Mechanistic Basis for Decreased Antimicrobial Susceptibility in a Clinical Isolate of Neisseria gonorrhoeae Possessing a Mosaic-Like mtr Efflux Pump Locus

Historically, after introduction of an antibiotic for treatment of gonorrhea, strains of N. gonorrhoeae emerge that display clinical resistance due to spontaneous mutation or acquisition of resistance genes. Genetic exchange between members of the Neisseria genus occurring by transformation can cause significant changes in gonococci that impact the structure of an antibiotic target or expression of genes involved in resistance. The results presented here provide a framework for understanding how mosaic-like DNA sequences from commensal Neisseria that recombine within the gonococcal mtr efflux pump locus function to decrease bacterial susceptibility to antimicrobials, including antibiotics used in therapy of gonorrhea.

effective antibiotics some gonorrheal infections in the future may be untreatable (4,5). Currently, a dual antibiotic treatment regimen of ceftriaxone (Cro) (single intramuscular injection of 250 to 500 mg) and azithromycin (Azi) (single oral dose of 1 to 2 g) is used in many western countries (6,7), but their continued efficacy for use in curing gonorrheal infections is threatened as strains resistant to either or both antibiotics have emerged in the past decade (8)(9)(10)(11).
The gonococcus has adapted numerous strategies to survive attacks by antimicrobials, including the use of multidrug efflux pumps to export toxic compounds (3,12,13). Five gonococcal efflux pumps that export a wide range of substrates have been described (13). Of these, the best-studied efflux pump is MtrCDE, which belongs to the resistance-nodulation-division family possessed by many Gram-negative bacteria. MtrCDE captures and exports structurally diverse, but generally amphipathic, antimicrobial agents, including macrolides, beta-lactams, cationic antimicrobial peptides, dyes, and detergents (13). The contribution of the MtrCDE efflux pump in antimicrobial resistance expressed by gonococci can be enhanced by cis-or trans-acting mutations that result in overexpression of the mtrCDE efflux pump operon (13). Importantly, overproduction of the MtrCDE efflux pump due to relief of transcriptional repression of mtrCDE can contribute to clinically relevant levels of resistance to beta-lactams and macrolides (13).
Expression of mtrCDE in wild-type (WT) gonococci is subject to repression by MtrR (14,15) and, in the presence of an inducer, activation by MtrA (16). Both MtrA and MtrR bind to regions within a 250-bp sequence that contains overlapping, divergent promoters for mtrR and mtrCDE transcription (17). Loss of MtrR repression of mtrCDE can result from point mutations in the MtrR-binding site (14), a single base pair deletion within a 13-bp inverted repeat sequence in the mtrR promoter (18), a point mutation that creates a new mtrCDE promoter (19), or missense/nonsense mutations in the mtrR gene (20)(21)(22). In addition to decreasing gonococcal susceptibility to antibiotics, these regulatory mutations can also enhance the fitness of gonococci during experimental infection of the lower genital tract of female mice (23), which supports the concept that the MtrCDE efflux pump is of importance during infection due to its ability to export host-derived antimicrobials such as cationic antimicrobial peptides and progesterone (24).
While single-site regulatory mutations impacting mtrCDE expression have been extensively studied (14,(19)(20)(21)(22)25), increasing evidence suggests that entry and recombination of donor DNA from commensal Neisseria spp. into the mtr locus can result in multiple nucleotide changes that can decrease gonococcal susceptibility to antimicrobials including Azi and Cro. Thus, the presence of mosaic-like sequences within the mtrR region likely resulting from transformation by DNA from N. lactamica or N. meningitidis has been reported in worldwide-isolated gonococcal strains (26)(27)(28)(29)(30). Recent work by Wadsworth et al. (30) showed that gonococci bearing diverse mosaic-like sequences within the mtrR/mtrCDE promoter region have elevated expression of mtr-associated genes and decreased susceptibility to Azi. Importantly, mosaic-like sequences within the mtrD inner membrane transporter protein-encoding gene showed strong linkage disequilibrium and epistatic effects that likely enhance the activity of the MtrCDE efflux pump (30). Taken together, the available information strongly suggests that mosaic-like sequences in the mtr locus can result in increased expression of the mtrCDE efflux pump operon as well as providing a gain-of-function property to MtrD that enhances its ability to export antimicrobials. In this study, we examined gonococcal clinical isolates that possess a mosaic-like mtr locus similar to other clinical isolates (30). We describe both a single nucleotide change in the overlapping mtrR/mtrCDE promoters and a likely mechanism for its impact on gene transcription, and amino acid changes in the C-terminal domain of MtrD that were linked to the decreased antimicrobial susceptibility phenotype expressed by gonococci with a mosaic-like mtr locus.

RESULTS
Importance of the MtrCDE efflux pump in reduced susceptibility to Azi and other antimicrobials in gonococcal clinical isolates bearing a mosaic-like mtr locus. Public health laboratories associated with the Gonococcal Isolate Surveillance Project (GISP) alert the Centers for Disease Control and Prevention (CDC) to N. gonorrhoeae isolates if the Azi MIC is 2 g/ml or greater. High-level MICs of Azi (Ն 256 g/ml [3]) are typically due to mutations in the four 23S rRNA genes (31). However, in recent years it has become apparent that gonococci can display a so-called "less-Azisusceptible" phenotype characterized by MIC values of 2 to 4 g/ml (3, 25-30) that does not involve 23S rRNA mutations. This less-Azi-susceptible property may help gonococci escape the action of this macrolide during treatment, especially at extragenital sites of infection (e.g., the pharyngeal mucosa) where the pharmacokinetic properties of antibiotics are not optimal (32). It is therefore important to define how gonococci can develop decreased Azi susceptibility in the absence of 23S rRNA mutations as this could result in clinical failure of this macrolide during certain infections.
In order to study emergence of gonococcal clinical isolates with reduced susceptibility to Azi and to ascertain the contribution, if any, of the MtrCDE efflux pump system to this phenotype, we analyzed eight clinical strains collected in 2014 that expressed a less-Azi-susceptible property (MIC of 2 g/ml); details of these strains are provided in Materials and Methods and Table S1 in the supplemental material. Whole-genome sequencing (WGS) performed on these eight strains and detailed bioinformatic analysis revealed that they lacked known 23S rRNA mutations associated with high levels of Azi resistance (data not presented). The sequences of the genes within the mtr locus of the 8 strains were identical, and they contained multiple nucleotide differences compared to antibiotic-sensitive reference strain FA19; the details of the WGS performed on these and other strains will be presented elsewhere (O. O. Soge and A. J. Abrams, unpublished data). Briefly, an alignment of the nucleotide sequence of the entire mtr locus in clinical strain LRRBGS0002 (here termed CDC2), displaying a less Azi-susceptible phenotype, with that of FA19 is shown in Fig. S1A. The nucleotide sequence of the mtr locus possessed by strain CDC2 was most dissimilar to FA19 (as well as three other gonococcal strains [MS11, FA1090, and H041]) in the mtrR/mtrCDE promoter region, the mtrD gene, and the noncoding region between mtrD and mtrE ( Fig. 1). A BLAST search against Neisseria (NCBI:txid482) nucleotide sequences in NCBI (https://blast.ncbi.nlm.nih .gov/Blast.cgi) determined that the entire mtr locus sequence possessed by CDC2 was most similar (95% identity) to that of N. polysaccharea M18661 (GenBank CP031325.1). Analysis of a phylogenetic tree based on the mtr loci of CDC2, FA19, and three clinical strains (GCGS0276, GCGS0402, and GCGS0834) with mosaic-like mtr loci studied by Wadsworth et al. (30) indicated that the mtr locus of CDC2 was most similar to that possessed by strain GCGS0402 (Fig. S1B). In fact, the nucleotide sequences of the mtr loci of CDC2 and GCGS0402 were identical. Importantly, however, CDC2 lacked the Correia element (CE) (33) that is positioned adjacent to the mtrR-mtrCDE promoter region found in mtr mosaic-like strain GCGS0276 (30) (Fig. S1A).
Since CDC2 contained numerous nucleotide sequence variations (with respect to FA19) in the mtrR coding region and the overlapping mtrR/mtrCDE promoter region, we hypothesized that it (as well as the seven other alert strains) might overproduce the MtrCDE pump, leading to decreased susceptibility to Azi and other antimicrobials. In order to confirm that this efflux pump is required for antimicrobial resistance in CDC2, as has been reported for non-mtr mosaic-like clinical isolates such as H041 (25), we created a mutant that lacked a functional MtrCDE efflux pump due to insertional inactivation (mtrD::kan) of the parental mtrD gene, which encodes the MtrD cytoplasmic membrane transporter. We also created other mutants that lacked functional MacAB or NorM efflux pumps to ascertain if their loss might also increase susceptibility of this strain to antimicrobials. Of these three efflux pumps, only loss of an active MtrCDE efflux pump rendered CDC2 and the other Azi alert isolates hypersusceptible to Azi (Table S1). Similarly, for CDC2 (Table 1) and the other Azi alert strains (data not presented), only the loss of the MtrCDE efflux pump (see CR.99 in Tables S1 and S2) resulted in hypersusceptibility to the tested antimicrobials.
Since the MtrCDE efflux pump was essential for cross-resistance of CDC2 to antimicrobials, we focused on defining the impact of mosaic-like sequences in its mtr locus on MIC values of antimicrobials exported by this pump. For this purpose, we studied expression of the mtrR repressor gene and mtrE gene, which encodes the outer membrane protein (OMP) channel of the pump and is the last gene in the mtrCDE operon ( Fig. 1) (3). Expression of these genes in FA19 and CDC2 was assessed at the levels of transcription and translation by qRT-PCR and Western immunoblotting, respectively. At the level of transcription, it was found that mtrR expression in FA19 was significantly higher than that of CDC2, while mtrE expression was higher in CDC2 ( Fig. 2A). With respect to the MtrR repressor protein, its level in CDC2 and the other seven clinical Azi alert strains was substantially lower than that in strain FA19 (Fig. S2A). As expected from the low level of the MtrR repressor in CDC2, the level of the MtrE OMP in this strain was higher than that in FA19. In this respect, the level of MtrE in CDC2 was  Table S1. Clustal Omega multiple sequence alignments were performed using N. gonorrhoeae strains FA19, CDC2, FA1090, H041, and MS11. Alignments were generated for each mtr gene using their nucleotide and amino acid sequences, and nucleotide sequences were aligned for the intergenic regions. Pairwise identity matrices were calculated, and the pairwise identity values for each alignment are shown. ® similar to that of strain JF1 (FA19ΔmtrR), but lower than that of KH15 (FA19 with single base pair deletion in the mtrR promoter, which is known to decrease mtrR expression and increase mtrCDE expression) (Fig. S2B). Increased expression of mtrCDE due to mosaic-like mtrR and mtrR-mtrCDE promoter regions can contribute to, but is not sufficient for, the decreased-Azi susceptibility phenotype of CDC2. We hypothesized that the less-Azi-susceptible phenotype of CDC2 was due, in part, to enhanced levels of the MtrCDE efflux pump resulting from cis-and/or trans-acting mutations generated by the mosaic-like sequence that influence mtrR and mtrCDE expression. To test this hypothesis, we replaced the FA19 mtrR coding and mtrR-mtrCDE promoter sequences with that possessed by CDC2 and tested if this would influence expression of these genes as well as increasing MICs of Azi and other antimicrobials recognized by the MtrCDE efflux pump. For this purpose, a PCR-generated DNA fragment containing the mtrR coding region, the mtrR-mtrCDE intervening region, and the 5= end of mtrC present in CDC2 (summarized in Fig. 3A) was used to transform strain FA19 for increased resistance to a known MtrCDE substrate, Triton X-100 (TX-100). A resulting transformant termed CR.100 (Tables 1 and 2) was selected for more detailed studies. DNA sequencing revealed that CR.100 had mtrR coding and mtrR-mtrCDE promoter sequences that contained most, but not all, of the CDC2 donor mosaic-like DNA in this region (Fig. 3A). With respect to the nucleotide changes in the noncoding sequence, CR.100 had the CDC2 mosaic-like sequence that included one nucleotide change within the Ϫ35 hexamer of the mtrR promoter (C to T) and one change within the Ϫ35 hexamer of the mtrCDE promoter (T to G) ( Fig. 3A and Table 2) that have been observed in other strains (27,30). In this respect, Wadsworth et al. found that the Ϫ35 mtrR promoter mutation was present in Mechanisms of Mosaic-Like mtr Sequences ® GCGS0402, while the Ϫ35 mtrCDE promoter mutation was present in all three of the mtr mosaic-like strains used in their study (30). Although a nucleotide difference was also noted in the previously identified transcriptional start site (TSS) for mtrCDE transcription (18), primer extension analysis revealed that transcription of mtrCDE in strains CDC2 and FA19 was similarly initiated (Fig. S3). CR.100 also contained the  CDC2-derived mosaic-like sequence in mtrR, which was characterized by missense mutations in codons 79 (D79N), 183 (S183N), and 197 (M197I) (Fig. 3B). As is shown in Table 1, transformant strain CR.100 was more resistant than parent strain FA19 to a panel of MtrCDE substrate antimicrobials but was 2-fold less resistant than donor strain CDC2 to macrolides Azi and erythromycin (Ery) and 4-fold less resistant to ethidium bromide (EB). An examination of transcript levels of mtrR and mtrE in FA19 versus CR.100 showed that mtrR expression was decreased in CR.100 while mtrE expression was increased (Fig. 2B). Although the MtrR repressor protein was readily detected in whole-cell lysates of FA19, it was much lower in transformant strain CR.100 (Fig. 4). This result indicated that acquisition of the mosaic-like sequence encompassing the mtrR coding and mtrR-mtrCDE promoter sequences resulted in transcriptional repression of mtrR and derepression of mtrCDE. However, it was unclear as to whether these gene expression differences and increased MICs of antimicrobials were due to the mtrR coding or promoter mutations present in CR.100.
In order to separate potential influences of cis-or trans-acting mutations in the overlapping mtrR/mtrCDE promoter regions from that of the mtrR coding region, respectively, on levels of antimicrobial resistance and gene expression in CR.100, we generated a pair of PCR products that singularly covered these regions in CDC2. We found that both regions could transform WT strain FA19 for decreased susceptibility to TX-100 or Azi. Recovered transformants were termed CR.101 and CR.102. DNA sequencing of PCR products from CR.101 and CR.102 was performed to determine the extent of recombination of the donor mosaic-like sequence in the mtrR coding or upstream noncoding region. With the promoter region-bearing PCR product, we found that complete replacement of the wild-type mtrR-mtrCDE intervening region by the donor sequence from CR.100 had occurred in CR.101 (Table 2 and Fig. 3A). In contrast, with the mtrR coding PCR product, only the MtrR D79N mutation was present in CR.102; the MtrR amino acid alignment information is provided in Fig. 3B and summarized in Table 2. Antimicrobial susceptibility testing results (Table 1) showed that the MICs of macrolides against CR.101 (mosaic-like mtrR-mtrCDE intervening region) were 2-fold higher than those of CR.102 (MtrR D79N). Interestingly, TX-100 resistance was Ͼ32-fold higher in CR.101 than CR.102 (MtrR D79N); the MIC against CR.101 resembled that seen with KH15 (Table 1), which has a single-base-pair deletion in the mtrR promoter that abrogates mtrR gene expression and shifts RNA polymerase recognition to the mtrCDE promoter (18). In contrast, the TX-100 MIC versus CR.102 was similar to that of JF1 (Table 1), which has a deletion of mtrR but retains a wild-type mtrR promoter (34).
The results from antimicrobial susceptibility testing suggested that mtrR coding and noncoding mosaic-like sequences have different impacts on expression of mtr- associated genes. Indeed, we found that while the MtrR D79N mutation in CR.102, which is also possessed by GCGS0402 and GCGS0834 (30), resulted in increased levels of mtrE expression compared to parental strain FA19, it also, unlike CDC2 and CR.100, had increased levels of the mtrR transcript (Fig. 2D) and the MtrR repressor protein (Fig. 4). Although position 79 of MtrR is outside the DNA-binding domain of this repressor (14,15), we suggest that this amino acid change causes a decrease in MtrR function, resulting in derepression of mtrCDE expression. Nevertheless, the consequence of this mutation did not endow gonococci with either the antimicrobial resistance profile or the mtrR gene expression profile observed for CDC2 or CR.100, which also have mtrR/mtrCDE promoter mutations (Table 1 and Fig. 2).
Based on the findings with CR.102, we next tested whether potentially cis-acting mutations in the mtrR-mtrCDE promoter region influence expression of mtrR or mtrCDE. As was observed with clinical isolate CDC2 and transformant strain CR.100, the presence of the mosaic-like mtrR-mtrCDE promoter region in CR.101 resulted in a decreased level of mtrR expression but increased amounts of the mtrE transcript compared to parental strain FA19 (Fig. 2C). We hypothesize that mutations upstream of mtrR, especially the single nucleotide changes in the adjacent Ϫ35 hexamers of the mtrR and mtrCDE promoters (Fig. 3A), negatively impact expression of mtrR. Such repression of mtrR would not occur in CR.102 (MtrR D79N) where the wild-type FA19 promoter sequence is present (Fig. 3A) and MtrR levels are elevated (Fig. 4). Coupled with the antimicrobial susceptibility data (Tables 1 and 2), this result indicated that the nucleotide changes in the promoter region were responsible for modulating expression of mtrR and mtrCDE in clinical isolate CDC2 and transformant strain CR.100.
A single nucleotide change in the mosaic-like mtrR/mtrCDE promoter region can impact gene expression. We hypothesized that the T-to-G change in the Ϫ35 hexamer region of the mtrCDE sigma-70 promoter in clinical isolate CDC2 and FA19 transformant strains CR.100 and CR.101 (red arrow, Fig. 3A) directly raised mtrCDE expression ( Fig. 2A) and the MICs of MtrCDE antimicrobial substrates (Table 1); a similar nucleotide change was observed by others in mtr mosaic-like strains (27,30). This nucleotide change would result in an improved Ϫ35 hexamer (5=-TTTTAT-3= to 5=-TT GTAT-3=) of the mtrCDE promoter. To test the importance of this T-to-G change, two pLES94 mtrCpromoter-lacZ fusion (P mtrC -lacZ) constructs containing either the CDC2 mtrCDE promoter (CR.102pLES2.2) or an identical sequence but with the T nucleotide (CR.102pLES4.1) were introduced into CR.102 (FA19 MtrR D79N) by transformation; the lacZ fusions were integrated in the proAB region of the gonococcal chromosome (35). After verification of selected transformants by DNA sequencing of PCR products, levels of beta-galactosidase (␤-Gal) were measured. As shown in Fig. 5, the ␤-Gal expression level was 3.5-fold higher in gonococci with the P mtrC -lacZ fusion that contained the G nucleotide in the Ϫ35 sequence of mtrCDE promoter compared to the variant that had the T nucleotide. This result suggests that the T-to-G change observed in the Ϫ35 hexamer of the mtrCDE promoter possessed by CDC2 as well as other isolates (30) results in a more effective mtrCDE promoter than that possessed by nonmosaic strain FA19. However, since MtrR can activate certain gonococcal genes (34), it was necessary to eliminate the (remote) possibility that the T-to-G nucleotide change facilitated binding of MtrR D79N to the mtrCDE promoter in an activating capacity. Accordingly, we introduced the P mtrC -lacZ fusion in pLES2.2 into strain JF1 (FA19 ΔmtrR). The results showed that compared to CR.102, expression of P mtrC -lacZ from the pLES2.2 fusion in JF1 was slightly elevated (data not presented), which is consistent with the MtrR D79N protein in CR.102 retaining a degree of transcriptional repressive activity as opposed to becoming an activator of mtrCDE.
Mutations in mtrD due to acquisition of a mosaic-like sequence can increase antimicrobial resistance. The results from the gene expression studies described above implicated the T-to-G change in the Ϫ35 mtrCDE promoter hexamer as having a strong cis-acting influence that derepresses mtrCDE expression, leading to decreased susceptibility of gonococci to antimicrobials, including Azi. Collectively, these findings likely explain the levels of mtrR and mtrCDE gene expression in mtr mosaic-like strain CDC2 and other clinical isolates (27,30) that possess this mutation. However, these transcriptional changes could not fully explain the higher MIC values of certain antimicrobials against CDC2. Thus, although transformant strain CR.100 had decreased susceptibility to all tested antimicrobials, only the MIC values of crystal violet (CV) and TX-100 matched that of CDC2 (Table 1). We hypothesized that amino acid differences in the MtrD efflux pump transporter protein possessed by CDC2 compared to FA19 and CR.100 might account for the higher Azi MIC seen with strain CDC2. Based on the structure of MtrD (PDB 4MTI) (36) and known similarity to AcrB (37), the following amino acid changes in MtrD possessed by strain CDC2 are predicted (E. Yu, personal communication) to impact its binding of antimicrobials: T42N, H46R, I48T, N101D, V662I, and K823E. To test whether these or other amino acid changes contribute to antimicrobial resistance in CDC2, three overlapping PCR products (summarized in Fig. S4) encompassing its entire mtrD gene were used together to transform strain CR.100 for resistance to 1 g/ml of Azi. A resulting transformant (CR.103) was examined in antimicrobial susceptibility assays and was found to have identical levels of resistance as CDC2 to Azi (MIC of 2 g/ml) and Ery (MIC of 4 g/ml) but was 2-fold less resistant than CDC2 to the dye EB (Table 1). DNA sequencing of mtrD PCR products from CDC2, CR.100, and CR.103 showed that the nucleotide sequence of mtrD in these strains was identical to that of FA19 until codon 738, after which the CDC2 sequence had recombined in CR.103 until codon 1020. This recombination event generated twenty-three amino acid replacements in the C-terminal end of MtrD (Fig. 6). The amino acid changes in this C-terminal region of MtrD are located in the DC subdomain of the docking domain as well as in the PC2 subdomain of the pore region and transmembrane domains TM8 and TM9 (36). As assessed by qRT-PCR, CR.100 and CR.103 did not differ in levels of the mtrD transcript (data not presented), indicating that differences in levels of antimicrobial resistance between these strains were linked to structural alterations of MtrD located at the C-terminal end.
In order to verify that amino acid changes in the C-terminal region of MtrD from CR.103 could increase MIC values of Azi, we prepared an mtrD 3=-end PCR product described above from CR.103 and used it to transform CR.100 with selection for resistance to 1 g/ml of Azi. Six individual transformants were assessed in antimicrobial susceptibility testing assays using Azi and Ery, and all showed an elevated level of resistance to these macrolides comparable to that observed with donor strain CR.103 (data not presented). Differences were detected in the nucleotide sequence of the donated 3= end of mtrD in the transformants, indicating that unique sites of recombination had occurred. Translation of the nucleotide sequence to amino acid sequence showed similar as well as unique amino acid changes in each transformant strain (data not presented). Importantly, however, only two amino acid changes, positioned at 821 (Ser to Ala) and 823 (Lys to Glu), were common in all six transformants. In fact, one transformant (CR.104) had only these two amino acid changes ( Fig. 6 and Table 2) compared to recipient strain CR.100. We also sequenced the entire mtr locus (6,793 bp) possessed by strains FA19 and CR.104 and found that except for the nucleotide differences in the mtrR coding region, mtrR-mtrCDE intervening region, and the mtrD allele of CR.104, the mtr loci in these strains were otherwise identical (data not presented). The MIC values of antimicrobials against CR.104 were identical to that of donor strain CR.103 (Table 1); the sole exception was with CV, to which CR.104 was 2-fold more resistant.

DISCUSSION
Recombination of commensal neisserial DNA sequences into the chromosome of N. gonorrhoeae and N. meningitidis has been well established (27,30,(38)(39)(40). Such horizontal transmission of DNA can endow gonococci or meningococci with decreased susceptibility to antibiotics that target the respective gene product. The best example of the consequence of this genetic exchange event is that of mosaic sequences in the pbp2 gene (also known as penA), which encodes the beta-lactam-sensitive target penicillin-binding protein 2 (PBP2) (41). The extensive remodeling of PBP2 that occurs due to the multiple (up to 60) amino acid changes due to a mosaic pbp2 decreases the acylation rate of penicillin and third-generation cephalosporins (cefixime and ceftriaxone) (42). pbp2 mosaic gonococcal strains also frequently contain cis-acting regulatory mutations that increase expression of the mtrCDE efflux pump operon (reviewed in reference 3). Loss of the MtrCDE efflux pump in gonococci with chromosomally mediated penicillin resistance can result in a return to a penicillin-sensitive phenotype  (25,43), while a similar loss can result in a 2-to 4-fold increase in susceptibility to Cro and Azi (25). Thus, mutations that derepress mtrCDE expression can work with mosaic pbp2 to decrease gonococcal susceptibility to beta-lactams.
The mosaic-like mtrR/mtrCDE promoter region described here and that reported by Wadsworth et al. (30) emphasize that in gonococci nucleotide changes due to acquisition of mosaic-like mtr sequences can elevate mtrCDE expression and MICs of antibiotics; a graphic model describing the influence of these nucleotide changes was recently presented in a commentary (44) by the corresponding author (W. M. Shafer) of the work by Wadsworth et al. (30). We propose that the single nucleotide change (T to G) within the Ϫ35 hexamer of the mtrCDE promoter provides increased expression of mtrCDE, as this change results in a consensus Ϫ35 hexamer sequence for sigma-70 promoters. Further, since this nucleotide change is also within the 13-bp inverted repeat positioned between the Ϫ10 and Ϫ35 regions of the mtrR promoter, it could alter the ability of this sequence to form a DNA secondary structure with regulatory activity. Thus, the position of this nucleotide change would maintain the spacing between the Ϫ10 and Ϫ35 hexamers of the mtrR promoter but would reduce the T:A bp repeat from 6 to 4. This reduction in the T:A bp repeat may impact promoter recognition by RNA polymerase, which preferentially shifts its recognition from the mtrR promoter to the mtrCDE promoter that has an improved Ϫ35 hexamer sequence. It should also be noted that mtr mosaic-like CDC2 exhibited a low level of mtrR expression, possibly due to the nucleotide change in the Ϫ35 hexamer of the mtrR promoter ( Fig. 2 and 3). We hypothesize that this may also contribute to the low levels of MtrR in this and other clinical isolates. Interestingly, when the MtrR D79N mutation from CDC2 was placed into strain FA19, which has a WT mtrR promoter, levels of MtrR increased significantly (Fig. 4), indicating that this mutation can cause dysregulation of autoregulation of mtrR expression.
While the Ϫ35 mtrCDE promoter mutation can endow gonococci with increased expression of mtrCDE (Fig. 2) and elevate gonococcal resistance to antimicrobials, it did not account for the Azi MIC observed with clinical isolate CDC2 (Tables 1 and 2). In this respect, the results presented here and those of Wadsworth et al. (30) show that amino acid changes in mtrD, which result from importation of commensal neisserial DNA, are also necessary for the less-Azi-susceptible phenotype of mtr mosaic-like strains such as CDC2 and GCGS0402. Based on the published MtrD structure (36), many of the amino acid positions in MtrD that are changed in CDC2 versus antibiotic-sensitive strain FA19 could influence drug binding, efflux activity, or protein stability. Results from our transformation experiments indicated, however, that amino acid changes in the C-terminal end of MtrD, especially at positions 821 (S821A) and 823 (K823E), are sites for gain-of-function mutations that can contribute to antimicrobial resistance seen in mtr mosaic-like strains. These two amino acids are within the PC2 region of MtrD that is part of the pore domain of this transporter and is located at the outermost region that faces the periplasm. Based on the structural comparison of MtrD with the orthologous AcrB protein (37), position 823 in MtrD is predicted to be a site for binding antimicrobials (E. Yu, personal communication). Amino acid changes located elsewhere in MtrD could also have a similar impact, as indicated by the higher EB MIC value against CDC2 versus transformant strains CR.103 and CR.104. We note that the S821A and K823E changes are also present in the MtrD protein of the GCGS0402 and GCGS0834 strains studied by Wadsworth et al. (30); interestingly, GCGS0276, which also possesses a mosaic-like mtrD, has a K823D change that may have a similar functional impact on MtrD activity as the K823E mutation (30).
Active international and national surveillance systems for determining trends in gonococcal resistance to antibiotics coupled with whole-genome sequencing and bioinformatic analyses have been instrumental in detecting gonococci with resistance determinants, including those generated by mosaic-like mtr sequences (45)(46)(47)(48)(49). The detailed molecular, genetic, and phenotypic analyses made possible by these efforts have provided new insights regarding the impact of the gonococcal MtrCDE efflux pump in determining levels of bacterial resistance to clinically important antibiotics.
The worldwide distribution of strains with mosaic-like mtr loci indicates that future diagnostic tests for resistance determinants should include functionally important nucleotide changes that were not described in earlier studies (summarized in reference 13). Further, the presence of such gonococcal strains reemphasizes that genetic exchange between gonococci and commensal Neisseria, which is likely to occur in the pharyngeal cavity, can be a major mechanism by which N. gonorrhoeae develops resistance to antibiotics. Thus, in order to enhance monitoring of antibiotic resistance trends, it would be prudent to routinely sample both genital and extragenital (especially the oral mucosae) sites for the presence of gonococci in patients suspected of having gonorrhea so as to better detect emergence of antibiotic-resistant strains in the community.
The resulting PCR was transformed into CR.100, and clones were selected on GC agar plates supplemented with 1 g/ml of Azi.
Quantitative reverse transcriptase PCR (qRT-PCR). RNA was extracted from strains FA19, CDC2, CR.100, CR.101, CR.102, and CR.104 at mid-logarithmic phase of growth in GC broth plus supplements by the TRIzol method as directed by the manufacturer (Thermo Fisher Scientific, Waltham, MA) and was performed as described previously (59). Briefly, genomic DNA (gDNA) was removed by RNase-free DNase treatment and gDNA Wipeout (Qiagen, Germantown, MD). The resulting RNA was then reverse transcribed to cDNA using the QuantiTect reverse transcriptase kit (Qiagen) as described previously (57). Primers 16Smai_qRTF and 16Smai_qRTR were used for 16S rRNA. Primers mtrEqPCR-F and mtrEqPCR-R were used for the mtrE gene, and primers mtrD8 and mtrD13 were used for the mtrD gene. Primers mtrR_qRT_F and mtrR_qRT_R were used for the mtrR gene. Sequences of primers are shown in Table S2. Results were calculated as normalized expression ratios (NERs) using 16S rRNA expression levels. Statistical significance was calculated by Student's t test.
Western immunoblot analysis. Whole-cell lysates were prepared from gonococcal strains grown overnight on GC agar plates with supplements as described previously and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (60). Coomassie brilliant blue (CBB) staining of duplicate SDS-PAGE gels was performed to calibrate and verify consistent loading of proteins (15 g) loaded into each well. The concentration of protein in whole-cell lysates was estimated by using a NanoDrop spectrometer at 280 nM. Western immunoblotting using a mouse anti-MtrE serum (kindly provided by A. Jerse) and a rabbit anti-MtrR serum was performed as described previously (23,34).
␤-Gal assays. lacZ fusions were constructed using the pLES94 system as previously described (35). Briefly, PCRs were performed on genomic DNA from strain CDC2 using primers C2 and C3PmtrC for pLES2.2 and C4 and C3PmtrC for pLES4.1. The resulting PCR products were cloned into the BamHI site of pLES94. pLES94, pLES2.2, and pLES4.1 and were introduced into strain CR.102 or JF1 by transformation with selection on GC agar plates supplemented with 1 g/ml of chloramphenicol. ␤-Gal assays were performed in triplicate as described by Folster et al. (34) from lysates of gonococcal strains after growth overnight on GC agar plates supplemented with chloramphenicol. The ␤-Gal specific activity was calculated using the formula A 420 ϫ 1,000/A 280 (mg/ml) ϫ time (min) ϫ volume (ml).
Primer extension analysis. Total RNA from strains FA19 and CDC2 was prepared from mid-and late-logarithmic phase GCB broth cultures by the TRIzol method as directed by the manufacturer (Thermo Fisher Scientific, Waltham, MA). Primer extension experiments were performed as described previously (18) on 6 g of total RNA with primer PEmtrC181. The primer extension transcription start site of the mtrC gene was determined by electrophoresis of the extension products on a 6% (wt/vol) DNA sequencing acrylamide gel adjacent to reference sequencing reactions.

ACKNOWLEDGMENTS
We thank E. Yu (Case Western University) for his insightful comments regarding MtrD structure and impact of mutations with respect to binding antimicrobials. We also thank the Gonococcal Isolate Surveillance Project (GISP) for the use of isolates and corresponding data for this analysis. Edward Bannister (Dallas, TX) and the Dallas, TX, GISP sentinel site collected, isolated, and provided epidemiological data for the isolates used in this analysis. The University of Washington GISP Regional Laboratory (Seattle, WA) determined and provided antimicrobial susceptibility result data for the isolates used in this analysis. We also thank Steven Johnson, Hsi Liu, Matthew Schmerer, Sandra Seby, Jesse Thomas, and Eshaw Vidyaprakash for thoughtful and informative discussions.
The contents of this article are solely the responsibility of the authors and do not necessarily reflect the official views of the National Institutes of Health, the Centers for