ESX-1-Independent Horizontal Gene Transfer by Mycobacterium tuberculosis Complex Strains

Abstract

Current models of horizontal gene transfer (HGT) in mycobacteria are based on "distributive conjugal transfer" (DCT), an HGT type described in the fast-growing, saprophytic model organism Mycobacterium smegmatis, which creates genome mosaicism in resulting strains and depends on an ESX-1 type VII secretion system. In contrast, only few data on interstrain DNA transfer are available for tuberculosis-causing mycobacteria, for which chromosomal DNA transfer between two Mycobacterium canettii strains was reported, a process which, however, was not observed for Mycobacterium tuberculosis strains. Here, we have studied a wide range of human- and animal-adapted members of the Mycobacterium tuberculosis complex (MTBC) using an optimized filter-based mating assay together with three selected strains of M. canettii that acted as DNA recipients. Unlike in previous approaches, we obtained a high yield of thousands of recombinants containing transferred chromosomal DNA fragments from various MTBC donor strains, as confirmed by whole-genome sequence analysis of 38 randomly selected clones. While the genome organizations of the obtained recombinants showed mosaicisms of donor DNA fragments randomly integrated into a recipient genome backbone, reminiscent of those described as being the result of ESX-1-mediated DCT in M. smegmatis, we observed similar transfer efficiencies when ESX-1-deficient donor and/or recipient mutants were used, arguing that in tubercle bacilli, HGT is an ESX-1-independent process. These findings provide new insights into the genetic events driving the pathoevolution of M. tuberculosis and radically change our perception of HGT in mycobacteria, particularly for those species that show recombinogenic population structures despite the natural absence of ESX-1 secretion systems.IMPORTANCE Data on the bacterial sex-mediated impact on mycobacterial evolution are limited. Hence, our results presented here are of importance as they clearly demonstrate the capacity of a wide range of human- and animal-adapted Mycobacterium tuberculosis complex (MTBC) strains to transfer chromosomal DNA to selected strains of Mycobacteriumcanettii Most interestingly, we found that interstrain DNA transfer among tubercle bacilli was not dependent on a functional ESX-1 type VII secretion system, as ESX-1 deletion mutants of potential donor and/or recipient strains yielded numbers of recombinants similar to those of their respective parental strains. These results argue that HGT in tubercle bacilli is organized in a way different from that of the most widely studied Mycobacterium smegmatis model, a finding that is also relevant beyond tubercle bacilli, given that many mycobacteria, like, for example, Mycobacterium avium or Mycobacterium abscessus, are naturally devoid of an ESX-1 secretion system but show recombinogenic, mosaic-like genomic population structures.

Keywords: DNA transfer; ESX-1; Mycobacterium canettii; Mycobacterium tuberculosis; conjugation; recombinant.

FIG 1

Optimized mating assay. (A) Example of results from the optimized mating assay showing a large number of double-resistant colonies obtained from a donor, carrying a hygromycin resistance cassette on integrated plasmid pYUB412, and a recipient, carrying a kanamycin resistance cassette on episomal plasmid pMRF1-dsRed, coculture as well as the absence of colonies when the monoculture of the recipient was plated onto 7H11 plates containing kanamycin and hygromycin. PCR confirmed the presence of the two antibiotic markers in 10 randomly selected recombinants as well as donor and recipient strains. (B) Number of recovered recombinants resulting from mating assays with STB-A, M. bovis AF2122/97, and M. africanum 65 donor strains and the STB-L recipient strain when cultures were grown in 7H9 medium with 10% ADC and 0.05% Tween 80, or 7H9 medium with 10% ADC, 0.05% Tween 80 and 0.2% glycerol prior to the assays. Note that M. bovis AF2122/97 and M. africanum 65 cultures were also routinely supplemented with 0.2% pyruvate. Bars represent means ± standard deviations (SD).

FIG 2

Circular representation of genomes from recombinants. The genomes of recombinant 1 (RC1) and recombinant 2 (RC2) shown in the top row were obtained when M. africanum 65 was used as the donor strain together with M. canettii STB-L as the recipient strain (M. africanum 65/STB-L RC1 and M. africanum 65/STB-L RC2). Similarly, the genomes of RC1 and RC2 shown in the bottom row were obtained when M. bovis AF2122/97 was used as the donor strain together with M. canettii STB-L as the recipient strain (M. bovis/STB-L RC1 and M. bovis/STB-L RC2). From the outer to the inner circle are (i) the density of detected variants calculated in 5-kb nonoverlapping windows between the recombinant and donor strains (orange), (ii) the donor strain reference genome (red), (iii) the best-scoring hits identified between the recombinant and donor strains (light red), (iv) the assembled recombinant genome (red, region assigned to the donor strain; blue, region assigned to the recipient strain; white, region of unknown origin), (v) the best-scoring hits identified between the recombinant and recipient strains (light blue), (vi) the recipient strain reference genome (blue), and (vii) the density of detected variants calculated in 5-kb nonoverlapping windows between the recombinant and recipient strains (purple). Black bars correspond to gap regions. Coordinates are indicated in megabases. attB, L5 integration site. Note that a considerable portion of the transferred sequences is usually localized in the proximity of the attB L5 integration site because at this site of the donor genome, vector pYUB412 is integrated, which carries an antibiotic resistance marker used for the selection of double-antibiotic-resistant recombinants.

FIG 3

Ability of selected slow-growing mycobacteria to act as donor strains in chromosomal DNA transfer. Numbers of recovered recombinants in mating assays with STB-L as the recipient strain are shown. At least two independent mating assays were performed per mating pair. Reproducibly, no double-antibiotic-resistant colonies were recovered when plating the recipient monoculture. Spontaneous kanamycin-resistant donor colonies, if any, were distinguished by the lack of DsRed production. Bars represent means ± SD.

FIG 4

Schematic representation of the esx-1 locus in different mutant strains used in mating assays. M. canettii strains STB-A ΔeccD₁, STB-A ΔRD1, STB-D ΔeccD₁, and STB-L ΔeccD₁ contain the gene conferring resistance to zeocin (represented by a full black arrow), replacing the entire coding sequence of gene eccD₁. In the case of STB-A ΔRD1, there is an additional deletion between genomic coordinates 4410645 and 4424524 of the STB-A reference genome, spanning genes espE to espI. An additional deletion upstream of eccD₁ is also found in STB-L ΔeccD₁ starting from position 220 of the esxA open reading frame (ORF) and ending at position 1364 of the espI ORF. STB-K ΔeccD₁ has the zeocin cassette inserted between positions 1247 and 1256 of the eccD₁ ORF. Each of the described deletions was confirmed by WGS, which also served to confirm the absence of any other mutations in these strains, which were then used in different mating assays.

FIG 5

DNA transfer efficiency in different ESX-1 mutant strains and the corresponding WT strains. (A) Donor ESX-1 mutant strains with STB-L as the recipient strain. The transfer efficiency is expressed as the number of recombinants per recovered donor cell. (B) STB-L ΔeccD₁ recipient strain with M. bovis AF2122/97 as the donor strain. The transfer efficiency is expressed as the number of recombinants per recovered recipient cell. (C) STB-K ΔeccD₁ donor strain with the STB-L ΔeccD₁ recipient strain. The transfer efficiency is expressed as the number of recombinants per recovered donor cell. Bars represent means ± standard errors of the means (SEM). Statistical analysis was performed using a Kruskal-Wallis test followed by Dunn’s multiple-comparison test when three or more groups were compared or a Mann-Whitney test when two groups were compared.

FIG 6

Examples of Artemis Comparison Tool (ACT) visualization of variants detected between recombinants (RC) and donor and recipient strains. SNPs that differ between genomes are represented by red and indels by blue lines. (A) BCG Pasteur/STB-L RC2, where sequences flanking the pYUB412 integration site were transferred from the donor strain. (B) Example of a region of BCG Pasteur/STB-L RC2 depicting microcomplexity (green dotted frame). (C and D) BCG Pasteur/STB-L RC1, where no apparent donor-derived segments are found flanking the pYUB412 sequence (C), and BCG Pasteur/STB-L RC3, where a chromosomal DNA transfer-related RD5 deletion had occurred (D). PCR confirmed the RD5 deletion in the recombinant (primers flanking the RD5 region are marked with purple arrows, and a primer inside the RD5 region is marked with a yellow arrow).

FIG 7

Transfer of large, specific genomic regions and their phenotypic consequences. (A) Analysis of PE_PGRS proteins in whole-cell lysates and culture filtrates of the BCG Pasteur donor, the STB-L recipient, BCG Pasteur/STB-L RC3, and M. tuberculosis H37Rv by immunodetection using anti-PE_PGRS antibodies. Due to the absence of ppe38-ppe71, the BCG Pasteur donor strain as well as the recombinant strain BCG Pasteur/STB-L RC3 do not secrete PE_PGRS proteins. SigA was used as a loading and cell integrity control. (B) Schematic representation of the type III-A CRISPR-Cas system found in M. bovis and the type I-C CRISPR-Cas system found in STB-L. PCR analysis of the presence of signature cas genes of the respective CRISPR-Cas systems (cas6 and cas10 for type III-A and cas3 for type I-C) in the M. bovis AF2122/97 donor, the STB-L recipient, and M. bovis/STB-L RC1 was performed. Black bars depict the regions amplified by PCR.