A systematic approach to classify and characterize genomic islands driven by conjugative mobility using protein signatures

Abstract

Genomic islands (GIs) play a crucial role in the spread of antibiotic resistance, virulence factors and antiviral defense systems in a broad range of bacterial species. However, the characterization and classification of GIs are challenging due to their relatively small size and considerable genetic diversity. Predicting their intercellular mobility is of utmost importance in the context of the emerging crisis of multidrug resistance. Here, we propose a large-scale classification method to categorize GIs according to their mobility profile and, subsequently, analyze their gene cargo. We based our classification decision scheme on a collection of mobility protein motif definitions available in publicly accessible databases. Our results show that the size distribution of GI classes correlates with their respective structure and complexity. Self-transmissible GIs are usually the largest, except in Bacillota and Actinomycetota, accumulate antibiotic and phage resistance genes, and favour the use of a tyrosine recombinase to insert into a host's replicon. Non-mobilizable GIs tend to use a DDE transposase instead. Finally, although tRNA genes are more frequently targeted as insertion sites by GIs encoding a tyrosine recombinase, most GIs insert in a protein-encoding gene. This study is a stepping stone toward a better characterization of mobile GIs in bacterial genomes and their mechanism of mobility.

Graphical Abstract

Figure 1.

GI classification from AtollGenDB. (A) Number of raw entries for each of the five GI sources centralized in AtollGenDB. Quality control is performed on the total number of GIs to obtain AtollGenDB data to be analyzed, completed by sequence and taxonomy information from the NCBI Nucleotide database (nuccore). The number of unique and independent islands are then extracted. (B) HMM signature sources used for the classification and characterization steps. (C) Description of the five classification rules. (D) Description of the classification decision tree. (E) Classification results of the 258193 independent GIs by source.

Figure 2.

Sequence length and inter-taxa distributions.(A) Sequence length distribution of independent GIs across classes (total numbers in parenthesis). Two data points from the GI class exceeding 300 kb are not shown. (B) Distribution of top-8 taxa per GI class. (C) Sequence length distribution of ICEs across top-8 taxa. Statistical significance was calculated using Welch's t-test and the Bacillota phylum data as the reference (ns = not significant; ***P < 0.001).

Figure 3.

Integration modules and insertion sites characterization.(A) Distribution of integrase types across the various GI classes. The few GIs (95) containing more than one integrase type in the same CDS are not shown in the bars. (B) Illustration of the insertion site categories (see Material and Methods for definitions). Black hemispheres correspond to GI boundaries. (C) Distribution of integrase types across insertion sites. (D) Distribution of insertion sites across GI classes.

Figure 4.

Diversity of antibiotic resistance mechanisms in cargo. (A) Distribution of GIs harbouring antibiotic resistance-related CDS per GI class. (B) Distribution of top-8 taxa per antibiotic resistance mechanism. (C) Distribution of antibiotic resistance mechanisms CDS count across GI classes. GIs containing multiple mechanisms are counted more than once. (D) Waffle chart of antibiotic resistance mechanisms distribution across all GIs bearing antibiotic resistance.

Figure 5.

Diversity of defense systems in cargo. (A) Distribution of distinct and complete defense systems found per GI class. (B) Distribution of top-8 taxa across the top-10 defense systems found in data. (C) Distribution of the top-6 defense systems found in each GI class. (D) Waffle chart of the top-10 defense system distribution.

Figure 6.

Coexisting antibiotic resistance mechanisms and defense systems in cargo. (A) Distribution of the absence or presence of at least one antibiotic resistance (AR) or defense system (DS) gene per GI class. (B) Distribution of AR, DS or both for each of the top-8 taxa.