Age and specific features of vcr
Nucleotide similarity is significantly higher between vcrABC cargo modules than can be expected if it was an orthologous locus present in the Dehalococcoides common ancestor (Figures 3, 4). In all cases vcrABC is located within a syntenic putatively-mobile element, vcr-GI, that is part of a broader class of ssrA-specific mobile elements that appear to be common among Dehalococcoides. In all vcrABC -containing strains except GT, the vcr-GI is located adjacent to the primary site of integration, ssrA, structural evidence that vcr-GIs are among the most recently integrated of the available Dehalococcoides ssrA-GIs. Within phylogenetic branches, integration modules are perfectly identical, except for a large identical deletion in the EV and PM vcr-GIs. The significantly unusual nucleotide signature of vcr-GIs [9, 20], as well as the discordance between the vcrA tree and the corresponding Dehalococcoides strain phylogeny, indicate that vcrABC has not been stably maintained in Dehalococcoides genomes since their divergence. Taken together, these observations suggest recent horizontal acquisition and dissemination of vcrABC across all Dehalococcoides ecotypes by way of a ssrA-specific mobile element with conserved attachment site and integration module.
Because anthropogenic release of chloroethenes into the environment is a relatively recent phenomenon (~100 years ), we are particularly interested in the recent bounds for estimates of the age of the MRCA of these vcrABC sequences as a proxy for their horizontal acquisition by Dehalococcoides. Using our highest estimated rates of mutation and chromosomal replication, the divergence of these vcrABC sequences appears to have occurred 4000 years ago. This value is in flated by the inexplicably high variation within the leader sequence of vcrA. If we remove the vcrA leader sequence from the calculation, the age of divergence decreases to 900 years. However, there is clear signal for positive selection in the remaining vcrAB sequence alignment: all 16 variant positions (15 in vcrA and 1 in vcrB) are predicted to result in amino acid substitutions. If positively selected, these mutations may have accumulated faster than the background rates assumed in our molecular dating calculations. Because the relative increase in substitution rate is unclear and the total information represented by just 16 variant positions is low, we cannot confidently distinguish the divergence of these vcrABC from the first industrial production of chloroethenes. By contrast, our most conservative estimate for the MRCA of contemporary Dehalococcoides strains is 40,000 years ago (ranging as high as 3 Mya, Table 1), long before industrial civilization had a chance to influence the evolution of Dehalococcoides and their streamlined genomes specialized for organohalide respiration.
It is important to note that these molecular dating estimates use the available vcrABC sequences to predict the first horizontal acquisition of vcrABC by Dehalococcoides. This analysis is not meant to predict the age of genesis of the first vinyl chloride reductase. We did not detect partial homology with other rdhA that would suggest vcrA is a chimera resulting from a recent homologous recombination event. Moreover, the existence of an alternate vinyl chloride reductase from strain BAV1, BvcA , that shares deeply branching ancestry with VcrA on a tree of available RdhA , suggests that vinyl chloride reductases have existed for a considerable period of time, just not within strains of Dehalococcoides for which sequence data is currently available. In fact, naturally occurring vinyl chloride has been detected in soils , providing a plausible source of selective pressure to explain the existence of vinyl chloride reductases in nature prior to human pollution. However, we have not identified any candidate lineages as the possible progenitor of vinyl chloride reductases, and we have no way of knowing whether the primary substrate for the ancestral VcrA or BvcA was consistently vinyl chloride, leaving their ancestral history unclear.
The phylogenetic discord between integration modules and their attached vcrABC indicates that homologous recombination - or perhaps a more directed form of 'module swapping' - has recently occurred between vcr-GIs (Figure 3). This additional inter-element recombination may be independent of ssrA-specific integration, but it would still require horizontal transfer so that 2 or more vcr-GIs are collocated within the same cell. Multiple vcr-GI variants have not been detected in the same complete genome. However, we did detect a low-coverage variant in the KB-1 metagenome assembly with 3 corroborating reads that perfectly match a different vcr-GI integration module found in VS, WL, GT, and WBC-2 cultures, providing preliminary evidence of the physical collocation of two vcr-GIs within the KB-1 culture (Additional file 5 Figure S4).
ssrA-GIs appear to be integrative and mobilizable elements
A subset of Dehalococcoides rdhAB were previously implicated in horizontal transfer [31, 32], including the trichloroethene reductase gene, tceAB . Although the selective conditions in chloroethene-contaminated environments favors maintenance of tceAB and vcrABC, the genes implicated in tceAB transfer  share no detectable homology with the ssrA-specific system described in detail here. We hypothesize that these Dehalococcoides ssrA-GIs behave as integrative and mobilizable elements ("IMEs") because they do not appear to encode conjugation, although they share many other features of the broadly defined class of integrative and conjugative elements ("ICEs") . It may be possible that conjugation is encoded by a surprisingly minimal gene set within the integration modules , similar to the small (10.9 kbp) integrating and conjugating element 'pSAM2' of Streptomyces ambofaciens, which requires only a single gene, traSA, for inter-mycelial (conjugal) transfer . Dehalococcoides core genes do include putative pil genes, the functions of which are unclear but may play a role in conjugation. Some strains of Dehalococcoides contain unambiguous prophages, providing an alternative hypothesis for the mechanism of ssrA-GI transfer, via illegitimate packaging of the excised ssrA-GI into a phage capsule. The length of Dehalococcoides ssrA-GIs is within the range of typical phage genomes. However, evidence for a complete prophage is not as ubiquitous among Dehalococcoides as the presence of ssrA-GIs, and there have been no descriptions to date of Dehalococcoides phage that also encode an rdhA, leaving the influence of phage on rdhA evolution unclear. Based on currently available evidence, we hypothesize that Dehalococcoides ssrA-GIs are mobilizable but not conjugating elements that sometimes mobilize adjacent tandem islands but in all cases rely on a host- or phage-encoded system for cell-cell transfer of a transient, presumably circular, intermediate.
Dehalococcoides also contains comEA, and it is unknown if Dehalococcoides is transiently competent for uptake of exogenous DNA. However, transfer via stochastic competence is an unsatisfying explanation, mainly because Dehalococcoides ssrA-GIs appear to lack genes encoding independent replication, and stable non-phage extrachromosomal elements have not been observed in Dehalococcoides [7–9].
Occasionally integrating and conjugating elements do have replicative forms , as in the case of rolling circle replication of pSAM2 in the donor cell . Maphosa et al. recently described a field site in which there were 1 to 2 orders of magnitude more vcrA copies detected than copies of tceA, bvcA, or Dehalococcoides 16S rRNA genes . vcrA was also more abundant than Dehalococcoides 16S rRNA genes in a dechlorinating bioreactor inoculated from the site , suggesting either (1) there exists a vcr-IME that can replicate independently or has integrated within an element that can replicate independently, or (2) they detected a non-Dehalococcoides population that also possesses vcrA, coexisting with a Dehalococcoides population.
It is important to note that, while a conspicuous and common feature, not all Dehalococcoides ssrA-GIs contain an integration module. We identified 15 ssrA-GIs without integration modules, containing a total of 38 rdhA as well as other genes. These might be 'cis-mobilizable elements' that encode neither integration nor transfer, but retain functional attL/attR sites  and are occasionally or constitutively mobilized with adjacent genomic islands through a process known as accretion . In some cases these tandem ssrA-GIs may have been previously mobile but are now fixed in the chromosome. For example, there is a region immediately downstream of the direct repeats furthest from ssrA that is similarly dense in rdhA while also syntenic across Dehalococcoides strains, phylogenetically coherent with whole genome estimates, and devoid of ssrA-GI signatures (Additional file 6 Figure S5); suggesting this region was present in the MRCA of available Dehalococcoides . Some or all of this region may have been acquired originally as an ssrA-GI, but deletion and amelioration has erased evidence of horizontal gene transfer.
Likely Roles within ssrA-GI Integration Modules
The first identified Dehalococcoides ssrA-specific integrase gene (dsiB) (DhcVS_1292) was sequenced following the original identification and characterization of VcrA, and noted for its proximity to vcrA on the chromosome . It is now clear that DhcVS_1292 is part of an integration module in an adjacent downstream ssrA-GI (GI 02 in VS, Figure 1), one of 16 dsiB homologs detected in Dehalococcoides genome sequences. The closest relative to dsiB in the public database is present on a fully-sequenced metagenomic fosmid from a deep (4000 m) ocean subsurface sample (EU016565, Figure 2), within an apparent integration module that also includes homologs to dsiA, parB, mom, and a putative tRNA embedded in mom, as well as an unambiguous ssrA-direct repeat at the homologous attL position embedded in dsiB (Additional file 7 Figure S6). This is especially intriguing in light of the recent sequencing of 32 novel rdhA amplified from various marine subsurface sediments , many of which appear phylogenetically within a major rdhA branch (Cluster I ) that is otherwise populated only by rdhA from Dehalococcoides or Dehalogenimonas. Given this indirect evidence and the large diversity of organohalogens detected in marine systems , it is tempting to speculate that Dehalococcoides plays a role in these settings. However, in the absence of direct observation of Dehalococcoides-like microorganisms in marine (subsurface) settings, this role remains unclear.
A more sensitive database search indicated that DsiB is a structurally similar homolog of CcrB, containing the serine-recombinase-catalytic domain at the N terminus, as well as similar motifs along its ~500 residue length (mean 22% ID, Figure 2A). CcrB specifically integrates/excises the so-called 'Staphylococcus Cassette Chromosome' (SCC ) family of mobile elements that are a vector of antimicrobial resistance (among other phenotypes [58, 59]), with major consequences for hospitals and the greater community [60–63]. CcrB was shown to have DNA-binding and recombination activity for attS of SCC , but SCC integration  and attB-specific excision both required the product of a smaller, co-transcribed serine recombinase gene, ccrA, that does not encode a DNA-binding domain . Similarly, Dehalococcoides integration modules encode on a putative operon a second, smaller serine recombinase, DsiA, that also lacks a detectable DNA-binding motif. Dehalococcoides ssrA-GIs and SCC also share overlapping size ranges and specifically integrate at a non-tRNA, single-copy essential gene. We hypothesize that integration/excision of Dehalococcoides ssrA-GIs occurs in a homologous mechanism to SCC, via DsiB in concert with DsiA, with other integration module elements likely playing a role in regulation of integrase/excisionase activity or modification of the excised element to facilitate transfer or maintenance. Unfortunately, the mode of SCC transfer among Staphylococcus is unclear , and so does not provide additional clues regarding a likely transfer mechanism.
Interestingly, dsiB is always found overlapping attL at its 3' end. A stop codon occurs only upstream of the genomic island, even if that means overlapping substantially with an adjacent genomic island or ssrA itself. Complimentary overlap of ssrA with small open reading frames has been detected in some bacteria with ambiguous implications . It seems unlikely in this instance that the 3' terminal ~70 bp of ssrA also encode a functional region of dsiB on its complementary strand. Accordingly, alignments of DsiB are divergent at this portion of their sequence, both in length and amino acid similarity. The majority of dsiB is upstream of ssrA or its direct repeat, and already comprises the expected length for homologs of ccrB (1600 bp). In addition to a trivial explanation in which dsiB undergoes low-efficiency translation that is variable at the C-terminus, it may be that dsiB is only fully functional when encoded on the circularized element, or alternatively when encoded on the chromosome downstream of an adjacent genomic island containing the requisite 3' gene fragment. In any case, the overlap of dsiB with attP/attL leaves the stop codon of dsiB unclear, and may have functional relevance or affect regulation of dsiB.