Local emergence and global evolution of Neisseria gonorrhoeae with high-level resistance to azithromycin

ABSTRACT

Antimicrobial resistance in Neisseria gonorrhoeae (Ng) has severely reduced treatment options, including azithromycin (AZM), which had previously been recommended as dual therapy with ceftriaxone. This study characterizes the emergence of high-level resistance to AZM (HLR-AZM) Ng in Baltimore, Maryland, USA, and describes the global evolution of HLR-AZM Ng. Whole genome sequencing (WGS) of 30 Ng isolates with and without HLR-AZM from Baltimore was used to identify clonality and resistance determinants. Publicly available WGS data from global HLR-AZM Ng (n = 286) and the Baltimore HLR-AZM Ng (n = 3) were used to assess the distribution, clonality, and diversity of HLR-AZM Ng. The HLR-AZM Ng isolates from Baltimore identified as multi-locus sequencing typing sequence type (ST) 9363 and likely emerged from circulating strains. ST9363 was the most widely disseminated ST globally represented in eight countries and was associated with sustained transmission events. The number of global HLR-AZM Ng, countries reporting these isolates, and strain diversity increased in the last decade. The majority (89.9%) of global HLR-AZM Ng harbored the A2059G mutation in all four alleles of the 23S rRNA gene, but isolates with two or three A2059G alleles, and alternative HLR-AZM mechanisms were also identified. In conclusion, HLR-AZM in Ng has increased in the last few years, with ST9363 emerging as an important gonococcal lineage globally. The 23S rRNA A2059G mutation is the most common resistance mechanism, but alternative mechanisms are emerging. Continued surveillance of HLR-AZM Ng, especially ST9363, and extensively drug-resistant Ng is warranted.

INTRODUCTION

The rates of gonorrhea continue to increase globally, with an estimated 82.4 million cases in 2020 (1). In the United States, the number of gonorrhea cases reported to the Centers for Disease Control and Prevention (CDC) has increased every year since the lowest point in 2009 (2). Neisseria gonorrhoeae (Ng) easily develops antimicrobial resistance (AMR) against commonly prescribed antimicrobial agents; the CDC and WHO have designated Ng as a high-priority antibiotic-resistant pathogen (3, 4).

To prevent the emergence and spread of AMR in Ng, dual therapy with antibiotics with different mechanisms of action, such as ceftriaxone (CRO, a cephalosporin) and azithromycin (AZM, a macrolide), has been recommended. However, the rapidly increasing rates of Ng isolates with decreased susceptibility or resistance to AZM in Europe (5) and the United States (6) prompted the CDC to discontinue the dual therapy recommendation (7), leaving CRO, at higher dosage, as the only recommended treatment for uncomplicated gonorrhea in the United States. The WHO, however, still recommends AZM as part of a dual therapy regimen when local resistance data are not available (8).

Reports of Ng clinical isolates exhibiting high-level resistance to AZM (HLR-AZM; MIC ≥ 256 µg/mL) have been infrequent and sporadic globally. In the United States, initial reports of HLR-AZM were from Hawaii in 2011 (9) with subsequent cases reported in California in 2014 (10), Hawaii in 2016 (11, 12), and North Carolina in 2018 (13). In Baltimore, Maryland, the CDC-sponsored Gonococcal Isolate Surveillance Program (GISP) has not reported any HLR-AZM Ng isolates through 2013; no surveillance data from Baltimore are available between 2014 and 2019 (14). Although sustained transmission of HLR-AZM Ng at the community level is rare, a cluster of 14 HLR-AZM Ng cases was reported in Indiana during a 13-month period in 2017–2018 (15). Globally, cases of HLR-AZM Ng are also rare, but a very large cluster (>70 cases) of HLR-AZM Ng was reported in England between 2014 and 2017 (16), with three smaller outbreaks in Barcelona, Spain, between 2016 and 2018 (17), and sustained transmission reported in Argentina between 2018 and 2022 (18). These recent outbreaks provide supportive evidence of sustained HLR-AZM Ng transmission and highlight the need for continued surveillance to monitor the evolving AMR Ng epidemic.

Our group has previously reported data on three Ng isolates with HLR-AZM recovered from patients attending sexual health clinics (SCHs) in Baltimore, Maryland, in 2016 (19). We now report the genomic characteristics of these isolates compared to AZM-susceptible isolates from Baltimore and other HLR-AZM Ng isolates collected in the United States. Additionally, using genomic data from global HLR-AZM Ng isolates, we describe and compare the characteristics of these isolates, resistance mechanisms, clonal relatedness, and global evolution of these highly resistant strains.

RESULTS

The Baltimore HLR-AZM Ng isolates shared identical antibiogram, were closely related, and belonged to multi-locus sequence typing sequence type 9363

The three HLR-AZM Ng isolates were recovered from urethral samples of three symptomatic patients at the Baltimore City Health Department (BCHD) SHCs in Baltimore, Maryland, in September 2016. The three men (36–61 years old) were African American, self-identified heterosexual, and HIV and syphilis uninfected at the time of the clinic visit. All of the Ng isolates collected at the BCHD SCHs between January and October 2016 (n = 142), including 24 collected in September, were susceptible to AZM (19). For this study, archived Ng isolates collected at the BCHD SCHs in 2014 and 2015 were analyzed to identify additional HLR-AZM Ng. Antimicrobial susceptibility testing (AST) of Ng isolates collected in 2014 (n = 164) revealed that all isolates were susceptible to AZM (MIC range = 0.023–0.5 µg/mL). Ng isolates from 2015 (n = 172), analyzed by PCR because they were not viable for phenotypic AST, did not harbor the 23S rRNA A2059G mutation, which is highly predictive of HLR-AZM in Ng.

The three HLR-AZM Ng isolates from Baltimore had identical antibiograms and were susceptible to CRO, cefixime, and ciprofloxacin and displayed intermediate resistance to penicillin and resistance to tetracycline. The isolates had the A2059G mutation in all four 23S rRNA alleles but did not harbor the −35A deletion in the mtrR promoter or the mtrR G45D mutation (Table S1). The three HLR-AZM Ng isolates belonged to multi-locus sequence typing (MLST) sequence type (ST) 9363, Ng multi-antigen sequence typing (NG-MAST) ST3935, and appeared to be related to other Baltimore strains (JM0316 and JM0903), which also belonged to MLST ST9363 (Fig. 1; Fig. S1). The Baltimore HLR-AZM Ng isolates were phenotypically and genotypically different from the 2016 HLR-AZM Ng isolates (n = 8) from Hawaii (MLST ST1901) (11, 12) (Fig. 2; Fig. S1; ). However, the Baltimore HLR-AZM Ng isolates were related to other HLR-AZM isolates from Norway (Fig. 1 and 2) and had the same MLST ST (9363) and NG-MAST ST (3935) as HLR-AZM isolates from Argentina, Italy, Norway, and Portugal (Fig. 1; Table S2; Fig. S1). The HLR-AZM isolates from Indianapolis, Indiana, USA, also belonged to MLST ST9363 but differed in their NG-MAST: ST5035 and ST4822 (Fig. 1). Phylogenetic analysis showed that the 30 Ng isolates from Baltimore were diverse, representing 13 different MLST STs (Table S1), which loosely clustered according to antimicrobial susceptibility profiles (Fig. S1; Table S1), and included globally observed genotypes, with local clusters of two to five strains (Fig. 2; Fig. S1). Phenotypic and genotypic data for all Ng isolates from Baltimore can be found in Table S1.

Fig 1.

Maximum-likelihood phylogeny of the 54 global HLR-AZM Ng MLST ST9363 included in the study (13, 15–18, 20–28). AZM-susceptible and non-MLST ST9363 Ng Isolates were included in the tree as they were the nearest neighbors. Isolates from Indianapolis, Indiana, USA, are in boxes. *This study. AZM, azithromycin; SRA, sequence read archive; mtrR, multiple transferable resistance repressor; mtrRP, mtrR promoter; NR, not reported; NP, not performed.

Fig 2.

Maximum-likelihood phylogeny of publicly available Ng genomes, including HLR-AZM Ng. @, HLR-AZM Ng (# of @ represents the number of isolates); AZM-susceptible Ng isolates from Baltimore (the number of isolates is shown). *Pharyngeal and penile isolate from the same patient. ^#All eight HLR-AZM Ng isolates also displayed decreased susceptibility to ceftriaxone. ^%Seven HLR-AZM Ng isolates also displayed decreased susceptibility/resistance to ceftriaxone: ERR2865779, ERR2865780, ERR2560140, ERR2560139 (WHO-Q), SRR21311934, SRR19905848, and SRR1661207. Two HLR-AZM Ng isolates from China with decreased susceptibility to ceftriaxone were excluded because sequencing data were not available.

Ng isolates with HLR-AZM are increasing in frequency and diversity globally

To further characterize the emergence and global evolution of HLR-AZM Ng, data from the Baltimore HLR-AZM Ng isolates (n = 3) and publicly available data from global HLR-AZM Ng (n = 286) were analyzed (Table S2). The HLR-AZM Ng isolates (n = 289) included in the analysis were from 17 countries and predominantly reported from across the United Kingdom (n = 97), China (n = 66), the United States (n = 45), Spain (n = 16), and Argentina (n = 14). Stratification of the number of HLR-AZM Ng isolates by year of isolation—prior to and after 2012 (based on dual CRO and AZM therapy recommendations adopted by many countries beginning in 2012)—identified 63 HLR-AZM Ng isolates from 9 countries between 2001 and 2012 and 226 HLR-AZM Ng isolates from 15 countries between 2013 and 2022. To account for the disproportionate number of HLR-AZM Ng isolates reported by different groups (e.g., the UK between 2014 and 2017), representative lineages based on MLST and/or NG-MAST STs were selected to explore the emergence of lineages prior to and after 2012; 7 MLST STs and 26 NG-MAST STs were reported prior to and including 2012. Between 2013 and 2022, 8 new MLST STs and 45 new NG-MAST STs were reported. Interestingly, 71.4% (5/7) of MLST STs, which existed between 2001 and 2012, was also described between 2013 and 2022, but only 23.1% (6/26) of NG-MAST STs was reported between 2013 and 2022.

Of the 289 HLR-AZM isolates, 71.3% (n = 206) had available whole genome sequencing (WGS) data. Phylogenetic analysis of the 206 HLR-AZM Ng isolates showed that the majority (86.9%%, n = 179) of HLR-AZM Ng genomes grouped into four distinct clades (Fig. 2). Clade 1 consisted of 93 HRL-AZM Ng isolates, the majority (93.5%; n = 87) from the United Kingdom, collected during an outbreak (16) between 2014 and 2017 (Fig. S2). Clade 2 consisted of 46 closely related HLR-AZM Ng isolates from Baltimore, other cities in the United States, and six other countries (Fig. 2). Clade 3 was the least diverse clade with only eight HLR-AZM Ng isolates, which also displayed decreased susceptibility to CRO, from Hawaii collected in 2016 (11, 12). Clade 4 consisted of 32 HLR-AZM Ng isolates from nine countries, including 7 HLR-AZM Ng isolates, which also displayed resistance to CRO (Fig. 2). The remaining (n = 27) HLR-AZM Ng isolates were represented on the phylogeny tree as either representative HLR-AZM Ng single isolates, descendants of AZM-susceptible Ng (data not shown), or grouped into smaller clusters of two to four HLR-AZM Ng isolates (Fig. 2).

Of the 289 HLR-AZM Ng strains, 78.2% (n = 226) had publicly available MLST results and/or WGS data available for MLST analysis and 95.2% (n = 275) had NG-MAST data. The HLR-AZM Ng isolates belonged to 15 MLST STs (Fig. 3A); 53.3% (8/15) of the STs were country specific, and of those, most (62.5%, 5/8) were represented by single isolates from five different countries (Fig. 3A insert). The most common MLST STs were ST1580, ST9363, and ST10899. NG-MAST classified the HLR-AZM Ng isolates into 72 NG-MAST STs (Fig. 3B), the majority (86.1%, 62/72) of STs were unique to specific countries, and 65.3% (47/72) of STs were represented by single isolates (Fig. 3B). The most common NG-MAST STs were ST9768, followed by ST649 and ST1866. ST3935, the fourth most prevalent NG-MAST ST in this analysis, was the most widely disseminated ST globally represented in six countries (Argentina, Italy, Norway, Portugal, Spain, and the United States, including the Baltimore HLR-AZM isolates) (Fig. 3B).

Fig 3.

Molecular typing of global Ng isolates with HLR-AZM. (A) MLST distribution of 226 isolates. (B) NG-MAST distribution of 275 isolates. ■, MLST/NG-MAST ST first reported from Ng isolates prior to and including 2012. ●, MLST/NG-MAST ST first reported during and after 2013. ▲, some of the 3, 5, 11, and 9 NG-MAST STs from Argentina, Australia, China, and the United Kingdom, respectively, were first reported prior to and including 2012, and some were reported during and after 2013. See Table S2 for additional details regarding MLST ST and NG-MAST ST profiles.

Given that dual therapy with CRO and AZM has been used for the treatment of gonorrhea, the data from the HLR-AZM Ng isolates were analyzed for evidence of resistance to CRO. Of the HLR-AZM Ng isolates, 99.7% (288/289) had available CRO and AZM susceptibility data, and of those, 18 (6.3%) isolates from the United States (n = 8), China (n = 4), Australia (n = 2), the United Kingdom (n = 2), Austria (n = 1), and Canada (n = 1) displayed HLR-AZM and decreased susceptibility/resistance to CRO (Fig. 2; Table S2). Fifty percent (n = 9) of isolates with dual AZM and CRO resistance belonged to MLST ST 1901, followed by ST 12039 (n = 4) (Table S2). Notably, one case (two isolates from one patient) from the United Kingdom and two cases from Australia were caused by a single clone of extensively drug-resistant (XDR) Ng with epidemiological links to Asia (29). Decreased susceptibility/resistance to CRO was also assessed by characterizing the penA gene as penA mosaic alleles have been associated with decreased susceptibility to extended spectrum cephalosporins, such as CRO and cefixime (CEF) (30). Characterization of the penA gene of 198 HLR-AZM Ng genomes revealed that 4.6% (n = 9) of genomes had a penA mosaic allele. Five of the nine isolates, two isolates from the United Kingdom (same patient) and two from Australia (29) and one isolate from Austria (31), harbored the penA Type 60 mosaic allele. The remaining four HLR-AZM Ng harbored the penA Type 34 mosaic allele (n = 3, all from Ireland) and the penA Type 131 mosaic allele (one isolate from the United Kingdom).

Ng MLST ST9363 is associated with sustained transmission events globally

HLR-AZM Ng MLST ST9363 accounted for 23.9% (54/226) of all global HLR-AZM Ng included in this analysis. Between 2016, when HLR-AZM Ng MLST ST9363 was first reported, and 2022, 42.5% (54/127) of all HLR-AZM Ng isolates belonged to MLST ST9363. Of the 15 HLR-AZM Ng MLST STs reported in this study, ST9363 was the most widely disseminated ST globally represented in eight countries (Argentina, Italy, Ireland, Norway, the Netherlands, Portugal, Spain, and the United States). Sustained transmission with HLR-AZM Ng MLST ST9363 was common as 57.4% (31/54) of isolates were collected during 2–3-year periods of sustained transmission in Argentina, Indianapolis, Indiana, USA, and Barcelona, Spain. Notably, the majority [75%, (27/36)] of HLR-AZM Ng MLST ST9363 collected during periods of sustained transmission in the United States and Argentina harbored a novel combination of the 23S rRNA A2059G mutation with a mosaic multiple transferable resistance (mosaic mtr) locus (Table 1); this combination of resistance determinants was only observed in HLR-AZM Ng MLST ST9363.

TABLE 1.

Distribution of AMR determinants associated with HLR-AZM in 218 global Ng strains^e

Year	Country	No. of isolates	No. of 23S rRNA A2059G alleles
2001, 2005–2007, 2009–2019, 2022	Argentina, Australia, Austria, Canada, Chile, China, Cyprus, Ireland, Italy, Japan, Norway, Portugal, Spain, Sweden, the Netherlands, United Kingdom, United States	159	4
2017–2022	Argentina, United States	37	4
2004, 2006, 2007–2008, 2015–2017	Ireland, Sweden, United Kingdom	14	3
2015	United Kingdom	1	2
2016	Ireland	1	2
2016	Norway	1	0a
2016	United States	3	0b
2019	United States	1	0c
2019	China	1	New alleled

^a Four alleles of the 23S rRNA C2611T identified during this study.

^b Four alleles of the 23S rRNA A2058G mutation.

^c Three alleles of the 23S rRNA A2058G mutation.

^d The new allele was not identified.

^e For additional details on each of the isolates, please see Table S2.

The 23S rRNA A2059G mutation is the major resistance determinant for HLR-AZM in Ng, but alternative resistance mechanisms are emerging

The distribution of HLR-AZM resistance determinants in the global HLR-AZM Ng isolates is shown in Table 1. Of the 218 HLR-AZM Ng isolates with AZM resistance determinant data, 89.9% (n = 196) harbored the A2059G (Escherichia coli numbering) mutation in all four alleles of the 23S rRNA gene; of those, 81.6% (160/196) had the 23S rRNA A2059G mutation with or without mutation(s) in the mtr repressor (mtrR) or mtrR promoter, which are commonly associated with decreased susceptibility to AZM. In addition to harboring the 23S rRNA A2059G mutation (four alleles), 18.3% (36/196) of HLR-AZM Ng isolates also harbored a nongonococcal mtrR locus (mosaic-mtr). Notably, isolates with the novel combination of 23S rRNA A2059G mutation and mosaic-mtr have only been reported in Argentina and the United States. Two HLR-AZM Ng isolates (one from Ireland and one from the United Kingdom) only harbored the A2059G mutation in two of the four alleles of the 23S rRNA gene. HLR-AZM Ng isolates with 23S rRNA mutations at another locus (A2058G), instead of A2059G, have now been reported in the United States (Table 1).

DISCUSSION

In this study, we characterize the local emergence of HLR-AZM Ng in Baltimore, Maryland, USA, and describe the global evolution of HLR-AZM in Ng. The three HLR-AZM Ng isolates from Baltimore were closely related and belonged to MLST ST9363. Prior to 2016, no HLR-AZM Ng isolates belonging to MLST ST9363 had been reported, but isolates belonging to this ST now account for over 40% of the reported HLR-AZM Ng. Additionally, HLR-AZM Ng MLST ST9363 have now been reported in over 50% of countries reporting HLR-AZM in Ng, highlighting the importance of this genotype in the evolution of AMR in Ng. To our knowledge, this study is the first to describe the global distribution and evolution of HLR-AZM Ng MLST ST9363.

While MLST ST9363 strains have been previously associated with decreased susceptibility to AZM (primarily mediated by resistance determinants in the mtr locus) in the United States (32), our study provides evidence of the ability of ST9363 strains to acquire additional resistance determinants resulting in HLR to AZM. Thomas et al. showed that Ng strains belonging to ST9363 have quickly evolved from AZM susceptible to AZM resistant (32). Additionally, Ng MLST ST9363 with reduced susceptibility to AZM has disseminated globally (20). Studies in Argentina and the United States have reported HLR-AZM MLST ST9363 Ng isolates with a novel combination of the 23S rRNA A2059G mutation with a mosaic-mtr, which may confer a fitness advantage over previously described HLR-AZM Ng (18, 21). This increased fitness hypothesis is supported by the observation that HLR-AZM isolates with this novel combination of resistance determinants were recovered during multi-year sustained transmission episodes in the United States and Argentina (15, 18). Given the potential for dissemination of ST963 strains and their capacity for sustained transmission, continued and enhanced surveillance is warranted to monitor the global evolution of Ng with this genotype.

Historically, HLR-AZM Ng isolates have been rare and appear sporadically (16, 33). However, these isolates are increasing in frequency, especially in the UK, where sustained transmission was confirmed over a 3-year period (16). In 2019, 245 HLR-AZM Ng cases were confirmed by the national reference laboratory in the United Kingdom; additional data from these isolates have not been reported (34). Despite the increases in frequency, our study revealed a very high number of NG-MAST STs that were unique to specific countries and represented as single isolates, suggesting that international transmission and spread of HLR-AZM Ng isolates remain rare. The large number of NG-MAST STs also suggests that emergence of HLR-AZM Ng can occur in a variety of genomic backgrounds. However, it is also possible, as suggested by the small number of MLST STs (n = 15) reported in this study, that some genomic backgrounds, such as the previously mentioned ST9363, are more likely to acquire resistance determinants resulting in AZM resistance and ultimately HLR. Further work is warranted to assess the predisposition of some Ng strains to acquire AMR especially resistance and HLR to AZM.

The three HLR-AZM Ng isolates from Baltimore (19) were phenotypically and genetically similar to some isolates recovered in Spain (2017–2018) highlighting the possibility of international transmission of these strains. However, without travel history, it is not possible to establish direct transmission of these isolates between the United States and Spain. Jennison et al. showed that cases of CRO-resistant and HLR-AZM Ng identified in the United Kingdom and Australia in 2018 were likely acquired in Asia (29). Furthermore, the linkage of these independent cases belonging to a specific clone of HLR-AZM Ng to Southeast Asia suggests that HLR-AZM Ng may be circulating in Asia, which is supported by the high number (n = 66) of HLR-AZM Ng cases from China described in our study (Table S2). In addition to the increasing rates of HLR-AZM Ng, the emergence of Ng with HLR-AZM and resistance to CRO (29), 6.3% in our study, are concerning as CRO is now the only recommended treatment option for gonorrhea. In regions like Asia where rates of CRO resistance could be higher than 25% (35, 36), concurrent high rates of HLR-AZM, as suggested by our analysis, reinforces high concerns about the future successful treatment of gonorrhea.

Our study also provides evidence of the evolution of AZM resistance mechanisms in Ng. We found that the majority of global HLR-AZM Ng isolates harbored four mutated A2059G alleles in the 23S rRNA gene confirming that this single mutation in all copies of the gene can result in HLR-AZM in Ng (37). However, our study also showed that the presence of two or three mutated A2059G alleles in combination with other mtr mutations can result in the same phenotype. Additionally, HLR-AZM Ng isolates with novel resistance determinants, including the 23S rRNA A2058G mutation, are beginning to emerge. Considering Ng’s capacity to quickly evolve, including the development of compensatory mutations to increase fitness (38), such as the mosaic mtr in HLR-AZM Ng described in this study, increasing capacity for the collection of Ng isolates is warranted to monitor emerging resistance mechanisms and develop a better understanding of the relationship between resistance markers and phenotypic resistance.

Our study has several limitations. First, we were unable to determine if transmission of HLR-AZM Ng in Baltimore had persisted because there were no isolates from Baltimore submitted to GISP in 2017 and 2018. However, analysis of 119 Ng isolates from Baltimore collected by GISP between 2019 and 2022 did not find evidence of HLR-AZM Ng in Baltimore, Maryland (14). Second, our analysis of previously reported HLR-AZM Ng was based on publicly available whole genome sequenced Ng, thus limiting our availability to make generalizable conclusions. Due to a global decrease in collection of Ng isolates for surveillance activities, it will be difficult to fully understand the global magnitude of HLR-AZM Ng. While our study summarizes the published data on HLR-AZM Ng, it likely underestimates the true magnitude of the rates of HLR-AZM Ng, especially in regions like Southeast Asia, which are now beginning to report higher rates of ceftriaxone-resistant Ng. Finally, due to the low number of HLR-AZM Ng isolates reported per year, it was not possible to perform a time trend analysis to determine if the introduction of dual treatment (CRO and AZM) recommendations in 2012 had an effect on the emergence of HLR-AZM Ng. Our results, however, found that the number of isolates and the number of countries reporting HLR-AZM Ng have increased since 2012.

The emergence and spread of HLR-AZM Ng pose another threat to the successful future treatment of gonorrhea especially in the setting of isolates with dual resistance to CRO and AZM. Surveillance activities should be strengthened, especially in regions with high rates of AMR, and focus not only on reporting susceptibility trends but on the emergence and spread of XDR Ng. Furthermore, research on factors affecting the emergence of AMR is critical to prevent future outbreaks and extend the lifetime of currently used treatment options. Additionally, a global consortium should be established to systematically collect, store, and disseminate WGS data from Ng to allow for additional analyses of AMR in Ng. Alternatively, the global infrastructure underlying surveillance consortia, such as the WHO Enhanced Gonococcal Antimicrobial Surveillance Programme (EGASP) that primarily reports on susceptibility trends with data that are typically limited to specific regions with established surveillance programs could potentially be expanded to accommodate standardized archival and characterization of isolates collected by other investigators in the field. Future studies are warranted not only to understand the evolution of AMR in Ng but to provide data to support future treatment recommendations.

MATERIALS AND METHODS

Ng isolates and susceptibility testing

A panel of 30 Ng isolates collected from men attending the BCHD SHCs between January and October 2016 was selected for genomic analysis as described below. This panel of isolates, which included the three HLR-AZM Ng isolates and 27 AZM-susceptible isolates, was phenotypically characterized using the E-test method as part of a previous study and was selected based on their antimicrobial susceptibility profiles and month of isolation in 2016 (Table S1) from a larger collection of previously described isolates (19). To determine if HLR-AZM Ng was circulating in Baltimore prior to 2016, archived isolates collected at the BCHD SHCs from 2014 and 2015 were tested for AZM susceptibility as follows. Antimicrobial susceptibilty testing (AST) for AZM was performed on 164 isolates collected in 2014 using the E-test method as previously described (19). The American Type Culture Collection (ATCC) 49226 and the WHO K, WHO L, and WHO V reference strains (39) were used as controls to ensure the accuracy of E-test results. Archived isolates from 2015 (n = 194) were cultured, but only 12 were viable for AST. Given the high number of non-viable isolates from 2015, genomic analysis was performed on all samples (viable and non-viable isolates), using a previously described assay (40), to determine if the 23S rRNA A2059G (E. coli numbering) mutation was present. The 23S rRNA A2059G mutation (four mutated alleles) is highly predictive of HLR-AZM in Ng (37). Briefly, DNA was extracted from the media containing the viable and non-viable isolates from 2015 using the QIAamp DNA Micro Kit (Qiagen). These samples have been previously analyzed in a study assessing ciprofloxacin susceptibility through detection of a mutation in the gyrA gene (41); our study concluded that these samples contained sufficient DNA for genomic analysis. The extracted DNA was tested by real-time PCR targeting the 23S rRNA gene and melt curve analysis used to differentiate the wild type (A2059) from mutant (2059G) 23S rRNA alleles. Of the 194 samples (isolates), 88.7% (n = 172) were included in the study; 22 samples were excluded either because of poor DNA yield or because the sample was not available.

Whole genome sequencing

Genomic DNA from the 30 isolates selected from the Baltimore collection (19) was subjected to library construction for whole genome shotgun deep sequencing on the Illumina HiSeq 4000 platform, yielding a minimum of 15M 150 bp paired-end reads per genome. Genome assembly with SPAdes v3.11.1 (42) resulted in draft genomes assembled into 87 to 136 contigs of size ≥ 500 bp. Assemblies were annotated with the Institute for Genome Sciences-automated annotation pipeline (43).

Global genomic data from HLR-AZM isolates

The PubMed electronic database was searched for published literature regarding publicly available whole genome sequenced HLR-AZM Ng isolates between 2000 and September 2023. The search was performed using the following search terms: Ng or gonorrhea and AZM resistance or high-level AZM resistance. Genomic data from HLR-AZM Ng isolates from published results were included in the analysis if they included an NCBI sequence read archive (SRA) accession number and/or typing data from either the MLST or NG-MAST method. The Neisseria isolate database in PubMLST was searched for Ng isolates with an AZM MIC ≥ 256 µg/mL or 2048 mg/L. Overall, 304 HLR-AZM Ng isolates matching the search criteria were identified, and of those, 18 (4.6%) were excluded from the analysis due to limited information. The excluded HLR-AZM Ng isolates were from Canada (5 isolates) (44) and China (13 isolates) (45). A total of 289 (286 global and 3 from Baltimore) isolates were included in the analysis.

Whole genome-based phylogenetic analysis

Publicly available assembled Ng genomes from NCBI RefSeq were downloaded on 29 April 2024. Unassembled genomes selected from SRA were assembled using SPAdes as described above. A matrix of pairwise distances between a total of 1,716 whole genome sequences was generated using Mash v2.3 (46) and a neighbor-joining phylogenetic tree constructed from this matrix using quicktree v2.5 (47) with default parameters. Trees were viewed and colored in Dendroscope v3.8.10 (48).

Molecular typing: MLST and NG-MAST

Raw sequencing reads from the Baltimore isolates and global HLR-AZM Ng genomes, without a MLST ST, were used to perform molecular epidemiologic analysis using the MLST method (49). MLST assigns STs based on a combination of alleles at seven housekeeping genes (49). Publicly available NG-MAST (50) data from the global HLR-AZM Ng isolates were used for this analysis. NG-MAST assigns STs based on the combination of alleles from the more variable internal segments of the porB and tbpB genes.

AMR gene determination

AMR determinants (Table S1) for the Baltimore Ng isolates were extracted from WGS data using MS11 and FA1090 as reference. AMR determinants for the global HLR-AZM Ng were obtained from published data. Additionally, for HLR-AZM Ng isolates with published genomes, variants in the 23S rRNA gene were identified and the copy number in the genome was determined as previously described (33) except that we used bowtie v1.2.2 for read mapping, samtools v1.11 for SAM/BAM file processing, and BBTools/BBMap’s callvariants.sh (51) for SNP calling (VCF file generation). Our analysis identified four HLR-AZM Ng isolates without the 23S rRNA A2059G mutation in their genome, but with four alleles of the 23S rRNA C2611T mutation. For three of those HLR-AZM Ng isolates, the previously reported results (23S rRNA A2059G mutation) have been used for this publication. No AMR determinants were reported for the fourth HLR-AZM Ng isolate (ERR3325482); therefore, the C2611T mutation identified by our analysis has been reported as the primary AZM-resistant determinant. For penA allele assignments, all assembled genomes were systematically annotated using Prokka v1.13 (52) (for consistency) and penA orthologs of the FA1090 penA gene (NGO1542) were identified using PanOCT v3.23 (53). All penA allele assignments for the nucleotide sequences of orthologs were made using blastn (54) alignments to the penA NG-STAR v2.0 database of alleles (downloaded in May 2024 from https://ngstar.canada.ca/alleles/penA), using the best single hit (-v 1, -b 1) per ortholog sequence with an e-value cutoff of 1e⁻⁵⁰ (all had an e-value of 0, i.e., all best hits had 100% identity to one of the alleles in the database).

DATA AVAILABILITY

Sequencing data for the 30 Ng isolates from Baltimore have been deposited at NCBI under BioProject accession number PRJNA1128699 (accession numbers JBFCOR000000000, JBFCOS000000000, JBFCOT000000000, JBFCOU000000000, JBFCOV000000000, JBFCOW000000000, JBFCOX000000000, JBFCOY000000000, JBFCOZ000000000, JBFCPA000000000, JBFCPB000000000, JBFCPC000000000, JBFCPD000000000, JBFCPE000000000, JBFCPF000000000, JBFCPG000000000, JBFCPH000000000, JBFCPI000000000, JBFCPJ000000000, JBFCPK000000000, JBFCPL000000000, JBFCPM000000000, JBFCPN000000000, JBFCPO000000000, JBFCPP000000000, JBFCPQ000000000, JBFCPR000000000, JBFCPS000000000, JBFLJW000000000, and JBFLJX000000000).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aac.00927-24.

Fig. S1. aac.00927-24-s0001.pdf.

Phylogenetic tree of isolates from Baltimore, Maryland and Oahu, Hawaii.

Fig. S2. aac.00927-24-s0002.pptx.

Maximum-likelihood phylogenetic tree of HLR-AZM Ng and closest neighbors in Clade 1.

Table S1. aac.00927-24-s0003.pdf.

Antimicrobial susceptibility profiles and genetic variants of Ng isolates collected in Baltimore in 2016.

Table S2. aac.00927-24-s0004.pdf.

All isolates included in this analysis.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.