Genotyping as a Tool for Antibiotic Resistance Surveillance of Neisseria gonorrhoeae in New Caledonia: Evidence of a Novel Genotype Associated with Reduced Penicillin Susceptibility

N. gonorrhoeae isolates.A total of 208 isolates of N. gonorrhoeae isolated from February 2005 through October 2007 were examined in this study. These strains were isolated at the bacteriology laboratory of the Institut Pasteur in New Caledonia from (i) persons attending a sexually transmitted disease clinic; (ii) their partner(s) when possible (even if asymptomatic); and (iii) symptomatic patients consulting private doctors or the hospital. Samples were collected from cervical or urethral swabs from female (111) and male (96) patients, respectively, and included one sample from an eye culture from a newborn baby. None of these strains was knowingly isolated posttreatment or twice from the same patient. The identity of the organism was confirmed by Gram staining, oxidase activity, and specific characteristics on API NH (BioMérieux SA) after culture on chocolate agar supplemented with PolyVitex plus colistin, vancomycin, amphotericin B, and trimethoprim antibiotics (BioMérieux SA, Marcy l'Etoile, France). The isolates were suspended into homemade 2-sucrose phosphate (sucrase, 200 mM; K₂HPO₄, 12 mM; KH₂PO₄, 8 mM) transport medium (supplemented with vancomycin, gentamicin, and fungizone) and stored at −20°C until DNA extraction.

Archival isolates in our collection classified as chromosomally mediated resistant N. gonorrhoeae (CMRNG) and 12 N. gonorrhoeae reference strains (WHO-C, WHO-G, WHO-L, WHO-M, WHO-O, WHO-P, and 07QA-01 through 07QA-06) kindly provided by John Tapsall, Australian Reference Center and WHO Collaborating Center, were used as controls for antimicrobial susceptibility testing.

Antimicrobial susceptibility testing and penicillin MICs.In vitro susceptibility to conventional antibiotics was tested by using a disk agar diffusion method, and β-lactamase production was analyzed by using nitrocefin discs (BioMérieux SA). MICs (μg/ml) for penicillin G were determined by using the E-test diffusion method (AB Biodisk, Solna, Sweden) on chocolate agar plus PolyVitex and were interpreted by directly reading the intercept of the penicillin gradient strip and the zone of inhibition. The Antibiogram Committee of the French Society for Microbiology (CA-SFM) recommendations for N. gonorrhoeae were followed (susceptible strain, MIC of ≤0.06 μg/ml, and resistant strain, MIC of >1 μg/ml) (28). MICs lower than 0.016 μg/ml (the lowest value on the E-test scale) were given a value of 0.004 μg/ml to allow statistical analysis as a continuous variable and, notably, the calculation of average MICs. The data were then analyzed by using StatView software for Windows (SAS Institute, Inc.).

Real-time PCR genotyping assay.Genomic DNA for PCR amplification and genotyping analysis was isolated from 0.2 ml of each thawed N. gonorrhoeae isolate by using a QIAamp DNA minikit (Qiagen, Doncaster, Victoria, Australia) according to the manufacturer's instructions. The DNA concentration for each specimen was adjusted to 1 μg/ml.

The additional aspartic acid residue (Asp-345a) in the penA gene and the single-base substitution in the ponA gene were identified by using a duplex PCR protocol, previously described (39), on a LightCycler 2.0 real-time PCR system (Roche Molecular Biochemicals, Mannheim, Germany). Primers and hybridization probes for genotyping the penB and mtrR determinants were designed by using LightCycler Probe Design software, version 2.0 (Roche Diagnostics), and were synthesized by Sigma-Proligo (Singapore Pty. Ltd.). The primers for penB, the mtrR promoter, and the region encompassing G45 in the coding sequence of mtrR amplified products of 754, 760, and 638 bp, respectively. The probes were designed to target the −T deletion in the inverted repeat of the mtrR promoter; mutations of the G45 codon in the DNA-binding region of mtrR; or codons 120 and 121 of the coding region of the porB1b gene (penB determinant), corresponding to amino acids 101 and 102 of the mature protein, respectively. An unusual system of three hybridization probes targeting the penB determinant was designed: (i) one single-anchor probe labeled with fluorescein at its 3′ end; (ii) a sensor probe specific to the sequence G101K/A102D that was labeled with LC Red-705 at its 5′ end; and (iii) a second sensor probe specific to the wild-type sequence of porB1b that was labeled with LC Red-640 at its 5′ end. For analysis of the −T deletion in the mtrR promoter, an anchor probe labeled with LC Red-670 at its 5′ end was combined with a sensor probe labeled with fluorescein at its 3′ end. For mtrR G45 codon genotyping, a sensor probe specific for G45D labeled with LC Red-640 at its 5′ end was combined with an anchor probe labeled with fluorescein at its 3′ end. The primer and probe sequences are summarized in Table 1.

Genotyping of both penB and the mtrR promoter was conducted in a single multiplex amplification run. Each PCR mixture (20 μl) contained 1× buffer from a LightCycler FastStart DNA master HybProbe kit (Roche Diagnostics), 4 mM MgCl₂, 0.5 μM of each forward primer, 1.0 μM of each reverse primer, 0.2 μM of each probe, and 2 μl template DNA. The amplification parameters were as follows: an initial enzyme activation step at 95°C for 10 min followed by 50 sequential cycles of denaturation at 95°C for 8 s, primer and probe annealing at 60°C for 8 s (55°C for 10 s for mtrR G45 codon mutations), and elongation at 72°C for 30 s. The subsequent melting-curve analysis consisted of heating PCR products up to 95°C for 5 s, cooling them at 40°C for 15 s, and finally, slowly heating them (0.1°C/s) up to 85°C with continuous reading of fluorescence. Controls in every experiment included a blank capillary, wild-type DNA, and mutated genotype DNA of known sequence. This protocol ensures multiplex real-time PCR genotyping after a compensation color has been activated to correct emission spectra overlaps of the dyes.

All genotyping assays were performed after a compensation color had been activated, following LightCycler software instructions, and this correction program was updated with each new batch of probes.

Sequencing of the porB1b and mtrR coding regions.A panel (n = 10 to 15 when possible) of DNAs from N. gonorrhoeae strains representative of each group of samples with identical melting-temperature (T_m) values after melting-curve analysis was amplified as described above, purified using a MinElute PCR purification kit (Qiagen, Australia) according to the manufacturer's instructions, and sequenced at the Waikato University sequencing facilities (Hamilton, New Zealand) using the same primers (Table 1). Multiple-sequence alignments were performed by using ClustalW on BioEdit software, version 7.0.9.0 (15).

Transformation of the novel porB1b GS allele.Transformation experiments were carried out essentially as described by Ropp et al. (25). Piliated colonies of the recipient strain, FA19 penA mtrR (FA19 previously transformed with the penA and mtrR alleles from strain FA6140, a CMRNG strain [6]), were passaged on a fresh GC broth (Difco, Detroit, MI) agar plate and grown overnight. The next day, the cells were swabbed from the plate, resuspended in GC broth with supplements I and II and 10 mM MgCl₂, and diluted to an optical density at 590 nm of 0.18. Cells were incubated for 5 h with 5 μg of a PCR product encompassing the entire porB1b coding sequence and ∼500 bp of flanking sequence at both the 5′ and 3′ ends amplified from a strain harboring the GS allele of porB1b. Aliquots of the cells were plated on GC broth plates containing penicillin at concentrations just above the MIC of the recipient strain (0.25 μg/ml) and allowed to grow overnight. The sequence of the porB gene from resistant transformants was verified by PCR amplification and sequencing, and the MIC of penicillin was determined as described previously (25).

Genotyping assays.The protocols using the fluorescent hybridization probe format were optimized by testing various primer and probe sequences in order to obtain clearly distinct and reproducible melting peaks after melting-curve analyses. These peaks, which reflect the T_m of each hybrid sensor probe-complement strand duplex of the PCR product, were automatically generated by LightCycler software from plotting of the negative derivative of the fluorescence ratio versus temperature. Representative profiles of the melting curves are shown in Fig. 1. According to the selected fluorescence channel, we identified specific T_m values that corresponded to a specific nucleotide sequence as described previously for detection of penA and ponA mutations (38, 39). For each gene, PCR products from a panel of 10 to 15 isolates classified according to their T_ms obtained with our multiplex PCR assay were sequenced and confirmed to be the genotypes evidenced by our hybridization probe genotyping assay.

• Open in new tab
• Download powerpoint

FIG. 1.

Melting peaks derived from negative derivatives of fluorescence ratios with respect to temperature of the amplicons obtained for penB and mtrR promoter genotyping. (A) Readings of wild-type versus mutated genotypes of porB1b alleles (705 nm/530 nm). (B) Discrimination between the wild-type (GA) and the G101D mutants (DA) (640 nm/530 nm). Both of these spectra were generated using color compensation as detailed in the text. (C) Discrimination between mtrR promoter genotypes (670 nm/530 nm).

Genotyping of the penB determinant.Following amplification, fluorescein- and LC Red-705-labeled probes for the porB1b alleles (penB) produced four distinct peaks, each of which indicated a gene sequence of interest. The amplified fragments containing sequences encoding G101K/A102D (KD) had the highest T_m (67.0 ± 0.5°C [mean ± standard deviation]), which reflects 100% homology between probe and target. Three other distinct peaks at 64.0°C (±0.2°C), 60.0°C (±1.0°C), and 49.0°C (±0.5°C) were shown to correspond to nucleotide sequences encoding G101K/A102N (KN); wild-type (GA) or G101D (DA) (these two had the same T_m value on the 705-nm channel); and A102S (GS), respectively. Variations in T_m values (0.2 to 1.0°C) within each genotype were always much lower than between genotypes. Thus, all melting peaks were easily grouped into specific genotypes, with the exception of the wild type (GA) and G101D (DA), which could only be differentiated with the LC Red-640 sensor probe. With this probe, the wild-type sequence had a T_m of 65.0 ± 1.0°C and was clearly distinct from the G101D sequence, which had a T_m of 57.0 ± 0.5°C. Isolates that did not display fluorescence either during amplification or during the melting-point analysis in the 640- and 705-nm channels were all confirmed to have porB1a alleles, based on the difference in size following agarose gel electrophoresis (692 bp instead of 754 bp).

Genotyping of the mtrR promoter −T deletion and mtrR G45 codon mutations.Using fluorescein and LC Red-670 probes, we observed two distinct melting peaks with reference strains that contained either wild-type sequence or the single-nucleotide −T deletion in the mtrR promoter. All isolates with the wild-type sequence produced a discernible peak with a T_m value of 67.0 ± 0.5°C, while those that contained the −T deletion produced a peak with a T_m value of 64.5 ± 0.3°C.

The assay for genotyping mutations at the G45 codon within the mtrR coding region allowed reproducible typing with clearly distinct melting peaks with T_m values of 61.7 ± 0.5°C for the wild type, 69.4 ± 0.2°C for G45D mutants, and 56.6 ± 0.2°C for G45S mutants (codon change of GGC to AGC demonstrated by sequencing) (data not shown).

Phenotypes and genotypes identified.All isolates from our collection covering an almost 3-year period were non-penicillinase producing and displayed susceptible to intermediate resistance to penicillin, with MICs ranging from <0.016 μg/ml to 0.38 μg/ml (Fig. 2). A total of 125 isolates (60.1%) were penicillin susceptible and inhibited by concentrations lower than 0.064 μg/ml, whereas 83 isolates (39.9%) were classified as intermediate-level resistant, for which the MIC of penicillin was in the range of 0.064 to 0.38 μg/ml. Additionally, all isolates were susceptible to quinolones (i.e., nalidixic acid and ciprofloxacin) and to expanded-spectrum cephalosporins (ceftriaxone).

• Open in new tab
• Download powerpoint

FIG. 2.

Penicillin G (Peni G) MICs of the N. gonorrhoeae isolates according to their year of isolation. Gray and black bars indicate susceptible and intermediate levels of MIC, respectively. MICs are expressed as μg/ml.

The genotypes of the 208 strains of N. gonorrhoeae, which are summarized in Table 2, led to the identification of 95 isolates harboring the Asp-345a codon insertion in penA (45.7%) and 33 isolates with the ponA L421P mutation (15.9%). The porB genotypes of our collection were as follows: the porB1a allele was observed in 30 isolates (14.4%), the wild-type porB1b (GA) allele in 100 isolates (48.1%), the single mutation G101D (DA) in 22 isolates (10.6%), and the single mutation A102S (GS) in 56 isolates (26.9%). The G101K/A102N (KN) and G101K/A102D (KD) alleles were not found in any of our isolates. No mtrR promoter −T deletion was observed in any isolate. Out of 83 isolates displaying an intermediate resistance to penicillin (MIC ≥ 0.06 μg/ml), none displayed the G45D mutation and only two harbored a G45S mutation. Out of the 208 isolates of our collection, 54 strains (26%) were identical to wild-type sequences in all of the assays carried out. The other genotypes were not randomly distributed over our strain collection; while there is no clear association between the porB1a or wild-type porB1b (GA) allele and the genotypes of other determinants, the G101D (DA) and A102S (GS) genotypes appear strongly linked with wild-type and Asp-345a penA genotypes, respectively (χ² = 81.993; P < 0.001). Therefore, the effect of these porB1b genotypes on penicillin susceptibility cannot be easily evaluated in our strain collection.

Interestingly, out of the 54 strains genotyped as wild type for all four determinants, 52 displayed penicillin-susceptible and 2 displayed intermediate penicillin-resistant phenotypes (MICs of 0.064 and 0.094 μg/ml). Taken together, the average MICs of strains with the porB1a allele and the wild-type (GA) and G101D (DA) porB1b genotypes were 0.031, 0.037, and 0.016 μg/ml, respectively, whereas the average MIC of strains with the A102S (GS) genotype (0.155 μg/ml) was significantly higher (F = 52.419; P < 0.001). Similarly, strains harboring the Asp-345a codon insertion in the penA gene exhibited a higher penicillin MIC (0.122 versus 0.019 μg/ml; P < 0.001; Student's t test, 11.694) than those without the insertion, whereas the MICs of strains with the ponA mutation were not different from those of the wild type (0.059 versus 0.067 μg/ml; P = 0.62; Student's t test, −0.488). Thus, a penA-porB GS (A102S) genotype predicts a significantly higher penicillin MIC than the other genotypes represented in our strain selection (Table 2).

In contrast to the strains in our collection, all of the archived CMRNG isolates displaying high-level penicillin resistance harbored both the −T deletion in the mtrR promoter and/or the G101K/A102N (KN) or G101K/A102D (KD) porB1b genotype. Interestingly, all of these archive isolates originated from infections obtained overseas.

Determination of the resistance phenotype of the porB1b GS allele (A102S).Although the porB1b GS allele is correlated with significantly higher penicillin MICs than the other genotypes, the A102 mutation has not been previously associated with resistance. To determine directly whether this allele is involved in resistance, we transformed strain FA19 penA mtrR with PCR products encompassing the porB1b gene from one of the intermediate resistant strains harboring the GS allele and selected for transformants on penicillin concentrations slightly higher than the MIC of the recipient strain (MIC = 0.25 μg/ml). Multiple transformants were obtained with PCR products amplified from the GS strain and from FA6140 (a CMRNG strain with a KD allele), but not from the porB1b gene from FA1090 (GA). The GS allele transformants exhibited increased penicillin MICs (0.63 μg/ml), but not to the level of the recipient strain transformed with the KD allele (MIC = 1.0 μg/ml). These data verify that the GS alleles are penB determinants, although they do not appear to have as strong a resistance phenotype as the KD allele.

Temporal trend.The mean MICs for penicillin were 0.100 μg/ml, 0.068 μg/ml, and 0.049 μg/ml for 2005 (35 isolates), 2006 (90 isolates), and 2007 (83 isolates), respectively (Fig. 2). Interestingly, there was a significant decrease in MICs between 2005 and 2006 and between 2005 and 2007 (analysis of variance and Fischer's protected least significant difference test, P = 0.0441 and P = 0.0015), but not between 2006 and 2007 (P = 0.108). The increased proportion of strains displaying MICs lower than 0.016 μg/ml confirms this trend, with this category representing 17%, 19%, and 36% of the isolates from 2005, 2006, and 2007, respectively. However, from a therapeutic point of view, there is no difference between the successive years, as the isolates being identified as having intermediate resistance accounted for 37%, 43%, and 37% of the total isolates in 2005, 2006, and 2007, respectively. Regarding genotypes, there is no evidence of a particular temporal evolution when considering either single-gene or multiple-gene genotypes. There is an apparent increased proportion of porB1a alleles in the 2007 isolates, reaching 18.1% in 2007, whereas the proportions were 14.3% and 11.1% in 2005 and 2006, respectively, but this difference is also not significant (χ² = 1.696; P = 0.4283).