Architecture of the superintegron in Vibrio cholerae: identification of core and unique genes

Introduction

Vibrio cholerae is a diverse, environmental, gram-negative bacterial species that can be pathogenic and can cause cholera, a severe diarrheal disease that occurs most frequently in epidemic form^1,2. The V. cholerae genome consists of two chromosomes. The largest chromosome of 2.96 Mbp encodes most essential genes. The 1.07 Mbp small chromosome contains few essential genes and the superintegron (SI), a large gene capture and excision system of ~120 kbp² (Figure 1). The SI is characterized by a site-specific integrase gene (IntI4) closely associated with a cognate recombination site attI and a promoter Pc followed by a large array of gene cassettes. Within the SI, the gene cassettes generally consist of a promoterless open reading frame (ORF) flanked by two recombination sites termed V. cholerae repeats (VCRs)³. Cassettes can be excised from any position in the array through VCR × VCR recombination mediated by the integrase. The resulting circular intermediate can then be integrated, preferentially through attI × VCR recombination by the integrase, bringing the cassette under control of Pc^4,5. Since gene cassettes are usually promoterless, only the first few cassettes are expressed by Pc and the rest of the array can be seen as a reservoir of standing genetic variation⁵.

Figure 1. Schematic organization of the Vibrio cholerae genome and the superintegron (SI).

The functional platform of the SI consists of an integrase gene, a cassette promoter (Pc), and a primary recombination site (attI). The system maintains an array of several cassettes, which generally consist of a promoterless ORF flanked by two recombination sites termed VCR (V. cholerae repeats).

The functions of the majority of the SI genes are unknown; however, a few genes have been characterized and it has been suggested that they are involved in adaptive functions such as toxin-antitoxin (TA) loci. TA loci consist of two genes in an operon encoding a ‘toxin’ and an ‘antitoxin’. The expression of the toxins reduces cell growth and prevents colony formation, thus exerting a bacteriostatic rather than bacteriocidal condition. However, cell viability can be rescued by later overproduction of the cognate antitoxins⁶.

The pangenome describes the complete repertoire of genes in a bacterial species, which includes the "core genome" containing genes present in all strains, a "dispensable genome" containing genes present in two or more strains, and "unique genes" specific to single strains⁷. Previous phylogeographic analysis, considering V. cholerae strains and its sister species Vibrio metecus⁸, showed that, in contrast to the core genome, the SI displays strong geographical differentiation, and cassettes from the V. cholerae group cluster with those of V. metecus from the same place rather than with cassettes from geographically distinct V. cholerae. It suggested that SI structure is influenced by geographic boundaries and in response to environmental conditions. The flexible nature of the SI that results from gene cassette capture, deletion and rearrangement is thought to make it a hotspot of V. cholerae diversity, but beyond the basic structure it is not clear if there is a core genome in the SI and if so how it is structured. The aim of this work was to explore the core genome structure and the differential gene content among strains of V. cholerae.

Methodology

Based on the complete genomes of seven V. cholerae and one V. mimicus representative strains (Table 1), we searched repeats above 10 nucleotides and used one VCR sequence (AAC AAA CGC CTC AAG AGG GAC TGT CAA CGC GTG GCG TTT CCA GTC CCA TTG AGC CGT GGT GGT TTC GGT TGT TGT GTT TGA GTT TAG TGT TAT GCG TTG TCA GCC CCT TAG GCG GGC G) to search for sequences with more of 45% nucleotide identity. The SI sequences were recovered using the locations of the structural gene IntI4 and VCRs identified with the UGENE software⁹. Cluster analysis of functional genes was performed using the pangenome analysis pipeline¹⁰, which searches for homologs or orthologs among multiple genomes using the MultiParanoid (MP) method (based on a 90% nucleotide identity threshold). For each pair of genes in the same cluster, the local matched region is no less than 25% of the longer protein coding sequence and the global matched region is no less than 50% of the longer protein coding sequence. The minimum score value and E-value in BLAST are 50 and 1e-8¹⁰. The gene content was converted to a presence/absence (0/1) matrix and then the core, dispensable and unique genes were identified by in-house R scripts. The phylogenetic tree based in gene content and split network for gene content were constructed with SplitsTree4¹¹ using the GeneContentDistance method¹². The SI structure and comparison of seven V. cholerae and, their sister species, V. mimicus were performed using genoPlotR¹³.

Results and discussion

SI regions were extracted from the seven V. cholerae and one V. mimicus genomes (Table 1). The 1285 genes recovered were clustered and a total of 408 clusters were detected (Figure 2A; Table S1). The pangenome of the SI of Vibrio strains evaluated was 408 genes, of which eight correspond to core genes, 196 are distributed or dispensable genes and 204 are unique genes. Six of the eight core genes are present in many copies (Table 2). The pangenome profile analysis shows that the cluster numbers of core genome are almost the same, when the SI considered reaches nine, while the pangenome is still increasing (Figure 2A). We infer that the V. cholerae SI has an open pangenome, which means that V. cholerae may have the ability to import new SI gene cassettes, which affect its plasticity and diversity. On the other hand, the set of SIs, from clinical and environmental lineages, used in this study are apparently representative of this species because allowed to establish that the core genome is close to being completed.

Figure 2.

(A) Pangenome plot of the SI region considering seven V. cholerae and one V. mimicus genomes. 1285 total genes, 408 pangenome clusters and eight core clusters were identified. (B) Word clouds of cluster function enrichment comparison according to clusters of orthologous groups (COG) for whole and core clusters identified are shown at the top and bottom, respectively. Clusters that are not assigned in the COG classification were excluded from the figure.

Function enrichment analysis of gene clusters were performed according to description of gene annotation (File S1) supplied to the pangenome analysis pipeline¹⁰. From the 408 clusters, 329 were unclassified by the function enrichment analysis. Following the categorization of Cluster of Orthologous Groups (COG), the characterized clusters were rich in the following categories: translation, ribosomal structure and biogenesis, transcription, replication, recombination and repair, cell cycle control, cell division, chromosome partitioning, defense mechanisms, cell wall/membrane/envelope biogenesis and posttranslational modification, protein turnover, chaperones, amino acid transport and metabolism, nucleotide transport and metabolism, lipid transport and metabolism, secondary metabolites biosynthesis, transport and catabolism (Figure 2B).

In the SI, random excisions occur throughout the cassette array to form nonreplicative circular intermediates containing one or several cassettes; integration events preferentially occur at the attI site⁵ and are subjected to selection. It is expected that SI core genes would be arranged and stay together; however, we found the core genes are distributed along the SI (Figure 3), apparently without any position effect.

We identified 204 unique genes, 94 belonging to V. mimicus MB451, nine to LMA3984, 45 to O395, nine to 2010EL1786, 14 to MJ1236, seven to IEC224, 20 to M66, and six to N16961 (Figure 3; Table S1). Considering only the V. cholerae SI, there are 21 core genes, most of them present in many copies and rich in the transcription, replication, recombination and repair, translation, ribosomal structure and biogenesis categories.

Figure 3. Superintegron (SI) structure and comparison of seven strains of V. cholerae and a strain of V. mimicus.

The core, dispensable and unique genes are indicated by red, cream and blue arrows, respectively. Vertical blocks between sequences indicate regions with more than 1 kb of shared similarity shaded according to BLASTn. A phylogenetic tree based on gene content of the SI is shown on the left.

Pandey and Gerdes¹⁴ identified 13 TA loci within the SI of the N16961 strain. Here we identified six TA genes as part of core SI genes (Table 2), of which the relB genes (VCA0349 and VCA0504) were present in all V. cholerae strains (including V. mimicus) SIs. The parE (VCA0359), relB (VCA0477) and relE (VCA0489) genes were present in all V. cholerae SIs. Moreover, we also identified two higBA loci (VCA0469 and VCA468), which encode mRNA cleaving enzymes and can stabilize plasmids⁶, as well as SI genes. The previous authors¹⁴ also identified higBA-1 TA loci (VCA0392 and VCA0391); in our results, these two TA loci are present in all clinical V. cholerae strains (Table S1). These results suggest that V. cholerae TA loci function as essential stress response elements that help cells survive⁶, as well as act to stabilize the massive arrays of SI cassettes, as reported previously¹⁵.

A previous study suggested that SI structure is influenced by geographic boundaries in response to environmental conditions⁸. Here, we found that the clinical nature of the V. cholerae and V. mimicus strains evaluated were not grouped together by the analyses performed. Therefore, the ability of V. cholerae to cause disease must be explained by other virulence factors found outside the SI region.

There are 199 clusters involved with indel or mutation events (Table S2). As for the non-synonymous/synonymous substitution (dN/dS) ratio, we found that 30 clusters were suffering positive selection pressure (dN/dS > 1). At the same time, we could also select those variable clusters as the markers for different strains. Based on pangenome profiles and single nucleotide polymorphism (SNP) information, gene content and phylogenetic trees were constructed (Figure 4). The SNP information from SI was useful for separating V. cholerae from V. mimicus, but nevertheless lacked the resolution to distinguish between the different lineages of V. cholerae. However, using gene content information (Figure 4), a good resolution was reached that was coherent with the evolution of the species and the environmental or clinical nature of the strains. These results indicate that the evolution of V. cholerae into different lineages is reflected in the diversity of the SI, which would be also influenced by horizontal gene transfer in these region, as proposed elsewhere^8,16,17.

Figure 4.

Top: Phylogenetic trees for the V. cholerae SI based on SNPs constructed by the Maximum Likelihood (A), Neighbor-Joining (B) and UPGMA (C) methods. The numbers indicate the bootstrap values. Bottom: Split network for gene content based on the 408 genes in seven V. cholerae and one V. mimicus genomes. The network was constructed with SplitsTree4 using the GeneContentDistance method¹².

Conclusions

In this study, we have revealed the genetic architecture of the V. cholerae SI, which contains eight core genes, many of them present in many copies. The V. cholerae SI has an open pangenome, which means that V. cholerae may have the ability to import new gene cassettes into the SI. The set of the dispensable SI gene cassettes is influenced by the niche and type species. The core genes are distributed along the SI, apparently without a position effect.