Bacterial natural transformation by highly fragmented and damaged DNA

Abstract

DNA molecules are continuously released through decomposition of organic matter and are ubiquitous in most environments. Such DNA becomes fragmented and damaged (often <100 bp) and may persist in the environment for more than half a million years. Fragmented DNA is recognized as nutrient source for microbes, but not as potential substrate for bacterial evolution. Here, we show that fragmented DNA molecules (≥ 20 bp) that additionally may contain abasic sites, cross-links, or miscoding lesions are acquired by the environmental bacterium Acinetobacter baylyi through natural transformation. With uptake of DNA from a 43,000-y-old woolly mammoth bone, we further demonstrate that such natural transformation events include ancient DNA molecules. We find that the DNA recombination is RecA recombinase independent and is directly linked to DNA replication. We show that the adjacent nucleotide variations generated by uptake of short DNA fragments escape mismatch repair. Moreover, double-nucleotide polymorphisms appear more common among genomes of transformable than nontransformable bacteria. Our findings reveal that short and damaged, including truly ancient, DNA molecules, which are present in large quantities in the environment, can be acquired by bacteria through natural transformation. Our findings open for the possibility that natural genetic exchange can occur with DNA up to several hundreds of thousands years old.

Keywords: DNA degradation; anachronistic evolution; early life; horizontal gene transfer; microbial evolution.

Fig. 1.

Natural transformation of A. baylyi by short DNA. (A) Chromosomal location of the single-nucleotide substitution marker (trpE27; SI Text) with sequence detail of the point mutation locus (bold) and proportional sizes of the donor DNA substrates (containing the wild-type residue G at the mutation locus). (B) Transformation efficiencies of wild type (circles; n = 3–7) and ∆recJ ∆exoX (triangles; n = 3–7; 26 for the 60-mer) calculated as transformants per marker-containing molecule with 100 ng/mL donor DNA of different lengths. (C) Transformation frequencies (mean with SD from three or more experiments) obtained with 100 ng/mL donor DNA of different lengths and calculated as transformants per recipient. Strains: wild type [circles; n = 3–7; frequencies were always higher than the mutational background (ANOVA: P = 0.033) determined in control experiments without DNA], ∆recA (squares; n = 3), and ∆recJ ∆exoX (triangles; n = 3–7; 26 for the 60-mer). The solid and dashed horizontal lines mark the background mutation frequency of the wild-type and ∆recJ ∆exoX strains, respectively. See also Table S1.

Fig. 2.

Natural transformation by damage-containing DNA. (A) Sequence details of end modifications, internal lesions, uracil (U)- and AP site (X)-containing donor DNA substrates. Position of the marker nucleotide is indicated by the dashed line. (B and C) Transformation frequencies of the ∆recJ ∆exoX and ∆recJ ∆exoX ∆nth (BER endonuclease III-deficient) strains (n = 3–7; 26 for the Oli-60 with ∆recJ ∆exoX) with DNA substrates shown in A. Transformation frequencies were calculated as in Fig. 1C. (D) Chromosomal location and sequence detail of the detection construct for mammoth mtDNA (hisC::′ND5i′ with double stop codons and nucleotide variations in bold), and sequence details of mammoth and human mtDNA (additional nucleotide polymorphisms are underlined). In all transformants obtained in control experiments with human mtDNA substrates, the boxed sequence was present. See also SI Text. (E) Diagram of the ancient DNA experiment. Woolly mammoth DNA was used as donor DNA for natural transformation of the hisC::′ND5i′ strain.

Fig. 3.

Recombination with very short DNA. (A and B) Donor DNA substrate illustrations: the top strand is depicted in red, and the bottom strand in blue. (A) Heteroduplex DNA with sequence detail and proposed integration mechanism. A bottom strand fragment can anneal with the discontinuously replicated strand and incorporate into the lagging strand at the replication fork. One hundred percent (n = 20) of wild-type transformants and >97% (n = 38) of ∆recJ ∆exoX transformants had the bottom strand integrated. (B) Schematic proportional sizes of double- and single-strand donor DNA substrates. Position of the marker nucleotide is indicated by the dashed line. (C) Transformation frequencies of the ∆recJ ∆exoX strain (n = 3–7; 26 for Oli-60) with various donor DNA substrates, and mutational background without DNA. Transformation frequencies were calculated as in Fig. 1C.

Fig. 4.

Double-nucleotide variants and polymorphism genome analysis. (A) Sequence details of nucleotide variation-containing donor DNA substrates (nucleotide exchanges in red). The dashed line indicates the position of the marker nucleotide. (B) Transformation frequencies of the ∆recJ ∆exoX, wild-type, and ∆recJ ∆exoX ∆mutS strains (n = 3–7; 26 for Oli-60 with ∆recJ ∆exoX) obtained with DNA substrates shown in A. Transformation frequencies were calculated as in Fig. 1C. (C–E) Distribution of DNPs, multiples, and SNPs in transformable and nontransformable bacterial species. (C) Bacterial genomes used for analysis. (D) Total polymorphism counts in transformable and nontransformable species. (E) Normalization of counts for direct visual comparison of the proportions of total polymorphisms of DNPs, multiples, and SNPs within nontransformable and transformable species, respectively.