As a Gram-negative diplococcus, Neisseria meningitidis is an obligate human pathogen and it usually inhabits the nasopharynx, often asymptomatically. Nonetheless, this pathogen is also a common cause of meningitis and/or sepsis and it can generate large outbreaks. Autotransporters (ATs), which are synthesized by pathogenic Gram-negative bacteria, have been implicated in forming biofilms and may also correlate with virulence (Arenas et al., 2015; Arenas and Tommassen, 2017). To date, eight types of ATs have been studied in N. meningitidis. They are proteins that are usually composed of an N-terminal signal sequence, a secreted passenger domain and a C-terminal translocator domain (Grijpstra et al., 2013). The iga gene of the immunoglobulin A1 (IgA1) protease of N. meningitidis is single-copy and polymorphic, which suggests that it was attained through interstrain horizontal gene transfer (Lomholt et al., 1992). App is an acronym for adhesion and penetration protein which is homologous to the Haemophilus adherence and penetration protein (Hap). Similarly, the NalP is a Neisseria autotransporter lipoprotein which was historically named AspA (autotransported serine protease A) (St Geme et al., 1994; Turner et al., 2002; Serruto et al., 2003; van Ulsen and Tommassen, 2006). The functions of AusI (also known as MspA) have not been elucidated and its protein expression is influenced by NalP (van Ulsen et al., 2003, 2006; Turner et al., 2006). The former 4 ATs are monomeric and have protease activity whereas NadA and NhhA (Neisseria hia homolog A) are trimeric ATs that are related to host cell adhesion (Peak et al., 2000; Capecchi et al., 2005; Scarselli et al., 2006; Arenas et al., 2013). Other than the former four monomeric ATs, AutA is not secreted into the milieu of the bacteria and it also participates in autoaggregation (Arenas et al., 2015; Arenas and Tommassen, 2017). In contrast, the passenger domain of AutB is secreted but the AutB is still attached to the cell surface due to the translocator domain (Arenas et al., 2016).

A detailed study has elucidated the expression of the autB gene in N. meningitidis, whereby the mechanism of AutB appears to affect biofilm formation and epithelial transmigration (Arenas et al., 2016). A comprehensive analysis of AutB has been conducted and it complements the study of AutB, which has firmly assessed the ATs (Arenas et al., 2016). The distribution of the autB gene is well-known in the two pathogenic Neisseria spp. N. meningitidis and N. gonorrhoeae. There is no autB gene in N. lactamica, N. flavescens, N. polysaccharea, and N. sicca while some strains of Haemophilus influenzae, H. haemolyticus, H. parainfluenzae, and H. aegyptius are autB positive (Arenas et al., 2016). The number of AAGC tetranucleotide repeats (Rn) in the autB gene has previously been published and studies have considered whether the autB gene is in frame or out of frame depends on these repeats (Arenas et al., 2016). Although the autB gene is intact in most of the meningococcal strains (105/117), it is often out of phase, suggesting conservation but a very strong negative selection for AutB expression (Arenas et al., 2016). The amino acid sequence of AutB can be divided into three parts: the N-terminal part, the linker and the C-terminal translocator domain. The variability of AutB can be demonstrated by phylogenetic analysis of the N-terminal component and the linker domain and has been designated AutB1, AutB2, and AutB3, most of the variants are associated with AutB1 from N. meningitidis (Arenas et al., 2016). The phase variation was studied in patient strains of sequence type ST-32 clonal complex (CC32) (49 isolates) and CC213 (53 isolates) collected in Netherlands and 207 genome sequences of carriage strains in public data base, only one strain 2081107 (disease isolate) of CC32 have an intact autB gene and demonstrated in Western blotting, five strains of carrier isolates were also have an undisrupted autB gene, no obvious differences in autB expression between carriage and patient strains were demonstrated (Arenas et al., 2016).

It is noteworthy that the autB gene is omnipresent in the pathogenic Neisseria spp. except for 053442 (Arenas et al., 2016). 053442 is a serogroup C strain with the unique ST-4821, which was isolated from a patient in an epidemic of cerebrospinal meningitis in China (Shao et al., 2006; Peng et al., 2007). CC4821 was first isolated in China during 2003–2004 and then rapidly spread to two-thirds of the provinces within China (Shao and Zhu, 2016). Although CC4821 is one of the dominant hyperinvasive CCs in China, it has rarely been isolated in other countries (Zhou et al., 2012). The meningococcal A+C polysaccharide-based vaccine was introduced to the Expanded Program on Immunization (EPI) in 2005 and N. meningitidis invasive serogroup B soon increased after 2006 (Zhou et al., 2012). Meanwhile, the population structure of N. meningitidis invasive serogroup B were used to be mainly CC11 and CC41/44 and it evolved into CC4821 in China, CC4821 is one of the dominant hyperinvasive CCs in China now. It has been verified that there were capsular switching from N. meningitidis CC4821 serogroup C strains to serogroup B strains (Zhu et al., 2015).

The DNA uptake sequence (DUS) is a 10-bp (5′-GCCGTCTGAA-3′) non-palindromic sequence, which mediates the transformation of species-specific DNA (Goodman and Scocca, 1988; Rotman and Seifert, 2014). Twelve base pairs sequences (5′-ATGCCGTCTGAA-3′) have also been noted in pathogenic Neisseria (termed DUS12) (Ambur et al., 2007). It was first to show the presence of DUS bordering autB gene in 2001, in which autB was characterized as lav (Davis et al., 2001). Studies have also demonstrated that the DUS is sufficient for DNA uptake and transformation in previously non-transformable plasmids, while it can also competitively inhibit transformation when a DUS is already present (Goodman and Scocca, 1988; Elkins et al., 1991; Obergfell and Seifert, 2015). Therefore, the purpose of our work is to assess the distribution and characterization of the autB gene in N. meningitidis in China, especially for CC4821.

Microcolony formation is one of the strategies of N. meningitidis to elude the host immune response (Sim et al., 2000). The autotransporter AutB has been demonstrated to play a part in biofilm formation (Arenas et al., 2016). The autB gene was first described as an orf2 downstream of lsi1 (rfaF), a LPS biosynthesis related gene, and was then demonstrated to be horizontally-transferred from H. influenzae to Neisseria immediately before the bifurcation of N. meningitidis and Neisseria gonorrhoeae, where it was characterized as lav (Jennings et al., 1995; Davis et al., 2001). Subsequently, Arenas et al. have extended the analysis of the distribution of the autB gene and it is worth noting that in Arenas' study, N. meningitidis strain 053442 was the only clinical isolate without the autB gene. 053442 was a CC4821 strain and this clonal lineage has only been detected in China (Zhu et al., 2015). CC4821 has been a dominant lineage in China since 2003 (Zhou et al., 2012; Shao and Zhu, 2016). On account of this, we analyzed the isolates in China to further enrich the study of the autB gene.

We have not found any explicit relationship between the autB gene and the meningococcal serogroup. In a former study, all strains of the highly pathogenic CC213 and CC32 were autB-positive and these strains were clinical isolates; all 49 isolates of CC32 were clustered into AutB1 and 52 of the 53 isolates of CC213 were AutB1, while only one CC213 isolate harbored an AutB2 variant (Arenas et al., 2016). In our study, the 67 clinical strains included 37 strains of AutB1 and 3 of AutB2, and the others were autB-null strains. The number of AAGC nucleotide repeats in our strains was 2–11 while in the former study was 3–19 (Arenas et al., 2016). With regard to the 207 carrier isolates in the former study, the number of the repeats was 2–22 and in most of the strains, the autB gene was disrupted by single-nucleotide insertions downstream of the repeats (Arenas et al., 2016). In the 111 isolates from carriers in our study, 25 isolates of CC4821 and 14 of UA were autB-null and isolated after 2005. The repeat number in the other 72 autB–positive isolates was also 2–11, where 60 harbored an AutB1 variant and the five isolates with 3 or 6 AAGC repeats were all disrupted by frame shift mutations. Meanwhile, another 12 AutB2 isolates belonged to CC174 (ST-6933) and UA. There were no AutB3 isolates in our strains (Table 1). Our study indicates that not all highly pathogenic strains are autB-positive and as expression is switched off in most N. meningitidis isolates, it can also be absent in some of the strains such as the CC4821 strains. Of 17 clinical strains of CC4821, 341215, and 440501, with the autB gene belonging to AutB1, two had repeat numbers of 8 and 4, respectively.

The sequences of the 66 autB-null strains between the smpB and glcD gene are all 968 bp long with no SNPs and these sequences are homologous to N. lactamica 020-06 (Figure 1C). The autB gene has been recognized as a pseudogene because it is not expressed in many meningococcal or gonococcal strains with the sequence (Ait-Tahar et al., 2000). Authors suggested this is probably due to the accumulation of mutations on autB, which with no need for coping with selective pressure (Ait-Tahar et al., 2000). Conversely, the high variability of the AutB passenger domain may be a reflection of the antigenic variability response to immune pressure, so that an immune response to AutB can be avoided, and where expression of the autB gene is switched off in most isolates (Arenas et al., 2016).

As has been verified in Arenas' study, the majority of N. meningitidis isolates equipped with the autB gene do not express AutB. One possible explanation is that some functions of AutB can be substituted by other ATs such as NalP, which cleaves the heparin-binding antigen NhbA or the α-peptide of IgA protease from the cell surface (Serruto et al., 2010; Roussel-Jazede et al., 2014). AutB is likely to be a backup mechanism when the concentration of these two proteins is low (Arenas et al., 2016). Nevertheless, the proportion of strains with the autB gene to autB-null strains is 24/27 (1956-2000), 20/31 (2001-2005), 49/85 (2006-2010), and 19/35 (2011-2015) in our study. In the 66 autB-null strains, only two have been isolated in 1985, one in 1988 and the others were all isolated after 2004. There are two hypothesis for this. First of this, it seems that the strains in China habitually lose the autB gene and the autB gene will not be retained as an intact gene in N. meningitidis. However, the number of strains in our study is relatively small so this hypothesis may be premature. Secondly, CC4821 strains are similar to N. lactamica, also shown in Figure 1A. In here, we have demonstrate that this is indeed a conservative feature of this CC, as we found in many isolates of this CC. Thus, in a common ancestor of this CC in China, the autB was lost by acquisition of N. lactamica sequences. Interestingly, two isolates of this CC seems to acquire autB, probably from another circulating strain as they harbor a typical and common AutB1 (see Table 1). This also supported by the similarity in the flanking regions with other CCs found in China and the predominance of this CC. Some isolates of UA, but no other CCs, do not have the autB gene but this is restricted uniquely to isolates of the same ST. Possibly, those isolates acquired sequences from the predominant clones of CC4821, which rarely have an autB gene.

It has been suggested that the autB gene in Neisseria was obtained from Haemophilus through domain shuffling and it has been proposed that the function of the 9 bp inverted repeat sequences of hUS at the beginning of the partial holB gene and the end of the partial tmk gene is recognition sites that enhance the transformation (Davis et al., 2001; Duffin and Seifert, 2010). In autB-null N. meningitidis and N. lactamica 020-06, the 10 or 12 bp DUS was located in the intergenic sequence (Figure 1C), according with the former study that most of the DUS (65%) in Neisseria is located in the intergenic sequences (Ambur et al., 2007). It has been postulated that the DUS may function to competitively inhibit transformation and this may be the mechanism of autB-null in these strains while strains without the DUS could gain the autB gene from other strains (Goodman and Scocca, 1988; Duffin and Seifert, 2010; Obergfell and Seifert, 2015).

In summary, we demonstrated the characterization and distribution of the autB gene and the variability of AutB in N. meningitidis isolates in China. Not all N. meningitidis strains were autB-positive in China. This study offers a new perspective on the autB gene and how it is possible that CC4821 and some of the UA N. meningitidis genomes could lose autB genes or acquire the alternative sequence from N. lactamica, the DUS may promote this process to some extent.

ZS, BZ, and AZ designed the experiments; AZ, FS, LX, YG, and NX carried out the experiments; AZ, PZ, BZ, and ZS analyzed the experimental results; AZ wrote the manuscript; PZ, BZ, and ZS modified the manuscript. All authors have read and approved the manuscript.