Five decades of genome evolution in the globally distributed, extensively antibiotic-resistant Acinetobacter baumannii global clone 1

Abstract

The majority of Acinetobacter baumannii isolates that are multiply, extensively and pan-antibiotic resistant belong to two globally disseminated clones, GC1 and GC2, that were first noticed in the 1970s. Here, we investigated microevolution and phylodynamics within GC1 via analysis of 45 whole-genome sequences, including 23 sequenced for this study. The most recent common ancestor of GC1 arose around 1960 and later diverged into two phylogenetically distinct lineages. In the 1970s, the main lineage acquired the AbaR resistance island, conferring resistance to older antibiotics, via a horizontal gene transfer event. We estimate a mutation rate of ∼5 SNPs genome^{- 1} year^{- 1} and detected extensive recombination within GC1 genomes, introducing nucleotide diversity into the population at >20 times the substitution rate (the ratio of SNPs introduced by recombination compared with mutation was 22). The recombination events were non-randomly distributed in the genome and created significant diversity within loci encoding outer surface molecules (including the capsular polysaccharide, the outer core lipooligosaccharide and the outer membrane protein CarO), and spread antimicrobial resistance-conferring mutations affecting the gyrA and parC genes and insertion sequence insertions activating the ampC gene. Both GC1 lineages accumulated resistance to newer antibiotics through various genetic mechanisms, including the acquisition of plasmids and transposons or mutations in chromosomal genes. Our data show that GC1 has diversified into multiple successful extensively antibiotic-resistant subclones that differ in their surface structures. This has important implications for all avenues of control, including epidemiological tracking, antimicrobial therapy and vaccination.

Keywords: Acinetobacter baumannii; antibiotic resistance; capsule; evolution; phylogenomics; recombination.

Fig. 1

Summary of genome variation in A. baumannii GC1. Heatmap (centre): density of SNPs identified in each genome compared with the A1 reference. Plot (top): density of recombination events detected within the GC1 genomes, across the A1 reference genome, inferred from the SNP data using Gubbins. Tree (left): ML phylogeny of genome-wide SNPS, excluding those attributed to recombination events in the Gubbins analysis. L1 and L2, lineages 1 and 2; red nodes have 100 % bootstrap support. Note the isolate highlighted in red has low confidence SNP calls and its position in the tree is uncertain.

Fig. 2

Temporal and phylogenetic analysis of 43 A. baumannii GC1. (a) Dated whole-genome phylogeny constructed using beast. x-axis, calendar years; grey bars, 95 % highest posterior density for divergence dates of selected nodes; tip colours, KL types; branch colours, clades sharing a recombination event resulting an exchange of the KL. Tips are labelled with isolate names and the KL and OC types, coloured to highlight those that differ from the inferred ancestral types KL1 and OCL1. (b) Heated timeline indicating time intervals in which antimicrobial resistance-associated mutations arose, inferred from the tree by overlaying x (time) coordinates of branches on which the mutational events occurred. Upreg. (3rd gen ceph), upregulation of ampC gene expression resulting in resistance to third-generation cephalosporins. (c) Scatter plot showing linear relationship between year of isolation and ML branch lengths (tree shown in Fig. 1). Red point, isolate for which genome data may be unreliable (excluded from beast analysis); black line, regression line excluding this isolate; red line, regression line including all isolates.

Fig. 3

ML tree and antimicrobial resistance determinants for 44 A. baumannii GC1. L1 and L2, lineage 1 and 2. Key resistance determinants are indicated with symbols, according to inset legend.

Fig. 4

Capsule biosynthesis gene clusters (KL) in GC1 genomes. Capsule gene clusters are drawn to scale from GenBank accession numbers listed in Table S1 (and GenBank accession number KP100029 for KL40), scale bar is shown. The horizontal bar above indicates the K locus region and the KL names are shown on the left. Genes are shown as arrows that are coloured according to the biological role of the encoded product, according to the colour scheme shown on the right. (a) Alignment of GC1 capsule gene clusters. Dark grey shading between gene clusters indicates DNA sequence that is >95 % identical, light grey shading indicates 80–95 % DNA identity. (b) Insertion variants of KL1. Positions of insertion sequence elements in each KL1 variant are shown with the identity and orientation of the insertion sequence below. For KL1c in Naval-83, the additional segment containing wzy, atr25 and three unknown ORFs and its position in the genome is shown on the right. Black flags indicate the position of a 9 bp duplication.

Fig. 5

Recombination events affecting KL in GC1. Left: tree showing relationships between isolates carrying KL types other than the ancestral type (KL1), extracted from the whole-of-GC1 beast tree (Fig. 3). Leaf nodes are coloured by KL type, branch colours indicate shared ancestry of the K locus (inferred from shared SNPs shown in the bar plot). Bar plots: density of SNPs compared with the A1 reference genome; black horizontal bars, regions that are non-homologous with A1 and thus no SNPs can be called; shared SNP alleles are indicated by use of the same colours across plots (i.e. each colour represents a single recombination event). x-axis, coordinates (Mbp) in the A1 reference genome; dashed lines indicate KL boundaries.

Fig. 6

Recombination events affecting the OCL in GC1. Left: tree showing relationships between isolates carrying OCL types other than the ancestral type (OCL1), extracted from the whole-of-GC1 beast tree (Fig. 3). Leaf nodes are coloured by OCL type, branch colours indicate shared ancestry of the OC locus (inferred from shared SNPs shown in the bar plot). Bar plots: density of SNPs compared with the A1 reference genome; black horizontal bars, regions that are non-homologous with A1 and thus no SNPs can be called; shared SNP alleles are indicated by use of the same colours across plots (i.e. each colour represents a single recombination event). x-axis, coordinates (Mbp) in the A1 reference genome; dashed lines indicate OCL boundaries.