Toward an Internally Consistent Model for Hg(II) Chemical Speciation Calculations in Bacterium–Natural Organic Matter–Low Molecular Mass Thiol Systems

To advance the scientific understanding of bacteria-driven mercury (Hg) transformation processes in natural environments, thermodynamics and kinetics of divalent mercury Hg(II) chemical speciation need to be understood. Based on Hg LIII-edge extended X-ray absorption fine structure (EXAFS) spectroscopic information, combined with competitive ligand exchange (CLE) experiments, we determined Hg(II) structures and thermodynamic constants for Hg(II) complexes formed with thiol functional groups in bacterial cell membranes of two extensively studied Hg(II) methylating bacteria: Geobacter sulfurreducens PCA and Desulfovibrio desulfuricans ND132. The Hg EXAFS data suggest that 5% of the total number of membranethiol functionalities (Mem-RStot = 380 ± 50 μmol g–1 C) are situated closely enough to be involved in a 2-coordinated Hg(Mem-RS)2 structure in Geobacter. The remaining 95% of Mem-RSH is involved in mixed-ligation Hg(II)-complexes, combining either with low molecular mass (LMM) thiols like Cys, Hg(Cys)(Mem-RS), or with neighboring O/N membrane functionalities, Hg(Mem-RSRO). We report log K values for the formation of the structures Hg(Mem-RS)2, Hg(Cys)(Mem-RS), and Hg(Mem-RSRO) to be 39.1 ± 0.2, 38.1 ± 0.1, and 25.6 ± 0.1, respectively, for Geobacter and 39.2 ± 0.2, 38.2 ± 0.1, and 25.7 ± 0.1, respectively, for ND132. Combined with results obtained from previous studies using the same methodology to determine chemical speciation of Hg(II) in the presence of natural organic matter (NOM; Suwannee River DOM) and 15 LMM thiols, an internally consistent thermodynamic data set is created, which we recommend to be used in studies of Hg transformation processes in bacterium–NOM–LMM thiol systems.


X-ray Absorption Spectroscopy Analyses
Sulfur K-edge XANES spectra were collected at Beamline 4B7A in Beĳing Synchrotron Radiation Facilities (BSRF), China. The experiment was conducted in fluorescence mode with a Si (111) double crystal monochromator at ambient temperature under high vacuum (10 −8 -10 −6 mbar). In order to protect the samples from oxidation, the culture media, buffer assays and any solution used were deoxygenated. The preparation processes (cells harvest, membranes extraction) were conducted in the glovebox, and the samples were protected from light by covering the vessels with aluminum foils. The freeze-dried samples were quickly put back into the glovebox for more than 2 h (with vessels lids opened) to build a N 2 (g) atmosphere for the samples. The samples were then stored in a −20°C freezer until analysis. Radiation damage was monitored by comparing successive scans. No radiation damage was observed. High self-absorption effects of several high sulfur concentration samples were observed and these samples were diluted in boron nitride (BN) and measured again.
Scans were taken at the energy range of 2462-2500 eV with a step size of 0.2 eV. Data averaging, normalization, and Gaussian curve deconvolution were conducted using Athena, WinXAS and Microsoft Excel, respectively, S6-S8 following the procedure in Song et al.. Mercury L III -edge EXAFS spectra were collected in fluorescence mode using a four-bounce Si(111) monochromator equipping with a 64-element solid state Ge detector on Beamline I20scanning at Diamond Light Source, U.K.. S10 The X-ray source is derived from a wiggler insertion S-3 device giving a spot size of 400 × 300 µm (ℎ × ) at the sample position. The sample was mounted in a flat PEEK holder, sealed with two Kapton ® foil windows and determined at 77 K in a liquid nitrogen (LN 2 ) cryostat (Optistat DN2, Oxford Instruments). The Hg L III -edge of 12 284 eV was calibrated at the Au L III -edge of 11 919 eV with a gold foil. EXAFS data were collected in steps of 0.3 eV from 12 245 to 12 340 eV. One to three scans were collected and averaged by software Athena. S7 Data were normalized in the energy range 12 200-12 600 eV and background was removed with a 7-or 8-knot spline function over the -range 2.7-13.5 Å −1 (see Table 1). Data were reduced and fitted in Fourier Transformed -space by a first coordination shell model using WinXAS S11 and FEFF-7. S12,S13 Models included Hg S and Hg O/N single paths in the first coordination shell, and the multiple scattering (MS) of Hg S path: four-legged Hg S Hg S Hg and three-legged Hg S S Hg, in agreement with models used for Hg(II)-NOM complexation. S14

Hg L III -Edge EXAFS Determination of Thiols in Bacteria Membranes
As shown in Table 1 and Figure S4, at sufficiently low Hg tot concentration (4 µmol g −1 C) not to saturate Mem-R II S functional groups, Hg(II) forms a two-coordinated, linear complex with Mem-R II S, Hg(Mem-R II S) 2 . This structure require the Mem-R II S functional groups to be close and flexible enough to form the Hg(Mem-R II S) 2 structure. At the increased Hg(II) additions of 28 and 55 µmol g −1 C, Hg EXAFS results suggests a significant contribution form RO/N functionalities in the complexation of Hg(II). Because the log of Hg(II) to RO/N functionalities is expected to be ∼20 orders of magnitude smaller than the Hg(Mem-R II S) 2 complex, this can only be explained by a saturation of the Mem-R II S functional groups. Hg(II) is always at least coordinated by two atoms, this may further suggest a formation of a mixed complex involving Mem-RSH groups and neighboring RO/N functional groups. These Mem-RSH groups are designated Mem-R I S and the mixed complex Hg(Mem-R I SRO). Finally, when all Mem-R I S functional groups are saturated, Hg(II) is forming the Hg(Mem-RO) 2 structure, composed of only RO/N functionalities. The saturated S-4 concentration of Mem-R II S may be calculated using the Hg L III -edge EXAFS results as follows: where 0 denotes the concentration of Hg(Mem-R I SRO) when Mem-RS I is fully saturated by Hg(II).
The Mem-RS I concentration equals to 0 . denotes the concentration of Hg(Mem-RO) 2 when Mem-RO is not fully saturated.

Experimental Losses of Hg(II)
We observed an average loss of ∼20% Hg tot in the CLE experiments ( Figure S2) with Geobacter membrane concentrations of 4-19 mg C L −1 (corresponding to a Hg(II)/Mem-RS tot molar ratios S-5 of 0.07-0.3). Similar to the CLE experiments, we also encountered significant losses of Hg tot (13-64%) in samples subjected to Hg EXAFS experiments (Table 1). Notably, Hg 0 was not detected by EXAFS in any sample. Thus, Hg 0 was likely formed prior to EXAFS measurements, similar to experiences from Hg-NOM studies. S9 As shown in Figure S3, Even if the mechanism of dark, abiotic reduction of Hg(II) at bacterial membranes remains to be studied, we know that phenolic, carboxylic and amide moieties are main functional groups of membranes S19-S21 and our observed increase in Hg ( oxidize Hg(0) to Hg(II), S24 but the oxidation mainly occurred in the spheroplast while the cell wall fragments barely oxidized Hg(0). S24

Form of Thiols
To   ), and the thiol ( SH) and amino groups ( NH + 3 ) protonated; Cysdenotes (S -)CH 2 CH(NH + 3 )COOwith both the carboxyl ( COO -) and thiol ( S -) groups deprotonated. † HgCys means the mix complex of Hg with one RS and one RO/N groups that are in the same Cys molecule. ‡ denotes the membranes samples of Geobacter. The log values for ND132 are 39.2 ± 0.2, 38.2 ± 0.1 and 25.7 ± 0.1, respectively.  Wet denotes wet weight and Dry denotes dry weight. § the average thiol concentration of inner and outer membranes.   Figure S7: Experimentally determined Hg(Cys) 2 concentrations (± SD, = 6) as a function of ND132 membrane concentrations in competitive ligand exchange experiments. Experiments were conducted at pH ∼4.0 and I = 10 mM NaClO 4 by pre-equilibrating 0.5 µM Hg(NO 3 ) 2 with different concentrations of bacterium membrane (1-27 mg C L −1 corresponding to 0.5-9.6 µM of Mem-RS tot ) for 24 h followed by Cys addition to yield 2.0 µM. Dashed black line represents the modeled Hg(Cys) 2 concentration. The model was optimized by R software using the PHREEQC package obtaining a minimum merit-of-fit, ( − ) 2 / 2 , of 5%. S-13