Identification of Neisseria meningitidis Genetic Loci Involved in the Modulation of Phase Variation Frequencies

Abstract

It has been proposed that increased phase variation frequencies in Neisseria meningitidis augment transmissibility and invasiveness. A Himar1 mariner transposon mutant library was constructed in serogroup A N. meningitidis and screened for clones with increased phase variation frequencies. Insertions increasing the frequency of slippage events within mononucleotide repeat tracts were identified in three known phase variation-modulating genes (mutS, mutL, and uvrD), as well as six additional loci (pilP, fbpA, fbpB, NMA1233, and two intergenic regions). The implications of these insertion mutations are discussed.

The ability of mucosal microorganisms to successfully colonize and survive within the human host is dependent on how well they can adhere to the mucosal surfaces, acquire and utilize local nutrients, and evade host immune systems. The obligate human pathogen Neisseria meningitidis is one such organism that has evolved the capacity to thrive in the human nasopharynx, as it can be isolated from up to 10% of the population (5). However, the incidence of meningococcal disease represents less than 1% of carriers (32), and the factors leading to invasive disease rather than asymptomatic carriage have yet to be fully elucidated.

One of the mechanisms believed to be involved in neisserial transmissibility and virulence is that of phase variation, which has been postulated to occur in over 60 N. meningitidis genes (21, 31, 34). This regulation of gene expression typically involves slippage events in repeated nucleotide tracts (slipped-strand mispairing) during DNA replication, altering reading frames or promoter strength (27). The frequency with which meningococci change the expression state of phase-variable genes differs greatly among clinical isolates, and it has been proposed that increased frequencies may augment fitness (2, 3, 10, 20, 25). In this study, we utilized in vitro Himar1 mariner transposition to identify previously unidentified genetic loci involved in the regulation of strain-specific frequencies of phase variation within mononucleotide repeat tracts.

Bacterial strains, plasmids, and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table 1. It has been shown that N. meningitidis is capable of altering the expression state of two outer membrane receptors involved in the uptake of iron from hemoglobin (HmbR and HpuAB) and that these phase variation events occur at strain-specific frequencies via slipped-strand mispairing of poly(G) tracts within the hmbR and hpuA coding regions (6, 7, 18, 19, 23, 24, 33). Because we chose to screen mutants for augmented hmbR switching frequencies, the meningococcal strain selected for mutagenesis (IR4162) is an HpuAB mutant (hpuB::Em). Thus, we could be confident that our assays were specific to hmbR phase variation. This strain was also chosen because we have previously determined that the phase variation frequency of this highly transformable serogroup A clone falls within the “slow” category (<2 × 10⁻⁵ CFU⁻¹) (2, 24), which is advantageous when attempting to identify Himar1 insertions leading to increased phase variation frequencies. Neisseria cells were routinely cultured on either GCB agar or broth (Difco Laboratories, Detroit, Mich.) containing Kellogg's supplements I and II (14) and incubated at 37°C with 5% (vol/vol) CO₂ or brain heart infusion media supplemented with 2.5% heat-inactivated fetal bovine serum unless otherwise noted.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
Strains
IR4048	N. meningitidis serogroup A clinical isolate	Laboratory collection
IR4162	hpuB::Em derivative of IR4048	2
IR4935	E. coli BL21 (DE3)/pET29C9	22
IR5480	hpuB::SpOFF Em derivative of IR4048	2
IR5602	IR5480 fbpB::Himar1 (Kmr)	This study
IR5647	IR4162 fbpA::Sp	This study
IR5650	IR4162 fbpA::Cm	This study
IR5703	629B Himar1 insertion-containing derivative of IR5480	This study
IR5704	739C Himar1 insertion-containing derivative of IR5480	This study
IR5705	530D Himar1 insertion-containing derivative of IR5480	This study
IR5706	759C Himar1 insertion-containing derivative of IR5480	This study
IR5707	820B Himar1 insertion-containing derivative of IR5480	This study
IR5708	698C Himar1 insertion-containing derivative of IR5480	This study
N16δFBPA	N. meningitidis B16B6 fbpA::Cm	15
Plasmids
pET29C9	pET29b+ carrying hyperactive Himar1 transposase	16
pSM1	Mini-Himar1 carrying aphA3 and neisserial uptake sequence	22

Construction of a Himar1 mariner mutant library.

Purification of Himar1 transposase and in vitro transposition reactions were described previously (16, 17, 22). One-half of the completed transposition reaction mixture was incubated with approximately 10⁷ N. meningitidis IR4162 cells in GC broth plus 10 mM MgCl₂. After 30 min of incubation, 500 μl of a mixture containing GCB plus Kellogg's supplements I and II was added, and incubation continued for 2 h as described above. Clones containing a chromosomal copy of the transposon were selected overnight on brain heart infusion medium-kanamycin (150 μg ml⁻¹). We desired a mutant library with a transposon insertion approximately every 1 kb. From the equation N = ln (1 − P)/ln (1 − f), where N is the number of mutants required, P is the probability desired, and f is the frequency of insertions within the genome (1 kb/2,200 kb), a saturation mutagenesis library would require ∼6,700 to 10,200 mutants for a confidence level of 95 to 99% (26). Therefore, 500 resultant Km^r colonies were picked from each of 20 transposition reactions and stored, creating a library of ∼10,000 insertion mutants.

Identification of mutants with increased phase variation frequencies.

Significant alterations in HmbR switching frequencies can be rapidly detected with hemoglobin (Hb) utilization disk assays (Fig. 1). In these experiments, ∼10⁷ Km^r N. meningitidis cells were plated onto GC agar containing Kellogg's supplement I and the iron chelator deferoxamine mesylate (Desferal) (50 μM; Ciba Geigy, Toms River, N.J.). Filter disks (diameter, 1/4 in) were soaked with 10 μl of 5-mg/ml human Hb (Sigma Chemical Co., St. Louis, Mo.) and added to the inoculated plates, which were then incubated overnight at 37°C with 5% (vol/vol) CO₂. The HmbR phase variation frequency of the parental strain (IR4162) on solid media is 1.9 × 10⁻⁶ CFU⁻¹. Therefore, by plating 10⁷ bacteria, approximately 5 CFU of the initial inoculum will be HmbR phase “on.” Such phase “on” bacteria will form single colonies around the Hb-soaked filter disks, as will those cells that vary from phase “off” to “on” during the incubation period (Fig. 1) (23). Thus, clones with augmented phase variation frequencies will have more single colonies around the filter disk than will the parental strain (Fig. 1). These assays were performed for each of the 10,000 transposon-containing clones, and assays for clones determined to have significantly increased phase variation frequencies were repeated a second time, finally resulting in 181 positive clones.

FIG. 1.

Representative phenotypes for various hmbR phase variation frequencies. The wild-type strain grows around Hb-saturated filter disks on iron-restricted media in a few distinct colonies (IR4162). Mutants exhibiting increased phase variation phenotypes have increased numbers of colonies representing more hmbR phase “on” cells (739C and 77D).

After the initial screen for mutants with increased switching rates, results were confirmed by a quantitative assay, which involved plating serial dilutions of a single meningococcal colony on selective GC agar containing Kellogg's supplement I, 100 μg of Hb ml⁻¹, and 50 μM deferoxamine mesylate and on nonselective media. The hmbR phase “off” to “on” switching rates were then calculated by dividing the number of colonies on selective media by the number on nonselective media (29). These assays were performed in triplicate and resulted in HmbR phase variation frequencies ranging from 1.72 × 10⁻⁷ to 1.51 × 10⁻⁴ CFU⁻¹ (parental frequency: 1.9 × 10⁻⁶ CFU⁻¹), indicating that the Hb utilization disk assays yielded numerous false positives (148 of 181 clones). We continued with the remaining 33 clones, which were at least threefold increased in phase variation frequencies relative to the parental strain.

Mapping and analysis of loci involved in phase variation modulation.

Himar1 insertion sites in the 33 positive clones were mapped by a ligation-mediated PCR technique developed specifically for this Himar1 element by Pelicic et al. (22). Linkage of the increased-phase-variation phenotype to the insertion was confirmed by amplifying each unique mariner insertion by PCR and transforming these aphA3-containing products into the parental strain, IR4162. Phase variation frequencies of the resultant clones were then measured a minimum of six times each, and the statistical significance of these data was determined by the Mann-Whitney test (P < 0.05). Using this approach, we linked phenotypes of increased hmbR phase variation frequency to single Himar1 mariner insertions in nine independent genes or intergenic regions (Table 2 and Fig. 2). The presence of a single Himar1 mariner insertion in each of these nine mutants was further confirmed by Southern blot hybridization analysis (Fig. 3). As expected, three of the identified genes, mutS, mutL, and uvrD, are the known components of the neisserial mismatch repair system, which has been shown to modulate phase variation frequencies (2, 24, 25). This provided confirmation that our mutagenesis and screening methods were effective in identifying loci involved in the regulation of switching frequencies.

TABLE 2.

Himar1 mariner transposon insertion mutants with increased phase variation frequencies

Gene designationa or location	Mutant no.	Position of Himar1 insertiona	Putative gene functiona	Fold p.v. increaseb (Hb/SP)
mutS	77D	233263	mmrc	51/NDd
	347A	233707	mmr	ND
	4B	233229	mmr	ND
	454B	233427	mmr	ND
	93C	231309	mmr	ND
	465C	231459	mmr	ND
mutL	631G	1569524	mmr	59/ND
	566E	1569329	mmr	ND
uvrD	41C	22300	mmr	46/ND
	638B	21524	mmr	ND
fbpAe	629B	826638	Periplasmic iron transport	10/2.4
fbpB	739C	824939	Periplasmic iron transport	5/4.1
	943H	824660	Periplasmic iron transport	ND
pilP	530D	638156	Pilus assembly	9/1.7
NMA1233	759C	1164369	Unknown	3/2.9
Intergenic	820B	1018795		20/2.4
Intergenic	698C	812141		4/2.6

^aAccording to the NMA Z2491 sequence.

^bCalculated as mutant/parental phase variation (p.v.) frequency; P < 0.05 (Mann-Whitney).

^cmmr, mismatch repair.

^dND, final frequencies not measured.

^eUpstream intergenic region.

FIG. 2.

Increased phase variation frequencies resulting from Himar1 mariner transposon insertions. Each bar represents the median of at least six independent measurements, with error bars depicting plus or minus quartiles.

FIG. 3.

Southern blot hybridization analysis of Himar1 mariner insertion mutants using a mariner-specific DNA probe. Genomic DNA was digested with ClaI. Lane kb, 1 kb DNA ladder (Promega); lane WT, IR4162.

We next wanted to examine whether the observed increases in phase variation frequency were hmbR specific, or if they had similar effects on an unlinked, phase-variable locus. To answer this question, we utilized the “universal rates of switching” (UROS) cassette, previously described by our laboratory (2). Similar to hmbR, the UROS cassette also contains a poly(G)₈ nucleotide tract. However, in this case, the locus is unrelated to iron utilization, as this mononucleotide repeat is within an antibiotic resistance cassette conferring spectinomycin (SP) resistance when in the correct reading frame (phase “on”) (2). Thus, these switching frequencies were determined by plating on selective GC media containing Kellogg's supplements I and II and 50 μg of SP ml⁻¹ rather than on GC-Hb-deferoxamine mesylate (for hmbR phase variation) (2). As was the case with hmbR, UROS phase variation frequencies were consistently and significantly (P < 0.05) increased in the six newly identified phase variation-modulating loci (Table 2), implicating each of these loci in genome-wide regulation of meningococcal mononucleotide repeat tracts. Although we have previously shown that phase variation frequencies for the UROS cassette are typically lower than for hmbR (2), it is interesting that mutants 629B, 530D, and 820B displayed much larger increases in hmbR phase variation frequencies than in UROS phase variation frequencies relative to those for mutants 739C, 759C, and 698C. These data suggest that not all of these mutants modulate phase variation frequencies by the same mechanism, further adding to the complexity of meningococcal gene regulation.

Two of the Himar1 insertions mapped to intergenic regions. The transposon in mutant 820B inserted in a noncoding region 3′ of two divergently transcribed hypothetical protein-encoding open reading frames (ORFs), designated NMA1060 and NMA1059, and caused a 20-fold increase in hmbR phase variation (Fig. 2). Due to this location, it is likely that the insertion does not disrupt the function of either gene product. However, it may exert a polar effect on downstream genes. Three ORFs with unknown or putative functions are located just downstream of NMA1059. NMA1058 encodes a hypothetical protein, NMA1057 encodes a putative glycosyltransferase, and the NMA1056 product appears to be a PhoP-like protein. The other intergenic insertion, in mutant 698C, had a much smaller effect on hmbR phase variation frequency (fourfold), but this effect was significant (P < 0.001). Unlike that of 820B, this locus is between the 5′ ends of two ORFs, NMA0829 and NMA0830, with unknown function and may affect either or both gene product(s) as well as downstream loci.

Mutant 759C also provides little insight into the mechanism by which the gene NMA1233 modulates phase variation. NMA1233 is homologous to part of the N. gonorrhoeae putative sucAB-lpd operon designated orfA but has no known function. Additionally, there do not appear to be any nearby ORFs that this insertion would affect.

Perhaps the most interesting transposon insertions were within the ferric binding protein operon (fbpABC). This locus encodes an ABC transporter that is reported to be responsible for the transport of ferric iron across the periplasmic membrane (1). One Himar1 element (that of 629B) was located just upstream of the first gene in the operon, fbpA, which encodes the iron binding protein (1, 9). The other two insertions (those of 739C and 943H) were within the proposed cytoplasmic membrane protein-encoding locus, fbpB (1). It is not surprising that we did not isolate a mutant element within the putative ATPase-encoding gene, fbpC, as it has been reported that this protein is not required for iron acquisition in Neisseria gonorrhoeae (1, 28). In addition to its role in iron assimilation, the fbp operon, like hmbR and numerous other genes, is regulated by iron concentration through the ferric uptake regulator protein (Fur) (4, 13, 29, 35).

The involvement of Fbp in transporting ferric iron across the periplasmic membrane led us to hypothesize that inactivation of the fbpABC operon could limit iron availability in the cytosol, thereby relieving Fur repression of iron-regulated genes. One or more of these derepressed genes, in turn, could impact phase variation. It has been reported, however, that, while an fbpABC mutant cannot utilize nonheme iron, it grows normally in the presence of heme or Hb (15). Thus, iron availability should not be affected in such a mutant when grown in the presence of Hb, as is the case with the hmbR phase variation assays (23). Conversely, an earlier study suggested that Fbp does in fact transport iron from heme and Hb (11). Due to these conflicting reports, we determined which iron sources two additional fbpA insertion mutants could assimilate. We constructed the first, IR5647, by disrupting the fbpA locus of IR4162 with a SP adenyltransferase AAD (9) cassette (fbpA::Sp) (36). The second consisted of IR4162 containing fbpA insertionally inactivated by a cat cassette (fbpA::Con) (15). The latter construct was reported by Khun et al. to render Neisseria incapable of utilizing nonheme iron (15) (IR5650). Both of these Fbp mutants displayed increased hmbR phase variation phenotypes similar to that produced by the Himar1 insertion that was just upstream of fbpA (629B) (Fig. 4). These clones were tested for their ability to utilize ferric nitrate, transferrin, heme, and Hb with filter disk assays on solid media (37) and growth assays in liquid culture (15). Under all conditions tested, the fbpA mutants, as well as the wild type, grew normally, indicating that they could use any of these iron sources (data not shown), further adding to the conflicting reports about the role of the Fbp in neisserial iron assimilation. At this point, we cannot rule out the possibility that another, functionally redundant ferric iron transport system(s) exists in Neisseria spp. It may be that the inactivation of Fbp restricts iron to such as small extent that another system is capable of relieving any potential growth defects. There may, however, be sufficient iron restriction to affect gene expression, leading to alterations in phase variation frequencies. It is interesting that this certainly is not the first report of iron availability impacting antigenic variation. Not only is hmbR both phase variable and iron regulated, but also iron availability has also been linked to DNA recombination (pilin antigenic variation), transformation efficiency, and DNA repair in the gonococcus (30).

FIG. 4.

Increased phase variation frequencies of Fbp mutants. Each bar represents the median of at least six independent measurements, with error bars depicting plus or minus quartiles.

Mutant 530D contains an insertion in pilP, a member of a cluster of pilus biogenesis genes (pilM-pilQ) that appears to be cotranscribed with the upstream gene, pilO, in N. gonorrhoeae (12). The downstream gene, pilQ, however, appears to have its own transcriptional start site, although mutations in pilO or pilP reduce levels of PilQ monomers in the cell (12). It has also been reported that pilP plays roles in both PilQ multimer formation and PilC localization (12). The gene order of this cluster is conserved among all published neisserial sequences, and, if the transcriptional arrangement is similarly conserved, then it is somewhat surprising that no transposons mapped to pilO, and we cannot rule out the possibility that we missed such an insertion in our screen. Interestingly, Chen et al. have recently identified a point mutation in N. gonorrhoeae pilQ (pilQ1) that suppresses the Hb utilization deficiency phenotype of a gonococcal HpuA mutant (8). It is possible that the Himar1 insertion in pilP results in a similar alteration of the downstream pilQ locus, further linking iron utilization to phase variation modulation. Additional experiments, however, are required to prove such a relationship.

Screening this Himar1 mariner mutant library for alterations in phase variation frequencies yielded some interesting findings. We have identified six genetic loci that modulate N. meningitidis phase variation frequencies. Three were in regions of unknown function, including two intergenic sequences and one putative ORF. Additionally, we linked phase variation frequency regulation to type IV pili (pilP) as well as iron transport and utilization (fbpA and -B), which may not be mutually exclusive (8). Determining the mechanisms by which the genes identified in this report modulate phase variation will lead to a better understanding of how pathogenic Neisseria regulates gene expression to enhance transmissibility and invasiveness.