Noncanonical cell-to-cell DNA transfer in Thermus spp. is insensitive to argonaute-mediated interference

Abstract

Horizontal gene transfer drives the rapid evolution of bacterial populations. Classical processes that promote the lateral flow of genetic information are conserved throughout the prokaryotic world. However, some species have nonconserved transfer mechanisms that are not well known. This is the case for the ancient extreme thermophile Thermus thermophilus. In this work, we show that T. thermophilus strains are capable of exchanging large DNA fragments by a novel mechanism that requires cell-to-cell contacts and employs components of the natural transformation machinery. This process facilitates the bidirectional transfer of virtually any DNA locus but favors by 10-fold loci found in the megaplasmid over those in the chromosome. In contrast to naked DNA acquisition by transformation, the system does not activate the recently described DNA-DNA interference mechanism mediated by the prokaryotic Argonaute protein, thus allowing the organism to distinguish between DNA transferred from a mate and exogenous DNA acquired from unknown hosts. This Argonaute-mediated discrimination may be tentatively viewed as a strategy for safe sharing of potentially "useful" traits by the components of a given population of Thermus spp. without increasing the genome sizes of its individuals.

FIG 1

Cell-to-cell DNA transfer in T. thermophilus. Shown is the growth on Hyg-Km double-selective TB plates of dilutions of strain CK1 (Km resistant) mated in a 1:1 ratio with an isogenic Hyg-resistant strain labeled in the megaplasmid (PH1) or the chromosome (CH1). As a control, CK1 was transformed with 200 ng of plasmid pMH::Pnqo::sgfp, and the growth of serially diluted transformants was assessed. The media for matings and transformation experiments were either supplemented with DNase I before mixing (left) or left unsupplemented (right).

FIG 2

Preference for megaplasmid genes. (A) Transfer frequencies were obtained after equal amounts of a strain labeled with the Hyg cassette in the chromosome (CH1) or in the megaplasmid (PH1, PH2) and strain CK1, labeled with the Km marker in the chromosome, were mixed. Frequencies (f) are averages of ratios between colonies grown on selective medium containing Km and Hyg and colonies grown with Km (tra/CK1) in 9 independent experiments. Error bars correspond to the standard deviations of the means. Differences in transfer frequencies between chromosomal (CH1) and plasmid (PH1, PH2) markers were statistically significant, as assessed by t tests (n = 9) for the CK1 × PH1 (P, <0.001) and CK1 × PH2 (P, 0.001) mating pairs versus the CK1 × CH1 mating pair. No statistical difference could be observed between megaplasmid-linked frequencies (PH1 versus PH2 [P, 0.134]). (B) Transfer frequencies were obtained as described above for matings between the Hyg-resistant strain labeled in the chromosome (CH1) and Km-resistant strains labeled either in the chromosome (CK1, CK5 to CK8) or in the megaplasmid (PK1 to PK6). Frequencies are averages of ratios between colonies grown on selective medium containing Km and Hyg and colonies grown with Hyg (tra/CH1) in 9 independent experiments. Error bars correspond to the standard deviations of the means. Note that the transfer frequencies for megaplasmid-linked genes were approximately an order of magnitude higher than those for chromosomal genes.

FIG 3

Competence genes are required for cell-to-cell DNA transfer. Selection plates with Hyg and Km were spotted in mating assays (1:1) between Km-resistant mutants in which the indicated components of the natural competence apparatus were affected (CK11 to CK21) and either of two competence-proficient strains, labeled in the chromosome (CH1) or the megaplasmid (PH1), or between CK11 to CK21 and another competence-deficient mutant (CH4). The results of transformation assays with a plasmid conferring Hyg resistance (pMH) are shown on the right. Note that most mutants are completely unable to take up DNA, whereas in others (e.g., CK16, CK20, and CK21), competence is still detectable. Other competence-defective mutants for which data are not shown here (pilO, pilW, pilM, pilN, and dprA mutants) yielded results similar to those for CK11.

FIG 4

The competence system is required in the recipient but not in the donor cells. (A) A transformation-deficient derivative of strain HB27 (CK11, CH2) or NAR1 (CH3) was mated with the transformation-proficient derivative CK1 (HB27) or CK3 (NAR1). A total of 10⁷ cells were spotted onto selective (Km and Hyg) plates either alone (−) or topped with the indicated mating counterpart or 200 ng of plasmid DNA conferring resistance to Km (+pMK) or Hyg (+pMH). (B) SDS-PAGE of whole membrane proteins from spots 2b (the CK3 × CH3 cross), 3b (the CK3 × CH2 cross), 2c (the CK1 × CH3 cross), and 3c (the CK1 × CH2 cross) of panel A. Note that the protein pattern always corresponds to the competence-proficient strain used in the mate: NAR1 for 2b and 3b, HB27 for 2c and 3c. Lane M, molecular size markers of 97.4, 66.2, 45, and 31 kDa.

FIG 5

The ttAgo interference pathway is not activated by cell-to-cell DNA transfer. (A) Transfer frequencies were obtained after mating of equal cell amounts of Δago strains (the CK22 × CH5 mating pair) as well as ago⁺ strains labeled in the same genes (the CK23 × CH6 mating pair). Parallel transformation of strains CH5 and CH6 with 10 ng of genomic DNA isolated from the respective counterparts (CK22 and CK23) was carried out to show the Ago-mediated interference against high-G+C isogenic lineal DNA acquired by natural competence. Frequencies {ratios of Km- and Hyg-resistant transconjugants to the Hyg-resistant partner [f(tra/Hyg^R)]} are averages of results from 9 independent experiments. Error bars correspond to the standard deviations of the means. Differences in transfer frequencies between Δago and ago⁺ matings were statistically nonsignificant (P, 0.501; n = 9). However, there were significant statistical differences between transformation frequencies (P, <0.001). (B) Transfer frequencies obtained after mating of equal cell amounts of CH5 (Δago) or CH6 (ago⁺) with Km-resistant mutants labeled in different chromosomal loci (CK11, CK16, CK24 to CK29) that were either Δago (shaded bars) or ago⁺ (filled bars). Frequencies are averages of results of 5 independent experiments. Error bars correspond to the standard deviations of the means. Differences in transfer frequencies between ago⁺ and Δago strains were not significant (P, 0.968). Likewise, no significant differences among locus groups or within each locus group could be detected (P, 0.339 and 0.105, respectively).