Evidence of Recent Genomic Evolution in Gonococcal Strains With Decreased Susceptibility to Cephalosporins or Azithromycin in the United States, 2014–2016

Abstract

Background

Given the lack of new antimicrobials or a vaccine, understanding the evolutionary dynamics of Neisseria gonorrhoeae is a significant public and global health priority. We investigated the emergence and spread of gonococcal strains with decreased susceptibility to cephalosporins and azithromycin using detailed genomic analyses of gonococcal isolates collected in the United States, 2014–2016.

Methods

We sequenced genomes of 649 isolates collected through the Gonococcal Isolate Surveillance Project. We examined the genetic relatedness of isolates and assessed associations between clades and various genotypic and phenotypic combinations.

Results

We identified a large and clonal lineage of strains (MLST ST9363) associated with elevated azithromycin minimum inhibitory concentration (AZI^em), characterized by a mosaic mtr locus (C substitution in the mtrR promoter, mosaic mtrR and mtrD). Mutations in 23S rRNA were sporadically distributed among AZI^em strains. Another clonal group (MLST ST1901) possessed 7 unique PBP2 patterns, and it shared common mutations in other genes associated with cephalosporin resistance.

Conclusions

Whole-genome sequencing methods can enhance monitoring of antimicrobial resistant gonococcal strains by identifying gonococcal populations containing mutations of concern. These methods could inform the development of point-of-care diagnostic tests designed to determine the specific antibiotic susceptibility profile of a gonococcal infection in a patient.

Whole-genome sequencing reveals evidence of recent genomic evolution of gonococcal strains collected in the United States from 2014 to 2016, directly following CDC’s 2012 change in treatment guidelines.

Gonorrhea, caused by the gram-negative organism Neisseria gonorrhoeae, is the second most commonly reportable condition in the United States. In 2017, 555 608 cases of gonorrhea were reported to the Centers for Disease Control and Prevention (CDC) [1]. Since the 1930s, N. gonorrhoeae has developed resistance to all recommended antimicrobial therapies, greatly limiting the number of available treatment options [2]. Reduced susceptibility of gonococcal isolates to either cephalosporins or macrolides, the two most clinically relevant antimicrobials, is primarily attributed to alterations in several chromosomal genes. Cis- or trans-acting mutations impacting function or expression of the mtrR-repressor encoding gene product (MtrR) can elevate expression of the mtrCDE efflux pump operon, resulting in decreased bacterial susceptibility to a broad range of antimicrobials, including beta-lactam and macrolide antibiotics [2]. Additionally, amino acid changes in the MtrD transporter component of the MtrCDE efflux pump can enhance such resistance and point mutations in 23S rRNA can confer high-level resistance to azithromycin [2, 3]. Mutations in ribosomal proteins L4 and L22, encoded by the rplD and rplV genes, may confer resistance to macrolides by altering the conformation of the 23S rRNA domains, as reported in a previous study on gonococcal isolates [4]. Mutations in porB can decrease the influx of antibiotics, while point mutations in ponA (PBP1) and penA (PBP2) can decrease the acylation rate for beta-lactams, including third-generation cephalosporins [2].

The Gonococcal Surveillance Isolate Project (GISP), a sentinel surveillance system of sexually transmitted disease (STD) clinics, monitors gonococcal antimicrobial resistance in the United States [5]. Reports from GISP indicated that isolates with elevated cefixime minimum inhibitory concentrations (MICs) (CFX^em; MIC ≥0.25 μg/mL) peaked at 1.4% nationally before CDC updated its treatment recommendations in 2012, but has since decreased to 0.4% in 2017 [5–7]. Since 2015, the first-line treatment option for uncomplicated gonococcal infections in the United States has been 250 mg of ceftriaxone injected intramuscularly along with 1g of azithromycin taken orally [8]. During 2012 to 2017, isolates with elevated ceftriaxone MICs (CRO^em; MIC ≥0.125 μg/mL) fluctuated between 0.05% and 0.2%, while isolates with elevated azithromycin MICs (AZI^em; MICs ≥2 μg/mL) demonstrated a steady increase [5]. Only 4 isolates (Hawaii, 2016) collected through GISP have been documented with high-level azithromycin resistance (HL-AZI^r; MICs ≥256 µg/mL) and CRO^em; however, no evidence of sustained transmission within Hawaii or elsewhere was observed [9]. Nevertheless, the recent global emergence of strains with resistance to extended-spectrum cephalosporins (ESCs; ie, cefixime and ceftriaxone), azithromycin, or both has raised concerns regarding a future of untreatable gonorrhea. History has shown that the decline and subsequent rapid emergence of newly resistant strains may be attributed to antibiotic and environmental pressures [2]. We hypothesized that uniquely resistant gonococcal strains may have emerged after a previous observational study [4].

As the available options for the effective treatment of gonorrhea wane, there is an increased urgency to better understand the underlying molecular epidemiology of the disease in the United States and other countries. Whole genome sequencing methods have been shown to provide enhanced resolution over traditional typing methods in studies aimed at identifying and monitoring the spread of resistant gonococcal lineages [4, 10]. Ultimately, data on the associations and stability of antimicrobial resistance markers can facilitate the development of rapid point-of-care diagnostic tests (eg, nucleic acid amplification tests, real-time polymerase chain reaction, etc.), which may inform treatment and prevention strategies.

MATERIALS AND METHODS

Isolate Selection and Antimicrobial Susceptibility Testing

This study included 649 male urethral N. gonorrhoeae isolates collected by GISP (January 2014 to December 2016). During this time period, there were 534 isolates identified with AZI^em (n = 448), CFX^em (n = 80), and CRO^em (n = 35), of which 267 (50%; 198 AZI^em, 45 CFX^em, and 25 CRO^em) were available for sequencing. See Supplementary Table 1, Supplementary Figures 1 and 2, for further details on selected isolates. Susceptibility to azithromycin, cefixime, and ceftriaxone, were determined using the agar dilution method, with azithromycin MICs ≥16 further analyzed by the Etest method (MIC range 0.016–256 µg/mL; bioMerieux, France) [11]. Elevated MIC interpretations were based on established criteria for GISP isolates [5].

Whole-Genome Sequencing and Sequence Analysis

DNA prepared from 649 N. gonorrhoeae isolates was sequenced (paired end, 2 × 250 bp) on an Illumina MiSeq sequencer (Illumina Denmark ApS, Copenhagen, Denmark), and whole-genome sequencing read data were submitted to the NCBI BioProject PRJNA317462 (Supplementary data). Quality assessment was performed using FastQC 0.10.1 and reads were de novo assembled using Spades 2.5.1 [12]. KmerGenie was used when the assemblies exceeded 150 contigs [13, 14]. Contaminants were identified using Kraken 0.10.5 [15]. A core genome single-nucleotide polymorphism (SNP) alignment was computed in ParSNP v1.2 using the FA19 reference strain (GenBank accession No. CP012026) [16]. A maximum likelihood phylogenetic tree was generated using RaxML v8.2.9 [17]. See the Supplementary methods for further details of analysis of multilocus sequence typing (MLST), N. gonorrhoeae multiantigen sequence typing (NG-MAST), antimicrobial resistance determinants, and phylogenies.

Statistical Analyses

Fisher’s exact or χ² test were used to determine the association between mutational patterns, sequence types (STs), sex of sex partner, or clades with respect to elevated MICs respective of each antibiotic [18]. P values and 95% confidence intervals were used to determine statistical significance. A Bonferonni correction was applied for multiple tests.

RESULTS

AZI^em Isolates Associated With Specific Clades and Distinct Mutational Combinations

The results of the phylogenetic analysis revealed 33 distinct clades (Figure 1). A summary of MLST STs and intraclade SNP distances observed in major clades/subclades of interest are presented in Table 1. AZI^em isolates represented nearly one-third of the examined isolates (198/649, 31%; Table 2). Clade A, the largest clade observed in the dataset, displayed significant associations with AZI^em (67%, 132/198; P < .0001), particularly MICs of 2 µg/mL (41%, 82/198; Figure 1). Despite varying by Health and Human Services (HHS) region, the majority of isolates in Clade A were closely related (96.55 ± 67.42 SNPs) and represented predominantly by MLST ST9363 (76%, 151/198; P < .0001) and NG-MAST ST3935 (17%, 34/198; P < .0001). The isolates appeared to be primarily circulating amongst men who have sex with men (MSM; 60%, 118/198; P < .0001; Supplementary Table 1). The predominant MIC range for AZI^em isolates in Clade A was 2–4 µg/mL (54%, 113/210). Although several azithromycin mutations were detected, only those with significant associations with known resistance mechanisms are discussed. Isolates in Clade A were characterized by the presence of an A→C substitution in the 13-bp inverted repeat of the mtrR promoter (58%, 114/198; P < .0001), a N. lactamica-like mosaic mtrR (40%, 80/198; P < .0001), the D79N mutation in mtrR (58%, 114/198; P < .0001), and a N. meningitidis-like mosaic mtrD (65%, 128/198; P < .0001; Supplementary Figure 3). Nearly all AZI^em isolates in Clade A contained V125A, A147G, and R157Q mutations in rplD (66%, 131/198; P < .0001; Supplementary Figure 4). Additionally, 24 of the 132 AZI^em isolates in Clade A with MICs of 16–32 µg/mL possessed the C2611T mutation in at least 1 of the 4 23S rRNA alleles. Isolates possessing at least 4 C2611T mutations, a C substitution, or A deletion in the mtrR promotor, and D79N or H105Y mutations in the mtrR coding region, had MICs of 16–32 µg/mL.

Figure 1.

Whole-genome core single-nucleotide polymorphism maximum likelihood phylogenetic tree of 649 Neisseria gonorrhoeae strains collected from 2014 to 2016 in the United States and their susceptibility/mutational profile with respect to azithromycin. Susceptibility to azithromycin is shown with colors representing the isolates with minimum inhibitory concentrations (MICs) in the susceptible (light orange) or elevated range (dark orange, light purple, purple, dark purple). Light orange for molecular determinants indicates the presence of a wild-type allele. The mtrR promoter column indicates the presence of a C substitution (light purple) or A deletion (dark purple). The mosaic mtrR column indicates the presence of a mosaic mtrR (purple). The mtrR D79N column represent the presence of a D → N mutation at position 79 in the mtrR coding region (purple). The mosaic mtrD column represents the presence of meningitidis-like mosaic mtrD. The rplD column represents the presence of amino acid substitutions at positions A125, A147, and R157. The colors in the column for either 23S C611T or A2059G mutations represent the cumulative number of mutations in the 4 23S rRNA alleles, as indicated by the key.

Table 1.

Summary of Clades and Subclades (Numbered) Identified in Dataset Including Intraclade SNP Distance, Counts of Isolates, Regions, Collection Year, and Predominant MLST ST

			No. of Isolates in Clade With Elevated MICs
Clade ID	SNP Distance, mean ± SD	No. of Isolates Within Clade/Subclade		AZI	CFX	CRO	Region	Isolate Collection Year	Predominant MLST
Clade A	96.55 ± 67.42	210	132			Multiple	2014–2016	9363
Clade B.1	66.32 ± 30.63	40	3	24	3	HHS:9,10	2014–2016	1901
Clade B.2	79.50 ± 17.42	8	4		4	HHS:9	2016	1901
Clade B.3	196.65 ± 77.73	39	5	1		Multiple	2014–2016	1901
Clade C.2	72.4 ± 48.02	4		2	1	Multiple	2014–2016	7367
Clade C.1	4.5 ± 1.78	2	4			Multiple	2016	7371
Clade D	116.18 ± 84.98	30	9			Multiple	2014–2016	8156
Clade E.2	133.11 ± 43.48	12				Multiple	2014–2016	1893
Clade E.1	189.83 ± 91.54	4				Multiple	2015–2016	1893
Clade F	34.67 ± 20.13	3				HHS:9	2016	1893
Clade G	15.33 ± 10.14	4		4	2	HHS:6	2014–2015	7822
Clade H.1	9.33 ± 6.47	3				HHS:6	2015–2016	7822
Clade H.2	4 ± 1.79	3	3			HHS:2	2014–2015	7822
Clade I	58.28 ± 61.66	20	1	3	8	Multiple	2014–2016	7827
Clade J	79.86 ± 66.45	14	1			Multiple	2014; 2016	11 181
Clade K	174.14 ± 102.59	8				Multiple	2014–2016	1601
Clade L	33.33 ± 15.78	19	9			Multiple	2014–2016	1579
Clade M	14.22 ± 11.37	5		4	1	HHS:9	2014–2016	1600
Clade N	73.55 ± 57.23	46	1	1		Multiple	2014–2016	7363
Clade O	6.33 ± 2.53	24	2			Multiple	2015–2016	8143
Clade P	59.54 ± 35.58	12	1	1	1	Multiple	2014–2016	1901
Clade Q	116.71 ± 57.77	16				Multiple	2014–2016	1588
Clade R	52.47 ± 75.14	8	8			HHS:9	2015–2016	10 932
Clade S.1	98.4 ± 26.7	5				Multiple	2014; 2016	10 931
Clade S.2	43.26 ± 10.15	8		1	1	Multiple	2014–2016	10 931
Clade T	147.67 ± 81.45	40	11			Multiple	2014–2016	1584
Clade U	166.29 ± 54.66	8				Multiple	2014–2016	8154
Clade V	164.3 ± 97.07	5				HHS:4	2015–2016	10 316

Abbreviations: AZI, azithromycin; CFX, cefixime; CRO, ceftriaxone; HHS, Health and Human Services; MIC, minimum inhibitory concentration; MLST, multilocus sequence typing; SNP, single-nucleotide polymorphism, ST; sequence type.

Table 2.

Molecular Markers Associated With Elevated Azithromycin MICs Among GISP Neisseria gonorrhoeae Isolates Collected From 2014 to 2016

No. of 23S rRNA Alleles With a Mutation/Total No. of Alleles in Genome		Presence of mtrR		No. of Isolates With Azithromycin MIC, µg/mL
A2059G	C2611T	Promoter Variant	Mosaic mtrR	≤0.5	1	2	4	8	16	32	≥256	Total
0/4	0/4	−	−	156	53	6						215
0/4	0/4	+	−	119	70	25	17	5				236
0/4	0/4	+	+	13	15	61	15	1				105
0/4	0/4	−	+	12	7	1						20
0/4	1/4	−	−	1	1							2
0/4	1/4	+	−	1	1	1	1					4
0/4	2/4	+	−	1			1		1			3
0/4	2/4	+	+					1				1
0/4	3/4	−	−			1	2					3
0/4	3/4	+	−					1				1
0/4	4/4	−	−				11	15	2			28
0/4	4/4	+	−		1		7	6	5	1		20
0/4	4/4	+	+						1			1
0/4	4/4	−	+				2	1				3
4/4	0/4	+	−								7	7
Total				303	148	95	56	30	9	1	7	649

Promoter variant includes A deletion (n = 246) and C substitution (n = 133) in inverted repeat. Mosaic mtrR coding region determined by alignment to the reference KT954125 mosaic mtrR coding sequence using a similarity threshold of 98%.

Abbreviations: GISP, Gonococcal Surveillance Isolate Project; MIC, multilocus sequence typing.

Outside of Clade A the majority of AZI^em isolates possessed an A deletion in the mtrR promoter (21%, 41/198; P < .0001) and the H105Y (17%, 34/198; P < .0001) mutation in the mtrR coding region, which are strongly associated with AZI^em. Clade B contained the second largest number of AZI^em isolates (14%, 12/87), represented predominately by MLST ST1901 (n = 7; Figure 1). All of the AZI^em isolates in Clade B were characterized by the possession of an A deletion in the mtrR promoter and the H105Y mutation in the mtrR coding region. Four of the 12 AZI^em isolates in Clade B had MICs 2–4 µg/mL and possessed mutations in the mtrR promoter only. The other 8 isolates contained additional point mutations in all 4 23S rRNA alleles. Four AZI^em isolates (MICs 8–16 µg/mL) possessed the C2611T mutation, while the other 4 with HL-AZI^r (MICs ≥256 µg/mL) possessed the A2059G mutation [9].

The remaining 54 AZI^em isolates, represented by several MLST STs, were singletons or part of smaller clonal groups dispersed throughout the phylogenetic tree (Figure 2 and Table 1). Although these isolates (MICs 4–8 µg/mL) were distantly related, the majority contained between 3 and 4 C2611T mutations in the 4 23S rRNA alleles (61.11%, 33/54), an A deletion in the mtrR promoter (64.29%, 45/54), and the H105Y mutation (54.29%, 38/54) in the mtrR coding region. Notable exceptions were as follows: 3 isolates in Clade H (MLST ST7827) with MICs ≥256 µg/mL possessed the A deletion in the mtrR promoter and A2059G mutations in all 4 23S rRNA alleles; and 9 isolates in Clade L (MLST ST1579) with MICs 2–8 µg/mL possessed the A deletion but no 23S rRNA variants. A summary of molecular determinants and the combinations significantly associated with AZI^em is displayed in Table 3.

Figure 2.

Whole-genome core single-nucleotide polymorphism maximum likelihood phylogenetic tree of 649 Neisseria gonorrhoeae strains collected from 2014 to 2016 in the United States and their susceptibility/mutational profile with respect to cefixime/ceftriaxone. Susceptibility to cefixime and ceftriaxone is shown with colors representing the isolates with minimum inhibitory concentrations (MICs) in the susceptible (light orange) or elevated range (dark orange, light purple, purple, dark purple). Light orange for molecular determinants indicates the presence of a wild-type allele. The PBP2 mosaicity column indicates whether an isolate’s penA allele is mosaic (purple). The A501 column indicates the presence of an alanine to threonine (light purple) or an alanine to valine (purple) substitution at position of 501 of penA. The mtrR promoter column indicates the presence of a C substitution (light purple) or A deletion (dark purple). The mtrR H105Y column represent the presence of an H → Y mutation at position 79 in the mtrR coding region (purple). The colors in the column for porB represent a single mutation (dark orange) in either G120 or A121 or a double mutation in both (purple). The colors in the column for ponA indicate the presence of an L → P mutation at position 421 in ponA (purple).

Table 3.

Amino Acid Substitutions in MtrR (Promoter and Coding Region) in Neisseria gonorrhoeae Isolates and Their Susceptibilities to Azithromycin, Cefixime, and Ceftriaxone

MtrR Mutation	Azithromycin Susceptibility Group, n (%)a		Cefixime Susceptibility Group, n (%)a		Ceftriaxone Susceptibility Group, n (%)a
	Susceptible	Elevated	Susceptible	Elevated	Susceptible	Elevated
WTb	45 (10)	13 (7)	57 (9)	1 (2)	58 (9)
WT;A39T	58 (13)	2 (1)	60 (10)		60 (10)
WT;A39T;R44H	60 (13)	12 (6)	72 (12)		72 (12)
WT;A39T;G45D	7 (2)		7 (1)		7 (1)
WT;R44H;H105Y		1 (1)	1		1
WT;G45D	17 (4)	12 (6)	29 (5)		29 (5)
WT;D79N;T86A;H105Y	32 (7)		32 (5)		32 (5)
WT;T86A	4 (1)			4 (9)	2	2 (8)
WT;T86A;H105Y	5 (1)		5 (1)		5 (1)
A deletion		2 (1)	2		2
A deletion;A39T		4 (2)	4 (1)		4 (1)
A deletion;G45D	37 (8)	3 (2)	36 (6)	4 (9)	31 (5)	9 (36)
A deletion;G45D;T86A;H105Y	5 (1)		1	4 (9)	4 (1)	1 (4)
A deletion;D79N;T86A;H105Y	45 (10)	1 (1)	45 (7)	1 (2)	46 (7)
A deletion;H105Y	116 (26)	33 (17)	119 (20)	30 (65)	136 (22)	13 (52)
C substitution;A39T;D79N;T86V		1 (1)	1		1
C substitution;D79N	17 (4)	113 (57)	130 (22)		130 (21)
C substitution;H105Y		1 (1)	1		1
Total no. of isolates per category	448 (100)	198 (100)	602 (100)	44 (100)	621 (100)	25 (100)

Bold text indicates a significant association with susceptibility group, based on Bonferroni adjusted P values (P < .05).

Abbreviations: AZI, azithromycin; CFX, cefixime; CRO, ceftriaxone; GISP, Gonococcal Surveillance Isolate Project; MIC minimum inhibitory concentration; WT, wild type.

^aSusceptibility groups defined based on GISP alert criteria: AZI^em (MIC ≤2 µg/mL), CFX^em (MIC ≤0.25 µg/mL), CRO^em (MIC ≤0.125 µg/mL).

^bWT, based on the FA19 (CP012026.1) mtrR promoter and mtrR coding region sequence.

CFX^em and CRO^em Isolates Associated With Specific Clades and Distinct Mutational Combinations

Of the 649 gonococcal isolates sequenced, 45 (7%) displayed CFX^em, 25 (4%) displayed CRO^em, and 3 (0.5%) displayed elevated MICs to both. Clade B, represented predominantly by MLST ST1901 (75%, 65/87) and NG-MAST ST1407 (24%, 21/87), contained the majority of CFX^em (49%, 22/45, P < .0001) and CRO^em (40.00%, 10/25, P < .001) isolates (Figure 2). Only CFX^em isolates displayed significant associations with MSM (42%, 19/45, P < .001; Supplementary Table 1). Isolates in Clade B were characterized predominantly by an A deletion in mtrR promoter, H105Y in mtrR coding region, ponA L421P, and porB G120K/A121N or A121D. The 4 isolates from Hawaii [9] with HL-AZI^r also possessed elevated ceftriaxone MICs. In addition to the A deletion in the mtrR promoter and A2059G mutation, all 4 isolates contained the G120K/A121D mutation in porB, a nonmosaic PBP2 pattern XVIII with the A501T mutation, and the ponA L421P mutation. These were the only isolates with a dually elevated MIC phenotype (HL-AZI/CRO^em). Elevated cephalosporin MICs were also observed in smaller clades, dispersed throughout the phylogenetic tree. The clades containing isolates with elevated cephalosporin MICs, including their average intraclade SNP distances, are summarized in Table 1.

The complete nucleotide sequences of the penA alleles were compared to reference sequence M32091.1 [19]. The frequency of PBP2 patterns for cefixime and ceftriaxone with respect to MICs are presented in Table 4 and Table 5. In total, 7 unique amino acid sequence patterns were identified for isolates with elevated MICs to either cefixime or ceftriaxone, including nonmosaic patterns (XII, XIII, XVIII, and LXVIII) and mosaic patterns (X, XXXIV, and LXXI) (Supplementary Figures 5 and 6). A significant association was observed between with CFX^em and PBP2 pattern XXXIV (53%, 24/45, P < .0001) and LXXI (9%, 4/45, P < .0001). PBP2 pattern XIII (12%, 3/25, P < .0001) and XVIII (16%, 4/25, P < .0001) were significantly associated with CRO^em. Only isolates with the PBP2 pattern X displayed a significant association with both elevated cefixime (24%, 11/45, P < .0001) and ceftriaxone MICs (28%, 7/25, P < .0001). Eight isolates with pattern XIII and the A501V alteration clustered clonally in Clade I, while 4 other isolates within Clade B contained pattern XVIII with the A501T alteration [20, 21].

Table 4.

Minimum Inhibitory Concentrations of Cefixime for the 649 Neisseria gonorrhoeae GISP Isolates With Various PBP2 Patterns

PBP 2 Pattern	No. of Isolates With Cefixime MIC, µg/mL
	0.015	0.03	0.06	0.125	0.25	0.5	Total
I	3						3
II	136	102	26	1			265
V	11	18	5	2			36
IX	15	25	14	7			61
Xa				1	8	3	12
XII	3	2	2	5	1		13
XIII		2	2	13	3		20
XIV	36	6		2			44
XV	6	1					7
XVIII	1	3	1	5			10
XIX	8	6		2			16
XXII	23	5					28
XXXIVa	1	1	2	20	23	1	48
XXXVa	3						3
XLIV	4	14	8	17	1		44
XLV	7						7
LXIII	6	5	2				13
LXVIII	3	2	2		1		8
LXIX	1						1
LXXIa				1	4		5
LXXXIII		1					1
XCIIIb		3	1				4
Total	267	196	65	76	41	4	649

Abbreviations: GISP, Gonococcal Surveillance Isolate Project; MIC minimum inhibitory concentration.

^aPBP2 pattern is mosaic.

^bPBP2 pattern is semimosaic.

Table 5.

Minimum Inhibitory Concentrations of Ceftriaxone for the 649 Neisseria gonorrhoeae GISP Isolates With Various PBP2 Patterns

PBP 2 Pattern	No. of Isolates With Ceftriaxone MIC, µg/mL
	0.008	0.015	0.03	0.06	0.125	0.25	Total
I	2	1					3
II	121	113	26	5			265
V	3	14	11	8			36
IX	17	23	15	6			61
Xa			2	3	7		12
XII	2	1	4	5		1	13
XIII	1	1	4	6	7	1	20
XIV	39	4		1			44
XV	7						7
XVIII		1	3	2	4		10
XIX	10	4		2			16
XXII	23	5					28
XXXIVa	2	3	17	23	2	1	48
XXXVa	3						3
XLIV	6	5	17	16			44
XLV	6		1				7
LXIII	13						13
LXVIII	5	2			1		8
LXIX	1						1
LXXIa			1	3	1		5
LXXXIII		1					1
XCIIIb	1	2	1				4
Total	262	180	102	80	22	3	649

Abbreviations: GISP, Gonococcal Surveillance Isolate Project; MIC minimum inhibitory concentration.

^aPBP2 pattern is mosaic.

^bPBP2 pattern is semimosaic.

An adenine deletion in the mtrR promoter was significantly associated with isolates with CFX^em (89%, 40/45; P < .0001) and CRO^em (92%, 23/25; P < .0001) MICs. Additionally, we examined combinations of mutations in the mtr locus with respect to either cephalosporin (Table 3). None of the CFX^em or CRO^em isolates carried the A39T, R44H mutations [22]. The G45D mutation was significantly associated only with CRO^em isolates that also possessed the A deletion in the mtrR promoter (36%, 9/25, P < .0001) [23, 24]. One CFX^em isolate possessed the D79N mutation whereas none of the CRO^em isolates did. All CFX^em and CRO^em isolates possessed double mutations at amino acid residues Gly-120 and Ala-121 of PorB [25]. The G120K/A121N (51%, 23/45) and G120K/A121D (38%, 17/45) amino acid mutations were significantly associated with CFX^em isolates (P < .0001), whereas G120K/A121D (56.00%, 14/25) mutations were significantly associated with CRO^em isolates (P < .0001).

Discussion

N. gonorrhoeae has shown an extraordinary capacity to rapidly develop resistance to all previously used antimicrobials [2]. Several studies have shown that strains with reduced susceptibility can rapidly spread by clonal expansion through outbreaks among sexual networks [4, 26]. Using whole-genome analyses in concert with antimicrobial susceptibility testing data from GISP 2014–2016, we demonstrated that several gonococcal strains, particularly MLST ST9363 (AZI^em) and ST1901 (elevated cephalosporin MICs, ESC^em), have independently formed lineages spanning multiple years and wide geographic regions in the United States, suggesting possible clonal expansion. Our analyses revealed that these lineages were characterized by unique phenotypic and genotypic antimicrobial resistance patterns.

Azithromycin

Our data demonstrated several important findings. The majority of AZI^em isolates (97%, 193/198) in our study could be explained by promoter and coding mutations in the mtr locus (mtrR promoter [58%, 114/198], a mosaic mtrR [42%, 83/198], mtrD [64%, 128/198]), mutations in rplD (72%, 143/198), 23S rRNA variants (C2611T [18%, 36/198] or A2059G [3%, 5/198]), or a combination of these mutations. Only 5 isolates were identified that lacked these mutations. While these data and the corresponding associations with respect to AZI^em is supported by the literature, notably for mutations in the mtr locus and 23S rRNA alleles, it stands in contrast to mutations in GISP isolates from previous years (2000–2013) as identified in an earlier study [4]. In that study it was reported that AZI^em was primarily attributed to mutations in 23S rRNA alleles (165/294, 56%), and not mtr locus variants (mosaic mtr locus: 1%, 29/294; mtrR promoter variants: 36%, 105/294). In addition, although we did not detect rarer mutations such as mtr₁₂₀ or imported genes such as ermB or ermC in any of our AZI^em isolates [4, 10] we did detect 8 unique mutational patterns in rplD (eg, G70D/A147G/R157Q), which may have an addictive effect on macrolide resistance [3, 27] (Supplementary Figure 4). Furthermore, the larger number of AZI^em strains in our study over a relatively smaller time period compared to that of the Grad et al study, reflects GISP reports that suggest an increase in AZI^em in all regions of the United States beginning around 2013 [4, 7].

We identified a very large and distinct lineage of AZI^em isolates with a high prevalence of a C substitution in the mtrR promoter, mosaic mtrR, D79N mutation, and a mosaic mtrD (42%, 83/198) [28, 29]. Recent work has shown the C substitution reduces mtrR expression [29]. Interestingly, this substitution also generates an improved −35 hexamer of the overlapping, divergent mtrCDE promoter, resulting in increased expression of the efflux pump operon. The MtrR D79N substitution reduces MtrR repressive activity but the molecular basis is not yet known as the amino acid change is outside of the known DNA-binding domain [29, 30]. Warner et al demonstrated that mutations in the mtr locus that affect expression of the mtrCDE efflux pump confer fitness advantages over wild-type strains in vivo [31]. Importantly, this may explain why the MLST ST9363 lineage exhibited a sustained temporal persistence and a high degree of spatial heterogeneity across multiple regions within the United States over all the years investigated. Highlighting a similar trend observed in GISP, many of the AZI^em isolates in this clade significantly associated with MSM (62%, 118/198; P < .0001), and were highly homogenous having 96 out of 132 AZI^em isolates with a relatively close SNP distance (39.94 ± 14.65 SNPs) [32]. These data may have important implications as recent studies have reported a higher prevalence of azithromycin-resistant gonococci amongst networks of MSM [33, 34].

Outside of Clade A (MLST ST9363) a majority of isolates with elevated MICs to AZI contained 23S rRNA mutations (C2611T [n = 66] and A2059G [n = 7]), in addition to mutations in the mtrR promoter and coding regions; however, this co-occurrence varied. The overall infrequency of 23S rRNA variants in AZI^em isolates suggests that these mutations may occur more sporadically compared to mutations in the mtr locus. Several smaller clades consisting of isolates with the 23S rRNA variant displayed clonal dissemination, albeit to a lesser degree than the MLST ST9363 strain. One clonal group (Clade T [n = 11/40]) represented by MLST ST1584, which possessed the C2611T mutation in all 4 23S rRNA alleles but no additional mutations elsewhere, appeared between patients from cities in the Midwest and the West in 2015–2016 (9.09 ± 6.90 SNPs). The only clonal groups to contain 23S rRNA alleles with the A2059G mutation were identified in Clades B (n = 4, 79.50 ± 17.42 SNPs) and H (n = 3, 4 ± 1.79 SNPs), represented by MLST ST1901 and ST7822, respectively [35]. Although, the 4 Hawaiian (2016) isolates (79.50 ± 17.42 SNPs) represented by MLST ST1901 were contained, the fact that it possessed HL-AZI^r is particularly worrisome because this multidrug-resistant gonococcal clone has been attributed to most of the elevated cephalosporin MICs observed globally [21]. Despite, the limited number of HL-AZI^r isolates observed between 2014 and 2016, a recent study [36] demonstrated that the A2059G mutation can enhance the in vivo biological fitness of N. gonorrhoeae, and when combined with the sporadic global emergence of HL-AZI clusters illustrates the growing concerns over continued use of azithromycin [9, 34, 37, 38].

Cefixime/ Ceftriaxone

The majority of CFX^em was observed in Clade B, while other CFX^em isolates appeared more sporadically distributed phylogenetically. CRO^em appeared less frequently (4%, 25/649) than CFX^em (7%, 45/649) despite sharing many of the same polymorphisms in the ponA, porB, and mtrR promoters. This could possibly be explained by the presence of additional alterations in penA alleles [2, 21, 38]. For example, 5 isolates in Clade I with elevated MICs to cefixime and ceftriaxone contained the single amino acid substitution A501V in penA, which can enhance ESC MICs [2, 20]. Overall, ESC^em showed clear associations with specific mosaic PBP2 patterns (eg, X and XXXIV), described in earlier studies, in addition to lesser known patterns such as LXVIII and LXXI [21, 39–41]. Similar to a previous study, CFX^em isolates with PBP2 pattern XXXIV (61%, 17/28) were observed circulating amongst MSM [42]. Isolates with nonmosaic PBP2 patterns, such as XII and XIII, which contain short modified segments likely originating from other Neisseria spp., in addition to mutations in other genes (ie, mtrR, porB, and ponA), also possessed the ESC^em phenotype [43, 44]. Although, resistance mutations in mosaic penA alleles can reduce the in vitro and in vivo fitness of cephalosporin resistant strains, compensatory mutations in other genes (ie, mtrR, acnB, and mleN) can be selected that restore fitness, thus allowing these strains to spread [45–47]. The combined effects of compensatory mutations in these secondary loci in addition to mutations in porB and ponA may explain why CFX^em and CRO^em isolates with nonmosaic penA alleles can also spread despite these alleles not being associated with ESC^em on an individual basis [44]. In support of this, a closer inspection of metabolic genes revealed that the acnB Q371K mutation was frequently present in CFX^em strains (53%, 24/45, P < .0001); however the acnB G348D as reported by Vincent et al was not detected (Supplementary Figure 7) [45]. A mleN A465 codon deletion was detected in five isolates, but all were susceptible to both ESCs (Supplementary Figures 7 and 8) [45]. Furthermore, we found that mutations in other genes (eg, ponA L421P substitution, A deletion in mtrR promoter, H105Y mutation in mtrR, and a double mutation in porB at the G120 and A121 positions) were often present (P < .0001) in ESC^em strains regardless of the mosaicity of the penA allele or lack thereof [25]. These results suggest that any future molecular diagnostic tests will need to consider several mutational combinations in order to reliably predict ESC^em.

There are several limitations to this study. First, isolates are collected only from men with symptoms of urethral gonorrhea diagnosed at participating STD clinics, and therefore may not fully represent all gonorrhea diagnosed in the United States. Second, although the final sequenced data set for this study was a sample of all GISP isolates collected during the time period, by design it over-represented isolates with elevated MICs (eg, 49% of sequenced isolates had elevated MICs to AZI versus 3% in the full GISP dataset). However, among susceptible isolates and those with elevated MICs, the sequenced sample provided a reasonable representation of isolates observed through GISP over the years and across Health and Human Services regions (Supplementary Table 1). Finally, because whole-genome sequencing of gonococcal isolates is conducted on subcultures from a single-colony isolated from subcultures of frozen isolates it does not provide any context to the susceptibilities or genetic diversity of the entire gonococcal population during infection. This may be especially pertinent in cases involving mixed infections [48, 49].

Overall, our analyses suggest a relatively recent evolution of several gonococcal strains that possess mutations that were not observed in the United States in 2000–2013 [4]. The emergence of these mutational patterns may be explained by recent interspecies recombination with other commensal Neisseria species (eg, N. lactamica and/or N. meningitidis), various selective pressures (eg, host environment and continued antibiotic usage), or both [3, 10, 30, 38]. This impact appeared to be most profound in AZI^em isolates, particularly of MLST ST9363, where we detected evidence of horizontally acquired variants throughout the mtr locus, mtrD, and elsewhere in rplD. Conversely, ESC^em isolates displayed multiple PBP2 patterns that had not been reported prior to 2014, therefore future studies utilizing whole-genome sequencing should consider prioritizing compensatory mutations identified in fitness studies for further examination. Changes in the diversity and distribution of several antimicrobial resistance determinants in a relatively short period highlight the need for routine molecular surveillance in order to identify and respond to circulating gonococcal strains with antimicrobial resistance.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Centers for Disease Control and Prevention (CDC)’s core sequencing facility (Atlanta, GA) and partners at the Maryland or Washington public health laboratories, which are both part of the Antibiotic Resistance Laboratory Network (ARLN). We also thank the Gonococcal Isolate Surveillance Project (GISP) for the contribution of isolate and epidemiologic data used in our analyses. We thank Steve Johnson for his insightful discussion regarding mosaic PBP2 sequences.

Neisseria gonorrhoeae Working Group members: Hillard Weinstock (CDC, Division of STD Prevention, National Center for HIV/AIDs, Viral Hepatitis, STD, and TB Prevention), Catherine Dominguez (Maryland ARLN, Maryland Department of Health), Sopheay Hun (Washington ARLN, Washington State Department of Health), and Katie Kneupper (Texas ARLN, Texas Department of State Health Services).

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC, the Department of Veterans Affairs, or the National Institutes of Health.

Financial support. This work was supported by the Centers for Disease Control and Prevention (CDC) Advanced Molecular Detection and Combating Antibiotic Resistant Bacteria programs (J. C. T., S. S., A. J. A., J. C., S. L., E. V., M. S., C. D. P., K. M. G., and E. N. K.) and an Intergovernmental Personnel Act (W. M. S.); US Department of Energy and CDC Research Participation Program administered by the Oak Ridge Institute for Science and Education (J. C. T, S. S., J. C., S. L., and E. V.); National Institutes of Health (grant number R37AI21150-33) awarded to W. M. S.; and the Biomedical Laboratory Research and Development Service of the US Department of Veterans Affairs (Senior Research Career Scientist Award to W. M. S.).

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.