Genomic surveillance of Neisseria meningitidis serogroup B invasive strains: Diversity of vaccine antigen types, Brazil, 2016-2018

Background Neisseria meningitidis serogroup B remains a prominent cause of invasive meningococcal disease (IMD) in Brazil. Because two novel protein-based vaccines against serogroup B are available, the main purpose of this study was to provide data on the diversity and distribution of meningococcal vaccine antigen types circulating in Brazil. Methodology Genetic lineages, vaccine antigen types, and allele types of antimicrobial-associated resistance genes based on whole-genome sequencing of a collection of 145 Neisseria meningitidis serogroup B invasive strains recovered in Brazil from 2016 to 2018 were collected. Results A total of 11 clonal complexes (ccs) were identified among the 145 isolates, four of which were predominant, namely, cc461, cc35, cc32, and cc213, accounting for 72.0% of isolates. The most prevalent fHbp peptides were 24 (subfamily A/variant 2), 47 (subfamily A/variant 3), 1 (subfamily B/variant 1) and 45 (subfamily A/variant 3), which were predominantly associated with cc35, cc461, cc32, and cc213, respectively. The NadA peptide was detected in only 26.2% of the isolates. The most frequent NadA peptide 1 was found almost exclusively in cc32. We found seven NHBA peptides that accounted for 74.5% of isolates, and the newly described peptide 1390 was the most prevalent peptide exclusively associated with cc461. Mutated penA alleles were detected in 56.5% of the isolates, whereas no rpoB and gyrA mutant alleles were found. Conclusion During the study period, changes in the clonal structure of circulating strains were observed, without a predominance of a single hyperinvasive lineage, indicating that an epidemiologic shift has occurred that led to a diversity of vaccine antigen types in recent years in Brazil.

Introduction unique alleles at each of the defined seven housekeeping gene loci. Sequence types (STs) are unique combinations of these seven alleles from the defined locus. Groups of related STs were grouped into clonal complexes (ccs) identified by MLST [27].
Meningococcal scheme nomenclature for antigenic variants. Different classification schemes have been proposed to describe allelic variation in the fhbp gene and the corresponding peptides: the Novartis/GlaxoSmithKline nomenclature classifies the protein variants of fHbp into three variant families, variants 1, 2, and 3 [28]; the Pfizer nomenclature classifies it into subfamilies A and B [29]; and the fHbp database at PubMLST proposes a unified nomenclature in which unique fHbp peptide and nucleotide sequences are arbitrarily assigned sequential numbers in order of discovery independent of subfamily/variant [30]. For this study, a combination of these systems was used: subB/var1 correlates to subfamily B or variant 1, and subA/var2-3 correlates to subfamily or variants 2 and 3. The NadA classification is based on peptide sequence homology in one of the four variants described: NadA-1, NadA-2/ 3, NadA-4/5, and NadA- 6 [31]. The NHBA peptide is assigned arbitrarily integer values to unique peptide sequences. PorA classification is based on nucleotide and peptide sequence homology and recognized the previous serologic classification: the prefix "P1." followed by the VR1 family name, followed by a dash and then the variant number, and the VR2 variant name in the same format [32]. All nomenclature is available on the PubMLST website (https:// pubmlst.org/). Statistical analysis. The association of fHbp subfamily/variant with patient age was evaluated using the chi-square test, and a p-value lower than 0.05 was used to denote statistical significance. Information on patient age is provided in S1 Table. Data availability statement. Nucleotide sequences have been deposited in GenBank under the accession numbers listed in S2 Table. This study was approved by the scientific (CTC 39-K/2018) and ethical committees (n. 3.448.268) of the Institute Adolfo Lutz (IAL), São Paulo, Brazil, and Institute of Tropical Pathology and Public Health, Federal University of Goiás, Goiânia, Brazil.

Genetic variability and distribution of PorA, PorB, and FetA
Thirty-five alleles of PorB2 (n = 2) and PorB3 (n = 33) were found among the variants.  Table).
A total of 17 rpoB alleles were found among the 145 isolates. Five rpoB alleles had not been previously assigned, and four of them were found only once. The four most common rpoB alleles (n = 100/145, 69.0%) were almost exclusively associated with a unique cc: alleles 2, 5, 34, and 72 were related to cc32, cc461, cc213, and cc35, respectively (S2 Table).

Discussion
Here, we present the detailed genomic analysis of a collection of MenB invasive strains from Brazil, focusing on vaccine antigens, antimicrobial resistance genes, and hyperinvasive lineage distribution. After three decades of the predominance of cc32 in the MenB clonal structure in Brazil [13,33], we observed the important circulation of cc213, cc35, and cc461, indicating that an epidemiological change has occurred. Our data show that there is no prevalent circulation of a single MenB cc in Brazil, which has currently been observed for serogroup B IMD epidemiology in some European countries and not in North America [15,17,34]. cc213, cc35, and cc461 have previously been reported in Argentina, Australia, Colombia, Paraguay, Tunisia, Turkey, and Europe [23,22,[35][36][37][38][39][40][41]. Additionally, clonal diversity exhibiting a decline in cc41/44 and cc32 and an increase in cc269 and cc461 was observed in the Republic of Ireland over 19 years [39]. Variations over time that led to a change in the distribution of cc were also reported in western Australia, England, Wales, and northern Spain [14,37,39,42,43]. It is well-documented in meningococcal epidemiology that certain meningococcal lineages are known to transmit globally [44], and this may be a possible explanation for the increase in cc213, cc35, and cc461 in Brazil. Thus, further studies are needed to better understand the emergence of these clones in Brazil and how they can impact transmission in the near future.
Additionally, the predominant circulation of serogroup B strains carrying fHbp peptide 1 for decades in Brazil has been described [13,22], and a strong association between fHbp subB/ var1 and peptide 1 within cc32 is well described in the literature [48]. We have shown the circulation of fHbp subB/A (var1/2-3) in almost equal proportions in recent years, in contrast to reports in China, where subA/var2 is prevalent [49], and in the United States and some European countries, where subB/var1 has prevailed among serogroup B IMD strains [50][51][52]. The epidemiology of fHbp subfamilies/variants has been related to different age groups. IMD cases caused by strains with fHbp subA/var2-3 have caused more disease in infants (< 1 year) than adolescents and young adults [53,54]. Our data reinforce the previous observation; 80.5% (p = 0,005) of IMD cases in patients less than 1 year of age were due to fHbp subA/var2-3 strains. Strains with fHbp peptides 24 (subA/var2) and 47 (subA/var3) were frequent causes of IMD cases in infants <1 year of age.
Regarding the nadA gene, only 26.2% (38/145) of the studied strains presented the potential for expression of a functional peptide; this low percentage has also been reported by others, including in a study on a collection of Brazilian MenB strains [13,43,51]. The majority (27/38, 71.0%) of these strains had the potential of expressing NadA peptide and were also predicted to encode the fHbp subB/var1 peptide. Only 11 strains harbored fHbp subA/var2/3 in the presence of the NadA peptide. Such an observation of a close correlation of NadA and fHbp subB/ var1 peptide within cc32 has also been described previously, including among Brazilian meningococcal strains [13,43,47,52,56]. cc32 was the only lineage-clustering isolate (25/38, 66.0%) presenting the potential for expression of NadA peptide, as prevalent peptide 1, as previously reported [36,47,55,56]. All strains belonging to cc213 were assigned to variant 4/5, peptide 3 or 79, but we found only four frameshift peptides, an event frequently reported for cc213 [16,36,43].
We also detected high heterogeneity in NHBA peptides, but these prevalent NHBA peptides have been widely reported; an exception is NHBA peptide 1390, which is first described in this study. We observed a close association with types of NHBA peptides within ccs as well as those already described [36,41,51,57]. In our study, cc213 includes strains that also present NHBA peptide 115, which was not observed in Spain [36]. Interestingly, we showed for the first time the association of cc461 with NHBA peptide 1390.
Considering the vaccine antigen components of the 4CMenB vaccine, fHbp peptide 1 (sub/ var1) was present in 21 strains (14.5%), NHBA peptide 2 was present in 1 strain (0.7%), Por-AVR2-4 was present in 15 strains (10.3%), and no strain presented NadA peptide 8. On the other hand, 10 strains (6.9%) carried fHbp peptide 45 (subA/var3), matching with one of the antigens of the MenB-FHbp vaccine. The genetic characterization of vaccine antigens is not sufficient to predict the strain coverage. In an attempt to fill this gap, two methods (MATS and MEASURE) were developed to measure the level of antigenic expression and provide predicted strain coverage by 4CMenB and MenB-FHbp vaccines, respectively [50,58]. Recently, a genetic MATS (gMATS) was proposed for predicting strain coverage by an association of antigen genotyping and MATS results [59].
Temporal changes in the prevalence of vaccine antigen variants can impact predicted MenB strain coverage. Recently, a decrease in the potential predicted coverage of MenB strains by the 4CMenB vaccine was observed in England, Wales, Northern Ireland, North Spain, and The Netherlands [37, 39,42]. MATS was performed in a Brazilian collection of 99 serogroup B IMD strains isolated in 2010, and the strain coverage predicted by MATS was 80.8%. There is no information about fHbp, NHBA, or NadA peptides in this studied collection. However, we can assume that the predicted coverage of MenB strains can be attributed to the predominance of cc32 mostly associated with fHbp peptide 1 (subB/v1) and to the prevalence of NHBA peptide cross-reactivity with the 4CMenB vaccine [60].
Half of the studied MenB strains belonged to three antigenic distinct emerging ccs: cc461, 35, and 213. Therefore, we identified a shift in the prevalent repertoire of the antigens of the serogroup B IMD strains. The predominance of fHbp subB/var1 decreased between 2004 (86.7%) and 2016-2018 (31.7%) due to circulation of cc461, cc35, and 213, which predominantly harbored fHbp subA/var2-3. It was observed that fHbp subA/var2-3 or subB/var1 antigens elicit mainly subfamily specific responses and that these are cross-protective against strains expressing variants from the same subfamily [58].
The emergence of these ccs carrying completely unknown peptides makes it difficult to estimate the real vaccine predicted coverage, since no experimental data are available. Most strains belonging to cc461 harbored fHbp peptide 47 (subA/var3), which is not covered by gMATS and had NHBA peptide 1390, which is unpredictable by gMATS or MATS (S2 Table) [59]. Most strains belonging to cc35 harbored fHbp peptide 24 (subA/var2), which is not covered by gMATS but did have NHBA peptide 21, which is covered by gMATS (S2 Table) [59]. All strains belonging to cc461, and cc35 did not have NadA or PorAVR2-4. Most strains belonging to cc213 harbor fHbp peptide 45 (subA/var3) and NHBA peptide 18, which are not covered by gMATS, or had NHBA peptide 115, which is unpredictable by gMATS (S2 Table) [59]. All strains belonging to cc213 had NadA 4/5, which has no cross-reactivity between NadA variant 2/3 [56]. On the other hand, the fHbp peptides 47 and 45 are indicated as being potentially susceptible to the MenB-FHbp vaccine, and there is insufficient data on fHbp peptide 24 to predict coverage by this vaccine (S2 Table) [61]. In light of the large antigenic variability found in this study, an ongoing investigation will provide more contribution to the cross-reactivity of this studied strain collection with available MenB vaccines based on genomic data. However, our data will be limited by the lack of cross-reactivity data from the new peptides with the available vaccines.
Although meningococcal resistance to ciprofloxacin remains uncommon, strains not susceptible to the drug have been reported sporadically in other countries, including Brazil [62][63][64][65]. Our results showed that Men strains remain extremely susceptible to ciprofloxacin, and we did not find a mutated gyrA allele. Currently, rifampicin is one of the recommended antimicrobials for chemoprophylaxis for IMD in Brazil [65]. Previous studies have already shown rifampicin resistance [66]. However, no mutated rpoB gene was characterized in our study. Therefore, our data strengthen the indication for the use of rifampicin as a drug that can be used for IMD chemoprophylaxis in Brazil.
Alterations in the penA gene encoding penicillin-binding protein 2 (PBP2) seem to be the most common mechanism of penicillin resistance [67]. Our findings indicate that the frequency of reduced susceptibility to penicillin remains high in Brazil. An upward trend of reduced susceptibility to penicillin has also been reported in other countries [68,69]. Interestingly, we observed the exclusive association of nonmutated penA allele 420 with cc461, which is in contrast to what was observed in Tunisia, where cc461 is associated with the mutated penA allele 33 [40]. Our data suggest a clonal expansion of cc461 associated with reduced susceptibility to penicillin. An association of serotype 19 within Brazilian serogroup B strains with this type of resistance has been previously described [70]. Other mechanisms for inducing reduced susceptibility to penicillin without mutation of the penA gene have been described [71]. Overall, more research is required to determine which type of mechanism is involved in cc461.
Important limitations of this study should be considered. The quality of epidemiological surveillance and meningococcal diagnosis differ among various geographic regions of the country. Although there has been an improvement with the incorporation of qPCR in our surveillance system, approximately one-third of IMD cases are still confirmed by clinical criteria, and only 47% of the IMD cases reported during the study period had serogroup information [2]. Therefore, we might have underestimated the real burden of IMD cases in Brazil. Nonetheless, we are aware that in the era of the dramatic spread of the COVID-19 pandemic in Latin America, continuous surveillance of IMD cases is key in monitoring the emergence of possible cases of IMD secondary to COVID-19, as recently observed in France [72].
A considerable amount of additional work will be needed to describe the antigenic diversity of the N. meningitidis strains circulating in Latin America. However, this study makes several noteworthy contributions to understanding the temporal changes in vaccine antigens as well as knowledge of emerging hyperinvasive lineages of N. meningitidis strains circulating in our region. The present data reinforce that continued laboratory surveillance over time supports local epidemiological data as the driver in planning and preparedness in making decisions about the best vaccination schedule strategy against serogroup B IMD.