Contribution of (cid:2) 70 and (cid:2) N Factors to Expression of Class II pilE in Neisseria meningitidis

Neisseria meningitidis expresses multicomponent organelles called type four pili (Tfp), which are key virulence factors required for attachment to human cells during carriage and disease. Pilin (PilE) is the main component of Tfp, and N. meningitidis isolates either have a class I pilE locus and express pilins that undergo antigenic variation or have a class II pilE locus and express invariant pilins. The transcriptional regulation of class I pilE has been studied in both N. meningitidis and Neisseria gonorrhoeae , while the control of expression of class II pilE has been eluci-dated in the nonpathogenic species Neisseria elongata . However, the factors that govern the regulation of the class II pilE gene in N. meningitidis are not known. In this work, we have bioinformatically and experimentally identiﬁed the class II pilE promoter. We conﬁrmed the presence of conserved (cid:2) 70 and (cid:2) N -dependent promoters upstream of pilE in a collection of meningococcal genomes and demonstrated that class II pilE expression initiates from the (cid:2) 70 family-dependent promoter. By deletion or overexpression of sigma factors, we showed that (cid:2) N , (cid:2) H , and (cid:2) E do not affect class II pilin expression. These ﬁndings are consistent with a role of the housekeeping (cid:2) D in expression of this important component of Tfp. Taken together, our data indicate that the (cid:2) -dependent network responsible for the expression of class II pilE has been selected to maintain pilE expression, consistent with the essential roles of Tfp in colonization and pathogenesis. as adhesion, transformation, and disease pathology. Meningococci express one of two distinct classes of Tfp (class I or class II), which can be distinguished antigenically or by the major subunit ( pilE ) locus and its genetic con-text. The factors that govern transcription of the class II pilE gene are not known, even though it is present in isolates that cause epidemic disease. Here we show that the transcription of class II pilE is maintained throughout growth and under different stress conditions and is driven by a (cid:2) 70 -dependent promoter. This is distinct from Tfp regulation in nonpathogenic Neisseria spp. and may confer an advantage during host-cell interaction and infection. using ML37/ML34. A fragment corresponding to the region 3 = of the pilE ORF with a kanamycin resistance marker inserted 81 bp downstream of the pilE stop codon was ampliﬁed from strain S4_ pilE kanR (R. Exley, unpublished data) using ML30/ML31. The two fragments were cloned into pCR2.1TOPO ﬂanking the lacZ gene, which was ampliﬁed from pRS415 (69) using primers ML32/ML33. The resulting plasmid, pCR2.1TOPO (cid:3) P pilE - lacZ , was used as a template to amplify DNA for transformation of wild-type S4 or S4Δ rpoN . N. meningitidis strains overexpressing (cid:2) factors were constructed by inserting the relevant ORF at the iga-trpB intergenic locus (41) of N. meningitidis S4, under the control of an IPTG inducible promoter. For this purpose, pNMC2 was generated. First, a fragment comprising part of the iga gene and terminator was ampliﬁed from S4 gDNA using primers ML151 and ML149. Next, a fragment comprising ermC , the lac regulatory region, and a multiple-cloning site (MCS) was ampliﬁed from pGCC4 (a gift from Hank Seifert) (Addgene plasmid 37058; http://n2t.net/addgene:37058; RRID:Addgene_37058) using primers ML148/ML150. The two fragments were fused by overlap PCR using primers ML148 and ML151. The resulting PCR product was then combined using Gibson assembly with a fragment ampliﬁed from

Meningococci express type four pili (Tfp), which play key roles during meningococcal carriage and disease. During colonization, Tfp meditate the initial adherence of N. meningitidis to epithelial cells (5). Progression to systemic disease and dissemination from the nasopharynx are proposed to be triggered by the detachment of a small number of bacteria from microcolonies on the epithelial surface following changes in Tfp expression (6). Subsequently Tfp mediate formation of microcolonies on endothelial cells, providing resistance against shear stress in the circulation (7), as well as reorganization of host cell components, leading to translocation of N. meningitidis into the cerebrospinal fluid (8). In addition, Tfp are required for twitching motility and competence for DNA uptake that allows horizontal gene transfer between bacteria (9,10).
The major component of Tfp is the pilin protein PilE. N. meningitidis isolates express one of two distinct classes of the type four pilin subunit: either a class I pilin, which undergoes high-frequency antigenic variation, or an invariant class II pilin (11). Based on genome analysis, meningococcal isolates have either a class I or a class II pilE locus but not both, and a number of features distinguish these loci (12,13). For example, class I pilE loci are flanked by sequences that enable intrastrain pilin variation, namely, a G-quadruplex-forming (G4) sequence upstream of the pilE open reading frame (ORF) and a downstream Sma/Cla sequence. These features facilitate nonreciprocal homologous recombination between pilE and silent pilS cassettes located immediately downstream of pilE (14)(15)(16)(17). Isolates that express invariant pilins harbor a class II pilE locus. This is situated at a different chromosomal site, and therefore, although pilS cassettes are present, they are not adjacent to pilE. Furthermore, the G4 and Sma/Cla elements are lacking from the regions adjacent to the class II pilE locus (12,13,18). Of note, strains with class II pilE mostly belong to hyperinvasive lineages that are responsible for epidemic disease in sub-Saharan Africa and for worldwide outbreaks (13,19,20).
A number of studies have investigated the transcriptional regulation of pilE in pathogenic and nonpathogenic Neisseria spp. (21)(22)(23)(24). These have shown that the number, arrangement, and activity of pilE promoters vary between species and strains. For example, the class I pilE promoter in N. meningitidis strain MC58 includes both Ϫ10/ Ϫ35 and Ϫ12/Ϫ24 sequences, but the Ϫ12/Ϫ24 sequence does not play a role in pilin transcription (21). The same is true in the related species Neisseria gonorrhoeae, in which the pilE promoter is similar to the meningococcal class I pilE promoter (25). In contrast, the pilE promoter of the nonpathogenic species Neisseria elongata lacks Ϫ10/Ϫ35 sequences, and transcription initiates from a Ϫ12/Ϫ24 sequence (23).
All bacterial promoters are recognized by sigma () factors. These can be divided into two groups: (i) the 70 family, which are structurally related factors that recognize Ϫ10/Ϫ35 sequences, and (ii) N , which recognizes a consensus Ϫ12/Ϫ24 sequence and has a noncanonical mechanism of transcription initiation, requiring interaction with an AAAϩ ATPase activator protein (26,27). The meningococcal genome contains four genes encoding potential factors: three 70 factors ( D , E , and H ) and N . D is considered the housekeeping factor, E is autoregulated by an anti-sigma factor and regulates 11 genes (28), and H is essential but as yet has no defined role in the meningococcus, although it has been linked with the response to heat shock in gonococci (29,30). N has been reported to be nonfunctional based on analysis of isolates in which the rpoN gene is similar to the rpoN-like sequence (RLS) of N. gonorrhoeae (25). The gonococcal RLS has a frameshift mutation that results in loss of the domains required for DNA binding, namely, the helix-turn-helix (HTH) motif and the RpoN box which are required to bind to the Ϫ12 and Ϫ24 sequences, respectively (23,25). More recently, genome sequence analysis has demonstrated that different meningococcal isolates encode distinct forms of N , some of which lack both of the key DNA binding domains and others of which lack only the HTH motif (23). Interestingly, in N. elongata rpoN encodes a protein with both DNA binding motifs and which activates pilE transcription (23).
The factors that govern transcription of meningococcal class II pilE have not been studied previously. In this work, we characterized the pilE promoter in an N. meningitidis isolate expressing class II pilin (31). Through bioinformatic analysis of whole-genome (32). The sequence of the second N -dependent promoter was conserved and exhibited 100% nucleotide sequence identity to the Ϫ12/Ϫ24 consensus in all but one of the 77 isolates (Fig. 1C). This sequence conservation and lack of overlap with the 70 binding site raised the possibility that the Ϫ12/Ϫ24 sequence is implicated in class II pilE regulation. Since the Ϫ12/Ϫ24 sequence requires N for transcription initiation, we next examined the presence and sequence of rpoN, which encodes N .
Meningococcal isolates contain atypical rpoN sequences. The rpoN gene in Escherichia coli (1,434 bp) is present in a single copy and encodes a protein of 477 amino acids (aa) with three domains. Region I is the activator binding domain, region II is a variable, nonessential region involved in DNA melting, and region III interacts with RNA polymerase (RNAP) and is implicated in promoter recognition and binding (26) (Fig. 2A). Region III contains two DNA binding/recognition motifs: an HTH motif (36) and the RpoN box (ARRTVAKYRE) (27). In N. gonorrhoeae the RLS is predicted to encode a protein of 277 aa which lacks the HTH motif and RpoN box; hence, any protein product is likely to be inactive (23,25). In N. meningitidis 8013, the rpoN ORF encodes a potential 289-aa protein also without any recognizable HTH motif and which lacks the RpoN box due to a frameshift mutation (25,37). However, more recently several N. meningitidis isolates were reported to harbor an rpoN sequence that encodes a protein which retains the RpoN box at the C terminus (23). We therefore investigated the presence and sequence conservation of rpoN in the collection of 290 N. meningitidis isolates.
The rpoN sequence from FAM18 (NEIS0212 allele 1 in PubMLST) was used as the query to extract sequences using BLASTn in PubMLST. We identified rpoN homologues in all 290 genomes. In total, there were 44 different alleles among the 290 isolates (see Table S1 and Fig. S1 in the supplemental material). The rpoN sequences were translated using EMBOSS Transeq, and predicted amino acid sequences were aligned using Clustal Omega. The 44 alleles encoded a total of 34 different peptides which range between 120 and 289 aa in length. Analysis of the 34 deduced protein sequences revealed three distinct types of N (Fig. 2B). In 17/290 (6%) isolates, the putative N protein is 289 aa and, similarly to the allele in N. meningitidis 8013, harbors a frameshift mutation and so lacks an RpoN box. In 4/290 (1%) of the isolates, the rpoN allele contains stop codons that result in truncated proteins. In the remaining 269 of 290 isolates (93%), the putative N protein was as described by Rendon et al. (23), i.e., a protein which lacks a recognizable HTH domain but has an RpoN box in the C terminus. Analysis of the distribution of each of the three types of N (frameshift, truncation, or deletion) in the isolates according to pilE locus demonstrated that 77/77 (100%) isolates with a class II pilE locus and 192/213 (90%) isolates with a class I pilE locus harbor an rpoN allele that encodes N with an RpoN box but lacking the putative HTH region (Fig. 2C). Given the presence of a conserved RpoN box, we hypothesized that this atypical form of N could retain binding to the Ϫ24 element and contribute to class II pilE transcription from the N -dependent promoter. N -dependent promoters are usually preceded by one or more upstream activator sequences (UAS) situated 80 to 150 bp upstream of the Ϫ12/Ϫ24 sequence (38,39). These sequences bind bacterial enhancer binding proteins (bEBPs), which are activated in response to certain stimuli and trigger a conformational change within the holoenzyme via ATP hydrolysis, resulting in the transition between closed and open complex formation and transcription initiation (38,40). Based on homology to sequences in Pseudomonas aeruginosa, a putative UAS has been identified in the promoter region of class I pilE in N. gonorrhoeae and N. meningitidis (21,22). We used the same UAS from P. aeruginosa (TGTGACACTTTTTGACA) to search for homologous sequences in the promoter regions of the 77 isolates with class II pilE. Sequences with 9/17 (TcTGACAa aaaacGtCA) or 10/17 (TcTGACAaaaaaTGtCA) nucleotide matches (unmatched nucleotides are lowercase) were identified approximately 112 bp upstream of the Ϫ12/Ϫ24 sequence in 50 and 24 of the genomes, respectively, but none of the isolates had an exact match to either the P. aeruginosa sequence or the putative UAS from N. meningitidis or N. gonorrhoeae (not shown). rpoN is expressed in N. meningitidis, but no protein is detected and it is not required for pilin transcription. Next, we examined whether N can be detected in class II pilin-expressing N. meningitidis by Western blotting. Using cc-11 strain S4, which expresses class II pilin, a mutant was constructed in which the endogenous copy of rpoN was replaced with the rpoN ORF in frame with a sequence encoding a C-terminal histidine tag. The presence of N in whole-cell lysates from bacteria grown overnight on solid medium was then assessed by Western blotting using an anti-His antibody. No protein of the predicted size (34 kDa) was present in extracts from S4rpoN_His, indicating that, under the conditions tested, N could not be detected in N. meningitidis (data not shown).
We therefore determined whether rpoN mRNA was detectable in N. meningitidis S4. Using quantitative reverse transcription-PCR (qRT-PCR) we measured rpoN transcript levels during growth of wild-type (WT) S4 in liquid medium at 37°C. RNA from strain S4 ΔrpoN, in which the entire rpoN coding sequence was replaced with an antibiotic resistance cassette, was used as a negative control for nonspecific amplification. Transcript levels were measured relative to those of transfer-messenger RNA (tmRNA). As shown in Fig. 3A, growth of wild-type S4 was similar to that of S4 ΔrpoN, indicating that the absence of rpoN does not impact bacterial fitness. rpoN transcript was detected throughout growth, albeit at low levels relative to those of the control tmRNA (Fig. 3B). rpoN transcript varied according to the growth phase, with highest levels detected during stationary phase (22 h).
To determine whether rpoN is required for pilE transcription, we next constructed reporter strains in which the class II pilE ORF was replaced by lacZ at the endogenous pilE locus in strains S4 and S4ΔrpoN (Fig. 3C). The resulting strains, S4P pilE -lacZ and S4ΔrpoNP pilE -lacZ, were plated onto solid medium containing 5-bromo-4-chloro-3indolyl-␤-D-galactoside (X-Gal), and the detection of blue colonies was used to qualitatively assess whether N is required for pilE expression. After overnight growth, blue colonies were detected for both S4P pilE -lacZ and S4ΔrpoNP pilE -lacZ, indicating that rpoN is not required for pilE expression. Finally we analyzed pilE transcript levels in wild-type S4 and in S4ΔrpoN. pilE mRNA was measured by qRT-PCR and expression levels normalized to those of tmRNA. As shown in Fig. 3D, pilE was expressed at similar levels throughout growth in the wild-type S4 and S4ΔrpoN, demonstrating that N is not required for class II pilin expression in S4 under the conditions tested.
Induced expression of N does not impact pilin protein levels in N. meningitidis. We considered the possibility that the absence of detectable N in wild-type S4 could account for the observed lack of any effect of the deletion of rpoN on class II pilE transcript levels. Therefore, we examined the impact of inducing rpoN expression on pilin levels. We constructed N. meningitidis S4 harboring rpoN with an in-frame FLAG tag at an ectopic locus (41) under the control of an isopropyl-␤-D-thiogalactopyranoside (IPTG)-inducible promoter (S4P lac -rpoN Nm ). In parallel, we generated a strain (S4P lac -rpoN Nel ) with IPTG-inducible N. elongata rpoN, which activates pilE transcription in N. elongata by binding the Ϫ12/Ϫ24 sequence and requires the activator protein Npa (23), and a control strain, S4 ery, which lacks the N -coding sequence at the ectopic locus. Pilin levels in S4P lac -rpoN Nm , S4P lac -rpoN Nel , and S4 ery were analyzed by Western blotting of cell lysates collected after overnight growth on solid medium with or without 1 mM IPTG to induce rpoN expression; the expression of N was confirmed with anti-FLAG antibody. There was no detectable difference in pilin levels in S4P lac -rpoN Nm in the presence or absence of inducer (Fig. 4A), demonstrating that expression of rpoN does not affect class II pilin levels. We therefore conclude that meningococcal N is unlikely to bind to the Ϫ12/Ϫ24 sequences in the pilE promoter to activate transcription of class II pilE.
Interestingly, the expression of N. elongata rpoN in N. meningitidis led to a substantial reduction in pilin expression (Fig. 4B), and this was also observed during growth in liquid medium (Fig. 4C). One possible explanation is that N. elongata N can engage with the Ϫ12/Ϫ24 promoter upstream of class II pilE but, in the absence of additional N. elongata specific factors, transcription does not initiate, and the bound sigma factor instead blocks transcription from adjacent promoters. This phenomenon has been described in E. coli, in which an excess of N can reduce RNA synthesis by 70 RNAP when the N binding site overlaps the region normally occupied by 70 RNAP (42). Furthermore, our finding is consistent with analysis of gonococcal pilE promoters in E. coli, which has demonstrated that N can reduce levels of transcription of PpilE-lacZ or PpilE-cat fusions, likely by steric hindrance at the overlapping 70 -dependent promoter (43,44). Analysis of the relative positions of the -10/Ϫ35 promoter and the N binding site upstream of class II pilE suggests that they are sufficiently close for competition to occur (42). Taken together, our data indicate that N in N. meningitidis is nonfunctional or does not function as a canonical N and that class II pilE transcription does not initiate from the Ϫ12/Ϫ24 promoter.
Class II pilE transcription initiates from the 70 -dependent promoter. Next, we mapped the class II pilE transcriptional start site (TSS). Total RNA was prepared from N. meningitidis S4 grown in liquid brain heart infusion (BHI) at 37°C to mid-exponential phase, and primer extension was performed using primers designed to capture transcripts generated from both the Ϫ10/Ϫ35 and the Ϫ12/Ϫ24 promoters (primer P2) or from only the Ϫ10/Ϫ35 promoter (primer P1) (Fig. 5A). Extension conditions were such that any TSS within nucleotide (nt) Ϫ70 to Ϫ250 of the pilE start codon would be detected. Totals of 1 g and 2.5 g of RNA were used to generate cDNA with radiolabeled P1 and P2. Extension with primer P1 generated a product of ϳ35 nt, corresponding to a transcript initiated from the Ϫ10/Ϫ35 sequence. Similarly, an ϳ74-nt product was generated with P2, which corresponds to the same TSS. Interestingly, an ϳ68-nt minor product was also generated with P2, which could be an alternative TSS or a degradation product. Importantly, using P2 there was no detectable product of ϳ38 nt that would indicate transcription from the Ϫ12/Ϫ24 promoter, supporting our hypothesis that class II pilE transcription initiates from the 70 -dependent promoter. Sequencing with primer P2 using a PCR fragment of the upstream region of class II pilE as the template identified a cytosine residue located 8 nt downstream of the Ϫ10 sequence as the TSS of class II pilE (Fig. 5B).
Class II pilE transcript initiation is not affected by stress conditions. factors are often associated with responses to a particular stress, and the level of a given factor increases rapidly following an appropriate stimulus (45). N was originally identified in the response to nitrogen starvation but has since been implicated in responses to diverse stimuli mediated by specific bEBPs (26). Therefore, to determine whether pilE promoter activity changes when meningococci are subjected to different stress conditions, we performed primer extension on RNA from S4 exposed to salt (5 M NaCl), acid (HCl, pH 2.5), oxidative (0.15% H 2 O 2 ), or envelope (5% ethanol or 5% Triton) stress for 10 min. N. meningitidis S4 grown to the same optical density at 600 nm (OD 600 ) but not exposed to any stress was used as a control. As previously, we detected products corresponding to transcript initiation from the Ϫ10/Ϫ35 promoter under all conditions as well as the ϳ68-nt product and a smaller ϳ47-nt product with the primer P2 from bacteria exposed to Triton. However, none of the stress conditions led to the appearance of an ϳ38-nt product with P2, indicating that the Ϫ10/Ϫ35 promoter is responsible for class II pilE transcription even under different environmental conditions.
Modulating levels of E and H does not impact class II pilin expression. In bacteria, Ϫ10/Ϫ35 promoter sequences are able to recruit different factors belonging to the 70 family (46,47). For instance, the Ϫ10/Ϫ35 sequence upstream of rpoD in E. coli can bind D , E , and S (48). Based on sequence similarity to orthologues in E. coli, three 70 factors ( D , H , and E ) have been identified in N. meningitidis. D (encoded by NEIS1466) is considered the "housekeeping" factor. In most bacteria this factor is essential and is responsible for the transcription of the majority of genes and essential pathways throughout growth (49). Based on the analysis of essential genes in N. meningitidis 8013 and N. gonorrhoeae FA1090, rpoD is also essential in pathogenic Neisseria species (50,51). The regulon of H (encoded by NEIS0663) and its role in heat shock adaptation and regulation of genes involved in cell adhesion in N. gonorrhoeae has been characterized (30,52), but the H regulon in N. meningitidis has not been described. Finally, E (NEIS2123) regulates 11 genes in N. meningitidis (28). The regulon does not include pilE, although the analysis was performed in a class I pilE-expressing strain. We hypothesized that if the pilE Ϫ10/Ϫ35 promoter binds different 70 members, changing the levels of the sigma factors relative to the level of D may impact pilin and Ϫ12/Ϫ24 promoter elements (boxed). The position where primers P1 and P2 bind is indicated. (B) Primer extension was performed using 1 g or 2.5 g of RNA prepared from N. meningitidis S4 grown at 37°C in liquid BHI to mid-exponential phase. Products of approximately 35 nt (with primer P1) and 74 nt (with primer P2) were detected, corresponding to transcripts initiating from the Ϫ10/Ϫ35 promoter. An additional product of ϳ68 nt was observed with P2. No product corresponding to transcript initiating from the Ϫ12/Ϫ24 sequence (ϳ38 nt) was detected. Lanes A, C, G, and T show sequences determined using a PCR product and primer P2. The transcription start sites (TSS) are indicated by arrows and highlighted by asterisks in the promoter sequence. (C) Primer extension of class II pilE transcript in bacteria subjected to different stresses as indicated. No difference in class II pilE TSS was observed. An additional extension product (ϳ47 nt) was detected with the P2 primer in Triton-treated samples. The experiment was performed twice using independent biological replicates. expression. Therefore, we constructed N. meningitidis mutants lacking or overexpressing H or E and examined pilin levels in these backgrounds.
Consistent with H being essential in N. gonorrhoeae and N. meningitidis (30,51), our attempts to construct S4ΔrpoH were unsuccessful. Therefore, we generated strain S4P lac -rpoH, in which H expression is inducible by IPTG and detectable due to the presence of a C-terminal FLAG tag. S4P lac -rpoH was grown overnight on solid medium with or without IPTG, and pilin levels were analyzed by Western blotting. There was no difference in pilin expression in the presence or absence of IPTG (Fig. 6).
As meningococcal E is not essential (28), we generated S4ΔrpoE by replacing the rpoE ORF with a kanamycin resistance cassette, as well as strain S4P lac -rpoE overexpressing FLAG-tagged E . Western blot analysis revealed that pilin levels remained unaltered in the absence or induced expression of E (Fig. 6). This is consistent with pilE not being part of the E regulon and indicates that E is not essential for pilin production. Since expression and/or deletion of rpoH and rpoE does not affect class II pilin levels, we conclude that class II pilE transcription is initiated from the Ϫ10/Ϫ35 promoter by D .

DISCUSSION
Among the factors that allow bacteria to colonize and infect the human host, Tfp play a key role by meditating adhesion, signaling, and interbacterial attachment (7,8,53). Alterations to Tfp, through either changes in sequence, posttranslational modification, or changes in expression, can have important consequences for bacterial behavior. For example, the absence of Tfp significantly reduces the ability of bacteria to attach to host cells (54)(55)(56), while modification of the pilin sequence, glycosylation, or phospho-modification can impact bacterial adhesion and aggregation and may contribute to dissemination within or between hosts (5,57,58). Therefore, understanding the mechanisms by which Tfp expression and function are modulated is important for understanding pathogenesis and may provide insights into novel ways to overcome meningococcal immune evasion strategies.
Two distinct classes of pilin have been identified in pathogenic Neisseria (11,12). Class I pilins can be altered at high frequency through gene conversion or phase variation (16,(59)(60)(61)(62), while class II pilins are invariant. As described in previous studies, the meningococcal class II pilE locus is distinguishable from the class I pilE locus based on the encoded pilin sequence, the flanking genes, and the presence of sequences required for antigenic variation (13). Our current work confirms that class I and class II pilE loci also have different promoter sequences, consistent with previous observations Western blot analysis of pilin levels in whole-cell lysates collected from bacteria lacking E (S4 ΔrpoE) or with inducible expression of H or E (S4 P lac -rpoH and S4 P lac -rpoE), or from a control strain (S4 ery), on solid medium at 37°C is shown. Expression of IPTG-induced E and H was detected using anti-FLAG antibody. No change in pilin levels was observed when factors were deleted or induced. RecA was used as a control, and the blot is a representative from two independent experiments. Numbers to the leftls indicate the positions of molecular weight markers (kilodaltons). (12). Mirroring the very low nucleotide diversity of class II pilin-coding sequences (13,19,20), the class II pilE promoter regions were highly conserved and, independent of clonal complex, harbored potential 70 and N binding sequences. These findings differ from those of Rendon et al. (23), who defined three types of pilE promoter, including one group (represented by FAM18) with N but not 70 recognition sequences; instead, we identified putative 70 -dependent promoters in all genomes, including FAM18.
Previous studies of pilE regulation in pathogenic Neisseria spp. showed that transcription of class I pilE relies on the 70 -dependent rather than the N -dependent promoter (21,22,24). The presence and position of highly conserved, consensus Ϫ12/Ϫ24 sequences upstream of class II pilE, as well as an rpoN gene encoding a sigma factor with a conserved Ϫ24 binding region, led us to investigate the role of the N -dependent promoter in class II pilE transcription. Based on structure-function studies in other organisms, we considered the possibility that this atypical form of N still activates transcription despite lacking the Ϫ12 binding HTH domain. For example, mutants which do not effectively recognize the Ϫ12 sequence can still produce RNA in vitro in an activator-independent manner known as N bypass transcription (63). Consistent with this, no GAFTGA-containing bEBPs have been identified in N. meningitidis (64), and Npa, the activator for N -dependent pilE transcription in N. elongata, is truncated in N. meningitidis (23,65). Furthermore, in E. coli N bypass transcription can occur at Ϫ12/Ϫ24 promoters with a noncanonical Ϫ12 element (42). However, our findings indicate that the Ϫ12/Ϫ24 promoter and N do not contribute to class II pilE expression; we were unable to detect any N protein expression, and neither deletion nor overexpression of rpoN had any significant effect on class II pilE transcript or protein levels. Furthermore, mapping the TSS of class II pilE demonstrated that the 70dependent promoter is responsible for class II pilE expression, even under a variety of stress conditions, some of which are known to induce N -dependent transcription in E. coli (66). Collectively, our results demonstrate that class II pilE transcription is independent of N and initiates from the Ϫ10/Ϫ35 promoter.
An important observation that emerged from our analysis is that in a variety of N. meningitidis mutants lacking sigma factors, as well in the presence of stress signals, pilE transcript and/or protein expression remained relatively stable. Considering the importance of Tfp for bacterium-bacterium interactions, competence, and adhesion to host cells, it is likely that class II pilE is constitutively expressed to ensure that the capacity to synthesize pilin is maintained. Consistent with this, changing levels of E , H , or N had no detectable impact on pilin levels. Although we cannot rule out that H or E may contribute to the expression of class II pilE under specific conditions, we found no evidence of N expression or activity.
Of note, we found alleles which encode the same N mutations or deletions in strains with either a class I or class II pilE locus, suggesting that the loss of functional rpoN occurred independently of the evolution of the distinct pilE loci. Interestingly, the commensal species Neisseria polysaccharea and Neisseria lactamica have class II pilE loci that are very closely related to the meningococcal class II pilE locus, and in our previous work we proposed that class II pilin-expressing meningococci may have either evolved from a common ancestor or acquired the class II pilE gene from these species (13). However, the rpoN gene in these nonpathogenic species does not harbor mutations and deletions and potentially encodes a functional N (23). Therefore, our findings suggest that loss of functional N is beneficial for the pathogenic Neisseria spp. and are consistent with the idea that a switch in pilE regulation from N -dependent to Ddependent control occurred during evolution and divergence of pathogens from commensals (23,25). As N -dependent transcription is usually tightly regulated by bEBPs which respond only to specific stimuli, loss of N -dependent transcription of pilE might therefore be an evolutionary pathway to achieve constitutive D -dependent expression from the Ϫ10/Ϫ35 promoter. In this way, transcription of the gene encoding the major component of Tfp is maintained, consistent with the significant contribution of this important organelle at multiple stages of colonization and pathogenesis.

MATERIALS AND METHODS
Bacterial strains and growth conditions. The bacterial strains used in this study are listed in Table  S2 in the supplemental material. E. coli was grown at 37°C, supplemented with antibiotics where appropriate, either on Luria-Bertani (LB) agar or in 5 ml LB liquid with shaking at 180 rpm. N. meningitidis was grown at the desired temperature in the presence of 5% CO 2 on brain heart infusion (BHI) (Oxoid) agar supplemented with 0.1% (wt/vol) starch, 5% (wt/vol) heat-denatured horse blood, and antibiotics as appropriate or in BHI liquid. To inoculate liquid cultures, N. meningitidis was grown overnight on BHI agar. A loop of bacteria was harvested from plates and resuspended in 600 l of phosphate-buffered saline (PBS), and DNA was quantified in lysis buffer (P2 buffer; Qiagen) by measuring the OD 260 . A total of 10 9 CFU was used to inoculate 25 ml of BHI in a 125-ml conical flask (Corning) and then incubated at appropriate temperatures with shaking at 180 rpm. The OD 600 of cultures was used to monitor growth. Antibiotics were used at the following concentrations: kanamycin, 50 g/ml and 75 g/ml for E. coli and N. meningitidis, respectively; erythromycin, 2 g/ml; and carbenicillin, 100 g/ml.
Bioinformatic analysis. The 290 isolates whose genomes were used in this study comprise 288 isolates described in our previous work (13) and two additional isolates: S4 and FAM18 (class II pilE). Among the 288 isolates, 56 belong to the collection of 107 meningococcal isolates originally used for multilocus sequence type (MLST) development in Neisseria (3). The remaining 232 are disease-associated strains isolated during the 2010 -2011 epidemiological year from patients in England, Wales, and Northern Ireland and whose genomes are publicly available through the Meningitis Research Foundation genome library (MRF-MGL) on PubMLST (https://pubmlst.org/neisseria/) (67). These isolates were chosen for pilE promoter and rpoN analysis based on our previous work where we identified them as having either a class I or a class II pilE locus (13). BLASTn analysis of genomes in the PubMLST Neisseria database was performed to identify pilE promoter and rpoN sequences. For pilE promoter analysis, up to 500 nt of sequence flanking the coding DNA sequence (CDS) was analyzed, and annotation was performed manually based on published pilE promoters (12,21,32) and bacterial consensus promoter sequences (33). Nucleotide sequences were aligned using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/ clustalo/) (68). Percent identity was determined using the percent identity matrix within Clustal Omega. rpoN (NEIS0212) of each isolate used in our study was identified by BLAST, results were manually inspected to define coding sequences, and alleles were assigned using PubMLST. Sequences were translated using EMBOSS Transeq (https://www.ebi.ac.uk/Tools/st/emboss_transeq/). N. meningitidis N putative domains were identified based on alignment with E. coli N using the EMBOSS Water pairwise sequence alignment tool with the default settings (https://www.ebi.ac.uk/Tools/psa/emboss_water/). SnapGene was used to construct plasmid maps and for automatic annotation of features and ORFs in whole-genome sequences.
Mutant construction. The primers used in this study are listed in Table S3 in the supplemental material. N. meningitidis strains lacking factors were constructed by gene replacement. First, DNA fragments corresponding to 479 bp upstream and 712 bp downstream of the N. meningitidis S4rpoN ORF (NEIS0212, homologue of FAM18 NMC_RS01150) and a kanamycin resistance cassette were amplified by PCR using primers ML20/21, ML22/23, and ML24/25, respectively. The upstream and downstream fragments were cloned into pGEM-T Easy flanking a kanamycin resistance cassette, resulting in pGEM-T Easy ΔrpoN. This plasmid was digested with BamHI and NotI, and the ΔrpoN fragment was purified by gel extraction prior to transformation into S4. Plasmid pUC19ΔrpoE was generated by amplifying 483-bp and 459-bp fragments corresponding to regions upstream and downstream of the N. meningitidis rpoE (NEIS2123, homologue of FAM18 NMC_RS11230) using primers ML284/ML285 and ML288/ML289, respectively. A kanamycin resistance cassette was amplified using primers ML286/ML287, and the three fragments were ligated into pUC19 using the NEB Builder HiFi DNA assembly kit (New England Biolabs). The construct for replacing rpoE with the kanamycin resistance gene was amplified from pUC19ΔrpoE by PCR and then purified and used for transformation into S4.
To generate S4 and S4ΔrpoN harboring P pilE -lacZ at the native pilE locus, first rpoN was replaced with ermC, encoding erythromycin resistance. Briefly, regions upstream (485 bp) and downstream (886 bp) of the rpoN start and stop codons, respectively, were amplified from existing mutant S4ΔrpoN (Kan r ) genomic DNA (gDNA) using primers ML20/ML46 and ML25/ML52, and overlap PCR was used to clone these fragments 5= and 3= of an erythromycin resistance gene, respectively. The product was cloned into pCR2.1TOPO, resulting in pCR2.1TOPOΔrpoN, which was then used as a template to amplify the deletion construct for transformation, generating S4ΔrpoN (Ery r ). Subsequently, the pilE ORF was replaced with lacZ as follows. A 673-bp region immediately 5= of the pilE ORF was amplified from S4 gDNA using ML37/ML34. A fragment corresponding to the region 3= of the pilE ORF with a kanamycin resistance marker inserted 81 bp downstream of the pilE stop codon was amplified from strain S4_pilEkanR (R. Exley, unpublished data) using ML30/ML31. The two fragments were cloned into pCR2.1TOPO flanking the lacZ gene, which was amplified from pRS415 (69) using primers ML32/ML33. The resulting plasmid, pCR2.1TOPOP pilE -lacZ, was used as a template to amplify DNA for transformation of wild-type S4 or S4ΔrpoN.
N. meningitidis strains overexpressing factors were constructed by inserting the relevant ORF at the iga-trpB intergenic locus (41) of N. meningitidis S4, under the control of an IPTG inducible promoter. For this purpose, pNMC2 was generated. First, a fragment comprising part of the iga gene and terminator was amplified from S4 gDNA using primers ML151 and ML149. Next, a fragment comprising ermC, the lac regulatory region, and a multiple-cloning site (MCS) was amplified from pGCC4 (a gift from Hank Seifert) (Addgene plasmid 37058; http://n2t.net/addgene:37058; RRID:Addgene_37058) using primers ML148/ML150. The two fragments were fused by overlap PCR using primers ML148 and ML151. The resulting PCR product was then combined using Gibson assembly with a fragment amplified from plasmid pNCC1 (70) with primers ML152 and ML153 (providing the origin of replication and kanamycin selection marker) and a fragment comprising the 3= end of the trpB gene amplified from S4 gDNA using primers ML146 and ML147. Genes encoding C-terminally tagged sigma factors were amplified as follows. The rpoN, rpoE, and rpoH (NEIS0663, homologue of FAM18 NMC_RS03540) genes were amplified from S4 gDNA using primers ML154/ML155, ML280/ML282, and ML184/ML188, respectively. The PCR products were ligated into HindIII-digested pNMC2 using the NEB Builder HiFi DNA assembly kit to generate pNMC2rpoN Nm , pNMC2rpoE, and pNMC2rpoH. pNMC2rpoN Nel was generated using Gibson assembly with a fragment amplified from pNMC2rpoN Nm using ML172/ML174 and a fragment comprising rpoN from N. elongata (NELON_RS02880) amplified using gDNA from strain 29315 and primers ML173/ML175. Plasmids were linearized by digestion with NcoI to ensure double-crossover recombination and prevent integration of the entire plasmid and were used to transform S4. DNA from single transformants selected on erythromycin was analyzed by PCR, and gDNA from PCR-positive colonies was backcrossed into the parental strain. Backcrossed colonies were pooled, and gDNA from the pooled stock was checked by PCR and sequencing of the region of interest.
RNA isolation. N. meningitidis was grown at 37°C in liquid BHI, and at the indicated times a volume equivalent to an OD 600 of Ϸ8 was centrifuged at 4,000 rpm for 10 min. The supernatant was removed, and total RNA was isolated from pellets using TRIzol extraction (Thermo Fisher). Briefly, cell pellets were resuspended in 2 ml of TRIzol and incubated at room temperature for 5 min before adding 200 l of chloroform. Samples were shaken by hand for 15 s, incubated at room temperature for 2 to 3 min, and then centrifuged at 12,000 ϫ g for 15 min at 4°C, and the aqueous phase was transferred to a fresh tube. RNA was precipitated by adding 500 l isopropanol, followed by centrifugation at 20,000 ϫ g for 30 min at 4°C. RNA pellets were washed with 75% ethanol and then air dried and resuspended in diethylpyrocarbonate (DEPC)-treated water. Following DNase treatment for 2 h at 37°C, RNA was extracted with phenol-chloroform (5:1). The aqueous phase was reextracted with DEPC-treated water and chloroformisoamyl alcohol. Samples were vortexed and centrifuged at 20,000 ϫ g for 5 min at 4°C. The aqueous phase was transferred to a tube containing 33 l of 3 M Na acetate (pH 4.5) and 812.5 l of 99.5% ethanol, and RNA was precipitated over 2 h or overnight at -20°C. Following centrifugation at 20,000 ϫ g for 30 min at 4°C, RNA pellets were washed with 75% ethanol, centrifuged again for 10 min at 20,000 ϫ g, air dried, and resuspended in 200 l DEPC-treated water.
qRT-PCR. All primers used in quantitative reverse transcription-PCR (qRT-PCR) are listed in Table S3. RNA (2.5 g) was used to synthesize cDNA with primers specific for the target gene containing a tag sequence (CCGTCTAGCTCTCTCTAATCG) which is not present in the N. meningitidis genome. RNA was reverse transcribed using reverse transcriptase III polymerase (Invitrogen) and treated with RNase H for 20 min at 37°C. cDNA was purified using the Wizard gel PCR purification kit (Promega). qRT-PCR was performed using Power SYBR green PCR master mix (Applied Biosystems). Primer efficiency was evaluated using serial dilutions of gDNA to generate a standard curve, and the slope of the standard curve for each primer pair was calculated. StepONEPlus real-time PCR software was used to collect qRT-PCR data; the ΔC T method (where C T is the threshold cycle) was used to analyze the data (71). Results are presented as the normalized R value (where the R value is calculated as 2 ϪCT ) relative to tmRNA. Experiments were repeated at least three times, each time with triplicate samples. The results shown are means and standard deviations (SD) of three biological replicates; tmRNA was used as a control.
Primer extension. Primer extension was carried out using 1 g and 2.5 g of RNA isolated from N. meningitidis grown to mid-log phase (OD 600 Ϸ 0.4 to 0.5). Primers (P1 and P2 [Table S3]) were radiolabeled using [␥-32 P]ATP (PerkinElmer). A mix of radiolabeled primers and RNA was incubated for 3 min at 95°C and rapidly cooled on ice for 4 min. Reverse transcription was performed using Superscript II (Invitrogen) by incubating samples at 45°C for 60 min then stopped by deactivating the enzyme for 10 min at 70°C. Formamide loading dye (10 l) was added to each reaction mixture, and samples were heated briefly at 95°C. Fifteen microliters of each sample was loaded on a 12% denaturing polyacrylamide sequencing gel, and electrophoresis was performed for 2 to 3 h at 20 to 25 W until bromophenol blue dye reached the bottom of the gel. Products were analyzed using a phosphorimager (FujiFilm). Sanger sequencing was performed using primer P2 and a DNA cycle sequencing kit (Jena Bioscience) with a PCR-amplified fragment of the promoter region of pilE from S4 as a template.
Western blot analysis. N. meningitidis whole-cell extracts were prepared from bacteria grown overnight at 37°C in the presence of 5% CO 2, on solid medium or in liquid with appropriate antibiotics with or without 1 mM IPTG. Suspensions were normalized to an OD 600 of 1 in 1 ml BHI and then centrifuged for 5 min at 12,000 ϫ g, and the pellets were resuspended in an equal volume of sterile water and 2ϫ SDS-PAGE lysis buffer (1 M Tris-HCl [pH 6.8], 10% SDS, 30% glycerol, 1% bromophenol blue) containing 200 mM ␤-mercaptoethanol. Samples were boiled for 10 min prior to electrophoresis. Proteins were separated on 12% polyacrylamide gels and transferred to nitrocellulose membranes (Hybond-C Extra [Amersham] or Immobilon-P polyvinylidene difluoride [PVDF] [Millipore]) using a Bio-Rad Trans-Blot Turbo system. Membranes were blocked for 1 h at room temperature or overnight at 4°C in PBS-5% milk. Western blotting was performed using anti-FLAG antibody (F1804; Sigma) at 1:5,000 and goat antimouse IgG-horseradish peroxidase (HRP) (final dilution, 1:10,000; Dako) or using antipilin (EP11270 antipeptide, 1:10,000) and goat anti-rabbit IgG-HRP (final dilution, 1:10,000; Santa-Cruz). Anti-RecA (1:5,000) (ab63797; Abcam) followed by goat anti-rabbit IgG-HRP (final dilution, 1:10,000; Santa-Cruz) or Coomassie blue staining was used as a loading control. All incubations with antibodies were for 1 h at room temperature in PBS-5% milk containing 0.1% Tween. Membranes were washed three times with PBS-0.1% Tween between incubations. Cross-reacting proteins were visualized using ECL detection reagent (GE Healthcare).