seeberg et al., 1995). In E. coli, BER is initiated by DNA glycosylases that nick the N-glycosylic bond and remove the damaged base (Lindahl, 1979). The DNA glycosylases MutY and MutM (formamidopyrimidine DNA glycosylase, Fpg) have been reported to reduce mutation rates in bacteria by removing adenine mispaired with 8-oxoG, or removing 8-oxoG when mispaired with cytosine, respectively (Cabrera et al., 1988; Nghiem et al., 1988; Michaels et al., 1992). Other major E. coli DNA repair pathways involved in correcting DNA lesions are the nucleotide excision repair (NER), recombinational repair, and translesion synthesis (TLS). NER generally repairs bulky lesions from exogenous sources interfering with normal basepairing and impairing transcription and replication (Pettijohn & Hanawalt, 1964; Seeberg, 1978). RecA, the major component of recombinational repair, is involved in the repair of double-strand breaks in DNA and control of the inducible SOS response in E. coli (Little & Mount, 1982). TLS polymerases conduct translesion synthesis past blocking lesions, allowing the cell to overcome such obstacles. However, these polymerases might generate mutations at high frequencies (Napolitano et al., 2000).

Despite the profound influence of genome instability on the occurrence of hypervirulent Mc lineages (Richardson & Stojiljkovic, 2001; Richardson et al., 2002), Mc DNA repair pathways have so far not been delineated in detail (Davidsen et al., 2005; Davidsen & Tonjum, 2006). The aim of this study was to characterize key DNA repair components in Mc clinical isolates to assess whether obvious defects, other than those in MMR, were associated with mutator phenotypes. The spontaneous mutation frequencies were determined in a collection of Mc disease-associated isolates. Cell lysates from these strains were tested for their in vivo DNA glycosylase activity. Moreover, Mc strains were assessed for survival under oxidative stress, and the mutation spectrum of rifampicin-resistant strains was determined. Selected Mc DNA repair genes with the potential to be part of the mutator-inducing pool were subjected to DNA sequence analysis. Multi-locus sequence typing (MLST) data were used as a reference for sequence polymorphisms and epidemiologic links of the Mc strains under study. On this basis, the sequence conservation and variation of DNA repair components in Mc wild-type and mutator isolates were compared.

Materials and methods

Results and discussion

The majority of hypermutable Mc isolates examined to date are genetically uncharacterized (Richardson et al., 2002). We have investigated a collection of disease-related Mc strains and employed functional DNA repair assays and DNA sequence analysis to define why some of these isolates display high mutation frequencies.

The spontaneous mutation frequencies of the Mc clinical isolates listed in Table 1 ranged from <0.1 to 9.9 per 10⁸ CFU (Table 2). Five isolates, Z1227, Z1269, Z3515, Z6244 and M1080, had spontaneous mutation frequencies that were significantly higher (1.3–9.9 × 10⁻⁸ CFU) (P<0.05) than those of the other isolates (<0.1 × 10⁻⁸ CFU) (Table 2). Mc strains Z1466 and Z5826 both had spontaneous mutation frequencies of 0.2 × 10⁻⁸ CFU, but could not be clearly distinguished from the isolates that exhibited ‘low’ spontaneous mutation frequencies (P>0.05) (Table 2). In addition, three strains were classified as having intermediate mutation frequencies by screening for rifampicin-resistant colonies only (Table 2).

To search for any uniformity in the pattern of mutation spectrum conferring rifampicin resistance, such as the GC→TA transversions one previously detected for Mc MutY (Davidsen et al., 2005), part of the gene encoding RpoB in selected Mc disease-associated isolates in the ‘high’ and ‘low’ mutation frequency groups were subjected to DNA sequencing (Table 3). All rifampicin-resistant strains, except Z1269 and Z4259, consistently exhibited CG→TA transitions; indeed, isolates Z1227, Z1466, M1080, Z3842 and Z7176 exclusively harboured this mutation (Table 3). Furthermore, Z1099, Z1269, Z5826, Z6244 and Z4259 harboured one CG→AT, AT→TA, AT→GC, TA→CG or GC→AT transversion/transition, while strain Z4756 harboured a triplet expansion (Table 3). The results indicate that Mc strains Z1099, Z1269, Z4756, Z5826 Z6244 and Z4259 exhibit a broader rpoB mutation spectrum than Z1227, Z1466, M1080, Z3842 and Z7176, suggesting the presence of more than one mutagenicity pathway in the six former strains. However, no relation between rpoB mutation pattern and high/low mutation rate could be observed.

In addition, the Mc clinical isolates were subjected to oxidative stress by exposure to H₂O_2. The Mc isolates exhibited differences in H₂O₂ sensitivity, as observed in the close relative Neisseria gonorrhoeae (Alcorn et al., 1994), and no clear correlation could be detected between high or low mutation rates and hyper-resistance or hyper-sensitivity to H₂O₂ (Table 4). In this context, it should be pointed out that Mc, in addition to catalase and a number of other scavenger enzymes (Seib et al., 2004), hosts no fewer than three DsbA homologues (Sinha et al., 2004), rendering this species particularly fit for handling oxidative stress.

The DNA glycosylase MutY of the BER pathway has previously been implicated in mutator activity in pathogenic species – four Pseudomonas aeruginosa mutator isolates from cystic fibrosis patients were found to have defects in MutY (Oliver et al., 2000). Moreover, we have recently demonstrated a striking increase in the spontaneous mutation frequency in MutY-defective Mc strains (Davidsen et al., 2005). Furthermore, Mc mutS and fpg mutants have been shown to give rise to increased mutation frequencies, but these were far below the rate induced by the mutY mutant (Davidsen & Tonjum, 2006; Davidsen et al., unpublished results). Fpg is, together with MutY, important in the protection against mutations caused by 8oxoG (Michaels et al., 1992; Davidsen et al., 2005). Defects in MutY and Fpg might thus contribute to the naturally occurring Mc mutator pool. However, MutY and Fpg were constitutively expressed in vivo in all of the Mc clinical isolates tested (Table 1), even in the strong mutator strains, as detected by their excision of A:8oxoG, 8oxoG:C and 5-hydroxy-C:G DNA lesions (Fig. 1). This is the first demonstration of in vivo expression of MutY and Fpg in Mc isolates. The findings might indicate that the excessive mutator phenotype seen in the mutY mutant (Davidsen et al., 2005) is infrequent in Mc natural populations. Such defects might be selected against, be very transient, or be rapidly repaired by transforming DNA (Davidsen et al., 2004). If the mutY gene is damaged, its uptake seems to be facilitated, because it exhibits the highest density of DNA uptake sequences (DUS) required for transformation of all Mc DNA repair genes (Davidsen et al., 2004).

DNA sequence analysis of genes encoding selected components representing major DNA repair pathways was performed. The deduced amino acid sequences of these DNA repair components from the Mc clinical isolates were aligned (Supplementary Material, Figs S1−S7). Interestingly, none of the amino acid changes detected in the DNA repair components occurred in recognized DNA binding sites, catalytic residues or important motifs. Amino acid changes were, however, found in the regions between these motifs, and some of the amino acid substitutions detected might influence protein structure, particularly when they involve a change in amino acid class. The patterns of amino acid substitutions in strain M1080 were distinctly different from those of the other strains with ‘high’ mutation rates, and M1080 might thus host a different mutator-inducing genetic background (see below).

The DNA repair genes were moderately to highly conserved with regard to both nucleotide and amino acid sequences, as were the MLST sequences (Table 5). The mutY gene exhibited the lowest average nucleotide variation (0.9%) among the isolates, while fpg, nth and uvrA exhibited the highest nucleotide variability (>3%) (Table 5). At the deduced amino acid level, however, RecA showed complete sequence conservation (no variability), while the MutY amino acid sequence was the second most conserved (0.5% variation). Despite the relatively high variability of the nth gene at the nucleotide level, the Nth deduced amino acid sequence was remarkably conserved (0.5% variation). The deduced amino acid sequences for MutS and Fpg exhibited the highest numbers of average amino acid substitutions (>1%) (Table 5). The sequence conservation observed in some of the DNA repair enzymes might underline the significance of these components in vivo. No clear relationship between frequencies of amino acid substitutions, functional assays and mutator phenotype was evident in the Mc isolates.

Interestingly, a putative correlation between the deduced amino acid sequence patterns and spontaneous mutation frequencies was found. Mc strains Z1099, Z1466, Z4746, Z5005, Z5826, Z7176 and Z3842 all exhibited ‘low’ mutation frequencies, while Z1318, Z3524 and Z5037 were classified as having intermediate mutation frequencies, and Z1227, Z1269, Z3515, Z6244 and M1080 were defined as mutators. Although the Mc strains were isolated from various geographical regions worldwide and at different times (Table 1), dendrograms of MLST data and DNA repair component sequences both indicated that strains Z1099/Z1466/Z4756/Z5005, Z7176/Z3842, Z1227/Z3515/Z3524 and Z1269/Z6244/Z1318 were linked (Fig. 2; Supplementary Material, Figs S8 and S9). Despite the small sample size, an association between sequence patterns and mutation phenotypes may exist. How much of this is the result of clonal expansion of hypermutagenic serogroup A lineages or selective pressures ensuring that only certain lineages of mutator strains will survive is unknown.

The evolution and adaptation of microorganisms are influenced by two opposing forces: the maintenance of genetic information and the generation of a suitable level of genetic variation upon which selection can act. Although the analysis of the potential role of Mc mutY, fpg, nth, mutS, uvrA, recA and dinB in the generation of mutator phenotypes has been initiated, a number of other putative mutator target genes remain (Horst et al., 1999). As a starting point, these findings represent a basis for understanding the mechanisms configuring the genetic profiles of Mc natural populations. In turn, this information could lead to models that elucidate how mutator phenotypes are linked to microbial fitness and hypervirulence.