Recent advances in understanding and combatting Neisseria gonorrhoeae: a genomic perspective

The sexually transmitted infection (STI) gonorrhoea remains a major global public health concern. The World Health Organization (WHO) estimates that 87 million new cases in individuals who were 15 to 49 years of age occurred in 2016. The growing number of gonorrhoea cases is concerning given the rise in gonococci developing antimicrobial resistance (AMR). Therefore, a global action plan is needed to facilitate surveillance. Indeed, the WHO has made surveillance leading to the elimination of STIs (including gonorrhoea) a global health priority. The availability of whole genome sequence data offers new opportunities to combat gonorrhoea. This can be through (i) enhanced surveillance of the global prevalence of AMR, (ii) improved understanding of the population biology of the gonococcus, and (iii) opportunities to mine sequence data in the search for vaccine candidates. Here, we review the current status in Neisseria gonorrhoeae genomics. In particular, we explore how genomics continues to advance our understanding of this complex pathogen.


Introduction
Neisseria gonorrhoeae, the gonococcus, is an obligate human pathogen that causes the sexually transmitted infection gonorrhoea 1 . Gonococcal infection often results in severe complications, ranging from disseminated infection to salpingitis or pelvic inflammatory disease, and gonococcal infections are found to be asymptomatic in over 50% of women and associated with a significant cause of infertility 2 . Therefore, prompt diagnosis and treatment are essential for a positive clinical outcome. However, the gonococcus has developed resistance against all available classes of antimicrobials and options for the effective treatment of this infection are becoming increasingly limited 2,3 . This is particularly concerning as gonorrhoea, after Chlamydia trachomatis, is the second most prevalent sexually transmitted infection and is globally associated with high levels of morbidity and economic cost 4 .
Infections are diagnosed by microbiological culture (with Gram stain in men with urethritis) or nucleic acid amplification tests. In settings with limited laboratory capacity, diagnosis is often based on clinical symptoms alone, followed by syndromic management. This commonly includes the use of doxycycline, an antibiotic that targets both N. gonorrhoeae and C. trachomatis, which present with similar clinical symptoms 5 . However, treatment has become complicated by the rapidly changing antimicrobial susceptibility patterns of N. gonorrhoeae, raising concerns about the development of untreatable gonorrhoea. Internationally, the prevalence of gonococci resistant to most antimicrobials recommended for treatment is high. For example, the recent occurrence of failures to treat gonorrhoea with the extended-spectrum cephalosporins (ESCs) cefixime and ceftriaxone and, the emergence of gonococci exhibiting high-level resistance to all other available therapeutic antimicrobials 6-8 , have caused great concern and highlight the need for a targeted approach to limit the spread of infection and prevent transmission globally.
Understanding the causative agents of disease is paramount to curing or preventing disease, and the availability of whole genome sequence (WGS) data offers novel opportunities to do this. In particular, a large amount of N. gonorrhoeae WGS data have become available, allowing an increasingly forensic examination of this bacterium. The knowledge gained from these analyses strengthens our capacity to detect antimicrobial resistance (AMR) while improving our understanding of the resistance phenotype and how it evolves. This knowledge also allows the population biology of the gonococcus to be resolved, facilitating global surveillance. Finally, WGS data provide opportunities for vaccine development through the characterisation of antigenic variability. Here, we review progress made in the field of gonococcal genomics, examining how WGS data have increased our knowledge of this pathogen. In particular, we focus on advances made in the detection of AMR, the genetic determinants conferring this phenotype, and the tools developed to identify AMR. We assess progress made in understanding the population biology of this pathogen and discuss the current status in vaccine development and how WGS can support this activity. To conclude, we review the challenges that remain and discuss the approaches needed to overcome them.

Antimicrobial resistance
In the absence of vaccines, antimicrobial therapy is critical to cure infection and limit its spread 9 . The effectiveness of antibiotics to treat gonorrhoea is becoming limited by the global emergence and spread of gonococci resistant to antimicrobials, combined with increasing rates of gonorrhoea. The total number of confirmed cases reported in 2018 in 28 European Union/European Economic Area (EU/EEA) member states was 100,673 (with an overall crude notification rate of 26.42 per 100,000 population), representing a 12% increase in the total number of reported cases compared with 2017 10,11 . In the UK alone, 61,775 confirmed cases were reported in 2018, representing a 26% increase compared with 2017 10 . In the US and its territories, 556,413 cases of gonorrhoea were reported in 2017, representing a rate of 168.9 cases per 100,000 population and a 19% increase compared with 2016 12 . In Australia, the number of confirmed cases rose from 23,875 in 2016 to 28,364 in 2017 with a rate of 118 per 100,000 population 13 . This was associated with a 63% increase in notification rates between 2012 and 2016. In China, 133,156 confirmed cases were reported in 2018, representing a rate of 10.06 cases per 100,000 population 14 . Similarly, a rate of 88.6 cases per 100,000 population was identified in 2015 in Japan 10,15 . These data are consistent with a global trend of increasing rates of gonorrhoea. However, there are differences in the reporting of incidence and prevalence data among countries, even where effective surveillance systems are in place. In addition, different jurisdictions have their own surveillance methodologies and diagnostic tests such that global estimates have to be interpreted with caution 3 . So it is challenging to assess effectively global rates of gonorrhoea prevalence.
This increasing trend in gonococcal prevalence is accompanied by global reports of gonococci exhibiting decreased susceptibility to several classes of antimicrobials 2,16 . The gonococcus employs multiple genetic mechanisms to overcome the effects of antimicrobial agents and is extremely adept in developing AMR (Figure 1). Most of the genetic determinants that confer AMR in N. gonorrhoeae are chromosomally encoded (Table 1), with plasmid-mediated AMR associated only with the blaTEM gene, which confers high-level resistance to penicillin [17][18][19] , and the tetM gene associated with resistance to tetracycline 20,21 (Table 1). Resistance to fluoroquinolones is conferred through the presence of non-synonymous mutations in DNA gyrase, gyrA, or the topoisomerase IV gene, parC, and amino acid mutations in both GyrA and ParC result in higher levels of resistance to fluoroquinolones 22,23 .
An increasing level of resistance or reduced susceptibility to beta-lactams, including penicillin and third-generation cephalosporins, is often the result of combinations of mutations in several genetic determinants (Table 1). These include (i) the Figure 1. Antimicrobial resistance mechanisms in the gonococcus. Schematic figure of a gonococcus depicting chromosomally and plasmid-mediated resistance determinants. Multiple antibiotics target the gonococcus against which the bacterium has in turn developed resistance. This includes (i) fluoroquinolones (e.g., ciprofloxacin), which inhibit the activity of topoisomerases involved in DNA replication. Resistance to fluoroquinolones is conferred by non-synonymous mutations in the topoisomerase-encoding genes gyrA and parC 24,25 ; (ii) beta-lactams target penicillin binding protein 1 (PBP1) and 2, which are required for peptidoglycan synthesis. Amino acid mutations in ponA and penA, which encode PBP1 and PBP2, respectively, confer resistance to beta-lactams 26-28 . Gonococci are increasingly developing reduced susceptibility to third-generation cephalosporins, including cefixime and ceftriaxone, with the presence of multiple genetic determinants required for this phenotype, including mosaic penA alleles 29-31 . More recently, mutations in rpoB and rpoD have been found to confer resistance to ceftriaxone only in the absence of mosaic penA alleles 32 ; (iii) spectinomycin and macrolides, including azithromycin, bind to ribosomes inhibiting protein synthesis. Mutations in the 16S and 23S rRNA confer resistance to spectinomycin and azithromycin, respectively 33-35 . Additional chromosomally mediated resistance determinants include amino acid mutations in loop 3 of the outer membrane porin PorB, which reduces permeability, thereby inhibiting the influx of antibiotics through the porin channel 36 . Mutations in efflux pump genes mtrRCDE, particularly in the promoter of mtrR, which encodes the MtrR repressor protein, lead to overexpression and increased activity of the efflux pump complex, conferring resistance 37-39 . Mosaic mtrD genes have been found to confer resistance to azithromycin 40 . Plasmidencoded resistance includes the plasmid-encoded TEM-1 penicillinase (bla TEM-1 ), which inactivates the beta-lactam ring of penicillin and the conjugative plasmid harbouring the tetM gene 19,21 . IM, inner membrane; OM, outer membrane; PG, peptidoglycan. Red stars indicate non-synonymous nucleotide mutations resulting in amino acid mutations; black stars nucleic acid mutations in the rRNA nucleotide sequences. Green lines indicate antibiotic targets; dashed red lines depict resistance mechanisms. presence of mosaic penA genes encoding penicillin binding protein 2, genetic fragments of which often are acquired through horizontal gene transfer (HGT) from commensal Neisseria species 41-43 ; (ii) non-synonymous mutations in the ponA gene, which encodes penicillin binding protein 1 26,27 ; (iii) the presence of mutations in the mtrR repressor gene or its promotor sequence, resulting in overexpression of the MtrCDE efflux pump system and secretion of antimicrobial agents 37,38 ; and

Beta-lactams, including penicillin (inhibits cell wall synthesis, bactericidal)
blaTEM Beta-lactamase plasmid containing the blaTEM gene encoding the enzyme beta-lactamase which hydrolyses the peptide bond of the four-membered beta-lactam, thus inactivating the antibiotic.
Jo (Johannesburg) (4.8 kb) and (vii) pbla.Au (Australia) (3.2 kb). Additional plasmids have been described, including the Toronto plasmid, a derivative of the Asian plasmid, and a novel Canadian plasmid variant of the Africa-type possessing a 6-bp deletion at the 5′ end of blaTEM.

Penicillins, beta-lactams and third-generation cephalosporins (bactericidal; inhibit cell wall synthesis)
penA Penicillin binding protein 2, PBP2: It is implicated in peptidoglycan biosynthesis and encodes a monofunctional transpeptidase. Amino acid mutations in the transpeptidase region are associated with reduced susceptibility. Several mosaic penA alleles with up to 70 amino acid alterations have been described with, in particular, mosaic allele XXXIV associated with decreased susceptibility to third-generation cephalosporins (when found concomitantly with other mutations). NEIS1753 alleles 266,281,498,660,1240,1503,1522,1523,1524,1525 and 1526 are XXXIV mosaic alleles. PorB is a porin that mediates ion exchange with the environment. It is essential for viability because of its ability to allow nutrient access to the periplasm. Non-synonymous mutations in loop 3 of PorB allele PIB (G120→K, G120→D, G120→N and/or A121→D, A121→ N, A121→G) reduce influx of antibiotics. NEIS2020 alleles: 204,524,539,634,792,832,1140,1608,1611,1901,1961,1993,2032,2083,2136,2200,2588,3767,3824,3833,3972  A single-nucleotide polymorphism in the spectinomycin region of 16S rRNA has been shown to result in high level resistance to spectinomycin. Deletion of a valine at residue 25 and a K26→E alteration in the 30S ribosomal protein gene S5, rpsE, have been shown to confer high-level spectinomycin resistance.

NEIS0149 (rspE) 16S_rRNA
(iv) mutations of the outer membrane channel porin PorB variant PIB, resulting in a decreased influx of penicillin, through reduced permeability of the porin 26,36 . In addition, mutations in rpoB, encoding RNA polymerase subunit B, and rpoD, encoding the RNA polymerase sigma factor, have been identified (Table 1) 32 . These mutations, which cause large-scale transcriptional changes, including an increase in ponA expression, were detected in clinical isolates which lacked penA mosaic alleles and yet exhibited reduced susceptibility specifically towards ceftriaxone only. The number of potential mutations implicated in conferring resistance to beta-lactams highlights the genetic diversity and complexity of this genotype and the possibility that other as-yet-unknown mutations are present or indeed will arise in the future.
The macrolides, azithromycin and erythromycin inhibit protein synthesis by binding to the 23S rRNA component of the 50S ribosome. Mutations in 23S rRNA can lead to resistance and include a C2611T change resulting in low levels of resistance or an A2059G mutation leading to increased resistance 61,62 . Levels of resistance are also dependent on how many of the four copies of 23S RNA contain either mutation such that, for example, A2059G mutations, if present in three or all four of the 23S rRNA alleles, result in high-level azithromycin resistance 6,62 . Mosaic mtr alleles are also associated with conferring resistance to azithromycin 40,57-59 . The mechanism of resistance is distinct from those associated with increased expression of the MtrCDE efflux pump and is due to the presence of mosaic mtr operons where, in particular, full-length mosaic mtrD genes combined with a mosaic mtr operon lead to much higher levels of resistance to azithromycin compared with isolates with mosaic mtr operons alone. Such mtr operons were found to result from interspecies HGT with other Neisseria species, including N. meningitidis and N. lactamica. Finally, resistance to spectinomycin is conferred through mutations in the rpsE gene, which encodes the S5 ribosomal protein, and T24P mutations that induce low levels of resistance 63 . A deletion of a valine residue at amino acid 25 and a K26E amino acid substitution have also been found to be associated with higher levels of resistance 60 . Resistance to spectinomycin can also occur through a C1192U nucleotide polymorphism in 16S rRNA 33 .
In addition to the genetic mutations described above, all of which expand in response to antimicrobial exposure, selection for resistance in gonococci may occur as a consequence of indirect exposure through the consumption of antibiotics intended to treat another pathogen 64 . For example, differences in the consumption of fluoroquinolones across 24 EU/EEA countries correlated with ciprofloxacin resistance in gonococci 65 . Seasonal variations in antibiotic consumption have also been associated with fluctuations in numbers of gonococci exhibiting resistance to azithromycin, possibly a consequence of increased use of macrolide to treat respiratory infections 66 . Therefore, a driver in the development of AMR in the gonococcus may also result from the bystander effect, which may be more prevalent than anticipated, particularly since gonorrhoea infections are often asymptomatic.
To monitor resistance, gonococci are classified as either multidrug-resistant (MDR-GC) or extensively drug-resistant (XDR-GC): MDR-GC is defined as a gonococcus with decreased susceptibility to one currently recommended therapy (cephalosporin or azithromycin) plus resistance to at least two other antimicrobials, whereas XDR-GC is a gonococcus with decreased susceptibility to two currently recommended therapies (i.e., cephalosporin and azithromycin) plus resistance to at least two other antimicrobials 67,68 . Rapid detection of MDR-GC or XDR-GC is crucial to limit their spread, and advances made in genomics play an increasingly central role in assisting in this where this technology is available. For example, WGS studies identified an association between the mosaic XXXIV penA allele and reduced susceptibility to the ESC cefixime when found in conjunction with additional resistance factors 29 . Genomic epidemiology studies have also allowed transmission outbreaks of azithromycin-resistant gonococci to be tracked in England 69 , and further studies have elucidated transmission networks in men who have sex with men in Australia 70 . WGS studies identified mosaicisms in the mtrD gene (a component of the mtrCDE efflux pump) that enhance the capacity of the protein to export antimicrobial agents 40 .
Genomic technologies therefore offer the tantalising prospect of enhancing surveillance of AMR with several bioinformatic tools developed to predict AMR from WGS data. These include (i) Antimicrobial Resistance Identification By Assembly (ARIBA), which identifies AMR genes and variants from paired sequencing reads through the use of a combined mapping/alignment and targeted local assembly approach 71 ; (ii) the Pathogenwatch AMR prediction module, PAARSNP, which can be used through the Pathogenwatch portal and which uses resistance databases developed in-house 72,73 ; and (iii) ABRICATE 74 . Performance of these in silico AMR prediction tools is dependent on the availability of accurate and curated AMR databases such as ResFinder 75 and CARD 76 and those reviewed by Hendriksen et al. 77 . Concordance between phenotype and genotype is stronger for some genetic determinants than others. For example, ciprofloxacin resistance can be readily determined from sequence data, whereas predicting beta-lactam-resistant genotypes is more challenging, particularly as this is often an additive effect resulting from multiple mutations. As a result, several studies have encountered difficulties when using WGS to predict such AMR phenotypes 78,79 .
Other approaches have employed multivariate linear regression models to identify genetic predictors of minimum inhibition concentrations (MICs). These models predict MICs for five gonorrhoea antimicrobials that exhibit different genetic mechanisms of resistance: cefixime, penicillin, azithromycin, ciprofloxacin and tetracycline 80 . More recently, a method called 'genomic neighbour typing', which allows the AMR phenotype of a bacterial sample to be predicted through the identification of its closest relatives in a database of genomes which have corresponding AMR metadata, has been described. This method was tested in both Streptococcus pneumoniae and N. gonorrhoeae datasets, and preliminary results indicate that this approach may work well for the pneumococcus given the greater sequence diversity observed in its core genome, which allows specific signature sequences to be more readily detected and matched to individual lineages. However, the lower diversity found in the gonococcal core genome results in the identification of multiple related neighbours which confounds obtained signals 81 .
An alternative approach for tracking AMR in the gonococcus is through the use of a multilocus sequence typing scheme, NG STAR (Neisseria gonorrhoeae sequence typing for antimicrobial resistance), which indexes variability found in nucleotide sequence fragments from seven genes associated with AMR (penA, mtrR, porB, ponA, gyrA, parC, and 23 rRNA) 82 .
Using an online, publicly available database (https://ngstar. canada.ca), users can submit new alleles for curation, allowing variation in genes conferring resistance to be defined and tracked. New allelic profiles are assigned NG STAR sequence types (STs), which are associated with corresponding isolate records. Since its publication, this tool has been used in several international studies assessing AMR, thus providing a tool with which to track the worldwide dissemination of resistant clones. All described genes associated with AMR in the gonococcus, both chromosomally and plasmid-mediated, have also been defined in the pubMLST.org/neisseria database with mutations annotated in allelic variants 83 . This database includes the NG STAR, multi-locus sequence typing (MLST) and N. gonorrhoeae multi-antigen sequence typing (NG MAST v2.0) schemes 84 . As a result, WGSs deposited in PubMLST are annotated in all of these schemes, allowing AMR to be evaluated in combination with conventional typing schemes and providing a publicly available resource for the analysis of gonococcal WGSs. Furthermore, WGSs can be directly queried for any of these schemes without the need for WGSs to be deposited. An additional publicly available tool, Gen2Epi, facilitates gonococcal WGS assembly followed by automatic retrieval of molecular epidemiological information, including NG STAR, MLST and NG MAST and AMR genotypes 85 .

Gonococcal population genomics
The study of genomics will play an increasingly important role in enhancing surveillance of AMR and in the development of molecular diagnostic tools for the detection of this phenotype. AMR emergence has to be studied alongside population structure and genome content since the dynamics of gonococcal population biology is complex and highly influenced by HGT, which results in diversification and reassortment of genetic variation over time 86,87 .
Several molecular typing tools have been developed to identify gonococcal isolates, including (i) MLST, which indexes the diversity found at seven housekeeping gene fragments, and (ii) the NG MAST scheme, where nucleotide sequence fragments of the outer membrane proteins PorB and TbpB are used to define NG MAST STs 88,89 . MLST is based on the characterisation of fragments from housekeeping genes under stabilising selection and remains the method of choice for typing many bacterial species, including Neisseria meningitidis, the meningococcus, for which MLST was first developed 90 . Meningococcal MLST STs provide a convenient means for tracking the epidemiology and population biology of this bacterium since STs can be grouped into coherent groups known as 'clonal complexes' (ccs) 91 . A number of these ccs, 'the hyperinvasive meningococci', are responsible for the majority of invasive disease cases worldwide and are stable over time and global spread 91,92 . The levels of linkage disequilibrium observed in meningococci result in the non-random association of MLST allelic profiles, producing discrete persistent meningococcal lineages, some of which exhibit an invasive phenotype 93 . Studies of the genetic diversity of gonococcal housekeeping genes have shown that, in addition to diversification arising from randomly distributed point mutations, these genes are frequently subject to HGT 94,95 . As a result, some gonococci, though possessing the same seven locus MLST STs, may have different ancestry in the majority of their loci and consequently MLST STs cannot be used to reliably evaluate gonococcal populations 84,96 . In addition, combinations of MLST alleles in gonococci are unlikely to be associated with transmission fitness or stabilising selection in contrast to the meningococcus 94 .
The high rate of HGT observed in the gonococcus led to the assumption that limited structure would be evident in its populations 95 . Genomic population studies have shown otherwise 96,97 . Hierarchical Bayesian analyses (Bayesian Analysis of Population Structure, or BAPS) uses nucleotide polymorphism-based alignments to generate clusters of related isolates after accounting for HGT identified by tools such as Gubbins 98,99 . With this method, gonococcal genome data from several studies have been found to consist of between six and 12 BAPS clusters, some of which associated with AMR phenotypes 96,97,100 . This is consistent with a semi-clonal population structure (i.e., one in which there is residual clonal signal in the population) 96,97,100,101 . The presence of distinct gonococcal lineages was also evident in studies of the gonococcal core genome. For example, more than 1,600 genes were identified as 'core' in WGS data from over 4,000 gonococci: these genes have been defined in the PubMLST database (https://pubmlst.org/neisseria) and form part of the N. gonorrhoeae cgMLST version 1.0 scheme 84 . Core genome STs are assigned for each isolate and, through single-linkage clustering and the use of increasing allelic difference thresholds, isolates can be grouped into related core genome groups ( Figure 2). Use of this approach identified discrete clusters of gonococci, some of which persisted over time and associated with AMR genotypes 84 . Thus, analyses composed of higher numbers of genes, and therefore higher genome content, improve resolution and are required to investigate the gonococcal population and detect structure in this highly recombining species.
In the meningococcus, non-overlapping antigenic repertoires are found among different genotypes, resulting in the circulation of distinct meningococcal lineages, the prevalence of . This visualisation also makes associations between core genome and NG STAR STs apparent. Numbers in brackets refer to the number of isolates belonging to that group. Legends depict only groups containing 10 to 20 (or more) isolates. Commonly used laboratory strains FA1090, F62 and MS11 are indicated as well as the resistant strain H041 first identified in Japan.
which fluctuates over time, probably as a consequence of changing natural immunity or vaccination or both 103-105 . However, it is apparent that other modes of selection are exerted on the gonococcal population, and AMR is a major influencing factor 96 . This was shown in the high prevalence of plasmid-mediated AMR in gonococci originating from low-and middle-income countries where, more often than not, syndromic management of gonorrhoea, including treatment with doxycycline, exists 106 . As a result, positive selection of gonococci harbouring conjugative plasmids expressing TetM occurs, resulting in the high prevalence of plasmid-mediated AMR 44 . That study also identified an association between plasmid types and the core genome, providing further evidence for the presence of structure within the gonococcal population.
The availability of WGS has had a major impact in enhancing our understanding of the population biology of the gonococcus and provides hope that it may be possible in the near future to curb transmission through the identification of gonococcal lineages circulating globally. In addition, the identification of loci constituting the core genome, which are common to all gonococci, provides an opportunity to assess antigenic diversity across the gonococcal population, equipping us with novel insights in vaccine development.

Vaccine development
The capacity of the gonococcus to become resistant to chemotherapy provides the alarming prospect that this infection will become untreatable in the near future. This has led to a renewed interest in preventing infection through immunisation and vaccine development. However, despite several decades of research in this field, no vaccine has yet been successfully developed for use in humans 107 .
Reasons for the difficulties encountered in gonococcal vaccine development stem from a combination of characteristics exhibited by the gonococcus that facilitate immune evasion 9 . A major challenge is the fact that the gonococcus causes infections solely in humans and it has been difficult to develop animal models of infection in which immune responses and vaccines can be evaluated 107 . However, improvements made in mouse models of infection, including BALB/C, C57BL/6, and genetically engineered mice, have provided insight into the complexities of pathogenesis and the host response elicited 107 . For example, the lack of protective immunity elicited following natural infection may be a consequence of the ability for gonococci to subvert host immune responses 108 . Indeed, studies of experimentally infected female mice indicate that gonococcal infection of the genital tract results in suppression of adaptive Th1-and Th2-governed responses and induction of Th17-driven innate responses, which the gonococcus is able to resist 107,109,110 .
Immune evasion is also facilitated by the ability of gonococci to exhibit extensive antigenic diversity in surface-exposed antigens, a consequence of mutation, HGT, and the modulation of expression through phase variation 107 . As a result, appreciable variability will be present both among gonococci and within the same gonococcus over time. For example, intra-chromosomal recombination of promoterless copies of pilS that consist of the variable regions of the complete (expressed) pilE gene, which encodes the major pilus subunit (pilin), provides a silent catalogue of sequences that can be reassorted into an expression locus to generate new pilE variants 111 . N. gonorrhoeae is also highly competent for genetic transformation, a process facilitated by a type IV secretion system which is present in over 90% of gonococci and which, in a contact-independent manner, secretes single-stranded DNA into the environment, providing a source of DNA for HGT 112 .
In spite of these challenges, several N. gonorrhoeae antigens, some of which are promising vaccine candidates based on their antigenic conservation, distribution in gonococcal populations, and immunogenicity, have been characterised ( Table 2). Several of these antigens have been identified using a combination of bioinformatic and proteomic analyses, which provide novel means of identifying vaccine antigens that would not have been detected using more conventional methods 113 . Indeed, through the use of proteomics-driven reverse vaccinology, the quantitative proteomic analysis of cell envelopes and naturally occurring vesicles has led to the discovery of several additional vaccine candidates, including BamA, LptD, TamA, NGO2054 and NGO2139 114 , sparking renewed interest in vaccine development (Table 2) 113 . However, it is worth noting that a combination of techniques, methods and tools will be required to evaluate vaccine candidates effectively: (i) proteomics to identify the relative abundance, post-translational modifications and protein-protein interactions of those vaccine candidates; (ii) genomics to assess the genetic diversity, prevalence and distribution in gonococcal populations; and (iii) animal models to examine in vivo host responses and the potential for an immune response to be elicited.
It is likely that vaccines containing a 'cocktail' of antigens will be required in order to generate broad protection against the N. gonorrhoeae population. In this respect, outer membrane vesicle (OMV) vaccines may be suitable since they contain many of the gonococcal surface antigens in their natural conformation 115,116 . Support for the use of OMV as vaccines comes with the epidemiological evidence provided by a retrospective case control study undertaken in New Zealand, which suggested that individuals vaccinated with the OMV vaccine MeNZB were less likely to contract gonorrhoea 117 . The MeNZB vaccine had been specifically designed in response to a serogroup B N. meningitidis epidemic in New Zealand, and although vaccine efficacy towards gonorrhoea was estimated to be 31%, ensuing studies have shown that cross-reactive anti-gonococcal antibodies are induced by MeNZB OMV proteins, providing hope that a gonococcal OMV vaccine is achievable 118 .
A vaccine targeting gonococci is now a possibility with the availability of WGS allowing an in silico appraisal of vaccine candidates to be undertaken prior to more costly and time-consuming vaccine development approaches. In particular, the gene-by-gene characterisation of the gonococcal core genome available on PubMLST allows the diversity of putative vaccine candidates to be catalogued across global gonococcal populations spanning decades 84,119,120 .

Conclusions
Gonorrhoea has been long considered to be an ancient human disease on the basis of clinical descriptions, but recent research has suggested that the emergence of gonorrhoea may date back to as recently as the sixteenth century, calculated through the comparison of genome sequence data and through inference of the time to most recent common ancestor 96 . However, it is conceivable that a much longer association, potentially spanning millennia and consistent with descriptions in ancients texts, has existed 121 .
In any case, the coexistence of humans and gonococci has had, and will continue to have, a profound effect on both populations. In humans, effects include reduced birth rates resulting from infertility caused by ascending gonococcal infections, in addition to increased mortality rates through coinfection with HIV 122 . The coexistence of gonococci with humans has led to the development of multiple mechanisms of genetic and antigenic variation, allowing adaptation, persistence and evasion of both environmental and host challenges. In addition, it is most likely that human migration has widened the gonococcal Genomics plays an increasingly important role in combatting gonococcal disease, allowing the definition and characterisation of complex characters, including the genetic elements associated with AMR, persistence and antigenic diversity. However, at the time of writing, only a very small proportion of gonococci causing infection were analysed by WGS. For example, 56,259 infections were diagnosed in England in 2018 and it would be both expensive and impractical to sequence or culture all of these in one year using current sequencing platforms. Therefore, a large proportion of the N. gonorrhoeae isolates circulating globally will remain uncharacterised at the WGS level for the near future. Our current understanding will be based on a subset of the gonococcal population and at the time of writing this was strongly biased towards isolates with an AMR phenotype. To further our ability to cure or prevent gonorrhoea, larger and more representative datasets need to be assembled and sequenced. Sequencing technology has made major advances in the last decade and what seems impossible now may be achievable in the future. Until then, it is likely that the gonococcus will continue to elude us.