Genomic analysis of the meningococcal ST-4821 complex–Western clade, potential sexual transmission and predicted antibiotic susceptibility and vaccine coverage

Introduction The ST-4821 complex (cc4821) is a leading cause of serogroup C and serogroup B invasive meningococcal disease in China where diverse strains in two phylogenetic groups (groups 1 and 2) have acquired fluoroquinolone resistance. cc4821 was recently prevalent among carriage isolates in men who have sex with men in New York City (USA). Genome-level population studies have thus far been limited to Chinese isolates. The aim of the present study was to build upon these with an extended panel of international cc4821 isolates. Methods Genomes of isolates from Asia (1972 to 2017), Europe (2011 to 2018), North America (2007), and South America (2014) were sequenced or obtained from the PubMLST Neisseria database. Core genome comparisons were performed in PubMLST. Results Four lineages were identified. Western isolates formed a distinct, mainly serogroup B sublineage with alleles associated with fluoroquinolone susceptibility (MIC <0.03 mg/L) and reduced penicillin susceptibility (MIC 0.094 to 1 mg/L). A third of these were from anogenital sites in men who have sex with men and had unique denitrification gene alleles. Generally 4CMenB vaccine strain coverage was reliant on strain-specific NHBA peptides. Discussion The previously identified cc4821 group 2 was resolved into three separate lineages. Clustering of western isolates was surprising given the overall diversity of cc4821. Possible association of this cluster with the anogenital niche is worthy of monitoring given concerns surrounding antibiotic resistance and potential subcapsular vaccine escape.


Introduction
The ST-4821 complex (cc4821) is a leading cause of serogroup C and serogroup B invasive meningococcal disease in China where diverse strains in two phylogenetic groups (groups 1 and 2) have acquired fluoroquinolone resistance. cc4821 was recently prevalent among carriage isolates in men who have sex with men in New York City (USA). Genome-level population studies have thus far been limited to Chinese isolates. The aim of the present study was to build upon these with an extended panel of international cc4821 isolates. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Introduction factor H-binding protein (fHbp) variant 1 peptide, and/or a Neisseria adhesin A (NadA) variant NadA-1 or NadA-2/3 peptide, and/or a cross-reactive neisserial heparin-binding antigen (NHBA) peptide. Coverage by MenB-fHbp is dependent on sufficient expression of fHbp. Potential coverage of individual isolates is determined using the ELISA-based Meningococcal Antigen Typing System (MATS; 4CMenB) [10] or the flow-cytometry based Meningococcal Antigen Surface Expression (MeASurE; MenB-fHbp) [11] assays. A genetic version of MATS (gMATS) has also been developed based on a large international MATS dataset [12]. The emergence of a serogroup W cc4821 strain in both carriage and disease is also a cause for concern [13], although highly effective serogroup W conjugate vaccines are available.
Although invasive disease due to cc4821 is rare outside of Asia, Guo et al. (2018) previously noted the expansion of non-Chinese cc4821 isolates within the PubMLST Neisseria database [6]. In a recent meningococcal carriage study in 706 men who have sex with men (MSM) in New York City (USA), cc4821 was the most prevalent clonal complex identified accounting for 14.3% (24/168) isolates [14]. Members of another strain, ST-11 complex lineage 11.2, have caused multiple outbreaks of IMD in MSM in Europe and the USA and a large multistate outbreak of urethritis among predominantly heterosexual men in the USA [15]. The latter strain (US_NmUC) has recently been detected in proctitis in MSM in the United Kingdom [16]. The strains responsible have shown several adaptations to the anogenital niche including the acquisition of active/efficient denitrification genes (aniA and norB) to enable growth in an anaerobic environment [17,18]. A recent study in gonococci found an association between loss of function (LOF) mutation in the efflux pump component, MtrC, and cervical infection. This extended to US_NmUC in which 8.7% of isolates possessed LOF mutations versus 0.62% in a large invasive disease collection [19].
In the present study, we expanded upon earlier work by sequencing additional Chinese genomes and performing high resolution phylogenetic analyses on these and an extended panel of all cc4821 genomes from multiple countries on several continents that are currently available on the PubMLST Neisseria database (accessed 12 th September 2019). We also charted the distribution of ciprofloxacin resistance-associated gyrA mutations and penA mutations associated with reduced susceptibility to penicillin; the nitrite reductase gene aniA; and potential cc4821 strain coverage by subcapsular (MenB) vaccines. Predicted functionality of MtrC was assessed in a set of anogenital isolates.

Genome sequencing and database searches
New genome sequences were obtained for 138 Chinese cc4821 isolates collected between 1978 to 2017, inclusive (S1 Table). DNA was extracted from overnight cultures using the Wizard genomic DNA purification kit (Promega, WI, USA) according to the manufacturer's instructions. Paired end libraries were generated with average insert lengths of 350 bp or 500 bp. Sequencing was performed on Illumina HiSeq 2000, 2500 or 4000 platforms (Illumina, CA, USA).Genome assembly was performed using SOAPdenovo (release 1.04; http://soap. genomics.org.cn/soapdenovo.html).
Associated metadata and genotypic data were exported using the 'export dataset' function. Stated capsular groups are from the 'capsule_group' field which is determined automatically from the contents of the serogroup and/or genogroup fields in PubMLST.
Putative recombinations were identified by comparing genomes in term of all NEIS loci and identifying blocks of differing genes between closely related isolates.

BLAST searches and nucleotide/amino acid sequence alignments
Allelic sequence data were obtained from the PubMLST Neisseria database using the 'sequence attribute search' tool. BLAST searches were performed using the PubMLST BLAST tool. Nucleotide and amino acid sequence alignments were performed using BioEdit (version 7.0.9.0) [23].

Classification of antibiotic susceptibility
Minimum inhibitory concentration (MIC) breakpoints for antibiotic sensitivities differ according to publication and by the organisations that define them, including occasional revisions over time. Current EuCAST (v10.0) breakpoints for intravenous administration of penicillin are susceptible �0.06 mg/L and resistant >0.25 mg/L. Studies in animal models, however, suggest an upper breakpoint of >1 mg/L for resistance [24]. Current EuCAST (v10.0) breakpoints for ciprofloxacin for use in the prophylaxis of meningococcal disease are susceptible �0.03 mg/L and resistant >0.03 mg/L.
For the present study, isolates possessing up to five mutations (F504L, A510V, I515V, H541N and I566V) within the deduced amino acid sequence for a 402 bp fragment (penA) of the penicillin binding protein 2 gene (PubMLST locus NEIS1753) were predicted to have reduced susceptibility to penicillin (0.094 to 1 mg/L) in accordance with Taha et al. (2007) [25]. Isolates with alterations to residues T91 or D95 of the deduced amino acid sequence of the DNA gyrase gene (gyrA; PubMLST locus NEIS1320) were predicted to be resistant (MIC, �0.03 mg/L) to ciprofloxacin in accordance with Hong et al., 2013 [7]. Alterations to residues D86, S87, S88 or E91 of the deduced amino acid sequence of the DNA topoisomerase IV subunit A gene (parC; PubmLST locus NEIS1525) were also noted having previously been associated with incrementally higher ciprofloxacin MICs when observed in tandem with gyrA mutations in N. gonorrhoeae [26].

Antibiotic resistance determinant genotyping
The different lineages/sublineages contained varying distributions of isolates with fluoroquinolone resistance-associated (MICs � 0.03 mg/L) gyrA and parC alleles and penA alleles associated with reduced penicillin susceptibility (0.094 to 1 mg/L) (Fig 3).

Subcapsular vaccine antigen genes
None of the cc4821/ST-14840 isolates possessed alleles for PorA P1.4 or NadA. Two isolates had incomplete nhba sequences and one, a UK invasive rectal swab isolate from 2017, had a frameshifted nhba allele. The remaining 185 isolates possessed alleles for one of 32 intact NHBA peptides, 22 of which were unique to China, and 27 of which were only, or predominantly, observed in cc4821 on the PubMLST Neisseria database (accessed 24/06/20). 4CMenB strain coverage of all but one of the encoded NHBA peptides was unpredictable according to gMATS. The remaining peptide, peptide 10 was predicted by gMATS to be covered and occurred in a single lineage 2a MenB carrier isolate from 2013 (this isolate possessed an allele for a non-covered variant 2 fHbp peptide). The majority of isolates 144/188 possessed fHbp variant 2 or 3 alleles and a further two isolates, a Chinese invasive MenC lineage 1 (non-epidemic clone) isolate from 2011 and a Chinese lineage 2c MenC carrier isolate from 2017, possessed a truncated allele (nonsense mutation) and a complete deletion of fhbp, respectively. The remaining 42 isolates possessed alleles for one of 13 fHbp variant 1 peptides, eight of which were only, or predominantly, found in cc4821, and six of which were unique to China on the PubMLST Neisseria database (accessed 24/06/20). All 13 peptides afforded unpredictable 4CMenB strain coverage according to gMATS. Given the general absence of gMATS-predictable strain coverage, potential strain coverage was estimated based on the most conservative criteria that any NHBA peptide and any fHbp variant 1 peptide is potentially covered. Potential antigen-specific strain coverage of cc4821 lineages/sublineages by 4CMenB is shown in Fig 4. A detailed breakdown of lineage-specific potential 4CMenb strain coverage is as follows:  Lineage 2a. All 49 lineage 2a isolates were potentially covered by NHBA though this was heavily reliant on peptides 910 and 688 being covered since these accounted for a half and a quarter of the isolates, respectively. Among the invasive isolates, 2/5 were also potentially covered by virtue of one of two different fHbp peptides.
Lineage 2b. All of the lineage 2b isolates were potentially covered by NHBA, though this was reliant on peptide 697 being covered. Additionally, 1/13 carrier isolates (MenW, 2014) and the two invasive MenC isolates were potentially covered by virtue of fHbp.
Lineage 2c. All of the 40 Chinese lineage 2c isolates were potentially covered by NHBA though this was mainly reliant on coverage of peptides 669 (n = 22) and 697 (n = 9). In addition, 4/25 carrier isolates and 2/10 invasive isolates (MenB, 2011 and MenC, 2005) were potentially covered by fHbp. An fhbp gene was lacking in one MenC carrier isolate (2017) and was replaced by N. lactamica sequence as previously reported among proposed ST-286 complex isolates [30]. Among the lineage 2c RoW isolates, 16/17 were potentially covered by virtue of NHBA, in particular peptide 669 (n = 15). One MenB isolate from a rectal swab from an invasive disease patient possessed a frameshifted nhba allele. None of the RoW isolates were covered by fHbp.
Diffuse isolates. Potential 4CMenB coverage of the 15 diffuse isolates was limited to NHBA, in particular peptide 669 (n = 10).

Distribution of the nitrite reductase (aniA) gene previously associated with adaptation to the anogenital niche
The acquisition of a working nitrite reductase (aniA) gene has previously been implicated in the adaptation of meningococci to the anogenital niche [17,18]. The distribution of aniA alleles within the cc4821 population structure is shown in S6 Fig. A number  A BLAST search of all genomes (>2Mb) on the PubMLST Neisseria database (n = 30,044; accessed 1 st September 2020) for the putatively acquired 2022 bp sequence identified n = 103 exact, full-length matches, all of which were meningococcal. Thirty eight of these belonged to, or were closely related to, cc4821, including n = 29 cc4821 lineage 2c isolates from the current study with aniA allele 8 (n = 28) or aniA allele 733 (n = 1); five cc4821 carriage study isolates from the UK and Sweden that were not included in the current study (all aniA allele 8); and four clonal complex-unassigned isolates with <4 MLST alleles in common with ST-4821 (all aniA allele 8). The latter four isolates each had 4 to 5 loci in common with ST-3200. The cc8 isolate descended of cc4821 (aniA allele 8) was also identified. The remaining BLAST hits were for MenA ST-4 complex (n = 14;1915to 1972 and MenA ST-5 complex (n = 48; 1966 to 2014) isolates and two isolates with incomplete MLST profiles but matching cc5 at six loci.

Discussion
High-resolution phylogenetic analyses have revolutionised our understanding of meningococcal population structures and our ability to identify and track existing and emerging strains and outbreaks on a global level [31,32]. Such analyses benefit from the inclusion of geo-temporally diverse isolates to provide the broadest possible view of the corresponding diversity and to put the constituent strains in context. To this end, our study included all currently available cc4821 genomes on the PubMLST database (accessed 12/09/19).
A key finding of our study is the existence of four well defined lineages, versus the two that were previously identified [5,6]. Lineage 1 corresponds to the previously identified group I/1 which includes the predominantly MenC epidemic clone with PorA P1.7-2,14 [6]. Lineages 2a, 2b, and 2c, and a number of diffuse isolates/singletons, collectively correspond to the previously designated group II/2. It was previously noted that group 2 was relatively diverse but the inclusion of more than five times as many genomes in the present study has helped to resolve the constituent lineages. The star-like arrangement of the four lineages along with the non-cc4821 outgroup on the phylogeny suggests that the four lineages have diverged concurrently from an early common ancestor, however, the relative closeness of lineages 2a, 2b and 2c to one-another and the benefits of maintaining continuity with previous publications justifies the chosen lineage designations (2a, 2b and 2c).
Interestingly, with the exception of a lineage 1 epidemic clone MenC case isolate from Japan in 2017, all of the remaining 17 non-Chinese isolates, from nine countries on three continents (2007 to 2018), exclusively belonged to a distinct cluster (the RoW cluster) within lineage 2c. Furthermore, whilst PorA subtype P1.17-6,23 and closely-related subtypes were only associated with the RoW cluster among our genome panel, they also accounted for 14/27 PorA subtypes among non-Chinese, non-genomic isolate submissions on the PubMLST Neisseria database (accessed 13 th July 2020; S2 Table). The lineage-specific distribution of several predominant Sequence Types suggests that the previously reported 'ST-3200 group' of invasive MenB isolates in Taiwan from 1996 to 2002 [33] would also form part of the broader lineage 2c. The global expansion of this particular sublineage may be due to an unknown feature of the strain making it more transmissible in these populations or may simply be opportune e.g. due to an international mass gathering [34].
More than a third (6/17) of the lineage 2c RoW cluster isolates were from an anogenital site among MSM. Although two groups of isolates (a group of three and a group of two) were from patients attending sexual health services that utilised common microbiology laboratories (at opposite ends of the UK), the corresponding isolates were distributed over time (2011, 2012 and 2013; and 2017 and 2018, respectively) and within the population structure of the corresponding cluster. Furthermore, the second group also differed from one-another by serogroup. This, along with the finding of the recent msm carriage study in New York (USA) [14], may be indicative of a propensity for this cluster to infect the anogenital site as has previously been observed with ST-11 complex lineage 11.2 strains [15,16,18]. One factor in the emergence of anogenital-associated cc11 strains was the acquisition of an active nitrite reductase (aniA) gene [15,18]. Among the cc4821 lineages, only lineage 1 lacked aniA functionality, however, 14/17 lineage 2c RoW isolates, including the anogenital isolates, and 15/40 Chinese lineage 2c isolates shared a putative ancestral recombination involving partial sequences of the divergent aniA and norB genes along with the intergenic/promoter regions. Although the corresponding sequence was exclusively meningococcal on the PubMLST database, its initial origin was likely in MenA-associated lineages dating back to 1915. It is possible that the newly generated alleles or the newly acquired intergenic region (containing promoter regions) where many of the changes occurred, are more efficient than the existing ones. This, however, requires further investigation. Formal bioinformatic analyses are also required to confirm the origin of the putatively horizontally acquired sequence. Another possible explanation for this particular cluster alone being observed in anogenital sites is that Chinese anogenital meningococci from other lineage 2c clusters or the other cc4821 lineages may simply not be collected, sequenced, posted or identified as such on PubMLST. Interestingly, lineage 2c also included a rare isolate lacking an fhbp gene and therefore unable to express fHbp which has also previously been implicated in adaptation to the anogenital niche [18]. None of the anogenital isolates possessed LOF mutations in mtrC that have been associated with cervical infection in gonococci and urogenitally adapted meningococci from uretheritis outbreaks among heterosexual males [19]. One possible explanation for this is that cc4821 isolates were all isolated from MSM and the cervix may not have been encountered in recent ancestral isolates.
The absence of the lineage 1 epidemic clone outside of Asia is surprising given that it has caused most cases of cc4821 disease within China. This may, in part, be due to a lack of corresponding genomes. Canada, for example, has reported IMD due to a P1.7-2,14:ST-4821 isolate [31,35] but no genome is currently available. On our phylogeny, with the exception of a single diffuse isolate from China in 1980, PorA P1.7-2,14 was only observed within the discrete cluster representing the epidemic clone. This supports PorA P1.7-2,14 as a specific marker for the clone among cc4821 isolates. In addition, PorA VR1 P1.7-2 alone was a strong marker for the epidemic clone cluster accounting for 22/23 (96%) isolates. Three of the epidemic clone isolates possessed a different VR2. In light of this, the lack of the epidemic clone outside of China is further supported by available PorA data among non-Chinese, non-genomic, cc4821 isolate submissions on the PubMLST Neisseria database (accessed 13 th July 2020). None of the isolates from Europe  Table).
The widespread distribution of fluoroquinolone resistance in cc4821 is well documented [36]. The present study confirms the distribution of diverse resistance-associated (MICs > 0.03 mg/L) mutant gyrA alleles throughout the four cc4821 lineages and, with the exception of lineage 1, these were interspersed with isolates possessing susceptibility-associated (MICs �0.03 mg/L) WT alleles. A small number of well distributed isolates also possessed mutant parC alleles that may further increase fluoroquinolone MICs. Notably, the RoW cluster does not possess a mutant gyrA, however, it may be susceptible to stable acquisition of fluoroquinolone resistance given the situation in the broader clonal complex and the gradual emergence of fluoroquinolone resistance in Western countries [35,[37][38][39]. Interestingly, all of the RoW isolates possessed a mutant penA allele associated with reduced penicillin susceptibility (0.094 to 1 mg/L). Although diverse mutant penA alleles were found among Chinese isolates in each of the four lineages, they were relatively rare. As the present study constituted a genomic survey, we focussed on antibiotic resistance/reduced susceptibility determinants commonly associated with cc4821, rather than phenotypic MICs. Further work may include consideration of other potential resistance determinants such as multidrug efflux systems or beta lactamase, the latter of which has recently been reported among meningococci in North America and Europe [37,40,41].
Potential 4CMenB strain coverage of lineages 2a, 2b and 2c, including the RoW cluster was heavily reliant on NHBA. As the corresponding peptides are rare among existing MATS datasets there is a critical need for MATS analysis of isolates representing each of the main peptides. This may then be used to extend the current list of covered, not covered, and unpredictable isolates as defined by gMATS [12]. Perhaps reassuringly, more than half of the lineage 1 isolates, including 85% (28/33) of invasive isolates were also potentially covered by fHbp. This vaccine may, therefore, be an option if an fHbp lineage 1-possessing MenB strain should emerge to cause a large outbreak or epidemic. The majority of isolates without a variant 1 fhbp possessed a variant 2 or 3 gene and so the majority of isolates are potentially covered by Men-fHbp, though this would need confirming with the MeASurE assay [11]. Of concern, two isolates lacked a functional fhbp and one isolate lacked a function nhba. Such isolates and their corresponding strains have infrequently been reported before [30,34,42] and should be monitored due to the potential for escape of subcapsular vaccines. They should also be considered when devising new vaccine formulations.
In conclusion, cc4821 can now be considered in terms of four distinct lineages and the phylogeny herein can form a template with which to analyse genomes from prevailing cases. A small and distinct cluster of lineage 2c is responsible for the majority of cases outside of China and may have an affinity for anogenital sites. In light of emerging fluoroquinolone and penicillin resistance, uncertain strain coverage by subcapsular vaccines, and the repeated emergence of MenB cc4821 strains through capsule switching, there is an urgent need to test isolates representative of each lineage in the MATS and MeASurE assays.