The mtr (multiple transferable resistance) gene complex in Neisseria gonorrhoeae encodes an energy-dependent efflux pump system that is responsible for export of anti-bacterial hydrophobic agents. Expression of the mtrCDE operon in gonococci is negatively regulated by the MtrR protein. Hydrophobic agent resistance mediated by the mtr system is also inducible, which results from an AraC-like protein termed MtrA. In this work, we identified and characterized a pump similar to the gonococcal mtr system in various strains of Neisseria meningitidis. Unlike the situation with gonococci, the mtr system in meningococci is not subject to the MtrR or MtrA regulatory schemes. An analysis of the promoter region of the mtrCDE operon in a panel of meningococcal strains revealed the presence of one or two classes of insertion sequence elements. A 155–159 bp insertion sequence element known as the Correia element, previously identified elsewhere in the gonococcal and meningococcal genomes, was present in the mtrCDE promoter region of all meningococcal strains tested. In addition to the Correia element, a minority of strains had a tandemly linked, intact copy of IS1301. As described previously, a binding site for the integration host factor (IHF) was present at the centre of the Correia element upstream of mtrCDE genes. IHF was found to bind specifically to this site and deletion of the IHF binding site enhanced mtrC transcription. We also identified a post-transcriptional regulation of the mtrCDE transcript by cleavage in the inverted repeat of the Correia element, as previously described by Mazzone et al. [Gene278: 211–222 (2001)] and De Gregorio et al. [Biochim Biophys Acta 1576: 39–44 (2002)]for other Correia element. We conclude that the mtr efflux system in meningococci is subject to transcriptional regulation by IHF and post-transcriptional regulation by cleavage in the inverted repeat of the Correia element.
Infections by Neisseria gonorrhoeae and Neisseria meningitidis remain significant worldwide health problems despite advances in antibiotic treatments and the availability of a vaccine that protects against certain meningococcal capsular serogroups. The gonococcus typically causes lower and/or upper genitourinary tract disease, but other forms of infection (e.g. rectal, oral-pharyngeal, or invasive bloodstream infections) are not uncommon. In contrast, the meningococcus can be transiently carried in the nasopharynx where it can be transmitted through droplet nuclei to a susceptible individual and then enter the bloodstream or cross the blood brain barrier, resulting in sepsis and/or meningitis.
The capacity of both the gonococcus and the meningococcus to develop resistance to clinically useful antibiotics is of concern because this requires the use of more costly antibiotics, resulting in an economic burden in developing countries. Both specific and non-specific means of antibiotic resistance have been described as emerging problems in the pathogenic Neisseria (Yagupsky et al., 1993). Non-specific mechanisms of resistance are especially worrisome because they can often provide cross-resistance to structurally distinct anti-microbial agents. These agents include not only those antibiotics used in disease treatment and/or chemoprophylaxis, but also certain host-derived anti-microbial agents (free fatty acids, bile salts, steroidal hormones and anti-bacterial proteins or peptides) that frequently bathe mucosal surfaces and have presumed in vivo activity against invading pathogens. The ability of Gram-negative bacteria to resist the toxic action of such structurally diverse compounds often results from the synergistic effects of the impermeability barrier imposed by the outer membrane (OM) and the energy-dependent export of these agents as they traverse the OM (Nikaido, 1994). Energy-dependent efflux pumps can be classified in five families: the major facilitator (MF) family, the small multidrug resistance family, the resistance–nodulation–division (RND) family (Paulsen et al., 1996), the ATP-binding cassette (ABC) family (Nikaido and Hall, 1998) and the multidrug and toxic compound extrusion (MATE) family (Brown et al., 1999).
The mtr efflux system of the gonococcus belonging to the RND family has been the subject of several recent investigations (see below), and one report (Abadi et al., 1996) suggested its presence in N. meningitidis, a prediction recently verified by two independent genome sequencing projects (Parkhill et al., 2000; Tettelin et al., 2000). In gonococci, this efflux pump is composed of the MtrC–MtrD–MtrE proteins, which are encoded by the tandemly linked mtrCDE genes (Pan and Spratt, 1994; Hagman et al., 1995; 1997; Delahay et al., 1997); MtrC is the fusion membrane protein, MtrD is the transporter and MtrE, the OM protein, channel of the pump. Loss of one or more of these proteins as a result of mutations results in hypersusceptibility of gonococci to structurally diverse hydrophobic agents including those with a non-ionic detergent-like activity (e.g. free fatty acids, bile salts, steroidal hormones), antibiotics and anti-bacterial peptides (Lucas et al., 1995; Shafer et al., 1998; Veal et al., 1998).
Transcription of the mtrCDE operon in gonococci is negatively regulated by the product of the mtrR gene (Pan and Spratt, 1994; Hagman et al., 1995; Shafer et al., 1995), which is positioned 250 bp upstream and divergently transcribed from mtrCDE. The MtrR repressor binds (Lucas et al., 1997) to a DNA sequence that overlaps the promoter used for mtrCDE gene transcription (Hagman and Shafer, 1995). Missense mutations that cause radical amino acid replacements in the helix–turn–helix (HTH) motif or in a downstream region in the MtrR repressor are common in gonococcal clinical isolates that express elevated levels of resistance to hydrophobic agents. Recently, we obtained evidence (Rouquette et al., 1999) that the mtrCDE gene complex in gonococci was inducible during growth in sublethal concentrations of Triton X-100 (TX-100) and that such cultivated gonococci expressed elevated resistance to hydrophobic agents. This inducible resistance was dependent of an intact mtrCDE gene complex and a transcriptional activator protein (MtrA) that belongs to the AraC/XylS family (Gallegos et al., 1997).
Rifampin (Rif) is a hydrophobic antibiotic recognized by the mtrCDE-encoded efflux pump in gonococci (Table 1; Shafer et al., 1984). It has been the principal antibiotic used in prophylactic treatment of individuals having close contact with an index case of meningococcal disease. Rif-resistant (RifR) strains of the meningococcus have been reported (Yagupsky et al., 1993) and Abadi et al. (1996) showed that such resistance could result from a mechanism dependent on mutations in the rpoB gene. They also examined whether the mtr system might be involved in RifR. They found that the promoter region of the mtrR gene from RifR mutants of two clinical isolates contained an additional 158 bp sequence that was absent in the corresponding gonococcal sequence. Additionally, compared with the MtrR protein of gonococcal strain FA19, which is susceptible to Rif, the meningococcal proteins produced by isogenic Rif-sensitive (RifS) and RifR strains were highly homologous (97% reported identity), but indistinguishable from each other. Interestingly, Abadi et al. (1996) found that the MtrR protein possessed by their meningococcal strains would have a His-105 to Tyr replacement, which in gonococci results in resistance to hydrophobic agents (Pan and Spratt, 1994).
Table 1. Hydrophobic agent (HA) susceptibility of gonococcal and meningococcal strains.
We now report that meningococci have a functional mtrCDE-encoded efflux system but its expression is independent of MtrR or MtrA. Instead, we noted the presence of a 155–159 bp Correia element, previously found at multiple sites in the gonococcal and meningococcal chromosomes (Correia et al., 1986; 1988), uniquely placed immediately downstream of the mtrC promoter. This was an interesting finding because Black et al. (1995) showed that the insertion of a Correia element upstream of the uvrB gene created a σ70 promoter. Correia element is small insertion sequences (IS) (≈ 100–155 bp in length), which have long-terminal inverted repeats (≈ 26–27 bp) (Mazzone et al., 2001) and a TA target site duplication (Mazzone et al., 2001; Buisine et al., 2002). Mazzone et al. (2001) classified Correia element into four distinct subfamilies (26L/26R, 27L/27R, 26L/27R and 27L/26R) depending on the sequence (L/R) and length of their inverted repeats. Mazzone et al. (2001) and De Gregorio et al. (2002) showed that transcripts that are read through a Correia element are cleaved by ribonuclease III, either one or twice, within the inverted repeats. Buisine et al. (2002) reported the presence of an integration host factor (IHF) binding site at the centre of Correia element that may play a role in modulating gene expression. IHF is a site-specific DNA-binding protein, which, upon interaction with its recognition site, induces a significant bend in the DNA. In Escherichia coli, IHF acts as a helper protein in a wide range of processes including DNA replication, site-specific recombination and transcription (Goosen and van de Putte, 1995). In gonococci, IHF has been shown to bind proximal of the three pilE promoters and function as a transcriptional co-activator (Hill et al., 1997). Hill et al. (1998) also showed that the ihf mRNA levels decline as N. gonorrhoeae enters the stationary growth phase. We propose that the Correia element has a negative impact on mtrC transcription in meningococci and that this negative regulation is mediated by IHF.
Meningococci encode a functional mtr efflux pump system
Abadi et al. (1996) provided evidence for the existence of the mtrR and mtrC genes in meningococci. However, they did not report the presence of other mtr genes (mtrD or mtrE) in their test strain that would be needed to form a functional efflux pump or whether regulation of the efflux pump genes was similar to that described for gonococci (Hagman and Shafer, 1995; Lucas et al., 1997; Rouquette et al., 1999). In order to determine whether meningococci have the genetic capacity to encode a functional mtr efflux pump, we initially asked whether in addition to mtrC, meningococci contain homologues of the gonococcal mtrD and mtrE genes, which encode the membrane transporter and OM channel protein components of the mtr efflux pump respectively. The mtr efflux pump gene complex possessed by a capsular serogroup B strain (NMB) and two serogroup Y strains (2633 and 0929) was determined using gene-specific oligonucleotide primers (designed on the sequences of the gonococcal mtrC, mtrD and mtrE genes) in polymerase chain reactions (PCRs). For all three meningococcal strains, we were able to obtain PCR products for all three mtr genes that were electrophoretically indistinguishable from their counterparts possessed by N. gonorrhoeae strain FA19 (data not presented). Moreover, using combinations of oligonucleotides that anneal within the mtrCDE genes, we were able to ascertain that the order of these three genes is identical to that of gonococci (Fig. 1A) (Hagman et al., 1995; 1997; Delahay et al., 1997), an observation that is consistent with the recently published genome sequences for N. meningitidis serogroup A strain Z2491 (Parkhill et al., 2000) and serogroup B strain MC58 (Tettelin et al., 2000). A comparison of the mtr genes from gonococcal strain FA1090 with the mtr genes from meningococcal strain MC58 showed 94%, 95.7% and 94% identity for mtrC, mtrD and mtrE genes respectively. The amino acid sequence for the MtrC, MtrD and MtrE proteins from gonococcal strains FA1090 are, respectively, 95.8%, 97.6% and 96.3% identical to their conterpart proteins from meningococcal strain MC58. We also compared the MtrC, MtrD and MtrE proteins from gonococal strain FA1090 with that of gonococcal strain FA19 and found that they were, respectively, 99.5%, 98.9% and 95.2% identical.
Although the PCR data indicated that meningococci harbour an mtrCDE gene complex similar to that of gonococci, it did not address whether the encoded proteins would form a functional efflux pump. We reasoned that if the pump was functional, then insertional inactivation of the gene complex should enhance the susceptibility of meningococci to hydrophobic agents in a manner similar to that observed previously in gonococci (Table 1; Pan and Spratt, 1994; Hagman et al., 1995, 1997; Delahay et al., 1997). Accordingly, we introduced by transformation the mtrC::Km sequence from gonococcal strain KH12 (as FA19 but mtrC::Km) into the test meningococcal strains; insertional inactivation of the mtrC gene in selected transformants was confirmed by PCR (data not presented). Selected transformants from each meningococcal parental strain were tested for their susceptibility to hydrophobic agents [crystal violet (CV), erythromycin (Ery), Rif and TX-100]. Insertional mutants of all three strains displayed an enhanced susceptibility to TX-100, CV, Ery and Rif (Table 1). Taken together with the PCR data, we concluded that the test meningococcal strains possessed an intact and functional MtrC–MtrD–MtrE efflux pump system.
Expression of mtrCDE in meningococci is independent of MtrR and MtrA
After we determined that the test meningococcal strains could produce a functional MtrCDE efflux pump, we investigated whether the mtrCDE operon was subject to MtrR- and MtrA-mediated regulation as previously reported for gonococci (Pan and Spratt, 1994; Hagman et al., 1995; Rouquette et al., 1999). Interestingly, the MtrR repressor that would be produced by the meningococcal strain studied by Abadi et al. (1996) was predicted to have the His-105 to Tyr-105 substitution that was previously shown to abrogate MtrR function in gonococci (Pan and Spratt, 1994; Hagman et al., 1995), resulting in increased resistance to hydrophobic agents. An inspection of the mtrR nucleotide sequence for meningococcal strains Z2491 (Parkhill et al., 2000) and MC58 (Tettelin et al., 2000) revealed the presence of the missense mutation at codon 105 that would cause the His-105 to Tyr-105 substitution as well. In order to learn the nature of the MtrR protein produced by other meningococcal strains, we sequenced the PCR-amplified mtrR gene from strains NMB (serogroup B), 2633 (serogroup Y) and 0929 (serogroup Y) and deduced the amino acid sequence for their respective MtrR protein. All three MtrR proteins differed substantially from that of the MtrR protein of gonococcal strain FA19 (Fig. 2). Thus, the mtrR-coding region from strain NMB (GenBank Accession No. AY490994) contained five missense mutations that would cause amino acid replacements at positions 77, 78, 86, 105 and 108 (Fig. 2). The coding region of the mtrR gene possessed by meningococcal strains 2633 (GenBank Accession No. AY490995) and 0929 (GenBank Accession No. AY490996) contained the His-105 to Tyr-105 substitution and nonsense mutations that would result in the production of truncated MtrR proteins of 135 and 150 amino acids respectively (Fig. 2).
Because a full-length MtrR protein is needed for regulation of mtrCDE in gonococci (Hagman and Shafer, 1995; Shafer et al., 1995; see Table 1), we hypothesized that the MtrR protein produced by strains 2633 and 0929 would be inactive. To confirm that the MtrR protein did not negatively regulate mtrCDE in meningococci, we inserted a Km cassette in the mtrR gene of strain NMB, the only studied isolate that was predicted to produce a full-length MtrR protein. We evaluated its susceptibilities to hydrophobic agents (TX-100, CV, Ery and Rif) exported by the mtrCDE-encoded efflux pump. We did not detect any difference in levels of susceptibilities between NMBmtrR::Km and its parental strains NMB (Table 1), suggesting that the MtrR protein encoded by strain NMB was indeed non-functional. Taken together, these cumulative results indicated that the presence of mutations that would typically abrogate MtrR function in gonococci was a common property of N. meningitidis isolates.
We previously observed that resistance to hydrophobic agents mediated by the mtrCDE operon in gonococci was inducible during growth in sublethal concentrations of TX-100 and that this induction was dependent on a transcriptional activator protein termed MtrA (Rouquette et al., 1999). Accordingly, we examined the MIC of TX-100 against meningococcal strains NMB, 2633 and 0929 with or without induction (overnight growth in the presence of 20 µg ml−1 TX-100). In both cases, the MIC of TX-100 was 50 µg ml−1 (data not presented). We next monitored the presence of MtrC membrane fusion lipoprotein in cultures grown in the absence or the presence of 20 µg ml−1 TX-100 by Western blot analysis. Contrary to the results obtained with the gonococcal strains FA19 (Rouquette et al., 1999), we could not detect any difference in MtrC production in these two conditions in the meningococcal strains (data not presented).
We considered the possibilities that the lack of induction in these meningococcal strains could result from the lack or from the presence of deletion mutations in the mtrA gene that would abrogate MtrA function, as observed previously (Rouquette et al., 1999) with non-inducible isolates of N. gonorrhoeae, or that the gene was not transcribed. We were able to obtain by PCR a full-length mtrA gene from these strains and DNA sequence analysis of the mtrA gene of strains NMB, 2633 and 0929 (GenBank Accession No. AY490997, AY490998 and AY490999 respectively) indicated that they would encode a full-length MtrA protein, which were 94.4% identical to the 301 amino acid MtrA protein of gonococcal strain FA19 (data not presented). It is important to note that the amino acid differences were outside of the AraC/XylS family consensus sequence (Gallegos et al., 1997), suggesting that they would unlikely impact MtrA function. We also eliminated the possibility that the mtrA gene was not transcribed in the test meningococcal strains because reverse transcription polymerase chain reaction (RT-PCR) analysis of total RNA permitted the detection of an mtrA-specific transcript (data not presented). Thus, we concluded that while the meningococcal mtrCDE operon would encode a functional MtrC–MtrD–MtrE efflux pump, the gene complex was not inducible as previously noted for gonococci (Rouquette et al., 1999).
Presence of insertion sequences between the mtrR and mtrC genes in meningococci
Transcriptional regulation of genes encoding efflux pumps proteins appears to be a common property in bacteria (Grkovic et al., 2002). Because the expression of mtrCDE operon in meningococci was independent of the MtrR and MtrA proteins, we therefore sought to determine whether this operon was regulated by other mechanisms. Because we noted that the PCR products encompassing the intergenic region between the mtrC and the mtrR genes, obtained with genomic DNA from strains NMB and 2633, were at least 150 bp larger than the corresponding product obtained with gonococcal DNA from strain FA19, we thought that the meningococcal DNA sequence within this 150 bp region might contain regulatory elements. Accordingly, we determined the nucleotide sequence of the intervening region between mtrR and mtrC genes in the test meningococcal strains and found it to contain a 155 bp (strain NMB) and 159 bp (strain 2633) insertion that corresponded to the previously described Correia element (Correia et al., 1986, 1988), which was also present in the mtrR/mtrC intervening region in the meningococcal strain studied by Abadi et al. (1996). The Correia element was positioned just downstream (Fig. 1A) of the promoter that drives transcription of mtrC in gonococci (Rouquette et al., 1999; Fig. 1B) and meningococci (see below). Importantly, the sequence of the Correia element contained a putative IHF binding site (Buisine et al., 2002). PCR amplification of the mtrR/mtrC intergenic region in strain 0929 generated a nearly 1.2 kb product. DNA sequence analysis of this product revealed the presence of not only a Correia element, but also an intact copy of IS1301 (Hilse et al., 1996), which was positioned between the Correia element and mtrCDE genes (Fig. 1A).
We had not previously observed either the Correia element or IS1301 elements in the mtrR/mtrCDE intergenic region in over 30 gonococcal strains studied to date (Hagman et al., 1995; Shafer et al., 1995; Veal et al., 1998; Zarantonelli et al., 1999). One report by Johnson et al. (2003) identified gonococcal isolates from Kansas City, MO, USA, harbouring a Correia element upstream of the mtrC gene. However, it seems to be a very rare phenomenon in gonococci. We suspected that their presence in this region might be common with meningococcal isolates. Accordingly, we examined 10 additional meningococcal clinical isolates representing capsular serogroups A, B, C, Y and W-135, obtained from various sites (e.g. bloodstream, genitourinary tract and rectum). Using chromosomal DNA from these strains and the PCR strategy described above, we found that the PCR products from these 10 clinical isolates were electrophoretically indistinguishable (460 bp) from those of strains NMB and 2633 (data not presented), suggesting that they also contained a Correia element within this region. Furthermore, analysis of the meningococcal genome sequence database from strains Z2491 (Parkhill et al., 2000) and MC58 (Tettelin et al., 2000) revealed that both contain a Correia element upstream of the mtrCDE operon. Thus, we concluded that the presence of a Correia element within the mtrR/mtrC region is a common trait in meningococcal isolates, while the presence of an IS1301 occurs very rarely. We did not observe any difference of susceptibility to hydrophobic agents between meningococcal strains harbouring only a Correia element versus meningococcal strains harbouring an IS1301 in addition to the Correia element in the mtrC promoter region (Table 1). Consequently, we decided to focus our work on the possible impact of the Correia element and, more precisely, of the IHF binding site located at the centre of the Correia element, on the mtrC expression and not to study the IS1301.
Regulation of mtrCDE expression in meningococci
Because the Correia element was positioned downstream of the promoter that would drive transcription of mtrCDE in gonococci (Fig. 1A) (Hagman et al., 1995), we hypothesized that it could modify mtrCDE transcription in meningococci. We used RT-PCR to initially map the mtrC promoter region in meningococci. Using RNA from meningococcal strains NMB (Correia element only), 2633 (Correia element only) and 0929 (Correia element and IS1301), we asked whether an amplification product could be obtained using an oligonucleotide primer that anneals within mtrC in the reverse transcription reaction and primers that anneal at the 5′ end of the Correia element and in the mtrC gene in the PCR reaction. We reasoned that if a product could be generated under these conditions, then the promoter that drives transcription of mtrC in meningococci would probably map upstream of the Correia element. Indeed, under these conditions we obtained RT-PCR products of the expected size with RNA from all three strains (data not presented). As this result suggested the presence of a promoter sequence upstream of the Correia element, we next used primer extension analysis to identify the mtrC transcriptional start point. Using oligonucleotide primer CR3 or CR1 (see Experimental procedures), we were able to detect an extension product (data not presented) with RNA from the three strains that corresponded to an A residue located upstream of the Correia element (Fig. 1B). This transcription start point was identical to that observed previously (Rouquette et al., 1999) for mtrC gene transcription in gonococci (Fig. 1B). A 29-nucleotide sequence containing near consensus −10 and −35 hexamers, separated by 17 nucleotides, was identified seven nucleotides upstream of the transcription start point. We hypothesized that this promoter (termed P1) was probably used for mtrC gene transcription in meningococci. We also identified two additional signals corresponding to smaller extension products (Fig. 3). Mazzone et al. (2001) and De Gregorio et al. (2002) described the presence of RNase III cleavage sites termed p and d, located in the inverted repeats of the Correia element. Correia elements were classified in three distinct families depending on the sequence of their inverted repeats: 26R/26L, 26L/26R and 27L/27R. Our DNA sequence analysis revealed that the Correia element upstream of the mtrC gene in strain NMB belonged to the 26L/26R family and confirmed the presence of the p and d cleavage sites (Fig. 3).
As described by Buisine et al. (2002) for other Correia element, the Correia element upstream of the mtrC gene in meningococci contained a putative IHF binding site at its centre. In order to evaluate the impact of the Correia element on mtrC transcription in meningococci, we constructed chromosomal β-galactosidase fusions of the mtrC promoters using the pLES94 vector (Silver and Clark, 1995) integrated at the chromosomal proAB site of strain NMB. The DNA fragments cloned upstream of the lacZ gene are presented in Fig. 1A. These constructs contained the entire mtrC promoter region encompassing the Correia element (PNMB), the same fragment but with the Correia element deleted (ΔCE), the mtrC promoter region but with the IHF binding site replaced by an EcoR1 restriction site (ΔIHF) so as to maintain the length of the Correia element, the mtrC promoter region but with an EcoR1 restriction site instead of the left inverted repeat of the Correia element (ΔIR) so as to maintain the length of the Correia element and finally the mtrC promoter region with the promoter P1 deleted. We found that strain NMB containing the two deleted fragments (ΔCE and ΔIR) had a level of β-galactosidase activity that was lower (Fig. 4) than the strain with the PNMB construct (wild-type meningococcal promoter of mtrC). In contrast, the strain with the ΔIHF construct had a level of β-galactosidase activity that was more than twice higher than the level observed with the PNMB construct. This result suggested that the presence of the IHF binding site had a negative impact on mtrC transcription (Fig. 4) and that the left inverted repeat is important for the stabilization of the mtrC mRNA. To make sure that P1 was the only promoter used to drive mtrC transcription in meningococci, we deleted it and cloned the remaining mtrC upstream region into the pLES94 vector. Because the transformant bearing this fragment (ΔP1) lacked β-galactosidase activity, we concluded that P1 was indeed the promoter used in meningococci for mtrCDE transcription (data not presented).
Binding of IHF to the Correia element
Because deletion of the IHF binding site within the Correia element enhanced mtrCDE expression, we investigated whether IHF could in fact specifically bind to this site. We performed electrophoretic mobility shift assays with an IHF protein preparation purified from gonococci (Hill et al., 1997). The Correia element located downstream of the mtrC promoter of strain NMB and the same fragment with the IHF binding site deleted were the target sequences for IHF binding. Electrophoretic mobility shift assays using these two probes showed a binding of IHF protein to the Correia element but not to the Correia element deleted for the IHF binding site (Fig. 5). To make sure that this binding was specific, we performed competition electrophoretic mobility shift assays using as the labelled DNA probe a PCR fragment encompassing the Correia element harbouring the IHF binding site, and as a competitor the same unlabelled DNA fragment. The results showed that the binding of the IHF protein was decreased in the presence of the specific competitor DNA (data not presented). In contrast, when we used the DNA fragment encompassing the Correia element but with the IHF binding site deleted as a competitor, we did not observe any decrease of the shifting by IHF protein (data not presented). Taken together, these results showed that the IHF binding site at the centre of the Correia element located upstream of the mtrC gene in meningococci was indeed functional because it could specifically bind the IHF protein and its replacement by an EcoR1 restriction site decreased mtrC transcription.
The results presented here indicate that although meningococci contain an mtrCDE gene complex (Fig. 1A) that encodes a functional MtrC–MtrD–MtrE efflux pump, the regulation of its expression is different from that observed previously for the related pathogen N. gonorrhoeae. Specifically, while there is clear evidence that the MtrR protein acts as a negative transcriptional regulator in gonococci (Hagman and Shafer, 1995; Lucas et al., 1997), it appears from our data that the production of an inactive MtrR repressor is a common trait among meningococci. Thus, the three meningococcal strains used in our investigation, the isolate studied by Abadi et al. (1996) and the sequences from strains MC58 and Z2491 all contained missense or nonsense mutations in their mtrR gene that would abrogate its function in gonococci, typically resulting in elevated resistance to hydrophobic agents in this pathogen (Pan and Spratt, 1994; Hagman et al., 1995; Shafer et al., 1995). We also determined that expression of mtrCDE in our test meningococcal strains was not inducible as previously shown for gonococci (Rouquette et al., 1999). Several hypotheses can be advanced as to explain the lack of induction of the meningococcal mtrCDE operon. The simplest would be that a protein(s) required for induction in gonococci are missing in meningococci. Another explanation could be that, in meningococci, a regulatory protein that functions in gonococci could not bind upstream of the gene(s) that it normally regulates. The binding of this regulatory protein could be negatively impact by mutations in the binding site or in the case of the mtrC gene, because of the insertion of the Correia element.
Primer extension analysis and β-galactosidase assays showed that the same promoter (P1) was used in gonococci and meningococci to drive transcription of the mtrCDE operon. We were also able to show by primer extension that the majority of mtrC transcripts were cleaved in both inverted repeats of the Correia element in meningococci (Fig. 3). Mazzone et al. (2001) and De Gregorio et al. (2002) suggested that these cleavages would stabilize the mRNA; this would explain why the β-galactosidase activities of the ΔCE and the ΔIR constructs were lower than that of the PNMB construct.
Inspection of the DNA sequence of the Correia element inserted upstream of the mtrC gene in strain NMB revealed the presence of a putative IHF binding site. We determined that this site was functional for IHF binding in vitro (Fig. 5) and it negatively regulated mtrCDE expression in vivo (Fig. 4). It was necessary to use the procedures described here to test the influence of IHF in regulating mtrCDE in meningococci because we and others (J. Davies, pers. comm.) have been unable to isolate IHF-deficient mutants of gonococci and meningococci. IHF is known to bend DNA (Thompson and Landy, 1988), so a putative model for this negative regulation of mtrCDE expression could be that after the binding of IHF at the centre of the Correia element, the inverted repeats of the Correia element come in contact and a DNA secondary structure is formed. This loop could act as a weak terminator of transcription. An alternative hypothesis is that the binding of IHF prevents the progression of the RNA polymerase by itself or in conjunction with another regulator and therefore decreases transcription. We noticed (by primer extension and β-galactosidase assays; data not shown) that the level of mtrC transcription is much more higher in exponential phase than in stationary phase. Consequently, we determined the levels of himA and himB transcripts (genes encoding IHF) in exponential and stationary phases but we could not detect any difference of expression that correlated with growth phase (data not presented).
Taken together, our results indicate that the negative and positive transcriptional regulatory processes that modulate mtrCDE gene expression in gonococci are ineffectual in meningococci. However, IHF, via a binding site located at the centre of a Correia element inserted upstream of the mtrCDE genes, seemed to have a negative impact on mtrC transcription. In addition, mtrCDE transcripts could be stabilized by cleavage in the inverted repeats of the Correia element. N. meningitidis and N. gonorrhoeae multiply and colonize very different environments which may differ in the type and amount of natural anti-microbials recognized by the pump so they probably require different regulations of their respective mtrCDE operons. To our knowledge, this is the first indication that transcriptional regulation by IHF and a post-transcriptional regulatory process have been shown to be involved in efflux pump gene expression in bacteria.
Both gonococci and meningococci harbour multiple copies of the Correia element in their respective genomes (Correia et al., 1986; 1988). As suggested by Mazzone et al. (2001), De Gregorio et al. (2002) and our present work, RNase III could be a global regulator in Neisseria. Transcriptional regulation mediated by IHF resulting from the insertion of a Correia element could also be a common phenomenon in Neisseriae. Insertions of Correia elements could be responsible for different mechanisms of regulations of homologous genes in meningococci and gonococci. With respect to the mtr efflux system in meningococci, the transcriptional regulatory process imposed by IHF probably counteracts the impact of mutations within mtrR, which in gonococci result in high-level resistance to multiple anti-microbial agents. This may explain why meningoccocal isolates expressing such resistance have not been identified.
Strains used, growth conditions and determination of susceptibility to anti-microbial agents
Isogenic N. gonorrhoeae strains FA19, KH11 (ΔmtrR) and KH12 (mtrC::Km), used in this study, have been previously described (Hagman et al., 1995). N. meningitidis strains NMB, 2633 and 0929 were provided by D. Stephens (Emory University School of Medicine, Atlanta, USA), while other clinical isolates were provided by C. Ison (St. Mary's Hospital, London, UK). Strains were grown overnight at 37°C under 3.8% (v/v) CO2 on GCB agar containing defined supplements I and II (Shafer et al., 1984) with or without sublethal concentrations of TX-100. The MICs of TX-100, Ery, CV and Rif were performed as described previously (Shafer et al., 1984); all of these agents were purchased from Sigma Chemical.
PCR analysis of the mtrRCDE locus in meningococci
Using genomic DNA prepared by the method of McAllister and Stephens (1993), the mtrR, mtrC, mtrD and mtrE genes in meningococcal strains 0929, 2633, NMB and gonococcal strain FA19 were PCR amplified using primers CEL3B (5′-GCCTGTTGATTGGCAAAACCC-3′), KH9♯4 (5′-CATCACT TGTGCGGACGCACC-3′), KH9♯15 (5′-CGTTCAGACGG CATCTGAAGC-3′), mtrE1 (5′-GATGGAAGAAACCGATGT GTCG-3′), mtrE2 (5′-CCTTTGCATTGTCTGCCTGCAC-3′) and mtrE3 (5′-CGGTTTGGGTATCCCGTTTCAATC-3′), respectively, as described previously (Hagman et al., 1995). The intergenic region between mtrC and mtrR was PCR amplified using primers KH9♯3 (5′-GACGACAGTGCCAATG CAACG-3′) and KH9♯2 (5′-CGTTTCGGGTCGGTTTGACG-3′).
Construction of the mutants
Genomic DNA from gonococcal strain KH12 (as FA19 but mtrC::Km; Hagman et al., 1995) was transformed into meningococcal strains NMB, 2633 and 0929. Transformants were selected on BHI agar containing 100 µg ml−1 Km and 2.5% (v/v) fetal calf serum. The insertion of the Km cassette into mtrC was verified by PCR using primers KH9♯11B (5′-GGTACGCAGCGATTCCAAACG-3′) and CEL2B (5′-CGAA CATTCGGGTATCAAAGC-3′). The mtrR gene from NMB was amplified using primers CEL1 (5′-GACAATGTTCATGCGAT GATAGG-3′) and KH9♯11B. This PCR fragment was cloned into pBAD (Invitrogen). A non-polar Km cassette (Ménard et al., 1993) was then inserted into the Nae1 restriction site of mtrR. The resulting construct was tranformed into NMB. The transformants were selected on BHI agar plates supplemented with 2.5% (v/v) fetal calf serum and 100 µg ml−1 Km. The insertion of the Km cassette was identified by PCR.
Analysis of the mtrC/mtrR intergenic region and DNA sequencing of the mtrR and mtrA genes
Oligonucleotides KH9♯2 and KH9♯3 (see above) were used to amplify the intergenic regions between mtrC and mtrR genes of meningococcal clinical isolates. The resulting PCR products were analysed by agarose electrophoresis. Oligonucleotides CEL1 and KH9♯3 were used to amplify a DNA fragment of genomic DNA from strains 2633, NMB and 0929. The resulting PCR products were manually sequenced using the cycle sequencing procedure (Hagman et al., 1995). Primers KH9♯2 and KH9♯3 were used to amplify the mtrC/mtrR intergenic region of the meningococcal strains 2633, NMB and 0929, and the resulting PCR products were manually sequenced. Oligonucleotides C4 (5′-GACGCGGTACGCA AATGAAG-3′) and C15 (5′-AATGCCGTCGGTAAGGTGC-3′), based on the FA19 genomic sequence (Rouquette et al., 1999), were used to amplify a 1640 bp region from 2633, 0929 and NMB genomic DNA. The resulting PCR product was manually sequenced using the cycle sequencing procedure (Hagman et al., 1995).
RT-PCR and primer extension analysis
Total RNA was prepared from broth cultures of meningococci by the method of Baker and Yanofsky (1968). RNA concentrations were determined by UV spectrometry at 260 nm. RT-PCR analysis of mtrC was performed using RNA from meningococcal strains 2633 and NMB. Primer KH9♯13B, which anneals within mtrC, was used for the reverse tran scription reactions. Primers KH9♯13B and CE2 (5′-GCTCTA GAGCAAGGCGACGAAGCCGCAG-3′), which anneals at the left end of the Correia element, were used for the PCR reaction. Oligonucleotide C2 (5′-GGCGAAACGTGATAT TGCCG-3′) was used for the reverse transcription of mtrA. PCR was then performed, after the RT reaction, with primers C2 and C1 (5′-TCGATCTCGCCCAATTGACG-3′). RT-PCRs were performed as described by Wang et al. (1989).
Primer extensions were performed as described previously (Hagman and Shafer, 1995) on 20 µg of total RNA with primers CR3 (5′-GGGTTAAAATAGGGCTCTTTTGGGCTTATTC-3′) and CR1 (5′-GCCATTATTTATCCTATCTGTCTGGTTT GATG-3′) (for mtrC), ihf4 (5′-ACGTTGTTTGTCATAGTAGTG-3′) (for himA) or ihf7 (5′-CGGCACCACCAGCTTGGG-3′) (for himD). The AMV Reverse Transcriptase Primer Extension System from Promega was used as described by the manufacturer. The transcription start point was determined by electrophoresis of the extension products on a 6% (w/v) DNA sequencing acrylamide gel adjacent to reference sequencing reactions.
The mtrC promoter region from strain NMB was amplified with PmtrC1 (5′-CGGGATCCCGAGCCATTATTTATCCTATC TG-3′) and PmtrC2 (5′-CGGGATCCCGCGTTTTCGTTT CGGGTCGGTTTGACG-3′) primers. PCR with primers PmtrC1, PmtrC2, PmtrC3 (5′-CGGAATTCCGATTTCAG GATATAAAAACCGCCT-3′) and PmtrC4 (5′-CGGAATTCC GAGAATGCCTGTTGATTGGCAA-3′) were performed on NMB genomic DNA to create a deletion of the Correia element in the mtrC promoter of NMB. PCR with primers PmtrC1, PmtrC2, PC6 (5′-CCGGAATTCCGGGGAGAATCG TTCTCTTTGAGCTA-3′) and PC7 (5′-CCCGGAATTCCGGC ACCAAGTGAATCGGTTCCG-3′) were performed on NMB genomic DNA to delete the IHF binding site at the centre of the Correia element located on the mtrC promoter region. PCR was performed on NMB genomic DNA with primers PmtrC1, PmtrC2, PmtrC6 (5′-CCCCGGAATTCCGGCCCAG GACAAGGCGACGAAGCCG-3′) and PmtrC7 (5′-GGCCGG AATTCCGGGGTATATTTCAGGATATAAAAACC-3′) to delete the left inverted repeat of the Correia element and replace it by an EcoR1 restriction site on the mtrC promoter region. PCR was performed on NMB genomic DNA with primers PmtrC5 (5′-CGGGATCCCGGAAATTGAGACTACATCTCA AC-3′) and PmtrC1 to eliminate the P1 promoter upstream of mtrC. The resulting PCR fragments (Fig. 1) were cloned into the BamH1 site of the pLES94 vector (Silver and Clark, 1995). All the lacZ fusions were then integrated at the chromosomal proAB site of strain NMB by transformation and controlled by sequencing. β-Galactosidase assays were performed as described by Snyder et al. (2003) after growth overnight on GC agar plates supplemented with 1 µg ml−1 chloramphenicol or in GCB broth supplemented with supplements I and II and with 1 µg ml−1 chloramphenicol, at 37°C with agitation.
Electrophoretic mobility shift assays
The Correia element (with or without the IHF binding site) located upstream of the mtrC gene in strain NMB was amplified by PCR using primers CE1 (5′-GCTCTAGAGCGGCGT TGCCTCGCCTTAGC-3′) and CE2 (5′-GCTCTAGAGCAAG GCGACGAAGCCGCAG-3′) and the pLES94 constructs PNMB and ΔIHF as templates. The resulting PCR products were end-labelled with [γ-32P]-d-ATP using T4 polynucleotide kinase (New England Biolabs). The labelled DNA fragment (40 ng) was incubated with 1 µg of IHF in 30 µl of reaction buffer [10 nM Tris-Hcl (pH 7.5), 0.5 mM DTT, 0.5 mM EDTA, 4% (v/v) glycerol, 1 mM MgCl2, 50 mM NaCl, poly-(dI-dC) (0.5 µg ml−1), salmon sperm (200 µg ml−1)] at room temperature for 20 min. For the competition assays, 250 ng of the same non-labelled DNA fragments were added in the reaction. Samples were subjected to electrophoresis in a 6% native polyacrylamide gel at 4°C followed by autoradiography.
Western immunoblot analysis
Whole-cell lysates were prepared from strains 2633 and 0929 grown on GCB agar plates containing 0 or 20 µg ml−1 TX-100. Western immunoblotting and detection of MtrC, using an anti-MtrC monoclonal antibody (Judd et al., 1991), was performed as described by Hagman et al. (1995).
We are grateful to T. Hrobowski and J. Harmon for technical assistance, L. Pucko for help in manuscript preparation, R.C. Judd for the gift of the anti-MtrC monoclonal antibody, C. Ison and D. Stephens for the gift of meningococcal clinical isolates. This work was supported by PHS Grants AI-21150 (W. Shafer) from the NIH. W. Shafer is the recipient of a Senior Career Scientist Award from the VA Medical Research Service.