The Neisseria gonorrhoeae population structure is not clonal [1–5], its panmictic structure being the result of horizontal genetic exchange [6]. The pathogenic Neisseria spp. are naturally transformable [7] and chromosomal changes are largely mediated by homologous recombination. Natural competence for transformation and homologous recombination allow the generation of mosaic genes, such as the divergent opa genes [8, 9], antigenic variation through recombination with silent cassettes, such as in the pilE/pilS system [10, 11], and the horizontal exchange of whole genes or groups of genes in Minimal Mobile Elements (MMEs) [12]. A Minimal Mobile Element (MME) is defined as a region between two conserved genes within which different genes are found in different strains. These elements are described as 'minimal' reflecting the fact that they do not have features suggesting mobilization by associated transposases, nor do they currently appear to have features by which remotely encoded systems would facilitate mobilization. These were initially identified through comparisons of the available Neisseria spp. genome sequences, on which basis the MMEpheS-pheT was the first to be studied in an extended set of strains [12]. In this region, flanked by the highly conserved pheS and pheT genes, eight different intergenic regions where found in the Neisseria strains that were investigated. Polymorphisms within the flanking gene sequences are indicative of the points of homologous recombination between the imported DNA and the native DNA. Further, assessments of five additional MMEs in the genome sequences show these to be present in different combinations in different strains, supporting the model that these are horizontally, rather than vertically, transferred elements [12].

N. gonorrhoeae may show less diversity than Neisseria meningitidis because it is more likely to be isolated away from other Neisseria spp. in its primary host niche [1]. The exchange of highly similar sequences between strains would prevent sequence analysis revealing the full extent of horizontal exchange. It is quite possible that recombination within genes coding for surface exposed proteins, or those under direct selection, may reveal a higher degree of recombinatorial diversity than is currently established from the predominantly housekeeping gene sequences investigated to date [10, 13–15].

There are four commonly used laboratory strains of N. gonorrhoeae. The complete genome sequence is available for N. gonorrhoeae strain FA1090 [16], for which there is also a physical map [17, 18]. This strain was originally recovered from a patient with disseminated gonococcal infection (DGI) in 1983 [19]. There is also a physical map of the N. gonorrhoeae strain MS11 chromosome [20]. It is from strain MS11 that the Gonococcal Genetic Island (GGI) has been sequenced [21]. The GGI is composed of 59 predicted coding regions, including genes believed to be involved in the secretion of DNA. This strain was isolated from a patient with uncomplicated gonorrhoea [22]. N. gonorrhoeae strain F62 was also isolated from a patient with uncomplicated gonorrhoeae [23], but it does not contain the GGI [21]. The last of the four commonly used laboratory strains of N. gonorrhoeae is strain FA19, which was isolated from a patient with both gonorrhea and DGI. Additionally, it has caused DGI in a laboratory worker, following a mouth-pipetting accident [24]. While other strains are certainly used experimentally, these four represent those most commonly used and for which there is the greatest body of existing literature available.

Since these strains have been widely used in experimental studies it is important to be able to compare them, and to provide a means of relating each to the available genome sequence of N. gonorrhoeae strain FA1090. Some differences between these four commonly used laboratory strains, in addition to the presence/absence of the GGI [21], have been noted previously. The expression of ICAM-1 by human mucosal epithelial cells is upregulated in response to strain FA1090, but not to strain MS11 [25]. Truncations of the protegrin PG-1 showed different potencies against strains FA19 and F62 [26]. There are two immunological classes of the major outer membrane protein Protein I (encoded by porB), designated PIA and PIB [2]. Strains F62 and FA19 have PIA, while strains FA1090 and MS11 have PIB [2, 27, 28]. While this is not an exhaustive list, it does indicate that some difference in gene complement between these strains may be expected, and defining the presence and absence of gene-complement differences may provide keys for the community to address their significant functional differences.

The Pan-Neisseria microarray was constructed by an international consortium of neisseriologists (Saunders, Davies, et al – in preparation). The probes printed onto this microarray represent the coding sequences of three complete neisserial genomes (N. gonorrhoeae strain FA1090 [16], N. meningitidis strain MC58 [29], N. meningitidis strain Z2491 [30]) as well as the coding regions from the GGI [21]. For those coding regions of sufficient length, each probe is complementary to 150–450 bp of the most conserved regions of the CDS between the different genome sequences, that are also best able to discriminate between paralogous genes. Basing the design upon the most conserved regions between the different strains and species makes the array minimally strain specific facilitating the use the Pan-Neisseria microarray in analyzing strains to which it was not designed.

This paper addresses the gene complement differences between the four most commonly used laboratory strains, utilizing the Pan-Neisseria microarray, comparative genomic analysis, and sequencing of the newly identified divergent MMEs revealed by the microarray data. This is the first use of a multi-species, multi-genome microarray to conduct gene complement studies. Additionally, this work supports and expands our observation of Minimal Mobile Elements in the Neisseria spp. [12].

The Pan-Neisseria microarray, coupled with MME analysis of sites of divergence indicated by the microarray analysis, has revealed five new neisserial genes not represented on the microarray, as well as identifying the 59 genes of the GGI in N. gonorrhoeae strain FA19. There are also 36 genes that are present within strain FA1090 that are not universally represented in the other main experimental strains, 16 of which are located within MMEs that are identifiable from comparison of the complete genome sequences of gonococcal strain FA1090 and the available complete meningococcal genome sequences. The ability to identify this range of differences is in part due to the unique nature of the Pan-Neisseria microarray (Saunders, Davies, et al. – in preparation). Its design incorporates probes to the most conserved regions of genes, having been evaluated through the comparison of N. gonorrhoeae strain FA1090 sequences [16] with those of N. meningitidis strains MC58 [29] and Z2491 [30]. Genes from the two meningococcal genomes that were not optimally probed by the gonococcal probes ((≥ 90% homology over 150 bp) or where not represented by the gonococcal probes, had additional probes added to the microarray. The 59 predicted coding sequences of the Gonococcal Genetic Island, as sequenced from N. gonorrhoeae strain MS11 [21], are also represented on the microarray. This microarray composition meant that for the three gonococcal laboratory strains evaluated here, it was possible to determine, solely from the microarray results, that 36 strain FA1090 genes are absent from the other gonococcal strains, while 130 genes are present in these strains that are absent from strain FA1090. This second group of genes could not have been identified with a single-genome microarray based upon strain FA1090 alone, emphasizing the strength of this multi-species, multi-genome design strategy. The microarray results additionally indicated regions of the chromosomes likely to contain divergent sequence, analysis of which revealed five new genes.

In strain MS11, 17 probes designed to strain FA1090 genes did not hybridize, including nine genes associated with five MMEs. Sequencing of two of these MMEs confirmed the absence of the strain FA1090 genes and identified four new genes in strain MS11. One of the five MMEs is the Gonococcal Genetic Island, the sequence of which is known from strain MS11. The remaining two MMEs (hypo-aspA and ribAB-hypo) may simply be too large to amplify using the techniques described here, but may therefore represent the location of additional genes present in strain MS11 and absent in FA1090, and vice versa. There are 19 strain FA1090 probes that do not hybridize to strain FA19 DNA, with seven of these being contained in five MMEs, one of which is the Gonococcal Genetic Island. Sequencing of three of the MMEs confirmed the absence of the strain FA1090 genes and identified one new neisserial gene in strain FA19. The remaining MME (hypo-aspA) was also identified in strain MS11. No differences were observed between N. gonorrhoeae strains FA1090 and F62. Discounting those genes that are likely to display differences due to divergence (pilC, lbpB, tspB, maf-associated, prophage-associated, and the NMA1360 gene fragment) a total of 80 genes were found to contribute to inter-strain gene-complement differences, of which 59 comprise the Gonococcal Genetic Island and one represents the hypothetical strain FA1090 gene that is present in its place (XNG0741; Figure 1). This does not take into account those genes that may be found in strains MS11 and FA19 in the large hypo-aspA MME and in strain MS11 in the large ribAB-hypo MME.

Although this approach to strain comparison, utilizing a multi-species, multi-genome microarray is novel, powerful, and has identified differences between these strains, there are some additional forms of divergence that this approach cannot address. This means that although we have an indication of the significant similarity of strains FA1090 and F62, which will facilitate the use of the available genome sequence in this strain, they cannot be assumed to be identical. There may be other differences between these and the other strains such as: (a) the presence of additional genes that are not currently on the microarray, although these cannot be associated with the absence of MME-associated genes as they are in strains MS11 and FA19; (b) divergence outside of the probe regions used for construction of the microarray, which were selected due to their maximal conservation across the species (≥ 90% ID over a probe of 150–450 bp), therefore the divergence in the fbp iron acquisition genes (strain F62 sequence: U33937) is not detected because the probes hybridize to the conserved regions of the genes at ≥ 98% over the length of the probe; (c) differences in transcriptional regulation, post-transcriptional regulation, and phase variation, for example as in pgtA, which is phase variable in strain FA1090 but not in strain F62 where the gene lacks the phase variation-mediating poly-G tract [38]. Despite these caveats this study clearly shows some of the significant advantages and power of this microarray design approach in its application to genomotyping, and provides a new basis upon which to inter-relate the data obtained from the many studies performed using these strains.

The multi-strain, multi-species microarray design strategy can be used to identify the presence and absence of genes in a more informative fashion than arrays based upon single genomes, or even single species. This approach can be used to identify those genes that are trans-species, rather than species associated, as demonstrated from limited complete genome sequenced examples. This approach is also able to identify chromosomal discontinuities which when coupled to MME analysis can identify new sequences and genes not present in any of the sequences used for microarray design. This study identified many differences, which are likely to account for at least some of their recorded differences in behaviour between the gonococcal experimental strains. This included the identification of five new neisserial genes. The recently completed N. meningitidis serogroup C strain FAM18 complete genome sequence ftp://ftp.sanger.ac.uk/pub/pathogens/nm/ identified 35 new neisserial genes, other than those associated with a large prophage region. This array-based approach is therefore a comparatively cheap and efficient way of finding new genes within a species for which existing genomes are available. The range of differences that have been identified in this study emphasize the utility of a pan-species microarray, with the design characteristics used, in the investigation of diverse bacterial populations.