There are three major Neisseria gonorrhoeae β-lactamase plasmid variants which are associated with specific lineages and carry distinct TEM alleles

Neisseria gonorrhoeae is a significant threat to global health with an estimated incidence of over 80 million cases each year and high levels of antimicrobial resistance. The gonococcal β-lactamase plasmid, pbla, carries the TEM β-lactamase, which requires only one or two amino acid changes to become an extended-spectrum β-lactamase (ESBL); this would render last resort treatments for gonorrhoea ineffective. Although pbla is not mobile, it can be transferred by the conjugative plasmid, pConj, found in N. gonorrhoeae . Seven variants of pbla have been described previously, but little is known about their frequency or distribution in the gonococcal population. We characterised sequences of pbla variants and devised a typing scheme, Ng_pblaST that allows their identification from whole genome short-read sequences. We implemented Ng_pblaST to assess the distribution of pbla variants in 15 532 gonococcal isolates. This demonstrated that only three pbla variants commonly circulate in gonococci, which together account for >99 % of sequences. The pbla variants carry different TEM alleles and are prevalent in distinct gonococcal lineages. Analysis of 2758 pbla-containing isolates revealed the co-occurrence of pbla with certain pConj types, indicating co-operativity between pbla and pConj variants in the spread of plasmid-mediated AMR in N. gonorrhoeae . Understanding the variation and distribution of pbla is essential for monitoring and predicting the spread of plasmid-mediated β-lactam resistance in N. gonorrhoeae .


INTRODUCTION
The sexually transmitted infection (STI) gonorrhoea is a major public health concern, with a yearly global incidence of over 80 million cases [1]. Untreated, Neisseria gonorrhoeae can cause serious complications such as disseminated infection, pelvic inflammatory disease, and infertility. Without an available vaccine, the management of gonococcal disease depends on effective antimicrobial treatment. However, strains of N. gonorrhoeae have developed resistance against all first-or second-line antibiotics [2]. Currently, monotherapy with the β-lactam ceftriaxone is the recommended treatment for gonococcal infection in the UK and USA [3,4]. Resistance to β-lactam antibiotics in the gonococcus can be mediated by chromosomal genes such as mosaic penA alleles [5], or plasmid-encoded TEM β-lactamase. Whilst chromosomal resistance to β-lactams is more frequent in high-income countries (HICs), there is a high prevalence of plasmid-mediated resistance in low and middle-income countries (LMICs) [6].

ACCESS
The gonococcal β-lactamase plasmid (pbla) encodes TEM β-lactamases, which confer resistance to penicillin, ampicillin and cephaloridine [7,8]. TEM-1 is the most frequently found β-lactamase in gonococci [6] and requires only two amino acid substitutions to become an extended spectrum β-lactamase (ESBL) [9]; one of these substitutions is already present in TEM-135 [10]. The emergence and spread of a transmissible ESBL is a significant public health threat, as it would render current the recommended treatment ineffective and undermine our ability to cure and control gonococcal infection. pbla is a 3.2-9.3 kb plasmid, which carries its own origin of transfer (oriT), MobA relaxase and MobC [11], allowing its transfer by the gonococcal conjugative plasmid, pConj [12]. pbla carries three distinct origins of replication, ori1, ori2 and ori3, and two replication initiation proteins RepA and RepB, which facilitate replication from ori1 and ori2/ori3, respectively [13], allowing its replication in N. gonorrhoeae, Escherichia coli, Salmonella enterica serotype Minnesota, and Haemophilus influenzae. Seven variants of pbla have been described [14][15][16] and are thought to have emerged from the prototypical 7.4 kb plasmid usually referred to as the Asia variant (pbla.As). The variants differ in the presence of replication and mobilisation genes [14] and could vary in their capacity to be transferred. Additionally, it has been reported that some variants (e.g. pbla.Rio) are primarily associated with TEM-135 [16,17].
Previously, pbla variants have been identified by their molecular weight, restriction digestion pattern [18,19] or PCR [20,21] which require DNA from isolates, hindering large-scale analysis. Therefore, the spread of pbla variants in the gonococcal population, and their association with TEM alleles and pConj variants are unknown. Furthermore, specific PCR assays developed to distinguish between the three earliest variants, do not identify later variants [16], and have led to discrepancies between sequencing data and PCR typing results even for WHO reference strains [22]. In this work, we established and implemented a typing scheme for pbla based on the characteristic gene presence/absence pattern of variants and propose a numeric nomenclature for pbla variants, avoiding the use of geographic locations as recommended by the WHO [23]. The typing scheme, Ng_pblaST, is publicly available on the PubMLST [24] Neisseria database (https://pubmlst.org/neisseria/) and can be used to identify the presence of plasmid variants in gonococcal isolates. We analysed the distribution of pbla in 15 532 isolates and found that only three of the seven previously described pbla variants circulate in gonococci at any appreciable frequency. Furthermore, we show that pbla variants are associated with TEM alleles, pConj types, and particular gonococcal lineages.

Whole genome sequence data
Gonococcal whole genome sequence (WGS) data in the PubMLST database (n=15 664, accessed 28 July 2022) were analysed on the BIGSdb platform [24], discarding WGS from isolates with ambiguous ribosomal multilocus sequence type (rMLST, e.g. incorrect assignment of species) [25] and removing duplicated entries. This resulted in a dataset of 15 532 isolates from 66 countries, including all six WHO regions, and spanning the years 1928 to 2022 (Table S1 and 2, Fig. S1).

Impact Statement
Plasmid-mediated AMR contributed to the discontinuation of penicillin and tetracycline for treating gonococcal infection. This article is, to our knowledge, the most extensive analysis of the occurrence and distribution of variants of the β-lactamase plasmid, pbla, in the WHO priority pathogen N. gonorrhoeae, and their carriage of TEM alleles. The TEM β-lactamase is only a few changes away from becoming an ESBL, which would render last resort treatments for gonorrhoea ineffective. Previously, identification of pbla variants relied on PCR or restriction digestion patterns which hinders the analysis of large numbers of isolates, while pbla fails to assemble from short-read, whole genome sequences due to repeat sequences on the plasmid. We report the development of the pbla-typing scheme, Ng_pblaST, to identify variants from short-read sequence data. Ng_pblaST is now available on the PubMLST Neisseria database (https://pubmlst.org/neisseria/) and can be used to unambiguously identify the plasmid and its variants in gonococcal isolates. realignment threshold: 0.3). Hydrophobicity plots were generated with ProtScale [35] using the scale Hphob./Kyte and Doolittle. The conserved neighbourhood of NEIS2964 was assessed using WebFlaGs [36] with BLASTp against NCBI RefSeq database (updated 29 January 2021, e value 1e-10) and three jackhmmr iterations. Initial hits of conserved neighbours (WP_140450222.1, WP_010904468.1 and WP_012881329.1) were investigated as above to assess their co-occurrence with NEIS2964.
pbla-carrying gonococci (n=2758, accessed 28 July 2022) were identified by the presence of NEIS2960, which encodes a hypothetical protein and is found in all plasmid variants [6]. Gene absence was confirmed by blasting available alleles against plasmid sequences using the sequence tag scan function on PubMLST with default parameters (minimum identity: 60 %, minimum alignment length: 50 % and BLASTn word size: 20). Plasmids were typed according to the characteristic gene presence/absence pattern of variants with incompletely assembled loci considered as present.
To check the accuracy of the typing scheme, we used pbla short-read sequences of pbla-containing WHO reference strains (WHO M, WHO N, WHO O), previously characterised by long-read sequencing [37]. We confirmed the pbla types in these strains by PCR using previously described primers [20] (0.25 µM) and Herculase II Fusion DNA Polymerase (Agilent) with the following cycling conditions: 92 °C for 2 min, followed by 30 cycles of 92 °C for 20 s, primer annealing at 50 °C for 20 s, elongation for 3 min at 68 °C, and a final extension at 68 °C for 8 min. Additionally, 16 gonococcal isolates, from nine Ng_cgc_400 lineages, eight countries and isolated from 1986 to 2015, containing different pbla variants were randomly selected from the gonococcal dataset (Tables S2 and S3) and typed using the scheme. pbla-containing contigs were then extracted from the sequencing bin and aligned to pJD4 (NC_002098.1) to define deleted regions and confirm the completeness of the remaining sequences. Sequencing quality was assessed by comparing sequencing parameters, such as number of contigs, N50, L50, percentage of alleles designated, percentage of loci tagged of isolates, which are accessible on BIGSdb.
For further analyses of the distribution of TEM alleles amongst pbla variants, typable plasmids with incomplete loci were removed from the analysis, resulting in a dataset of 2157 pbla sequences.

Analysis of the distribution of pbla variants
N. gonorrhoeae population structure was resolved using core genome MLST (cg-MLST) [38]. This scheme groups gonococcal isolates according to allelic designations in 1668 loci core to the gonococcus. Each different allele at every locus is given a unique integer and isolates are grouped into single-linkage clusters according to the number of allelic differences. A cut-off of 400 allelic differences between individual isolates resolves the population into discrete N. gonorrhoeae core genome clusters (Ng_cgc_400) which are stable over time [38]. A minimal spanning tree was generated with GrapeTree [39], clustering isolates according to core genome allelic differences. NEIS2220 (trbM) was used to detect the presence of pConj in individual isolates [6]. pConj types were defined based on the pConj typing scheme with a cut-off of five allelic differences (Ng_cp_5) [6]. Where the Ng_cp_5 could not be assigned, plasmids were clustered based on pConj core genome loci [6] using GrapeTree [39], and pConj types were identified manually.

Statistical analysis
Data manipulation and analysis were performed in R version 4.1.1 using base R and the tidyverse package [40]. To calculate odds ratios (OR), the odds. ratio function from the epitools package [41] was used. Plots were generated with ggplot2 [42]. A P value of <0.01 was considered statistically significant.

Characterisation of gonococcal pbla
Seven variants of pbla have been described in the gonococcus [14,16,43], ranging in size from 3.2 to 9.3 kb due to duplication and deletion events. The prototypical 7.4 kb plasmid, often referred to as the Asia variant [14], contains nine loci that can be functionally grouped into Tn2-derived, replication, and mobilisation regions (Fig. 1a). The plasmid contains 34 % of Tn2 [44], including bla TEM , 84 % of the tnpR resolvase gene, and the right inverted repeat [45,46].
The replication region of pbla contains NEIS2358 (repA) and NEIS2360 (repB) encoding replication initiation proteins (Rep) that are required for initiation of replication at ori1 and ori2/ori3, respectively [13]. The three origins belong to the iteron family [13], with approximately half of the sequence consisting of an array of ~20 bp repeats. These serve as binding sites for the Rep proteins [47]. pbla has been classified as an IncW plasmid with a silent IncFII replicon which is used when the ori2/ori3 region is absent [13]. However, RepA shares less than 17 % amino acid identity with RepA from the characterised IncFII plasmid R1 [28], and RepB has only 17.9 % amino acid identity with the IncW plasmid R388 RepA [29]. MOB-typer [31,32], which probes a collection of plasmid-derived replication protein sequences against the query sequence, instead classifies pbla as an IncP plasmid.
The mobilisation region of pbla contains an origin of transfer (oriT), flanked by genes encoding a relaxase (MobA, NEIS2961) and MobC (NEIS2962) [11]. Mob typing characterises plasmids according to their N-terminal relaxase domain [48], and identified pbla as a MobQ family plasmid. MobQ plasmids have been reported to be efficiently mobilised by helper plasmids belonging to a variety of Inc groups, leading to their spread across different phyla [48]. Therefore, we investigated the host range of pbla and found it is largely limited to N. gonorrhoeae, with only two clinical Neisseria meningitidis isolates possessing the plasmid [49]. Interestingly, we identified pbla sequences in the chromosome of seven of the 30 available Haemophilus ducreyi sequences in the NCBI nucleotide database. H. ducreyi strains DMC64 (Genbank accession: CP011230.1) and 33 921 (Genbank accession: CP011228.1) [50] are annotated with intact pbla in their chromosome adjacent to phage-related genes (Fig. 1b). Unlike in pbla, Tn2 is intact in the H. ducreyi sequences and includes the transposase gene which is absent from gonococcal pbla. Furthermore, pbla GC-content (38.3 %) is similar to H. ducreyi chromosomal DNA (37.9-38.6 %) [50], whilst the N. gonorrhoeae chromosome GC-content is 52.5 % [51].
NEIS2964 is present in two copies on the 7.4 kb pbla and is predicted to encode an 86 amino acid protein of unknown function. Hydrophobicity plots of NEIS2964 indicate it has two domains; an N-terminal transmembrane segment and a charged C-terminal domain (Fig. S2). Neighbourhood analysis revealed a conserved localisation of NEIS2964 with mob and rep (Fig. S3). However, whilst repB is frequently found without NEIS2964-related genes, mobA and mobC show a consistent association with NEIS2964 (Fig. S3). This suggests a role for NEIS2964 in plasmid mobilisation.

pbla typing reveals that three major variants circulate in gonococci
Understanding the variation and prevalence of pbla is essential for monitoring and predicting the spread of plasmid-mediated β-lactam resistance in the gonococcus. However, repeat sequences on the plasmid hinder pbla assembly from short-read sequencing data. Therefore, we developed a typing scheme (Ng_pblaST; available on https://pubmlst.org/neisseria/) based on the characteristic gene presence/absence pattern of pbla variants, which does not require plasmid assembly (Table 1). We propose a new pbla nomenclature which avoids the use of geographic regions, as this can convey misleading information about origin and could be stigmatising [52]. This implements the recent WHO recommendations on renaming infectious diseases [23] and number the variants in the order of when they were first reported (Table 1). Therefore, pbla.1 corresponds to pbla.Af, pbla.2 to pbla.As, pbla.3 to pbla.Rio, pbla.4 to pbla.Jo and pbla.5 to pbla.Au.
We first analysed published sequences of pbla variants and identified their differences based on the presence/absence of genes in the 7.4 kb pbla.2 (previously described as the Asia variant) [26]. Two insertion variants and four deletion variants of pbla.2 have been described. The insertion variants [18,19], previously referred to as New Zealand and Nîmes variants, have a duplication or IS5 insertion downstream of blaTEM, respectively. Both have only been reported on a single occasion [18,19]. We analysed available gonococcal and pbla sequences (n=15 532) and the pbla sequences in 2758 of these isolates (Table S1) for the presence of the duplicated sequence and IS5. We obtained a single match for the duplicated sequence in pbla-containing gonococcal isolates. However, closer investigation revealed that this isolate (GHA-TMH-537) shows an uncharacteristic duplication of NEIS2960, NEIS2357, NEIS2359 and NEIS2360 and a deletion of NEIS2961 and NEIS2962. Additionally, IS5 is not present in Neisseria spp. or in Haemophilus spp., but is found on the chromosome of the E. coli strain from which the insertion variant was isolated following transformation with gonococcal plasmid DNA [19]. Therefore, pbla-insertion variants are rare and may be experimental artefacts [19]. pbla.1 (5.6 kb, previously the African variant) [37] and pbla.4 (5.4 kb, previously the Johannesburg variant) [43] harbour characteristic deletions in the replication region. pbla.1 is structurally identical to pbla.2 except for the loss of a 1 826 bp fragment (Fig. 2, Table 1), which spans NEIS2960 (repB), NEIS2964, ori2 and ori3. pbla.4 has a 2 560 bp deletion that includes ori2 and ori3. In addition to genes absent from pbla.1, pbla.4 lacks NEIS2962 (mobC) and the second copy of NEIS2964 (Fig. 2, Table 1).  (Fig. 2, Table 1). pbla.5 is the smallest pbla variant reported to date and has lost a further 1 885 bp comprising ori1 and NEIS2358 (repA), as well as NEIS2962 and NEIS2964.
To check the accuracy of Ng_pblaST, we examined available whole genome, short-read sequences of WHO reference strains which are known to harbour pbla and have been characterised by long-read sequencing [37]. Ng_pblaST correctly identified the pbla variant in every instance from the short-read sequences alone. Additionally, we analysed short-read sequences from 16 gonococcal isolates containing pbla (Table S2) by extracting plasmid-loci containing contigs from the sequencing bin and mapping them against the pbla.2 reference sequence; this confirmed the exact variant-specific deletions in all tested variants (not shown).
With Ng_pblaST, we analysed the 2758 pbla-carrying isolates in our dataset which originate from 50 countries representing all six WHO regions and were recovered between 1979 and 2022 (Table S1). Although it is possible to distinguish between pbla variants based on only four genes (NEIS2961, NEIS2962, NEIS2964 and NEIS2360), all plasmid loci were included in our scheme to avoid incorrect allocation of variants due to low sequencing quality. The pbla variant was assigned in 74.6 % of isolates. The remaining sequences showed uncharacteristic gene presence/absence patterns due to poor sequencing quality. The sequences of isolates with uncharacterised pbla had a significantly lower N50 than isolates with typed pbla (average N50 untyped=59 578.6, average N50 typed=78 270.6, two sample t-test P<0.001). When assembling genomes from the longest to the shortest contig, the N50 is the length of the shortest contig when half the total genome has been assembled [54]. Of the typed isolates (2 054), only nine and three isolates carry pbla.4 and pbla.5, respectively (  [33].

pbla variants carry distinct TEM alleles
All pbla variants carry blaTEM, conferring resistance to penicillins. There are 59 blaTEM alleles in the 2157 pbla with complete blaTEM sequences in the PubMLST database. TEM-1 (encoded by allele 3) is the most prevalent, accounting for 59.6 % (1286/2157) of sequences, followed by TEM-135 (allele 2; 24.3 %, 525/2157) and allele 6 (11.1 %, 240/2157). TEM-135 has an M182T substitution, which increases the stability of the enzyme and often precedes further mutations extending the range of substrates [55,56]. Individual isolates carry alleles with substitutions in addition to M182T at positions P14, R61, L201, A224 and A248. These substitutions have not been characterised or are not known to confer increased resistance to β-lactams. Allele six has a P14S substitution in its leader sequence and is found in a subpopulation of pbla.1.

pbla variants are associated with specific gonococcal lineages
We next analysed the distribution of pbla variants amongst 15 532 N. gonorrhoeae isolates originating from 66 countries and isolated from 1928 to 2022 (Table S1, Fig. S1). pbla is present in 17.8 % of isolates (2 758/15 532), with significant differences in prevalence between geographic regions (chi 2 -test, P<0.001, Table 2). Significantly high rates of pbla-carriage were found in isolates from Africa (58.4 %, 185/317; OR=6.9, P<0.001). To define the distribution of pbla amongst gonococcal lineages, we grouped isolates into N. gonorrhoeae core genome clusters (Ng_cgc_400) [38]. Strikingly, isolates from Ng_cgc_400 21 and 34, which are prevalent in Africa but also found in Asia, Europe, and North America, show a high pbla prevalence (>50 %) irrespective of their origin, indicating pbla carriage is a feature of a lineage, rather than the region from where a strain had been isolated.

DISCUSSION
The potential of the pbla-encoded TEM β-lactamase to become an ESBL highlights the importance of monitoring the spread of pbla variants in the gonococcal population. However, the presence of repeat regions on the plasmid [14] hinders its assembly from short-read sequences. Additionally, the plasmid sequence is rarely recovered from long-read sequencing data [37], likely due to its small size and low copy number [57]. Therefore, plasmid variants have previously been characterised by their restriction digestion pattern [14] or PCR [20,21], which only allow small-scale analyses. In this study, we analysed published sequences of pbla to design the typing scheme Ng_pblaST to identify plasmid variants from short read WGS according to their characteristic gene presence/absence (Fig. 6). The typing scheme allows plasmid classification from short-read sequencing data and enables unambiguous analysis of large datasets. Current nomenclature of pbla variants is based on the geographical region from where they were first identified. However, this scheme is not necessarily informative of their site of origin, and uses geographic names which is no longer recommended [58]. Therefore, we propose a numeric nomenclature for pbla, numbering them in the order in which they were first described. We have implemented Ng_pblaST on the PubMLST database and have included both current and proposed nomenclatures to avoid confusion and allow comparison.  We found pbla-related sequences that are annotated to be on the chromosome of seven H. ducreyi strains. This contrasts with the previous hypothesis that pbla was originally a small cryptic plasmid in from Haemophilus spp. which gained TEM the β-lactamase through insertion of Tn2 [59]. Of note, Tn2 is intact in the pbla-related sequences in H. ducreyi, indicating that these sequences pre-dated N. gonorrhoeae pbla in which Tn2 is truncated. N. gonorrhoeae is naturally competent [60] so uptake of linear H. ducreyi chromosomal DNA followed by recombination events at repeat sequences could have led to pbla appearing in N. gonorrhoeae; alternatively, pbla was already present as a plasmid in H. ducreyi as proposed previously [59].
We analysed the occurrence of pbla variants across 15 532 isolates and found that, of the seven described variants, only pbla.1, pbla.2 and pbla.3 are present in appreciable numbers in the gonococcal population. The other plasmid variants were found in 12 isolates only and may either be unstable [18] or are experimental artefacts [19]. Our dataset is the largest quality-checked collection of publicly available isolates, and includes strains from over 60 countries, spanning almost 100 years. However, LMICs, especially those in Africa [22], are under-represented in our and all current datasets. Nevertheless, our dataset does include 317 isolates Table 3. Odds ratios of pbla variants in distinct gonococcal core genome clusters. The odds ratios were calculated as the odds of pbla-carrying isolates in the core genome clusters (cgc) to carry the indicated variant divided by the odds of any pbla-carrying isolate to harbour the specific variant, and the P-value is given in brackets. from ten African countries, expanded by the available WGS data of 49 novel Kenyan isolates. Consistent with previous findings [6], pbla prevalence was highest in LMICs. For example, 58.3 % of isolates from countries in Africa harbour the β-lactamase plasmid, compared with only 7.9 % of isolates from the USA and 20 % of isolates from Europe. This was hypothesised to be a result of the extensive spread and maintenance of resistance plasmids in isolates in low-income countries [6]. However, we found that prevalent Ng_cgc_400 s in Africa, such as Ng_cgc_400 21, also have high plasmid carriage when isolated in countries in other geographical regions such as Europe (77.2 %). Therefore, plasmid carriage is an Ng_cgc_400-specific trait and independent of the country from where a strain was isolated. This indicates that pbla forms long-term, stable relationships with certain lineages, which may differ in their ability either to acquire and/or maintain the plasmid. Several other factors might contribute to this phenomenon. Compensatory mutations could occur in certain lineages that reduce the metabolic burden imposed by the plasmid [61,62], while sporadic selection pressure imposed by penicillin, a widely used antibiotic, may select for pbla in certain lineages [63]. In contrast, low transfer rates could lead to the absence or loss of the plasmid from other lineages [64]. The preponderance of certain pbla variants in geographic regions could result from fitness costs or a selective advantage in specific environmental conditions, leading to the establishment of variants in different areas.
Our analysis also reveals that pbla variants co-occur with specific pConj types. Of note, pbla is rarely found with pConj variants 2, 4, 6 and 7. These pConj variants are present at lower frequencies than the pConj variants that co-occur with pbla, and pConj.2, 4 and 7 are Ng_cgc_400-specific [65]. Therefore, the co-occurrence between pbla and pConj variants could merely reflect the wider distribution of pConj variants 1, 3, and 5. Alternatively, specific interactions between pbla and pConj variants could explain their co-occurrence. For example, pbla MobA has a relaxase domain which nicks the oriT sequence on pbla [11], but then must interact with the pConj-encoded Type 4 Secretion System for its transfer [66]. Therefore, differences in the mpf region of pConj variants [6] may lead to distinct interactions with pbla and favour their co-transfer.
pbla.3 is associated with TEM-135 and confined to three Ng_cgc_400 s that cluster together on a minimal-spanning tree (Fig. 4). The deletion of the mob region indicates this variant might not be mobilised by pConj, with the observed distribution resulting from clonal expansion of the Ng_cgc_400. pbla.3 transfer has been reported [67], and co-integration into pConj, mediated by IS1, was described as a transfer mechanism of pbla.3 in E. coli [53,68]. However, IS1 is not found in the gonococcal genome, and further experiments are required to investigate possible mechanisms for pbla.3 transfer. Nevertheless, pbla.3 is found in gonococcal isolates at comparable numbers to pbla.2. This might reflect the success of Ng_cgc_400 25, the Ng_cgc_400 with the third most pbla-containing isolates in HIC after Ng_cgc_400 21 and 33.
TEM-135 is the second most prevalent allele on pbla (in 24.3 % of plasmids) and relative to TEM-1, carries an M182T substitution, which stabilises the enzyme [69] and facilitates further amino acid changes in the active site that would convert the enzyme into an ESBL. We found TEM-135 at almost twice the frequency described before in gonococcal pbla [6]. Previous reports indicate an association of pbla.3 with TEM-135 [17], while TEM-135 carrying pbla.2 has been isolated on a few occasions [57,70]. Our results confirm these observations but show that almost half of pbla.2 carry TEM-135. In contrast, pbla.1 encodes mainly TEM-1, with TEM-1 P14S found in a subpopulation of pbla.1 associated with Ng_cgc_400 33, a prevalent Ng_cgc_400 in HICs. Nothing is known about the effect of this substitution on TEM activity, even though it is common among isolates with pbla.1.
As pbla long-read sequence data is sparse, we manually validated the typing scheme with pbla-containing WHO reference strains, as well as aligning pbla contigs to the reference plasmid. This confirmed the typing scheme in every case, albeit with a limited number of isolates. Ng_pblaST is now available on PubMLST, allowing further independent validation by the research community.  6. Several variants of the gonococcal β-lactamase plasmid have been described, but little is known about their frequency or distribution in the gonococcal population. We have established and implemented the pbla typing scheme Ng_pblaST allowing identification of variants from short read sequencing data. Analysis of 17 827 gonococcal isolates revealed three major variants circulating in gonococcal populations which are associated with distinct TEM alleles.
The typing scheme identifies the five reported deletion variants but should also classify novel variants as untypeable; these should be investigated by PCR or long read sequencing to monitor for the emergence of novel variants which are not included in the current scheme.
In summary, we devised and implemented the typing scheme Ng_pblaST for gonococcal pbla and revealed associations of plasmid variants with TEM alleles, pConj variants and Ng_cgc_400 s. Ng_pblaST is available on PubMLST and should aid in monitoring of the spread of the plasmid and TEM alleles, resolve ambiguities [22], and inform research on the adaption of AMR plasmids to their gonococcal host. The high prevalence of plasmid-mediated resistance in LMICs and the potential of pbla-encoded TEM to become an ESBL highlights the importance of understanding the variation and distribution of pbla. Further research is needed to investigate the interactions of pbla with the host genome and co-resident plasmids to understand the underlying factors of its distribution and spread, and to develop interventions to interrupt the emergence of plasmid-mediated β-lactam resistance in this priority pathogen.