Evolution and emergence of Mycobacterium tuberculosis

Abstract

Tuberculosis (TB) remains one of the deadliest infectious diseases in human history, prevailing even in the 21st century. The causative agents of TB are represented by a group of closely related bacteria belonging to the Mycobacterium tuberculosis complex (MTBC), which can be subdivided into several lineages of human- and animal-adapted strains, thought to have shared a last common ancestor emerged by clonal expansion from a pool of recombinogenic Mycobacterium canettii-like tubercle bacilli. A better understanding of how MTBC populations evolved from less virulent mycobacteria may allow for discovering improved TB control strategies and future epidemiologic trends. In this review, we highlight new insights into the evolution of mycobacteria at the genus level, describing different milestones in the evolution of mycobacteria, with a focus on the genomic events that have likely enabled the emergence and the dominance of the MTBC. We also review the recent literature describing the various MTBC lineages and highlight their particularities and differences with a focus on host preferences and geographic distribution. Finally, we discuss on putative mechanisms driving the evolution of tubercle bacilli and mycobacteria in general, by taking the mycobacteria-specific distributive conjugal transfer as an example.

Keywords: Mycobacterium tuberculosis; evolution; genetic diversity; host–pathogen relationship; population dynamics; virulence.

Figure 1.

Phylogenetic topology of mycobacteria. Genomes of selected Mycobacterium genus strains were downloaded from the GenBank database and analyzed using PanACoTA v1.4.0 (Perrin and Rocha 2021). Genomes were annotated using Prokka v1.14.5 (Seemann 2014) and their pan-genome was inferred using MMseqs2 v14-7e284 (Steinegger and Söding 2017) based on a minimum sequence identity of 80% at the protein level. Genes conserved across all selected mycobacteria were aligned using MAFFT v7.522 (Katoh and Standley 2013). Maximum-likelihood phylogenetic reconstruction was performed using RAxML-NG v1.2.0 (Kozlov et al. 2019) with the generalized time reversible (GTR) substitution model, mean GAMMA distribution of rate heterogeneity with four categories (G), a maximum-likelihood estimate of stationary frequencies (FO), and 1000 bootstrap replicates. Bipartition support of the best-scoring tree rooted using M. abscessus was computed using the transfer bootstrap expectation metric from BOOSTER v0.1.2 (Lemoine et al. 2018). The resulting maximum-likelihood phylogenetic tree was drawn as a cladogram with the daylight layout and no branch length scaling using the R package ggtree v3.6.2 (Yu et al. 2017). Taxonomy IDs of selected mycobacteria are indicated in brackets and bootstrap support values are depicted in gray as percentages.

Figure 2.

Genomic representation of the fumarate reductase locus in modern (TbD1-deleted) and ancestral (TbD1-intact) M. tuberculosis strains, M. decipiens (ATCC TSD-117), M. lacus (DSM 44577), M. riyadhense (DSM 45176), M. shinjukuense (DSM 45663), M. kansasii (ATCC 12478), and M. marinum (E11). Comparisons were performed using the Artemis Comparison Tool (Carver et al. 2005) and the MicroScope database (Vallenet et al. 2017). Genes surrounding the genomic locus containing the orthologues of rv1556, mmpL6, and mmpS6 in M. lacus, M. riyadhense, and M. shinjukuense were compared to the M. tuberculosis H37Rv genome and percentages of amino acid identities with M. tuberculosis genes determined by the MaGe tool (Vallenet et al. 2006) are indicated.

Figure 3.

(A) Phylogenetic topology of M. canettii strains and members of the MTBC. Genomes of selected M. canettii and MTBC strains were downloaded from the GenBank database and analyzed using PanACoTA v1.4.0 (Perrin and Rocha 2021). Genomes were annotated using Prokka v1.14.5 (Seemann 2014) and their pan-genome was inferred using MMseqs2 v14-7e284 (Steinegger and Söding 2017) based on a minimum sequence identity of 95% at the protein level. Genes conserved across all selected mycobacteria were aligned using MAFFT v7.522 (Katoh and Standley 2013). Neighbor-Net network was computed from pairwise distances estimated with the Jukes and Cantor substitution model (JC69) using the R package phangorn v2.11.1 (Schliep 2011). The resulting unrooted phylogenetic network was drawn using the R packages tanggle v1.4.0 and ggtree v3.6.2 (Yu et al. 2017). (B) Higher magnification of the Neighbor-Net topology of the MTBC from (A). Genomic loss events such as the deletion of RD7-RD8-RD10, RD9, and TbD1 regions are indicated by arrows. The presence of the cobF gene in M. canettii and L8 genomes is indicated as “cobF+”, all other depicted genomes are cobF-deleted. Scale bars represent the number of substitutions per site. NB: We note a discrepancy in the MTBC phylogeny regarding both M. tuberculosis strains T46 and GM 1503, which were originally classified as belonging to the L1 and L4 lineages, respectively, based on selected genes (Hershberg et al. 2008) and whole-genome (GenBank: ACHO00000000.1; ABQG00000000.1) sequencing. More recently released genome sequences for T46 and GM 1503 strains (GenBank: JLCS00000000.1; JLCR00000000.1) were used here and the present topology depicts GM 1503 as part of the lineage L1 and T46 as part of the lineage L4 (denoted with *), which is consistent with the presence of an intact and a deleted TbD1 region in their genome sequence, respectively. We assume that the newer genome versions have been mislabeled in the database and that the original classification of T46 and GM 1503 within the lineages L1 and L4, respectively, is correct.