Comparative and functional genomic analysis of foreign DNA defense mechanisms in Enterococcus faecium

Abstract

Enterococcus faecium is notoriously difficult to study genetically due to the poor understanding of barriers preventing foreign DNA uptake, such as the proteins that modify type I restriction modification (RM) system activity. Here, we compared E. faecium repertoires of the HsdS specificity subunit (dictating the DNA motif that is adenine methylated) from type I RM systems among 805 globally reported E. faecium isolates. We showed there were eight distinct HsdS types, with four dominant variants that were also significantly enriched in the hospital-associated clade A1 E. faecium lineage. Adenine methylome analysis of a subset of eight representative E. faecium strains revealed that only two exhibited functional type I RM systems, with the activity corroborated by the construction of type I RM deletion mutants. To investigate this surprising finding, we assessed the contribution of the anti-restriction protein ArdA that specifically inhibits type I RM function. The E. faecium ST796 clinical isolate AUS0233 has one intact type I RM system, no adenine methylation, and two distinct ardA paralogs. When heterologously expressed in Staphylococcus aureus JE2, both E. faecium ardA variants were functional, each inhibiting the function of the two type I RM systems in S. aureus. However, the deletion of one or both versions of ardA in E. faecium AUS0233 did not change the transformation efficiency with exogenous DNA, suggesting ArdA in E. faecium AUS0233 is not controlling type I RM. This study highlights the complexity of DNA defense mechanisms in E. faecium and suggests that unidentified factors control the acquisition of foreign DNA.IMPORTANCEEnterococcus faecium has mechanisms of DNA methylation and targeted DNA degradation (called restriction modification [RM]) that hinder foreign DNA uptake, thus influencing the acquisition of important phenotypes such as antibiotic resistance. Restriction barriers also frustrate efforts for laboratory genetic manipulation used to study this pathogen. From PacBio analysis of E. faecium strains, it was observed that the majority of E. faecium do not adenine methylate DNA despite genome analysis indicating they have intact type I RM methylation systems. One explanation for this observation is that E. faecium produces anti-restriction factors such as ArdA, which can inhibit type I RM systems. However, the deletion of both ardA alleles did not improve the efficiency of DNA uptake. These findings build our foundational knowledge of how E. faecium controls foreign DNA and show there is additional complexity surrounding these systems to be discovered.

Keywords: DNA defense mechanisms; Enterococcus faecium; genomics; hsdS; restriction modification system.

Fig 1

Global core-single nucleotide polymorphism (SNP) based phylogeny of 805 E. faecium strains. The 805 strains are representative of 121 different MLSTs across clade B (dark blue), clade A2 (blue), and clade A1 (light blue). Sequence types ST80 (light orange), ST203 (orange), ST796 (red), and ST1421 (yellow) are colored. The 20 test panel strains are highlighted on the tree by stars, and the letter code (A to T) is linked to the test panel strain ID in Table S2. The phylogenetic positions of eight PacBio-sequenced isolates are overlaid onto the phylogenetic tree. The large bubble represents the chromosome, and the smaller surrounding bubbles illustrate the plasmid content of each strain, with the size being a semi-quantitative depiction of plasmid size. This tree was recombination filtered using Clonal Frame ML, and nodes with less than 70% bootstrap support were collapsed. Branch distance is shown as SNP per site.

Fig 2

Examination of type RM systems found in the 20-strain E. faecium panel. (A) A panel of 20 clinical E. faecium strains was transformed with plasmid pIMC8(Pgap-YFP) (1 µg), with the optimal glycine percentage used shown. E. faecalis strain V583 is shown as a comparator. The data is from three independent batches of competent cells transformed, with the mean and standard deviation (SD) shown. (B) Rooted core-SNP-based phylogenetic tree of VRE strains with the presence of HsdS alleles shown, purple (HsdS_1), green (HsdS_2), yellow (HsdS_3), and red (HsdS_4), (T) indicates a truncated HsdS protein (Table S4). The type IIG system is shown in pink. (C) Genomic organization of RM systems and methylation profiles of the eight PacBio sequenced strains. The methylated adenine residue in the methylation profile is highlighted in bold, with the fraction proceeding representing the adenine methylated versus the total number of sites across the genome. Genes are not drawn to scale.

Fig 3

Transformation efficiency of type I RM mutants. The wild-type and isogenic mutant were electroporated with 1 µg of pIMC8(Pgap-YFP) isolated from Escherichia coli DC10B. The graph is representative of three independent transformations from different batches of competent cells, with the mean and SD shown.

Fig 4

(A) Global representation of the four main hsdS alleles: purple (HsdS_1), yellow (HsdS_2), green (HsdS_3), and red (HsdS_4). The pie charts represent the proportion of the four main types of HsdS for each country. Shown are the eight countries (out of 32) that had >20 strains from the clade A1 681 E. faecium genomes. Pie charts are size-scaled based on the number of strains present in the study. (B) Proportion of hsdS allele carriage in 805 global E. faecium strains. The strains are from clade B, clade A2, and clade A1, with HsdS subunits 1–8. These data are representative of 38 clade B strains, 86 clade A2 strains, and 681 clade A1 strains.

Fig 5

Impact of ArdA on the transformation of E. faecium AUS0233 and S. aureus JE2. (A) The ardA1 and/or the ardA2 genes were deleted from AUS0233, and the transformation efficiency with plasmid pIMC8(Pgap-YFP) was measured. (B) Role of ardA1 or (C) ardA2 on the inhibition of type I restriction in S. aureus. S. aureus JE2 was transformed with the tetracycline-inducible vector pRAB11 (empty) or pRAB11 containing (B) ArdA1 or (C) ArdA2. Electrocompetent cells of JE2+pRAB11 or JE2+pRAB11(ardA1 or ardA2) were made in the presence of anhydrotetracycline (aTc, 100 ng/mL), and the transformation efficiency was determined with plasmid (pCN34*RBS) extracted from either DC10B (no adenine methylation) or E. coli IM08B (JE2 type I adenine methylation profile). These data are from three independent batches of competent cells transformed, with the mean and standard error of the mean shown.