Horizontal Gene Transfer and Drug Resistance Involving Mycobacterium tuberculosis

Abstract

Mycobacterium tuberculosis (Mtb) acquires drug resistance at a rate comparable to that of bacterial pathogens that replicate much faster and have a higher mutation rate. One explanation for this rapid acquisition of drug resistance in Mtb is that drug resistance may evolve in other fast-replicating mycobacteria and then be transferred to Mtb through horizontal gene transfer (HGT). This paper aims to address three questions. First, does HGT occur between Mtb and other mycobacterial species? Second, what genes after HGT tend to survive in the recipient genome? Third, does HGT contribute to antibiotic resistance in Mtb? I present a conceptual framework for detecting HGT and analyze 39 ribosomal protein genes, 23S and 16S ribosomal RNA genes, as well as several genes targeted by antibiotics against Mtb, from 43 genomes representing all major groups within Mycobacterium. I also included mgtC and the insertion sequence IS6110 that were previously reported to be involved in HGT. The insertion sequence IS6110 shows clearly that the Mtb complex participates in HGT. However, the horizontal transferability of genes depends on gene function, as was previously hypothesized. HGT is not observed in functionally important genes such as ribosomal protein genes, rRNA genes, and other genes chosen as drug targets. This pattern can be explained by differential selection against functionally important and unimportant genes after HGT. Functionally unimportant genes such as IS6110 are not strongly selected against, so HGT events involving such genes are visible. For functionally important genes, a horizontally transferred diverged homologue from a different species may not work as well as the native counterpart, so the HGT event involving such genes is strongly selected against and eliminated, rendering them invisible to us. In short, while HGT involving the Mtb complex occurs, antibiotic resistance in the Mtb complex arose from mutations in those drug-targeted genes within the Mtb complex and was not gained through HGT.

Keywords: antibiotic resistance; horizontal gene transfer; isoniazid; phylogenetic incongruence; refampin.

Figure 1

Phylogenetic methods for detecting horizontal gene transfer and recombination in viral and bacterial genomes. (A) A species tree T_S. (B) A gene tree T_G resulting from species S4 horizontally acquiring the gene from S2. (C) Expected evolutionary distance (${\overline{d}}_{ij}$) between S4 and other species for eight genes (1 to 8), given T_S. The two subscripted i and j in ${\overline{d}}_{ij}$ are species indicators. The observed rRNA distance may fluctuate stochastically above and below $\overline{d}$. (D) Distance $d_{i,4}$ between S4 and other species for the eight genes, given that species S4 gained genes 4 and 5 from species S2. For all species descended from the common ancestor of S2 and S4, $d_{i,4}$ would differ among genes because S4 gained genes 4 and 5 from S2 so T_G from these two genes would differ from T_S. (E) Distance $d_{i,2}$ between S2 and other species for the eight genes. Only $d_{4,2}$ differs among genes, but other $d_{i,2}$ distances do not, because S2 does not change its phylogenetic positions.

Figure 2

Approximation of the species tree with genes encoding ribosomal proteins. Species names are in the format of “GenBank accession”_”Species Name”. (A) Phylogeny from 21 aligned and concatenated RPL (large ribosome protein) genes. (B) Phylogeny from 18 aligned and concatenated RPS (small ribosomal protein) genes. The two trees are highly concordant, with only three species (colored red) differing slightly in their phylogenetic positions. The trees are unrooted. Bootstrap values equal to 1.00 are not shown, but those smaller than 1.00 are next to the node. The green-colored branch lengths, one leading to Mycobacteroides abscessus, and the other separating the Mtb complex from the rest, serve the function of a scale bar. Corresponding clades in (A,B) are shaded in the same color.

Figure 3

TN93 distances between the Mtb H37Rv strain and the other 42 species in Figure 2. The 21 RPL genes (and their aligned lengths) are rplA(720), rplB(843), rplC(702), rplD(735), rplE(600), rplF(540), rplI(462), rplJ(657), rplK(435), rplL(405), rplM(444), rplN(369), rplO(447), rplP(417), rplQ(702), rplR(414), rplS(342), rplT(396), rplU(324), rplV(780), and rplX(327). The 18 genes (and their aligned lengths) are rpsA(1467), rpsB(906), rpsC(873), rpsD(606), rpsE(777), rpsF(291), rpsG(471), rpsH(399), rpsI(543), rpsJ(306), rpsK(444), rpsL(375), rpsM(375), rpsO(270), rpsP(633), rpsQ(456), rpsS(282), rpsT(267). The red arrow points to the distance curve between Mtb H37Rv and Mycobacterium canettii. The distance between Mtb H37Rv and other Mycobacterium variants is effectively zero for all ribosomal proteins.

Figure 4

Genomic integrity as seen from rRNA genes. (A) Phylogeny from concatenated 23S + 16S rRNA genes. DistPlot of the 23S rRNA (B) and 16S rRNA (C) genes between Mtb H37Rv and its close relatives within the G4 clade. The G2 clade is not monophyletic as seen in Figure 2. Species that changed their phylogenetic position relative to tree in Figure 2 are colored red. The branch length to M. abscessus (colored green) is comparable to that in Figure 2, but the branch length (colored green) separating the Mtb complex from the rest is much shorter than that in Figure 2. TN93 distances were calculated over a sliding window of 500 nt.

Figure 5

Phylogenetics of β (A) and β′ (B) subunits of RNA polymerase. Species with phylogenetic position different from those in Figure 2 are colored in red. The branch lengths (colored green) leading to M. abscessus and separating the Mtb complex from the rest were shown.

Figure 6

DistPlots for β (A) and β′ (B) subunits of the RNA polymerase in Mycobacterium species. TN93 distances were calculated between Mtb H37Rv and each of the other species over a sliding window of 500 nt. All members in the Mtb complex were colored red in (B) except for M. canettii, which was colored black.

Figure 7

Phylogenies from inhA (A) and katG (B) genes. The branch lengths (colored green) leading to M. abscessus and separating the Mtb complex from the rest were shown next to the branch. Species with multiple versions of katG are highlighted with different colors, with an appended “_1”, “_2” and “_3” to distinguish different versions of the gene within the same genome. The numbering does not imply relationship.

Figure 8

The phylogeny of mgtC is compatible with three gene duplication events. Species with multiple versions of mgtC are colored and appended with “_1”, “_2” and “_3” to distinguish different versions of the gene within the same genome. The numbering does not imply relationship. The Mtb complex is colored in green and blue. Other species with multiple versions of mgtC are colored red.

Figure 9

Phylogeny of the 16 IS6110 sequences from Mtb H37Rv (numbered from 1 to 16) and two homologues from M. smegmatis (colored red). The Mtb H37Rv sequences numbered 1–8, 10, 12–14 and 16 are identical and represented by one sequence. The tree is midpoint-rooted, with the number 0.5211 indicating the branch length.

Figure 10

A tree with a fast-evolving clade (S1 to S3) and a slow-evolving clade (S4 to S6).