Microbial DNA records historical delivery of anthropogenic mercury

Abstract

Mercury (Hg) is an anthropogenic pollutant that is toxic to wildlife and humans, but the response of remote ecosystems to globally distributed Hg is elusive. Here, we use DNA extracted from a dated sediment core to infer the response of microbes to historical Hg delivery. We observe a significant association between the mercuric reductase gene (merA) phylogeny and the timing of Hg deposition. Using relaxed molecular clock models, we show a significant increase in the scaled effective population size of the merA gene beginning ~200 years ago, coinciding with the Industrial Revolution and a coincident strong signal for positive selection acting on residues in the terminal region of the mercuric reductase. This rapid evolutionary response of microbes to changes in the delivery of anthropogenic Hg indicates that microbial genomes record ecosystem response to pollutant deposition in remote regions.

Figure 1

Depth profiles within sediment cores of three lakes. Depth profiles of total Hg concentrations, merA, merT, merP and glnA gene copy numbers collected in Aquatuk (a), Hawley (b) and North Raft (c) Lakes in Northern Ontario. Coloured boxes in panel (a) (red: 2010, green: ca. 1900 and blue: prior to 1800) correspond to the three depths chosen for rpoB and merA gene fragments sequencing.

Figure 2

Model selection and paleo-demographic reconstructions. Models are ranked from top to bottom by increasing estimates of marginal log probabilities of each model, based on two runs; error bars: 2 standard deviations. Compared models are: All.equal.const: all three depths are contemporaneous (equal prior dates) with constant population size (demography); Two.dates.const: two prior dates (0 for superficial samples; 500 y for all other samples), constant demography; Three.dates: three prior dates (0 for superficial samples; 125 y for intermediate depth; trailing number is prior age for deepest samples); sky: skyline demographic model. Bayesian skyline demographic reconstruction under the best model is shown in the inset; dashed vertical line in inset shows timing of the Industrial Revolution, ca. 200 years ago.

Figure 3

MerA and rpoB gene phylogenies. Example of phylogenetic reconstructions, showing the nonlinear relationship between sediment core depth and divergence times for merA (a) and rpoB (b) genes in Aquatuk lake sediments. The evolution of merA sequences is highly contingent: only a small fraction of sequences persist through time. Coloured lines (red: 2010, green: ca. 1900 and blue: prior to 1800) correspond to the three depths chosen for rpoB and merA gene fragments sequencing.

Figure 4

Evidence for selection on the mercuric reductase N-terminus. Location of site A375L that shows strongest signal of positive selection. (a) Three-dimensional model (PDB identifier: 4K7Z; chosen by SwissModel); secondary structure are indicated (green: β sheets; red: α helices), as well as position 375; the left and right structures are identical, but views differ by a rotation of 180 degrees; the region sequenced in this study is circled. (b) This position is shown in a subset of the protein sequence alignment for deepest (*DEP*), middle (*MID*) and surface (*SUR*) sequences; this position changes from an A to a T and to an L with time; AQSUR12 and AQSUR8 are among the three sequences where no positive selection could be detected (Supplementary Figure S3,Supplementary Information). (c) The identified Pfam domain is involved in the dimerisation of the MerA protein, here represented in its totality with its two Pfam domains, PF07992 and PF02852.