1 Introduction

Neisseria meningitidis causes approximately 500,000 cases of meningitis and septicemia worldwide annually, with a case-fatality rate of approximately 10% (World Health Organization 1998). Most disease in temperate countries is caused by capsular group B organisms (Halperin et al. 2012). A number of group B vaccines based on different combinations of subcapsular antigens have been developed, including several outer membrane vesicle (OMV) vaccines (Sadarangani and Pollard 2010). Many OMVs in vaccines used in clinical trials have been derived from strains which have been genetically modified by bacterial transformation to alter the expression of specific surface components in order to improve immunogenicity. Therefore, cellular events influenced by transformation are highly relevant to vaccine development.

The Opacity-associated (Opa) adhesin proteins are major phase-variable proteins found in the outer membrane of N. meningitidis. Each bacterium contains four opa genes (opaA, opaB, opaD and opaJ), which may encode identical or different Opa proteins, and can therefore express up to four different Opa variants at any one time (Tettelin et al. 2000). Opa proteins are critical in meningococcal pathogenesis, mediating bacterial adherence to the nasopharynx and modulating human cellular immunity via interactions with T cells and neutrophils (Virji et al. 1993; Gray-Owen 2003).

Phase variation (PV) describes the phenomenon whereby pathogens can genetically switch ‘on’ and ‘off’, or vary, the expression level of specific components, and is a major mechanism of variability for N. meningitidis (Snyder et al. 2001). Several meningococcal surface structures exhibit PV, including the capsule, pili and outer membrane proteins PorA, NadA, Opc and Opa. PV allows the meningococcus to adapt to the different host environments (i.e. nasopharynx, blood, meninges) encountered during infection. PV is believed to be vital for the organism to evade the host immune response by regular alteration of its surface, which is especially important for N. meningitidis since it is usually carried asymptomatically in the nasopharynx for up to several months. Understanding the dynamics of meningococcal PV will therefore enable the appropriate development of vaccines targeting subcapsular antigens, such as OMVs. PV of Opa proteins is thought to predominantly occur through slipped-strand mispairing within a pentanucleotide coding repeat (CR) sequence 5′-CTCTT-3′ present within the opa open reading frame (within the signal peptide), resulting in frame shifting during DNA replication and the translation of non-functional, truncated proteins (Stern et al. 1986).

We have previously described the construction of Opa-deficient derivatives of N. meningitidis strain H44/76, produced by sequential disruption of opa genes using antibiotic resistance cassettes (Sadarangani et al. 2012). In this study, strains from this unique library were used to analyse the PV of different Opa proteins to identify factors which may influence Opa PV rates.

2 Materials and methods

2.1 Bacterial strains and growth conditions

Wild type N. meningitidis strain H44/76 and derivative Opa-deficient strains, which have previously been described, were used (table 1) (Sadarangani et al. 2012). N. meningitidis was grown on brain heart infusion (BHI) agar (Merck, Darmstadt, Germany) supplemented with Levinthal’s base (10% v/v) at 37°C in a humidified 5% CO2 atmosphere for 16–18 h. Selective media was supplemented with kanamycin (100 μg/mL) (Sigma-Aldrich, Gillingham, UK).

Table 1 Strains of Neisseria meningitidis used in this study
Full size table

2.2 Transformation of Neisseria meningitidis

N. meningitidis was transformed using the spot transformation technique, as previously described (Sadarangani et al. 2012). Briefly, approximately 108 colony forming units (cfu) of bacterial suspension from overnight growth was incubated with approximately 1 μg of pT7-E2-F2-kan plasmid DNA, plated over a 1–2 cm diameter region on BHI agar. pT7-E2F2-kan contains the LPS biosynthesis gene lpt3 disrupted with a kanamycin resistance cassette (Mackinnon et al. 2002). After incubation at 37°C, 5% CO2 for 4–8 h bacteria were plated onto selective and non-selective BHI agar and incubated for a further 16–18 h. Experiments were performed in duplicate.

2.3 Colony blotting

Colony blotting with anti-Opa monoclonal antibodies (mAbs) 15-1-P5.5 and MN20E12.70 (de Jonge et al. 2003) was performed to identify the expression of surface-expressed Opa proteins. mAb 15-1-P5.5 targeted OpaA and OpaJ of H44/76 and mAb MN20E12.70 bound to OpaB and OpaD. Prior to blotting, bacteria were diluted in phosphate buffered saline (PBS) and plated onto BHI agar to achieve 1500–3000 colonies per plate. Plates were incubated at 37°C, 5% CO2 for 16–18 h, following which a nitrocellulose membrane (pore size 0.45 μm, GE Healthcare, Buckinghamshire, UK) was used to adsorb the colonies. The nitrocellulose membrane was air-dried, then blocked for 1 h with 5% skimmed milk in PBST (PBS containing 0.1% Tween 20 and 7.7 mM sodium azide). The membrane was then washed several times with wash buffer (154 mM NaCl, 0.05% Tween 20) and the mAb added after dilution to 1:50,000 in PBST containing 1% bovine serum albumin (BSA). After 1–2 h, the membrane was washed several times prior to incubation with goat anti-mouse IgG (Fab specific)-alkaline phosphatase antibody (Sigma-Aldrich), diluted 1:10,000 in PBST (without sodium azide) containing 1% BSA, for 1–2 h. Following further wash steps, alkaline phosphatase activity was detected by the addition of 5-bromo-4-chloro-3-indolyl phosphate (BCIP)/nitro blue tetrazolium (NBT) (Sigma-Aldrich). The membrane was incubated for 10 min and the reaction stopped by rinsing with water.

2.4 Estimation of opa mutation rates

While PV describes the alteration in Opa surface expression, the term ‘mutation rate’ is used here to describe the genetic changes occurring in the cell which underlie PV. The mutation rates of opa genes relevant to PV (i.e. the mutation rate of the CR sequences) were estimated using the following equation:

$$ \alpha =\frac{2}{1+\mathbf{x}}\left[1{-}^n\sqrt{1-\left(1+x\right){p}_n}\right], $$

where α represents the mutation rate per cell per generation, the back-variation rate is x times the forward rate, n is the number of generations of bacterial cell division and p n is the proportion of mutants at n generations (Saunders et al. 2003). This equation assumes no fitness advantage between the two phase-variable states, which was assumed in this study given that it was performed in vitro without any selection pressures. Therefore, x=1 was used in this equation. In an initial experiment, colonies were plated out using 10-fold serial dilutions between 10−2 and 10−4, which demonstrated that each colony contained approximately 20 generations (n=20). To estimate the mutation rates from Opa expression experiments, it was assumed that the mutation rate was double the PV rate because for an opa gene in the ‘off’ state, 50% of alterations in the length of the CR tract would result in a detectable ‘on’ state, while the other 50% would not be detected by colony blotting as the expression would remain ‘off’.

2.5 Opa expression in UK meningococcal disease isolates

The Meningitis Research Foundation (MRF) Meningococcus Genome Library (MGL) was used to analyse the expression of Opa proteins in UK disease-causing meningococci ( http://pubmlst.org/perl/bigsdb/bigsdb.pl?db=pubmlst_neisseria_mrfgenomes ). The library contained whole genome sequence data from UK disease-causing isolates (predominantly from blood and cerebrospinal fluid (CSF)) of N. meningitidis which had caused infections between 2010 and 2013.

3 Results

3.1 opa mutation rate

To assess the opa mutation rate, analysis was performed using anti-Opa mAbs 15-1-P5.5 and MN20E12.70. To circumvent the partial cross-reactivity of the mAbs and specifically assess individual Opa proteins, mutant strains were utilised – each strain was derived from H44/76 and had a single opa gene disrupted. In a given experiment, therefore, the mAb used would only be able to bind to one of the Opa proteins – for example, to assess PV of OpaA, strain M001/02 (ΔopaJ) was used with mAb 15-1-P5.5. In addition the Opa protein being assessed was not expressed at the start of the experiment due to the number of CR sequences present (table 1). opaD had the highest mutation rate of 6.9×10−3 per cell per generation, followed by opaA (1.9×10−3 per cell per generation) (table 2). opaB and opaJ had lower mutation rates of 6.4×10−4 and 7.9×10−4 per cell per generation, respectively. A longer CR tract was associated with an increased mutation rate (r 2=0.77, p=0.1212), although this did not achieve statistical significance.

Table 2 Mutation rates of opa genes by nucleotide sequence analysis and colony blotting
Full size table

3.2 The effect of transformation on opa mutation rate

To assess the effect of transformation on opa mutation rate, strain M001/02, which expressed no Opa and had opaJ deleted (table 1), was transformed with pT7-E2F2-kan, and OpaA expression determined by colony blotting with mAb 15-1-P5.5. When the total number of bacteria were considered, the opaA mutation rate was 5.6×10−4 per cell per generation (95% CI 4.1×10−4 to 7.1×10−4). In successfully transformed bacteria (transformation rate 1.5×10−7), the mutation rate was 180 times higher at 1.0×10−1 per cell per generation (95% CI 8.4×10−2 to 1.2×10−1). Overall, 32.1% of bacteria had switched from the ‘off’ state to the ‘on’ state. Given that only 50% of alterations in the length of the CR tract result in a switch from an ‘off’ state to a detectable ‘on’ state, it can be estimated that 64% (95% CI 62.1 to 66.4) of the transformants were therefore hypermutable, at least transiently.

3.3 Opa expression in UK meningococcal disease isolates

To investigate the influence of Opa PV at a population level, we interrogated a database of UK meningococcal disease isolates to assess Opa expression. There were a total of 1,381 isolates in the MRF MGL database, of which 451 contained sequence data for a total of 463 opa genes (annotated as NEIS1719). Overall, 315/463 (68.0%) opa genes would result in expression of Opa on the bacterial surface. There were a total of 82 unique opa sequences when the variability of the 5′ sequence upstream of the sequence encoding the mature surface-expressed protein was excluded, of which 52 were found once only and 30 were present on multiple occasions. Further analysis of the duplicates among these confirmed that variability between genes encoding the same Opa protein in different isolates was almost entirely due to differences in the length of the CTCTT repeat.

4 Discussion

Our finding that transformation of N. meningitidis with unrelated plasmid DNA resulted in a 180-fold increase in Opa PV rate has potential significance for meningococcal pathogenesis. Increased PV following transformation is likely to confer a survival advantage to bacteria during natural infection. Transformation is a powerful mechanism for generating genetic diversity, spreading advantageous allelic variants, and mediating some forms of antigenic variation. During colonisation of the human nasopharynx multiple bacterial species co-exist, resulting in relatively large amounts of extracellular DNA and consequently frequent transformation events, although transformation is most likely to occur using DNA from the same species. Increased population diversity aids immune evasion by modification of potential targets on the bacterial surface, and may occur due to increased PV in a single organism following DNA uptake and/or by selection of ‘hypermutable’ strains at the population level (Alexander et al. 2004; Bayliss et al. 2008). The former is supported by a previous study which demonstrated a 24- to 73-fold increase in PV following transformation with chromosomal neisserial DNA due to inhibition of the mismatch repair (MMR) genes by single stranded DNA (Alexander et al. 2004). Mutations in the MMR genes mutS or mutL can produce ‘hypermutable’ strains, which have a 10-fold increase in basal mutation rates and a 1000-fold increase in PV rates of genes containing mononucleotide repeat tracts (Bayliss et al. 2008; Martin et al. 2004; Richardson and Stojiljkovic 2001). In our study we were unable to confirm whether these loci were mutated by the transformation process and therefore the mechanism of the increased PV rate.

The PV rates we found are similar to those previously determined for gonococcal Opa proteins (Mayer 1982). The data suggested a relationship between repeat tract length and PV rate, which has previously been described for the polyG tract of N. meningitidis hmbR and in Haemophilus influenzae, and is thought to be due to increased instability of the region of DNA (De Bolle et al. 2000; Richardson et al. 2002). Other important factors in determining PV rates include the type of repeat sequence (poly-G/C tracts are less stable than poly-A/T (Warmlander et al. 2002)), the location of a gene within the chromosome (in particular repetitive DNA in close proximity and transcription levels of that region of the chromosome) and gene promoter strength. To more fully assess Opa PV it would be ideal to assess different Opa proteins in a range of N. meningitidis strains, and directly measure both forward and backward switching rates. The presence of multiple Opa proteins per organism and availability of only the specific monoclonal antibodies described means we were limited to using H44/76 and the derivative strains we have constructed. While it is possible that these findings are specific to this strain, our results are consistent with previous studies.

Our finding of increased Opa PV following transformation has potential implications for development of meningococcal OMV vaccines because a number of such vaccines are based on strains which have been genetically modified via transformation (Cartwright et al. 1999; de Kleijn et al. 2000; Pettersson et al. 2006; Finney et al. 2007; van den Dobbelsteen et al. 2007; Weynants et al. 2007, 2009; Koeberling et al. 2009; Bonvehi et al. 2010; Zollinger et al. 2010; Keiser et al. 2010, 2011). Characterisation of the modified bacteria and resulting OMVs usually focuses on the vaccine antigens, without a detailed assessment of other antigens. To confirm this phenomenon of transformation-induced increased PV, it would be ideal to conduct these transformation experiments with multiple plasmids, as well as whole genomic DNA, and assess several phase variable proteins. It is possible that disruption of the LPS biosynthesis gene lpt3 had a direct effect on transformation rate, although this has not been described previously. We have previously shown that transformation with opa-containing plasmid DNA did not modify the expression of other major outer membrane proteins PorA, PorB, RmpM or factor H binding protein (fHbp) (Sadarangani et al. 2012), but the phenomenon of increased PV following transformation has been previously described (Alexander et al. 2004) and the narrow confidence intervals we found suggest this effect also occurs for Opa proteins.

Results of the analysis of the MRF MGL suggested there may be a population level fitness advantage conferred by Opa expression in blood and CSF, which could be related to increased adhesion and colonization by Opa+ variants, or the immunomodulatory effects of Opa contributing to survival of the organism during invasive disease (Sadarangani et al. 2011). An important role for Opa during infection with all pathogenic Neisseria is supported by the observation that Opa expressing bacteria are recovered during natural gonococcal infection (James and Swanson 1978) and following inoculation of humans with Opa non-expressing bacteria (Swanson et al. 1988; Jerse et al. 1994). A major limitation of analysis of opa genes in the MRF MGL is that coverage of opa using whole genome sequencing is relatively poor due to it being a multi-copy gene, and being surrounded by significant repetitive DNA. However there is no clear explanation why the successfully sequenced genes would result in a biased dataset and biological plausibility exists as to the importance of Opa in meningococcal pathogenesis. In addition the treatment and in vitro culture of these clinical strains between isolation and genome sequencing is unknown and may vary between strains - it is therefore possible that there may be in vitro selection with respect to Opa expression, although this would be minimised due to the absence of specific selection pressure during in vitro culture.

PV of Opa proteins is of particular importance for a number of reasons. Opa PV may be a key mechanism of immune evasion by the meningococcus, since Opa proteins induce bactericidal antibodies following meningococcal infection and after immunization with serogroup B OMV vaccines (Mandrell and Zollinger 1989). PV of other outer membrane proteins included in meningococcal vaccines have been shown to mediate escape from bactericidal antibodies, and further investigation of this phenomenon will aid the future design of such vaccines (Bayliss et al. 2008; Tauseef et al. 2013; Alamro et al. 2014). Many OMV vaccines in clinical trials have been derived from genetically modified strains (created by transformation) and characterization of the modified bacteria and resulting OMVs usually focuses on the vaccine antigens, without a detailed assessment of other antigens. The process of transformation to construct these vaccine strains may result in changes in expression of other phase variable meningococcal proteins, leading to the possibility that modification of just one or two proteins may significantly alter the immunogenicity or reactogenicity of the resulting OMV, or have an effect on bacterial growth and therefore influence vaccine production. Further study of the major phase variable proteins during vaccine development may therefore aid the development of more refined products with better immunogenicity and reactogenicity profiles, which may also be more easily produced. Future studies of meningococcal Opa PV would ideally include the examination of a number of diverse strains, including those from invasive disease and carriage.