Emergence and evolution of internationally disseminated cephalosporin-resistant Neisseria gonorrhoeae clones from 1995 to 2005 in Japan

Background Neisseria gonorrhoeae strains with resistance to extended-spectrum cephalosporins (ESCs), last options for first-line monotherapy of gonorrhoea, likely emerged and initially disseminated in Japan, followed by international transmission. In recent years, multi-locus sequence typing (MLST) ST1901 and N. gonorrhoeae multiantigen sequence typing (NG-MAST) ST1407 isolates with the mosaic penicillin-binding protein (PBP) 2 XXXIV have accounted for most ESC resistance globally. Our aim was to elucidate the initial emergence and transmission of ESC-resistant strains by detailed examination of N. gonorrhoeae isolates from 1995 to 2005 in Kanagawa, Japan. Methods N. gonorrhoeae isolates were examined phenotypically (n = 690) and genetically (n = 372) by agar dilution method (cefixime, ceftriaxone and ciprofloxacin), penA gene sequencing, MLST and NG-MAST. Results Already in 1995, one cefixime-resistant (CFM-R) isolate was found, which is the first CFM-R isolate described globally. After 1996, the prevalence of CFM-R and CFM-decreased susceptibility (CFM-DS) isolates significantly increased, with the peak resistance level in 2002 (57.1 % CFM-R). In 1997–2002, the CFM-R MLST ST7363 strain type with the mosaic PBP 2 X was predominant among CFM-R/DS isolates. The first CFM-R/DS MLST ST1901 clone(s), which became the predominant CFM-R/DS strain type(s) already in 2003–2005, possessed the mosaic PBP 2 X, which was possibly originally transferred from the MLST ST7363 strains, and subsequently acquired the mosaic PBP 2 XXXIV. The first MLST ST1901 and NG-MAST ST1407 isolate was identified in Kanagawa already in 2003. Conclusions The two main internationally spread cefixime-resistant gonococcal clones, MLST ST7363 and ST1901 (NG-MAST ST1407 most frequent internationally) that also have shown their capacity to develop high-level ceftriaxone resistance (superbugs H041 and F89), likely emerged, evolved and started to disseminate in the metropolitan area, including Kanagawa, in Japan, which was followed by global transmission.

Over the past two decades, gonococcal strains with decreased susceptibility and resistance to ESCs have been suspected to have emerged in Japan and subsequently spread internationally [2][3][4]. For example, from 1995 to 2000, in Fukuoka, Japan, the MIC peak of CFM and CRO in gonococcal isolates increased from 0.008 mg/L to 0.25 mg/L and from 0.015 mg/L to 0.064 mg/L, respectively [16]. Furthermore, from 1999 to 2002 in six hospitals in central Japan, the proportion of gonococcal isolates with decreased susceptibility or resistance to CFM (MIC ≥ 0.5 mg/L) and CRO (MIC ≥ 0.5 mg/L) increased from 0 % to 30.2 % and from 0 % to 0.9 %, respectively [17]. The first treatment failures with CFM were also described in Japan, i.e. in the late 1990s and early 2000s [18,19]. Accordingly, already in the late 1990s ESC-resistant N. gonorrhoeae strains emerged and were disseminated in Japan. However, the genetic relatedness between the ESC-resistant N. gonorrhoeae strains spreading in Japan 1-2 decades ago and the ESCresistant gonococcal strains currently spreading internationally has not been investigated appropriately. It is crucial to understand the emergence and spread of ESCand MDR-resistant gonococcal strains globally in order to develop and implement evidence-based strategies for prevention and control of gonorrhoea.
The aim of this study was to examine 690 (372 genetically) N. gonorrhoeae isolates cultured from 1995 to 2005 in Japan (the initial phase of emergence and dissemination of ESC-resistant strains) using antimicrobial susceptibility testing, penA gene sequencing, MLST and NG-MAST.

Neisseria gonorrhoeae isolates
In total, 690 N. gonorrhoeae isolates, cultured at the Kanagawa Prefectural Institute of Public Health, Japan from 1995 to 2005, were included. N. gonorrhoeae isolates were cultured, species identified and stored as described previously [20,21]. All examined gonococcal isolates were cultured and preserved as part of the routine diagnostics (standard care) and no patient identification information was available in the present study. Accordingly, ethical approval was not required for this study.

Antimicrobial susceptibility testing
The MICs (mg/L) of CFM, CRO and ciprofloxacin were determined for all isolates immediately after initial isolation by the agar dilution method, according to the instructions from the Clinical and Laboratory Standards Institute (CLSI) [22]. However, the MIC values for all the isolates with resistance (R) or decreased susceptibility (DS) to CFM (CFM-R/DS) were also verified in the present study by repeated testing using the identical agar dilution method. The resistance breakpoints stated by the European Committee on Antimicrobial Susceptibility Testing (EUCAST; www.eucast.org) were applied. However, additionally CFM-DS and CRO-DS was defined as MIC = 0.125 mg/L, because treatment failures with both CFM and CRO have been caused by gonococcal isolates with this MIC [3,8,[10][11][12]. The N. gonorrhoeae international reference strains WHO A, B and E were used for quality control.

DNA extraction
The bacterial isolates were suspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0) and boiled for 10 min. After centrifugation to remove cell debris, the supernatant was promptly used as DNA template for the PCRs.
penA gene sequencing The penA gene was PCR amplified and sequenced using the previously described primers penA_F and penA_R [13]. Briefly, the PCR mixtures were incubated for 2 min at 96°C, followed by 30 cycles of 10 s at 96°C, 10 s at 65°C and 2 min at 72°C. The PCR products were subsequently purified with the ExoSAP IT kit (GE Healthcare Limited, Buckinghamshire, United Kingdom). Both DNA strands of the PCR products were sequenced with an ABI BigDye Terminator cycle sequencing kit (version 3.1) on an ABI 3130 xl sequencer, in accordance with the instructions from the manufacturer (Applied Biosystems, Foster City, CA, USA). The penA gene that encodes the PBP 2 X sequence variant has a mosaic-like structure that includes regions relatively similar to the corresponding regions of the penA genes of N. perflava, N. sicca and N. cinerea [23]. Subsequently, many different mosaic PBP 2 sequences have been described, which have up to 60 to 70 amino acid changes compared to a wild type PBP 2 sequence [7]. In the present study, all these sequences are named as mosaic PBP 2 sequence variants and wild type PBP 2 (e.g., GenBank accession no. M32091) and highly similar sequences are defined as non-mosaic PBP 2 sequences.
The nucleotide sequences of penA determined in the present study have been deposited in the DDBJ sequence library and assigned the accession numbers LC055782 and LC055783.

Molecular epidemiological characterisation
Molecular epidemiological characterisation by means of MLST (based on partial alleles of seven loci, abcZ, adk, aroE, fumC, gdh, pdhC and pgm) and NG-MAST (based on partial alleles of porB and tbpB) was performed as described elsewhere [24]. New alleles (and STs) identified in this study were named and deposited in the MLST and NG-MAST database, respectively (Tables 1 and 2). The diversity index for the MLST and NG-MAST was calculated as described earlier [24]. UPGMA trees based on partial porB gene sequences (490 bp), i.e. obtained in the NG-MAST, were generated using the MEGA 4 software.

Statistical analysis
The distribution of MLST ST7363 in CFM-S and CFM-R/DS strains was analysed by Fisher's exact test.

Results
Emergence of cephalosporin resistance in Neisseria gonorrhoeae in Japan Already in 1995, among 34 N. gonorrhoeae isolates one CFM-R isolate (MIC = 0.25 mg/L), which is the first CFM-R isolate ever described, and one CFM-DS isolate (MIC = 0.125 mg/L), were identified. No CFM-R/DS isolates were found in 1996. However, from 1997 (one CFM-R isolate with MIC = 0.25 mg/L) and onwards, the prevalence of CFM-R and CFM-DS isolates significantly increased, with the peak resistance level in 2002, where 57.1 % of the examined isolates were CFM-R. The first isolate with CRO-DS was identified in 2000. CRO-DS isolates were subsequently found annually in 2001-2005, with a peak prevalence of 29.1 % in 2003 (Table 3).
penA (encoding PBP 2) sequences of CFM-R/DS Neisseria gonorrhoeae isolates in Japan The CFM-R isolate from 1995 was not possible to recover. However, the CFM-DS isolate from 1995 contained the non-mosaic PBP 2 XIII [7,25], whereas the CFM-R isolate from 1997 possessed the mosaic PBP 2 X [3,7,13]. These two isolates from 1995 to 1997 were considered the original CFM-R/DS isolates in the present study.
Among the CFM-R and CFM-DS isolates from 1998 to 2005 (n = 166), 149 (90 %) isolates were possible to analyse genetically; of these 149, 12 PBP 2 amino acid sequence variants were found. Nine were mosaic PBP 2 and three were non-mosaic PBP 2 sequence variants (VII, XI, XIII). The most prevalent sequence type was the mosaic PBP 2 X (86.6 %, 129/149). The additionally identified mosaic PBP 2 sequence variants were variants of X, i.e. with single amino acid substitutions (XXIV, XXX, XXXI), XXVI, XXXIV, and three XXXIV variants with single amino acid substitutions (Fig. 1). Most noteworthy, the first CFM-DS isolates with the mosaic PBP 2 XXXIV (n = 2) were identified already in 2001. All the non-mosaic PBP 2 sequence variants contained the A501V alteration and additionally alterations in G542 or P551 ( Fig. 1, Table 4).       Table 4). One of these isolates (from 2003) was also assigned as NG-MAST Fig. 1 Neisseria gonorrhoeae penicillin-binding protein 2 (PBP 2) amino acid sequences in strains with resistance or decreased susceptible to cefixime. a The dendrogram analysis of amino acid sequences included 12 PBP 2 sequences from N. gonorrhoeae with resistance and decreased susceptibility to cefixime and a wild-type (WT) PBP 2 sequence (M32091). Lower half contains the wild-type PBP 2 sequence and amino acid alterations in the non-mosaic PBP2 XI, XIII and VII, which possessed two amino acid substitutions compared to WT. Upper half displays the mosaic PBP 2 X, XXXI and XXXIV and their single amino acid variants that were found in this study. The numbers of isolates in this study are shown in parentheses. b Amino acid sequence similarities of mosaic PBP 2 (X, XXXIV and XXVI) and WT are shown. The boundary of N-and C-terminal domain is from the crystal structure of PBP 2 derived from the penicillin-resistant strain FA19 [33]. The N-terminal domain (1-239) of the mosaic PBP 2 sequences are similar to WT (over 96.7 %), but the C-terminal domains of mosaic PBP 2 show lower similarity: 86.4 % for PBP 2 X and PBP 2 XXXIV and 87.7 % for XXVI compared to WT. PBP 2 X is identical to PBP 2 XXXIV, except for the C-terminal end (549-582, where seven amino acids differ). The C-terminal end of PBP 2 XXXIV is identical with that of WT, whereas that of PBP 2 X is identical to PBP 2 XXVI, although the C-terminal domain of PBP 2 XXVI differs from the mosaic PBP 2 X and XXXIV (97.7 % identity). For detailed amino acid sequences of these and other PBP 2's, see Ohnishi et al. [7] ST1407 (porB908 and tbp110). This is the first isolate described worldwide of this MDR clone that has subsequently accounted for most of the reported ESC resistance globally [3,4,8,9,12,14,15,24,26].
NG-MAST analysis of CFM-R/DS Neisseria gonorrhoeae isolates in Japan The CFM-DS isolate from 1995 was assigned as the NG-MAST ST4045 (porB2445 and tbpB29), which was not subsequently identified during 1997-2005. The CFM-R isolate from 1997 belonged to the NG-MAST ST4127 (porB2520 and tbpB10), which two CFM-R isolates were also assigned to in 1998. However, ST4127 isolates, even isolates with porB2520, were not found after 1998.
To describe to a great extent the evolution of the isolates examined in the present study the porB gene sequences in isolates from 1998 to 2005 (only CFM-R/DS isolates from 2003 to 2005; n = 370) were compared (Fig. 3). For comparison, the six major porB alleles from isolates cultured in Kyoto and Osaka from 2010 to 2012 (porB4, porB206, porB908, porB1059, porB1785 and porB2569) [24] and porB254 and porB628, which were reported in CFM-R isolates in Sweden and the USA in 2002 and 2003 [27], respectively, were also included in the analysis (Fig. 3). Of these eight porB alleles, four of the porB alleles from the Kyoto and Osaka collection (porB4, porB206, porB908, porB1059) were also found in the present study. Seven clusters were generated by the porB sequences: clusters (CL) A-1, A-2, A-3, B-1, B-2, C and D ( Fig. 3; Table 5). The CL A-1 cluster was the largest cluster, containing 56 porB alleles from 132 isolates. The CL A-1 cluster included porB2520, which was from the first putative CFM-R isolates found in 1997 and 1998. Sixty-six (66.7 %) of the 99 MLST ST7363 CFM-R/DS isolates from 1998 to 2005 had one of the CL A-1 porB sequences and 64 (64.6 %) had the mosaic PBP 2 X. Furthermore, 64.5 % (49/76) of all the CFM-R/DS isolates from 2003 to 2005 belonged to the CL A-1 cluster. As mentioned above, the CFM-R/DS isolates with porB2520 from 1997 to 1998 did not appear to be widely disseminated. Instead, isolates with other porB sequences, such as porB1059, porB917 and porB206 in the CL A-1 cluster, appeared to become dominant (Fig. 3).
In the present study the initial emergence and dissemination of these internationally transmitted ESCresistant gonococcal clones in Japan from 1995 to 2005 were investigated. We showed that the first CFM-R isolate was cultured as early as 1995 in the Kanagawa area, Japan, which is four years earlier than previously recorded. This CFM-R isolate (MIC = 0.25 mg/L) possessed a non-mosaic PBP 2 XIII sequence variant [7,25], with the amino acid alterations A501V and P551S that increase the ESC MICs [3,4,24,28,29]. This first CFM-R strain did not appear to spread widely; however, two years later (in 1997) one CFM-R MLST ST7363 isolate with the mosaic PBP 2 X [3,7,13] was found. During 1998-2002, this was the predominant CFM-R/DS strain type in the Kanagawa area, which is a neighbouring area of Tokyo. In contrast, in 1998-2002 MLST ST1901 isolates accounted for 11 % (32/294) of the isolates, but only two were CFM-R. Furthermore, no isolate was typed as NG-MAST ST1407 and the two CFM-R MLST ST1901 isolates possessed the mosaic PBP 2 X, and not the nowadays frequent mosaic PBP 2 XXXIV [3,4,8,9,12,24,26]. This observation indicates that the CFM-R/DS MLST ST1901 clone(s) were originally possessing the mosaic PBP 2 X, which possibly was originally transferred from the MLST ST7363 strains spreading widely, and subsequently, acquired the mosaic PBP 2 XXXIV that presently is the most frequent PBP 2 sequence variant in ESC-R/DS MLST ST1901 and NG-MAST ST1407 isolates [3,4,8,9,12,24,26]. In the present study it was also shown that the first MLST ST1901 and NG-MAST ST1407 isolate was cultured already in 2003. This is substantially earlier than the previously first described NG-MAST ST1407 strain, which possessed the mosaic PBP 2 XXXIV, identified in the USA in 2008 [30]. Since then, this strain type has been found to account for most of the ESC-R/DS isolates globally [3,4,8,9,12,24,26] and this clone has also shown Two isolates possessed the non-mosaic PBP 2 VII and XI [7] (See figure on previous page.) Fig. 3 Comparison of the porB gene sequences in Neisseria gonorrhoeae isolates obtained from 1998 to 2005 in the Kanagawa area, Japan. Dendrogram was constructed using 163 partial porB sequences (490 bp) by UPGMA. One -hundred -fifty-nine porB sequences were from N. gonorrhoeae isolates examined in this study, including four of the major porB alleles in the Kyoto/Osaka strains isolated during 2010-2012 [24], (depicted in the shaded frame). The two remaining of the major porB alleles in the Kyoto/Osaka strains (porB1785 and porB2569) [24] and porB254 and porB628, from strains with cefixime resistance in Sweden (2002) and the USA (2003) [27], were also included with the porB type framed. Each box illustrates an individual strain. When more than four isolates with the identical porB type were cultured in the same year, the number of isolates is shown. Red and black boxes indicate MLST ST7363 strains and non-ST7363, respectively. Filled boxes indicate CFM-R/DS strains. Red and black filled rectangles, ST7363 and non-ST7363 strains with PBP 2 X, pink and gray are mosaic PBP 2 other than PBP 2 X and non-mosaic PBP 2 its capacity to develop high-level ceftriaxone resistance, i.e., develop into a superbug such as F89 [14,15]. All 11 CFM-R/DS isolates from 1997 to 1999 were assigned to MLST ST7363 (mosaic PBP 2 X), but the NG-MAST analysis revealed two major types of CFM-R/DS isolates, i.e. due to the diversification of the porB gene (porB2520 in CL A-1 and porB2465 in CL B-2). These two clones were considered to have emerged and started to disseminate in the Metropolitan area, including Kanagawa. Introduction from some other area(s) was highly unlikely because there is no recorded isolation of CFM-R gonococcal strains before 1999 in any other place. The reasons for the initial emergence of ESC-resistance in Kanagawa and in general gonococcal antimicrobial resistance in Japan have still not been completely resolved, however, they have been hypothesized elsewhere [3].
In general, during 1998-2002, the MLST ST7363 (31.6 %), ST7359 (13.9 %) and ST1901 (10.9 %) were the three most prevalent STs. During 2003-2005, the proportion of MLST ST1901 strains significantly increased, particularly among the CFM-R/DS isolates (from 2.6 % to 15.8 % of isolates). According to a recent study from 2012-2012 [24], these three MLST STs have remained the most prevalent MLST STs in the Kyoto/Osaka area; however, ST1901 has taken over as the significantly most prevalent ST (ST1901: 40.9 %, ST7359: 19.2 % and ST7363: 17.1 %). In the present study all 41 ST7359 isolates possessed NG-MAST tbpB241 and 90.2 % contained a porB sequence in CL C-1, including porB2569, which was also the major allele among the MLST ST7359 Kyoto/Osaka strains in 2010-2012 [24]. Accordingly, MLST ST7359 strains have also been frequently isolated in many years. However, these strains have been highly susceptible to CFM and initially to also ciprofloxacin. In contrast, a major proportion of the MLST ST7363 and ST1901 strains have also been resistant to ciprofloxacin for decades. It is not evident whether CFM-R/DS acquired the ciprofloxacin-R phenotype, or vice versa, but these types of MDR strains had significant advantages, and accordingly, could rapidly and efficiently be disseminated, first locally in Japan and then globally.

Conclusions
The two main internationally spread cefixime-resistant gonococcal clones, MLST ST7363 and ST1901 (NG-MAST ST1407 most frequent internationally) that also have shown their capacity to develop high-level ceftriaxone resistance (superbugs H041 and F89 [7,14,15]), likely emerged, started to disseminate and evolved in the metropolitan area, including Kanagawa, in Japan, which was followed by global transmission. A grave concern is that we might face a similar scenario in the future, i.e. that strains with resistance to both ceftriaxone and azithromycin, which are today used widely internationally in dual antimicrobial treatment regimens [31,32], start to spread. It is crucial to understand the emergence and spread of ESC-and MDR-resistant gonococcal strains globally to develop and implement evidence-based strategies for prevention and control of gonorrhoea.