ABSTRACT
Two Neisseria gonorrhoeae isolates from Seattle and two isolates from Uruguay were resistant to erythromycin (MIC, 4 to 16 μg/ml) and had reduced susceptibility to azithromycin (MIC, 1 to 4 μg/ml) due to the presence of the self-mobile rRNA methylase gene(s)ermF or ermB and ermF. The two Seattle isolates and one isolate from Uruguay were multiresistant, carrying either the 25.2-MDa tetM-containing plasmid (Seattle) or a β-lactamase plasmid (Uruguay). Sixteen commensal Neisseria isolates (10 Neisseria perflava-N. sicca, 2 N. flava, and 4N. mucosa) for which erythromycin MICs were 4 to 16 μg/ml were shown to carry one or more known rRNA methylase genes, includingermB, ermC, and/or ermF. Many of these isolates also were multiresistant and carried thetetM gene. This is the first time that a complete transposon or a complete conjugative transposon carrying an antibiotic resistance gene has been described for the genus Neisseria.
Neisseria gonorrhoeaeisolates obtained in Denver, Colo., and Edinburgh, United Kingdom, and having high-level resistance to erythromycin (MIC, >8 μg/ml) and reduced susceptibility to azithromycin (MIC, 2 to 4 μg/ml) have been described (4, 33). The MICs for these isolates were higher than those normally associated with chromosomal mtrmutations (3, 8, 13, 15, 30). Unfortunately, these isolates were not available for examination of the mechanisms of resistance. During a gonorrhea outbreak in 1994 to 95 in Seattle, Wash., caused by strains containing the 25.2-MDa plasmid encoding tetracycline resistance of the Pro−/IA-1,2 class, two isolates resistant to both tetracycline and erythromycin (MICs, ≥16 μg/ml) were identified. Two additional gonococci, isolated in 1991 and 1995 in Uruguay and for which an erythromycin MIC (4 μg/ml) higher than that previously found in this setting was determined, were available for study. In addition to the identification of these isolates with high-level erythromycin resistance (4, 33), plasmids carrying an ermC gene (34) and conferring erythromycin resistance to both N. gonorrhoeae andN. meningitidis have been created. These findings prompted us to evaluate whether these four N. gonorrhoeaeisolates had acquired one or more of the erm genes known for other urogenital species (2). Investigations further sought to define the location of these methylase genes (plasmid versus chromosome), to determine whether their location was on conjugative units, as have been found in many other species (2, 18, 19, 26,29), and to examine the transfer of such genes to other isolates and species. The methylase gene composition of oral commensalNeisseria spp. for which the erythromycin MICs were 4 to 16 μg/ml was compared to that of the gonococcal isolates.
MATERIALS AND METHODS
Bacterial isolates.Erythromycin-resistant (Emr) Pro−/IA-1,2 N. gonorrhoeae isolates were isolated in Seattle during the 1994-1995 gonococcal outbreak (Table1). Both Seattle isolates (94-965 and 95-1) were also tetracycline resistant. The 1995 Uruguay isolate, 581, was Pro−/IB-3, and the 1991 Uruguay isolate, 1101, was nonrequiring Proto/IB-3. Strain 1101 carried the 3.2-MDa β-lactamase plasmid and was resistant to penicillin in addition to erythromycin, while all the other N. gonorrhoeae isolates did not carry a β-lactamase plasmid. Six other TcrPro−/IA-1,2 N. gonorrhoeae isolates that were from the Seattle outbreak but that were not resistant to erythromycin were available for comparison with the two EmrTcr N. gonorrhoeae isolates. The NRL (Neisseria Reference Laboratory, University of Washington, Seattle) strains were isolated prior to 1986. The N. gonorrhoeae isolates were confirmed by biochemical methods (11). Auxotypes, protein I serovars, and plasmid contents of the gonococcal isolates were determined by established methods (5,6, 9, 32).
MICs for and antibiotic resistance determinants ofN. gonorrhoeae and commensal Neisseria spp.
We also examined 16 isolates of commensal Neisseria spp., including 10 N. perflava-N. sicca, 2 N. flava, and 4 N. mucosa (Table 1). These isolates were confirmed by biochemical methods. The commensal Neisseriaspp. were clinical isolates collected from the periodontal pockets of seven periodontitis patients seen at the Graduate Periodontics Clinic at the University of Washington, Seattle, between 1991 and 1995, isolates collected from oropharyngeal specimens from six patients attending the DeKalb County Sexually Transmitted Disease Clinic in 1986 (designated CTM before the number [7]), and type strains (NRL strains) obtained from Joan Knapp (22, 23).
Media.GC base or GCP broth (Difco Laboratories, Detroit, Mich.) supplemented as previously described (9, 20) was used for routine culturing of N. gonorrhoeae,Neisseria spp., and Enterococcus faecalis.
Antimicrobial susceptibilities.Mueller-Hinton medium (Difco) was used to determine the MICs for the commensalNeisseria spp., N. meningiditis, andE. faecalis transconjugants, and supplemented GC medium base (Difco) was used for N. gonorrhoeae, as recommended by the National Committee on Clinical Laboratory Standards for aerobic bacteria (12). The antibiotic concentrations tested were as follows: erythromycin, 0.06 to 32 μg/ml; azithromycin, 0.03 to 8 μg/ml; and tetracycline, 0.06 to 32 μg/ml. MIC plates were incubated at 36.5°C for 24 h with CO2 for N. gonorrhoeae and without CO2 for commensalNeisseria spp., N. meningitidis, and E. faecalis transconjugants.
Proteinase K treatment.Isolates and transconjugants were treated with proteinase K as previously described (2) and used as templates for the PCR assays. Each proteinase K-treated sample was not used more than three times, since repeated freezing and thawing has been shown to degrade DNA samples (2).
PCR of the ermF gene.The PCR primers used in the study were F1 (5′ CGGGTCAGCACTTTACTATTG 3′, starting at bp 1235) and F2 (5′ GGACCTACCTCATAGACAAG 3′, antisense sequence ending at bp 1700). The expected size of the PCR fragment was 466 bp (2, 14,28). Each 100-μl reaction mixture contained 2 U ofTaq polymerase (Boehringer Mannheim Indianapolis, Ind.), 200 mM deoxynucleoside triphosphate, 1× PCR buffer I (1.5 mM MgCl2), and 100 ng of each primer. Ten to 40 ng of DNA or 1 to 2 μl of proteinase K-treated bacteria were used as the DNA template. The PCR conditions were as follows: denaturing at 94°C for 30 s, annealing at 50°C for 30 s, and elongation at 72°C for 2 min. The cycle was repeated 35 times. Plasmid pBF4 (2), containing the cloned ermF gene, and water were used as positive and negative controls, respectively. The PCR products were dried on a lyophilizer, resuspended in 10 μl of sterile H2O, run on 1.5% agarose gels, and stained with ethidium bromide for visualization. Southern blots of these gels were hybridized with labeled ermF-containing plasmid probes for confirmation of PCR products as previously described (2).
PCR primers and conditions for the ermA,ermB, and ermC genes.AF(5′ CTTCGATAGTTTATTAATATTAGT 3′) and AR (5′ TCTAAAAAGCATGTAAAAGAA 3′), BF (5′ AGTAACGGTACTTAAATTGTTTAC 3′) and BR (5′ GAAAAGGTACTCAACCAAATA 3′), and CF (5′ GCTAATATTGTTTAAATCGTCAAT 3′) and CR (5′ TCAAAACATAATATAGATAAA 3′) have been described previously (2). The PCR conditions for the ermB reaction were the same as those for the ermF reaction. The PCR assay used for ermA consisted of denaturing at 94°C for 30 s, annealing at 48°C for 1 min, and elongation at 72°C for 2 min; that used for ermC consisted of denaturing at 94°C for 30 s, annealing at 43°C for 1 min, and elongation at 72°C for 2 min.
DNA hybridization.DNA was extracted from N. gonorrhoeae, commensal Neisseria spp., N. meningitidis, and selected transconjugants as previously described (9, 20). Uncut whole-cell DNA was visualized on a 0.7% agarose gel stained with ethidium bromide, and Southern blots were prepared. Fragment probes prepared from rRNA methylase genes from the cloned plasmids pEM9592, pJIR229, pBR328:33RV, pBF4, and pJI3, which carried the genes ermA, ermB, ermC,ermF, and tetM, respectively, or oligonucleotide probes for the appropriate genes were used (2, 16). The DNA probes were labeled with the appropriate Genius 3 chemiluminescence kit as recommended by the manufacturer (Boehringer). Hybridization under stringent conditions and detection were done according to the manufacturer’s instructions as previously described (2,19). Positive and negative controls were included in each Southern blot.
Hybridization of PCR products.Plasmids pEM9592, pJIR229, pBR328:33RV, and pBF4 or oligonucleotide probes for ermA,ermB, ermC, and ermF were labeled with nonradioactive Genius kits as recommended by the manufacturer (Boehringer). The labeled plasmids were used for hybridization with Southern blots of the appropriate PCR product or purified whole-cell DNA. The hybridization and wash steps were performed at stringent temperatures according to the manufacturer’s instructions. Detection was done with a CDP-Star detection kit at a reagent concentration of 1:1,000 as described by the manufacturer (Boehringer).
Sequencing.The ermF PCR products from N. gonorrhoeae and commensal Neisseria spp. were sequenced separately with primers ermF 1 andermF 2. A Taq Dye Deoxy terminator cycle sequencing kit (Applied Biosystems, Foster City, Calif.) was used for PCR amplification, and the filtered PCR products (Nuclean D50 filters; Kodak, Rochester, N.Y.) were examined on a model 373A sequencer (Applied Biosystems) (2, 19). The two sequences for each isolate were overlapped, aligned and compared with the known GenBank sequence of ermF (accession no. M14730) by use of GCG software (Genetics Computer Group, Madison, Wis.). The putative amino acid sequences were determined from the DNA sequences and also compared with the known GenBank sequence of ErmF( 19).
Mating experiments.Recipients included N. gonorrhoeae F62, with chromosomally mediated resistance to rifampin (25 μg/ml), streptomycin (250 μg/ml), and nalidixic acid (25 μg/ml) (21, 22, 27); N. gonorrhoeae CDC36N, with chromosomally mediated resistance to nalidixic acid (25 μg/ml) and carrying the 4.4-MDa β-lactamase plasmid (20);E. faecalis JH2-2, resistant to rifampin (25 μg/ml) and fusidic acid (25 μg/ml) (2, 19, 24, 26); N. meningitidis NRL9205 (serogroup A), resistant to streptomycin (250 μg/ml) and rifampin (20 μg/ml) (23); and N. mucosa CTM 1.1, with chromosomally mediated resistance to streptomycin (250 μg/ml) and rifampin (20 μg/ml). Donors included Emr N. gonorrhoeae isolates and selected isolates from each of the commensal Neisseria species. Donors and recipients were grown separately for 24 h on agar plates. The donor and recipient isolates were each resuspended in 0.5 ml of GCP broth to form turbid suspensions (>108/ml), mixed together, and plated on a GC agar (Difco) plate without antibiotics as previously described (22, 23, 27). The mixture was incubated at 36.5°C in 5% CO2 for 24 h.N. meningitidis, N. mucosa, and N. gonorrhoeae F62 transconjugants were selected on medium containing streptomycin (150 μg/ml) and erythromycin (10 μg/ml). The transconjugants were confirmed by growth on rifampin (25 μg/ml) (21-23). N. gonorrhoeae CDC36N transconjugants were selected on medium containing penicillin (10 μg/ml) and erythromycin (10 μg/ml). The transconjugants were verified by growth on nalidixic acid (25 μg/ml) and the presence of the 4.4-MDa β-lactamase plasmid (21-23, 27). JH2-2 transconjugants were selected on medium containing rifampin (10 μg/ml) and erythromycin (10 μg/ml). The E. faecalis transconjugants were confirmed by growth on medium supplemented with streptomycin (150 μg/ml) and by use of chromosomal DNA probe specific for E. faecalis (2). N. meningitidistransconjugants were confirmed by growth on medium supplemented with rifampin (20 μg/ml).
The identity of erm genes in the transconjugants was confirmed by PCR and hybridization of the PCR products as described above (2).
PFGE.Pulsed-field gel electrophoresis (PFGE) was used to compare the Emr N. gonorrhoeae isolates to six Tcr Pro−/IA-1,2 Seattle N. gonorrhoeae isolates which were part of the outbreak. The isolates were digested with NheI or SpeI (Promega, Madison, Wis.) as previously described (31, 32). The PFGE patterns were compared and assumed to be genetically related if they were identical or had three or fewer band differences.
PFGE was also used to compare the N. meningitidistransconjugants with the donor and recipient N. meningitidisisolates by use of the N. gonorrhoeae protocol and one enzyme (31, 32). This procedure allowed us to verify that the N. meningitidis transconjugants were related to the recipient rather than the donor N. meningitidis.
RESULTS
Characterization of macrolide-resistant N. gonorrhoeae. The two Seattle Pro−/IA-1,2 isolates were identified because of the high MICs of erythromycin (16 μg/ml) and azithromycin (4 μg/ml) for them (Table 1). Both Seattle isolates carried 25.2-MDa plasmids (tetracycline MIC, 16 μg/ml) which hybridized with thetetM probe (data not shown). For the two Uruguay isolates, the erythromycin MIC was 4 μg/ml and the azithromycin MIC was 1 μg/ml (Table 1). Uruguay isolate 1101 was resistant to penicillin and carried a 3.2-MDa β-lactamase plasmid (data not shown). All fourN. gonorrhoeae isolates carried an ermF gene, which encodes a known rRNA methylase, and one isolate (95-1) also carried ermB (Table 1). The other isolates did not hybridize with ermA, ermB, or ermC gene probes, while 95-1 did not hybridize with ermA or ermCgene probes (data not shown). The ermF probe hybridized with the chromosomal fraction of the gel when whole-cell DNA was used for the Southern blots, suggesting a chromosomal location for theermF gene. In addition, the 2.6- and 3.2-MDa β-lactamase plasmids or the 25.2-MDa tetM-containing plasmid common toN. gonorrhoeae was found, but no other plasmids were found in any of the four isolates.
PCR fragments of the ermF genes from two strains of N. gonorrhoeae (94-965 and 1101) were sequenced. The DNA sequence and amino acid homologies between the PCR fragment from 94-965 and theermF gene originally identified in colonicBacteroides spp. were both 99% identical over 374 bp (Fig.1 and 2). Results were similar (95% identity) for N. gonorrhoeae 1101 from Uruguay (data not shown). The G+C content of the PCR fragment was approximately 35%, which differs from the 50% G+C content found in the N. gonorrhoeae chromosome, suggesting a non-Neisseria origin for these genes. PCR fragments of theermB gene from strain 95-1 were also sequenced. The DNA sequence and amino acid homologies between the PCR fragment and theermB gene previously characterized from Clostridium perfringens (GenBank accession no. X58285) were over 99% identical over 342 bp.
DNA sequence homology between the GenBankermF sequence (listed as m14730) and the PCR product fromN. gonorrhoeae 94-965 (listed as 94-F1) (99% identity over 374 bp).
Amino acid homology between the GenBank ermFsequence (listed as m14730) and the PCR product from N. gonorrhoeae 94-965 (listed as 94.pep) (99% identity over 123 amino acids).
The six Tcr Pro−/IA-1,2 N. gonorrhoeae isolates obtained during the 1994-1995 Seattle outbreak had NheI PFGE patterns (Fig.3, lanes 1 to 6) that were indistinguishable from the NheI PFGE patterns of the two Emr Tcr N. gonorrhoeae isolates (Fig. 3, lanes 7 and 8), but the Seattle Emr isolates differed from the two Emr N. gonorrhoeaeisolates from Uruguay (Fig. 3, lanes 9 and 10). The two Uruguay isolates appeared unrelated to the Seattle isolates or to each other (the PFGE patterns differed by more than three fragments) (Fig. 3). Similar results were found when SpeI was used for PFGE analysis (data not shown). Both enzymes gave identical patterns for the eight Seattle isolates, strongly suggesting a very close relationship between the two Emr and the six Ems N. gonorrhoeae isolates from the outbreak.
PFGE with NheI. Lanes 1 to 6, Tcr N. gonorrhoeae, with isolates in lanes 1 to 3 being isolated before and isolates in lanes 4 to 6 being isolated after the Emr Tcr N. gonorrhoeae isolates from Seattle; lanes 7 and 8, Emr Tcr N. gonorrhoeae isolates from Seattle; lanes 9 and 10, Emrisolates from Uruguay; lane 11, λ standard. Numbers at right are molecular weight standards.
Characterization of macrolide resistance in commensalNeisseria spp.The erythromycin MICs for the otherNeisseria spp. ranged from 4 to 16 μg/ml (Table 1). Compared to the N. gonorrhoeae isolates studied, these commensal species contained a more heterologous group of knownerm genes (ermB, ermC, andermF). Among the 10 N. perflava-N. sicca isolates, 4 carried the ermB gene and had erythromycin MICs ranging from 4 to 8 μg/ml; 2 carriedermC and had an erythromycin MIC of 16 μg/ml; 2 carried both ermB and ermC and had erythromycin MICs of 4 to 16 μg/ml; and 2 carried ermB, ermC, andermF and had erythromycin MICs of 4 to 16 μg/ml (Table 1). Among the four N. perflava-N. sicca strains isolated before 1990, three carried one erm gene, while three of six strains isolated after 1990 carried multipleerm genes. One N. flava strain carriedermC, and the other strain carried ermB(erythromycin MICs, 8 to 16 μg/ml). Among the four N. mucosa strains, two carried both ermB andermC, one carried ermB, and one carriedermC; the erythromycin MIC for all four strains was 8 μg/ml (Table 1).
To confirm the presence of the erm genes, we used PCR sequencing. The PCR fragment of the ermF gene fromN. perflava-N. sicca 10915 was sequenced; the DNA sequence homology between the PCR fragment and theermF gene from Bacteroides spp. showed 97% identity over 374 bp, and the amino acid homology was 94% (data not shown).
Transfer of erythromycin resistance.All four of the N. gonorrhoeae isolates and seven of the commensalNeisseria sp. isolates were examined for their ability to transfer the Emr phenotype to Neisseria andE. faecalis recipients (Table2). N. gonorrhoeae donors transferred the ermF gene at frequencies of 10−6/recipient with the two different N. gonorrhoeae recipients, 10−7/recipient with N. meningitidis as the recipient, and 10−7 to 10−8/recipient with E. faecalis as the recipient.
Mobility of erm genes in representative transconjugants
The commensal species carried a variety of erm genes and were able to transfer ermF, ermC andermF, ermB and ermC, all threeermC, or ermB (Table 2) at frequencies ranging from 10−5 to 10−9 for E. faecalis and N. meningitidis. Matings were done at least twice, and only a portion of the transconjugants were characterized and described in Table 2. The donor N. mucosaCTM 2.2 could move the ermB gene but not the ermCgene to the recipient N. mucosa CTM 1.1 (Table 2) at a frequency of 10−8/recipient. The other N. mucosa donor (CTM 8.1) and the various N. perflava-N. sicca and N. flava donors used in the matings, which transferred erm genes to E. faecalis and/or N. meningitidis recipients, could not transfer erm genes at measurable frequencies (>10−9/recipient) to the recipient N. mucosaCTM 1.1 (Table 2). Both E. faecalis and N. meningitidis recipients were able to acquire one or moreerm genes. No N. meningitidis withermF was isolated from the transconjugants with the commensal donors, but ermF was found in N. meningitidis transconjugants when N. gonorrhoeaecarrying the ermF gene was used as the donor (Table 2).
The 25.2-MDa plasmid, conferring tetracycline resistance, was transferred from a Seattle N. gonorrhoeae donor to N. gonorrhoeae and N. meningitidis recipients but not toE. faecalis recipients (data not shown). No plasmids carrying the tetM gene were found in the commensal species, and we were unable to transfer tetracycline resistance from these species to either N. meningitidis or E. faecalis. However, this result was anticipated, since we have previously shown that commensal Neisseria sp. isolates carry an incompletetetM transposon in the chromosome and were unable to transfer tetM by conjugation (16, 17, 25).
DISCUSSION
This is the first description of a known erm gene(s) in the genus Neisseria, since both the TEM β-lactamase and the tetM genes have incomplete transposons in N. gonorrhoeae, N. meningitidis, and the commensalNeisseria spp. (5, 16, 17). The data indicates that the ermF genes have been in N. gonorrhoeaesince at least 1991, the ermB genes have been in N. gonorrhoeae since 1995, and various erm genes have been in three commensal species (N. perflava-N. sicca,N. flava, and N. mucosa) since at least the 1980s (Table 1). Whether erm genes are relatively new (last 20 years) in Neisseria spp. or whether they predate the identification of the β-lactamase plasmids in N. gonorrhoeae (5) is currently under investigation. Donors carrying the 25.2-MDa plasmid with the tetM gene transferred this gene into N. gonorrhoeae and N. meningitidis recipients but not into E. faecalis (data not shown), indicating that the ermF gene had a wider host range than the gonococcal 25.2-MDa plasmid (17, 22) or the gonococcal 24.5-MDa and β-lactamase plasmids (17, 23). Although 10 (63%) of the commensal Neisseria isolates carried the tetM gene (Table 1), none could move this gene, as has previously been described (16, 17).
N. gonorrhoeae with reduced susceptibility to erythromycin (MICs, 2 to 4 μg/ml) has been reported since the 1960s (1,15). Some studies have shown a positive association between reduced susceptibility to penicillin, erythromycin, chloramphenicol, and tetracycline and mtr mutations (3). It was hypothesized that resistant N. gonorrhoeae isolates for which erythromycin MICs were 2 to 4 μg/ml were due to the presence of mtr mutations (3,8). However, the maximum azithromycin MICs for these isolates generally were 0.25 to 0.5 μg/ml (unpublished observations). Based upon our finding with the two Uruguay isolates, for which the erythromycin MIC was 4 μg/ml and the azithromycin MIC was 1 μg/ml (Table 1), it is tempting to speculate that other N. gonorrhoeae isolates for which erythromycin MICs are 2 to 4 μg/ml also may carry erm genes with or withoutmtr mutations. We are currently examining isolates obtained during different decades and for which erythromycin MICs range from 0.5 to 8 μg/ml. It will be of interest to determine whether the characteristics attributed to the mtr mutations are due to the combination of mtr mutations and ermgenes or whether reduced susceptibility to penicillin and tetracycline is associated with mtr mutations but reduced susceptibility to erythromycin is associated not with mtr mutations but with the presence of erm genes (3). Clinically, this information will be of interest because many infections in homosexual and bisexual men in Seattle-King County (10) are due to N. gonorrhoeae isolates with a phenotypic pattern (reduced susceptibility to erythromycin, penicillin, and tetracycline) suggesting mtr mutations. Some of these isolates have been shown to carry an mtr mutation by sequencing of PCR products. One can speculate that one or more of the four Emr N. gonorrhoeae isolates in this study may carry both erm genes and mtr mutations. However,mtr mutations cannot be transferred by conjugation, nor do they influence the transfer of coresident β-lactamase plasmids. Since the ermF-tetQ transposons in colonic Bacteroidesspp. are able to transfer mobilizable plasmids betweenBacteroides spp., the ability of mobile ermF to conjugally transfer gonococcal β-lactamase plasmids is under investigation (2).
Three of the N. gonorrhoeae isolates and over 60% of the commensal species isolates were multiresistant (Table 1). TheermF gene in N. gonorrhoeae and theermB, ermC, and ermF genes in the commensal species (Table 2) were able to move themselves by conjugation to other Neisseria spp. and to E. faecalisrecipients. This finding implies that these erm genes are associated with complete conjugative elements, and this is the first description of complete transposable elements in Neisseria. Previously described TEM β-lactamase genes and the tetMgene are both on incomplete elements (5, 9, 17).
There was greater diversity among the erm genes carried by the 16 commensal Neisseria isolates than by the 4 N. gonorrhoeae isolates. Within this study, there was no association of erythromycin MIC with the number of erm genes found or with a particular gene among the commensal Neisseria spp. Further studies are needed to determine whether clinical isolates ofN. gonorrhoeae carrying the ermC gene can be found or whether carriage of this gene is unique to the commensal species. Studies are needed to determine the influence of theseerm genes on the treatment of gonococcal disease with the newer macrolides. It also remains to be determined if the commensalNeisseria spp. are reservoirs for these ermgenes, how long these genes have actually been in the genus, whether most erm genes are on mobile conjugative elements, and whether isolates carrying erm genes are more likely to be multiresistant than the general Neisseria population.
ACKNOWLEDGMENT
This work was supported in part by National Institutes of Health grants AI-131448 and DE-10913.
FOOTNOTES
- Received 1 October 1998.
- Returned for modification 20 January 1999.
- Accepted 12 March 1999.
- Copyright © 1999 American Society for Microbiology
