Phasevarions Mediate Random Switching of Gene Expression in Pathogenic Neisseria

1The pathogenic Neisseria are host-adapted human pathogens that pose a significant health problem worldwide. Neisseria meningitidis colonizes the upper respiratory tract and causes meningitis and septicemia. Neisseria gonorrhoeae colonizes the genitourinary tract and can cause a spectrum of disease ranging from uncomplicated mucosal infection to disseminated gonococcal infection. There is no N. gonorrhoeae vaccine, and no fully protective vaccine for N. meningitidis. Vaccine development has been hampered due to the high frequency of antigenic and phase variation of surface structures typical of these organisms.

2Phase variation is the high frequency reversible on/off switching of gene expression and is commonly mediated by mutations in simple tandem DNA repeats in the open reading frame or promoter region of genes encoding surface expressed virulence determinants. The independent, random switching of these genes results in phenotypically diverse populations that enables rapid adaptation to host environments and evasion of immune responses. While phase variation is typically associated with genes encoding surface structures, several host-adapted bacterial pathogens have methyltransferases (mod genes) associated with type III restriction modification (R-M) systems that contain simple tandem DNA repeats that have been proven to phase vary (Pasteurella haemolytica, Haemophilus influenzae and Helicobacter pylori ) or predicted to phase vary (N. meningitidis, N. gonorrhoeae,, and Moraxella catarrhalis ), as reviewed in Fox et al.

3R-M systems are ubiquitous in bacteria and confer protection to the bacterial host against invasion by foreign DNA. R-M systems are classified into three groups; Types I, II or III on the basis of subunit composition, DNA cleavage position, sequence-specificity and co-factor requirements. Type III systems are composed of a methyltransferase (modification, mod) gene and an endonuclease (restriction, res) gene, whose products form a two-subunit enzyme – Mod and Res. Type III systems are unusual in that Res must form a complex with Mod to be functional, however, Mod can function independently of Res. The Mod subunit contains several conserved motifs in the N- and C-terminal regions and the central region contains the DNA-recognition domain that dictates sequence specificity.

4In H. influenzae, the random switching of the modA gene controls expression of a phase variable regulon of genes (a “phasevarion”), via differential methylation of the genome in the modA ON and OFF states. This was the first report of the coordinated random switching of a “regulon” of genes and, considering the wide distribution of phase variable type III R-M systems, may represent a widely used mechanism in bacterial pathogens. In this study we investigate the phase variable type III R-M systems of pathogenic Neisseria to determine whether they play a role in gene regulation and virulence.

5To investigate whether the type III R-M systems of the pathogenic Neisseria behave as a phasevarion, we first carried out a phylogenetic analysis of mod genes associated with type III R-M systems of N. meningitidis and N. gonorrhoeae. A comparison of the available genome sequences revealed that each strain contains two distinct phase variable mod genes, which we define as modA and modB, that share only 37% similarity to each other along the full length of the Mod deduced amino acid sequence. Both genes contain tracts of simple tandem repeats, 5′-AGCC-3′ (modA) and 5′-CCCAA-3′ (modB), that mediate phase variation of mod gene expression (Figure 1). ModA is highly homologous (>90% identity along the length of the Mod deduced amino acid sequence excluding the variable DNA recognition domain) to the mod gene of H. influenzae strain Rd (HI1058/56),. Differences in the modA DNA recognition domain (Figure 1) have previously been observed in H. influenzae with 17 distinct mod alleles defined in this organism (modA1–17; ). The Neisseria modA alleles present in the genome strains surveyed have the designations modA11, 12 and 13 and share >94% similarity to each other along the length of the Mod deduced amino acid sequence, excluding the variable DNA recognition domain. Our recent work also shows that the modA gene of H. influenzae and Neisseria are essentially the same gene with evidence of horizontal transfer of this gene in both directions between these organisms. Unlike modA, the modB gene appears to be specific to Neisseria species. Two distinct modB alleles, modB1 and 2, distinguished by differences in their DNA recognition domain, were also observed (Figure 1). ModB1 and 2 share >95% similarity to each other along the length of the Mod deduced amino acid sequence, excluding the variable DNA recognition domain (which shares <33% identity).

6To investigate whether additional alleles of modA and modB are present in these organisms, and to look at the distribution of mod alleles and their repeat sequence type and number, sequence analysis of a large, genetically diverse set of N. meningitidis and N. gonorrhoeae isolates was performed. This analysis revealed that all strains examined contained both modA and modB genes. Sequencing of the repeat region of the mod alleles revealed that the repeat numbers vary in length between different strains, resulting in the mod genes being in-frame (ON) or out-of-frame (OFF) for expression, consistent with phase variation of the mod genes (Table S1, Table S2). The N. gonorrhoeae strains contained either the modA12 or modA13 allele, and only the modB1 allele. One strain was found not to have a modB gene (Figure 1, Table S2).

7A complete analysis of modA allele distribution was conducted in N. meningitidis, which has a well characterized population structure defined by multi locus sequence typing (MLST; ). The complete 107 strain MLST modA survey revealed that the majority of N. meningitidis strains had either the modA11 or modA12 allele, with modA15 found in two strains and modA4 and modA18 found in one isolate each (Figure 1, Figure 2A and 2B, Table S1). The most notable associations were in capsule type, where 100% of serogroup A strains and 92% of serogroup C strains contained the modA12 allele (Table 1, Figure 2B). Some association with clonal complex was also observed, with meningococci belonging to the ST-32 clonal complex predominantly harbouring the modA11 allele. Further clustering could be seen among ST-41/44 and ST-8 clonal complexes. Unlike N. gonorrhoeae, which contained only one type of modB allele, modB1, N. meningitidis strains contained either modB1 or modB2. There were seven strains, all from the ST-32 group, which contained point mutations in modB1 suggesting the gene is inactive in these strains (Figure 1, Table S1).

8The modA genes of N. meningitidis and N. gonorrhoeae have two alternate initiation codons (Distal ATG and Proximal ATG) that are predicted to code for proteins of either 589 aa/640 aa or 707 aa/758 aa, depending on the number of tetranucleotide repeats that are present (Figure 3A). As this study is focused on modA phase variation and expression, a clear understanding of the relationship between tetranucleotide repeat number and modA expression was established. ModA expression was examined in the three frames; Distal, Proximal and OFF (which has no candidate ATG and a stop codon immediately after the 5′-AGCC-3′ repeats), by constructing a modA::lacZ reporter fusion in the N. meningitidis strain MC58 chromosome (Figure 3A and Figure S1). Maximal expression was found to be from the Distal ATG only and unlike H. influenzae strain Rd, minimal expression was observed from the Proximal ATG and the OFF frame (Figure 3B and 3C).

9Natural modA ON and OFF colonies of N. meningitidis strain MC58 and N. gonorrhoeae strain FA1090 were required for microarray analysis and biological characterization experiments. N. meningitidis strain MC58 has 21 5′-AGCC-3′ repeats resulting in the modA11 gene being out-of frame (OFF) from the Distal ATG. Single colonies of MC58 were picked and screened by PCR and sequencing to find modA11 in-frame (ON) with the Distal ATG (see Figure 3). Similarly, single colonies of N. gonorrhoeae strain FA1090 were picked and screened by PCR and sequencing to find modA13 in-frame and also out-of-frame with the Distal ATG. During this process, ModA13 repeat tracts ranging from 13 (ON) to 26 (OFF) and also 37 (ON) were observed, demonstrating phase variation of N. gonorrhoeae strain FA1090 modA13 (results not shown).

10Having established the relationship between modA repeats and modA expression (see Figure 3), we were in a position to conduct studies to determine whether phase variation of the various modA alleles in pathogenic Neisseria resulted in changes in gene expression. These studies were initiated with N. meningitidis strain MC58 modA11 gene, where the modA11 gene was inactivated by insertion of a kanr cassette to make the mutant strain MC58 modA11::kan (Figure S2). Wild-type MC58 modA11 ON and MC58 modA11::kan were compared by microarray analysis using N. gonorrhoeae/meningitidis genome arrays (Materials and Methods). Initially, microarray analysis was performed using RNA isolated from wild-type MC58 modA11 ON and MC58 modA11::kan strains grown under standard culture conditions. However, these studies revealed no statistically significant difference in gene expression. Experiments were then performed in which N. meningitidis were cultured under iron-limiting conditions to more closely reflect in vivo conditions. Using this more physiologically relevant culture condition, many genes were found to have an expression ratio of 1.5-fold and over, with 162 genes up-regulated in MC58 modA11::kan relative to wild-type and 123 genes down-regulated, confirming modA11 phase variation has an influence on gene expression (Table 2, Table S3). Five of these genes encode surface exposed proteins, including NMB1540 (lbpA) and NMB1541 (lbpB), encoding LbpA and LbpB respectively, which are part of the lactoferrin receptor that allows acquisition and binding of iron from lactoferrin containing compounds. LbpA is the TonB-dependent integral outer membrane lactoferrin receptor and iron transport channel. LbpB is an accessory lipoprotein anchored to the outer membrane that contributes to lactoferrin binding/use,. The lactoferrin receptor is a potential vaccine candidate in N. meningitidis. Quantitative real time PCR (RT-PCR) was performed to confirm that the lbpA and lbpB genes were expressed at a higher level in MC58 modA11::kan compared to the MC58 modA11 ON parent strain (see Figure 4A). Altered expression was further confirmed by an lbpB::lacZ fusion (Figure S3) located on the chromosome of each strain, which showed ∼2-fold higher expression in the modA11::kan mutant strain compared to the modA11 ON parent strain (see Figure 4B). Consistent with this, western blot analysis confirmed the effect of modA11 phase variation on expression of LbpA, with an apparent reduction in expression when modA11 is ON (Figure 4C). The same effect is seen when comparing a wild-type modA11 ON strain to either a modA11::kan mutant or a natural phase variant in which the modA11 gene had switched OFF due to an alteration in the 5′-AGCC-3′ repeat tract (from 22 to 21 AGCC repeats; see Figure 4C), confirming the regulation of LbpA by ModA11 is not related to the use of a kanr inactivated modA11 gene.

11A similar microarray study was conducted using a N. meningitidis modA12 clinical isolate, B6116/77. To determine whether phase variable expression of the N. meningitidis strain B6116/77 modA12 gene also led to alteration in global gene expression, the modA12 gene was inactivated by insertion of a kanr cassette to make the mutant strain B6116/77 modA12::kan (Figure S2). Wild-type B6116/77 modA12 ON and B6116/77 modA12::kan were compared by microarray analysis using N. gonorrhoeae/meningitidis genome arrays. Experiments were performed under the same conditions as described above. Twenty six genes were found to have an expression ratio of 1.5-fold and over, with 14 genes up-regulated in B6116/77modA12::kan relative to wild-type and 12 genes down-regulated, confirming modA12 phase variation has an influence on gene expression (Table 2, Table S4). The set of genes differentially expressed in the modA12 mutant were different to the modA11 set of genes. This is consistent with the differences in the DNA recognition domain between modA11 and modA12, and confirms that these distinct alleles control different phasevarions in N. meningitidis.

12An additional phasevarion study was conducted in N. gonorrhoeae. In this case, a modA13 knockout mutant was constructed by interrupting the modA13 gene with a kanr cassette (Figure S2). Comparison of the phenotype of the FA1090 modA13::kan mutant strain with wild-type FA1090 modA13 ON formed the basis of expression and phenotypic studies.

13Global gene expression was compared between wild-type FA1090 modA13 ON and FA1090 modA13::kan under iron-limiting conditions. 34 genes were up-regulated in FA1090 modA13::kan relative to wild-type, and 20 genes were down-regulated (Table 2, Table S5). Five of the differentially regulated genes have obvious roles in virulence; four in oxidative stress and one in antimicrobial resistance. NGO0929 (metF) and NGO0928 (metE) are part of the MetFE operon, which plays a role in the methylation of homocysteine, the final step of methionine biosynthesis, and is involved in defence against oxidative stress. NGO0554 encodes a gonococcal-specific hypothetical protein that is shown to protect against damage caused by high levels of H2O2. NGO0650 (recN) encodes the DNA repair protein RecN. The gonococcal RecN protein is demonstrated to be involved in DNA repair and DNA transformation and plays an important role in H2O2 damage protection as well as resistance to killing by polymorphonuclear leukocytes. NGO1368 (mtrF) encodes the inner membrane protein, MtrF, and has been shown to have a role in antimicrobial resistance.

14Our phylogenetic analysis revealed that N. gonorrhoeae strains have one of two distinct modA alleles (modA12 or 13; see Figure 1), indicating that different phasevarions may exist within N. gonorrhoeae and that strains with the same mod allele may regulate similar sets of genes. To determine if a strain with the same DNA recognition domain as FA1090 (modA13 allele) would result in the same set of genes being regulated, we chose a N. gonorrhoeae clinical isolate, strain O1G1370, from a representative set of N. gonorrhoeae strains (Table S2) that also contains a modA13 allele. A modA13::kan knockout mutant was made using the same approach as described for FA1090 (Materials and Methods). Quantitative RT-PCR on the metF, mtrF and NGO1581 genes confirmed that metF and mtrF, which are up-regulated in expression in the FA1090 modA13::kan mutant, are also up-regulated in the O1G1370 modA13::kan mutant (Table 2). NGO1581, which was down-regulated in the FA1090 modA13::kan mutant, is also down-regulated in the O1G1370 modA13::kan mutant (Table 2). Furthermore, similar results were seen with quantitative RT-PCR on the same set of genes when comparing a wild-type FA1090 modA13 ON strain to a natural phase variant of FA1090 in which the FA1090 modA13 gene had switched OFF due to an alteration in the 5′-AGCC-3′ repeat tract (from 37 to 26 AGCC repeats) (Table 2), confirming the regulation of this set of genes by ModA13 is not related to the use of a kanr inactivated modA13 allele.

15When we conducted microarray analysis of a N. gonorrhoeae modA12 strain 96D551, comparison of 96D551 modA12 ON and 96D551 modA12::kan (OFF) strains revealed a distinct set of genes being regulated compared to the modA13 data above (Table 1, Table S6). These results are consistent with the differences in the DNA recognition domain between modA13 and modA12 (see below), and supports the idea that these distinct alleles control different phasevarions in N. gonorrhoeae. However, unlike the modA11, modA13 and modA12 (N. meningitidis) expression studies described above, the 96D551 modA12 ON and 96D551 modA12::kan (OFF) strains showed a significant difference in growth rate for the cultures used to make RNA (Figure S4D). We cannot rule out the possibility that these differential growth rates may have influenced the gene expression data in Table 1 and Table S6.

16In all cases, phase variation or mutagenesis of modA of pathogenic Neisseria results in altered gene expression, defining these systems as functional phasevarions. In order to determine whether the observed changes in modA expression correspond to global changes in DNA methylation, thereby indicating this as the likely mechanism of gene control, it was necessary to identify one or more of the modA target sites. In addition to confirming global changes in methylation, target site identification would also facilitate future studies on the molecular mechanisms operating at individual promoters within the phasevarion. In order to identify methylation target sites, a strategy based on inhibition of DNA restriction was used. In initial studies, plasmid pCmGFP was isolated from N. gonorrhoeae or N. meningitidis modA11, modA12 or modA13 ON strains and their corresponding modA::kan mutants, and digested with a range of restriction enzymes known to be inhibited by methylation of an adenine within their recognition sequence (see Materials and Methods). Differences in digestion patterns between plasmid extracted from modA ON cells (ModA methylated DNA) and modA::kan cells (DNA not methylated by ModA) would indicate an overlap of the respective ModA methylated target and the restriction enzyme used. No such fortuitous inhibition pairs were seen with modA11 or modA12 strains, but were with modA13. Figure 5A shows an obvious difference in the restriction pattern of plasmid extracted from modA13 ON and modA13::kan cells, indicating overlap between the ModA13 site and ApoI. The recognition sequence of ApoI is 5′-RAATTY-3′. The specific ApoI site displaying inhibition (5′-AAATTC-3′) was not unique on the plasmid. Comparison between the inhibited ApoI site and other ApoI sites whose sequence was also 5′-AAATTC-3′ revealed that the overlap with ModA13 must be 5′ to the ApoI site. As seen in Figure 5B, methylation of the different adenines in the ApoI recognition site inhibits digestion to varying degree. Since over-digestion did not result in the 722 bp band being cleaved into the two smaller fragments, it can be assumed that the first adenine of the ApoI sequence was not the methylation target, as methylation of this adenine would result in only 10% inhibition of restriction, an effect which could potentially be overcome by over-digestion. Therefore, depending on which of the other two adenines was methylated, the ModA13 recognition sequence must be found within 5′-CAGAAA-3′.

17To confirm which adenine is the ModA13 methylation target, and to further specify the ModA13 recognition sequence, overlaps of the putative ModA13 recognition sequence and ApoI restriction sites were identified on the FA1090 chromosome. Chromosomal DNA was extracted from FA1090 modA13 ON and FA1090 modA13::kan cells, digested with ApoI and examined by Southern blot. Inhibition of ApoI restriction at the internal ApoI/ModA13 overlap results in the presence of one large band at 3408 bp, while the unmethylated DNA (modA13::kan) is cleaved at this site by ApoI into two smaller bands of 2500 and 908 bp (Figure 5C). This confirms that ApoI restriction can be inhibited in DNA methylated by ModA13, as shown by the plasmid digest (Figure 5A). Similar studies were done with three other ApoI restriction sites in the genome that overlap ModA13 with 5′-TAGAAA-3′, 5′-GAGAAA-3′ or 5′-AAGAAA-3′. In each of these cases, ApoI restriction of ModA13 methylated DNA was inhibited compared with DNA extracted from modA13::kan cells (data not shown). This confirms the ModA13 recognition sequence to be 5′-AGAAA-3′.

18To identify which of the two potential adenines in 5′-AGAAA-3′ is methylated, ModA13 recognition sequences in the FA1090 chromosome were identified which overlapped with restriction enzymes other than ApoI and were known to be inhibited by methylation of adenines. Two of these enzymes were HindIII (5′-AAGCTT-3′) and AluI (5′-AGCT-3′). Although the recognition sequence of HindIII contains two adenines (Figure 5E), both these adenines are part of the ModA13 recognition sequence. Methylation of the adenine in the AluI recognition sequence is known to result in complete inhibition. When chromosomal DNA digested with AluI is probed using a PCR product containing an AluI/ModA13 overlap (Figure 5D) no difference in restriction is seen between the modA13 ON and modA13::kan lanes, indicating that the common adenine in this overlap is not methylated by ModA13. This suggests that ModA13 methylates AGAAA on the second most 3′ adenine of recognition site 5′-AGAAA-3. Information on the sensitivity of HindIII to hemimethylation is only known for the 5′ adenine of the HindIII recognition site. Hemimethylation of this adenine results in a 95% inhibition of restriction. Using a random site in the FA1090 genome where the overlap between the ModA13 target site and HindIII resulted in the 5′ adenine of HindIII site corresponding to the second last adenine of the AGAAA (see Figure 5E) we were able to determine whether this was the residue methylated by ModA13. The results shown in Figure 5E confirm the expected restriction inhibition phenotype allowing us to conclude that the ModA13 methylation site is AGAAmA, with the methyl group being added to the third adenine in the sequence.

19Having established the ModA13 target sequence, we tested DNA derived from a modA12 strain and confirmed that the ModA12 target site was distinct as there is no difference between modA12 ON and modA12::kan (OFF) DNA in a ModA13/ApoI inhibition assay (Figure 5F). Analysis of the FA1090 genome has revealed a total of 5135 ModA13 target sites.

20To determine whether the phasevarion mediated changes in gene expression correspond to altered phenotypes in model systems, we chose to focus on the modA13 allele of N. gonorrhoeae. Several model systems were available from our previous studies on oxidative stress, biofilm formation and bacterial - host cell interactions –. Furthermore, strains FA1090 and O1G1370 provided an opportunity to test the reproducibility of key phenotypes in two independent modA13 strains.

21Previous studies using N. gonorrhoeae strain FA19 demonstrate that mtrF is required for induction of high-level antimicrobial resistance to Triton X-100 by gonococci. Our data show that MtrF expression is up-regulated in the modA13 mutant relative to the wild-type under iron-limiting conditions. To test whether differences in antimicrobial-resistance could be observed between wild-type FA1090 modA13 ON and the FA1090 modA13::kan mutant, an antimicrobial-resistance assay was performed using increasing Triton X-100 concentrations (Figure 6A). The FA1090 modA13::kan mutant was found to be more resistant than wild-type FA1090 modA13 ON, consistent with the higher level of expression of MtrF in this modA13 OFF strain. As modA13 ON is free to phase vary to OFF, and OFF cells appear to be fitter in this assay, the status of modA13 expression was monitored by PCR with fluorescent primers across the repeat region to determine whether ON to OFF phase variants had been selected in the survivor colonies at various Triton X-100 concentrations. This analysis revealed that the FA1090 modA13 ON culture plated on zero Triton X-100 remained ON, with only 11.21% OFF cells. However, cells plated on increasing Triton X-100 concentrations changed to 46.99% OFF, 80.29% OFF and 80.15% OFF over the course of the assay for 40, 50 and 60 µg/ml Triton X-100, respectively (Figure 6B).

22A number of studies have shown that N. gonorrhoeae can form a biofilm in a continuous-flow chamber and over primary human genital tract epithelial cells in culture,. Biofilms provide a number of advantages in survival of the bacteria. It is suggested that biofilm formation by N. gonorrhoeae may contribute to its ability to persist in an asymptomatic state in the female genital tract. In addition, bacteria within biofilms show increased resistance to antimicrobial agents, and links between biofilm formation and oxidative stress defenses have been observed in N. gonorrhoeae.

23The ability of N. gonorrhoeae O1G1370 modA13 ON, O1G1370 modA13 OFF and O1G1370 modA13::kan (OFF) to form a biofilm was evaluated after two days of growth under continuous flow conditions. Three-dimensional images of these biofilms were created in Volocity (Materials and Methods). These images show that O1G1370 modA13::kan and modA13 OFF form a thick and dense biofilm, while O1G1370 modA13 ON forms an extremely weak biofilm with a few sparse patches of cells scattered across the surface of attachment (Figure 7A). The O1G1370 modA13 ON strain also formed biofilms with lower maximum thicknesses than the O1G1370 modA13::kan and O1G1370 modA13 OFF strains. (Figure 7D). Scanning electron microscopy of the surface of the biofilm taken at 5,000× magnification shows that there are gaps between clusters of biofilm in the O1G1370 modA13 ON strain, unlike the O1G1370 modA13 OFF and O1G1370 modA13::kan strain biofilms, where there are no areas where the glass surface of attachment is visible. There are also large areas where no biofilm is present in the O1G1370 modA13 ON samples (Figure 7B). Scanning electron microscopy taken at 15,000× magnification shows that O1G1370 modA13::kan and O1G1370 modA13 OFF form biofilms that are tightly enmeshed in an extracellular material that obscures the structure of individual cells, while cells in the modA13 ON biofilm are clearly distinguishable (Figure 7C). Transmission electron microscopy shows that O1G1370 modA13::kan forms a biofilm where individual cells are shedding copious amounts of membrane, as seen in the numerous enclosed membrane blebs on the surface of the cells, while there is no evidence of blebbing in the O1G1370 modA13 ON biofilm. Cells in the O1G1370 modA13 OFF biofilm also appear to be blebbing, like those in O1G1370 modA13::kan biofilm, as numerous blebs can be seen forming on the surface of the O1G1370 modA13 OFF strain. These electron micrographs suggest that the extracellular matrices of the O1G1370 modA13::kan and O1G1370 modA13 OFF biofilms may be at least partially composed of fused membrane blebs (Figure 7C). COMSTAT was used to quantitatively assess the biomass, and average and maximum thickness of confocal z-series photomicrographs taken for each flow chamber. COMSTAT analysis showed that the O1G1370 modA13::kan and O1G1370 modA13 OFF strains form significantly thicker biofilms with significantly more biomass than the O1G1370 modA13 ON strain. Specifically, O1G1370 modA13 ON had 3.5% of the biomass and 4.2% of the thickness of the O1G1370 modA13::kan mutant on average and 4.7% of the biomass and 5.2% of the thickness of the O1G1370 modA13 OFF (Figure 7E). Similar results were observed using N. gonorrhoeae strains FA1090 modA13 ON and FA1090 modA13::kan (Figure S5).

24The use of primary human cervical epithelial (pex) cells as a model system of gonococcal cervicitis is well established and has been used in a number of studies, such as the examination of the role of oxidative stress regulators in host-pathogen interactions,. To determine the biological significance of O1G1370 modA13 expression using this pex cell culture model, we performed quantitative association, invasion, and survival assays using O1G1370 modA13 ON, O1G1370 modA13 OFF, and O1G1370 modA13::kan mutant gonococci (Figure 8). These data revealed that there was no significant (P≥0.2338) difference in the ability of the O1G1370 modA13 OFF and O1G1370 modA13::kan strains to adhere to, invade, or survive within pex cells. This confirmed that the modA13::kan knockout allele behaves in the same way as a natural phase variant modA13 OFF strain. In contrast, behavior of the O1G1370 modA13 ON strain was significantly (P≤0.001) different from that obtained with the use of either the O1G1370 modA13 OFF or O1G1370 modA13::kan strains in parallel assays. In this regard, a modA13 ON phenotype resulted in the increased ability of gonococci to associate with pex cells, whereas a modA13 OFF configuration augmented the ability of gonococci to invade (Figure 8A, invasion index) and survive within pex cells following invasion (Figure 8A, survival index). These data suggest a possible role for Mod-dependent phase variation in promoting the adaptive changes required for gonococci to switch from an extracellular to an intracellular existence. This idea is supported by our observation of selection for a switch from ON to OFF in the O1G1370 modA13 ON strain. Fragment analysis confirmed that the O1G1370 modA13 ON inoculum, which contains only 5.86% OFF, changes to ∼49.84% OFF by the time the 3 hour intracellular survival sample was taken (Figure 8B, Table S9). An independent N. gonorrhoeae modA13 strain, 1291, displayed the same intracellular survival and modA13 switching phenotype (Figure S6).

25The pathogenic Neisseria are the archetypal organisms for the study of phase variation. Simple tandem repeats are typically associated with individual genes involved in biosynthesis of a surface component, such as an outer membrane protein, or a poly- or oligosaccharide. The consequence of hyper-mutation of these simple tandem repeats is phase variable expression of these genes, i.e., the presence or absence of a single component on the surface of the cell. Independent, random switching of many different phase variable genes encoding these surface structures leads to a combinatorial effect generating a huge number of alternate combinations of surface components. Phase variation, in conjunction with antigenic variation of the major antigen pili, leads to evasion of host immune responses. The distinction between the phasevarion and typical phase variation of genes encoding surface factors is that the ON/OFF switching of the phasevarion methyltransferase mediates expression changes in multiple genes in a coordinated manner.

26Our phylogenetic studies on the mod genes of a collection of pathogenic Neisseria strains reveal that differences in the DNA recognition domain within the mod gene results in distinct mod alleles. R-M systems show extreme diversity in their DNA sequence recognition specificities. DNA sequence specificity in type III R-M systems is conferred by the Mod subunit. Based on differences in the DNA recognition domain, three major modA alleles were found - modA11, modA12, modA13, and two distinct modB alleles were found - modB1 and modB2. This suggests the possibility that multiple phasevarions exist within the pathogenic Neisseria, each regulating a different set of genes. Furthermore, as each strain has both modA and modB, and these genes switch independently, there are four potential combinations of mod gene expression (ON/ON;ON/OFF;OFF/ON;OFF/OFF). We confirmed that two strains with the same DNA recognition domain (modA13 allele) regulated the same set of genes, while, N. meningitidis modA11 and modA12 were found to regulate the expression of different sets of genes, consistent with differences in their DNA recognition domain. In this study we also identified the recognition sequence for ModA13 as 5′-AGAAmA-3′. In all, five randomly selected sites were tested for ModA13 inhibition of digestion in genomic DNA from modA13 ON and OFF strains. All five sites tested displayed the expected inhibition of digestion phenotype (with either ApoI or HindIII), supporting the hypothesis that when expressed, ModA13 methylates all AGAAA sites in the genome, and thereby indicating this as the likely mechanism of gene control. Identification of the ModA13 target site is facilitating current studies on the molecular mechanism of regulation operating in the promoters of genes controlled by the modA13 phasevarion (see Figure S7).

27Gene regulation through the methylation of specific DNA sequences by methyltransferases has been reviewed,, but has focused on the role of Dam methylation. Dam methylation has been reported to be essential for bacterial virulence. In Salmonella species, dam mutants are highly attenuated for virulence and have been proposed as live vaccine candidates –. In addition, mutations in Dam attenuate the virulence of several other pathogens –. In these studies the mechanisms of attenuation (genes regulated) are unknown. In contrast, there are a few well-established examples of Dam mediated phase variation of genes encoding individual virulence factors–, for example the pyelonephritis-associated pilus (pap) operon in uropathogenic Escherichia coli,,. The fundamental characteristic of these DNA-methylation-dependent phase variable systems is that the target site's methylation state affects the DNA binding of a regulatory protein, which directly regulates transcription. The key point to note is that the Dam methyltransferase itself does not phase vary, nor are there any examples of Dam itself being regulated by an environmental signal. These systems are not analogous to phasevarions, but do provide examples of how DNA methylation may alter gene expression at a bacterial promoter. In the case of N. meningitidis, most strains have been found to be dam negative, as are all of the N. meningitidis and N. gonorrhoeae strains used in expression profile analysis and functional assay in this study (result not shown).

28The question of whether the phase variable mod genes are associated with a functional type III restriction system remains to be fully resolved. In H. influenzae we have reported inactivating mutations in the res gene that is required for restriction function in strains containing phase variable mod genes. We propose that in these cases the R-M system function has been lost and that the modA gene is dedicated to a gene regulation function. We have observed a similar inactivating mutation in the res gene associated with modA11 of N. meningitidis (Table S2), and in N. gonorrhoeae, a 250 aa in-frame deletion has been observed in the res gene associated with modA11 and modA12 (see Table S3), supporting a dedicated gene regulation function for mod genes in pathogenic Neisseria.

29Several of the genes regulated by the modA11 phasevarion of N. meningitidis strain MC58 are outer membrane proteins, including the vaccine candidates LbpA and LbpB. These are typical of the class of gene that have evolved phase variation mechanisms under immune selection. It is clear that phasevarion mediated phase variation of candidate vaccine antigens has the potential to mediate escape from a vaccine primed immune response. In contrast, none of the typical genes encoding outer membrane structures were influenced by the modA13 phasevarion in strain FA1090. Instead, the genes under phasevarion control were involved in functions such as oxidative stress, antibiotic resistance, and transport of nutrients. For example, the MetFE operon, which plays a role in the methylation of homocysteine, the final step of methionine biosynthesis. MetE catalyses the methylation of homocysteine using a methyl group that is donated by the metF gene product, 5-methyltetrahydrofolate. In E. coli, a correlation is shown between oxidative stress, methionine availability, and MetE, where MetE is inactivated under conditions of oxidative stress. In addition, MtrF, which is required for high-level, hydrophobic agent-resistance that is mediated by the MtrC-MtrD-MtrE efflux pump,, is controlled by the modA13 phasevarion. The active efflux of antimicrobial agents from the cell by this systems is recognized as a major contributor to bacterial resistance to antibiotics,. Altered expression of this group of genes is more consistent with a switch between cell types that are more suited to alternate physical environments, rather than switching to enable evasion of a particular primed immune response.

30Phenotypic analyses of modA13 ON, modA13::kan mutant or modA13 OFF revealed distinct behavior in the model systems tested: modA13::kan and modA13 OFF mutant cells were superior at formation of a biofilm. Bacteria within biofilms display increased resistance to antimicrobial agents,. In addition, links between biofilm formation and oxidative stress defenses have been observed in N. gonorrhoeae. Consistent with this observation, genes involved in oxidative stress and antimicrobial susceptibility were found to be up-regulated in expression in a modA13::kan mutant. Furthermore, a modA13::kan mutant was also found to be more resistant to antimicrobial agents in a Triton X-100 killing assay. Finally, the modA13::kan mutant and modA13 OFF strains were more fit in an intracellular survival assay in the pex model system, and this assay selected for a switch from ON to OFF during the course of the 3 hour assay. These observations are consistent with random generation of two populations containing different cell types with distinct niche specialization. The observation of common biofilm and intracellular survival phenotype in independent modA13 strains suggests that these can be attributed directly to modA13 phase variation, rather than an independent phase variation event in an unrelated gene, and that these may be key aspects of gonococcal - host interactions. This study shows that all modA alleles regulate gene expression of many genes, and that we observe distinct behavior of, and switching between, modA ON and OFF states in model systems. It is clear that any future study of pathogenic Neisseria that investigates gene expression or behavior of strains in model systems should take modA phase variation into account.

31Our initial microarray studies resulted in data with no statistically significant difference in the regulation of any gene. Changing the culture conditions to iron limitation resulted in the differential expression of 54 genes (Table S5). This change in expression was not due to differential growth rates (Figure S4) or a direct effect of iron on expression of the Mod methyltransferase (Figure S8). Clearly, a difference in gene expression can only be detected if the genes in question are being expressed. It is well established that many genes are expressed under iron-limiting conditions in pathogenic Neisseria, via the Fur regulon, either directly, or due to cascade effects resulting from activation of the Fur regulon,. One limitation of the data presented in this study is that the gene expression profile analysis of the phasevarions was only done under one culture condition. Using different physiologically relevant conditions may enable other virulence-associated genes under the control of the phasevarion to be discovered. For example, it is established that the interaction of N. meningitidis with epithelial cells induces changes in the expression of 347 genes.

32The results presented in this paper, in conjunction with our recent studies in H. influenzae,, provide confirmation of a role for phase variable mod genes associated with type III R-M systems in gene regulation. The widespread distribution of phase variable R-M systems in host-adapted pathogenic bacteria suggests that this novel mechanism of coordinated random switching of multiple genes may be a commonly used strategy for generation of distinct, “differentiated”, cell types with distinct niche specialization in host adapted bacterial pathogens.

33N. meningitidis and N. gonorrhoeae strains were grown at 37°C with 5% CO2 in either GC broth or GC agar with IsoVitaleX (Becton Dickinson). E. coli strains DH5α and JM109 (Promega) were used to propagate plasmids and were grown at 37°C in Luria-Bertani (LB) broth supplemented with either ampicillin (100 µg/ml) or kanamycin (100 µg/ml).

34All enzymes were sourced from New England Biolabs. Sequencing was performed on PCR products using QiaQuick PCR purification kit (Qiagen) and Big-Dye (Perkin Elmer) sequencing kits. Data was analysed using MacVector v9.0 (Accelrys).

35PCR products specific for the DNA recognition domain and repeat regions of modA and modB were generated using the primers listed in Table S7. N. meningitidis isolates and N. gonorrhoeae DGI and MI clinical isolates were used as templates. The reaction was performed in 50 µl using 1× Taq buffer, 1.5 mM MgCl2, and 1 unit of Taq DNA polymerase (Promega) with the following cycling conditions for the DNA recognition domain: 30 cycles of 94°C for 30 sec, 57°C for 30 sec, 72°C for 2 min and 1 cycle of 72°C for 7 min with 5 µM of the primer pair ModADRDF and ModADRDR for modA or ModBDRDF and ModBDRDR for modB. For modA, a 597 bp region containing the DNA recognition domain (393 bp downstream of ModADRDF and 101 bp upstream from ModADRDR) was compared to the genome strains to determine the modA allele group. For modB a 537 bp region containing the DNA recognition domain (461 bp downstream from ModBDRDF and 285 bp upstream from ModBDRDR) was compared to the genome strains to determine the modB allele group. The following cycling conditions were used for the repeat region: 30 cycles of 94°C for 30 sec, 57°C for 30 sec, 72°C for 30 sec and 1 cycle of 72°C for 7 min with 5 µM of the primer pair ModAF and ModAREPEATR or ModBREPEATF and ModBREPEATR. PCR products were cleaned using the QIAquick PCR Purification Kit (Qiagen).

36PCR products specific for the res gene were generated using the primers listed in Table S7. N. meningitidis isolates and N. gonorrhoeae DGI and MI clinical isolates were used as templates. The reaction was performed in 50 µl using 1× Taq buffer, 1.5 mM MgCl2, and 1 unit of Taq DNA polymerase (Promega) with the following cycling conditions for the DRD: 30 cycles of 94°C for 30 sec, 55°C for 30 sec, 72°C for 3 min and 1 cycle of 72°C for 7 min with 5 µM of the primer pair ResF and ResEDR2. PCR products were sequenced using ResF, ResR, ResEDF2 and ResEDR2.

37The modA nucleotide sequences were assembled using the Staden sequence analysis package and all sequences aligned manually in the Seqlab alignment program (Genetics Computer Group, Madison, Wis.). Phylogenetic analysis was undertaken using the software package ClonalFrame version 1.1, which is a statistical model-based method initially described for inferring bacterial clonal relationships using multilocus sequence data. Inference is performed in a Bayesian framework and a neutral coalescent model is assumed based on the hypothesis that the bacteria in the sample come from a constant-sized population in which each bacterium is equally likely to reproduce, irrespective of its previous history. The key assumption of ClonalFrame is that recombination events introduce a constant rate of substitutions to a contiguous region of sequence with the end result that a clonal frame can be inferred. In the present study, over 50,000 iterations were performed with every hundredth tree sampled after which, a 95% majority-rule consensus tree was derived. ClonalFrame is available at available at http://bacteria.stats.ox.ac.uk. The modA gene is composed of relatively conserved N and C-terminal regions with the DNA recognition domain in between. Consequently, sequence input into ClonalFrame was undertaken by firstly adding the N-terminal region starting at bp 359 in the modA gene belonging to the reference N. meningitidis isolate MC58, followed by the DNA recognition domain occurring from 416 to 795 bp and ending with the C-terminal region 796 to 1242 bp. Annotation was then undertaken by importing the tree into the Molecular Evolutionary Genetics Analysis software package (MEGA ver 4.0).

38Plasmid pCmGFP (Source M A Apicella) was extracted from N. meningitidis strain C311. Primers used to sequence this plasmid are listed in Table S7. Sequencing reactions were prepared using the plasmid as template and Big-Dye sequencing kit (Perkin-Elmer). Samples were analysed using a 3130×l Capillary Electrophoresis Genetic Analyser (Applied Biosystems International). Data were analysed and plasmid map constructed using MacVector (version 9.0). The plasmid sequence is deposited in GeneBank under accession number FJ172221).

39Plasmid pCmGFP was extracted from N. gonorrhoeae strain Fa1090 modA13 ON and FA1090 modA13::kan, N. gonorrhoeae strain 96D551 modA12 ON and modA12::kan, and N. meningitidis strain MC58 modA11 ON and modA11::kan cells using the Qiagen Plasmid Midi Kit (Qiagen, Doncaster, Vic., Au). The modA13 expression status of the ON cultures was verified by fragment analysis. 2 micrograms of each plasmid was digested overnight with a range of restriction enzymes (AcuI, AluI, ApoI, BsgI, BsmI, DdeI, DpnII, DraI, HinfI, HpyI88I, HpyI88III, MboI, MboII, MseI, MslI, NlaIII, TaqI, Tsp509I) according to manufacturer's instructions and the resulting fragments were separated on a 1.5% agarose gel with TBE at 120 V for 1 hour and visualised under UV illumination.

40Chromosomal DNA was extracted from FA1090 modA13 ON, modA13 OFF and FA1090 modA13::kan. Mod expression status of the modA13 ON and modA13 OFF cultures were verified by fragment analysis. 1.6 micrograms of DNA was digested overnight with ApoI, AluI and or mix of HindIII and RsaI according to manufacturer's instructions. Fragments were separated on 0.8% agarose gels in TBE at 100 V for 1–2 hours and visualised under UV illumination. Southern transfer and hybridisation analysis was carried out as described by Sambrook et al using DIG-labelled (Roche) PCR products as probes. Probes were amplified using the primers listed in Table S7 to investigate restriction sensitivity at the respective ModA13/restriction site overlaps.

41A modA::lacZ fusion was constructed in N. meningitidis MC58. The gene fusion was initially constructed in E. coli with subsequent transformation into the N. meningitidis chromosome. In the fusion construct, the codons for LacZ are in the same translational frame as ModA resulting in an in-frame Mod-LacZ fusion protein. A 4 kb fragment of a promoterless lacZ::kan fragment was amplified by PCR using the primer pair LacZStyI1 and KanStyI. The plasmid pBluescriptlacZ::kan was used as template. Following digestion with StyI, the 4.0 kb lacZ::kan fragment was then ligated into the XbaI site of pGEMmodA. The ligation mixture was transformed into E. coli JM109 and transformants were selected on LB agar plates supplemented with kanamycin (50 µg/ml; Sigma). The orientation and sequence of the insert were checked and found to be correct. The resulting construct was named pGEMmodA::lacZ::kan. This plasmid was linearized with SacII and used to transform competent N. meningitidis. The MC58modA::lacZ::kan transformants were streaked on BHI plates containing Levinthal supplement and X-gal (40 µg/ml).

42The modA open reading frame (ORF) was amplified using PCR with primers ModAF and ModAR (see Figure S1). N. meningitidis strain MC58 was used as template. The PCR product was cloned into vector pGEM-Teasy (Promega) and named pGEMmodA. The pGEMmodA construct was digested with XbaI and blunt ended using Klenow Polymerase (New England Biolabs). The Tn903 kan resistance gene from the pUC4K vector (Pharmacia) was excised using HincII and inserted into the blunt XbaI site. Previous work has demonstrated that the pUC4Kan kanamycin cassette has no promoter or terminator that is active in Neisseria and will neither affect transcription nor have a polar effect on expression of adjacent genes. The resulting plasmid, pGEMmodA::kan was linearized by digestion with SphI and used to transform competent N. meningitidis strains MC58 and B6116/77 or N. gonorrhoeae strains FA1090, O1G1370 or 96D551. MC58 modA11::kan, B6116/77 modA12::kan, FA1090 modA13::kan O1G1370 modA13::kan and 96D551 modA12::kan transformants were selected on BHI plates containing Levinthal supplement and 100 µg/ml kanamycin. Transformants were confirmed by PCR and sequence analysis using primers ModAF2 and kanamycin specific primers. RNA midi-preps of both the wild-type (MC58 modA11 ON, B6116/77 modA12 ON, FA1090 modA13 ON and O1G1370 modA13 ON) and mutant (MC58 modA11::kan, B6116/77 modA12::kan, FA1090 modA13::kan and O1G1370 modA13::kan) were made using the RNeasy Midiprep kit (Qiagen). Wild-type colonies, from which RNA was isolated for microarray analysis, were sequenced using primers ModAF and ModAREPEATR to check that the mod repeat region was in-frame.

43Triplicate cultures of N. meningitidis strain MC58 modA11 ON and the MC58 modA11::kan mutant, N. meningitidis strain B6116/77 modA12 ON and the B6116/77 modA12::kan mutant or N. gonorrhoeae strain FA1090 modA13 ON and the FA1090 modA13::kan mutant, O1G1370 modA13 ON and the O1G1370 modA13::kan and 96D551 modA12 ON and 96D551 modA12::kan mutant were grown to exponential phase (optical density at 600 nm = 0.5 to 0.6) with 30 µM desferal (Sigma) in GC broth prior to RNA extraction. Growth rates of strain pairs used to make RNA for microarray comparison were determined (Figure S4) and were found to be equivalent ensuring that the samples taken were in the same growth phase. Only 96D551 modA12 ON and 96D551 modA12::kan (OFF) strains showed a significant difference in growth rate (see Figure S4). Culture media for RNA preps was free of antibiotics as once the modA::kan mutation is transferred to the chromosome by double crossover we observed that it is stable without selection. Approximately 100 µg of total RNA was prepared from each sample using the RNeasy Maxi Kit according to the manufacturer's instructions (Qiagen). The triplicate samples were pooled and the integrity and concentration of RNA was determined via micro-fluidic analysis on a bio-analyser (Agilent Technologies).

44All microarray analysis was performed on N. gonorrhoeae/meningitidis genome arrays (TIGR; http://pfgrc.tigr.org/). Each microarray consists of 6,389 70mer oligonucleotides representing open reading frames (ORFs) from N. gonorrhoeae strains FA1090 and ATCC 700825 (reference strain), as well as N. meningitidis strains Z2491 (serogroup A) and MC58 (serogroup B). Methods and analysis were as previously described. All primary data was imported into an in-house installation of the comprehensive microarray relational database, BASE (accessible at: http://kidney.scgap.org/base login: Nmmod, password: Nmmod, login: NmmodA12, password: NmmodA12, login: NgmodA12, password: NgmodA12 or login: Ngmod, password: Ngmod).

45Oligonucleotides (Table S7) were designed using Primer Express 1.0 software (ABI Prism; PE Biosystems) and are named according to the ORF being amplified. All real-time PCR reactions were performed in a 25 µl mixture containing 1/5 volume of cDNA preparation (5 µl), 10XSYBR Green buffer (PE Applied Biosystems) and 2 µM of each primer. We used 16S RNA as the standard control in each quantitative PCR. Amplification and detection of specific products were performed with the ABI Prism 7700 sequence-detection system (PE Applied Biosystems) with the following cycle profile: 95°C for 10 min, followed by 45 cycles of 95°C for 15 sec and 60°C for 1 min. Data was analysed with ABI prism 7700 (version 1.7) analysis software. Relative gene expression between the MC58 modA11::kan mutant and wild-type MC58 modA11 ON, N. meningitidis strain B6116/77 modA12 ON and the B6116/77modA12::kan mutant or the FA1090 modA13::kan mutant and wild-type FA1090 modA13 ON was determined using the 2ΔΔCT relative quantification method.

46Total RNA was isolated using the RNeasy kit (Qiagen). The equivalent of 1 µg of the total RNA preparation was treated with RQ1 RNase-free DNase (Promega). RT-PCR was performed using the TaqMan RT-PCR kit (PE Applied Biosystems) as recommended by the manufacturer. PCR was carried out in 50 µl using 1× Taq buffer, 1.5 mM MgCl2, and 1 unit of Taq DNA polymerase (Promega) and cDNA amplified using gene specific primers designed for Real-time PCR (Table S7) with the following cycling conditions: 30 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s and 1 cycle of 72°C for 7 min. 16S rRNA internal standards for comparison were used with amplification resulting in a 200 bp RT-PCR product. PCR products (20 µl) were run on a 3% agarose gel.

47Growth experiments were carried out in GC medium supplemented with IsoVitaleX, at 37°C with 5% CO2, under iron-limiting conditions (30 µM desferal). Triplicate cultures of the strain pairs being compared were adjusted to an identical initial OD600. One milliliter of culture was removed at fixed times to measure the OD600.

48Wild-type MC58 modA11 ON, wild-type MC58 modA11 OFF and MC58 modA11::kan mutant bacterial cells were grown under iron-limiting conditions to an optical density at 600 nm = 0.55–0.6. Cells were spun down at 5000 rpm for 5 min and then washed once in PBS, pH 7.2. Cells were then re-suspended in PBS to an optical density at 600 nm = 2.5 and separation was carried out on a 4–12% Nu-PAGE Novex Bis-Tris gel (Invitrogen) according to the manufacturer's instructions. The Nu-PAGE semi-dry system was used to transfer protein from gel to nitrocellulose membrane (0.22 µM pore, Bio-Rad), as recommended by Invitrogen. Immunoblotting of membranes was carried out in a 1∶2000 dilution of LbpA specific monoclonal 269-H1 in 5% skimmed milk powder in TBS. Bands were visualized following incubation in 1∶5000 dilution of alkaline phosphatase-conjugated anti-mouse IgG secondary antibody (Sigma).

49An lbpB::lacZ fusion was constructed in N. meningitidis strain MC58. The gene fusion was initially constructed in E. coli with subsequent transformation into the N. meningitidis chromosome. In the fusion construct, the codons for LacZ are in the same translational frame as lbpB resulting in an in-frame LbpB-LacZ fusion protein. A 1.7 kb DNA fragment was amplified by PCR using the primer pair LbpBF and LbpBR. MC58 was used as the template. The reaction was performed in 50 µl using 1× Taq buffer, 1.5 mM MgCl2, and 1 unit of Taq DNA polymerase (Promega) with the following cycling conditions: 30 cycles of 94°C for 30 sec, 57°C for 30 sec, 72°C for 1 min and 1 cycle of 72°C for 7 min. The fragment was then cloned into vector pGEM-Teasy (Promega). A 4 kb fragment of a promoterless lacZ::kan fragment was amplified by PCR using the primer pair LacZStyI+1 and KanStyI. The plasmid pBluescriptlacZ::kan (M. Dieckelman, personal communication) was used as template. Following digestion with StyI, the 4.0 kb lacZ::kan fragment was blunted using Klenow Polymerase and then inserted into the EcoRV site of the lbpB construct. The ligation mixture was transformed into E. coli JM109 and transformants were selected on LB agar plates supplemented with kanamycin (50 µg/ml). The orientation and sequence of the insert were checked and found to be correct. The resulting construct was named pGEMlbpB::lacZ::kan. This plasmid was linearized with NcoI and used to transform competent N. meningitidis strain MC58 with a naturally derived number of mod ON and OFF repeats. The MC58lbpB::lacZ::kan mod ON and MC58lbpB::lacZ::kan mod OFF transformants were streaked on BHI plates containing Levinthal supplement and X-gal (5-bromo-4-chloro-3-indolyl-D-galactopyranoside; 40 µg/ml).

50MC58lbpB::lacZ::kan modA11 OFF and MC58lbpB::lacZ::kan modA11 ON strains were grown on GC plates with 15 µM desferal at 37°C over night. The next day triplicate cultures of iron-starved strains were grown to exponential phase (optical density at 600 nm = 0.55 to 0.6) with 30 µM desferal in GC broth. Cells were spun down at 15,000×g for 10 min, resuspended in PBS and lysed by repeated freeze-thaw cycles. The cells debris was spun down at 15,000×g for 5 min. The amount of protein was calculated by using the BCA protein assay reagent kit (Pierce). The amount of β-galactosidase in the cell extracts was measured in Miller units, in triplicate, as described. Miller units were calculated as follows: Units (1000×A420)/(t×v×C), where t is the time of the assay (in mins), v is the volume of cell extract used in the assay, and C is the total protein concentration (in µg/ml).

51The antimicrobial resistance assay was adapted from a method described by Dougherty et al.. In brief, N. gonorrhoeae FA1090 wild-type modA13 ON and FA1090 modA13::kan mutant colonies were re-suspended in PBS to a density of 106 colony forming units (CFUs), and 5 µl of serial ten-fold dilutions were spotted in triplicate onto GC agar plates containing 15 µM desferal, supplemented with IsoVitaleX and increasing concentrations of Triton X-100 (40, 50, 60, and 80 µg/ml). The plates were then incubated at 37°C under 5% CO2 for 24 h. Colony counts were used to compare wild-type FA1090 modA13 ON to the FA1090 modA13::kan mutant by plating each dilution in triplicate. The experiment was repeated on three separate occasions. The ratio of FA1090 modA13 ON to FA1090 modA13 OFF at the following concentrations of Triton X-100 (40, 50 and 60 ug/ml) was calculated as follows. Colonies were taken from the triplicate samples of the original inoculum and each of the increasing concentrations of Triton X-100 from FA1090 modA13 ON and used as PCR template. The percentage of modA13 ON and modA13 OFF from the starting inoculum and the three different Triton X-100 concentrations was verified via fragment analysis using primers ModAF6Fam and ModAREPEATR (Table S7). A Student's t-test was used to determine the statistical significance between the percentage of modA13 ON and modA13 OFF from the original inoculum of modA13 ON and the percentage of modA13 ON and modA13 OFF from the three different Triton X-100 of modA13 ON.

52For examination of biofilm formation via confocal microscopy, the N. gonorrhoeae FA1090 modA13::kan and wild-type FA1090 modA13 ON strains and N. gonorrhoeae strains O1G1370 modA13::kan, modA13 OFF and modA13 ON were transformed with a plasmid encoding a green fluorescent protein, pCmGFP. Formation and analysis of biofilms was as described previously, except the cells were grown under the same iron-limiting conditions as for the microarray analysis. Colonies used to inoculate cultures for biofilm assays were assessed for morphology to ensure equivalent level of piliation. Biofilms images are three-dimensional reconstructions of stacked z-series taken at 200× magnification, which were rendered by Volocity.

53Biofilms of N. gonorrhoeae strain FA1090 modA13::kan and modA13 ON and N. gonorrhoeae strains O1G1370 modA13::kan, modA13 OFF and modA13 ON were grown at in glass flow chambers at 37°C and a flow rate of 180 µl/min in 1∶10 GC broth diluted in PBS with 10 ml/L IsoVitaleX, 3 µM desferal, and 100 µM sodium nitrite. The modA13 status of the starting inoculum was verified via fragment analysis using primers ModAF6Fam and ModAREPEATR. After 48 hours of growth, biofilms were prepared for scanning electron microscopy (SEM) and transmission electron microscopy (TEM) as follows. Glass coverslips, which served as the surface of attachment for biofilm, were removed from the chambers and fixed in 1% osmium perfluorocarbon for 1 h. The coverslips were then gently rinsed for 15 min with pure perfluorocarbon three times. To avoid destruction of the biofilm, rinse solution was gently added to coverslips in a 100 mm Petri dish, allowed to incubate at room temperature for 15 min, then the rinse was aspirated and another rinse was applied. The samples were then dehydrated with 100% ethanol by performing another three 15 min rinses. At this point, the coverslips were cut in half and one half was processed for SEM, while the other half was processed for TEM. SEM samples were transitioned into HMDS for two 15 min washes and then allowed to air dry. SEM samples were then sputter-coated and viewed with the Hitachi S-4800 SEM. TEM samples were infiltrated with a 50% Eponate-12 resin (epon) in ethanol for 1 h. The coverslips were then inverted and imbedded in 100% epon at 42°C overnight. Thin-sections (70 nm) were prepared on an ultramicrotome, mounted on a grid, and then stained with uranyl acetate and lead citrate. TEM samples were viewed with the JEOL 1230 TEM.

54Surgical cervical biopsies were used to seed primary cervical epithelial (pex) cell cultures and were procured and maintained as described previously. Quantitative association, invasion, and survival assays were performed as previously described using a multiplicity of infection of 100 with modification as follows. Our previous studies demonstrate that pex cells produce a full alternative pathway of complement, and that iC3b serves as a critical opsonin for CR3-mediated gonococcus adherence to and invasion of these cells. Thereby, antibiotic-free medium was harvested from uninfected pex cell monolayers and treated overnight with 30 µM desferal (Sigma). Our previous (unpublished) studies have revealed that N. gonorrhoeae strain FA1090 uniquely becomes cytotoxic to human, primary cervical and male urethral epithelial cells within 2 to 3 hours post-challenge, which prohibits their confident use in gentamicin survival assays for time periods totaling greater than 90 min. Therefore, N. gonorrhoeae strains O1G1370 modA13 ON, O1G1370 modA13 OFF, and the O1G1370 modA13::kan mutant, and 1291 modA13 ON, 1291 modA13 OFF, and 1291modA13::kan were selected to elucidate the role of mod-dependent phase variation during pex cell challenge. Complement-containing, iron-depleted, primed medium was inoculated with 5×106 gonococci per ml. Colonies used to inoculate cultures for these assays were assessed for morphology to ensure equivalent level of piliation. Bacterial cultures were incubated (37°C, with shaking) for 2 h, after which the optical density of the gonococcal cultures was adjusted to 107 gonococci per ml and directly used to challenge (new) pex cell monolayers. Pex cell infections were then allowed to progress at 37°C, 5% CO2. Association (90 min infection), invasion (90 min infection plus a 30 min incubation in 100 µg/ml gentamicin), and survival (90 min infection, 30 min gentamicin treatment, plus a 3 h incubation in antibiotic-free medium) assays were performed using a modified gentamicin-resistance assay as described previously.Serial dilutions of the cervical cell lysates were plated to determine CFUs. The percent association, invasion, and survival were determined as functions of the original inoculum. From these data the invasion and survival indices were determine as follows: Invasion index, percent invasion/percent association; Survival index, percent survival/percent invasion. P-values were determined for the actual data points using a Kruskal-Wallis non-parametric analysis of variance. A Student's t-test was used to determine the statistical significance of the invasion and survival indices. The ratio of modA13 ON and modA13 OFF within the O1G1370 modA13 ON and O1G1370 modA13 OFF original inoculum, association, invasion, and survival time points were determined as follows. Samples were taken from the original inoculum, association, invasion, and survival time points from three independent assays and chromosomal DNA extracted. The percentage of modA13 ON and modA13 OFF from the starting inoculum was verified via fragment analysis using primers ModAF6Fam and ModAREPEATR (Table S7). A Student's t-test was used to determine the statistical significance between the percentage of modA13 ON and modA13 OFF from the original inoculum of O1G1370 modA13 ON and the percentage of modA13 ON and modA13 OFF from the invasion time point and survival time point of O1G1370 modA13 ON. Similarly, a Student's t-test was used to determine statistical significance between the percentage of modA13 ON and modA13 OFF from the original inoculum of O1G1370 modA13 OFF and the percentage of modA13 ON and modA13 OFF from the invasion time point and survival time point of O1G1370 modA13 OFF.

55The mod gene was amplified from H. influenzae strain Rd chromosomal DNA using primers listed in Table S7. The NcoI restriction site at the 5′ end and the BamHI site at the 3′ end of the mod gene were introduced. The resulting PCR fragment was subsequently digested with NcoI and BamHI and cloned into the digested pET16b expression vector (Novagen & EMD, San Diego, CA, USA) carrying the same enzyme cutting sites, leading to the construct, pET16b::mod. The sequence of the insert was confirmed and then used for generating the recombinant Mod protein with the (His)10-tag (MGHHHHHHHHHH) attached at the N-terminal end. For generating the recombinant Mod protein, the construct, pET16b::mod, was transformed into E. coli strain BL21(DE3) and the cells were grown in LB broth at 20°C. Induction of the expression was initiated by adding IPTG to the final concentration of 0.1 mM and then incubated at 10°C for 3 days. After harvesting the bacteria by centrifuging at 6,000 rpm for 30 min at 4°C, the bacterial pellet was lysed with the lysis buffer (25 mM Tris-HCl, 300 mM KCl, 5 mM imidazole, pH 7.5) plus protease inhibitor, Complete cocktail EDTA-free (Roche, Switzerland). Soluble proteins were obtained from the supernatant by centrifuging at 20,000 rpm for 20 min at 4°C to remove the cell debris and precipitates. The Mod protein was purified using the Ni2+-nitilotriacetic acid (Ni-NTA) column (Amersham Biosciences, Piscataway, NJ, USA) with an elution gradient from 25–500 mM imidazole in the buffer solution (25 mM Tris-HCl, 300 mM KCl, pH 7.5). The purity of the eluted protein was examined by SDS–PAGE analysis and the concentration determined by Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA). The pure fractions were collected and transferred to 25 mM Tris, pH 7.5 by the HiPrep 26/10 Desalting column (Amersham Biosciences, USA) and store at −80°C.

56Rabbits (New Zealand White strain, weighing 3–3.5 kg, were immunized by intrasplenic injection with the soluble recombinant Mod protein at 300 µg per immunization. The antigen was administered together with an equal amount of Gold TiterMax adjuvant (CytRx, Norcross, GA, USA). The rabbit antisera were collected from weeks 4∼9 and the titers of rabbit sera from weeks 4∼6 were analyzed using Western blot assays. Antiserum of week 6 had a high titer of 5,000,000 against 1 µg of the Mod protein. The antisera recognized a single band in wild type modA ON H. influenzae strain RD and N. meningitidis strain MC58, but not in their corresponding modA::kan mutants (not shown). For the subsequent Western blot experiments, 1/1,000 dilution of the antiserum of week 6 was used.