Novel Genes Related to Ceftriaxone Resistance Found among Ceftriaxone-Resistant Neisseria gonorrhoeae Strains Selected In Vitro

Abstract

The emergence of ceftriaxone-resistant Neisseria gonorrhoeae is currently a global public health concern. However, the mechanism of ceftriaxone resistance is not yet fully understood. To investigate the potential genes related to ceftriaxone resistance in Neisseria gonorrhoeae, we subcultured six gonococcal strains with increasing concentrations of ceftriaxone and isolated the strains that became resistant. After analyzing several frequently reported genes involved in ceftriaxone resistance, we found only a single mutation in penA (A501V). However, differential analysis of the genomes and transcriptomes between pre- and postselection strains revealed many other mutated genes as well as up- and downregulated genes. Transformation of the mutated penA gene into nonresistant strains increased the MIC between 2.0- and 5.3-fold, and transformation of mutated ftsX increased the MIC between 3.3- and 13.3-fold. Genes encoding the ABC transporters FarB, Tfq, Hfq, and ExbB were overexpressed, while pilM, pilN, and pilQ were downregulated. Furthermore, the resistant strain developed cross-resistance to penicillin and cefuroxime, had an increased biochemical metabolic rate, and presented fitness defects such as prolonged growth time and downregulated PilMNQ. In conclusion, antimicrobial pressure could result in the emergence of ceftriaxone resistance, and the evolution of resistance of Neisseria gonorrhoeae to ceftriaxone is a complicated process at both the pretranscriptional and posttranscriptional levels, involving several resistance mechanisms of increased efflux and decreased entry.

INTRODUCTION

Gonorrhea is one of the most prevalent sexually transmitted infections (STIs), and it is associated with many severe clinical symptoms such as prostatitis, seminal vesiculitis, pelvic inflammation, infertility, and ophthalmia neonatorum (1). Although epidemiological data show a decreased trend in gonorrhea incidence as a whole, in some countries, the incidence is increasing, especially in specific population groups (adolescents, men who have sex with men, or sex workers) or specific venues (brothels) (2–5). The pathogen responsible for this disease is Neisseria gonorrhoeae. According to worldwide surveillance reports, N. gonorrhoeae has developed resistance to several kinds of antibiotics over the past decades (6–10). Of increasing concern is the prevalence of ceftriaxone (CRO)-resistant N. gonorrhoeae (11), because CRO is the first-line empirical drug for gonorrhea treatment in dual therapy with azithromycin or doxycycline recommended by the WHO (covering the Western Pacific region and the Southeast Asian region) and the CDC (12–14). Several CRO-resistant gonococcal strains, such as H041 and F89, were isolated recently from patients who had a treatment failure with CRO (H041) and cefixime (F89) (15–18). Highlighting the seriousness of these resistant strains, Ohnishi et al. proposed the concept of an N. gonorrhoeae “superbug” in 2011 (17). Uncovering the mechanisms of CRO resistance is now a critical topic of N. gonorrhoeae research.

Previous studies have shown that the reduced susceptibility of N. gonorrhoeae to CRO is associated with the genes penA, mtrR, and penB (19–23). Tanaka et al. reported that a clinical gonococcal isolate with reduced CRO susceptibility and a multidrug-resistant phenotype could be explained by the presence of multiple-locus mutations of penA, ponA, mtrR, penB, gyrA, and parC (20). Lindberg et al. discovered gonococcal isolates with higher MICs to cefixime and CRO that contained a nearly identical penA mosaic allele and polymorphisms in mtrR, penB, and ponA (19). Ameyama et al. found that transformation of CRO-sensitive strain ATCC 19424 with mosaic penA derived from a clinical isolate with reduced cefixime susceptibility increased the CRO MIC of the transformant 8-fold compared with that of the isogenic parent (24). By comparing 28 N. gonorrhoeae isolates with reduced CRO susceptibility with 32 isolates that are sensitive to CRO, Liao et al. found that the reduced susceptibility of N. gonorrhoeae to CRO was mediated by porB1b alleles and was associated with penicillin binding protein 2 (PBP2) and MtrR (25). To define the individual contributions of these determinants to reduced cephalosporin susceptibility (Ceph^r), Zhao et al. created isogenic strains containing the mosaic penA allele from Ceph^r strain 35/02 (penA35) together with one or more of the other resistance determinants (mtrR, penB, and ponA1). The results showed that the majority of cefixime resistance is conferred by the penA35 allele, and CRO resistance is nearly equally dependent upon mtrR and penB, but ponA1 does not appear to be important for Ceph^r. Moreover, those authors found that the CRO MICs of transformants containing all four of these determinants could not reach the CRO MIC of the donor strain, suggesting that there was an additional unknown determinant required to reach donor levels of resistance (21).

Since the first report of a gonococcal strain, H041, with high-level resistance to CRO, investigators have gained a further understanding of the molecular mechanisms of CRO resistance in N. gonorrhoeae. By sequencing penA, mtrR, penB, ponA, and pilQ of H041, Ohnishi et al. found that this strain contained a new penA mosaic allele (penA_H041) rather than several mutated genes reported previously (including mtrR, penB, and ponA1), but pilQ was not altered (17). The mutated genes that emerged in the H041 strain are similar to the mutated genes that emerged in N. gonorrhoeae isolates with high-level CRO resistance found in Spain (15). In the study conducted by Ohnishi et al. (17), when penA_H041 was transformed into nine recipient strains, Ohnishi et al. observed that the CRO MICs of the transformants increased between 16- and 500-fold; however, eight of the nine transformants had lower CRO MICs than that of donor strain H041. Theoretically, the transformants with the same determinants of penA_H041, mtrR, penB, and ponA should have had the same resistance levels, but there were remarkable differences among these transformants. This phenomenon was also seen in similar transformation assays conducted by Unemo et al. (18). In another previous study, Unemo et al. also indicated that pilQ polymorphisms are unlikely contributors to the decreased susceptibility of clinical gonococcal strains to CRO (26), results that are similar to those reported by Ohnishi et al. Both studies suggest that there are unknown determinants related to CRO resistance. Specifically, Ohnishi et al. pointed out that there is at least one unknown resistance determinant and warned that although the biological fitness of CRO resistance in N. gonorrhoeae remains unknown, N. gonorrhoeae may soon become a true superbug causing untreatable gonorrhea (17). Other studies have also shown that the penA mutation was important in the population genetics of N. gonorrhoeae isolates (27, 28).

It is clear that CRO resistance in N. gonorrhoeae is not entirely explained by mosaic penA or the mutation of other known genes. All the studies mentioned above were based on clinical gonococcal isolates, which gain CRO resistance probably mainly by transformation. However, antimicrobial pressure, which frequently occurs due the abuse of antimicrobial drugs, may also be an important factor in the emergence of CRO resistance in N. gonorrhoeae. Therefore, to understand further the molecular mechanism of CRO resistance in N. gonorrhoeae caused by selective pressure of antimicrobial treatment, we applied increasing concentrations of CRO to one clinical and five reference gonococcal strains that are sensitive to CRO in vitro. One resistant strain emerged, and we used this strain to perform PCR to detect frequently reported genes associated with CRO resistance, and differential analysis of genomes and transcriptomes between the pre- and postselection strains was performed.

MATERIALS AND METHODS

Neisseria gonorrhoeae strains.

One gonococcal clinical isolate, GD1337, obtained from Guangdong Provincial Dermatology Hospital (GDDH), and five reference strains of N. gonorrhoeae, WHO-A, WHO-B, WHO-C, WHO-D, and WHO-E, presented by the WHO, were included for selection (29). Two other gonococcal strains, ATCC 43069 (reference strain obtained from the ATCC) and GD1372 (clinical isolate obtained from GDDH), were also included for transformation assays. All of the strains passed species verification according to previously described methods (17). The CRO MICs of these strains were determined by the agar dilution method as described previously (17, 30), and resistance to CRO was determined based on criteria described by the Clinical and Laboratory Standards Institute (31). The initial CRO MICs of gonococcal strains were limited to <0.016 μg/ml.

In vitro selection of resistance to CRO using increasing concentrations of CRO.

Clinical isolate GD1337 and five gonococcal reference strains, WHO-A, -B, -C, -D, and -E, were cultured on Thayer-Martin (T-M) medium (Oxoid, England) containing different concentrations of CRO at 36°C with 5% CO₂ (32–34). CRO-resistant gonococci were selected with increasing concentrations of CRO starting at the half-initial MIC. As a control, each strain was cultured without drug pressure alongside the drug-selected cultures. Strains were subcultured 3 to 10 times before beginning drug selection. After 18 to 24 h of growth in the initial drug-containing medium, gonococcal cells were passed into two flasks: one containing T-M medium with the same concentration of drug and one with double the amount of drug. Gonococcal cells in the medium containing the same concentration of drug were passaged several more times. Gonococci with the higher concentration of drug were subcultured 3 to 10 times before the drug concentration was increased. When gonococci grew with the higher concentration, this was marked as a step up on the graph (Fig. 1). Species verification of these strains was carried out every 5 to 7 subcultures. All operations were carried out in a biosafety level 2 (BSL-2) laboratory, and the whole process was performed strictly according to requirements of biological safety to prevent contamination by CRO-resistant gonococci.

FIG 1.

In vitro selection of CRO resistance in N. gonorrhoeae. Resistance in postselection gonococcal strains increased after 100 to 103 subcultures. Strain WHO-D showed the highest ceftriaxone MIC (64-fold increase) compared with other WHO reference strains.

Stability test of CRO MICs for the postselection strains.

The postselection strains were subcultured on T-M plates without CRO 20 times continuously, followed by MIC determination to evaluate the stability of CRO resistance without selection pressure. To test the stability of CRO resistance after cryopreservation, the postselection strains were stored in a −80°C ultra-low-temperature freezer for 4 weeks, followed by reproductive culture and MIC determination.

Detection of β-lactamase in strains.

β-Lactamase production was detected by using nitrocefin discs. Strains WHO-A and -E (penicillinase-producing Neisseria gonorrhoeae [PPNG]) were used as a negative control and a positive control, respectively.

Nucleotide sequencing and analysis of the penA, ponA, mtrR, penB, gyrA, and parC genes.

Three clones of the postselection ceftriaxone-resistant strain were selected and subcultured for further analysis. Each gene was amplified by PCR using primers, reaction mixtures, and amplification parameters described previously (20, 35). Agarose gel electrophoresis was performed with 5 μl of PCR products to verify the approximate molecular weight of DNA segments. The remaining PCR mixture was purified by using a Jetquick PCR Product Purification Spin kit (Genomed, Germany), and the purified products were sequenced by using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, USA) on an ABI Prism 377 DNA sequencer (Applied Biosystems, USA). The NCBI BLAST program was used to execute alignments of the DNA and amino acid sequences between the postselection resistant strain and wild-type strain LM-306 (GenBank accession number M32091) as well as between the pre- and postselection strains.

Genomic sequencing and analysis of the preselection sensitive and postselection resistant strains.

Genomic DNA was extracted from both the preselection and postselection strains. DNA quality was determined to ensure that it met all requirements for follow-up experiments. Genomic DNA was cut randomly into fragments of a requisite length and then purified by gel electrophoresis. Short adaptors were ligated onto the ends of the fragments for preparation of clusters and construction of a DNA library. Sequencing was performed with the Illumina HiSeq 2000 system. Raw sequence reads were filtered by quality score, and the data were then assembled by using the SOAPdenovo short oligonucleotide analysis package (version 1.05). Final DNA sequences were assembled with Glimmer software (version 3.0) and aligned with the sequence of reference strain FA1090 (GenBank accession number NC002946) from the PubMed database to ascertain functional annotation of the assembled genomes. Sequence alignments between the preselection and postselection strains were also performed, and the differential genes were further searched in PubMed to screen for CRO resistance-related loci. Additionally, to determine when the mutations of the CRO resistance-related differential genes emerged, the genes were amplified by PCR and sequenced as described above before and after each major jump in resistance.

Transformation assays.

To determine whether the increased resistance to CRO was due to the mutated genes found by genomic analysis, the purified PCR-amplified full-length mutated genes or the DNA mixture (molar ratio of 1:1) from postselection strain WHO-D was transformed into new gonococci according to procedures described previously (17, 18, 36). Three recipients were used: preselection strain WHO-D, N. gonorrhoeae reference strain ATCC 43069, and clinical isolate GD1372. These recipients displayed different CRO MICs and compositions of penA alleles, mtrR promoter mutations, penB, ponA, and the mutated genes found by genomic analysis (Table 1). Each transformation assay was repeated 3 times, and the mean MIC values were determined.

TABLE 1.

Neisseria gonorrhoeae strains with different CRO MICs and gene compositions used as recipients in transformation experiments with full-length novel genes

Strain	CRO MIC (μg/ml)	Genotypea
		penA	mtrR	penB	ponA	ftsX	phbP	farB	hfq
ATCC 43069	0.001	penA XV (WT)	WT	WT	WT	WT	WT	T239V	WT
WHO-D (preselection)	0.016	penA XII	A-del	WT	L421P	WT	WT	WT	WT
GD1372	0.064	penA XVIII	A-del	WT	L421P	R251H	WT	T239V	WT

^aWT, wild type; penA XV, penA sequence pattern XV; A-del, a characteristic single-nucleotide (A) deletion in the inverted repeat of the promoter region of mtrR.

Transcriptional sequencing and analysis of the preselection sensitive and postselection resistant strains.

Total RNA was isolated with the EZNA bacterial RNA kit (Omega Bio-Tek, USA) from both the preselection and postselection strains according to the manufacturer's instructions. Quality detection was carried out to ensure that RNA met all requirements for follow-up experiments. rRNA was removed with the Ribo-Zero rRNA removal kit for Gram-negative bacteria (Epicentre of Illumina, USA), and mRNA was then cut with fragmentation buffer. The first cDNA strand was synthesized with random hexamers using mRNA as the template, followed by synthesis of the second cDNA chain after treatment with a buffer solution, deoxynucleoside triphosphates (dNTPs), RNase H, and DNA polymerase I. cDNAs were purified with a QIAquick PCR purification kit (Qiagen, Germany) and eluted with ethidium bromide buffer solution, followed by terminal repair and ligation with sequencing adaptors. After selection by gel electrophoresis, fragments were amplified by PCR for the construction of a cDNA library, and sequencing was performed with the Illumina HiSeq 2000 system. Raw sequencing reads were filtered, resulting in clean reads. Sequences were aligned with the sequence of reference strain FA1090 (GenBank accession number NC002946) for functional annotation. Sequence alignments of the preselection and postselection strains were also performed. Differential genes were annotated by using Gene Ontology (GO) functional analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, and PubMed searches to screen for CRO resistance-related loci.

Nucleotide sequence accession number.

Genomic sequences of preselection and postselection WHO-D strains are found under EMBL accession number ERP001822.

RESULTS

Variation of CRO MICs from the preselection to postselection strains.

As shown in Fig. 1 and Table 2, the CRO MICs of the postselection gonococcal strains increased 4-fold to 64-fold after the selection of 100 to 103 subcultures. Among the postselection strains, strain WHO-D acquired the largest increase (64-fold), and its CRO MIC reached 1.024 μg/ml, indicating that it gained resistance to CRO. After the observation endpoints, strain WHO-D was selected additionally for 33 subcultures, but the MIC remained constant. Additionally, when selected with increasing concentrations of CRO, the gonococcal strains, particularly strain WHO-D, also displayed increased MICs for penicillin and cefuroxime. This result indicates that N. gonorrhoeae can produce cross-resistance to penicillin and cefuroxime when developing resistance to CRO.

TABLE 2.

MIC variations of several antimicrobial drugs from preselection to postselection strains^a

Strain	No. of subcultures	Postselection MIC (μg/ml)/preselection MIC (μg/ml) (increase rate [%])
		CRO	Penicillin	Cefuroxime	Tetracycline	Ciprofloxacin	Spectinomycin
WHO-A	103	0.064/0.001 (64)	0.128/0.012 (11)	0.256/0.016 (16)	0.256/0.256 (1)	0.016/0.016 (1)	>512/>512 (1)
WHO-B	100	0.016/0.004 (4)	0.128/0.128 (1)	0.016/0.016 (1)	0.256/0.256 (1)	0.016/0.016 (1)	8/8 (1)
WHO-C	100	0.064/0.016 (4)	0.512/0.512 (1)	1.024/0.128 (8)	1.024/1.024 (1)	0.016/0.016 (1)	16/16 (1)
WHO-D	136	1.024/0.016 (64)	16.38/2.048 (8)	>16.38/2.048 (>8)	4.096/4.096 (1)	0.06/0.03 (2)	16/16 (1)
WHO-E	103	0.128/0.002 (64)	16.38/16.38 (1)	0.256/0.032 (8)	0.512/0.512 (1)	0.016/0.016 (1)	16/16 (1)
GD1337	100	0.128/0.016 (8)	4.096/2.048 (2)	4.096/2.048 (2)	2.048/2.048 (1)	8.192/8.192 (1)	16/16 (1)

^aIt was more difficult for N. gonorrhoeae isolates to produce drug resistance to erythromycin than to tetracycline and ciprofloxacin in our previous study (54); therefore, erythromycin was not included for MIC determination.

Stability test of the CRO MIC in postselection strain WHO-D.

Postselection strain WHO-D was cultured on T-M plates without CRO continually for 20 subcultures, MIC determination was then carried out, and the data showed no obvious decline. Furthermore, postselection strain WHO-D was stored at −80°C for 4 weeks, with no significant change in the CRO MIC. This demonstrated that the increased MIC of postselection strain WHO-D was stable, with little probability of pressure-driven loss of resistance.

Detection of β-lactamase in strain WHO-D.

Both the pre- and postselection strains tested negative for β-lactamase, suggesting that the increased MIC of postselection strain WHO-D was unlikely to be mediated by β-lactamase production.

Nucleotide sequencing and analysis of the penA, ponA, mtrR, penB, gyrA, and parC genes of preselection and postselection WHO-D.

Three clones of the postselection ceftriaxone-resistant strain were selected and subcultured for further analysis, and all the results were consistent. As shown in Fig. 2, PBP2 of preselection strain WHO-D was of the XII type, containing 582 amino acids (aa), and after selection, it acquired an A501V mutation, known as the XIII type (37). Mosaic penA did not emerge in strain WHO-D with significantly increased CRO MICs. Moreover, there was no variation of ponA, mtrR, penB, gyrA, and parC between the preselection and postselection WHO-D strains. These findings were observed for all three clones.

FIG 2.

Multiple-sequence alignment of the amino acid sequences of PBP2 from preselection and postselection strain WHO-D with reference to LM-306. PBP2 in preselection strain WHO-D was of the XII type, containing 582 amino acids. After selection, PBP2 acquired an A501V mutation, known as the XIII type.

Genome sequencing and analysis of preselection sensitive and postselection resistant WHO-D.

As shown in Fig. 3 and Table 3, the sequencing reads were of high accuracy and quality, and there was good coverage of the genome for both pre- and postselection strains. Based on differential analysis of genomes between the preselection and postselection WHO-D strains, we found 92 mutated genes with lengths that varied from 105 to 4,020 bp. The gene consistency ranged from 85.6% to 99.98%, and the corresponding number of mismatched bases ranged from 1 to 24. Twenty-three of the 92 mutated genes were unannotated in GenBank. The remaining 69 mutated genes could be divided into three categories: 6 genes encoding cellular structure proteins, 58 genes associated with biochemical metabolism and signal transduction, and 5 genes related to drug resistance (Table 4). The 5 genes related to drug resistance encoded PBP2, the ABC transporter subunit FtsX (termed FtsX), ABC transporter periplasmic histidine-binding protein (termed PhbP), FarB, and Hfq. All 5 genes carried point mutations with no base insertions or deletions. Moreover, it was observed that the hfq mutation emerged in strain WHO-D after it was subcultured 19 times, farB and phbP mutations emerged after 33 subcultures (the phbP mutation arose before the jump, and the farB mutation arose after the jump, indicating that the farB mutation caused the jump in resistance), the penA mutation emerged after 47 subcultures, and the ftsX mutation emerged after 86 subcultures.

FIG 3.

Quality detection of genome sequencing reads of preselection and postselection strain WHO-D. Almost all of the quality values of the reads were >20, indicating the high accuracy of these reads. (A) Quality detection of preselection strain WHO-D. (B) Quality detection of postselection strain WHO-D.

TABLE 3.

Alignment statistics for the genome sequencing products

Strain	No. of mapped reads of reference gene	Length of mapped-read reference gene region (bp)	Gene no. coverage (%)	Gene length coverage (%)	Genome coverage (%)
Preselection WHO-D	1,983	1,667,389	97.83	97.09	98.54
Postselection WHO-D	1,984	1,666,971	98.03	96.68	99.14

TABLE 4.

Mutated genes in postselection resistant strain WHO-D^a

Length (bp)	No. of mismatches	Gene annotation	Mutation(s)
1,749	1	PBP2	A501V (vs LM306)
918	1	ABC transporter subunit FtsX	T31P
807	1	ABC transporter periplasmic histidine-binding protein (PhbP)	G192V
816	2	FarB	T239V
201	4	Hfq	R62P, S63P, V64P

^aThere were no gaps in the sequences of the mutated genes.

Transformation assays confirm that mutated penA and ftsX genes contribute to increased CRO MICs in postselection resistant strain WHO-D.

The results showed the mutated penA and ftsX genes conferred significant resistance to CRO (Fig. 4). Transformation of the mutated penA or ftsX gene into the recipient strains resulted in CRO MICs that were increased 2.0- to 5.3-fold and 3.3- to 13.3-fold, respectively, while the penA_H041 positive control resulted in CRO MICs that were increased by between 26.1- and 412.0-fold. Transformation with phbP, farB, or hfq did not significantly increase the CRO MICs of the recipient strains. Remarkably, transformants of ATCC 43069, which has wild-type alleles of all known CRO resistance determinants, displayed the highest MIC ratios compared with transformants of the other two recipients. Furthermore, transformation of the DNA mixture of mutated penA and ftsX into the preselection WHO-D strain resulted in the CRO MIC being increased 16-fold (0.256 μg/ml).

FIG 4.

Transformation of the purified PCR-amplified full-length novel genes from postselection strain WHO-D into three recipient strains with different CRO MICs and gene compositions. The CRO MICs were determined by using the agar dilution method, and the MIC ratios of the transformant to the recipient strain (T/R) are the means of data from three transformation experiments.

Transcriptome sequencing and analysis of preselection sensitive and postselection resistant WHO-D strains.

The total mapped-read coverage of preselection strain WHO-D was 93.10%, and that postselection strain WHO-D was 80.28% (Table 5). The distribution of gene coverage was also good for both preselection and postselection WHO-D (Fig. 5). Using differential analysis of transcriptomes of strain WHO-D preselection and postselection, we found 306 differentially expressed genes (DEGs). The gene length ranged from 126 to 5,934 bp, and the log₂ ratio of the RNA quantity of postselection strains to that of preselection strains varied from −3.33 to 3.47. These 306 DEGs consisted of 78 downregulated genes and 228 upregulated genes. The 78 downregulated DEGs comprised 24 genes without functional annotation and 54 with functional annotation. The 54 downregulated DEGs with functional annotation were further divided into three categories: 3 genes encoding cellular structural proteins, 46 genes involved in biochemical metabolism and signal transduction, and 5 genes related to drug resistance (Table 6). The 5 genes related to drug resistance contained genes encoding an ABC transporter ATP-binding protein, an ABC transporter permease protein, PilM, PilN, and PilQ. The 228 upregulated DEGs consisted of 76 genes without functional annotation and 152 with functional annotation, and the 152 upregulated DEGs with functional annotation were also divided into three categories. These genes included 20 genes encoding cellular structural proteins, 123 genes involved in biochemical metabolism and signal transduction, and 9 genes related to drug resistance (Table 6). The 9 genes related to drug resistance consisted of two genes encoding an ABC transporter permease protein; two genes encoding an ABC transporter ATP-binding protein; the gene encoding the ABC transporter periplasmic binding protein; and genes encoding FarB, Tfp, Hfq, and ExbB.

TABLE 5.

Alignment statistics for the transcriptome sequencing products

Strain	Total no. of reads	Total size of reads (bp)	Total mapped-read coverage (%)
Preselection WHO-D	6,574,652	591,718,680	93.1
Postselection WHO-D	6,434,662	579,119,580	80.28

FIG 5.

Distribution of gene coverage in transcriptome sequencing of preselection and postselection strain WHO-D. (A) Distribution of gene coverage in preselection strain WHO-D. (B) Distribution of gene coverage in postselection strain WHO-D.

TABLE 6.

DEGs in postselection resistant strain WHO-D

Gene product	Log2 ratio	Regulation type
PilN	−1.82	Down
ABC transporter ATP-binding protein	−1.36	Down
PilM	−1.26	Down
ABC transporter permease protein	−1.20	Down
PilQ	−1.14	Down
ABC transporter permease protein	2.19	Up
ABC transporter periplasmic binding protein	2.08	Up
ABC transporter ATP-binding protein	2.05	Up
FarB	1.95	Up
ABC transporter permease protein	1.87	Up
Type IV pilus (termed Tfq)	1.74	Up
Hfq	1.46	Up
ExbB	1.28	Up
ABC transporter ATP-binding protein	1.11	Up

GO functional analysis demonstrated that GO terms, which showed a significant enrichment in DEGs of postselection compared to preselection WHO-D, included mainly integral membrane proteins (such as ABC transporters) and proton-transporting proteins in ontology of cellular components. In ontology of molecular function, enriched GO terms consisted primarily of intramolecular oxidoreductase activity, interconverting aldoses and ketoses, chloride channel activity, and wide-pore channel activity. In ontology of biological processes, enriched GO terms mainly comprised glucose catabolic processes and pyrimidine deoxyribonucleoside triphosphate biosynthetic processes.

DISCUSSION

In this study, analysis of frequently reported genes associated with CRO resistance in N. gonorrhoeae revealed that only penA was mutated in the CRO-resistant WHO-D strain. Specifically, we detected an A501V mutation of penA, which contributes to decreased susceptibility to CRO (38, 39). The A501V mutation that arose spontaneously in our study was also observed to arise spontaneously in a separate study conducted by Takahata et al. (40). In contrast, mosaic penA was mainly responsible for CRO resistance based on studies of clinical isolates H041 and F89 (15, 17, 18, 37). Gonococcal mosaic penA originated from gene fragment exchange and reorganization of the transpeptidase coding region of penA with other Neisseria bacteria (24, 41), which is thought to have occurred by frequent commensalism of Neisseria bacteria clinically. In this study, we prevented the evolution of mosaic penA by selecting gonococci in a closed environment, and this resulted in an A501V mutation in penA. However, we further verified that transformation of penA-A501V into preselection sensitive strain WHO-D increased the MIC of CRO by only 3.3-fold, and postselection gonococcal strain WHO-D reached the level of resistance to CRO (increasing by 64-fold), suggesting that there are probably other unknown molecular mechanisms involved in the resistance of N. gonorrhoeae to CRO.

To explore these unknown molecular mechanisms, we combined differential analysis of genomes of the preselection sensitive and postselection resistant strain WHO-D with transformation assays of the mutated genes described above. We found that mutation of the ftsX gene, encoding a subunit of an ABC transporter, increased the CRO MIC of the preselection sensitive WHO-D strain 6.7-fold and also increased CRO resistance significantly if the A501V mutation of the penA gene was already present. This result suggests that ftsX could be another determinant of resistance to CRO in addition to mutated penA. Previous studies demonstrated links between ABC transporter proteins and drug resistance. For example, the enterococcal ABC transporter gene lsa(E) conferred resistance to lincosamides in Staphylococcus aureus (42), and a predicted heterodimeric ABC transporter, TetAB, conferred tetracycline resistance in Streptococcus australis (43). Moreover, ABC transporters together with neighboring two-component systems, which regulate their expression, could coevolve to form self-sufficient detoxification modules against antimicrobial peptides in Firmicutes bacteria (44). A significant correlation between penicillin resistance and upregulation of ABC transporters was also reported (45). These studies indicate that the overexpression of ABC transporters might also be involved in CRO resistance by enhancing the activity of drug transport.

Because the magnitude of increased CRO resistance in strain WHO-D could not be accounted for by both penA and ftsX mutations, a differential analysis of transcriptomes of preselection sensitive and postselection resistant strain WHO-D was performed. We found 306 DEGs consisting of 78 downregulated genes (including PilMNQ) and 228 upregulated genes (including ABC transporters, FarB, Tfq, Hfq, and ExbB). Most of the 306 DEGs were involved in biochemical metabolism and signal transduction, and the enriched GO terms mentioned above also suggested an increased biochemical metabolic rate in postselection strain WHO-D. As predicted, the overexpression of ABC transporters was enriched. Additionally, a few other abnormally expressed genes that are involved in CRO resistance were found, such as farB, hfq, exbB, and pilQ. It has been reported that the farAB-encoded efflux pump could result in resistance to antibacterial fatty acids in gonococci (46), and Hfq regulates the efflux system at the posttranscriptional level in Escherichia coli (47) and Salmonella enteritidis (48). A study by Rouquette-Loughlin et al. revealed that the TonB-ExbB-ExbD system was required for full selection with hydrophobic agents in gonococci (49), consistent with our discovery of overexpressed ExbB during selection with a high concentration of CRO. Similar to our finding that pilQ was not mutated in CRO-resistant strain WHO-D, a recent study of the relationship between a pilQ polymorphism and decreased susceptibility to extended-spectrum cephalosporins (ESCs) in clinical gonococcal isolates indicated that the mutation of pilQ had no significant correlation with CRO resistance (26). However, in our differential analysis of transcriptomes, the expressions of PilM, PilN, and PilQ were all reduced, suggesting decreased pore sizes in the outer membrane. This might lead to the overexpression of Tfq because downregulated PilQ creates positive feedback for Tfq (50). Moreover, this might be linked with a potential increase in resistance due to decreased influx through PilQ pores.

CRO resistance is commonly produced by extended-spectrum beta-lactamases (ESBLs) or AmpC beta-lactamase, and two primary kinds of mechanisms contribute to antimicrobial resistance to β-lactam antibiotics in gonococci. The first mechanism is mediated by a resistance plasmid that produces β-lactamase. This kind of resistance is usually high level, quickly acquired, and easy to transfer and diffuse among strains (51). The second mechanism is mediated by chromosomal genes. This kind of resistance often takes a relatively long time for the gradual accumulation of multiple resistance-related gene mutations to occur, and the degree of resistance often depends on the loci and quantity of the mutations (21). In this study, the CRO MICs of gonococci increased gradually, there was no β-lactamase production in the postselection gonococcal strains, and there was no sequence encoding β-lactamase in the differential genes, which all suggest that CRO resistance in N. gonorrhoeae was mediated by chromosomal genes. Furthermore, we demonstrated that antimicrobial pressure could lead to the emergence of CRO resistance in N. gonorrhoeae, the ability to resist CRO varies among the different gonococcal strains, and N. gonorrhoeae can produce cross-resistance to penicillin and cefuroxime when developing resistance to CRO. Therefore, drug abuse during treatment of gonorrhea should be avoided, and monitoring of CRO-resistant isolates needs to be strengthened to attempt to screen out latent high-risk gonococci.

Interestingly, we discovered that when N. gonorrhoeae began to grow on increasing concentrations of CRO, the growth time was prolonged, even though the biochemical metabolic rate was increased. The decrease in growth time could be partially recovered after several subcultures of N. gonorrhoeae in medium with the same CRO concentration (Fig. 6). A similar phenomenon was reported by Vincent et al., who created FA19^penA41 and FA19^penA89 mutants by replacing the wild-type penA allele of strain FA19 with penA^H041 or penA^F89 and found that both mutants grew significantly more slowly than FA19, demonstrating a fitness defect (52). More importantly, another fitness defect is found in the downregulation of PilM, PilN, and PilQ, because this could affect proper pilus formation in gonococcal strains, making it harder for these strains to infect hosts. These fitness defects might partially account for the most recent surveillance results showing the there was no dissemination of an N. gonorrhoeae strain with high-level CRO resistance identified in Japan after H041 was isolated (53).

FIG 6.

Time curve of entering plateau during the selection of CRO resistance in N. gonorrhoeae. The growth time was prolonged when N. gonorrhoeae began to grow on increasing concentrations of CRO. The decrease in growth time was partially recovered after several subcultures of N. gonorrhoeae in medium with the same CRO concentration.

In conclusion, antimicrobial pressure can result in the emergence of CRO resistance. Furthermore, the evolution of resistance of N. gonorrhoeae to CRO is not just a simple change of some frequently reported genes but a complicated process at both the pretranscriptional and posttranscriptional levels involving several resistance mechanisms of increased efflux and decreased entry. We were surprised to discover several novel genes related to CRO resistance and that the ability of N. gonorrhoeae to develop resistance to CRO varied; N. gonorrhoeae produced cross-resistance to other drugs; the biochemical metabolic rate was increased; the growth time was prolonged; and PilM, PilN, and PilQ were downregulated. All of these findings may help deepen our understanding of CRO resistance in N. gonorrhoeae and manage CRO-resistant N. gonorrhoeae appropriately. Moreover, to further address whether the resistance developed from the gene mutations and/or whether other changes in gene transcription may be necessary, the use of a stepwise process of transformation is the best way. This experiment is now being tried, and we will present the results in the future.