Identification and characterization of the farAB efflux system in gonococci
The mtrCDE-encoded efflux pump has been shown to mediate energy-dependent export of structurally diverse hydrophobic antimicrobial agents (Lucas et al., 1995; Shafer et al., 1998), and mutations in these genes impact levels of gonococcal susceptibility to FAs (Delahay et al., 1997; Hagman et al., 1997). To date, this is the sole efflux pump reported in gonococci. However, an mtrCDE-independent mechanism of resistance to a subset of hydrophobic agents, long-chained FAs, was originally proposed by McFarland et al. (1983), but the nature of this resistance has not been described before. Using the online (www.genome.ou. edu) genomic sequence data for N. gonorrhoeae strain FA 1090, it was possible to search for additional efflux pumps. Through such a search, the presence of an open reading frame (ORF) that would encode a protein similar to EmrB of E. coli was identified (D. Dyer, personal communication). EmrB is the cytoplasmic membrane-bound transporter protein (Lomovskaya and Lewis, 1992) that also contains EmrA as the membrane fusion protein component of an efflux pump in E. coli and TolC as the outer membrane protein channel for export of uncoupling agents and hydrophobic antibiotics (Nikaido, 1996). We successfully isolated from a λZAPII library of FA 19 chromosomal DNA (Hagman et al., 1995) an emrB-like ORF. With a series of polymerase chain reaction (PCR) amplification reactions that used oligonucleotides synthesized on the basis of the FA 1090 genome sequence information, we ultimately identified and sequenced a locus from strain FA 19 that contained two tandemly linked ORFs (Fig. 1) that would encode proteins similar to those expressed by the emrAB operon in E. coli (see below). These two ORFs were separated by 23 bp. A likely ribosome binding site (AAGAA) was noted 11 bp upstream of orf1 (the emrA homologue), and a putative promoter region containing near consensus −10 and −35 hexamers (TAAAAT and TTGATT respectively) was located 30 bp upstream of orf1. In addition, a third ORF (orf3 ) that would encode a 17.3 kDa protein was positioned upstream of orf1 but on the opposite strand. The putative protein encoded by orf3 was not similar to the EmrR transcriptional regulatory protein in E. coli, which is encoded by a gene just upstream of emrAB (Lomovskaya and Lewis, 1992), nor did it resemble other proteins in the databases (data not presented).
Figure 1. . Organization of the farAB operon in N. gonorrhoeae strain FA 19. The order and transcriptional orientation of the farAB locus are shown. The insertion site of KmR cassette in the farB gene in strain EL-I is indicated.
Download figure to PowerPoint
An analysis (data not presented) of the predicted amino acid sequences of the orf1- and orf2-encoded proteins confirmed their similarity to EmrA and EmrB of E. coli, respectively, and to other members of these families (Dinh et al., 1994; Paulsen et al., 1996a,b). Thus, orf1 (1185 bp) would encode a 394-amino-acid protein (42.5 kDa; theoretical pI 8.76) that displays strong similarity (data not presented) not only to EmrA of E. coli (44.5% identity over 380 amino acids), but also to EmrK of E. coli (43.2% identity over 315 amino acids), EmrA of Haemophilus influenzae (42.7% identity over 379 amino acids) and VceA of V. cholerae (40.8% identity over 380 amino acids); the C-terminal region of these proteins showed the strongest degree of homology. Similar to EmrA of E. coli and VceA of V. cholerae, the gonococcal protein would lack a signal peptide at its NH2-terminus.
An alignment of the protein encoded by orf2 revealed its high degree of similarity (data not presented) to a number of known or presumed efflux pump transporter proteins possessed by other bacteria (Saier et al., 1994; Paulsen et al., 1996a,b). Thus, orf2 (1524 bp) would encode a 507-amino-acid protein (54.8 kDa; theoretical pI 8.40) that is highly similar to EmrB of E. coli (57.7% identity over 503 amino acids) and VceB of V. cholerae (38.7% identity over 507 amino acids). Both EmrB and VceB are cytoplasmic membrane proteins that serve as transporters, by an energy-dependent process, of antibacterial uncoupling agents and certain hydrophobic antibiotics. They belong to the major facilitator superfamily (MFS) of transporter proteins (Saier et al., 1994; Paulsen et al., 1996a,b). Like other members of the MFS protein family, the gonococcal protein is likely to be an integral membrane protein, as TopPred2 analysis of its amino acid sequence revealed the presence of 14 α-helical, transmembrane domains (data not presented).
In order to ascertain whether the orf1- and orf2-encoded proteins were of any functional significance in gonococci, we created an insertional mutation in the orf2 sequence in strain FA 19 (see Experimental procedures for details). A transformant of strain FA 19 containing a kanamycin resistance (KmR) cassette in orf2 was examined for its susceptibility to uncoupling agents and hydrophobic antibiotics associated with the emrAB- and vceAB-encoded efflux systems, as well as antimicrobial agents associated with the gonococcal mtrCDE-encoded efflux pump. Unlike similar mutants in E. coli (Lomovskaya and Lewis, 1992) and V. cholerae (Colmer et al., 1998), a gonococcal mutant (strain EL-1) was at best only twofold more susceptible than its parental strain (strain FA 19) to uncoupling agents (carbonyl cyanide-m-chlorophenylhydrazone or phenylmercury acetate) or antibiotics, such as nalidixic acid (data not presented). However, an expanded screen of additional antibacterial compounds that consisted of hydrophobic antimicrobial agents, such as free FAs, bile salts, drugs, dyes and detergents, revealed that transformant strain EL-1 was hypersusceptible, compared with parental strain FA 19, to long-chained FAs such as palmitic acid (C16:0), oleic acid (C18:1) and linoleic acid (C18:2) (Table 1) but not the short-chained (C10:0) FA, capric acid [minimum inhibitory concentration (MIC) versus both FA 19 and EL-1 equal to 25 μg ml−1; data not presented]. Strain EL-1 was not, however, greater than twofold more susceptible (Table 1) to other antimicrobial agents [erythromycin (Ery) and Triton X-100 (TX-100)] that are recognized by the MtrC–MtrD–MtrE efflux pump (Hagman et al., 1995; 1997; Shafer et al., 1998). This was evident when transformants of strain FA 19 bearing an insertionally inactivated mtrD gene (strain KH 14) or mtrE gene (strain RD 1) were tested against a number of hydrophobic antibacterial agents because, unlike strain EL-1, these strains were hypersusceptible not only to long-chained FAs but also to TX-100 and Ery (Table 1). The hypersusceptibility of strains KH 14 and RD 1 to FAs is caused by the requirement for MtrE in the export of FAs mediated by the orf1 and orf2 gene products (see below). As insertional inactivation of orf2 seemed significantly to impact gonococcal susceptibility to long-chained FAs but not to hydrophobic agents (e.g. Ery and TX-100) recognized by the MtrC–MtrD–MtrE or EmrA–EmrB efflux pumps, we termed orf2 as farB and the upstream orf1 as farA to signify their involvement in mediating fatty acid resistance.
Table 1. . Susceptibility of gonococcal strains to hydrophobic agents. Ery, erythromycin; LA, linoleic acid; OA, oleic acid; PA, palmitic acid; TX-100, Triton X-100; ND, not determined.
The lack of consensus −10 and −35 hexamers upstream of farB suggested that, like emrAB (Lomovskaya and Lewis, 1992) and vceAB (Colmer et al., 1998), farAB in gonococci is a single transcriptional unit. This hypothesis was confirmed by reverse transcriptase (RT)–PCR analysis of total RNA prepared from strain FA 19. In this experiment, an oligonucleotide that would anneal within the 5′ end of the farB transcript was used for first strand synthesis, and other oligonucleotides that would anneal within farA and farB, but on different strands, were used in the subsequent PCR reaction. In the absence of RT, a PCR product was not generated. However, in the presence of RT, a 942 bp fragment that would span farA and farB was readily observed (data not presented). As this product could only be generated if farAB represented a single transcriptional unit, we concluded that these genes are an operon similar to emrAB and vceAB.
Expression of farAB is dependent on the MtrR transcriptional regulator in strain FA 19
As a homologue of emrR, which negatively regulates the expression of emrAB (Lomovskaya and Lewis, 1992), could not be found in the FA 1090 genome sequence database (data not presented), we thought that farAB expression might be controlled by an alternative transcriptional regulator. We noted that the putative promoter region for farAB transcription contained a sequence (TTTTGCCGCCTGAAGCGTTGTTTTTTGAATA) resembling the MtrR binding site (see underlined nucleotides) previously identified within the mtrCDE promoter region (Lucas et al., 1997). Based on this information, we hypothesized that MtrR might also negatively regulate the transcription of farAB. If so, then FA resistance levels in gonococci should increase because of mutations that result in the loss of MtrR in a manner seen with hydrophobic agents such as Ery and TX-100 (Pan and Spratt, 1994; Hagman et al., 1995; see also Table 1). Accordingly, we measured levels of FA resistance in isogenic transformant strains of FA 19 that contained a deletion in the mtrR-coding sequence (strain KH 11), a missense mutation in the helix–turn–helix coding sequence (strain KH 16) that abrogates MtrR binding to its target DNA (Lucas et al., 1997) or a promoter mutation (strain KH 15) that abrogates mtrR transcription (Hagman and Shafer, 1995). Surprisingly, compared with parental strain FA 19, all of these mutant strains were more susceptible to FAs even though they were more resistant to other hydrophobic agents (Ery and TX-100) (Table 1). Because of these opposing resistance properties in the mtrR mutant strains, we hypothesized that MtrR might act directly or indirectly as a positive regulator of farAB gene expression. In order to test this hypothesis, we compared the levels of the farB and mtrD transcripts in RNA preparations from strains FA 19 and KH 15. It is important to note that, in order to detect the farB transcript reliably, it was necessary to use 2.5 μg of RNA as opposed to 1 μg of RNA for detection of the mtrD transcript. We interpreted this to mean that the mtrCDE expression in strain FA 19 was higher than that of farAB expression. However, compared with parental strain FA 19, MtrR-deficient strain KH15 had an elevated level of the mtrD transcript but a decreased level of the farB transcript (Fig. 2). Taken together with the FA sensitivity data (Table 1), we concluded that the presence of both MtrR (lacking in KH 11 and KH 15) and its DNA-binding activity (lacking in strain KH 16) are required for maximal farAB gene expression and FA resistance in gonococci.
Figure 2. . Detection of rmp, mtrD and farB mRNA level by RT–PCR. Total RNA was prepared from strains FA 19 and KH 15. Total RNA (1 μg) from each strain was applied for rmp and mtrD gene expression, while 2.5 μg of total RNA was used for farB gene expression. Oligonucleotide primers mtrD7 (5′-ATATACAGGGG- AACCACGCCC-3′, anneals 208 nucleotides upstream of mtrD stop codon on the non-coding strand) and mtrD10 (5′-AGCATCAACC- TGCAAGACCGC-3′, anneals 2028 nucleotides downstream of the mtrD start codon on the coding strand) were used for detection of the mtrD transcript. Oligonucleotide primers rmp2 (5′-GTGTTGG- TGATGATTGCGTGCC-3′, anneals 1 nucleotide upstream of the rmp stop codon on the non-coding strand) and rmp3 (5′-ACGCA- ACAACTATGGAGAATGC-3′; anneals 119 nucleotides downstream of the rmp start codon on the coding strand) were used to detect the rmp transcript. Oligonucleotide primers farB5 (5′-GCCGTTGATTCCCCTGTCGC-3′) and farB6 (5′-GCCGAACCTGCCGATTAACG-3′) were used to detect the farB transcript. Lane 1, FA 19 rmp ; lane 2, KH 15 rmp ; lane 3, FA 19 mtrD ; lane 4, KH 15 mtrD ; lane 5, FA 19 farB ; lane 6, KH 15 farB.
Download figure to PowerPoint
The MtrE outer membrane protein serves as a component of the FarA–FarB efflux pump
The 50 kDa MtrE lipoprotein serves as the outer membrane protein channel for the mtr efflux pump (Delahay et al., 1993). We hypothesized that it could also serve a similar function for the farAB-encoded efflux pump, especially as the TolC outer membrane protein is thought to be the outer membrane channel protein for the emrAB- and acrAB-encoded efflux pumps (Fralick, 1996; Nikaido, 1996). In order to test whether MtrE functions as part of the FarA–FarB efflux pump, we introduced an mtrE::Km sequence into strain AP 697 using donor DNA from a genetic derivative (strain RD 1) of strain FA 19. Strain AP 697 has a 2 bp deletion in its mtrC gene (nucleotide positions 351 and 352) that would result in premature truncation of MtrC (data not presented). This mutation can explain (Veal et al., 1998) its hypersusceptibility to Ery and TX-100. The mtrE gene in strain RD 1 has a 392 bp deletion that has been replaced (Delahay et al., 1997) by the non-polar aphA-3 cassette described by Ménard et al. (1993). A number of transformants that contained an insertionally inactivated mtrE gene, as determined by PCR (data not presented), were screened for levels of susceptibility to FAs, TX-100 and Ery. As shown in Table 1, a representative transformant strain (EL-3) was more sensitive than parental strain AP 697 to linoleic acid, but its susceptibility to TX-100 and Ery was unchanged. Based on this observation and previous studies dealing with MtrE (Delahay et al., 1997), we concluded that MtrE serves as the outer membrane channel for both the mtr and far efflux systems in gonococci. This result explains why previous studies (Hagman et al., 1995; 1997; Delahay et al., 1997) found that gonococci with polar mutations in mtrC or mtrD (strain KH 14, Table 1) that would abrogate mtrE expression or insertional inactivation of mtrE (strain RD 1, Table 1) in strain FA 19 resulted in hypersusceptibility of gonococci to long-chained FAs. It is interesting to note that, although MtrE is required for the FarA–FarB pump function, its level would be reduced in strains with a wild-type MtrR regulator. We hypothesize that, as farAB expression is apparently lower than that of mtrCDE (see above), sufficient levels of MtrE would be present in wild-type strain FA 19 for maximal efflux of FAs by the FarA–FarB efflux pump.
Previously, the mtrCDE-encoded efflux pump system was thought to be responsible for the export of antibacterial FAs (Delahay et al., 1997; Hagman et al., 1995, 1997). Interestingly, a homologue of this pump (AcrA–AcrB) can protect E. coli from decanoate (Ma et al., 1995). However, both Pan and Spratt (1994) and Hagman et al. (1995) in their initial reports of the mtr system recalled the earlier report of McFarland et al. (1983) that predicted an mtr-independent mechanism of FA resistance in gonococci. The results presented herein strongly suggest that this mechanism of FA resistance results from a second efflux pump system (farAB ) that consists of a membrane fusion protein (FarA) and cytoplasmic transporter protein (FarB) that resembles the EmrAB and VceAB efflux pumps of E. coli (Lomovskaya and Lewis, 1992) and V. cholerae (Colmer et al., 1998) respectively. The gonococcal FarA–FarB efflux system does not, however, seem to recognize substrates (uncoupling agents and certain antibiotics) exported by these homologous systems, nor is there evidence that these pumps export FAs.
While the far efflux mechanism of FA resistance seems to be the mtr-independent system that McFarland et al. (1983) predicted to exist in gonococci, both may be subject to control by the MtrR regulator. It is now clear that MtrR is a transcriptional repressor of mtrCDE, as its absence in strains KH 11 and KH 15 or loss of function in strain KH 16 results in enhanced resistance to hydrophobic agents such as Ery and TX-100 (Pan and Spratt, 1994; Hagman et al., 1995) and mtrCDE gene expression (Hagman and Shafer, 1995). MtrR binds within the promoter used for mtrCDE expression (Lucas et al., 1997) and a similar MtrR binding site upstream of farAB. We hypothesized that both the decreased FA resistance (Table 1) and expression of farB (Fig. 2) among the MtrR− variants of strain FA19 could be explained if MtrR binding to this site results in gene activation instead of repression. However, in a number of binding experiments that used the MtrR fusion proteins employed in our earlier experiments of MtrR binding to the mtrCDE promoter (Lucas et al., 1997), we found that, in gel mobility bandshift experiments, large amounts of MtrR were required to observe its binding to the putative binding site upstream of farAB, but this binding was non-specific (data not presented). Thus, MtrR may act indirectly on farAB gene expression through its capacity to regulate other genes, perhaps a transcriptional repressor of farAB. We did not detect an emrR homologue upstream of farAB, nor have we identified one in the gonococcal genome sequence database. As transcriptional regulatory proteins that negatively control expression of efflux pump operons seem to be a common feature in bacteria (Ma et al., 1994; 1995), we are presently examining the mechanism(s) by which farAB is regulated and how, as our experimental results would suggest, MtrR is involved in such control.