Neisseria meningitidis Uses Sibling Small Regulatory RNAs To Switch from Cataplerotic to Anaplerotic Metabolism

ABSTRACT Neisseria meningitidis (the meningococcus) is primarily a commensal of the human oropharynx that sporadically causes septicemia and meningitis. Meningococci adapt to diverse local host conditions differing in nutrient supply, like the nasopharynx, blood, and cerebrospinal fluid, by changing metabolism and protein repertoire. However, regulatory transcription factors and two-component systems in meningococci involved in adaptation to local nutrient variations are limited. We identified novel sibling small regulatory RNAs (Neisseria metabolic switch regulators [NmsRs]) regulating switches between cataplerotic and anaplerotic metabolism in this pathogen. Overexpression of NmsRs was tolerated in blood but not in cerebrospinal fluid. Expression of six tricarboxylic acid cycle enzymes was downregulated by direct action of NmsRs. Expression of the NmsRs themselves was under the control of the stringent response through the action of RelA. Small sibling regulatory RNAs of meningococci, controlling general metabolic switches, add an exciting twist to their versatile repertoire in bacterial pathogens.

referred to as the stringent response (3,4). This response is mediated by the alarmones guanosine 5=,3=-bispyrophosphate and guanosine pentaphosphate, ppGpp and pppGpp, collectively referred to here as (p)ppGpp (5). In Escherichia coli, (p)ppGpp is synthesized from GTP and ATP via the action of two paralogous enzymes, RelA and SpoT (4,6). The transcriptional changes occur mainly as a result of the direct effects of (p)ppGpp and its cofactor, the (protein) transcription factor DksA, on RNA polymerase (4). In addition to (p)ppGpp and regulatory proteins, among which are the transcription factors (TFs) small regulatory RNAs (sRNAs) are also involved in the switch from nutrient-rich (feast) to nutrient-limiting (famine) growth conditions of bacteria (7,8).
sRNAs are important players in many cellular processes and prominent in those involving adaptive physiological changes. They can function as posttranscriptional regulators of gene expression to orchestrate stress responses and metabolism. Many sRNAs are synthesized upon nutritional stresses encountered by pathogens. They often regulate expression of target mRNAs that form part of a single nutritional regulatory circuit or network. sRNAs usually act by occupying or freeing up ribosomal entry sites of target transcripts as well as by regulating the accessibility of transcripts for RNases in an antisense fashion (9)(10)(11)(12). The RNA chaperonin protein Hfq is frequently involved, enhancing these processes (13,14).
We identified two highly conserved sRNAs, designated sibling Neisseria metabolic switch regulators (NmsRs), in N. meningitidis which are functionally involved in the regulation of tricarboxylic acid (TCA) cycle activity by antisense mechanisms. These novel sibling sRNAs extend the stringent response in meningococci, thereby connecting metabolic status to colonization and, possibly, virulence. (Continued on next page) Sibling sRNAs of N. meningitidis as Metabolic Switches ® pressing NmsRs did start to grow (Fig. 2). However, sequence analysis of the region encoding NmsRs showed part of nmsR B to be deleted in the escape variant.
TCA cycle enzymes of N. meningitidis under the control of NmsRs. To further assess the biological significance of the NmsRs, we compared the protein expression profile of the wt strain with that of the ΔnmsR A ΔnmsR B strain by mass spectrometric analysis of whole-cell lysates. Transcription of flanking genes of the ΔnmsR A ΔnmsR B strain remained unaffected after replacement of nmsR A and nmsR B with the erythromycin resistance cassette (not shown). Of all 2,300 annotated open reading frames in the N. meningitidis MC58 genome (17), 515 proteins (22%) were identified at the protein level. Of these, 387 yielded reliable quantification of relative expression comparing wt and ΔnmsR A ΔnmsR B strains (see Table S1 in the supplemental material). Differentially expressed proteins ranged from 7-fold upregulated to 6-fold downregulated. Using a 1.5-fold up-or downregulation as a cutoff for differential expression, a total of 18 upregulated and 10 downregulated proteins were identified (Table 1). Among the 18 proteins with increased expression in the ΔnmsR A ΔnmsR B strain, 10 (56%) were either involved in the TCA cycle directly or linked to it, such as acetate kinase (AckA-1), involved in acetyl coenzyme A (CoA) synthesis, or PrpC and PrpB, involved in propanoate metabolism feeding into the TCA cycle through succinyl-CoA. Other upregulated proteins belonged to the glycine cleavage system (GlyA), part of glycine/serine meta-

FIG 1 Legend (Continued)
the white circles indicate predicted Rho-independent terminators. Note that the region visualized encodes two distinct transcripts, indicated as NmsR A and NmsR B . Coverage of the NmsR B transcript is 5-fold (~7,500 reads/nt); that of the transcripts of NmsR A is 1,500 reads/nt. (C) Predicted secondary structures of siblings NmsR A and NmsR B . Secondary structures were predicted using Mfold. Unique sequences of the NmsRs are indicated by boxes. Putative ␣-SD sequences are circled; stem-loops are indicated as SL.
Pannekoek et al. ® bolism, or were involved in valine, leucine, and isoleucine degradation (3-hydroxyacid dehydrogenase; NMB1584). In contrast, proteins involved in ATP synthesis-coupled proton transport (AtpG and AtpC), a protein involved in the pentose pathway (Zwf), and proteins involved in biosynthesis of valine/leucine and isoleucine (IlvD and IlvA) are downregulated without NmsRs ( Table 1). Complementation of the ΔnmsR A ΔnmsR B strain with a plasmid encoding both NmsRs led to normalization of protein levels for a slight majority of the overexpressed proteins identified (10/18; results not shown). Together, these results strongly suggest that in meningococci without NmsR activity, metabolism has been switched to higher TCA cycle activities, which are less strongly coupled to respiration. As we also observed notable expression of the NmsRs in transcriptome analyses of meningococci grown in nutrient-rich medium, this implies relatively low TCA cycle activity in meningococci grown in media with abundant nutrients. In the absence of NmsRs, the role of the TCA cycle in meningococcal metabolism increases, shifting to anabolism with, e.g., breakdown products of branched-chain amino acids as anaplerotic substrates and synthesis of components beneficial for growth under nutrient-poor conditions ( Table 1).
NmsR A and NmsR B translational downregulation of TCA cycle enzymes is mediated by anti-Shine-Dalgarno sequences. Results indicate that mRNAs encoding TCA cycle enzymes are potential targets for the NmsRs. In silico analysis (19) indeed revealed putative interactions between both NmsRs and 5= untranslated regions (UTRs) of PrpB, PrpC, GltA, and SucC mRNAs. In addition, SdhC and FumC were identified as putative targets of NmsRs (Fig. S1 in the supplemental material). To obtain experimental evidence for the interaction between the NmsRs and potential target mRNAs, we used a well-established gfp reporter system in Escherichia coli (20). The 5= UTR of the h Fold change of Ն1.5. Downregulation is expressed as the reciprocal with added "Ϫ." i Independent t test, two-tailed, equal variable. All samples, P Յ 0.005 (except in red); bold values are significant after correction for false discovery rate according to the work of Benjamini and Hochberg (66). Genes indicated in green are confirmed as true targets of NmsRs in the gfp reporter system. potential target mRNA and its first 7 to 13 codons were fused in frame to a gfp coding region (target-gfp fusion) which is constitutively expressed in E. coli together with the NmsRs from another plasmid vector. However, transformation of E. coli with plasmids harboring both nmsR A and nmsR B or only nmsR B failed, even when a strain (JVS-2001) was used in which the sRNA chaperonin gene hfq was deleted or when a low-copynumber vector was used (20). However, E. coli could be transformed with the plasmid harboring only nmsR A , though it displayed attenuated growth (not shown) in all cases. Reduced fluorescence of target-gfp fusion in the presence of NmsR A expression, but not in the presence of expression of a control nonsense sRNA, indicates a direct interaction between NmsR A and the 5= UTR of the target mRNA. In this way, direct translational control by NmsR A was demonstrated for six out of eight tested putative target mRNAs (prpB, prpC, sdhC, gltA, sucC, and fumC [P Ͻ 0.005]) ( Fig. 3A and B). The observation that fluorescence levels of the target-gfp fusion of two putative mRNA targets (acnB and cbbA) remained unaffected upon NmsR A expression in trans indicates that the slowergrowth phenotype of E. coli upon NmsR A expression is not interfering with expression and/or proper folding of green fluorescent protein (GFP) as such (Fig. 3B).
Sequence comparison of the 5= UTR of the target mRNAs with proven direct NmsR A interaction showed homology around the Shine-Dalgarno sequence motif (SD), part of the ribosome binding site (Fig. S1) (21). The NmsRs are predicted to fold into similar secondary structures consisting of three stem-loops (SLs) (Fig. 1C). The single-stranded region between SL1 and SL2 exposes a UC-rich sequence. This region together with the UC-rich single-stranded loop of SL2 is (partly) complementary to the SD of the target mRNAs (Fig. S1). Both of these regions, referred to as ␣-SDs, are completely conserved among more than 7,335 meningococcal genomes analyzed (accessed at http://pubmlst .org/neisseria/) (16). Mutagenesis of either of the ␣-SD sequences in NmsR A (using mutations designed to preserve the secondary structure of the NmsRs) abrogated the downregulation of all targets but GltA (Table S2). For the latter, mutagenesis of ␣-SD1 did not influence downregulation, but downregulation of GltA disappeared upon ␣-SD2 mutation (Table S2). Mutations outside ␣-SD regions had no effect on regulation (not shown). Replacement of nucleotides in the SD regions of the target mRNAs resulted in fluorescence levels of the cells below the level of detection. Consequently, it was not possible to investigate whether downregulation could be restored by the introduction of compensatory mutations. Taken together, these results strongly argue that NmsR A inhibits synthesis by an antisense mechanism that involves direct base pairing to 5= UTRs of six out of eight target-gfp fusions assessed, presumably by preventing ribosomal entry.
NmsRs alter expression of transcript levels of TCA cycle enzymes in meningococci. The effect of NmsR expression on the expression of genes of the TCA cycle targeted by NmsR A was assessed in meningococci grown under two culture conditions, TSB (nutrient rich) and Jyssum medium (glucose as the sole carbon source [18]), in which we anticipated differential expression. Transcript levels of all NmsR A targets were indeed (1.5-to 8-fold) higher in meningococci with the NmsRs deleted and grown in TSB than in wt (P Ͻ 0.005). In Jyssum medium, transcript levels of prpC, gltA, and sucC were (5-to 3-fold) higher (P Ͻ 0.001) in the ΔnmsR A ΔnmsR B strain than in the wt strain. Transcript levels of prpB and fumC in ΔnmsR A ΔnmsR B and wt strains were not significantly different in Jyssum medium (Fig. 4). Of note, in all cases (except sdhC), the transcript levels of the targets were significantly lower in meningococci overexpressing NmsR A (P Ͻ 0.05) or NmsR B (P Ͻ 0.01) (or 2-fold lower in the case of prpB [P ϭ 0.26]) or after overexpressing both sRNAs (P Ͻ 0.01) in the NmsR A and NmsR B deletion strain and in all these cases (except sdhC) became comparable to target levels found in wt meningococci when grown in TSB (Fig. 4). Transcript levels of sdhC in Jyssum medium were not significantly different in the wt strain from those in the NmsR A and NmsR B deletion strain of overexpression isogenic variants (Fig. 4).
NmsRs are connected to the stringent response and controlled by RelA. In Neisseria gonorrhoeae, RelA is the sole producer of (p)ppGpp (22), which acts with DskA in interacting with RNA polymerase to regulate transcription. Whether a given promoter is directly controlled by (p)ppGpp and DksA is dictated by a DNA sequence motif, the so-called discriminator. Repressed targets typically contain GC-rich 7-nucleotide discriminators between the Ϫ10 box hexamer and the transcriptional start site, whereas activated promoters harbor AT-rich discriminators at this position (3). Of note, such a GC-rich nucleotide region can be identified between the putative Ϫ10 site and the transcriptional start site of NmsR A (Fig. 1A). To investigate whether NmsRs are directly controlled by the stringent response, we created a relA-knockout strain of H44/76 (the ΔrelA strain) by replacing relA with an erythromycin resistance cassette and assessed NmsR A and NmsR B levels after growth in TSB or Jyssum medium by reverse transcription-quantitative PCR (RT-qPCR). We did not obtain viable meningococci when relA was expressed in trans. Of interest, upon deletion of relA, NmsR A transcript levels . Meningococci were cultured in TSB (nutrient rich) and Jyssum medium (glucose as the sole carbon source [18]). (B) Relative expression levels of NmsR targets. Transcript levels assessed by RT-qPCR in wt meningococci and in meningococci in which nmsR A and nmsR B are deleted (Δ) and in the Δ strain overexpressing NmsR A (ΔϩnmsR A ), overexpressing NmsR B (ΔϩnmsR B ), and overexpressing nmsR A and nmsR B (ΔϩnmsR A nmsR B ) (error bars, standard errors of the means; technical replicates, n ϭ 8, over biological, n ϭ 3). Meningococci were cultured in TSB (nutrient rich) and Jyssum medium (glucose as the sole carbon source [18]). were 10-fold higher than wt levels (P Ͻ 0.0001), reaching levels that were comparable to NmsR B levels in wt cells grown in TSB. NmsR B levels were also significantly higher in ΔrelA cells and increased 5-(P Ͻ 0.0005) and 2.5-fold (P Ͻ 0.05) in TSB and Jyssum medium, respectively (Fig. 5). No significant difference in levels of NmsRs was observed between cells grown in medium with glucose as sole carbon source and cells grown in nutrient-rich medium (Fig. 5).
We next investigated whether transcript levels of the NmsR A targets were affected upon deletion of relA. Transcript levels of all NmsR A targets except sdhC in cells grown in TSB were relatively low and comparable in wt and ΔrelA cells. In Jyssum medium, the transcript levels of all targets in the wt strain, again with the exception of sdhC, were 2-to 17-fold higher (P Ͻ 0.0005) than levels in cells cultured in TSB (Fig. 6). However, upon deletion of relA, transcript levels of prpB, prpC, gltA, and sucC were inversely correlated with levels of NmsR A and NmsR B . In Jyssum medium, the transcript levels of the targets in ΔrelA meningococci were 2-to 7-fold lower (P Ͻ 0.0001) than in wt meningococci and comparable to levels in wt cells or ΔrelA cells grown in TSB (Fig. 6). Upon deletion of nmsR A and nmsR B in the ΔrelA strain (ΔrelA ΔnmsR A ΔnmsR B ) in cells cultured in TSB, the transcript levels of all targets, with the exception of sdhC, were 2to 10-fold higher than those in ΔrelA and wt strains. Of note, transcript levels of prpB, prpC, gltA, and sucC in the triple mutant ΔrelA ΔnmsR A ΔnmsR B cells grown in Jyssum medium were 2-to 5-fold higher (P Ͻ 0.0001) than in the single ΔrelA mutant. These results confirmed relA-mediated downregulation of NmsRs, irrespective of the culture conditions used. Transcript levels of sdhC and fumC in the triple-knockout ΔrelA ΔnmsR A ΔnmsR B strain remained unaffected in Jyssum medium compared to the ΔrelA single knockout (Fig. 6).

DISCUSSION
In this study, we identified novel sibling regulatory sRNAs of N. meningitidis that establish a connection between the stringent response and the riboregulatory network. Our data showed regulation of the TCA cycle activity by direct action of sibling sRNAs in N. meningitidis. The expression of NmsRs themselves is under the control of the stringent response via RelA. The activity of the sibling sRNAs is crucial when meningococci encounter different host environments with variable nutrient supplies, such as blood and CSF. High sibling sRNA expression allows replication and survival in blood but impairs growth in CSF. Whether this is solely due to differential expression of TCA cycle enzymes or whether other, so-far-unknown targets are involved is the subject of further study. Sibling sRNAs of N. meningitidis as Metabolic Switches ® NmsRs are the first sibling sRNAs and only the third sRNA species in Neisseria for which target genes are experimentally confirmed (23). NrrF, a Fur-regulated sRNA, has been identified in meningococci (24,25) and gonococci (26) and is upregulated under iron-depleted conditions. This sRNA has been shown to be involved in regulation of sdhA, belonging to the operon encoding the succinate dehydrogenase complex (sdhCDAB). Recently, an sRNA was identified in N. gonorrhoeae that acts in cis and influences antigenic variation of pilin (27). Other sRNAs, among them AniS in meningococci and FnrS in gonococci, are synthesized under oxygen limitation, but their targets remain elusive (28,29). The same is true for a E -dependent sRNA that has been identified in N. meningitidis (15).
By employing proteomics, putative targets of the NmsRs were identified. We experimentally validated direct interaction for four mRNAs coding for enzymes belonging to the TCA cycle (sdhC, gltA, sucC, and fumC) and for two mRNAs encoding enzymes producing intermediates of the TCA cycle (prpB and prpC) (schematically represented in Fig. 3C). Of interest was our identification of sdhC as a direct target of NmsRs. sdhC is the first gene of the cluster sdhCDAB, coding for the succinate dehydrogenase complex. This complex, which generates fumarate from succinate during the TCA cycle, concomitantly feeds electrons to the respiration chain (30). As mentioned before, in meningococci, NrrF has been shown to be involved in the Fur-dependent regulation of genes belonging to this cluster (24,25). Although we showed a direct interaction between NmsR A and sdhC resulting in downregulation of GFP by using the gfp reporter system, in the genetic background of meningococci we did not observe regulation. Likewise, we were also not able to confirm regulation for FumC in meningococci. This might imply either that both mRNAs are not true targets of NmsR A or that expression of these transcripts, in the genetic background of meningococci, is more complex, e.g., under the control of other regulators as well. The latter is not unlikely since it has been shown that transcript levels of sdhC are controlled by the two-component regulator MisR (31), Relative expression levels of NmsR targets, assessed by RT-qPCR, in wt meningococci, in meningococci in which relA is deleted (ΔrelA), and in meningococci in which relA nmsR A , and nmsR B are deleted (ΔrelAΔ) after growth in TSB (nutrient rich) and Jyssum medium (glucose as the sole carbon source [18]) (error bars, standard errors of the means; technical replicates, n ϭ 5, over biological, n ϭ 1). and as mentioned above, sdhA levels are controlled by NrrF. If true, sdhCDAB might be the first example of a cistronic mRNA in meningococci that is subject to regulation by two different sRNA species as well as a two-component regulator.
We identified prpB and prpC as targets of NmsRs. It has been shown that expression of these genes enables survival of the meningococci in the "normal" habitat (i.e., the adult nasopharynx) by allowing utilization of propionic acid as a supplementary carbon source (32). Thus, use of propionate becomes crucial under conditions of nutrient limitation. These observations are nicely in line with our data, as overexpression of NmsRs under nutrient-limiting conditions (e.g., Jyssum medium or liquor) leads to growth arrest. Tightly regulating NmsR expression is an essential prerequisite to support growth under divergent in vivo conditions, exemplified by colonization of the nasopharynx and replication in CSF.
We convincingly demonstrated that the expression of at least four different mRNA species is controlled by NmsRs. This makes NmsRs, as far as we know, the first example of sibling sRNAs in N. meningitidis acting on multiple trans-encoded targets, thus rewiring interconnected transcriptional networks, possibly including the MisR and Fur regulon. The unexpected transcript levels of sdhC and fumC observed with some strain-growth condition combinations could reflect such complex regulation.
Many small RNAs are known to contain one single-stranded domain that is able to interact with multiple target mRNAs (33)(34)(35)(36)(37). Other sRNAs have several functional domains that base pair with different sets of target mRNAs (38)(39)(40)(41). Using in vivo experiments, we demonstrated that NmsR A represses synthesis of its mRNA targets most likely by an antisense mechanism. This involves base pairing of predicted singlestranded ␣-SD regions (UC-rich) of the NmsRs to a sequence stretch overlapping the SD in the targets. Basically, this antisense mechanism is shared by many other sRNAs (7,42). Of interest, the NmsRs contain two ␣-SD regions apparently acting on the same set of mRNAs by duplex formation with the region encompassing the SD. Both ␣-SD regions are characterized by UC-rich stretches but differ slightly from each other in sequence. In five out of six cases, action of both ␣-SDs is required for downregulation, suggesting coordinated activity, while in one case (gltA) downregulation is abrogated only after mutating ␣-SD2. This suggests that, for this target, ␣-SD2 acts independently of ␣-SD1 and that only duplex formation with ␣-SD2 is essential. Calculation of the minimum free energy (MFE) of the duplexes before and after mutagenesis of the ␣-SDs using RNAhybrid (43) could not accurately predict the in vivo outcome of this regulation (not shown). However, similar modes of action have recently been described for the LhrC family of sibling sRNAs of Listeria monocytogenes (44,45), and future experiments are necessary to investigate whether, for example, less conserved flanking sequences of the region of interaction of NmsRs with their targets might contribute to a different affinity and subsequent outcome of the duplex formation. The finding that in silico predictions of duplex formation based on complementarity of target and sRNA sequences do not always match in vitro observations, e.g., the predicted target cbbA apparently not being controlled by NmsR A , is also important in this context.
We observed that expression of both NmsRs or NmsR B in E. coli did not result in viable cells, while expression of NmsR A in E. coli showed attenuated growth. Possibly, the expression of both NmsRs is toxic for E. coli. Alternatively, E. coli encodes (an) NmsR A and NmsR B target(s), which will be interesting to identify as well. This interpretation is strengthened by the observation that the slower-growth phenotype of E. coli disappeared after mutagenesis of the ␣-SDs of NmsR A (not shown).
The activity of many sRNAs in bacterial pathogens depends on the RNA chaperone Hfq (13,14,46). Among the direct targets of NmsR A identified, three proteins (GltA, PrpB, and PrpC) were also found to be overexpressed in an Hfq deletion strain of N. meningitidis (Δhfq) (47). This overlap between some of the NmsR A targets and Hfqdependent mRNAs might indicate that for these cases NmsRs act in concert with Hfq. In general, two signatures in the sequences of sRNAs are reported as preferred binding sites for Hfq. The first is a typical A/U-rich single-stranded stretch that precedes the predicted Rho-independent terminator. The second signature consists of terminal U residues (14,48,49). The second signature is obviously present, but the first signature seems absent from NmsRs. Thus, whether the observed overlap between the Hfq and NmsR A regulon in N. meningitidis represents a joint action of the chaperone and sRNAs, or represents a more indirect phenomenon, awaits further experiments.
The continuous discovery of more sRNAs has resulted in the identification of several examples of homologous sRNAs, "sibling" sRNAs (50). We identified a novel sibling member of this expanding class of sRNAs. The NmsRs are encoded in tandem in an intergenic region. Equal expression levels of the two sRNAs were observed under nutrient-rich and nutrient-poor conditions, but relative expression levels of NmsR A were very low compared to those of NmsR B . The relatively high expression levels of NmsR B might suggest that NmsR B acts redundantly in a compensatory manner on the same targets, as has been described for sRNAs of other pathogens (36,51,52). The system is more complicated, however, as illustrated by the fact that target levels were significantly downregulated upon overexpression of single NmsRs but expression of both was required for complete repression (demonstrating combined NmsR action). In addition, the action of both sRNAs was also required for growth inhibition of meningococci under nutrient-limiting conditions. Thus, these observations suggest classification of NmsRs as riboregulators that act cumulatively, each contributing in a different degree to overall adaptation. Homologous sRNAs acting together have also been described for other pathogens (50,53). It should be noted that although single NmsR A levels were low and single NmsR B levels were significantly lower when they were expressed in ΔnmsR A ΔnmsR B cells, they were still sufficient to downregulate 5= UTR gfp fusion products in E. coli and mRNA target levels in meningococci.
How the meningococcal NmsRs are regulated themselves and whether they are fine-tuned individually with regard to their own expression levels as well as their target preferences have to be elucidated further. We could show the expression of NmsRs to be elevated in a relA knockout, indicating that it is connected to the stringent response. A direct interaction of (p)ppGpp with the putative negative discriminator in the 5= UTR of NmsR A looks tempting. Of interest, many gammaproteobacterial genes shown to be direct targets of (p)ppGpp contain typical 70 -dependent promoters (3). Indeed, inspection of the predicted promoter region of NmsR A shows a high similarity to -dependent promoters with the 70 signatures identified in E. coli (Ϫ35 element TTGACA [E. coli consensus TTGACA] and Ϫ10 element GATAAT [E. coli consensus TATAAT]) (54)(55)(56). Also, the 5= UTR of NmsR B has an, albeit weaker, 70 signature. This might suggest -dependent NmsR A transcription directly controlled by (p)ppGpp. Expression levels of NmsR B are much higher when the two single sRNAs are coexpressed, reflecting less restricted transcription for this sRNA. Cotranscription of the two sRNAs might also be needed for their stabilization. How the possible mutual stabilization and interregulation might work is under investigation. Alternatively, the NmsRs are indirectly controlled by (p)ppGpp. In some cases, regulators of sRNA expression are located in the close vicinity of the particular sRNA to be regulated (57). The gene encoding Lrp (NMB1650) is located directly downstream of the locus encoding the sibling sRNAs (though in opposite orientation). The expression of Lrp might be stimulated by (p)ppGpp (58). However, NmsR levels did not significantly change upon either deletion or overexpression of lrp (not shown).
In conclusion, we identified sibling sRNAs targeting genes encoding TCA cycle enzymes, stressing their importance in the adaptation to changing environments in the host. The riboregulated network of the sibling sRNAs is part of the RelA-regulated stringent response. NmsRs of N. meningitidis form a crucial part of the riboregulatory network monitoring metabolic status, translating this into colonization with likely implications for pathogenesis.

MATERIALS AND METHODS
Bacterial strains and culture conditions. N. meningitidis strain H44/76, B:P1.7,16: F3-3:ST-32 (cc32), is closely related to the sequenced serogroup B strain MC58, belonging to the same clonal complex (59). Meningococci were grown for 16 to 18 h on GC plates (Difco) supplemented with 1% (vol/vol) Vitox (Oxoid) or on PVX plates (bioMérieux) at 37°C in a humidified atmosphere of 5% CO 2 . Broth culturing was performed in tryptic soy broth (TSB) (BD), GC medium supplemented with 1% (vol/vol) Vitox, or Jyssum medium (18), on a gyratory shaker (180 rpm) at 37°C. When appropriate, plates or broths were supplemented with erythromycin (Erm) (5 g/ml) and/or chloramphenicol (Cm) (5 g/ml) and/or kanamycin (50 g/ml). Expression of recombinant DNA in meningococci was induced by IPTG (isopropyl-␤-D-thiogalactopyranoside) (0.5 mM). Growth in broth was monitored by measuring optical density of cultures at 530 nm (OD 530 ) at regular time intervals. Growth of meningococci in human blood and CSF was monitored as follows. Blood was collected in 4-ml Vacutainers with 17 IU/ml sodium heparin (BD) from a healthy male volunteer approximately 2 h prior to use. CSF was extracted, with informed consent, from patients with suspected normal-pressure hydrocephalus, either used fresh (within Ͻ24 h stored at 4°C) or aliquoted and stored at Ϫ80°C. CSF white blood cell count and glucose and protein levels were within normal range. Heparinized human blood and CSF, the latter diluted prior to use to 50% (vol/vol) with phosphate-buffered saline (PBS), was inoculated with the equivalent of approximately 1 ϫ 10 5 meningococci, which were obtained from precultures in TSB (OD 530 ,~0.2 to 0.4). Aliquots of 220 l were incubated in sterile 96-well plates (Corning) and incubated at 37°C in a humidified atmosphere of 5% CO 2 . At regular time intervals, 20-l samples were serially diluted and plated on PVX plates (bioMérieux), and colonies were counted after 16 to 18 h of growth at 37°C in a humidified atmosphere of 5% CO 2 .
E. coli strain Top10 (Invitrogen) was used to clone gfp fusions and in experiments involving coexpression of gfp fusions and sRNAs. E. coli strain Top10F= (Invitrogen) was used to clone sRNA expression plasmids and pCR2.1 (Invitrogen) and pEN11-pldA (60) Table S3 in the supplemental material. DNA was gel purified using a GeneJET gel extraction kit (Thermo Scientific). Digestion and ligation were carried out using enzymes supplied by New England BioLabs or Thermo Scientific. Plasmid pCR2.1 was used for cloning and sequencing of PCR products. Plasmids pXG-0 (control for autofluorescence), pXG-1 (control for sRNA effect on gfp expression), and pXG-10 (standard plasmid for gfp fusion cloning) were kindly provided by J. Vogel (Würzburg, Germany) and have been described previously (20). The nmsR A gene was amplified using primers RHsRNA25CFW11 and RHsRNA25GFPRV13b and inserted into the sRNA-expressing plasmid pZE12-luc, thereby creating pNmNmsR A using the cloning strategy described previously (20). The shuttle vector pEN11-pldA was used to express sRNAs in meningococci (60).
N. meningitidis was transformed as described previously (61). Transformants were plated on selective plates containing appropriate antibiotics and checked by PCR for integration and orientation of the erythromycin or kanamycin resistance cassette. All constructs were verified by Sanger sequencing.
Fluorescence measurements of gfp E. coli reporter strains and data processing. E. coli Top10 cells expressing gfp fusions were streaked on standard LB plates supplemented with appropriate antibiotics. After overnight growth, colonies were photographed in a Syngene Bio Image analyzer using a Lumenera camera with a 510-nm emission filter and excitation at 460 nm. Fluorescence measurements in 96-well plates were carried out as described previously (20). In brief, single colonies (in quadruplicate) of E. coli strains harboring a target-gfp fusion and sRNA-expressing plasmids were inoculated in 200 l LB in a 96-well microtiter plate, and cultures were grown at 37°C. The OD was measured at 600 nm in an enzyme-linked immunosorbent assay (ELISA) reader (Anthos Labtec), and fluorescence was measured (optical excitation filter, 485/20 nm; emission filter, 530/25 nm) in a CytoFluor II multiwell plate reader (PerSeptive Biosystems). The linear range of increasing fluorescence during growth covered by all members of a quadruplicate was selected to obtain the specific fluorescence. To calculate the specific fluorescence, the total fluorescence of a given strain expressing NmsR A and a target-gfp fusion gene (the mean fluorescence of the quadruplicate at a chosen time point within the linear range) was corrected for the autofluorescence measured in strains harboring an NmsR A expression plasmid or control sRNA in combination with the negative-control plasmid pXG-0 (expressing luciferase, i.e., no gfp). The regulatory effect of NmsR A on a target-gfp fusion was expressed as fold regulation (mean of the quadruplicate values). This is calculated by dividing the unregulated gfp fusion specific fluorescence (negative-control sRNA pJV300) by the regulated gfp fusion specific fluorescence (sRNA of interest) (20).
Construction of ⌬nmsR A ⌬nmsR B and ⌬relA mutants of N. meningitidis strain H44/76. N. meningitidis H44/76 knockout mutants of nmsR A and nmsR B and of relA (NMB1735) were constructed using the PCRligation-PCR method as described previously (62). PCR products were generated with primer pairs YPsRNA25FWKO1-YPsRNA25RPKO2 and YPsRNA25FWKO3-YPsRNA25RPKO4 to create the ΔnmsR A ΔnmsR B strain and primer pairs KSrelAF1-KSrelAR2 and KSrelAF6-KSrelAR7 to create the ΔrelA strain and ligated. The ligation products were reamplified with primer pair YPsRNA25FWKO1-YPsRNA25RPKO4 (for ⌬nmsR A ⌬nmsR B ) and primer pair KSrelAF1-KSrelAR7 (for ΔrelA). Resulting PCR products were cloned into pCR2.1 (Invitrogen). The EcoRI-digested Erm resistance cassette from pAErmC= was introduced into the created unique MfeI restriction site, yielding plasmids pCR2.1-sibling sRNA and pCR2.1-NMB1735 (relA) (62). The ΔnmsR A ΔnmsR B and ΔrelA knockout strains were generated by natural transformation of strain H44/76 with pCR2.1-NmsR A NmsR B and pCR2.1-NMB1735, respectively, with selection for Erm resistance. Replacement of NmsR A and NmsR B and NMB1735 by the Erm cassette was confirmed by PCR with primer pairs that were used for amplification of the ligation products and sequence analysis. Mutant strains in which the transcriptional direction of the Erm cassette was in accordance with the original transcriptional direction of the deleted genes sum of all proteins quantified for each individual sample. Subsequently, the average of the normalized femtomoles from the 4 biological replicates was calculated if detected in Ͼ1 biological replicate and used to calculate the fold change between ΔnmsR A ΔnmsR B and wt strains if a value for both strains was obtained (see Table S1 in the supplemental material). If a change was Ն1.5-fold up-or downregulated, a t test was performed to ascertain whether the change in protein expression was significant, i.e., whether it had a P value of Յ0.05 following adjustment for multiple testing according to the method of Benjamini and Hochberg (66). Proteins which were detected in only one of the two strains were reported as uniquely detected only if all of the quadruplicate injections of the biological replicates of one strain yielded quantitative data for this protein while the protein was not detected in any injections of the other strain. When a protein was detected in all injections from one strain and also in one of the replicate injections of the other, the value in normalized femtomoles is given for that single detection (i.e., no fold change value), as this would give the impression that the protein was consistently detected in both strains (Table 1).
Accession number(s). The mass spectrometry proteomics data have been deposited in the Proteome-Xchange Consortium (http://www.proteomexchange.org) via the PRIDE partner repository (67) with the data set identifier PXD000891.

ACKNOWLEDGMENTS
We thank R. B. Pouw and M. L. Ferrando for technical assistance. We acknowledge the PRIDE Team for their efforts in making the mass spectrometry data of this paper publicly available.
This publication made use of the PubMLST website (http://pubmlst.org/) developed by Keith Jolley (16) and hosted at the University of Oxford. The development of that website was funded by the Wellcome Trust.