Common physiological processes control mercury reduction during photosynthesis and fermentation

Mercury (Hg) is a global pollutant and potent neurotoxin that bioaccumulates in food webs as monomethylmercury (MeHg). The production of MeHg is driven by anaerobic and Hg redox cycling pathways such as Hg reduction, which control the availability of Hg to methylators. Anaerobes play an important role in Hg reduction in methylation hotspots, yet their contributions remain underappreciated due to how challenging these pathways are to study in the absence of dedicated genetic targets and low levels of Hg0 in anoxic environments. In this study we used Hg stable isotope fractionation to explore Hg reduction during anoxygenic photosynthesis and fermentation in the model anaerobe Heliobacterium modesticaldum Ice1. We show that cells preferentially reduce lighter Hg isotopes in both metabolisms leading to mass-dependent fractionation, but mass-independent fractionation commonly induced by UV-visible light is absent. We show that isotope fractionation is affected by the interplay between pathways controlling Hg recruitment, accessibility, and availability alongside metabolic redox reactions. The combined contributions of these processes lead to isotopic enrichment during anoxygenic photosynthesis that is in between the values reported for anaerobic respiratory microbial Hg reduction and abiotic photoreduction. Isotope enrichment during fermentation is closer to what has been observed in aerobic bacteria that reduce Hg through dedicated detoxification pathways. These results demonstrate that common controls exist at the atomic level for Hg reduction during photosynthesis and fermentation in H. modesticaldum. Our work suggests that similar controls likely underpin diverse microbe-mediated Hg transformations that affect Hg’s fate in oxic and anoxic habitats. IMPORTANCE Anaerobic and photosynthetic bacteria that reduce mercury affect mercury delivery to microbes in methylation sites that drive bioaccumulation in food webs. Anaerobic mercury reduction pathways remain underappreciated in the current view of the global mercury cycle because they are challenging to study, bearing no dedicated genetic targets to establish physiological mechanisms. In this study we used stable isotopes to show that common physiological processes control mercury reduction during photosynthesis and fermentation in the model anaerobe Heliobacterium modesticaldum Ice1. The sensitivity of isotope analyses highlighted the subtle contribution of mercury uptake towards the isotope signature associated with anaerobic mercury reduction. When considered alongside the isotope signatures associated with microbial pathways for which genetic determinants have been identified, our findings underscore the narrow range of isotope enrichment that is characteristic of microbial mercury transformations. This suggests that there exist common atomic-level controls for biological mercury transformations across a broad range of geochemical conditions.

are challenging to study, bearing no dedicated genetic targets to establish physiological 48 mechanisms. In this study we used stable isotopes to show that common physiological processes 49 control mercury reduction during photosynthesis and fermentation in the model anaerobe 50 Heliobacterium modesticaldum Ice1. The sensitivity of isotope analyses highlighted the subtle 51 contribution of mercury uptake towards the isotope signature associated with anaerobic mercury 52 reduction. When considered alongside the isotope signatures associated with microbial pathways 53 for which genetic determinants have been identified, our findings underscore the narrow range of 54 isotope enrichment that is characteristic of microbial mercury transformations. This suggests that 55 there exist common atomic-level controls for biological mercury transformations across a broad 56 range of geochemical conditions. 57 58 INTRODUCTION 59 Mercury (Hg) is a global pollutant and potent neurotoxin that bioaccumulates in aquatic and 60 terrestrial food webs as monomethylmercury (MeHg) (1). Anaerobic microbes with the hgcAB 61 gene cluster, which encodes for the metabolic machinery responsible for Hg methylation (2), 62 contribute to MeHg production in habitats such as aquatic sediments, water columns, and 63 flooded soils (3-6). Hg methylation is thought to occur intracellularly and as such, it is ultimately 64 controlled by the bioavailability of Hg to methylators in anoxic habitats (7). 65 Hg redox cycling plays a key role in determining the inorganic Hg substrates available to 66 methylators. Its role is two-fold: anaerobic Hg 0 oxidation can supply dissolved Hg II required to 67 generate MeHg and evasion of Hg 0 due to reduction can limit MeHg production by removing 68 inorganic Hg substrates. Although it is well established that abiotic photochemical Hg reduction 69 dominates Hg redox cycling in oxic surface systems where light is present (8-16), Hg reduction 70 pathways in anoxic zones where Hg methylation occurs have mostly been characterized under 71 laboratory conditions and remain challenging to assess in the field. 72 Anoxic Hg reduction can occur via abiotic redox reactions with dissolved organic carbon 73 (17) and magnetite (18,19), but also through biotic pathways mediated by chemotrophic and 74 phototrophic microorganisms (20-24). Unlike many aerobes that reduce Hg using dedicated 75 enzymatic machinery encoded by the mer operon (25, 26), Hg reduction in anaerobes occurs 76 through cometabolic pathways tied to anaerobic respiration, fermentation, and anoxygenic 77 photosynthesis (20-24). In contrast to mer-mediated Hg reduction and demethylation, dedicated 78 genetic determinants have yet to be identified for anaerobic Hg reduction, making these 79 pathways challenging to study from a mechanistic standpoint. Furthermore, while some studies 80 have shown that Hg 0 can significantly contribute to Hg speciation in anoxic habitats (27, 28), it is 81 8 Total Hg analyses and sample preparation for stable isotope measurements 150 All samples for aqueous Hg analyses were oxidized with 10 % (v/v) bromine chloride (BrCl) in 151 line with previous work (40) and THg concentrations were determined by stannous chloride 152 reduction coupled to cold vapor atomic fluorescence spectrometry (42). During THg analysis, 153 duplicates and matrix standard spikes were analyzed every 10 samples; analyses showed less 154 than 10 % relative percent difference between duplicates and spike recoveries of 90-110%. XXX is used to signify the isotope of interest. Hg also undergoes MIF of both even and odd 185 isotopes. Here, odd-MIF is described by Δ 199 Hg and even-MIF by Δ 200 Hg. MIF is calculated as: 186 Δ xxx Hg ≈ δ xxx Hg -(δ 202 Hg * β). (Eq 2) 187 β denotes the mass dependent scaling constant, which is determined by the laws of mass 188 dependence (45). Isotope enrichment effects based on the ratio of products to reactants (i.e. ε p/r, 189 herein referred to as εAP for anoxygenic phototrophic reduction and εFM for fermentative 190 reduction) were calculated using Rayleigh fractionation models to account for the Hg 0 that was 191 removed following methods from previous work on microbial Hg reduction (34). 192 193 This method involved fitting a linear regression (shown in Fig S3, Fig S4, and Table S3) based 194 on the change in relative isotope ratios, ln (R/R0), as a function of ln(fr) where: 195 R = (δ 202 Hg/1000) +1 for Hg II (reactant) or Hg 0 (product) at a given time point and 196 R0 = (δ 202 Hg/1000) +1 at the beginning of the experiment (Eq 3); 197 fr= Hg II in the reactor at a given time / Hg II in the reactor at the start of the experiment (Eq 4); 198 and the slope of the linear regression, which is equivalent to εAP or εFM: 199 ε = ln (R/R0)/ ln(fr) (Eq 5).

200
We chose to use the total Hg analyses from the bioreactor to establish fr as this is a true 201 representation of the instantaneous isotope fractionation rather than a time integrated sample 202 following the methods outlined in previous work (34). Note that in our slope calculations we do 203 not have data for 202 Hg 0 when fr=1 (the 0h sampling point) because no isotope fractionation was 204 taking place prior to the start of Hg reduction (

217
Hg 0 production during anoxygenic photosynthesis and fermentation 218 In our experiments Hg 0 production relied on the presence of live cells for both chemotrophic and 219 phototrophic growth conditions. Cumulative Hg 0 production was an order of magnitude higher 220 for live cell treatments compared to abiotic controls (>0.60 vs <0.05 nmol) (Fig 1). 221 Chemotrophic cells produced double the amount of Hg 0 compared to phototrophic cells (1.33 to 222 1.64 nmol vs 0.60 to 0.81 nmol, respectively) despite having been supplied with slightly less Hg II 223 (Fig 1B, E). In relative terms, phototrophic cells reduced 12 to 15% of the initial Hg II supplied 224 whereas chemotrophic cells reduced 31 to 40% (Fig S2). 225 Relative Hg 0 production in these experiments was lower, yet consistent with our previous 226 work, which showed chemotrophically-grown cells reduced 15% more Hg II than 227 phototrophically-grown cells, reducing between 60 to 75% of the initial Hg II supplied (23). The 228 relative amount of Hg 0 production did not increase with higher Hg II exposure (10 nM) versus 229 previous experiments (250 pM) (23). It is unlikely that the lower reduction rate observed in the 230 experiments presented here was due to the toxicity of 10 nM Hg supplied, given that H. 231 modesticaldum grows well under the same conditions at 50 times higher Hg concentrations (23). 232 Instead, we suspect that the lack of constant bubbling of the reactor may be responsible for the 233 observed results. It is possible that the accumulation of CO2 in the bioreactor following organic 234 carbon oxidation may compete with Hg II as an electron sink, decreasing the amount of Hg 0 235 produced. Inorganic carbon is required for anaplerotic reactions that fulfill the biosynthetic needs 236 of H. modesticaldum and CO2 could have competed with Hg II for reducing power, as previously 237 shown (47). These results could also be due to Hg 0 oxidation allowed by the increased residence Despite carefully controlling the growth conditions in our experiments, chemotrophic 249 cells from replicate #1 exhibited a lag phase of 12 hours (Fig 1F). Although this lag did not 250 affect final cell density, the slow-growing culture from replicate #1 initiated Hg 0 production 251 earlier than the culture in replicate #2 (Fig 1E, F). These results may be attributable to cells in 252 replicate #1 maintaining a higher Hg to cell ratio in the first 12 hours of the experiment, which 253 suggests that Hg is being reduced intracellularly in line with our previous work on Heliobacteria 254 (23, 50). Indeed, even at low initial densities, cells can maintain µM levels of intracellular 255 reducing power, which could be used to reduce nM levels of Hg II (24). 256 257

Mass dependent fractionation during photosynthetic and fermentative growth 258
Hg reduction in phototrophically grown H. modesticaldum resulted in consistent positive MDF 259 whereas fermentatively grown cultures showed positive MDF with variability in fractionation 260 patterns across replicates (Fig 2). Abiotic controls showed minimal fractionation for both MDF 261 and MIF, confirming that Hg reduction and subsequent isotopic fractionation were 262 predominantly driven by cellular processes (Fig 2). Live cell experiments were devoid of MIF 263 (Table S1 and Table S2). 264 For phototrophically-grown cultures, the reactant pool δ 202 Hg II increased steadily over time 265 suggesting cells preferentially reduced lighter Hg II (Fig 2A). A mirror trend was observed for 266 δ 202 Hg 0 , which was depleted in 202 Hg 0 at the beginning of the experiment, became enriched with 267 heavier isotopes as the reaction proceeded, and eventually approached the initial isotopic 268 composition of the NIST 3133 standard (0‰) (Fig 2B). Curiously, the δ 202 Hg 0 for fr = 0.94 at 3h in the two phototrophic live cell replicates showed 279 that the product pool was initially enriched with the 202 Hg 0 isotope (Fig 2B). A similar pattern 280 has been observed in iron reducing bacteria that preferentially reduced heavier Fe III (51) and the 281 model anaerobic Hg methylator and reducer Geobacter sulfurreducens PCA (37, 40). The most 282 recent work on G. sulfurreducens has shown that uptake of Hg selects for lighter Hg isotopes but 283 cells can also access an isotopically heavier pool of Hg II from the dissolved phase (37). It is 284 possible that a similar process is occurring for phototrophically-grown H. modesticaldum 285 wherein cells access an isotopically heavier pool of Hg during equilibrium binding of Hg II to the 286 outside of the cell, followed by an alternative uptake process that selects for lighter pools of Hg. 287 Based on these results, we included all time points in isotopic enrichment calculations for Hg II 288 but omitted the 3h time point in calculations for Hg 0 given that other cellular fractionation 289 processes could be occurring (Fig S3, Fig S4, and Table S3). 290 Positive MDF was also observed for chemotrophic cultures but fractionation patterns differed 291 strongly between replicates. Replicate #1 showed an initial depletion in reactant pool δ 202 Hg II and 292 became progressively enriched with 202 Hg II relative to the NIST 3133 suggesting lighter isotopes 293 were still preferentially reduced (Fig 2C). Note that the second δ 202 Hg II data point available for 294 replicate #1 occurs at fr=0.72 due to the rapid rate of Hg II reduction leading to substantially less 295 Hg remaining in the reactor (Fig 1E and Fig 2C). Results for replicate #2 further suggested that 296 lighter Hg II isotopes were preferentially reduced with δ 202 Hg II slightly increasing at the 297 beginning of the experiment before rising to a maximum of 0.72 ‰ between fr = 0.85 at 12h and 298 0.53 at 24h (Fig 2C). The trends for δ 202 Hg 0 in chemotrophic cultures mirrored what was observed for δ 202 Hg II . 301 δ 202 Hg 0 from replicate #1 showed that the product pool was initially enriched with heavy 202 Hg 0 , 302 further suggesting cells may have accessed a readily bioavailable pool of heavy Hg II , before 303 undergoing a pronounced depletion and progressive enrichment with heavy isotopes similar to 304 phototrophic cultures (Fig 2B, D). In contrast, the δ 202 Hg 0 values for cells in replicate #2 suggest 305 that the product pool was depleted in 202 Hg 0 before undergoing enrichment for heavier Hg 0 306 isotopes reaching a maximum of 1.73 ‰ (Fig 2D). 307 The enrichment in 202 Hg 0 exceeded the initial composition of the NIST 3133 standard (0 ‰) 308 suggesting fractionation deviated from the predicted Rayleigh fractionation for Hg reduction 309 (Table S4). Recent work with G. sulfurreducens has shown that subcellular partitioning 310 processes contribute to Hg isotope fractionation and cannot be detected at the level of THg pool 311 (37). In our work, Hg reduction provided such a strong isotopic shift that we could detect 312 fractionation in the THg pool, however we cannot discount the contributions of subcellular 313 partitioning processes, which are smaller in magnitude (˂0.10 ‰). Such processes could be 314 driving the depletion of 202 Hg II early on for chemotrophic cells in replicate #1 where low cell 315 densities may have amplified uptake and adsorption-driven fractionation (Fig 1F). It is more 316 challenging to discuss contributions of such processes for replicate #2, where the low cumulative 317 Hg 0 production at fr = 0.94 and 0.88 (3h and 6h time points, respectively) precluded our ability to 318 measure Hg 0 isotope fractionation (Fig 1E and Table S2). Despite this limitation, the high 319 δ 202 Hg 0 value obtained for the final timepoint for replicate #2 reinforces that additional processes 320 are contributing to fractionation under chemotrophic conditions that warrant further exploration. 321

322
Our experimental results suggest that Hg 0 oxidation is not a major contributor at earlier 323 stages of reduction (i.e., 3-12 hours) but may be contributing to isotope fractionation at later time 324 points associated with longer Hg residence times in the reactor. Under phototrophic conditions, 325 the measured values of δ 202 Hg II at 48h were isotopically heavier than predicted values, which 326 suggests Hg 0 oxidation may be occurring (Table S4). Under fermentative conditions the 327 potential contribution of Hg oxidation is unclear, since a consistent enrichment was not observed 328 between replicates ( Table S4). The variability in isotope results for later time points in both 329 replicates suggests processes other than reduction may be occurring. 330 Although Hg 0 oxidation has never been observed in anoxygenic phototrophs, it has been 331 observed in other model chemotrophic anaerobes (49). Recent work showed that thiol-bearing 332 molecules preferentially oxidize heavy 202 Hg 0 , leading to negative MDF, in addition to a MIF 333 signal (52). Although we did not observe any negative MDF or MIF in our experiments (Table  334 S1 and Table S2), we acknowledge that such processes have the potential to contribute to net 335 fractionation alongside internal partitioning related to uptake or adsorption processes. These 336 results are an important first step to resolving the physiological processes that drive these Hg 337 transformations in H. modesticaldum and we plan to carry out additional experiments to assess 338 the contributions of uptake, adsorption, and redox transformations to the net Hg isotope 339 fractionation observed in the future. 340 341 342

19
The enrichment factor could not be calculated using the product Hg 0 pool for fermentative 363 cultures and overestimated the value in phototrophic cultures (-2.86 ± 1.32 ‰) (Fig S4 and  364 Table S3). Previous work has shown that using the product pool can overestimate the 365 fractionation factor due to difficulty obtaining analytical replicates of gaseous Hg 0 and the 366 complexity of quantitatively collecting this pool (53). As such, we carried out comparisons for 367 enrichment factors using the reactant Hg II pool. Ideally, we would have been able to obtain more 368 replicates of our experiments but were unable to do so due the challenges stated previously. It is 369 important to note that low numbers of replicates are routinely discussed in isotope geochemistry 370 and the variability from our results is in line with previous work (Fig 3 and Table S5). 371 Our initial motivation for this study was to test if phototrophic and fermentative Hg 372 reduction led to markedly different isotope enrichment factors and establish whether different 373 underlying processes supported different metabolic Hg reduction pathways. When comparing ε 374 values, which are used as an index to distinguish between different isotope fractionation 375 processes, we report a difference of 0.33 ‰ between εAP and εFM (Fig 3 and Table S5). This 376 small difference in enrichment factors supports that common processes are controlling Hg 377 reduction during anoxygenic photosynthesis and fermentation in H. modesticaldum. 378 In our previous work with H. modesticaldum we demonstrated that inhibiting pyruvate 379 ferredoxin oxidoreductase, the enzyme responsible for the production of reduced ferredoxin, 380 greatly hampered Hg II reduction during anoxygenic photosynthesis and fermentation (23). It is 381 possible that the similarity between εAP and εFM is the result of reactions originally involving the 382 same redox cofactors (e.g. ferredoxins) but with Hg II being reduced by different electron donors 383 specific to each metabolism. Such reactions could involve direct electron transfer from 384 ferredoxin to Hg II or from enzymes directly or indirectly relying on ferredoxin (23, 54). It is also 385 20 possible that the net isotopic enrichment stems from a combination of multiple processes 386 involved in Hg transport (e.g. adsorption, uptake, and efflux) that would occur regardless of the 387 intracellular transformations in question. Our results suggest that the relative contributions of 388 these processes may be more pronounced during fermentation compared to anoxygenic 389 photosynthesis although more controlled experiments are required to constrain the variability in 390 the isotope fractionation observed. 391 At the broader scale, the isotopic enrichment observed for both metabolisms tested in H. 392 modesticaldum falls between the values observed for abiotic and microbial Hg reduction 393 pathways (Fig 3 and Table S5). The isotopic enrichment observed for phototrophic H. 394 modesticaldum is slightly lower than photoreduction in the presence of dissolved organic carbon 395 (-0.80 ± 0.56 ‰ vs -0.60 ± 0.20 ‰, respectively) (32) but higher than photoreduction in the 396 presence of marine exudates and UVB light (-0.80 ± 0.56 ‰ vs -1.47 ‰, respectively) (41) (Fig  397   3 and Table S5). When comparing Hg reduction in the anoxygenic phototroph H. modesticaldum 398 to the oxygenic phototrophic green alga Isochrysis galbana, the isotopic enrichment observed is 399 also slightly lower (-0.80 ± 0.56 ‰ vs -0.14 to -0.60 ‰, respectively) (41) (Fig 3 and Table S5). 400 Though this similarity suggests anoxygenic and oxygenic photosynthetic Hg reduction share a 401 common physiological pathway, the presence of MIF in I. galbana due to free radical generation 402 within the cell rules out this possibility (41). (Table S1) When compared with previous work on microbial Hg reduction in pure cultures, the isotopic 407 enrichment observed for phototrophically-grown H. modesticaldum was higher than what has 408 previously been reported for aerobic mer-mediated reduction and anaerobic respiratory Hg 409 reduction (-0.80 ± 0.56 ‰ vs -1.80 to -1.20 ‰, respectively) (34, 36, 37) (Fig 3 and Table S5). 410 Interestingly, the isotopic enrichment observed in fermentatively-grown H. modesticaldum (-1.13 411 ± 0.50 ‰) was closer to aerobic mer-mediated reduction (-1.40 to -1.20 ‰) (34) than to 412 anaerobic respiratory Hg reduction in Shewanella oneidensis MR-1 and a modified strain of G. 413 sulfurreducens PCA incapable of Hg methylation (-1.80 to -1.60 ‰) (Fig 3 and Table S5) (36, 414 37). An additional line of evidence supporting that Hg 0 oxidation is unlikely to be contributing to 415 the isotope signature observed in our system are the ε values for thiol mediated Hg 0 oxidation in 416 the dark. These values are considerably larger than what we observed for H. modesticaldum 417 (1.51 ± 0.20 ‰ for cysteine and 1.47 ± 0.24 ‰ for glutathione) (Fig 3 and Table S5). 418 These comparisons illustrate that isotopic enrichment for microbial Hg reduction 419 pathways that only display MDF fall within a narrow range (i.e. -0.80 to -1.80 ‰) (Fig 3 and  420 Table S5). We find this to be a strikingly small difference given the ecological diversity of the 421 model organisms and variety of growth conditions used to study microbial Hg reduction. The 422 similar ε values reported for fermentation in this study and the mer operon are also noteworthy as 423 they suggest nearly identical isotope enrichment can occur in aerobes harbouring dedicated 424 enzymatic machinery to reduce Hg and anaerobes where mer-based strategies are largely absent. 425 Our study also shows that similar isotopic enrichment can be observed for different (40); anaerobic respiratory reduction (36, 37); anoxygenic photosynthetic reduction (this study); 683 fermentative reduction (this study); oxygenic photosynthetic reduction (41). The raw data in this 684 figure are in Table S5.