Duquesne University Duquesne Scholarship Collection
Electronic Theses and Dissertations
Spring 5-8-2020
The Physiology and Evolution of Selenite Respiration in Bacteria
Michael Wells
Follow this and additional works at: https://dsc.duq.edu/etd
Part of the Biochemistry Commons, Environmental Microbiology and Microbial Ecology Commons, Evolution Commons, and the Microbial Physiology Commons
Recommended Citation Wells, M. (2020). The Physiology and Evolution of Selenite Respiration in Bacteria (Doctoral dissertation, Duquesne University). Retrieved from https://dsc.duq.edu/etd/1871
This One-year Embargo is brought to you for free and open access by Duquesne Scholarship Collection. It has been accepted for inclusion in Electronic Theses and Dissertations by an authorized administrator of Duquesne Scholarship Collection.
THE PHYSIOLOGY AND EVOLUTION OF SELENITE RESPIRATION IN BACTERIA
A Dissertation
Submitted to the Bayer School of Natural and Environmental Sciences
Duquesne University
In partial fulfillment of the requirements for
the degree of Doctor of Philosophy
By
Michael Wells
May 2020
Copyright by
Michael Wells
2020
THE PHYSIOLOGY AND EVOLUTION OF SELENITE RESPIRATION IN BACTERIA
By
Michael Wells
Approved January 28, 2020
______Dr. John F. Stolz Dr. Partha Basu Professor of Environmental Microbiology Chair, Department of Chemistry and (Committee Chair) Chemical Biology Professor of Chemistry (Committee Member)
______Dr. Jan Janecka Dr. Nancy Trun Assistant Professor of Biology Associate Professor of Biology (Committee Member) (Committee Member)
______Dr. Philip Reeder Dr. Joseph McCormick Dean, Bayer School of Natural and Chair, Department of Biological Environmental Sciences Sciences Professor Professor of Biology
iii ABSTRACT
THE PHYSIOLOGY AND EVOLUTION OF SELENITE RESPIRATION IN BACTERIA
By
Michael Wells
May 2020
Dissertation supervised by Dr. John F. Stolz
The selenium oxyanions selenate (Se(VI)) and selenite (Se(IV)) can be utilized by some bacteria and archaea as terminal electron acceptors in anaerobic respiration. Se(VI) and Se(IV) respiration is mediated by a phylogenetically and ecologically diverse array of organisms, suggesting that selenium respiration is ubiquitous in natural environments. Several respiratory
Se(VI) reductases have been characterized in bacteria, revealing that Se(VI) respiration has evolved independently several times in this domain. Se(IV) respiration, in contrast, has yet to be characterized. I have purified and characterized the first respiratory Se(IV) reductase from
Bacillus selenitireducens MLS10. The Se(IV) reductase appeared to purify as a single 80 kDa enzyme and contained both iron and molybdenum. The enzyme was highly specific for Se(IV).
The genome of MLS10 demonstrated that this enzyme was part of an operon encoding a putative respiratory Se(IV) reductase (Srr) complex. Srr was electrophoretically purified from MLS10
iv periplasm using non-denaturing gels, and identified using in-gel enzyme assays for Se(IV) reducing activity. Multiple subunits from Srr were identified using liquid chromatography tandem mass spectrometry, confirming the operon codes for a Srr complex. The enzyme was designated SrrA. Phylogenetic analysis determined that SrrA was a member of the polysulfide reductase catalytic subunit (PsrA) and thiosulfate reductase catalytic subunit (PhsA) lineage of molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD)- containing molybdoenzymes. Analysis of the operons associated with each catalytic subunit in the phylogeny revealed that putative PsrA, PhsA, and SrrA homologs did not form monophyletic clades with respect to one another. Furthermore, while putative Srr operons were not observed in archaea, Srr was present throughout the PsrA/PhsA lineage in bacteria. To determine if Srr is a reliable marker for Se(IV) respiration, I attempted to ascertain if two organisms, Bacillus beveridgei MLTeJB and Desulfitobacterium hafniense PCP-1, expressed Srr when cultivated in the presence of Se(IV). MLTeJB was shown to express SrrA when grown on both Se(VI) and
Se(IV). Definitive identification of the Se(IV) reductase expressed by PCP-1 when grown on
Se(VI) was not possible due to the low expression levels of the enzyme. This work represents the first physiological and evolutionary studies of Se(IV) respiration.
v DEDICATION
To my sister Annie, a continual source of support and guidance as I have transitioned from one life to another. I am perpetually in awe of her selflessness, compassion, and wisdom, and am so grateful every day for the privilege of being her brother. To İstanbul, to Türkiye, I remain forever indebted to a city and a culture that showed me how to be the man I aspire to be.
So many people, too many to innumerate, let me into their culture, their homes, and their lives. I have been forever changed because of it.
vi TABLE OF CONTENTS
Page
Abstract ...... iv
Dedication ...... vi
Acknowledgement ...... vii
List of tables ...... ix
List of figures ...... x
Chapter 1: Introduction, hypothesis, and specific aims...... 1
Chapter 2: A review of the selenium biogeochemical cycle and of selenium metabolism...... 19
Chapter 3: Materials and methods ...... 103
Chapter 4: The respiratory selenite reductase of Bacillus selenitireducens strain MLS10...... 119
Chapter 5: Selenite metabolism in Thermus scotoductus SA-01, Desulfitobacterium hafniense
PCP-1, and Bacillus beveridgei MLTeJB ...... 143
Chapter 6: Phylogenetic analysis of SrrA and other Dimethyl Sulfoxide Reductase (DMSOR) family member catalytic subunits ...... 163
Appendix 1: MLS10 and MLTeJB media and trace elements solution ...... 204
Appendix 2: PCP-1 medium and trace elements and vitamins solutions ...... 206
Appendix 3: SA-01 media and trace elements and vitamins solutions ...... 208
References ...... 211
vii LIST OF TABLES
Page
Table 4.1...... 127
Table 5.1 ...... 151
Table 5.2 ...... 158
Table 6.1 ...... 172
Table 6.2 ...... 189
viii LIST OF FIGURES
Page
Figure 4.1 ...... 126
Figure 4.2 ...... 128
Figure 4.3 ...... 129
Figure 4.4 ...... 131
Figure 4.5 ...... 133
Figure S4.1 ...... 140
Figure 5.1 ...... 150
Figure 5.2 ...... 154
Figure 5.3 ...... 156
Figure 5.4 ...... 157
Figure 6.1 ...... 174
Figure 6.2 ...... 179
Figure 6.3 ...... 181
Figure 6.4 ...... 186
Figure S6.1 ...... 195
Figure S6.2 ...... 196
Figure S6.3 ...... 197
Figure S6.4 ...... 198
Figure S6.5 ...... 199
Figure S6.6 ...... 200
Figure S6.7 ...... 201
ix Figure S6.8 ...... 202
Figure S6.9 ...... 203
x CHAPTER 1
INTRODUCTION, HYPOTHESES, AND SPECIFIC AIMS
Introduction and historical overview of selenium research
Our recognition of selenium as a vital micronutrient that can sustain life via assimilation into biological macromolecules and as a source of energy during anaerobic respiration emerged slowly. Biological interactions with selenium were observed throughout the 19th and early 20th centuries, though much of that literature remains unavailable in the English language. These findings are summarized elsewhere in depth for bacteria (Levine 1925) and plants and animals
(Levine 1915), and describe the reduction of an extensive array of selenium compounds, including the selenium oxyanions selenite (Se(IV)) and selenate (Se(VI)), and their respective acids, selenious and selenic acid, by a variety of aerobic bacteria and eukaryotes. At that time, it was unclear what metabolic function, if any, was served by these reductive transformations of selenium, much less if metabolic pathways specific for selenium were involved.
The first indication that selenium might be of biological importance was reported by
Brenner (1916), who isolated a bacterium, Micrococcus selenicus, that appeared to be capable of autotrophic growth using selenide (Se(-II)) as an electron donor, presumably during aerobic respiration (Levine 1925). Later, Lipman and Waksman (Lipman and Waksman 1923) reported a bacterium that could utilize elemental selenium (Se(0)) as an electron donor, oxidizing Se(0) to
Se(VI) during aerobic respiration. Tantalizing though these two reports may be, evidence that these bacteria could indeed grow autotrophically on selenium oxyanions was never provided, and no isolated bacterium to date is known to perform this metabolic feat.
Instead, the prevailing biological narrative concerning selenium for much of the 20th century focused on the potent toxicity of the element. This narrative emerged with the discovery
1 that alkali disease, which afflicted domesticated livestock across the Great Plains states of the
United States, was caused by the consumption of seleniferous grasses (Franke et al. 1934).
Subsequent research on biological interactions with selenium from the 1930s to 1950s largely sought to document the physiological maladies caused by exposure to selenium (e.g., (Schneider
1936; Manville 1939)), and the role of selenium as a potential mutagen (e.g., (Seifter and Ehrich
1946)). Contrary to the thrust of these findings, Trelease and Trelease (1938) found that several plants of the genus Astragalus required selenium to grow, though no possible biological role for selenium utilization was known at that time. A specific biological function was first ascribed to selenium by Pinsent (Pinsent 1954), who found that cells of Escherichia coli required Se(IV) to express the nitrate-inducible formate dehydrogenase N during anaerobic respiration on nitrate.
The nature of this requirement remained unknown for several decades.
Subsequently, several lines of evidence suggested that selenium could have a nutritive function in organisms. Rats fed a diet consisting only of yeast, sugar, vitamin E, cysteine, and methionine developed lethal necrosis of the liver. Supplementation of this diet with an unknown biological compound, Factor 3, could arrest the onset of liver necrosis, and this compound was shown to contain selenium (Schwarz and Foltz 1957). Presciently, the authors noted that the supplemented selenium seemed to function in redox reactions in vivo, as rat liver cells that did not receive supplementation of Factor 3 showed a decreased respiratory rate prior to the onset of necrosis. The authors later demonstrated that the rescuing effect of Factor 3 could be obtained by supplementing the diet of rats with small quantities of inorganic forms of selenium, including
Se(VI) and Se(IV) (Schwarz and Foltz 1958). Several studies also demonstrated that a number of diseases in livestock was caused by a deficiency of selenium (e.g., white muscle disease in sheep and other livestock (Muth et al. 1958)). Finally, it was demonstrated that several selenium
2 accumulating plants, upon supplementation with Se(IV), produced the seleno-amino acid methylselenocysteine (Shrift and Virupaksha 1963). These observations spurred Shrift (Shrift
1964) to postulate the existence of a selenium biogeochemical cycle, the first mention of the cycle in the literature at a time when the basic contours of the selenium biogeochemical cycle remained unknown.
1973 proved to be a banner year for research on the role of selenium in biology, as multiple independent lines of inquiry identified specific biochemical functions for selenium assimilation in both mammals and bacteria. The mammalian enzyme glutathione peroxidase, which couples the oxidation of glutathione to the reduction of various toxic organic and inorganic H2O2 reactive oxygen species, was shown to require selenium for activity in vitro, and that this activity could be restored by addition of selenite to the reaction (Rotruck et al. 1973).
Concurrently, it was demonstrated that selenium was incorporated into glutathione peroxidase at a stoichiometry of 1 mol selenium per each 1 mol monomer subunit of the homo-tetrameric enzyme (Flohe, Günzler and Schock 1973).
Two teams of researchers, working on different bacteria from the clostridia, identified two novel selenium-containing enzymes. Protein A (now termed glycine reductase A) of the glycine reductase complex, which catalyzes the fermentative reduction of glycine to acetate, from the bacterium Clostridium sticklandii was found to specifically incorporate selenium into the enzyme (Turner and Stadtman 1973). An additional selenoenzyme, the NAD+-dependent formate dehydrogenase (Andreesen and Ljungdahl 1973), expressed during acetogenesis by the acetogen Clostridium thermoaceticum (now christened Moorella thermoacetica (Collins et al.
1994)), was identified. Shortly thereafter, the first indication as to the species of selenium incorporated into these enzymes was provided, as the purified glycine reductase A of C.
3 sticklandii was shown to incorporate selenium as an organic selenide species, specifically a selenocysteine (Sec) residue (Cone et al. 1976). Finally, the 1970s saw the first report of a Sec- containing protein from the Archaea domain, with the discovery that the F420-dependent formate dehydrogenase of the archaeon Methanococcus vannielii, expressed during hydrogenotrophic methanogenesis, contained a Sec residue (Jones, Dilworth and Stadtman 1979a; Jones and
Stadtman 1981)
Research in the 1980s into Sec utilization revealed that Sec was not a product of post- translational modification, or of the non-specific substitution of a selenium atom for sulfur, as occurs for the other known seleno-amino acid, selenomethionine (SeMet) (Stadtman 1987).
Rather, it was shown that Sec was encoded for specifically in a hitherto unprecedented expansion of the genetic code. This was first suggested in 1986, with two independent reports that the nucleotide sequences of the mouse glutathione peroxidase gene (Chambers et al. 1986) and the
E. coli formate dehydrogenase H subunit of the formate hydrogen lyase complex (Zinoni et al.
1986) both contained a UGA opal codon at the site of the Sec residue. Shortly thereafter, it was shown that the UGA opal codon directed the co-translational insertion of the Sec residue, confirming the remarkable finding that the UGA codon simultaneously serves both as a stop codon and a Sec codon in organisms capable of synthesizing Sec. Proteins containing this residue are referred to as selenoproteins. The late 1980s inaugurated a series of reports that laid down the basic molecular architecture of Sec synthesis and co-translational insertion in Bacteria using E. coli as a model organism (Forchhammer, Leinfelder and Böck 1989; Leinfelder et al.
1988a, 1988b, 1990; Zinoni, Heider and Böck 1990; Forchhammer, Rücknagel and Böck 1990;
Forchhammer and Böck 1991).
4 Research in the 1980s also uncovered two novel forms of selenium metabolism. One new form of selenium metabolism involves assimilation of selenium into tRNA nucleosides.
Earlier reports had shown that E. coli produced selenium-containing tRNAs when grown on medium supplemented with Se(IV) in the form of 4-selenouridine ((Hoffman and McConnell
1974; Prasada Rao and Cherayil 1974). These findings were corroborated in 1980 for the bacterium C. sticklandii (Chen and Stadtman 1980) and the archaeon M. vannielii (Ching et al.
1984), which incorporated selenium specifically in the form of 5-methylaminomethyl-2- selenouridine. It was subsequently shown that the incorporation of selenium into tRNAs was highly specific in E. coli, occurring only in lysine- and glutamate-accepting tRNA species, and that this incorporation was mediated by a pathway distinct from the incorporation of the sulfur- containing analogue 4-thiouridine (Wittwer 1983). It was also shown that both E. coli and C. sticklandii synthesized selenium-containing tRNAs primarily in the form of 5- methylaminomethyl-2-selenouridine (mnm5Se2U), rather than 4-selenouridine (Wittwer et al.
1984). Intriguingly, this species of tRNA was shown to be incorporated into lysine- and glutamate-accepting tRNAs exclusively at the wobble position of the anti-codons (Ching,
Alzner-DeWeerd and Stadtman 1985; Wittwer and Ching 1989).
Concomitant with these discoveries, the dissimilatory reduction of selenium oxyanions during anaerobic respiration was described for the first time. Selenium pollution, in the form of
Se(VI) and Se(IV), had become a problem in the San Joaquin Valley, as selenium-contaminated irrigation wastewater was exposed to the Kesterson Reservoir of the Kesterson National Wildlife
Refuge, resulting in the accumulation of toxic quantities of selenium in mammalian (Clark
1987), avian (Presser and Ohlendorf 1987; Ohlendorf et al. 1988), and fish (Tanji, Läuchli and
Meyer 1986; Presser and Ohlendorf 1987) populations. Avian and fish populations were
5 particularly devastated, with widespread deformations in their embryos reported. Bacteria from the Kesterson Reservoir were shown to be resistant to high quantities of Se(VI) (Maiers et al.
1988) and Se(IV) (Burton et al. 1987), with these bacteria detoxifying both oxyanions by reducing them to Se(0).
The first indication that Se(VI) reduction could be linked to energy conservation came from Zehr and Oremland (1987). These researchers demonstrated that small quantities of Se(VI) could be reduced by sulfate-respiring bacteria. This led to the discovery that dissimilatory
Se(VI) reduction during anaerobic respiration occurred both in the Kesterson Reservoir and environments not impacted by selenium pollution, and that this dissimilatory reduction was independent of dissimilatory sulfate reduction (Oremland et al. 1989, 1991). The first bona fide
Se(VI) respiring organism was later isolated from selenium polluted water, a Pseudomonas species that could grow on millimolar quantities of Se(VI), reducing Se(VI) to Se(IV) during energy conservation (Macy, Michel and Kirsch 1989). This organism was later renamed
Thauera selenatis (Macy et al. 1993).
Research in the 1990s and early 2000s elucidated the central components of the Sec synthesis and insertion architecture of Archaea and Eukarya (Berry et al. 1991; Kim and
Stadtman 1995; Rother et al. 2000, 2001; Carlson et al. 2004; Kaiser et al. 2005; Yuan et al.
2006). These studies revealed that the machinery of Sec synthesis and insertion is broadly similar among the three domains of life, though the archaeal and eukaryotic domains share features of Sec synthesis that differ markedly from bacteria. Biochemical techniques also expanded the repertoire of known selenoproteins, underscoring the importance of selenoproteins in energy conservation (Sorgenfrei et al. 1993; Vorholt, Vaupel and Thauer 1997; Kabisch et al.
1999; Wagner et al. 1999; Self and Stadtman 2000) and antioxidant defense in prokaryotes
6 (Andreesen et al. 1999a; Söhling et al. 2001) and in antioxidant defense and redox homeostasis in eukaryotes (Berry, Banu and Larsen 1991; Vendeland et al. 1993; Gladyshev, Jeang and
Stadtman 1996; Gromer, Schirmer and Becker 1997; Kryukov et al. 2002). The molecular mechanisms of mnm5Se2U synthesis were also elucidated in Bacteria (Veres and Stadtman 1994;
Wolfe et al. 2004a) and, a decade later, in Archaea (Su et al. 2012). As with the Sec synthesis machinery, the mechanisms of mnm5Se2U synthesis are broadly conserved between the prokaryotic domains.
A third form of selenium assimilatory metabolism was also discovered in the 1990s.
Researchers in the 1970s had uncovered evidence for the presence of acid-labile Se(-II) in proteins from rat liver (Lucy, Diplock and Caygill 1970; Diplock et al. 1973). Later, indications emerged that the nicotinic acid hydroxylase of Clostridium barkeri (later renamed Eubacterium barkeri (Collins et al. 1994)), expressed during the fermentation of nicotinic acid, also contained inorganic Se(-II) (Dilworth 1982). The presence of inorganic Se(-II) as a co-factor at the catalytic site of the nicotinic acid hydroxylase of E. barkeri, rather than Sec, was subsequently confirmed (Gladyshev, Khangulov and Stadtman 1994). The xanthine dehydrogenase of E. barkeri, expressed during the fermentation of the purine xanthine, was also shown to incorporate inorganic Se(-II) (Schräder, Rienhöfer and Andreesen 1999), as were the xanthine dehydrogenase and purine hydroxylase of Clostridium purinolyticum (Self and Stadtman 2000).
C. purinolyticum was subsequently reclassified as Gottschalkia purinolyticum (Poehlein et al.
2015). This third selenium utilization trait has only been observed in bacteria.
The advent of the “omics” era, and the concomitant development of bioinformatic tools for gene annotation, has proven to be a boon for research on selenium assimilation, despite the complications posed by the dual function of the opal codon in Sec utilizing organisms.
7 Comparative genomics has identified putative selenoproteins in eukaryotes (Kryukov, Kryukov and Gladyshev 1999; Lescure et al. 1999; Kryukov et al. 2003; Castellano et al. 2005), and a number of bioinformatic tools have been developed to identify putative selenoproteins in eukaryotic genomes (Castellano et al. 2008; Mariotti et al. 2013; Romagné et al. 2014). These advances have allowed researchers to construct and test evolutionary hypotheses regarding the evolution of Sec usage and loss in eukaryotes, which are sparsely and unevenly distributed among eukaryotic taxa, entirely absent in land plants and yeast, and particularly concentrated in metazoans (Lobanov et al. 2007; Jiang, Ni and Liu 2012).
Comparative genomics has identified putative selenoproteins in prokaryotes, many involved in antioxidant defense and redox homeostasis (Kryukov and Gladyshev 2004; Zhang,
Fomenko and Gladyshev 2005; Zhang et al. 2006). Far fewer bioinformatic tools are available for Bacteria (Zhang and Gladyshev 2005), and none exist for Archaea. Comparative genomics has also identified putative genes associated with the incorporation of inorganic Se(-II) into enzymes (Zhang et al. 2008). Ecological and evolutionary analysis of the three known traits of selenium utilization in bacteria reveals that selenium utilization is encoded in approximately 1/5th of genomes, and is unevenly distributed across the bacteria, with some phyla (e.g., the
Firmicutes and Deltaproteobacteria) possessing a substantial number of organisms encoding all three selenium utilization traits and exploiting an extensive repertoire of selenoproteins (Zhang et al. 2006; Zhang and Gladyshev 2008, 2010; Peng et al. 2016). These analyses have offered several insights into ecological features selecting for selenium utilization, as well as evolutionary hypotheses for the origin and evolution of selenium utilization in bacteria. Finally, a recent bioinformatic discovery is the likely ability of organisms from the Lokiarchaeota to utilize Sec, a trait previously thought to be confined to methanogens in this domain (Mariotti et al. 2016).
8 Most research on selenium respiration in the 1990s and 2000s has centered on expanding the number of selenium respiring prokaryotes. A second Se(VI) respiring bacterium, strain SeS-
3 was isolated from estuarine sediments in Nevada (Oremland et al. 1994). This strain was later christened Sulfurospirillum barnesii (Stolz et al. 1999). Subsequently, a substantial number of
Se(VI) respiring bacteria and archaea have been isolated from a wealth of different environments, from soda lakes to hydrothermal vents (Fujita et al. 1997; Switzer Blum et al.
1998; Huber et al. 2000; Blum et al. 2001; Nakagawa et al. 2006; Narasingarao and Häggblom
2006, 2007a, 2007b; Fisher and Hollibaugh 2008; Baesman et al. 2009; Slobodkina et al. 2015;
Abin and Hollibaugh 2017), demonstrating that Se(VI) respiration is mediated by a phylogenetically diverse assemblage of prokaryotes. The molecular mechanisms of Se(VI) respiration in T. selenatis (Schröder et al. 1997; Krafft et al. 2000) were elucidated, and a respiratory Se(VI) reductase of Bacillus selenatarsenatis was identified a decade later (Kuroda et al. 2011b). Curiously, the Se(VI) reductases of these two bacteria are only distantly related
(Grimaldi et al. 2013), suggesting that Se(VI) respiration has evolved multiple times in the
Bacteria.
Two novel forms of selenium respiration were also identified. The bacterium Bacillus selenitireducens was isolated from Mono Lake, a soda lake in California, and was shown to be able to utilize Se(IV) as an electron acceptor in anaerobic respiration, reducing Se(IV) to Se(0)
(Switzer Blum et al. 1998). Shortly thereafter, an archaeon, Pyrobaculum aerophilum, was shown to be capable of Se(IV) respiration (Huber et al. 2000). Four other prokaryotes have been isolated that are capable of Se(IV) respiration (Baesman et al. 2009; Rauschenbach,
Narasingarao and Häggblom 2011; Slobodkina et al. 2015; Abin and Hollibaugh 2017). It was later demonstrated that B. selenitireducens could also conserve energy via Se(0) respiration,
9 reducing Se(0) to Se(-II) (Herbel et al. 2003). The authors also found that enrichment cultures from various environmental settings could also reduce Se(0) to Se(-II) during anaerobic respiration, but additional Se(0) respiring prokaryotes have yet to be isolated.
Hypothesis and specific aims
Se(IV) respiration remains a poorly characterized form of selenium metabolism. While few Se(IV) respiring prokaryotes are known, three bacteria capable of Se(IV) respiration have been isolated from the Firmicutes phylum (Switzer Blum et al. 1998; Baesman et al. 2009; Abin and Hollibaugh 2017) and one from the Chrysiogenetes phylum (Rauschenbach, Narasingarao and Häggblom 2011). Additionally, two hyperthermophilic archaea from the Pyrobaculum genus in the Crenarchaeota phylum can respire Se(IV) (Huber et al. 2000; Slobodkina et al.
2015). Thus, Se(IV) respiration is mediated by a phylogenetically diverse assemblage of organisms from both prokaryotic domains of life. Moreover, these organisms have been isolated from locales as varied as the hypersaline and alkaline Mono Lake in California (Switzer Blum et al. 1998; Baesman et al. 2009; Abin and Hollibaugh 2017) to geothermal springs in Kamchatka,
Russia (Slobodkina et al. 2015). Thus, respiratory Se(IV) reduction has been observed in equally diverse geochemical environments. Both lines of evidence implicate this reductive transformation of Se(IV) to Se(0) as a ubiquitous and crucial reaction in the global selenium biogeochemical cycle.
Furthermore, Se(IV) respiration, in contrast to Se(VI), could represent a form of anaerobic respiration that predates the divergence of life into three domains. Physiological and phylogenetic characterization of respiratory Se(VI) reductases have already shown that multiple independent mechanisms of Se(VI) respiration have evolved in bacteria (Schröder et al. 1997;
Stolz et al. 2006; Kuroda et al. 2011b). This is fully consistent with thermodynamic data
10 indicating that Se(VI) would only be stable in the presence of strong oxidants (e.g., substantial quantities of O2) (Stüeken 2017), and selenium isotope fractionation data indicating that Se(VI) was not present in environments to an appreciable extent until the Neoproterozoic Era ~ 1.0 billion years ago (Stüeken et al. 2015a; Stüeken 2017). Thermodynamic data, however, indicate that Se(IV) would be the predominant species of selenium in environments with only mild oxidants, and can even be stable in reducing environments characterized by alkaline pH conditions (Stüeken 2017). Experimental data has shown that reduced selenium species oxidize spontaneously to Se(IV) in sediments with a redox potential of 0 mV, roughly concomitant with the oxidation of ferrous iron to ferric iron (Masscheleyn, Delaune and Patrick 1990). The redox state of Archean Eon (~4.0 – 2.5 billion years ago) environments remains unconstrained, though the presence of banded iron formations throughout this geological eon (Knoll, Bergmann and
Strauss 2016) suggests that Se(IV) could have been available to organisms as a source of selenium for assimilation into macromolecules and as a source of energy in anaerobic respiration.
Thus, there is currently an active debate as to whether Se(IV) was a salient feature of the
Archean selenium biogeochemical cycle. Some researchers contend that selenium isotope fractionation data demonstrates that only the reduced selenium species Se(0) and Se(-II) were present in Archean environments (Stüeken et al. 2015a; Stüeken 2017). This model, if correct, is difficult to reconcile with evolutionary data indicating that Sec utilization was present in the last universal common ancestor (Mariotti et al. 2015; Weiss et al. 2016), as both Se(0) and inorganic
Se(-II) are thought to have limited bioavailability (Masscheleyn, Delaune and Patrick 1990,
1991). Other researchers argue that rapid biological utilization of Se(IV) could mute isotope
11 fractionation patterns indicative of biological Se(IV) reduction in the sedimentary record
(Mitchell et al. 2016).
Evolutionary study of the mechanisms of Se(IV) respiration could independently test these competing hypotheses for the evolution of the selenium biogeochemical cycle. If respiratory Se(IV) reductases are conserved between bacteria and archaea, this would be powerful evidence that Se(IV) respiration was an adaptation of the last universal common ancestor and served as a source of energy for microbial communities throughout the Archean
Eon. This would also be demonstrate that Se(IV) was likely an essential source of bioavailable selenium prior to the Great Oxygenation Event ~2.5 billion years ago (Mitchell et al. 2016). If multiple mechanisms of Se(IV) respiration have evolved in both domains of life, as has been observed for Se(VI) respiration in bacteria, this would be fully consistent with isotope fractionation data indicating that a substantial reservoir of Se(IV) was not available for organisms until after the Great Oxygenation Event. A further possibility is that Se(IV) respiration evolved once in bacteria and once in archaea, early in the diversification of these two lineages. This scenario would indicate that Se(IV) was present at some point in the Archean, but would not decisively favor either of these two competing models. I evaluated the likelihood of these three scenarios by testing the hypothesis that Se(IV) respiration is mediated by conserved mechanisms in bacteria. This hypothesis was tested using three specific aims:
Hypothesis 1: The respiratory Se(IV) reductase of B. selenitireducens is a member of the
DMSOR family of mononuclear enzymes.
Specific Aim 1: Biochemical characterization of the Bacillus selenitireducens Se(IV) reductase.
Both of the respiratory selenate reductases identified in T. selenatis (Schröder et al. 1997;
Dridge et al. 2007) and B. selenatarsenatis (Kuroda et al. 2011b) are members of the dimethyl
12 sulfoxide reductase (DMSOR) family of mononuclear molybdoenzymes. Furthermore, DMSOR family members are also the terminal reductases of other geochemically similar chalcophilic substrates including the respiratory arsenate reductase (ArrA) (Krafft and Macy 1998; Afkar et al. 2003; Saltikov and Newman 2003) and the sulfur intermediates polysulfide reductase (Krafft et al. 1992), thiosulfate reductase (Heinzinger et al. 1995), and tetrathionate reductase (Hensel et al. 1999). Thus, it is reasonable to suppose that a DMSOR family member would be a likely candidate for the respiratory Se(IV) reductase of B. selenitireducens. This possibility can easily be evaluated using a combination of proteomics, biochemistry, enzymology, and sequence analysis. A metals analysis of the purified or enriched Se(IV) reductase would contain iron and either molybdenum or tungsten. A sequence analysis of enzyme should also contain binding motifs associated with the distinctive molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD) cofactor. If Se(IV) respiration is conserved in bacteria, then the
B. selenitireducens enzyme should also be highly specific for Se(IV) and have a Km for the substrate in the M range, reflecting reasonable concentrations of the oxyanion that would be available in natural environments.
Evolutionary analysis of Se(IV) respiration requires the identification of specific proteins involved in mediating respiratory reduction. The lack of known respiratory Se(IV) reductases in the literature currently precludes this. To address this gap, I purified the respiratory Se(IV) reductase of Bacillus selenitireducens MLS10 and investigated the kinetics, substrate specificity, and metal content of enzyme. B. selenitireducens was the first organism shown to be capable of respiratory Se(IV) reduction (Switzer Blum et al. 1998) and is also the only organism known to utilize Se(0) as a terminal electron acceptor (Herbel et al. 2003). Additionally, the genome of B.
13 selenitireducens contains genes encoding nine putative selenoproteins, making this an ideal model organism in which to investigate Se(IV) respiration.
Candidates for the respiratory Se(IV) reductase were identified by electrophoretically separating proteins from crude periplasmic, insoluble, and cytoplasmic fractions in non- denaturing gels. Following this separation, in-gel enzyme assays using methyl viologen as an artificial electron donor and Se(IV) as an electron acceptor was performed in an anaerobic glove box. Proteins with Se(IV) reducing activity coupled methyl viologen oxidation to Se(IV) reduction, producing distinct bands of clearing. The protein content of these bands was analyzed directly using LC MS/MS or soaked in a Laemmli buffer and run out on SDS-PAGE gels to determine the molecular weight and subunit composition of the enzyme. Furthermore, I obtained enriched fractions of the respiratory Se(IV) reductase for biochemical characterization of the enzyme using the methyl viologen assay to spectrophotometrically monitor Se(IV) reducing activity in enriched and pure fractions. Fractions were incubated in assays with varying concentrations of Se(IV) to determine the affinity of the enzyme for this substrate. I also investigated the substrate specificity of the Se(IV) reductase by testing the ability of the enzyme to couple the oxidation of reduced methyl viologen to the reduction of other electron acceptors
(e.g., As(V), thiosulfate, Se(VI), etc.). Finally, purified Se(IV) reductase was analyzed by inductively coupled plasma mass spectrometry (ICP-MS) to determine the metal content of the enzyme.
Hypothesis 2: The genes encoding the respiratory Se(IV) reductase of MLS10 are highly conserved and thus a reliable marker for identifying Se(IV) respiring bacteria.
Specific Aim 2: Identification of Se(IV) reductases in bacteria with homologs to the MLS10 respiratory Se(IV) reductase
14 If Se(IV) respiration is conserved among bacteria, the gene or genes encoding for the B. selenitireducens respiratory Se(IV) reductase should be reliable markers for the ability of distantly related bacteria to exploit Se(IV) as a terminal electron acceptor in anaerobic respiration. Moreover, the genomes of all known Se(IV) respiring bacteria should contain genes encoding homologs of this enzyme. Comparative genomics was utilized to ascertain that all
Se(IV) respiring bacteria indeed harbor homologs of this enzyme and to identify bacteria that are not currently known to utilize Se(IV) as a terminal electron acceptor. One organism that is known to respire Se(IV) and two organisms in which this capability has yet to be assessed were chosen for further physiological and biochemical characterization.
Both organisms were cultivated in anaerobic media on Se(IV), with the ability to couple
Se(IV) reduction to the oxidation of an electron donor determined by the successful cultivation of each organism through three sequential transfers into media with 10mM Se(IV). In-gel enzyme assays using non-denaturing gels were exploited, as described in Specific Aim 1, to assess whether homologs of the B. selenitireducens Se(IV) reductase possess Se(IV) reducing activity in organisms capable of respiratory growth on Se(IV). I used both crude fractions
(periplasm, insoluble, and cytoplasmic) and enriched fractions (e.g., fractions obtained via ammonium sulfate precipitation) for electrophoretic separation. Bands with Se(IV) reducing activity were excised from the non-denaturing gels and soaked in a Laemmli buffer. The proteins eluted from the Se(IV) reducing band were then run down an SDS-PAGE gel to confirm that the putative Se(IV) reductase from these organisms has the appropriate molecular weight and subunit composition as the MSL10 homolog. Proteins were identified using LC MS/MS.
Hypothesis 3: The respiratory Se(IV) reductase is an ancient enzyme and part of the metabolic repertoire of the last common ancestor (LUCA).
15 Specific Aim 3: Phylogenetic analysis of the respiratory Se(IV) reductase of B. selenitireducens
The identification of a bona fide respiratory Se(IV) reductase creates an opportunity for a rigorous evolutionary analysis of Se(IV) respiration as a bioenergetic adaptation. The first goal of this aim was to determine the family or superfamily the enzyme from B. selenitireducens is a member of. It should be noted that respiratory enzymes are frequently composed of multi- subunit complexes (e.g., see (Rothery, Workun and Weiner 2008; Hille, Hall and Basu 2014)) with each subunit having a specific function in facilitating the transfer of electrons from the membrane quinone pool to the substrate. The Se(IV) reductase is similarly a complex of multiple subunits, and the phylogenetic analysis was thus confined to the catalytic subunit that specifically transfers electrons to Se(IV) to produce Se(0).
The phylogenetic analysis did not merely determine which family or superfamily the B. selenitireducens respiratory Se(IV) reductase is a member of, but also accurately positioned the enzyme and putative homologs within the dimethyl sulfoxide reductase (DMSOR) family. This is essential to ascertain the ancestry of Se(IV) respiration. If Se(IV) respiration is conserved between bacteria and archaea, then putative homologs will likely form a coherent monophyletic clade that is deeply branched within the DMSOR family. Furthermore, bacterial and archaeal homologs would form monophyletic clades with respect to one another. If Se(IV) respiration represents a derived adaptation that evolved long after current phyla within the two domains diversified from one another, then putative Se(IV) reductase homologs should form a monophyletic clade in recently branched lineages. These homologs should come from genomes representing a relatively limited phylogenetic distribution
Identifying appropriate homologs involved using the DELTA-BLAST tool (Boratyn et al.
2012) to identify putative homologs of this oxidoreductase in genomes from both bacteria and
16 archaea. Proteins were considered homologs if the B. selenitireducens query covers 95% or more of the protein sequence, and if the identity is 30%. This search was confined to bacterial and archaeal phyla in which organisms have been isolated in pure cultures, particularly focusing on phyla represented in the Genomic Encyclopedia of Bacteria and Archaea (Wu et al. 2009;
Mukherjee et al. 2017). Sequence alignments of the Se(IV) reductase homologs were constructed using a number of alignment programs, such as Muscle (Edgar 2004) and MAFFT
(Katoh, Rozewicki and Yamada 2017), that are appropriate for aligning amino acid sequences.
The metals analysis and primary sequence analysis provided insight as to cofactor binding sites or structural and/or functional domains and motifs that facilitated assessment of the quality of alignments. The alignment program and settings that produced the most robust alignments (i.e., accurately align important binding motifs and domains in the large multiple sequence alignments) was the MAFFT using the GINS-1 settings and was used for the subsequent phylogenetic analysis. A number of software programs were used to identify an appropriate amino acid substitution model including ProtTest3 (Darriba et al. 2011) and ModelFinder
(Kalyaanamoorthy et al. 2017). ModelFinder proved to be the best software program to accomplish this task due to the large number of sequences in the alignment.
A number of phylogenetic methods were used to construct trees. Bayesian phylogenies were constructed using MrBayes 3.2 (Ronquist et al. 2012) and maximum likelihood phylogenies were constructed using RAxML version 8.0 (Stamatakis 2014) and IQ-TREE version version 1.6.10 (Nguyen et al. 2015). Note that Bayesian phylogenetic analysis proved to be inappropriate for the large data set I constructed, as this method is computationally intensive.
Similarly, non-parametric bootstraps were not computationally feasible either, thus the parametric bootstrap option was used for RAxML v. 8.0 and the UltraFast boostrap
17 approximation (Hoang et al. 2018) for IQ-TREE for the phylogenetic analysis. The substantial size of the dataset required the use of the supercomputers available on the CIPRES (Miller,
Pfeiffer and Schwartz 2010) platform. Nodes were considered robustly supported if the posterior probability of the node was 98 for Bayesian phylogenies or if the bootstrap support was 80.
Collectively, these three specific aims offered a detailed assessment of the antiquity of Se(IV) respiration using protein biochemistry, enzymology, proteomics, comparative genomics, and phylogenetic analysis.
18 CHAPTER 2
REVIEW OF THE SELENIUM BIOGEOCHEMICAL CYCLE AND SELENIUM
METABOLISM
What has emerged slowly from nearly two centuries of research into the selenium biogeochemical cycle is the realization that the cycle bears many similarities to the biogeochemical cycle of its sister chalcogen, sulfur, as the element sustains life both via assimilation into macromolecules and as a source of energy during respiration. A number of reviews exist which detail the ecology and molecular biology of selenium utilization (Stolz and
Oremland 1999; Stolz et al. 2006), and reviews that also focus heavily on the utility of selenium respiring and detoxifying prokaryotes in bioremediation (Nancharaiah and Lens 2015; Eswayah,
Smith and Gardiner 2016). This review aspires to not just provide updated information on the substantial progress that has been made on selenium metabolism, but to integrate this work within broader concerns in geobiology. It will therefore be confined to the ways that selenium sustains life in bacteria and archaea. First, we will provide an overview of the selenium biogeochemical cycle in oxic and anoxic environments, and how this information, coupled with what is currently known about bacterial and archaeal utilization of selenium, can inform hypotheses as to how the selenium biogeochemical cycle has changed over deep time. Then we will discuss the molecular biology, biochemistry, ecology, and evolution of selenium assimilation and respiration, emphasizing areas of inquiry that warrant further investigation.
The selenium biogeochemical cycle
The literature contains a number of excellent reviews that provide an overview of the global selenium biogeochemical cycle (Fernández-Martínez and Charlet 2009; Winkel et al.
2012; Sharma et al. 2014), and the selenium biogeochemical cycle in a variety of specific
19 environmental compartments (Wen and Carignan 2007; Floor and Román-Ross 2012). Selenium is a group 16 element, located on the same column of the periodic table of elements as oxygen and sulfur. Selenium exists in nature in four common oxidation states: Se(-II), Se(0), Se(IV), and Se(VI), while Se(-I) is a rare oxidation state seldom encountered in nature (Sharma et al.
2014). Selenium shares very similar chemical and physical properties with sulfur, including identical oxidation states (Reich and Hondal 2016).
The most substantive difference between the two elements is simply the larger size of the selenium atom, which makes the element significantly more reactive (i.e., selenium is both more nucleophilic and electrophilic) than sulfur (Reich and Hondal 2016). From a geochemical perspective, this produces a biogeochemical cycle in which all four common oxidation states of selenium have wide regions of thermodynamic stability, whereas sulfide and sulfate are the only stable species of sulfur in the sulfur biogeochemical cycle, with elemental sulfur, sulfite, and multivalent sulfur molecules being transient species (Pogge von Strandmann et al. 2015; Stüeken
2017). This makes the speciation of selenium, and thus the element’s mobility and bioavailability, in any particular environmental compartment acutely sensitive to local fluctuations in redox state and pH.
Selenium in the atmosphere
Wen and Carignan (2007) provide a comprehensive overview of the atmospheric selenium biogeochemical cycle. Atmospheric selenium represents a transient phase in the selenium biogeochemical cycle. Selenium emitted into the atmosphere consists of particulate selenium (e.g., Se(0)) and volatile gaseous phases of selenium (e.g., methylated Se(-II)s and
SeO2). Several reports have estimated the total flux of selenium into the atmosphere each year to be between 13,000 and 19,000 tons (Mosher and Duce 1987; Nriagu and Pacyna 1988). These
20 selenium emissions come from both natural and anthropogenic sources. Mosher and Duce
(1987) estimated that approximately 60% of global selenium emissions come from natural sources. Ross (1985), however, estimated that roughly 70-80% of selenium emissions in the
Northern Hemisphere come from anthropogenic sources. Both of these studies suggest that the selenium biogeochemical cycle, as with other biogeochemical cycles (e.g., the nitrogen biogeochemical cycle), has been profoundly transformed by the Industrial Revolution.
There is substantial agreement in the literature that the predominant natural input of selenium into the atmosphere is biogenic selenium gases produced by the marine biosphere
(Mosher and Duce 1987; Nriagu and Pacyna 1988; Nriagu 1989; Amouroux et al. 2001; Wen and Carignan 2007). This biogenic selenium is comprised of a variety of methylated Se(-II) species (e.g., dimethyl selenide and dimethyl diselenide). These methylated Se(-II) gases are quite volatile, with an atmospheric residence time on the order of hours. Volcanism is another significant source of natural selenium emissions. The species of selenium emitted by volcanoes has not been empirically determined, but the predominant species is thought to be H 2Se, and the exposure of these reduced volcanic plumes to the atmosphere is predicted to rapidly oxidize
H2Se to SeO2 (Wen and Carignan 2007; Floor and Román-Ross 2012). Among anthropogenic sources of selenium emissions, the two industrial processes responsible for the majority of emissions are metal refining and coal combustion, with the coal industry comprising nearly half of anthropogenic selenium emissions (Wen and Carignan 2007). The majority of the selenium emitted by coal combustion appears to be in the form of particulate Se(0) (Andren, Klein and
Talmi 1975). Volatile inorganic selenium gases are produced during coal combustion as well, though the specific species produced during this process have only been modelled from thermodynamic data (Monahan-Pendergast et al. 2008).
21 How selenium is transformed in the atmosphere is a largely unexplored area of study.
Particulate Se(0) and SeO2 transform into Se(VI) and especially Se(IV) when these species are exposed to water or aerosols in the atmosphere (Wen and Carignan 2007; Floor and Román-Ross
2012). Wen and Carignan also highlight a number of chemical reactions that are hypothesized to occur between volatile selenium gases and atmospheric sulfoxy radicals, producing transient multivalent species of selenium and sulfur (e.g., selenoselenate and selenosulfate) (Wen and
Carignan 2007). The latter suggests that selenium transformations in the atmosphere may intersect with other biogeochemical cycles. Particulate Se(0) eventually undergoes deposition, and Se(VI) and Se(IV) are precipitated by rainfall (Wen and Carignan 2007; Floor and Román-
Ross 2012).
Selenium in aquatic environments
One of the earliest reports of the distribution and speciation of selenium in marine environments established that the concentration of Se(VI) and Se(IV) displayed a depth profile similar to biologically essential trace elements (Measures and Burton 1980), as the concentrations of both oxyanions were depleted at surface waters of the Atlantic Ocean and increased with depth. Subsequent research would elucidate further details in the marine selenium biogeochemical cycle. A recurring finding is that selenium in surface waters is predominantly in the form of organic Se(-II) (operationally defined as the proportion of selenium not present as soluble Se(VI) and Se(IV)), which is rapidly depleted with depth (Cutter and
Bruland 1984; Cutter and Cutter 1995, 2001). A contrary report exists in the literature, however, documenting that surface waters near New Zealand have higher concentrations of Se(VI) and
Se(IV) than organic Se(-II) (Sherrard, Hunter and Boyd 2004).
22 The dearth of Se(VI) and Se(IV) in surface waters has been attributed to the assimilatory reduction of these oxyanions to Se(-II) by phytoplankton for incorporation into selenoproteins
(Hu et al. 1997; Baines et al. 2001; Mason et al. 2018), with Se(IV) being the preferred oxyanion for incorporation (Hu et al. 1997). Particulate organic Se(-II) and Se(0), which is a small component of the total selenium of marine environments (Mason et al. 2018), can also be directly assimilated by some bivalves (Luoma et al. 1992; Schlekat et al. 2000). Multiple lines of evidence support the hypothesis that selenium assimilation by organisms is the primary factor governing the mobility and speciation of selenium in surface waters. The distribution of organic
Se(-II)s in sites in subtropical and equatorial waters of the Pacific Ocean strongly correlates with primary productivity, bioluminescence, chlorophyll concentrations, etc. (Cutter and Bruland
1984). The concentration of selenium in Atlantic Ocean surface waters is higher in oligotrophic conditions compared to mesotrophic conditions (Cutter and Cutter 2001). Phytoplankton cell mass in subtropical and subantarctic surface waters have been found to contain significant quantities of selenium, and algal blooms in surface waters result in decreased concentrations of
Se(VI) and Se(IV) (Sherrard, Hunter and Boyd 2004).
In deeper oxic waters, the mobility and speciation of selenium is primarily driven by ocean circulation patterns. Cutter and Bruland (1984) postulated a model where organic Se(-II)s became depleted in deeper waters by the oxidative regeneration of Se(-II)s to Se(IV). Se(IV) would subsequently be oxidized to Se(VI), though the kinetics of Se(IV) oxidation to Se(VI) were thought to be substantially slower. This model was supported in later observations in subtropical and subantarctic waters of the Atlantic Ocean, where slower circulating deep waters were found to harbor higher concentrations of Se(VI) relative to waters where ocean circulation is more swift (Cutter and Cutter 1995, 2001). Additionally, the concentration of organic Se(-II)s
23 increases in deep waters where a substantial suboxic zone is present, consistent with slower rates of oxidation of organic Se(-II)s to Se(IV) (Cutter and Bruland 1984; Cutter and Cutter 1995). It is not known whether the oxidation of organic Se(-II) to Se(IV) or of Se(IV) to Se(VI) is strictly abiotic or if organisms could also mediate the oxidation of selenium oxyanions in these waters.
Little is known about the mobility and speciation of selenium in marine environments within anoxic basins. The only comprehensive studies come from the Saanich Inlet near
Vancouver Island, British Columbia, Canada (Cutter 1982) and the Black Sea (Cutter 1992).
Both studies found that the selenium present in surface waters was primarily Se(VI), with concentrations of Se(VI) decreasing with depth until the suboxic zone, where the concentration of Se(VI) fell below the detection limit and the concentrations of both Se(IV) and organic Se(-II) increased with depth. The concentration of organic Se(-II) increased throughout the suboxic and the anoxic zones, whilst the concentration of Se(IV) fell below the detection limit as the suboxic
- zone transitioned to the anoxic zone. A small quantity of inorganic H2Se and HSe was detected in the anoxic zone of the Saanich Inlet, but the bulk of the Se(-II) present was in the form of organic Se(-II)s. It isn’t clear whether the disappearance of Se(VI) and Se(IV) in the suboxic zone of the depth profile reflects abiotic or dissimilatory reduction of Se(VI) and Se(IV).
The research discussed above found that the concentration of selenium in marine environments was in the nM range. The predominant input of selenium into marine ecosystems appears to be atmospheric Se(VI) and Se(IV) deposited by rainfall and aerosols (Wen and
Carignan 2007). Of the two oxyanions, Se(IV) appears to be the predominant species of selenium delivered to marine ecosystems through precipitation (Cutter and Church 1986).
Anthropogenic emissions of volatile selenium gases appears to be a growing input of selenium into marine environments, as the total dissolved selenium content of Pacific Ocean mid-depth
24 waters has increased substantially over a 30 year interval (Mason et al. 2018). Cutter and Cutter
(2001) found that rivers can be an important input of selenium into marine ecosystems, as they estimated that the selenium flux into the Atlantic Ocean from the Amazon River basin represented approximately 16% of the total flux of selenium into the Atlantic. Currently, the flux of selenium into marine ecosystems from hydrothermal vent systems remains unconstrained, even though hydrothermal vent precipitates are often enriched in selenium (Rouxel, Fouquet and
Ludden 2004). Similarly, little is known about the mobility or speciation of selenium in hydrothermal vent fluids, particularly in alkaline hydrothermal vent systems (i.e., “white smokers”). Volatile methylated Se(-II) gases produced by phytoplankton appears to be the primary flux of selenium from marine ecosystems (Mosher et al. 1987; Mason et al. 2018).
The earliest studies of the cycling of selenium within estuarine environments found that selenium behaved conservatively, suggesting that biotic reduction and uptake of selenium wasn’t a key feature of selenium cycling in these environments (Measures and Burton 1978; Takayanagi and Wong 1984). Moreover, both studies found that selenium concentrations in estuaries were consistently in the nM range, contained a substantial fraction (~10%) of Se(IV), and that the speciation and concentration of selenium species seemed primarily dependent on riverine inputs.
Subsequent research, mainly in the San Francisco Bay estuary, has painted a much more complex biogeochemical cycle, as the speciation and mobility of selenium in estuarine environments can be profoundly impacted by anthropogenic activities and inputs.
Selenium in the South Bay area of the San Francisco Bay estuary displayed a conservative mixing pattern, and the primary input of selenium in the area was from municipal sewage waste effluent (Cutter 1989a). The predominant species of selenium were Se(VI) and organic Se(-II). The northern reach of the San Francisco Bay estuary displayed a more complex
25 pattern, as selenium inputs consisted of both riverine inputs from the San Joaquin and
Sacramento rivers and industrial effluent (Cutter 1989a). Selenium in both rivers consisted mainly of Se(VI) and organic Se(-II), though ~10% of the total selenium was Se(IV). Selenium did not display a conservative pattern of mixing. Industrial effluents along the northern reach, as well as the oxidation of organic Se(-II) and Se(IV) from resuspended particulate Se from turned over sediments, effected the speciation of selenium.
Throughout the estuary system, the concentration of total selenium was in the nM range.
The concentration of organic Se(-II) was correlated with the concentration of chlorophyll a, suggesting that biotic assimilation of selenium is a component of the estuarine selenium biogeochemical cycle obscured by bulk measurement of selenium concentrations. The importance of organisms in the cycling of selenium in estuarine environment is supported by work in the Changjiang Estuary, as phytoplankton blooms in the summer months were associated with a non-conservative mixing pattern for selenium (Chang et al. 2016). Curiously, the concentration of Se(IV) was strongly associated with industrial activities in the San Francisco
Bay. Not only was Se(IV) enriched in industrial effluents, but Se(IV) concentrations throughout the estuary system increased during times of low riverine input (Cutter 1989a; Cutter and San
Diego-McGlone 1990). This raises the possibility that Se(IV) may be a signature for industrial pollution in freshwater and estuarine environments. Subsequent work in the San Francisco Bay
Estuary found that state regulations controlling the selenium content and concentration in industrial effluents had a dramatic impact on the speciation of selenium in the estuary, as Se(IV) concentrations declined precipitously (Cutter and Cutter 2004), and the primary selenium inputs into the estuary system became riverine inputs from the San Joaquin and Sacramento rivers.
26 Particulate selenium was found to be an important component of the San Francisco
Estuary, comprising an estimated 7-13% of the total selenium concentration in the system (Cutter
1989a). Subsequent work found that the amount of particulate selenium in the estuary ecosystem was largely unaffected by state regulations, and that particulate selenium consisted about 45% organic Se(-II), 30% Se(IV) and Se(VI) that had absorbed onto metals or organic particles, and
25% Se(0) (Doblin et al. 2006). Much of this particulate selenium came from phytoplankton biomass. The concentration and composition of particulate selenium therefore is primarily determined by the speed of river and estuarine currents, as well as variation in the composition of the population of phytoplankton inhabiting the San Francisco Bay, as different species of phytoplankton display marked variability in their ability to assimilate selenium (Doblin et al.
2006). Thus, the concentration and composition of particulate selenium in the estuary is determined at least partially by biological uptake and transformation of selenium.
In freshwater systems, characterized by pH values in the range of 5.0 – 9.0, the
- 2- 2- predominant forms of Se(IV) are HSeO3 and SeO3 , Se(VI) mainly as unprotonated SeO4 , and
Se(-II) is predicted to be primarily in the form of HSe- (Sharma et al. 2014). Selenium concentrations in riverine systems range from pM to nM and is comprised of a small portion of
Se(IV), with the remainder of the selenium in the form of Se(VI) and organic Se(-II) (Yee,
Measures and Edmond 1987; Cutter 1989a, 1989b; Zhang, Feng and Larssen 2014). The main input of selenium into freshwater systems is oxidative weathering of seleniferous sediments, which is promoted at elevated pH (Yee, Measures and Edmond 1987). Small quantities of particulate organic Se(-II) and colloidal Se(0) are most likely present in freshwater columns, as has been documented in marine and estuarine environments. Most of the colloidal Se(0) would likely settle into sediments, but a proportion of the Se(0) would be oxidized to Se(IV), which
27 would be either immobilized by binding to organic matter or remain in the water column according to laboratory studies (Zhang, Zahir and Frankenberger 2004). Most of the selenium in the sediments of freshwater systems appears to be in the form of organic Se(-II) and is associated with organic matter (Martens and Suarez 1997; Oram et al. 2008), making the oxidative weathering of organic matter the predominant process governing the mobility of selenium in these environments. Some freshwater environments, particularly swamps and bogs, appear to be effective sinks for selenium, as selenium appears to become immobilized in these environmental compartments (Lidman et al. 2011).
A number of anthropogenic activities are possible inputs of selenium into freshwater environments. Irrigation of seleniferous soils can lead to the oxidative weathering and release of substantial quantities of Se(VI) into irrigation waste waters that can then be discharged into freshwater environments, as has been carefully documented in the Kesterson National Wildlife
Refuge of California (Lemly, Finger and Nelson 1993; Lemly 1994). Coal and fly ash impoundments can also be a significant input of Se(IV) into freshwater environments when these impoundments fail (Lemly 2004, 2015). Mining is an anthropogenic activity that can also be an important input of selenium into freshwater systems (Lemly 2004). Several studies have reported that the acidic conditions prevalent in aquatic environments impacted by acid mine drainage promotes the oxidative weathering of selenium to Se(IV), but not Se(VI) (Bujdoš et al.
2005; Oram et al. 2010). At surface waters of acid mine drainage impacted environments,
Se(IV) predominates, but Se(IV) appears to decrease with depth in pore waters, as the Se(IV) becomes rapidly reduced to Se(0) or Se(-II), suggesting that Se(IV) in these environments is being metabolized by organisms (Oram et al. 2010). Aquatic environments impacted by alkaline mine drainage, however, promote the oxidative weathering of selenium in sediments to Se(VI),
28 suggesting that the impacts of alkaline mine drainage in freshwater environments can only be discerned through increases in Se(VI) concentrations, rather than changes in selenium speciation
(Zhang, Feng and Larssen 2014).
Selenium in terrestrial environments
The average concentration of selenium in the Earth’s crust is estimated to be between
0.2-0.5 mg kg-1, though the distribution of selenium is highly heterogenous, with some sediment types being depleted in selenium, and others highly enriched (Lemly 2004; Fernández-Martínez and Charlet 2009). Clays, carboniferous and pyritic shales, phosphatic rocks, mineralized organic matter, and coal deposits are highly enriched in selenium, with observed concentrations of selenium ranging anywhere from 1 mg kg-1 to nearly 700 mg kg-1 (Fernández-Martínez and
Charlet 2009). Se(-II) can be found in a variety of pyrites and chalcopyrites, as Se(-II) substitutes readily with S(-II) (Masscheleyn, Delaune and Patrick 1991; Fernández-Martínez and
Charlet 2009). Se(-II), however, is most commonly found in sediments in association with organic matter, and Se(IV) is also readily immobilized by organic matter, particularly humic acids. The interaction of Se(-II) and Se(IV) with organic matter is an extremely complex geochemical process, and is reviewed elsewhere (Li et al. 2017). Se(IV) is also immobilized readily by aluminum oxides and silicates (Fernández-Martínez and Charlet 2009) and ferromanganese deposits (Stüeken 2017). Ferromanganese deposits may be the largest sink for
Se(IV) in marine sediments. Se(VI) tends to become immobilized only by phosphatic rocks, via the non-specific substitution of Se(VI) with phosphate (Fernández-Martínez and Charlet 2009).
As in other environmental compartments, the speciation of selenium in terrestrial environments is strongly governed by both sediment pH and redox conditions. Generally, aerated alkaline sediments and soils promote the formation of Se(IV) and Se(VI), making
29 selenium more mobile and bioavailable, whereas anoxic, acidic sediments and soils promote the speciation of Se(-II) and Se(0), reducing the mobility and bioavailability of selenium
(Fernández-Martínez and Charlet 2009). In a series of laboratory experiments, Masscheleyn,
Delaune, and Patrick (1990) established how the speciation of selenium in sediments is affected by specific redox conditions in association with other elements. Oxidation of Se(-II) and Se(0) to Se(IV) occurred between 0mV and 100mV, roughly concomitant with the oxidation of Fe(II) to Fe(III), and after oxidation of S(-II) to S(VI), which occurred between -200mV and -100 mV.
Oxidation of Se(IV) to Se(VI) occurred between 200mV and 300mV concurrently with the oxidation of ammonium to nitrate and dinitrogen, and slightly after the oxidation of Mn(II) to
Mn(IV). The oxidation of Se(-II) and Se(0) to Se(IV) was more rapid than the oxidation of
Se(IV) to Se(VI). The authors thus found that generally higher redox conditions promoted greater mobility for selenium, but found that incubations under more alkaline conditions (pH 9.0) resulted in less mobility under redox conditions favoring Se(IV), as Se(IV) was readily immobilized by Fe-bearing minerals. These experiments suggest that the kinetics of abiotic
Se(IV) reduction and oxidation are substantially more rapid than abiotic Se(VI) reduction and oxidation. This has been corroborated by several observational studies, which have documented that the reduction of Se(IV) in anoxic sediments is rapid, whereas Se(VI) reduction in anoxic sediments proceeds slowly (Bruggeman et al. 2005; Bruggeman, Maes and Vancluysen 2007).
The primary input of selenium into terrestrial environments appears to be the deposition and mineralization of organic matter (Oram et al. 2008), though the deposition of selenium from the atmosphere either as particulate Se(0) and Se(-II) or as Se(IV) and Se(VI) in rainwater constitute substantial inputs as well (Lidman et al. 2011). The chief means of mobilizing selenium in terrestrial compartments is oxidative weathering (Stüeken 2017). Anthropogenic
30 activities, particularly coal mining, processing, and combustion, also constitute a substantial flux of selenium from terrestrial compartments to atmospheric and aquatic compartments (Lemly
2004). Selenium can also be mobilized from sediments into groundwater through leaching
(Fernández-Martínez and Charlet 2009). The slow kinetics of abiotic Se(VI) reduction strongly suggest that the rapid reduction of Se(VI) in anoxic sediments that has been observed previously
(Oremland et al. 1989, 1991) is due primarily to the biotic reduction of Se(VI). Despite the rapid kinetics of abiotic Se(IV) reduction, the Se(0) and Se(-II) produced during Se(IV) reduction in a number of environments has been shown to be associated with organic matter, indicating that biotic reduction of Se(IV) is again the primary process driving Se(IV) reduction in sediments
(Gustafsson and Johnsson 1994; Zhang and Moore 1996).
The evolution of the selenium biogeochemical cycle through deep time
The acute sensitivity of selenium speciation to local redox conditions has made the element a promising proxy for the presence of molecular oxygen over geologic time (Stüeken
2017). Interest in this possibility was stimulated by the discovery that selenium isotopes display a pattern of mass dependent fractionation in sediments through deep time, with lighter isotopes tending to fractionate with the reduced species Se(-II) and Se(0) during reduction of Se(IV) or
Se(VI) (Stüeken et al. 2015a). As with sulfur, a number of stable selenium isotopes exist, including 74Se, 76Se, 77Se, 78Se, 80Se, and 82Se (Stüeken 2017). These stable isotopes have natural abundances of 0.87%, 9.36%, 7.63%, 23.78%, 49.61%, and 8.73%. The concentration and isotopic fractionation of selenium in organic and pyritic shales could therefore be used to infer whether Se(IV) or Se(VI) (and thus selenium oxyanion reduction) could have been present in marine environments during the Archean Eon (~4.0-2.5 billion years ago (Gya)). This would also provide meaningful constraints on the possible redox state of Archean environments.
31 This has led to the proposal by Stüeken et al. (2015) of an Archean Eon selenium biogeochemical cycle where the Paleoarchean Era (~3.6-3.2 Gya) and the Mesoarchean Era
(~3.2-2.8 Gya) selenium biogeochemical cycle was largely comprised of immobilized and biologically unavailable Se(-II) and Se(0), as determined from the relatively low abundance of selenium in marine sediments and their heavier isotopic composition compared to marine sediments of a more recent provenance. Fluvial sediments, however, showed evidence for
Se(IV) reduction, suggesting that in some local environments oxidative weathering of Se(-II) and
Se(0) to Se(IV) occurred, but that the resultant Se(IV) was rapidly reduced during fluvial transport. Marine and non-marine sediments from the Neoarchean Era (~2.7-2.5 Gya), however, displayed a significant increase in the concentration of selenium, while only fluvial sediments displayed the isotopically lighter signature consistent with increased Se(IV) reduction. These observations are consistent with a late Archean Eon selenium biogeochemical cycle where increased oxidative weathering promoted the formation of Se(IV). This Se(IV) was reduced during fluvial transport, making it likely that lacustrine and riverine environments would have had a ready supply of bioavailable selenium.
Curiously, the selenium biogeochemical cycle did not seem to change substantially in response to the Great Oxygenation Event (~2.5 Gya). Marine sediments do not display isotopic fractionation patterns consistent with marine Se(IV) or Se(VI) reduction in the Archean immediately prior to the onset of the GOE or in the Paleoproterozoic Era (~2.5-1.6 Gya)
(Stüeken, Buick and Anbar 2015; Stüeken et al. 2015a), though the researchers note that Se(IV) reduction may have occurred in shallow marine basins unlikely to be well-preserved in the geologic record. Over the course of the Neoproterozoic Era (~1.0-0.541 Gya) and the Paleozoic
Era (0.541-0.251 Gya), increasingly lighter selenium isotope fractionation patterns suggest that
32 the basic contours of the selenium biogeochemical cycle were largely identical to the modern selenium biogeochemical cycle.
This hypothesis, if supported by further investigation, implies that selenium isotope studies could provide unprecedented detail into the timing, dynamics, and heterogeneities of the growing oxidation state of Earth environments over this crucial time interval. Indeed, a number of papers have exploited precisely this potential to estimate molecular oxygen concentrations over the Neoarchean Era and the Proterozoic Eon (~2.5-0.541 Gya) (Pogge von Strandmann et al. 2015; Stüeken, Buick and Anbar 2015; Kipp et al. 2017). Selenium isotope fractionation patterns and concentration levels in ancient marine sediments have even been used to trace the euxinic and suboxic conditions that prevailed in oceans concomitant with mass extinction events over the Phanerozoic Eon (~0.541 Gya-present) (Stüeken et al. 2015b; Long et al. 2016).
Nevertheless, this model faces a substantive challenge from the fact that selenium utilization is an ancient trait that almost certainly was a feature of the last universal common ancestor (LUCA) of the Bacteria and Archaea (Mariotti et al. 2015; Weiss et al. 2016), and could have been available to the earliest forms of life at alkaline hydrothermal vents (Nitschke and Russell 2009). Moreover, there is broad agreement in the evolutionary literature that selenium utilization by organisms has decreased over time, not increased, as lineages of prokaryotes have steadily lost the ability to metabolize selenium (Zhang et al. 2006; Rother and
Krzycki 2010). Given the predicted thermodynamic (Masscheleyn, Delaune and Patrick 1991) and experimental (Masscheleyn, Delaune and Patrick 1990) data available on how readily both
Fe(II) and S(-II) (both abundant in Archean environments) immobilize Se(-II) and Se(0), we are left with two possibilities. Either organisms possess currently unknown mechanisms for
33 mobilizing unavailable species of selenium, or bioavailable Se(IV) was metabolized by Archean organisms, but did not leave a discernible signature in marine sedimentary data.
This exact contention has been advanced by Mitchell et al. (2016), who have noted that isotope fractionation patterns consistent with Se(IV) reduction and assimilation could have been muted in the Archean Eon by widespread utilization and efficient recycling of selenium by organisms in marine environments. Such a scenario would be consistent with the present evolutionary literature on selenium metabolism, as well as with the experimental data of
Masscheleyn, Delaune, and Patrick (1990) indicating that the Fe(III) present throughout the
Archean (Knoll, Bergmann and Strauss 2016) would promote the formation and stability of
Se(IV). This possibility would not invalidate the crux of the above model for Archean and
Proterozoic selenium biogeochemical cycling, as selenium isotopic patterns do convey meaningful information about the growing oxidation state of the Earth. Nonetheless, the extraordinary complexity of the modern-day selenium biogeochemical cycle, and the dominant role of biological organisms in cycling selenium through modern environments, suggests that evolutionary analysis of selenium metabolism needs to be considered in devising models for how selenium cycling may have evolved over geologic time scales.
Selenium assimilatory metabolism
Selenophosphate is the organoselenide species that is the definitive selenium substrate for the synthesis of Sec and mnm5Se2U (Glass et al. 1993), and is thought to be the source of Se(-II) for the co-factors of a number of molybdenum hydroxylases (Haft and Self 2008; Zhang et al.
2008). Selenophosphate is generated in vivo by the enzyme selenophosphate synthetase (SelD), which catalyzes the phosphorylation of Se(-II) using Se(-II) and ATP as substrates (Veres et al.
1994). The ultimate source of Se(-II) for this reaction is thought to be hydrogen selenide (H2Se
34 and/or HSe-) (Cupp-Sutton and Ashby 2016). Thus, all species of selenium that are assimilated by organisms must first be transformed to inorganic Se(-II).
Transformation of selenium to hydrogen selenide and delivery to selenophosphate synthetase
Selenoamino acids can be metabolized by organisms in all three domains of life to
- generate inorganic Se(-II). H2Se and/or HSe is thought to be generated from the two most common non-proteinogenic selenoamino acids, SeMet and methylselenocysteine, by identical pathways as their sulfur-containing analogs (Cupp-Sutton and Ashby 2016; Tobe and Mihara
2018). Sec, however, is recycled for selenium assimilation by the specific enzyme selenocysteine -lyase, which catalyzes the degradation of Sec to inorganic Se(-II) and alanine in
Bacteria (Chocat et al. 1985; Mihara et al. 1999) and Archaea (Stadtman 2004).
It has not yet been rigorously demonstrated that cultivation of either bacteria or archaea with Se(VI) stimulates selenium metabolic pathways. However, Se(VI) is widely detoxified by prokaryotes (discussed below) in a 2 electron reduction to Se(IV), and Se(IV) is the most frequently utilized selenium source in cell cultures in all three domains of life (Tobe and Mihara
2018). This suggests that at minimum Se(VI) can indirectly stimulate selenium assimilation.
Glutathione has been discussed as a potential endogenous Se(IV) reductase for both selenium assimilation and detoxification (Nickerson and Falcone 1963; Kessi and Hanselmann 2004).
Glutathione spontaneously reduces Se(IV) to selenodiglutathione, which glutathione reductase can subsequently reduce to generate the selenopersulfide-containing molecule glutathioselenol.
Glutathioselenol can then spontaneously decompose to Se(0) and reduced glutathione, or can be further reduced to Se(-II) via the glutaredoxin and thioredoxin systems (Tobe and Mihara 2018).
It should be noted that in E. coli it was thioredoxin reductase, not glutathione reductase, that was found to be essential for the production of the Sec-containing formate dehydrogenase H of the
35 formate hydrogen lyase complex (Takahata et al. 2008), indicating that thioredoxin reductase is the predominant means of generating Se(-II) from Se(IV) in at least some bacteria. It has not yet been determined whether Se(0) can undergo assimilatory reduction, though the ability of bacteria to utilize Se(0) as a terminal electron acceptor in anaerobic respiration suggests that the possibility should be considered.
There is little information available in the literature as to how cells deliver volatile species of Se(-II) to SelD. There is evidence in eukaryotes that the Se(-II) produced by the selenocysteine -lyase enzyme remains bound to the active site in the form of selenopersulfide, and coimmunoprecipitation assays suggest that the enzyme interacts directly with SelD (Tobe and Mihara 2018). In bacteria, a number of putative cysteine desulfurase NifS-like homologs have been identified in vitro as possible selenium delivery proteins to SelD during the recycling of Sec (Lacourciere and Stadtman 1998; Lacourciere et al. 2000). A number of proteins have also been identified in vitro as candidate delivery proteins for Se(-II) produced by the reduction of Se(IV) with glutathione, and these proteins include rhodanese domain-containing sulfurtransferases, 3-mercaptopyruvate sulfurtransferases, and glyceraldehyde-3-phosphate dehydrogenases (Ogasawara, Lacourciere and Stadtman 2001; Ogasawara et al. 2005). While in vivo identification of a selenium binding and delivery system for SelD has yet to be determined in either Bacteria or Archaea, the result of in vitro biochemical research strongly suggests that cells contend with the volatility of Se(-II) by sequestering it in organic molecules (i.e., selenopersulfide, selenodigluathione) until delivery to SelD.
36 Selenophosphate synthetase
No SelD from Archaea has been characterized either biochemically or structurally, though it has been confirmed that SelD is required for Sec synthesis in Sec utilizing archaea
(Stock, Selzer and Rother 2010; Hohn et al. 2011). However, SelD homologs from a number of bacteria have been studied extensively. Random mutagenesis first demonstrated that the selD gene was required for the synthesis of Sec and mnm5Se2U in E. coli (Leinfelder et al. 1990).
Subsequent overexpression and biochemical characterization of SelD established that the enzyme required Mg2+ and ATP for activity, and that the products of the reaction were some species of selenium bound to phosphate, and orthophosphate (Ehrenreich et al. 1992). This suggested that
SelD did not participate directly in the synthesis of Sec and mnm5Se2U, but rather provided a selenium donor for the synthesis of these selenium-containing macromolecules. Further biochemical characterization of purified SelD demonstrated that the selenium species that serves as the substrate was in fact Se(-II), and that the E. coli SelD had a Km for Se(-II) of 46 µM
(Veres et al. 1992a). SelD showed low activity with sulfide as a substrate, confirming that SelD specifically participates in selenium metabolism. The identity of the Se(-II)-containing product of SelD was then shown to be selenophosphate (Glass et al. 1993), demonstrating that selenophosphate was the actual selenium donor for Sec and mnm5Se2U synthesis.
Further biochemical characterization of the reaction sequence definitively established the reactants and products of the reaction catalyzed by SelD (Veres et al. 1994) (Reaction 1):
Se(-II) + ATP → SePO3 + Pi + AMP
It was also determined that a K+ was required for SelD activity and that other purine and pyrimidine triphosphates could not be used by SelD to drive selenophosphate synthesis. The analysis of the reaction sequence established that the -PO4 of the ATP molecule phosphorylated
37 Se(-II) to generate selenophosphate, and that the orthophosphate was formed from the -PO4 during ADP hydrolysis. Selenophosphate, orthophosphate, and AMP were formed in a 1:1:1 stoichiometry over the course of the reaction. Stringent attempts to prevent exposure to O2 during the purification of SelD could lower the Km of the enzyme for Se(-II) by up to an order of magnitude.
Early attempts to elucidate the structural basis for the reaction catalyzed by SelD involved site-directed mutagenesis studies in E. coli. These studies focused on a glycine-rich N- terminal region of SelD that was thought to serve as a possible ATP binding site. This work implicated the Cys17 (Kim, Veres and Stadtman 1992) and Lys20 (Kim, Veres and Stadtman
1993) residues (numbering from the E. coli SelD protein sequence) as being essential for catalysis, as mutant SelD enzymes without these residues could successfully bind ATP homologs, but displayed no selenophosphate synthetase activity. The hypothesis that the Cys17 residue was essential for catalysis was complicated by the arrival of genomic data, which revealed that a number of SelD homologs in all three domains of life contain a Sec residue at a homologous position, rather than Cys (Guimarães et al. 1996). The presence of Sec in SelD was hypothesized to be an autoregulatory mechanism to ensure that cells had an adequate supply of selenium before investing energy in synthesizing selenoproteins. Purification of the Sec- containing SelD from Haemophilus influenzae revealed that the Km for Se(-II) in SelD was comparable to the Cys-containing SelD of E. coli (Lacourciere and Stadtman 1999), consistent with the hypothesis that Sec was incorporated into SelD as a form of autoregulation. These residues therefore likely serve as Se(-II) binding sites, rather than participating in catalysis.
Unprecedented insights into how SelD functions to synthesize selenophosphate came from crystal structures of the Sec-containing SelD of Aquifex aeolicus (Itoh et al. 2009) and a
38 Cys17Ser mutant SelD from E. coli (Noinaj et al. 2012). The crystal structures of these enzymes produced significant consensus in models of how SelD synthesizes selenophosphate from Se(-II) and ATP. Both crystal structures revealed that SelD exists in vivo as a homodimer, with the two subunits held together by hydrophobic interactions. Two Mg2+ cations are bound by Asp residues at the SelD active site, with a single Mg2+ cation binding to each subunit. A single polar amino acid at the active site forms hydrogen bonds with the adenosine moiety of ATP (Ser in the mutant E. coli SelD and Thr in the SelD of Aquifex aeolicus). Both models agree that the binding of Mg2+, K+, and ATP stimulate conformational changes in the N-terminal region of the first subunit of SelD from a disordered conformation to an ordered conformation. This positions the homologous Cys and Sec residues as well as the essential Lys residue at the SelD active site.
Neither structure included SelD bound to Se(-II), though Itoh et al. (2009) propose that
Se(-II) likely binds to Sec/Cys, which produces perselenide or selenosulfide, respectively. The formation of the perselenide/selenosulfide intermediate would then promote a conformational change driving the phosphorylation of Se(-II) from the PO4 of ATP. The authors propose that an unknown reducing agent would break the perselenide/selenosulfide bond, releasing selenophosphate and stimulating ADP hydrolysis by a nucleophilic attack from a water molecule positioned by a conserved Asn residue. A significant disagreement between the models proposed by Itoh et al. (2009) and Noinaj et al. (2012) concerns whether or not Se(-II) binding precedes ATP binding or vice versa. Itoh et al. propose that Se(-II) binds to Sec/Cys after metal and ATP binding triggers conformational changes in the N-terminal Gly-rich region. Noinaj et al. note that this is highly improbable if Se(-II) is delivered to SelD via a selenium delivery protein, as the positioning of the Sec/Cys residue in the active site would make an interaction between SelD and a delivery protein unfavorable.
39 Selenocysteine synthesis and incorporation into selenopolypeptides
The first component of the Sec synthesis architecture to be elucidated was the product of the selC gene, which encodes a tRNA molecule (Leinfelder et al. 1988b). Sequence analysis of the selC gene indicated that the tRNA molecule contains all the structural features characteristic of other tRNA molecules, including a D arm, T arm, variable arm, and a CCA acceptor stem.
The predicted structure of SelC also contained an anticodon loop complimentary to the UGA codon. However, the predicted SelC tRNA structure also displayed a number of unique structural characteristics. tRNAs for the canonical twenty proteinogenic amino acids are 88 nucleotides long, whilst SelC is 95 nucleotides long. The SelC CCA acceptor stem appeared to be 8 nucleotides long, whereas other tRNA molecules invariably have CCA acceptor stems of 7 nucleotides. The variable arm also seemed to be substantially longer than other tRNA molecules. These features were later confirmed when the solution structure of SelC was determined (Baron et al. 1993). Curiously, SelC purified from E. coli was charged with a Ser residue, rather than a Sec residue, suggesting that mechanisms must exist for converting the Ser residue to a Sec residue (Leinfelder et al. 1988b).
The enzyme mediating the conversion of seryl-tRNASec to selenocysteyl-tRNASec was shortly thereafter purified from E. coli (Forchhammer et al. 1991). This selenocysteine synthase
(SelA) was found to be a protein of ~50 kDa, though the molecular weight of native SelA was
~600 kDa, suggesting that SelA forms a multi-subunit homomeric complex in vivo. SelA was also found to contain a pyridoxal 5’-phosphate co-factor. Electron microscopy later confirmed that SelA indeed forms a large homomeric complex, as native SelA forms a decameric ring with fivefold symmetry around a central pentagonal cavity (Engelhardt et al. 1992). The fundamental unit of the SelA complex is a dimer of two SelA subunits.
40 The replacement reaction catalyzed by SelA was elucidated shortly after the enzyme was purified (Forchhammer and Böck 1991). After a seryl-tRNASec molecule binds to SelA, a
Schiff base forms between the -amino group of the Ser residue and the formyl group of the pyridoxal 5’-phosphate co-factor. This results in a elimination of the hydroxyl functional group from the Ser residue, producing an aminoacrylyl-tRNASec intermediate. Selenophosphate then serves as the selenium donor to replace the hydroxyl functional group with a selenol functional group. Analysis of the kinetics of seryl-tRNASec to selenocysteyl-tRNASec conversion indicated that one seryl-tRNASec molecule binds to two SelA dimers, producing a 1:2 stoichiometry. However, later analysis of seryl-tRNASec binding to SelA using fluorescence anisotropy would demonstrate that one seryl-tRNASec molecule binds to each one of the subunits in the SelA decamer, producing an assembly stoichiometry of 1:1 for the SelC:SelA complex
(Manzine et al. 2013). Further biochemical characterization of SelA demonstrated that the pyridoxal 5’-phosphate co-factor is coordinated by a conserved Lys residue in SelA proteins
(Tormay et al. 1998). The Km of SelA for selenophosphate was found to be 0.3 M. The sulfur analogue, thiophosphate, could also bind efficiently to SelA (with a Km of 4.0 M), leading the authors to hypothesize that the use of selenophosphate as a selenium donor may have evolved as a way of ensuring efficient discrimination between selenium and sulfur in vivo.
As with SelD, recent structural characterization of the SelA:SelC complex has substantially expanded our knowledge of the physiology of Sec synthesis in bacteria. A crystal structure for the decameric SelA complex in association with SelC has been obtained from A. aeolicus (Itoh et al. 2013). Each SelA subunit is composed of an N terminal domain, a disordered linker region, a core domain, and a C terminal domain. The N terminal domain of is positioned by the linker region over the central pentagonal cavity of the decamer. Both the
41 pyridoxal 5’-phosphate co-factor and the coordinating Lys residue are located in the core domain. Each SelA subunit contains half of the active site for the dimer that comprises the functional unit of the decameric complex. While the dimer alone can efficiently bind to SelC, decamerization is required for the conversion reaction to proceed. Thr191, Thr192, Asp199, and
Tyr220 (numbering adopted from the A. aeolicus primary sequence) residues are essential for the dimer-dimer interactions promoting decamerization (Itoh et al. 2013). Decamerization is driven by alterations in the secondary structure of each monomer at the interface of two dimers (Itoh et al. 2014).
The conversion reaction requires a SelC molecule to interact with two dimers (Itoh et al.
2013). One dimer unit positions SelC, whilst another dimer mediates the conversion of seryl- tRNASec to selenocysteyl-tRNASec. One subunit within a dimer interacts with the D arm and the
T arm of SelC at the N terminal domain. Nucleotides unique to the SelC D arm allow for efficient discrimination between SelC and other tRNA molecules. The C terminal domain of the other subunit then positions the CCA arm of SelC over the catalytic site of the other dimer for conversion. Thiosulfate was used as an analog of selenophosphate, and it appears that a dimer is also required for selenophosphate to bind to the SelA complex. Selenophosphate appears to bind to an Arg86 residue in the linker region of one subunit and to Arg312 and Arg315 residues in the C terminal region of the other. Analysis of the A. aeolicus SelA:SelC complex indicates that the
Lys residue coordinating the pyridoxal 5’-phosphate co-factor may also directly participate in the
elimination reaction step, serving as a proton donor for the removal of the hydroxyl functional group of the seryl-tRNASec molecule (Itoh et al. 2014).
It has recently been shown that SelD interacts directly with the SelA:SelC complex to deliver selenophosphate for Sec synthesis (Silva et al. 2015). This is the first report that SelD
42 functions as a selenium delivery protein during the assimilation of selenium-containing macromolecules. The SelA decamer, SelC, and SelD forms a 1.3 MDa ternary complex with a stoichiometry of one SelA decamer, ten SelC tRNA molecules, and five SelD dimers. The dissociation constant of ternary complex formation indicates that formation is driven by transient interactions. The N terminal Gly-rich region of SelD is required for the formation of the complex. The SelA decamer and ten SelC molecules must associate together before SelD can interact to form the ternary complex.
The first indication that specific mechanisms existed for the incorporation of Sec into selenopolypeptides came when mutagenesis of the E. coli selB gene produced a mutant strain that built-up selenocysteyl-tRNASec and could not synthesize selenoproteins (Leinfelder,
Stadtman and Böck 1989). Sequence analysis of the selB gene proved consistent with the hypothesis that the SelB protein participated in the incorporation of Sec into selenopolypeptides.
The predicted N terminal amino acid sequence of selB showed substantial homology with the initiation factor 2 (IF-2) and elongation factor Tu (EF-Tu) proteins involved in translation
(Forchhammer, Leinfelder and Böck 1989; Kromayer et al. 1996), including a conserved G box binding motif and Mg2+ coordination sphere (Hilgenfeld, Böck and Wilting 1996) characteristic of GTPases. The primary sequence of the translation factor indicated that the most striking difference between EF-Tu and the N terminal region of SelB was that SelB lacked any sequence features for interacting with Elongation Factor Ts (EF-Ts), and therefore did not require a nucleotide exchange factor to promote the dissociation of GDP after GTP hydrolysis (Hilgenfeld,
Böck and Wilting 1996). Purification of SelB confirmed that it possesses GTP and GDP binding activity with a Kd for GTP of 1.7M and a Kd for GDP of 10M (Forchhammer, Leinfelder and
Böck 1989). GTP and GDP both bind to SelB with a 1:1 stoichiometry.
43 The SelB protein was also shown to be able to bind selenocysteyl-tRNASec, but not seryl- tRNASec, in a GTP independent manner, suggesting that part of the specificity for SelC is mediated by the selenol group of the Sec residue (Forchhammer, Rücknagel and Böck 1990).
EF-Tu, in comparison, cannot bind efficiently to either selenocysteyl-tRNASec or seryl-tRNASec
(Förster et al. 1990). While the selenol group participates in the specificity of SelC for SelB, other features of the tRNA molecule are also vital for recognition of SelC. Deletion of the 8th nucleotide in the CCA arm of SelC to mimic the canonical twenty tRNA molecules rendered
SelB incapable of binding selenocysteyl-tRNASec, while promoting the efficient binding of SelC to EF-Tu regardless of whether the tRNA molecule was aminoacylated with either a Ser or Sec residue. (Baron and Böck 1991). SelB in E. coli was shown to be expressed in vivo with ribosomes at a ratio of 1:18 and a ratio of 1:180 with respect to EF-Tu (Forchhammer,
Rücknagel and Böck 1990).
The final element required for Sec incorporation into selenopolypeptides is found in the secondary structure of selenoprotein mRNA itself. This was demonstrated when mutating the
TGA codon of the E. coli formate dehydrogenase H (FdhH) gene, fdhF, to either a TGC or TGT codon still resulted Sec incorporation in FdhH (Zinoni et al. 1987) (albeit at reduced levels).
Subsequently, a region of approximately forty nucleotides downstream from the UGA codon of the fdhF mRNA was shown to be indispensable for Sec incorporation into FdhH (Zinoni, Heider and Böck 1990). Curiously, the predicted secondary structure of the mRNA over this region indicated that a stem-loop structure was positioned immediately downstream from the UGA codon. A predicted stem-loop structure in the fdhG mRNA from the formate dehydrogenase N
(FdhN) catalytic subunit was also shown to be indispensable for Sec incorporation (Berg, Baron and Stewart 1991). This stem-loop is referred to as the Sec insertion sequence (SECIS) element.
44 The ability of SECIS elements to facilitate Sec incorporation appeared to be sequence specific, as some mutating some nucleotides in the element could substantially reduce rates of Sec incorporation (Berg, Baron and Stewart 1991; Heider, Baron and Böck 1992). The loop region of the SECIS element was particularly sensitive to mutations (Heider, Baron and Böck 1992).
The importance of the SECIS element in Sec incorporation began to be elucidated when in vitro assays demonstrated that the SelB protein binds both the fdhF SECIS element (Baron,
Heider and Böck 1993) and the fragments of the fdhF (Baron, Heider and Böck 1993) and fdhG elements (Ringquist et al. 1994) containing only the apical loop. Moreover, SelB binds to
SECIS elements with a high affinity, as the Kd of the E. coli SelB protein for the fdhF and fdhG
SECIS elements has been estimated to be in the nM range (Hüttenhofer, Westhof and Böck
1996; Thanbichler, Bock and Goody 2000). SelB could successfully bind to fdhF or fdhG mRNA SEICS elements even when the mRNA had been pre-saturated with ribosomes (Ringquist et al. 1994). In vitro assays also confirmed that SelB binding to SECIS elements in the presence of GTP and selenocysteyl-tRNASec resulted in the formation of a quaternary complex (Baron,
Heider and Böck 1993).
This resulted in an initial model for Sec incorporation wherein the SECIS element of the mRNA transcript functioned to recruit the SelB-GTP-selenocysteyl-tRNASec ternary complex to the transcript. GTP hydrolysis would occur in response to the correct codon-anticodon interaction between the UGA codon and the anticodon loop of SelC, promoting the dissocation of the ternary complex from the mRNA transcript so translation of the remainder of the transcript could proceed. Work done using synthetic mRNA transcripts of various lengths refined this model by demonstrating that ribosome binding to transcripts could not occur when SECIS elements were within 16 nucleotides of the ribosomal P site (Hüttenhofer, Heider and Böck
45 1996). This demonstrated that SelB must recognize the SECIS element prior to the element entering the ribosome track, as the inability of the ribosome to bind transcripts with SECIS elements near the P site suggests that SelB is required to stimulate conformational changes i to facilitate successful translation.
The solution structures for both the fdhF and fdhG SECIS elements from E. coli have been solved (Hüttenhofer, Westhof and Böck 1996). Both SECIS elements are approximately 40 nucleotides long and consist of two domains: a base stem composed of an extended helix and an apical tetraloop. Curiously, the UGA codon is part of the stem in both SECIS elements. There is little sequence homology between the fdhF and fdhG SECIS elements, but nevertheless a guanine and uridine base, G23 and U24 were found to interact directly with SelB. Subsequent work defined a minimal region of seventeen nucleotides (including the apical loop) in the fdhF
SECIS that was indispensable for successful binding to SelB (Klug et al. 1997; Liu et al. 1998;
Liu, Reches and Engelberg-Kulka 1999; Li, Reches and Engelberg-Kulka 2000), including a distinctive uracil base that “bulges” from the extended helix stem (Li, Reches and Engelberg-
Kulka 2000; Fourmy, Guittet and Yoshizawa 2002). NMR spectroscopy of both SECIS elements determined that the elements are A-form helices (interrupted by the bulged U nucleotide), and determined that an additional guanine nucleotide G22, was also important for
SECIS recognition by SelB (Fourmy, Guittet and Yoshizawa 2002). Both G22 and G23 bases are in the major groove of the helix, whilst the neighboring U24 base is in the minor groove. This conformation exposes the phosphate backbone of both elements near the apical loop, suggesting that a number of the interactions critical for recognition between SelB and SECIS elements involve electrostatic interactions with the phosphate backbone, in addition to base specific interactions.
46 This is consistent with the results of detailed study of SECIS elements from other bacteria. Predicted SECIS elements from Eubacterium acidaminophilum revealed that the elements displayed substantial variation in the length of the apical loop, distance of the apical loop from the UGA codon, and substantial variation in whether a bulge was present in the base stem and the extent to which the bulge disrupted the secondary structure (Gursinsky et al. 2000).
NMR spectroscopy of the Mo. thermoacetica NAD+-dependent formate dehydrogenase fdhA
SECIS element revealed that the element is comprised of nineteen nucleotides and forms a hairpin structure with a base stem lacking irregular bulges and an apical pentaloop (Beribisky et al. 2007). A guanine and uracil base, G9 and U10 are found at the top of the pentaloop and are exposed to the solvent.
Comparing the NMR spectra of unbound Mo. thermoacetica SECIS elements and elements bound by the Mo. thermoacetica SelB confirmed that SelB binding drives conformational changes in the SECIS structure. The U10 base rotates from the major groove of the mRNA to the minor groove, driving a conformational change in the G9 base that promotes a specific interaction between the guanine base and SelB. The A helix stem structure undergoes a number of conformational changes in response to SelB binding. A neighboring uracil base, U12, bulges out of the stem in a manner that is reminiscent of the E. coli unbound SECIS elements. A guanine and cytosine base pair that caps off the stem changes from a non-canonical base pair to a
Watson-Crick base pair. On the whole, detailed study of SECIS elements have revealed that
SelB recognition of SECIS elements is far more subtle than simply recognizing conserved bases in bacteria. A loop structure is essential for SelB recognition, but the majority of SelB contacts with the element come from non-specific interactions with the phosphate backbone. Thus the only constraints on the evolution of SECIS elements in bacteria appear to be to maintain a loop
47 structure that will expose the phosphate backbone with a single guanine base in the apical loop that can interact directly with SelB.
Subsequent research on the physiological function of SECIS elements in bacteria have substantially refined the initial model that these elements function exclusively to tether the SelB-
GTP-selenocysteyl-tRNASec complex to mRNA transcripts for Sec incorporation. SELEX experiments produced a number of synthetic RNA aptamers that could bind to the E. coli SelB as efficiently or more efficiently than native E. coli fdhF SECIS elements, but could not mediate
Sec incorporation in vitro (Klug et al. 1997). This suggested that the SECIS element participates in Sec incorporation in a more profound way than as a mere anchor. Research has also demonstrated that these elements have physiological functions that extend beyond Sec incorporation. Translation of the FdhH selenoprotein was monitored in E. coli in the absence of selenium, and a downstream cytosine base adjacent to the UGA codon and a six nucleotide stretch immediately upstream of the UGA codon were found to be essential for preventing the erroneous readthrough of the UGA codon as a Trp codon (Liu, Reches and Engelberg-Kulka
1999). Additionally, a SECIS-like element was found in the 5’ untranslated region of the E. coli selAB operon that efficiently recruited SelB (Thanbichler and Böck 2002). During conditions of selenium deficiency, this SECIS-like element competed effectively with selenoprotein mRNA transcripts, reducing the ability of E. coli to translate selenoproteins when a cellular supply of
Sec is unavailable. Thus, at least some bacteria exploit SECIS elements as a level of translational control in response to the environmental availability of selenium.
The sequence analysis of the SelB protein immediately suggested a natural division of labor. The N terminal region, comprising three domains homologous to EF-Tu, would bind to
GTP and selenocysteyl-tRNASec, whilst the distinctive C terminal region unique to SelB would
48 bind to SECIS elements of selenoprotein mRNA transcripts. Expression of the N terminal and C terminal fragments of the E. coli SelB protein confirmed that these fragments have the predicted biochemical activity (Kromayer et al. 1996; Tormay, Sawers and Böck 1996). Curiously, only the latter half of the C terminal fragment is required for SECIS recognition (Kromayer et al.
1999; Li, Reches and Engelberg-Kulka 2000). A crystal structure of the C terminal fragment from Mo. thermoacetica revealed that this fragment is L-shaped and actually consists of four winged helix domains (Selmer and Su 2002), which are common structural domains in DNA and
RNA binding proteins. SelB thus consists of seven distinct domains. Comparison of bacterial
SelB amino acid sequences found that the sixth and seventh domains contain a number of conserved basic Arg and Lys residues, and a number of these residues (in addition to a Ser residue) had bound sulfate ions in the crystal structure, consistent with predictions made from
NMR spectra that recognition of SECIS elements by SelB largely involve non-specific interactions with the phosphate backbone. A salt bridge was found linking the fifth and sixth domains of the fragment, suggesting that dissolution of the salt bridge, stimulated by binding to a
SECIS element, could drive conformational changes in SelB necessary for Sec incorporation.
Crystal structures of the sixth and seventh domains of the Mo. thermoacetica SelB alone
(Yoshizawa et al. 2005) and all four of the winged helix domains (Ose et al. 2007) bound to the fdhA SECIS element confirmed that a number of the conserved basic residues in the seventh domain, as well as the Ser residue, form hydrogen bonds with the mRNA phosphate backbone.
Consistent with NMR spectra, the G9 guanine base was coordinated by a number of Arg and Leu residues via hydrogen bonds and van der Waals interactions in both structures. Both crystal structures found that only the seventh domain interacted directly with the SECIS element, though
NPR spectra indicate that the sixth domain does form transient interactions that would be
49 unlikely to be preserved during crystallization (Beribisky et al. 2007). While the seventh domain did not seem to undergo conformational changes in response to binding to the SECIS element, the linker region between the fifth and sixth domains contracted to form an acute angle (Ose et al. 2007). This gave the four domains a V-shaped conformation. This conformational change appeared to promote an interaction with the 16S rRNA of the 30S ribosomal subunit.
As with other components of the Sec synthesis and incorporation architecture, the crystal structure of the A. aeolicus full-length SelB protein provided substantial insights into how SelB functions to mediate Sec incorporation (Itoh, Sekine and Yokoyama 2015). The crystal structure confirmed that SelB does indeed consist of seven domains, with three domains homologous to the three domains of the EF-Tu protein and four winged helix domains unique to SelB.
Moreover, linker regions exist between domains 1 and 2, 3 and 4, and 5 and 6. GTP was shown to bind SelB at domain 1, with multiple amino acid residues interacting directly with the guanine
2+ base. A Ser residue and a Mg cation coordinate the -PO4 of GTP at the first GTPase switch.
Selenocysteyl-tRNASec molecules bind SelB at the intersection of domains 1 and 2 at an aminoacyl binding pocket, with a conserved Arg residue (Arg241 in the A. aeolicus SelB protein sequence) in the 2nd domain interacting specifically with the selenol group of Sec. Indeed, electrostatic interactions with the selenol group was shown to be the means by which SelB assures specificity for selenocysteyl-tRNASec over seryl-tRNASec. Domain 3 interacts with the T and variable arms of SelC. Selenocysteyl-tRNASec also interacts with the first GTPase switch. A conserved Phe residue (Phe45 in A. aeolicus) from the first GTPase switch forms an electrostatic interaction with the selenol group of Sec and a conserved Arg residue (Arg30 in A. aeolicus) interacts with SelC. In striking contrast to crystallographic studies of the C term fragment from
Mo. thermoacetica SelB, the fourth domain (first winged helix domain) was found to be highly
50 disordered. This provides strong evidence that SelB binding to GTP and selenocysteyl-tRNASec stimulates conformational changes in the C terminal fragment that promotes a more efficient interaction with the 70S ribosome.
Unlike Sec synthesis, Sec incorporation involves a number of transient interactions, and these interactions are difficult to probe with traditional techniques in structural biology.
Therefore, dissection of the kinetics of SelB associations with GTP, GDP, selenocysteyl- tRNASec, and SECIS elements, and dynamic structural biology of Sec incorporation at ribosomes have been indispensable for formulating a full model for how Sec incorporation occurs in bacteria. A kinetic study of SelB binding to GTP, GDP, selenocysteyl-tRNASec, and the E. coli fdhF SECIS element confirmed that SelB has an affinity for GTP that is several orders of magnitude greater than GDP (Thanbichler, Bock and Goody 2000). Calculation of the association and dissociation constants of GTP and GDP binding to SelB established that this affinity stems from the rapid dissociation of GDP from SelB, consistent with the finding that
SelB does not require a nucleotide exchange factor to remove GDP during Sec incorporation.
The affinity of SelB for the fdhF SECIS element was increased substantially by binding to GTP and selenocysteyl-tRNASec, and additionally increased the stability of the SelB-SECIS complex.
This is consistent with the model that GTP and selenocysteyl-tRNASec binding by SelB precedes recognition of SECIS elements and drives conformational changes in SelB that facilitate Sec incorporation. Further evidence that GTP binding stimulates conformational changes in SelB was provided by isothermal titration calorimetry (Paleskava, Konevega and
Rodnina 2012). This work demonstrated that conformational changes occurred in approximately
43 amino acids in SelB in response to the binding of GTP, and hydrolysis of GTP to GDP stimulates conformational changes in 25 amino acids. Thus, GTP binding to SelB appears to
51 stimulate conformational changes promoting associations with the ribosome, whilst GTP hydrolysis promotes conformational changes that lead to the dissociation of SelB from the ribosome after Sec incorporation occurs.
~1,000,000 cryo-EM images of synthetic mRNA including a UGA codon and an fdhF
SECIS element bound by ribosomes and either unbound or bound by the SelB-GTP- selenocysteyl-tRNASec ternary complex were classified into six distinct structural intermediates along the pathway of Sec incorporation using extensive hierarchical computational classification
(Fischer et al. 2016). These images, in conjunction with molecular dynamics simulations, provided unprecedented details as to how SelB interacts with the SECIS element and ribosome to drive Sec incorporation into selenopolypeptides. Cryo-EM images of the initial complex, with the ternary SelB-GTP-selenocysteyl-tRNASec complex unbound to the SECIS element, confirmed that the three EF-Tu like domains of SelB interact with selenocysteyl-tRNASec as revealed in the full-length crystal structure of the A. aeolicus SelB protein bound to GTP and selenocysteyl-tRNASec (Itoh, Sekine and Yokoyama 2015). The initial binding of the ternary complex (at the 7th domain of SelB) to the SECIS element stimulated two universally conserved adenine bases in the 16S rRNA of the 30S subunit (A1492 and A1493 in the E. coli 16S rRNA sequence) to adopt a “flipped-in” conformation that opened up the aminoacyl site for selenocysteyl-tRNASec. The positioning of the selenocysteyl-tRNASec into the aminoacyl site was also promoted by a shift in the 30S lower ribosomal subunit from the 50S subunit towards the 2nd domain of SelB.
Once the correct codon-anticodon interaction occurs between the UGA codon and SelC, multiple conformational changes occur in the ribosome to promote the activation of GTPase activity by SelB. The A1492 and A1493 adenine bases adopt a “flipped-out” conformation that
52 closes the aminoacyl site and interact with the minor groove of the 1st and 2nd positions of the
SelC anticodon. The elbow of SelC, which is initially docked onto the sarcin rich loop (SRL) of the 23S rRNA in the 50S upper ribosomal subunit, begins to be moved along the 23S rRNA molecule in a relay until the 1st domain of SelB interacts with SRL. Both conformational changes in the 30S and 50S ribosomal subunits are necessary for GTPase activation to occur.
This pathway to GTPase activation is broadly similar with that observed for EF-Tu.
Furthermore, the pathway of GTP hydrolysis in SelB was also identical for that of EF-Tu.
GTPase activity promoted peptide bond formation between the Sec residue and the upstream fMet residue on SelC at the aminoacyl site and the dissociation of SelB from the SECIS element.
Sec synthesis differs markedly between Bacteria and Archaea. The only component that is shared between the two domains is SelC. The first indication that Sec synthesis in archaea involves a Sec-specific tRNA came from the genome of the archaeon Methanococcus jannaschii
(later renamed Methanocaldococcus jannaschii (Graham et al. 2001)) (Bult et al. 1996). The predicted structure of the Me. jannaschii SelC molecule confirmed that the putative tRNA had a typical D arm, T arm, variable arm, anticodon loop, and CCA arm (Ioudovitch and Steinberg
1999). Unlike bacterial SelC, the predicted archaeal SelC contained distinct structural features that are more characteristic of eukaryotic SelC molecules. These include a 9 nucleotide base pair long CCA arm (compared to the eight nucleotide base pairs of bacterial SelC or the seven nucleotide base pairs of canonical tRNA molecules), a shortened T stem (4 base pairs long compared to the 5 base pairs typical for bacterial SelC and canoncial tRNA molecules), and a longer D stem (six or seven base pairs long compared to the four base pairs found in bacterial
SelC). The structure of the Me. jannaschii SelC was subsequently solved and confirmed the predicted structure, and the tRNA molecule was shown to participate in Sec synthesis and
53 incorporation (Rother et al. 2000). SelC homologs have subsequently been identified in all other
Sec utilizing archaea (Mariotti et al. 2016; Rother and Quitzke 2018).
As is the case with Sec synthesis in bacteria, archaeal SelC is initially charged with a Ser residue to yield seryl-tRNASec. It was initially assumed that a homolog of the bacterial SelA would be utilized to convert seryl-tRNASec molecules to selenocysteyl-tRNASec in archaea.
However, no obvious SelA homologs were observed in archaeal genomes and attempts to biochemically characterize a putative SelA homolog in Me. jannaschii demonstrated that the enzyme cannot mediate the conversion of seryl-tRNASec to selenocysteyl-tRNASec (Kaiser et al.
2005). The first step in the elucidation of the Sec synthesis architecture in archaea came with the discovery of an O-phosphoseryl-tRNASec kinase (PSTK) (Carlson et al. 2004). The mouse
PSTK was purified and characterized and found to mediate the phosphorylation of the Ser residue of seryl-tRNASec to yield O-phosphoseryl-tRNASec. This reaction is coupled to ATP hydrolysis, and Mg2+ is an essential co-factor. Homologs of PSTK were found exclusively in
Sec utilizing eukaryotes and archaea. Subsequently, the PSTK homolog of Me. jannaschii was shown to catalyze the identical reaction in vitro (Kaiser et al. 2005).
The Me. jannaschii PSTK has been thoroughly characterized biochemically (Sherrer,
O’Donoghue and Söll 2008). The enzyme was found to have a high affinity for seryl-tRNASec with a Km of 40 nM. Curiously, PSTK demonstrated equal affinity for SelC regardless of
Sec whether the tRNA was charged or uncharged. PSTK had a Kd of 53.3 nM for seryl-tRNA and a Kd for uncharged SelC of 39.4 nM. This suggested that PSTK recognizes the SelC molecule rather than the Ser residue. Curiously, PSTK had little affinity for ATP, with a Km of 2.6 mM.
Consistent with this observation, PSTK could convert seryl-tRNASec to O-phosphoseryl-tRNASec using ATP, GTP, CTP, and UTP as a phosphate donor in vitro. PSTK did not have intrinsic
54 ATPase activity. ATPase activity could only be observed in vitro when the enzyme was bound to either seryl-tRNASec or to uncharged SelC. Sequence analysis of PSTK indicated that the enzyme was a member of the P-loop superfamily. The N terminal domain had a number of motifs characteristic of this superfamily. This included a Walker A and ArgX3Arg motif for
2+ ATP binding and positioning of the -PO4 for hydrolysis, and a Walker B motif for Mg coordination. The C terminal domain shared sequence similarity with only one other P-loop superfamily member, the tRNA binding domain of the Kti12 protein which functions exclusively in eukaryotes to modify nucleotides in tRNA molecules (Sherrer, O’Donoghue and Söll 2008).
This suggested that the N terminal domain functions as a kinase domain, whilst the C terminal domain functions to recognize and bind tRNASec.
Concurrently, the specific portions of SelC required for PSTK recognition were elucidated (Sherrer, Ho and Söll 2008). Mutating specific base pairs in the Me. jannaschii SelC molecule demonstrated that the D and T arms made only minor contributions for the specificity of Me. jannaschii PSTK for SelC. PSTK did not seem to form specific interactions with the anticodon stem or variable arm of SelC. The acceptor arm, however, was vital for recognition by
PSTK. Two universally conserved base pairs in the acceptor arms of archaeal SelC, G2-C71 and
C3-G70, were essential for recognition. Curiously, a single universally conserved base pair in the
Ser acceptor arm of the Me. jannaschii tRNA UGA codon, A5-U68, was anti-determinant for recognition by PSTK. This suggests that PSTK forms base specific interactions with SelC that simultaneously render interactions with tRNASer molecules unfavorable, ensuring that only seryl- tRNASec is phosphorylated in vivo. Finally, it was demonstrated that PSTK does not recognize the Ser residue in binding to seryl-tRNASec. PSTK could still efficiently phosphorylate SelC that had been charged with a Thr residue in vitro to yield O-phosphothreonyl-tRNASec.
55 A crystal structure of the Me. jannaschii PSTK yielded additional insights into how the enzyme functions to efficiently phosphorylate seryl-tRNASec (Araiso et al. 2009). PSTK was shown to form a homodimer of two antiparallel subunits. In agreement with the sequence analysis above, PSTK forms two domains comprised of an N terminal domain, a flexible linker region, and a C terminal domain. The N terminal domain includes the characteristic P-loop structural motif and clearly functions as the kinase domain. The domain can be broken down structurally into a core region, a lid region, and an A76 binding region. The core region included the Walker A and ArgX3Arg motifs that are essential for ATP binding. A conserved Arg116 residue (numbering from the Me. jannaschii PSTK protein sequence) forms a single hydrogen bond with the adenine base of ATP. Other interactions at the ATP binding sites are with the -,
-, and -PO4 groups. The fact that most interactions with ATP are with the phosphate groups, rather than the adenine base, explains why PSTK is able to couple seryl-tRNASec phosphorylation to the hydrolysis of many purine and pyrimidine triphosphtases in vitro. The
2+ Mg is coordinated by a conserved Asp41 residue. The ATP binding site positions ATP so that
2+ the O atoms of the - and -PO4 groups interact with the Mg . A single H2O molecule is also
2+ coordinated with the Mg to facilitate hydrolysis of the -PO4.
The A76 binding region defines a portion of PSTK that putatively binds to the conserved
A76 nucleotide in SelC which is aminoacylated with Ser. Dimerization of the two subunits occurs at the A76 binding region and forms the A76 binding pocket. Dimerization is promoted by both electrostatic and hydrophobic interactions. The bottom of the A76 binding pocket is lined with the hydrophobic side chains of a number of amino acids, including conserved Met and Tyr residues. Models of seryl-tRNASec docking to this binding pocket indicate that the Ser residue is positioned by the Asp41 residue and a nearby Arg44 residue to accept the phosphate from the -
56 PO4 of ATP. The diameter of the binding pocket formed by the dimerization of the two subunits is sufficient to accommodate the acceptor arm of seryl-tRNASec.
Two crystal structures of the Me. jannaschii PSTK bound to SelC have been solved. One crystal structure features Me. jannaschii PSTK bound to Methanopyrus kandleri SelC (Chiba et al. 2010). The other crystal structure includes Me. jannaschii bound to a truncated Me. jannaschii SelC molecule lacking the anticodon stem and loop (Sherrer et al. 2011). Both structures provided models for how PSTK recognizes seryl-tRNASec that are in substantial agreement. Each subunit of the PSTK homodimer binds to a single seryl-tRNASec molecule, producing a 1:2 stoichiometry for PSTK and seryl-tRNASec in complex. The N terminal domain of each subunit binds the acceptor arm of SelC, whilst the C terminal domain recognizes the D arm, consistent with the division of labor indicated by sequence analysis of PSTK (Sherrer,
O’Donoghue and Söll 2008). Two -helices in the C terminal domain interact with the minor groove of the D arm. Asn, Arg, and Lys residues interact with the phosphate backbone and ribose groups. The only base specific interactions formed by the C terminal domain are from a conserved Arg199 residue (numbering adopted from (Sherrer et al. 2011)) that forms hydrogen bonds with the U59 base in the T loop and the C20 base in the D stem.
The N terminal domain forms base specific interactions with a number of nucleotide bases previously found to be essential for SelC recognition by PSTK (Sherrer, Ho and Söll
2008). Conserved Lys142 and Tyr143 residues in archaeal PSTK form hydrogen bonds with the
G2-C71 base pair in the SelC acceptor stem. Asn80, Ser81, and Asp133 residues (numbering adopted from (Chiba et al. 2010) form hydrogen bonds with the universally conserved G73 base in archaeal SelC. The multiple base specific contacts formed with the SelC acceptor arm by the
N terminal domain, compared to the non-specific interactions with the D arm by the C terminal
57 domain, agrees with previous work finding that the acceptor arm was the major determinant for
SelC recognition by PSTK (Sherrer, Ho and Söll 2008).
Shortly thereafter, an enzyme capable of converting O-phosphoseryl-tRNASec to selenocysteyl-tRNASec was discovered in Me. jannaschii and Methanococcus maripaludis (Yuan et al. 2006). This O-phosphoseryl-tRNASec selenocysteyl-tRNASec synthase (SepSecS) was not able to convert archaeal seryl-tRNASec to selenocysteyl-tRNASec, but could convert bacterial
SelC charged with O-phosphoserine to selenocysteyl-tRNASec. This indicated that SepSecS specifically recognizes the O-phosphoserine residue rather than the SelC molecule. Sequence analysis of SepSecS suggested that the enzyme contains a pyridoxal 5’-phosphate co-factor coordinated by a conserved Lys residue, similar to bacterial SelA. SepSecS was found only in the genomes of Sec utilizing archaea and eukaryotes, again underscoring that Sec synthesis mechanisms are shared between the two domains.
The solution of a crystal structure for SepSecS from Met. maripaludis emphasized that
SepSecS and SelA share structural homology and mediate similar biochemical reactions (Araiso et al. 2008), but nonetheless represent distinct mechanisms for Sec synthesis. The crystal structure confirmed that SepSecS indeed contains a pyridoxal 5’-phosphate co-factor coordinated by a conserved Lys278 residue (numbering from the Met. maripaludis SepSecS protein) as was observed in the SelA crystal structure. Unlike SelA, however, SepSecS forms a homotetramer, rather than a homodecamer. The four subunits of SepSecS have some structural homology with the subunits of SelA, as each subunit is L-shaped and composed of an extended N terminal domain, a core catalytic domain, and a C terminal domain. The extended N terminal domain is a feature shared by SepSecS and SelA (Araiso et al. 2008; Itoh et al. 2013), but is not present in other pyridoxal 5’-phosphate dependent enzymes (Araiso et al. 2008). Similar to bacterial SelA,
58 the functional unit of the SepSecS tetramer is a dimer. Tetramer assembly is promoted by hydrophobic interactions between the SepSecS subunits.
The active site of SepSecS is formed when two subunits dimerize. Dimerization of two
SepSecS subunits is promoted by hydrogen bond formation between Arg residues in one subunit and the pyridoxal 5’-phosphate of the other. Sulfate ions were used to mimic the phosphate group of O-phosphoserine. Sulfate was only coordinated by one subunit of each dimer and was
Sec bound by Arg94, Ser95, Gln105, and Arg307. Modelling of O-phosphoseryl-tRNA docking to the active site of SepSecS suggests that the mechanism of Sec synthesis is broadly analogous to that of SelA. Both feature replacement reactions when a Schiff base forms between the -amino group of the Ser or O-phosphoserine residue and the formyl group of the pyridoxal 5’-phosphate co-factor. The only difference is that the phosphate group of O-phosphoserine is the leaving group for the elimination step of the reaction in archaeal Sec synthesis, rather than the hydroxide ion of Ser in bacterial Sec synthesis. Three conserved Arg residues in the active site are essential for catalysis. The involvement of Arg residues in the elimination is shared with
SelA, but the Arg residues involved are not conserved between SelA and SepSecS (Itoh et al.
2013). It isn’t known why bacterial SelA forms a complex decameric structure, whilst SepSecS is able to efficiently synthesize Sec using a tetramer. However, the evolution of the bacterial decameric complex likely stemmed from the need to convert seryl-tRNASec to selenocysteyl- tRNASec, rather than O-phosphoseryl-tRNASec. Not only is the biochemical reaction involving a hydroxide leaving group less efficient, but the Ser residue in seryl-tRNASec necessitates that
SelA form specific interactions with bacterial SelC molecule in order to avoid the non-specific conversion of seryl-tRNASer to selenocysteyl-tRNASer (Itoh et al. 2014). SepSecS in archaea and eukaryotes, however, can use the biochemically superior phosphate as a leaving group, and can
59 recognize only the non-canonical amino acid O-phosphoserine to ensure no tRNASer molecules are mis-acylated with Sec.
Compared to Sec synthesis, the mechanisms of Sec incorporation in Bacteria and
Archaea are broadly conserved. A putative archaeal SelB was initially identified from the genome of Me. jannaschii based on substantial sequence homology with the three EF-Tu-like domains of the E. coli SelB translation factor (Rother et al. 2000). Encouragingly, this putative translation factor lacked homologous regions found on EF-Tu that allow for interaction EF-Ts.
Curiously, the C terminal region of the putative SelB shared no sequence homology with bacterial SelB. Purification of the translation factor confirmed that it was capable of binding
GTP and GDP. The KD of the protein for GTP was 0.1 M and the KD for GDP was 0.4 M.
This confirmed that the putative archaeal SelB displays similar affinity for GTP and GDP as the
E. coli SelB (Forchhammer, Leinfelder and Böck 1989; Thanbichler, Bock and Goody 2000).
The protein could also efficiently bind selenocysteyl-tRNASec, but not seryl-tRNASer. Gene knockouts of the putative SelB homolog of Met. maripaludis subsequently confirmed that this protein is required for Sec incorporation (Rother et al. 2003).
Curiously, the Me. jannaschii SelB could not bind to archaeal SECIS elements (Rother et al. 2000). The discovery of the location of the SECIS element on archaeal selenoprotein mRNA transcripts provide a mechanistic explanation for the inability of archaeal SelB to bind their
SECIS elements. A computational analysis of the selenoprotein mRNA transcripts of Me. jannaschii found evidence for conserved stem-loop structures within the transcripts that could serve a similar function as bacterial SECIS elements (Wilting et al. 1997). These stem-loop structures, however, were not located immediately downstream of the UGA codon. The elements were located in the 3’ untranslated region (UTR) of six of the seven selenoprotein
60 mRNA transcripts. The seventh transcript, from the gene encoding for the F420-dependent formate dehydrogenase involved in hydrogenotrophic methanogenesis, had its putative SECIS element located in the 5’ UTR. Analogous structures were also predicted to be transcribed into the 3’ UTR of the mRNA transcripts of two selenoproteins in Methanococcus voltae.
Shortly thereafter, it was confirmed that the SECIS element is transcribed into the 3’
UTR of the Me. jannaschii fruA selenoprotein mRNA (Rother et al. 2001). The fruA selenoprotein gene encodes for a subunit of the [NiFeSe] F420-reducing hydrogenase (Halboth and Klein 1992; Wilting et al. 1997). The solution structure of the SECIS element was consistent with the predicted structure, suggesting the SECIS element contained a base stem, two internal loops, and an apical loop (Rother et al. 2001). Deletions of the internal loops or the base stem abolished the expression of the FruA protein. Putative SECIS elements have been identified in the 3’ UTR region of selenoprotein genes of the methanogenic archaeon Met. maripaludis (Rother et al. 2001; Rother and Quitzke 2018).
SECIS elements are conserved in archaea, bacteria, and eukaryotes, but were previously thought to have no sequence or structural homology between the three domains (Krol 2002).
However, the discovery of putative selenoproteins and SECIS elements within Lokiarchaeota
(Mariotti et al. 2016) and Thorarchaeota (Liu et al. 2018) has revealed that sequence and structural motifs between members of Asgard archaea and eukaryotes have been conserved.
Analysis of Lokiarchaeota SECIS elements revealed a number of commonalities with eukaryotic
SECIS elements (Mariotti et al. 2016). All putative selenoprotein SECIS elements in
Lokiarchaeota are predicted to be part of the 3’UTR of the mRNA transcripts, and most are comprised of a 9 base pair long stem and an 11-nucleotide apical loop. A kink-turn core forms the base of the stem. An AUGA tetranucleotide motif typically precedes the base stem, and a
61 GA dinucleotide motif is found at the base of the apical loop. Apical loops are typically adenine- rich. SECIS elements for the putative Lokiarchaeota SelD forms a stem loop structure that is characteristic of Type II eukaryotic SECIS elements. Comparison of the only other known Sec utilizing archaea, methanogenic members of Methanococcales and Methanopyrales, revealed that the SECIS elements for putative vhuD transcripts share a number of structural and sequence motifs with Lokiarchaeota, including the tetranucleotide AUGA and dinucleotide GA motifs.
The vhuD gene encodes for a subunit of the [NiFeSe] F420 non-reducing hydrogenase (Halboth and Klein 1992; Wilting et al. 1997).
Locating the SECIS elements in the 3’ or 5’ UTR appears to confer an advantage to archaea and eukaryotes, as this allows these organisms to incorporate multiple Sec residues into a single selenoprotein. Within the archaea, this includes the VhuD protein, which can incorporate two Sec residues in Me. jannaschii (Wilting et al. 1997), Met. maripaludis (Rother et al. 2001) and Lokiarchaeota (Mariotti et al. 2016), a putative peroxiredoxin-like protein in
Lokiarchaeota (Mariotti et al. 2016) that incorporates two Sec residues, and a putative heterodisulfide reductase subunit A that can incorporate up to four Sec residues in Lokiarchaeota
(Mariotti et al. 2016). Examples of selenoproteins incorporating multiple Sec residues exist in eukaryotes as well (e.g., (Hill, Lloyd and Burk 1993; Shchedrina et al. 2007; Lee et al. 2011).
No bacterial selenoproteins have been described in the literature that are predicted to incorporate multiple Sec residues.
Support of the idea that a single SECIS element in the 3’ UTR can mediate the incorporation of multiple Sec residues into a selenopolypeptide comes from the metagenomic context of the vhuD and hdrA genes in Lokiarchaeota (Mariotti et al. 2016). The vhuD and hdrA genes were often located in tandem in Lokiarchaeota metagenomes. While both would contain
62 putative UGA codons, only the vhuD SECIS element would be present. This is the first evidence that a single SECIS element could potentially mediate the incorporation of up to six Sec residues in two different selenopolypeptides over a 2.0 kb span. In order to incorporate Sec residues into selenopolypeptides from the 3’ UTR, eukaryotes utilize a SECIS binding protein (SBP2) to recruit a distant SECIS element to the ribosome during Sec incorporation (Copeland et al. 2000).
Thus far, no SBP2 homolog has been identified in archaeal genomes (Mariotti et al. 2016;
Rother and Quitzke 2018). The need for a protein to bring a distant SECIS element in contact with the ribsome complex at the UGA codon explains why Me. jannaschii SelB is unable to bind to archaeal SECIS elements in vitro (Rother et al. 2000).
The lack of any known SECIS binding protein in archaea represents a significant gap in our knowledge of the physiology of Sec incorporation in this domain. It is unknown what regions of the SECIS element are recognized by a SECIS binding protein, nor is it known what the Sec incorporation machinery on the ribosome is comprised of. This makes it difficult to know what the function of the fourth domain of archaeal SelB may be during Sec incorporation.
What interactions may occur between the SECIS binding protein and SelB and the 80S ribosome is similarly unknown. Finally, it remains to be determined whether or not specific interactions between a putative SECIS binding protein, SelB, and the ribosome promote specific conformational changes to facilitate Sec incorporation as occurs between SelB, the SECIS element, and the 70S ribosome in Bacteria (Fischer et al. 2016).
Solution of the crystal structure of Met. maripaludis SelB confirmed that the protein consists of four domains (Leibundgut et al. 2005). The function of the first three EF-Tu-like domains and critical structural residues appear to be conserved between bacteria and archaea
(Leibundgut et al. 2005; Itoh, Sekine and Yokoyama 2015). The first domain binds to GTP and
63 harbors two GTPase switches. The second and third domains have an overall barrel conformation. The aminoacyl-binding pocket is situated at the interface of the first and second domains. As with bacterial SelB, conserved Phe51 and Arg247 residues (numbering taken from the Met. maripaludis SelB protein sequence) form part of the core of the aminoacyl-binding pocket. The Phe51 residue links the aminoacyl binding pocket to the first GTPase switch. The mechanism of Sec recognition is identical as well. In bacteria a number of basic Arg residues form electrostatic interactions with the negatively charged selenol group of Sec. In archaeal
SelB, this is mediated by two Arg residues (including Arg247) but additionally a His192 residue that is strongly conserved among archaeal and eukaryotic SelB proteins.
Modeling of the docking of selenocysteyl-tRNASec suggests that the third domain interacts with T arm of SelC, as was observed in the A. aeolicus crystal structure (Itoh, Sekine and Yokoyama 2015). The fourth domain comprises a single barrel structural motif and is linked the the three EF-Tu-like domains via a flexible linker region. The authors note that the fourth domain appears structurally similar to known RNA binding proteins, suggesting that interactions between archaeal SelB and the mRNA transcript do occur. This could function to stabilize the transcript or ensure proper orientation of the transcript for entry into the ribosome when the unknown SECIS binding protein binds to the element and bends the transcript to bring the element into spatial proximity of the translation complex.
The evolution of selenocysteine utilization
Phylogenetic analyses of SelD provide substantial evidence that the ability to synthesize selenophosphate from Se(-II) was inherited from LUCA (Mariotti et al. 2015; Weiss et al. 2016).
Moreover, the use of Sec in lieu of Cys was likely the ancestral state for SelD (Mariotti et al.
2015). This phylogenetic evidence is in agreement with the observed distribution of SelD in the
64 genomes of organisms from all three domains of life (Zhang and Gladyshev 2010; Mariotti et al.
2015). This constitutes strong evidence that Sec utilization was a key feature of the metabolism of LUCA. The structural and functional similarities of the N terminal domains of SelB in
Bacteria and Archaea, particularly the conservation of a number of amino acids critical for tRNASec binding and GTPase activity (Leibundgut et al. 2005; Itoh, Sekine and Yokoyama
2015), argue persuasively that SelB homologs between the two domains also share a common ancestry. However, similar topologies and selenol binding residues could reflect convergent evolution, rather than vertical descent. The only information available in the literature currently is that bacterial SelB is closely related to the eukaryotic and archaeal specific translation initiation factor eIF-2 within the GTPase superfamily of translation initiation and elongation factors (Keeling, Fast and McFadden 1998). The phylogenetic relationship between bacterial, archaeal, and eukaryotic SelB proteins remains to be clarified.
Mechanisms of Sec synthesis clearly differ between Bacteria and the Archaea and
Eukarya domains. SelA and SepSecS are both members of the fold type I pyridoxal 5’- phosphate-dependent (PLP-dependent) enzyme superfamily (Araiso et al. 2008; Itoh et al. 2014).
However, SelA and SepSecS do not form a coherent monophyletic clade within the superfamily.
SelA clusters with the cystathionine -synthase family and SepSecS clusters with the distantly related sugar aminotransferase family (Itoh et al. 2014). Curiously, selenocysteine -lyase is also a member of the fold type I PLP-dependent superfamily, though it is only distantly related to either SelA or SepSecS (Araiso et al. 2008; Itoh et al. 2014). PSTK is a member of the P-loop kinase superfamily (Yuan et al. 2006; Sherrer, O’Donoghue and Söll 2008). Both PSTK and
SepSecS co-occur in archaeal and eukaryotic genomes, suggesting that these proteins have experienced a significant degree of coevolution (Yuan et al. 2006). All of these phylogenetic
65 analyses, however, offer substantial support that SelA and SepSecS/PSTK have largely been inherited via vertical descent within their respective domains.
While Sec utilization may be a primordial metabolic adaptation, it is hardly universal.
Estimates of the prevalence of Sec utilization traits in bacterial genomes have consistently ranged between ~20%-26% of all available genomes for more than a decade (Zhang et al. 2006;
Zhang and Gladyshev 2010; Lin et al. 2015; Mariotti et al. 2015; Peng et al. 2016). Moreover, the number of selenoproteins expressed by Sec utilizing bacteria also varies widely among different phyla (Zhang et al. 2006; Zhang and Gladyshev 2010; Lin et al. 2015; Peng et al.
2016). In most phyla, the genomes of Sec utilizing bacteria include only one or two selenoproteins (typically SelD and/or the FdhN catalytic subunit). A number of deeply branched phyla, such as the Firmicutes and Deltaproteobacteria, feature organisms whose genomes encodes more than a dozen selenoproteins. Previous surveys of archaeal genomes for Sec utilization have only identified members of the Methanococcales and Methanopyrales orders as being capable of synthesizing and incorporating Sec (Zhang and Gladyshev 2010; Mariotti et al.
2015). In a recent comprehensive survey of archaeal genomes, only 21 out of nearly 500 genomes surveyed contained the selC gene (Santesmasses, Mariotti and Guigó 2017). The recent discovery of Sec utilization traits in the reconstructed genomes of members of the Asgard archaea (Mariotti et al. 2016; Liu et al. 2018), however, indicates that the number of archaea known to exploit Sec will likely increase in the near future.
Selenoproteins in bacteria and archaea are frequently involved in energy conservation
(Stock and Rother 2009; Mariotti et al. 2015; Rother and Quitzke 2018). Consistent with the deep antiquity of Sec utilization, a number of selenoproteins have been experimentally verified to catalyze crucial reactions in acetogenesis and/or hydrogenotrophic methanogenesis. A great
66 deal of evidence exists indicating that acetogenesis and hydrogenotrophic methanogenesis may represent the oldest forms of energy conservation (Martin W et al. 2008; Lane, Allen and Martin
2010; Sousa et al. 2013; Weiss et al. 2016). This implicates selenoproteins, then, as not just crucial components of the global carbon biogeochemical cycle but suggest an interrelationship between the selenium and carbon biogeochemical cycles that extends through deep time.
Sec-containing formate dehydrogenases are common to both biochemical pathways.
These include the NAD+-dependent formate dehydrogenases of Mo. thermoacetica (Andreesen and Ljungdahl 1973, 1974) and Eu. barkeri (Graentzdoerffer et al. 2003) among acetogens. Sec- containing F420-dependent formate dehydrogenases have been characterized in the hydrogenotrophic methanogens Met. vannielii (Jones, Dilworth and Stadtman 1979a; Jones and
Stadtman 1981) and Met. maripaludis (Wood, Haydock and Leigh 2003). These formate dehydrogenases function in acetogenesis to reduce CO2 to formate and to oxidize formate to CO2 during hydrogenotrophic methanogenesis (Ferry 1990). Recent work in our group has demonstrated that these formate dehydrogenases are members of a single evolutionary lineage that also includes the Sec-containing FdhH protein from E. coli that functions to decompose excess formate produced during fermentation to H2 and CO2 (Sawers 1994) and the Cys- containing FdrA protein from Cupriavidus necator that functions to oxidize formate to CO2 under aerobic conditions (Niks et al. 2016; Yu et al. 2017) (Wells et al. in preparation).
Crucially, our phylogeny provided robust support that Sec-use in these formate dehydrogenases represents the ancestral state.
A number of other selenoproteins are known to participate in hydrogenotrophic methanogenesis. This includes the molybdopterin-harboring subunit of the formyl-methanofuran dehydrogenase of Methanopyrus kandleri (Vorholt, Vaupel and Thauer 1997). Formyl-
67 methanofuran dehydrogenase catalyzes the first step in the reductive transformation of CO2 to methane (Börner, Karrasch and Thauer 1991). Subunits of two hydrogenases are also known to be selenoproteins. This includes the large subunit (FruA) of the F420-reducing hydrogenase
(Yamazaki 1982; Sorgenfrei et al. 1997). The F420-reducing hydrogenase couples the oxidation of H2 to the reduction of F420, the soluble electron carrier utilized during hydrogenotrophic methanogenesis (Stock and Rother 2009). Additionally, two small subunits of the F420-non- reducing hydrogenase are frequently selenoproteins. This includes the VhuU subunit (Halboth and Klein 1992; Sorgenfrei et al. 1993) and the VhuD subunit (Wilting et al. 1997; Rother et al.
2001), which frequently incorporates two Sec residues in Sec utilizing methanogens. The F420- non-reducing hydrogenase functions to transfer electrons from H2 to the heterodisulfide reductase complex (Costa et al. 2013).
The heterodisulfide reductase complex uses electron bifurcation to reduce ferredoxin and the heterodisulfide that forms between Coenzymes M and B during the reduction of methyl-S-
CoM to methane (Kaster et al. 2011; Wagner et al. 2017). The reduced ferredoxin can then serve as a soluble electron carrier during the fixation of CO2 and the recycled Coenzymes M and
B are available to catalyze the final step in methanogenesis. The electron bifurcating HdrA subunit of the heterodisulfide reductase has been shown to be a selenoprotein in Me. jannaschii
(Wilting et al. 1997). Curiously, the HdrA subunit and the F420-reducing hydrogenase appear to have been acquired by some members of the Deltaproteobacteria phylum from Sec-utilizing methanogens via horizontal gene transfer (Zhang et al. 2006). These proteins appear to have largely retained the Sec residues in these bacteria. Putative Sec-containing homologs of the
VhuU and VhuD subunits of the F420-non-reducing hydrogenase and the HdrA subunit of the
68 heterodisulfide reductase complex have been found in Lokiarchaeota (Mariotti et al. 2016) as well. This suggests that these selenoproteins do not exclusively function in methanogenesis.
It is unknown whether the Sec residue in these selenoproteins represents the ancestral state for these enzyme and hydrogenase subunits. Consistent with the hypothesis that hydrogenotrophic methanogenesis represents a basal metabolism for energy conservation is the wide phylogenetic distribution of some of the components mediating the fixation of CO2 to methane. Heterodisulfide reductases, for example, are widespread among various non- methanogenic members of both Bacteria and Archaea (Yan, Wang and Ferry 2017). They appear to mediate a staggering diverse array of metabolic reactions in vivo (e.g., ((Mander et al.
2004; Ramos et al. 2015; Koch and Dahl 2018)). And Sec-containing HdrA subunits are by no means confined to the Deltaproteobacteria in the bacterial domain. The CHY_0930 locus of the
Carboxydothermus hydrogenoformans Z-2901 genome putatively encodes a Sec-containing
HdrA selenoprotein (Wu et al. 2005). The phylogenetic relationship of these various selenoproteins to homologs in non-methanogenic organisms needs to be clarified to better understand whether Sec use represents the ancestral state and whether patterns of Sec loss in non-methanogenic lineages are correlated with the acquisition of novel physiological functions.
A number of other selenoproteins in Bacteria are known to catalyze central reactions in fermentative growth on amino acids. The GrdA (Cone et al. 1976; Sliwkowski and Stadtman
1988) and GrdB (Wagner et al. 1999) subunits of the glycine reductase complex are frequently selenoproteins. The glycine reductase complex allows peptidolytic fermenting bacteria to reduce glycine to ammonia and acetyl-phosphate via substrate-level phosphorylation (Sliwkowski and
Stadtman 1988). The methylated glycine derivatives betaine and sarcosine can also be fermented, and the PB subunits of the betaine and sarcosine reductase complexes have been
69 confirmed to be selenoproteins in Eu. acidaminophilum (Hormann and Andreesen 1989; Meyer,
Granderath and Andreesen 1995). In addition to fermentation, glycine reductase, betaine reductase, and sarcosine reductase can also participate in a unique acetogenesis pathway that utilizes glycine to produce acetate (Dürre and Andreesen 1982; Andreesen et al. 1999b; Stock and Rother 2009). Finally, the amino acid proline can undergo reductive cleavage to yield 5- aminovalerate during fermentative growth, catalyzed by the D-proline reductase complex
(Kabisch et al. 1999). The PrdB subunit of the complex can also be a selenoprotein. The use of
Sec in the fermentation of amino acids appears to be confined to bacteria. It unknown whether
Sec use represents a basal adaptation in the evolution of amino acid fermentation or a more recent innovation.
One of the central challenges in understanding the evolution of Sec utilization is that the adaptive advantage it confers on organisms is not yet known. Frequent and independent loss of
Sec utilization traits is common across lineages in all three domains of life (Zhang et al. 2006;
Lobanov et al. 2007; Rother and Krzycki 2010; Zhang and Gladyshev 2010; Mariotti et al. 2015,
2019). Known selenoproteins additionally all have Cys-containing homologs that are more widely distributed across genomes than the selenoproteins themselves. Sec incorporation is also highly inefficient compared to the incorporation of the 20 canonical amino acids (Suppmann,
Persson and Böck 1999). This begs the question of why Sec utilization evolved and has persisted for nearly 3.7 billion years. There is considerable evidence in mammalian selenoproteins that Sec residues have experienced substantial purifying selection (Castellano
2009; Castellano et al. 2009), underscoring that some selective advantage for Sec exists. A common hypothesis is that Sec residues confer on oxidoreductases superior catalytic activity
70 compared to Cys-containing homologs (e.g., ((Rocher, Lalanne and Chaudière 1992; Gromer et al. 2003; Kim et al. 2006, 2015)).
This hypothesis is unsatisfactory in that it is difficult to conceive of an objective way to demonstrate that a Sec-containing homolog provides a catalytic advantage to one organism sufficient to offset the costs associated with Sec synthesis and incorporation, while it failed to provide that same fitness advantage to a substantial number of other organisms that use a Cys- containing homolog. A far more compelling hypothesis is provided by Reich and Hondel
(2016), who note that the principle benefit of Sec appears to be protection from irreversible oxidation. The authors provide two pertinent observations. The first is that substituting Sec for
Cys in recombinant proteins or synthetic peptides, confers on these enzymes and peptides the ability undergo repeated cycling between oxidized and reduced states without irreversible oxidation. (Boschi-Muller et al. 1998; Sun et al. 2004; Casi, Roelfes and Hilvert 2008; Snider et al. 2013). Furthermore, Sec-containing NiFeSe hydrogenases from strictly anaerobic
Deltaproteobacteria can be purified aerobically without being irreversibly oxidized by O2, in contrast to their Cys-containing homologs (Teixeira et al. 1987; Parkin et al. 2008).
Testing this hypothesis would be a productive avenue of research, given that redox homeostasis and antioxidant defense are predicted functions for many selenoproteins in all three domains of life (Zhang et al. 2006; Lobanov, Hatfield and Gladyshev 2009; Zhang and
Gladyshev 2010; Labunskyy, Hatfield and Gladyshev 2014; Mariotti et al. 2016; Peng et al.
2016). There has been growing interest in the evolution of thiol-based antioxidant defense mechanisms as a response to the growing oxidation of the Earth‘s surface over the course of the
GOE (Fahey 2013; Fischer, Hemp and Valentine 2016). It is interesting to note, then, that putative peroxiredoxin-like selenoproteins have been identified in bacteria and archaea (Zhang et
71 al. 2006; Zhang and Gladyshev 2008; Mariotti et al. 2016; Peng et al. 2016). Putative thioredoxin-like, glutaredoxin-like, glutathione peroxidase-like, and alkyl hydroperoxide-like selenoproteins have also been observed in bacterial genomes (Zhang et al. 2006; Zhang and
Gladyshev 2008; Peng et al. 2016). Moreover, the use of Sec in these oxidoreductases is confined to selenoprotein-rich bacteria from deeply branching phyla such as the Firmicutes,
Synergistetes, and Deltaproteobacteria. Only two selenoproteins involved in redox homeostasis have been characterized to date. A peroxiredoxin-like selenoprotein was characterized in Eu. acidaminophilium, but the protein was expressed in E. coli as a recombinant Cys-containing homolog and did not demonstrate the predicted thiol-dependent peroxidase activity (Söhling et al. 2001). A Sec-containing methionine sulfoxide reductase A protein was also characterized from Alkaliphilus oremlandii OhILAs (Kim et al. 2009). Methionine sulfoxide reductases function to prevent methionine residues in proteins from becoming oxidized, which can negatively affect both the structure and function of a protein (Kim 2013).
Physiological and evolutionary study of these putative redox-active selenoproteins is essential to determine whether Sec use in these oxidoreductases constitutes an adaptation by organisms to the redox challenges posed by growing concentrations of O2 during the GOE and
Proterozoic Eon. Alternatively, these Sec-containing enzymes could represent components of primordial antioxidant defense mechanisms that would have protected ancient life from mild oxidants. Current estimates indicate that the Archean Earth would have been bombarded with several orders of magnitude more UV radiation than the modern Earth (Cnossen et al. 2007).
This UV radiation would have produced oxidants that could have been an important stimulus for the evolution of antioxidant defense mechanisms. This would provide a simple explanation for the patterns of Sec loss seen in prokaryotes. Organisms that have developed antioxidant
72 mechanisms robust enough to allow for aerobic respiration would presumably have less need to maintain a suite of energetically expensive selenoproteins.
5-methylaminomethyl-2-selenouridine synthesis
Selenium-containing nucleosides were first observed in tRNA molecules from E. coli
(Hoffman and McConnell 1974; Prasada Rao and Cherayil 1974), C. sticklandii (Chen and
Stadtman 1980), and Met. vannielii (Ching et al. 1984). It was initially assumed that the incorporation of selenium into these nucleosides represented the non-specific substitution of selenium for sulfur by enzymes synthesizing sulfur modified bases (Hoffman and McConnell
1974; Prasada Rao and Cherayil 1974). It was later demonstrated that the incorporation of selenium into tRNA nucleosides is mediated by selenium-specific molecular pathways (Chen and Stadtman 1980; Wittwer 1983). Moreover, these selenium-containing bases can represent a substantial fraction of the cellular pool of tRNA molecules. For example, selenium-containing tRNAs have been estimated to comprise 5-8% of bulk tRNA molecules in C. sticklandii (Chen and Stadtman 1980) and 13-20% of bulk tRNA molecules in Met. vannielii (Ching et al. 1984).
These selenium modified bases were originally identified as 4-selenouridine (Hoffman and McConnell 1974; Prasada Rao and Cherayil 1974). Subsequently, 70-90% of selenium modified bases in E. coli, 40-60% in C. sticklandii (Wittwer et al. 1984) and 60% of the selenium-containing tRNA molecules in Met. vannielii (Ching et al. 1984; Wittwer et al. 1984) were definitively identified as mnm5Se2U. A significant proportion of the remaining selenonucleosides in C. sticklandii and Met. vannielii were thought to comprise either an intermediate in the synthesis of mnm5Se2U or a second, currently unknown, selenium modified base. mnm5Se2U has been shown to be a constituent of tRNAGlu molecules in C. sticklandii
(Ching and Stadtman 1982), tRNAGlu and tRNALys molecules in E. coli (Wittwer 1983) and
73 tRNAGlu, tRNAGln, and tRNALys molecules in Met. vannielii (Ching et al. 1984; Politino et al.
1990) and Salmonella typhimurium (Veres et al. 1990). These selenium-containing tRNA molecules have a single mnm5Se2U base (Ching and Stadtman 1982; Wittwer et al. 1984) located specifically at the first anticodon position (Ching, Alzner-DeWeerd and Stadtman 1985).
Curiously, a sulfur-containing nucleoside analog, 5-methylaminomethyl-2-thiouridine
(mnm5S2U) is also present in tRNAGlu and tRNALys populations and occupies the same first anticodon position as mnm5Se2U in E. coli (Wittwer 1983; Wittwer and Stadtman 1986). Thus, bacteria and archaea capable of synthesizing mnm5Se2U appear to maintain two distinct populations of tRNAGlu, tRNAGln, and tRNALys molecules. One population contains mnm5Se2U at the first anticodon position, and the other mnm5S2U at the first anticodon position.
Early attempts to elucidate the pathways involved in mnm5Se2U synthesis and incorporation revealed a physiological link with Sec metabolic pathways. Gene knockout mutations in the selD gene of S. typhimurium and E. coli produced mutant strains that were incapable of synthesizing mnm5Se2U (Kramer and Ames 1988; Stadtman et al. 1989).
Furthermore, supplementation of Sec pathway products and intermediates (e.g., L-Sec and L-
Ser) could stimulate mnm5Se2U synthesis in in vitro assays containing bulk tRNA and cell lysates of S. typhimurium (Veres et al. 1990) and Met. vannielii (Politino et al. 1990). Of course, the linkage between Sec and mnm5Se2U synthesis and incorporation was shown to be the common selenium donor, selenophosphate (Veres et al. 1992a; Glass et al. 1993). Subsequently, an enzyme was purified from S. typhimurium that could catalyze the ATP-independent conversion of mnm5S2U to mnm5Se2U in bulk tRNA (Veres and Stadtman 1994), likely via the replacement of the sulfur atom for a selenium atom in the modified base. This putative tRNA 2- selenouridine synthase (SelU) could catalyze the reaction using either E. coli or S. typhimurium
74 bulk tRNA as substrate. The Km of SelU for selenophosphate was found to be 17.1 M. The purified enzyme was not sequenced, and the gene locus coding for SelU was not determined.
The gene coding for SelU was later established in E. coli (Wolfe et al. 2004b).
Comparative genomics had found a gene that frequently co-occurred with selD in bacterial genomes that putatively encoded for a protein consisting of an N terminal rhodanese domain with a single active site Cys residue, and a C terminal domain consisting of a P loop-like Walker
A motif and an anticodon loop binding site. These genes are homologs of the E. coli ybbB gene
(which does not form an operon with the E. coli selD gene). Gene knockouts of the E. coli ybbB gene confirmed that deletion of the gene abolishes the ability of cells to synthesize mnm5Se2U.
Purification of the ybbB gene product confirmed that the protein encoded at this locus is indeed
SelU. The estimated molecular weight of SelU is 43 kDa. Analysis of the selenium replacement reaction indicated that SelU binds mnm5S2U-containing tRNA molecules with a 1:2 stoichiometry. Site-directed mutagenesis of the single active site Cys residue, Cys97 in the E. coli SelU protein sequence, confirmed that this residue is indispensable for the replacement reaction. Sequence analysis of putative SelU homologs suggested that SelU is defined by a common CysXArgGlyGlyXArgSer motif, with the Cys residue corresponding to the Cys97 residue in E. coli. SelU homologs were identified in representatives from the Proteobacteria,
Firmicutes, and Cyanobacteria. A putative SelU homolog was identified in the Me. jannaschii genome, but the respective N terminal and C terminal domains were encoded by separate genes.
The identification of a putative SelU homolog in an archaeal genome suggested that mnm5Se2U synthesis and incorporation mechanisms could be conserved between Bacteria and
Archaea. A subsequent comparative genomics analysis specifically in archaeal genomes found
13 Methanococcales genomes that had two gene homologs for the N terminal and C terminal
75 domains of bacterial SelU (Su et al. 2012). The gene encoding for the putative rhodanese domain-containing protein shared the same CysXArgGlyGlyXArgSer motif with bacterial SelU and the gene encoding for the C terminal-like protein was predicted to harbor a P loop-like
Walker A motif. Deletion of both these genes in Met. maripaludis abolished the ability of this archaeon to synthesize mnm5Se2U. A neighbor joining phylogeny of bacterial SelU and the concatenated archaeal SelU-like proteins confirmed that mnm5Se2U synthesis mechanisms are conserved in prokaryotes.
Curiously, bacterial SelU was separately shown to be capable of catalyzing the geranylation of sulfur in mnm5S2U nucleosides (Chen et al. 2005; Dumelin et al. 2012). The ability of SelU to geranylate tRNA molecules was initially thought to implicate SelU in multiple physiological roles in bacteria. However, proposals followed shortly thereafter that the geranylation of sulfur in mnm5S2U actually represented an intermediate step in the selenium replacement reaction , with the S-geranyl moiety functioning as a leaving group (Bartos et al.
2014; Jäger, Chen and Björk 2016). A recent report has provided robust support for this hypothesis (Sierant et al. 2018). In vitro assays using E. coli SelU and synthetic 17 nucleotide oligomers representing the anticodon stem of E. coli tRNALys molecules established that SelU was unable to convert mnm5S2U to mnm5Se2U without the addition of geranyl phosphate. The
-3 min Lys sulfur geranylation reaction was found to have a kcat of 0.14 x 10 and a Km for the tRNA -
min like oligomer of 1.23 M. The selenium replacement reaction had a kcat of 0.53 and a Km for the oligomer of 2.49 M. Previous biochemical work had not observed the requirement for geranylated intermediates for the selenation reaction to proceed likely because the bulk pool of tRNA molecules used as substrate in these assays contained a supply of geranylated mnm5S2U- containing tRNA.
76 Biochemical characterization of SelU has so far established that the reaction sequence likely consists of the geranylation of the 2-thiouridine moiety of mnm5S2U, followed by the replacement of selenium for sulfur using S-geranyl as a leaving group (Sierant et al. 2018).
Furthermore, the successful use of synthetic oligomers representing only the anticodon stem indicates that this stem alone is sufficient for SelU recognition of mnm5S2U -containing tRNA molecules. Previous biochemical characterization of SelU has demonstrated that these enzymes can convert mnm5S2U to mnm5Se2U in tRNA molecules from diverse groups of organisms. The
S. typhimurium SelU could catalyze the selenium replacement reaction in vitro when provided with bulk E. coli or S. typhimurium tRNA (Veres and Stadtman 1994). Surprisingly, E. coli
SelU was even capable of catalyzing selenium replacement when provided with bulk Met. maripaludis tRNA (Su et al. 2012). All of this data indicates that SelU likely forms few, if any, base specific reactions with tRNAGlu, tRNAGln, and tRNALys molecules, and the 2-thiouridine moiety alone may be sufficient for recognition. Similarly, it is unknown whether both domains are involved in the geranylation and selenium replacement reactions, or if these two reactions are mediated by separate domains. The selenophosphate binding site is also unknown, as is whether
SelU interacts with SelD to form a transient complex as has been observed between SelD, SelA, and SelC in E. coli (Silva et al. 2015). It is also not clear whether bacteria and archaea specifically maintain separate populations of mnm5Se2U- and mnm5S2U-containing tRNA molecules by having SelU specifically recognize only a subset of these tRNAs for mnm5Se2U conversion. Alternatively, the inability of SelU to convert all tRNAGlu, tRNAGln, and tRNALys molecules to selenonucleosides could represent thermodynamic or kinetic constraints on the geranylation and/or replacement reactions of SelU that leave a substantial cellular repertoire of mnm5S2U-containing tRNA available in vivo.
77 The only phylogenetic analysis of bacterial SelU and the archaeal concatenated N terminal- and C terminal-like protein homologs indicated that the ability to synthesize mnm5Se2U was inherited from LUCA (Su et al. 2012). Consistent with this, comparative genomics has identified SelU homologs in representatives from most bacterial phyla and estimates of the prevalence of SelU in bacterial genomes are very similar to estimates for the prevalence of Sec utilization (Romero et al. 2005; Lin et al. 2015; Peng et al. 2016). Currently, the ability to synthesize mnm5Se2U appears confined to methanogenic archaea (Su et al. 2012;
Lin et al. 2015). Sec utilization and mnm5Se2U synthesis traits appear strongly linked in both domains of life (Lin et al. 2015; Peng et al. 2016). Nevertheless, there are bacterial genomes whose sole predicted selenium assimilation trait is mnm5Se2U synthesis (e.g., selenium utilizing representatives from the Cyanobacteria phylum). Thus, it appears that incorporating mnm5Se2U in tRNA molecules provides a substantial selective advantage to some organisms, even when the ability to utilize Sec has been lost.
Given that the modified mnm5Se2U base is found exclusively at the first wobble position in the anticodon loop, the most intuitive benefit to incorporating mnm5Se2U into selenonucleosides would be in facilitating codon recognition. Multiple studies of mnm5Se2U- and mnm5S2U-containing tRNAGlu and tRNALys molecules have found substantial differences in codon recognition by these selenium and sulfur modified bases (Ching, Alzner-DeWeerd and
Stadtman 1985; Ching 1986; Wittwer and Ching 1989). Glu is coded for by the GAA and GAG codons and Lys is coded for by the AAA and AAG codons. The recognition of the GAA and
AAA codons by mnm5S2U-containing tRNAGlu and tRNALys molecules was far more efficient than for the GAG and AAG codons. mnm5Se2U-containing tRNAGlu and tRNALys molecules, however, showed no bias towards codons with a third position adenine or guanosine. While the
78 efficiency of codon recognition by tRNAGln molecules has yet to be studied, it should be noted that Gln is coded for by CAA and CAG, and thus presumably mnm5Se2U-containing tRNAGln would similarly provide better recognition of CAG compared to mnm5S2U-containing tRNAGln.
A second selective advantage for mnm5Se2U-containing tRNA molecules over mnm5S2U- containing homologs has recently been proposed. The 2-thiouridine moiety in mnm5S2U- containing tRNA has been shown to become desulfurated in response to oxidative stress. This results in either a reduced ability to efficiently recognize appropriate codons, or a loss of function altogether, in sulfur-containing tRNA molecules (Sochacka et al. 2011, 2013). A synthetic analog of mnm5Se2U was recently shown to be resistant to irreversible oxidation and deselenization in response to treatment with oxidizing agents (Payne et al. 2017). This remains to be tested in vivo, but if these observations are confirmed by additional experiments this would provide further support for the possible coevolution of selenium utilization and antioxidant defense systems.
Selenium as an inorganic cofactor in molybdenum hydroxylases
It was initially shown that selenium supplementation is essential for fermentative growth on nicotinic acid (nicotinate) (Imhoff and Andreesen 1979) and various purines (Wagner and
Andreesen 1979; Dürre and Andreesen 1983) in a number of purinolytic clostridia. The nature of this selenium requirement became clear when enzymes involved in these pathways were purified and biochemically characterized. The nicotinic acid hydroxylase of Eu. barkeri
(Dilworth 1982; Gladyshev, Khangulov and Stadtman 1996), the xanthine dehydrogenases of
Clostridium acidi-urici (Wagner, Cammack and Andreesen 1984), Eu. barkeri (Schräder,
Rienhöfer and Andreesen 1999), and Clostridium purinolyticum (Self and Stadtman 2000), and the purine hydroxylase of Clostridium purinolyticum (Self and Stadtman 2000) have been shown
79 to contain a molybdopterin cytosine dinucleotide cofactor, an FAD cofactor, two [2Fe-2S] clusters, and a labile selenium atom. The lability of the selenium atom differentiates it from Sec as it is non-covalently bound to these selenoenzymes as an inorganic cofactor. Both Clostridium acid-urici and Clostridium purinolyticum have been recently reclassified as Gottschalka acidurici (Poehlein et al. 2017) and Gottschalkia purinolyticum (Poehlein et al. 2015).
Based on the subunit and cofactor composition, these various selenoenzymes all appear to be members of a collection of enzymes present in all three domains of life that are referred to alternatively as the xanthine oxidase family (Hille, Hall and Basu 2014), after the well- characterized mammalian bovine xanthine oxidase, or the molybdenum hydroxylase family
(Dobbek 2011). These selenoenzymes all function to hydroxylate heterocyclic rings. Nicotinic acid hydroxylase hydroxylates the 6-position of nicotinic acid to yield 6-hydroxynicotinic acid using NADP+ as an electron acceptor (Gladyshev, Khangulov and Stadtman 1994). Xanthine dehydrogenase hydroxylates the 8-position of xanthine to yield uric acid (Self and Stadtman
2000). Purine hydroxylase likely functions in vivo to hydroxylate the 2-position of hypoxanthine to yield xanthine using NADP+ as an electron acceptor (Self, Wolfe and Stadtman 2003).
The molybdopterin cytosine dinucleotide cofactor of other molybdenum hydroxylases has been viewed as a model for the nature of the Se moiety in these selenoenzymes. In most molybdenum hydroxylase family enzymes, the Mo atom is coordinated by two enedithiolate sulfur ligands and a functional hydroxyl ligand on the same geometric plane, an apical oxo ligand, and a terminal sulfido ligand (Hille, Hall and Basu 2014). This terminal sulfido ligand is readily replaced by an oxo ligand when enzymes are exposed to cyanide, yielding an inactivated enzyme. Similarly, all of these selenoenzymes are inactivated by treatment with cyanide, resulting in the loss of the labile selenium co-factor (Dilworth 1982; Gladyshev, Khangulov and
80 Stadtman 1996; Schräder, Rienhöfer and Andreesen 1999; Self and Stadtman 2000; Self, Wolfe and Stadtman 2003). This suggested that the Se moiety functions as a selenido ligand in the first coordination sphere of the Mo atom in an analogous manner to the terminal sulfido ligand. For the nicotinic acid hydroxylase of Eu. barkeri this was definitively shown to be the case, as EPR spectroscopy confirmed that the Se atom directly interacts with the Mo atom, but not the polypeptide chain (Gladyshev, Khangulov and Stadtman 1994, 1996). Additionally, a crystal structure of the selenoenzyme revealed the Mo atom was coordinated by the expected selenido ligand (Wagener et al. 2009).
Curiously, the selenium atom of the purine hydroxylase of G. purinolyticum likely occupies a different position in the selenoenzyme than in the Eu. barkeri nicotinic acid hydroxylase. EPR spectroscopy of the cofactors of purine hydroxylase confirmed that the Se atom is likely part of the molybdopterin cytosine dinucleotide cofactor, rather than the FAD or
Fe-S clusters (Self, Wolfe and Stadtman 2003). However, the EPR spectra did not find any evidence that the Mo atom directly interacts with the Se atom. This led the authors to propose that the Se atom was likely present in purine hydroxylase as an S-selanylcysteine residue that interacts with the terminal sulfido ligand, as had been observed in a crystal structure of the molybdenum hydroxylase family member CO dehydrogengase in Oligotropha carboxidovorans
(Dobbek et al. 1999). However, it was later discovered that the increased electron density around the terminal sulfido ligand moiety was not due to an S-selanylcysteine residue, but rather an unusual CuS terminal ligand (Dobbek et al. 2002). The model for Se coordination based on the older crystal structure is therefore likely incorrect. However, no further data on where the Se atom is positioned in purine hydroxylase is available.
81 Biochemical characterization of enzymes containing a labile inorganic selenium cofactor suggests that the use of selenium in molybdenum hydroxylase family members is confined to purinolytic clostridia. The presence of so-called “orphan” selD genes in the genomes of organisms lacking genes for either Sec or mnm5Se2U synthesis, however, suggested that SelD could provide selenophosphate as a selenium donor for Se cofactor insertion in molybdenum hydroxylases (Haft and Self 2008; Zhang et al. 2008). A gene coding for a hypothetical protein
(later named yqeB from an E. coli homolog (Lin et al. 2015)) was found to frequently co-occur with selD and molybdenum hydroxylases and molybdenum hydroxylase accessory proteins in organisms with “orphan” selD genes. A gene coding for a second hypothetical protein, later named yqeC from an E. coli homolog (Lin et al. 2015), was frequently found to co-occur with selD and yqeB in the genomes of organisms with “orphan” selD genes (Zhang et al. 2008).
Curiously, only the presence of selD, yqeB, and yqeC together in a genome appears to be a reliable predictor for the selenium cofactor trait. Organisms with either yqeB or yqeC homologs frequently lack any selD homolog, suggesting that neither functions exclusively in selenium metabolism (Zhang et al. 2008; Peng et al. 2016).
If these three genes together are indeed reliable indicators of the selenium cofactor trait, then this trait doesn’t appear to be common. Only about 6% of bacterial genomes, out of nearly
5200 sampled, harbored these three genes together (Peng et al. 2016). The trait seems most common in members of the clostridial and Enterobacteriales lineages (Lin et al. 2015). The use of selenium as an inorganic cofactor is also the only predicted selenium metabolism trait in halophilic archaea (Zhang et al. 2008). Gauging the reliability of these computational predictions is difficult, given that functions have not yet been assigned to the YqeB and YqeC proteins. Moreover, the primary sequence of these proteins does not suggest the presence of
82 conserved and well-characterized functional domains that could yield insights as to their potential role in selenium cofactor synthesis or incorporation. It should be noted that if the position of the selenium atom within the molybdopterin cytosine dinucleotide cofactor differs between these molybdenum hydroxylases, as appears to be the case, reliable markers for selenium insertion into the molybdenum cofactor may not exist.
In the case of nicotinic acid hydroxylase, computational modeling indicates that the likely selective advantage of using a selenido ligand over a sulfido ligand appears to be to lower the activation energy of the Mo(V) intermediate state formed during the reduction of Mo(VI) to
Mo(IV) (Wagener et al. 2009). This has the predicted effect of increasing the kcat of nicotinic acid hydroxylases using the selenido ligand several orders of magnitude over enzymes using the sulfido ligand in the first coordination sphere. Obviously, substantial questions concerning the nature of the selenium cofactor in xanthine dehydrogenase and purine hydroxylase remain. The mechanisms by which cells specifically insert selenium into the molybdopterin cofactor need to be elucidated. It also needs to be determined whether organisms outside of the purinolytic clostridia actually do incorporate selenium into molybdenum hydroxylases. Addressing these gaps in our knowledge of the biochemistry and physiology of selenium cofactor utilization are required before the evolution of this trait can be addressed.
Selenium respiratory metabolism
A substantial challenge in elucidating the physiology and evolution of selenium respiration is that the ability of a microorganism to reduce a selenium oxyanion is frequently conflated with the ability of that organism to utilize the oxyanion as a terminal electron acceptor in anaerobic respiration. For example, recent reviews have described the bacterium Enterobacter cloacae SLD1a-1 as a Se(VI) respiring organism (Nancharaiah and Lens 2015; Eswayah, Smith
83 and Gardiner 2016). The Se(VI) reductase of En. cloacae SLD1a-1, however, is expressed under both aerobic and anaerobic growth conditions, with maximum activity observed under microaerophilic conditions (Watts et al. 2003; Ridley et al. 2006). The reduction of Se(VI) under nitrate-depleted conditions, moreover, can only generate sufficient proton motive force to maintain viable cells, not to increase cell mass (Leaver, Richardson and Butler 2008). One of the same reviews, moreover, misidentifies a number of organisms as Se(IV) respiring bacteria
(Nancharaiah and Lens 2015). A review of the primary literature, however, demonstrates that
Se(IV) is being reduced by these organisms as a detoxification mechanism. These misidentified bacteria include Desulfovibrio desulfuricans (Tomei et al. 1995), Pseudomonas stutzeri NT-I
(Kuroda et al. 2011a), Shewanella oneidensis MR-1 (Klonowska, Heulin and Vermeglio 2005), and the purple non-sulfur bacteria Rhodospirillum rubrum and Rhodospirillum capsulatis (Kessi
2006).
Given that Se(VI) and Se(IV) reduction is a common mechanism for detoxification under both aerobic and anaerobic conditions, it is vital that researchers demonstrate that a putative
Se(VI) or Se(IV) respiring organism is capable of coupling this reduction to the generation of proton motive force to drive cell growth. Physiologically, this is confirmed by measuring the oxidation of a supplied electron donor and the reduction of a selenium oxyanion in defined media using pure cultures. A stoichiometric mass balance between the amount of electron donor oxidized and the amount of Se(VI) or Se(IV) reduced is confirmation that an organism is capable of selenium oxyanion respiration. Accurately determining the physiological function that selenium oxyanion reduction serves is crucial. The detoxification and respiration of selenium oxyanions may be mediated by different mechanisms, and thus the proteins involved in these two physiological processes may have fundamentally different evolutionary histories.
84 For example, organisms from all three domains of life are capable of detoxifying the arsenic oxyanion arsenate via reduction to arsenite, and numerous bacteria and some archaea can also utilize arsenate as a terminal electron acceptor (Stolz et al. 2006; Andres and Bertin 2016).
The two arsenate reductases mediating detoxification and respiratory reduction, however, come from unrelated protein families. Similarly, the ferric reductases allowing organisms from all three domains of life to reduce ferric iron to ferrous iron for assimilation into proteins and the ferric reductases allowing bacteria and archaea to utilize ferric iron as a terminal electron acceptor are members of unrelated lineages (Schröder, Johnson and de Vries 2003). This section will therefore focus exclusively on organisms that have been shown to be capable of growing via the respiratory reduction of Se(VI), Se(IV), and Se(0). Similarly, the terminal reductases of
Se(VI) and Se(IV) discussed below will only include enzymes that have been determined biochemically or genetically as being participating in an electron transport chain.
Respiratory selenate reduction
Bacteria capable of utilizing Se(VI) as a terminal electron acceptor in anaerobic respiration include organisms from the Betaproteobacteria (Macy et al. 1993),
Gammaproteobacteria (Nakagawa et al. 2006; Narasingarao and Häggblom 2006),
Deltaproteobacteria (Narasingarao and Häggblom 2007b), Epsilonproteobacteria (Stolz et al.
1999), Actinobacteria (von Wintzingerode et al. 2001), Chrysiogenetes (Rauschenbach,
Narasingarao and Häggblom 2011), Deferribacteres (Rauschenbach et al. 2013), and Firmicutes
(Fujita et al. 1997; Switzer Blum et al. 1998; Blum et al. 2001; Niggemyer et al. 2001; Baesman et al. 2009; Abin and Hollibaugh 2017) phyla. Additionally, several hyperthermophilic archaea from the Crenarchaeota phylum are capable of Se(VI) respiration (Huber et al. 2000;
Slobodkina et al. 2015). These organisms have been isolated from a diverse range of habitats
85 including Se(VI) contaminated wastewaters, geothermal springs, estuaries, and soda lakes.
Geographically, these sites range from Mono Lake, California to Kamchatka, Russia. This phylogenetic and ecological diversity is a clear illustration that Se(VI) respiring organisms are ubiquitous. The organisms capable of utilizing Se(VI) as a terminal electron acceptor include facultative anaerobes, microaerophiles, and strict anaerobes, further underscoring that the respiratory reduction of Se(VI) is a crucial component of the global selenium biogeochemical cycle that can sustain life in environments characterized by a wide range of redox regimes.
The biochemistry and physiology of Se(VI) respiration has mainly been dissected in the bacterium Thauera selenatis (Macy et al. 1993). T. selenatis was the first Se(VI) respiring organism to be isolated in a pure culture and reduces Se(VI) to Se(IV) in a two electron transfer reaction (Macy, Michel and Kirsch 1989). The Se(VI) reductase (Ser) was purified from the periplasm of T. selenatis and consists of a heterotrimeric complex with a 96 kDa subunit, a 40 kDa subunit, and a 23 kDa subunit (Schröder et al. 1997). A metals analysis confirmed that
Ser contains molybdenum and iron sulfur clusters, and spectroscopic characterization identified a b-type heme group. Biochemical characterization of Ser found that the enzyme is highly specific for Se(VI) in vitro. Ser also had an affinity for Se(VI) characteristic of a physiological reductase,
-1 -1 with a Km of 16 M and a Vmax of 40 M min mg .
Subsequently, the entire ser operon was sequenced (Krafft et al. 2000). The predicted amino acid sequence of the first gene encoding for the subunit, serA, identified a motif for coordinating a [4Fe-4S] cluster and a molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD) cofactor. The predicted amino acid sequence for the gene coding for the subunit, serB, had motifs for three [4Fe-4S] clusters and a single C-terminal [3Fe-4S] cluster. The fourth gene, serD, appears to be a homolog of the E. coli NarJ protein, which
86 functions in vivo as a chaperone mediating proper insertion of the Mo/W-bisPGD cofactor of the respiratory nitrate reductase subunit NarG (Blasco et al. 1998). All of these features are characteristic of members of a collection of molybdoenzymes that are variously referred to as the
DMSO reductase (DMSOR) (Hille, Hall and Basu 2014), complex iron-sulfur molybdoenzyme
(CISM) (Rothery, Workun and Weiner 2008), and the Mo/W-bisPGD enzyme (Grimaldi et al.
2013) family.
Canonical members of the DMSOR enzyme family additionally contain a subunit that functions as a membrane anchor, connecting the enzyme complex to the quinone pool of the electron transport chain (Rothery, Workun and Weiner 2008). The predicted amino acid sequence of the gene putatively encoding the subunit, serC, however, did not seem to harbor any transmembrane regions, though a putative heme b binding motif was present (Krafft et al.
2000). The lack of transmembrane regions in SerC raised questions as to how Ser is able to abstract electrons from the electron transport chain in the membrane for reduction of Se(VI).
This was later addressed in a detailed dissection of the cytochromes expressed by T. selenatis during growth on Se(VI) (Lowe et al. 2010). A high potential 24 kDa periplasmic diheme cytochrome c4 was purified that could function as an electron donor to Ser in vitro. Furthermore, treatment of T. selenatis cells with quionol-cytochrome c oxidoreductase inhibitors in the presence of Se(VI) substantially reduced growth rates. General quinol dehydrogenase inhibitors could abolish the ability of T. selenatis cells to grow on Se(VI). This provides a model of Se(VI) respiration where quinol-cytochrome c oxidoreductase and other quinol dehydrogenases abstracts electrons from the membrane quinone pool to donate to the cytochrome c4 protein.
This c-type cytochrome can then donate electrons to SerC for the reduction of Se(VI).
87 The cofactors of Ser have received detailed spectroscopic characterization. X ray absorption spectroscopy of the oxidized Mo/W-bisPGD cofactor indicated that the Mo atom is likely coordinated by two dithiolene sulfur ligands, an oxo ligand, and a des-oxo ligand (Maher et al. 2004). The reduced Mo/W-bisPGD cofactor had only two dithioline ligands and a des-oxo ligand, suggesting that Ser reduces Se(VI) via an oxygen atom transfer reaction. Curiously, no amino acid ligand for the Mo atom could be determined. Sequence alignments of SerA with other DMOSR enzymes identified a conserved Asp residue (Dridge et al. 2007) that serves as the
Mo atom ligand in some family members (e.g., nitrate reductase, chlorate reductase, ethylbenzene dehydrogenase) (Grimaldi et al. 2013).
Redox titration of Ser was also able to assign midpoint potentials to the four iron-sulfur clusters of SerB (Dridge et al. 2007). Iron-sulfur clusters are named according to a standard nomenclature, wherein the iron-sulfur cluster in the catalytic subunit is referred to as FS0 and the four iron-sulfur clusters of the electron transfer subunit are named (from N terminus to C terminus) FS1, FS2, FS3, and FS4 (Rothery, Workun and Weiner 2008). The FS1, FS2, and FS3
[4Fe-4S] clusters were assigned midpoint potentials of 183 mV, -356 mV, and -51 mV respectively. The FS4 [3Fe-4S] cluster was assigned a midpoint potential of 118 mV. EPR spectra did not record a discernible signal for the putative FS0 [4Fe-4S] cluster of SerA, but a signal characteristic of Mo(V) was observed, demonstrating that SerA mediates two single electron transfers during the reduction of Se(VI). Optical redox titrations later determined that the heme b of SerC had a midpoint potential of 234 mV (Lowe et al. 2010).
A second respiratory Se(VI) reductase was subsequently identified from Bacillus selenatarsenatis SF-1 using random transposon mutagenesis (Kuroda et al. 2011b). Mutants in which the Se(VI) reducing phenotype was abolished all had the transposon inactivate a single
88 gene, named srdA, whose sequence seemed characteristic of DMSOR family members. This led the authors to identify a srdBCA operon putatively encoding a heterotrimeric enzyme, Srd.
Sequence analysis of the predicted amino acid sequence of srdA indicated that the catalytic subunit was closely related to the tetrathionate reductase catalytic subunit (TtrA) within the family, with a predicted twin arginine translocation motif for export to the periplasm, a [4Fe-4S] cluster, and a Cys ligand for the Mo atom. The predicted amino acid sequence of srdB harbored motifs for four [4Fe-4S] clusters suggesting that SrdB functions as an electron transport subunit.
The putative SrdC protein appeared to have nine transmembrane regions, indicating that SrdC functions as a membrane anchor connecting the Srd catalytic and electron transfer subunits to the quinone pool.
Both sequence analysis and limited phylogenetic evidence indicated that the catalytic subunits mediating Se(VI) reduction, SerA and SrdA, are members of different lineages within the DMSOR family. SerA appeared most closely related to chlorate reductase, ethylbenzene dehydrogenase, and dimethyl sulfide dehydrogenase catalytic subunits (McDevitt et al. 2002;
Stolz et al. 2006). SrdA, however, seemed to be most closely related to the TtrA lineage
(Kuroda et al. 2011b). We have recently utilized robust maximum likelihood methods to demonstrate conclusively that SerA and SrdA homologs are indeed distantly related members of this family (Wells et al. in preparation). Our phylogeny moreover demonstrated that the SrdA subunit of B. selenatarsenatis was not related to the TtrA subunit lineage but was a member of this clade.
We also found that the srdBCA operon of B. selenatarsenatis also harbored a polysulfide reductase-like catalytic subunit (PsrA-like) and a four [4Fe-4S] cluster electron transfer subunit
(PsrB-like). We suggest that these Psr-like subunits could participate in Se(VI) reduction by Srd,
89 and moreover that this distinctive operon may be characteristic of this respiratory Se(VI) reductase. If this is the case, it should be noted that we found only three putative SrdA homologs out of 220 TtrA-like catalytic subunits. These three putative SrdA homologs were found in the genomes of three Bacillales species, B. selenatarsenatis SF-1, Anaerobacillus arseniciselenatis
E1H, and Desulfuribacillus stibiiarsenatis. All of these organisms have been shown to be capable of respiratory Se(VI) reduction (Fujita et al. 1997; Switzer Blum et al. 1998; Abin and
Hollibaugh 2017).
Moreover, a recent DELTA-BLAST (Boratyn et al. 2012) using the T. selenatis SerA protein as query returned only fifteen putative SerA homologs with a query cover of more than
95% and an identity of more than 50%. Of these fifteen sequences, only one came from a known
Se(VI) respiring bacterium, Sedimenticola selenatireducens (Narasingarao and Häggblom 2006).
Another sequence was the chlorate reductase catalytic subunit of Ideonella dechloratans, which shares an identical operon structure with the T. selenatis Ser (Thorell et al. 2003). This suggests that organisms utilizing Ser as a Se(VI) reductase have a very limited phylogenetic distribution as well, and it may not be possible to phylogenetically distinguish SerA catalytic subunits from chlorate reductase catalytic subunits. Consistent with all of these results, DELTA-BLAST searches against the genomes of the Se(VI) respiring organisms Bacillus beveridgei MLTeJB,
Sulfurospirillum barnesii SeS-3, and Desulfitobacterium hafniense do not yield any SerA homologs, and no TtrA homologs are found in Srd-like operons. Clearly, then, more respiratory
Se(VI) reductases await discovery, particularly in the archaeal domain. The evolution of multiple independent mechanisms for Se(VI) respiration is perfectly consistent with interpretations of the Se isotope fractionation data indicating a likely Neoproterozoic Era (~ 1.0
Gya) geochemical origin for Se(VI) (Stüeken et al. 2015a; Stüeken 2017).
90 Selenite and elemental selenium respiration
Only six organisms are known to utilize Se(IV) as a terminal electron acceptor in anaerobic respiration. These include Bacillus selenitireducens MLS10 (Switzer Blum et al.
1998), B. beveridgei MLTeJB (Baesman et al. 2009), and Desulfuribacillus stibiiarsenatis from the Firmicutes phylum, Desulfurisprillum indicum S5 (Rauschenbach, Narasingarao and
Häggblom 2011) from the Chrysiogenetes phylum, and the hyperthermophilic archaea
Pyrobaculum aerophilum (Huber et al. 2000) and Pyrobaculum ferrireducens (Slobodkina et al.
2015) from the Crenarchaeota phylum. During respiratory growth on Se(IV), Se(IV) is reduced to Se(0) in a four electron transfer reaction. It should be noted that while many more Se(VI) respiring organisms have been isolated, the phylogenetic and ecological diversity of Se(IV) respiring prokaryotes is similar to that observed in Se(VI) respiring prokaryotes. This suggests that Se(IV) respiration is as ubiquitous in natural environments as Se(VI) respiration. As with many Se(VI) respiring organisms, Se(IV) respiring prokaryotes include facultative anaerobes, microaerophiles, and strict anaerobes. Thus, respiratory Se(IV) reduction is likely also a central component of the global selenium biogeochemical cycle over a similar range of redox states.
Thus far, Se(IV) respiration has only been characterized biochemically and physiologically in a single organism, B. selenitireducens MLS10 (Wells et al. 2019). We were able to obtain enriched fractions of the respiratory Se(IV) reductase (Srr) catalytic subunit (SrrA) from crude periplasm. SrrA was ~80 kDa and was highly specific for Se(IV). The Km of SrrA for Se(IV) was approximately 145 M. LC MS/MS allowed us to determine that this enzyme was encoded by the Bsel_1476 locus in the B. selenitireducens genome. Sequence analysis of
SrrA strongly suggested that the catalytic subunit was a DMSOR enzyme. The N terminal region had a twin arginine translocation motif, a [4Fe-4S] cluster binding motif, and a Mo/W-
91 bisPGD cofactor binding motif with a Cys residue appearing to be the amino acid ligand for the
Mo atom. A metals analysis of SrrA from enriched fractions was fully consistent with this, finding both molybdenum and iron in the catalytic subunit.
Although SrrA alone appeared sufficient to mediate Se(IV) respiration in vitro, analysis of the neighboring genes around SrrA identified other putative subunits. Consistent with the identification of SrrA as a DMSOR, we found genes encoding a putative four [4Fe-4S] cluster (~
17.7 kDa), a membrane anchor subunit (~ 43 kDa), and a NarJ-like chaperone protein (~ 24 kDa). Adopting the usual nomenclature for DMSOR member subunits, we refer to these proteins as SrrB, SrrC, and SrrD, respectively. Srr likely connects to the electron transport chain in the membrane using the membrane anchoring subunit SrrC to abstract electrons from the quinone pool to SrrB and SrrA. However, we also found two genes that are not characteristic of this family of molybdoenyzmes. Both genes encoded for rhodanese domain-containing proteins with predicted molecular weights of ~ 38 kDa (SrrE) and ~ 45.6 kDa (SrrF). Each protein was predicted to have two rhodanese domains, each with a single active site Cys residue. Consistent with the hypothesis that these rhodanese domain-containing proteins function as part of the Srr complex in vivo, non-denaturing gels could be used to isolate the Srr complex from crude periplasmic fractions. LC MS/MS analysis of the isolated Srr complex found peptides from
SrrA, SrrB, and SrrE with high confidence.
Two phylogenetic analyses of the B. selenitireducens SrrA have established that this catalytic subunit is a member of the polysulfide reductase catalytic subunit (PsrA) and thiosulfate reductase catalytic subunit (PhsA) lineage of the DMSOR family (Wells et al. 2019, Wells et al., in preparation). Both phylogenetic analyses found putative SrrA subunits, but PsrA, PhsA, and
SrrA catalytic subunits did not form monophyletic clades with respect to one another. Putative
92 subunits were designated as members of Psr, Phs, or Srr complexes based on their operon organization. Putative PsrA and PhsA homologs were found in the genomes of both bacteria and archaea. Putative SrrA homologs were confined to bacteria but could be found in deeply branched members of the PsrA/PhsA/SrrA lineage. Crucially, SrrA homologs were found in all known Se(IV) respiring bacteria. No putative Srr operons were found in Se(IV) respiring members of the Pyrobaculum genus, suggesting that mechanisms of Se(IV) respiration may differ between the bacterial and archaeal domains. However, biochemical and physiological characterization of Se(IV) respiration is vital to ascertain that this is the case, as the rhodanese domain-containing proteins may simply not be part of the Srr operon.
A number of questions remain regarding Se(IV) respiration. The most obvious is to establish that the two distinctive rhodanese domain-containing proteins are indeed reliable indicators of the ability of a bacterium to respire Se(IV). It must also be determined whether or not mechanisms of respiratory Se(IV) reduction are conserved between bacteria and archaea. If these mechanisms are conserved, this would demonstrate that PsrA-type subunits likely were catalytically promiscuous in LUCA, capable of polysulfide, thiosulfate, and Se(IV) reductase activity. This would also be consistent with the possibility that a reservoir of bioavailable Se(IV) was present throughout the Archean Eon, as has been suggested recently (Mitchell et al. 2016).
If bacteria and archaea utilize different mechanisms for respiratory Se(IV) reduction, and Srr represents a conserved Se(IV) reductase in bacteria, this would be consistent with a Mesoarchean or Neoarchean Era geochemical origin for Se(IV) as a competing model for the evolution of the selenium biogeochemical cycle has posited (Stüeken et al. 2015a; Stüeken 2017).
93 A number of biochemical and physiological questions also require resolution regarding how Srr functions in vivo. The exact nature of the rhodanese domain-containing proteins requires investigation. Several functions for these proteins are possible. Polysulfide respiration in the bacterium Wolinella succinogenes requires a single rhodanese-domain containing protein with a catalytically active Cys residue (referred to as the Sud protein) that transports polysulfide to the Psr complex, increasing the affinity of the Psr for polysulfide by approximately an order of magnitude (Klimmek et al. 1998). Se(IV) exists in most natural environments in the low M range (see references discussed above for the selenium biogeochemical cycle), and polysulfide as a transient sulfur species is present in natural environments at similar concentrations (Hedderich et al. 1998; Findlay 2016a). Thus, one or both of the rhodanese domain-containing proteins could function to increase the Km of Srr for Se(IV) sufficiently to make physiological Se(IV) respiration viable. Alternatively, as noted above, rhodanese domain-containing proteins and 3- mercaptopyruvate sulfurtransferases have been found to function as efficient selenium delivery
- proteins for HSe and H2Se when bacteria are cultured with Se(IV) as a selenium source for selenium assimilatory metabolism (Ogasawara, Lacourciere and Stadtman 2001; Ogasawara et al. 2005). One or both of these rhodanese domain-containing proteins could function to transport inorganic Se(0) or Se(-II) from Srr after Se(IV) reduction.
Finally, a number of questions remain about how Srr biochemically functions to reduce
Se(IV). All known DMSOR family members mediate two electron reduction or oxidation reactions (Hille, Hall and Basu 2014). Yet respiratory reduction of Se(IV) to Se(0) is a four electron reduction reaction. If Se(IV) is indeed the physiological substrate of Srr, Se(IV) reduction requires either an inherently unstable Se(II) intermediate, or the Srr complex could consist of two functional SrrABCEF dimers. It must be kept in mind that Psr was originally
94 thought to function as a respiratory elemental sulfur reductase (Schröder, Kröger and Macy
1988), and it was only later shown that the physiological substrate was actually polysulfide
(Klimmek et al. 1991). Crystallographic studies of Srr in complex with the selenium substrate are required to determine if Se(IV) is indeed the physiological substrate, and whether intermediate Se(II) states are formed during catalysis. Finally, the Psr and Phs complexes mediate the reduction of sulfur intermediates with low midpoint potentials, as we have noted previously (Wells et al. 2019). The midpoint potentials of polysulfide and thiosulfate are -260 mV and -402 mV (Thauer, Jungermann and Decker 1977; Hedderich et al. 1998). Se(IV), in contrast, has a midpoint potential of 216 mV (Nancharaiah and Lens 2015). Thus, it is likely that the Srr complex would modulate the midpoint potentials of the Mo/W-bisPGD and [4Fe-4S] clusters to facilitate reduction of Se(IV). Spectroscopic characterization of these cofactors can determine whether such modulation has occurred.
Only one bacterium, B. selenitireducens, has been shown to be capable of Se(0) respiration, presumably reducing Se(0) to Se(-II) (Herbel et al. 2003). While only one organism is known to respire Se(0), the same authors observed active Se(0) reduction in anoxic sediments, and therefore respiratory Se(0) reduction is most likely a relevant reaction in the global selenium biogeochemical cycle. Curiously, B. selenitireducens demonstrated greater rates of Se(0) reduction when supplied with biogenic Se(0) nanoparticles rather than abiogenic Se(0). The preference for biogenic Se(0) could be a function of smaller particle size, or proteins associated with biogenic Se(0) that facilitates biological interactions with Se(0). An alternate possibility must also be considered. A crucial difference between Se(0) and elemental sulfur is that Se(0) exists in nature in multiple crystalline and amorphous allotropes (Fernández-Martínez and
Charlet 2009), whereas elemental sulfur exists exclusively as orthorhombic S8 rings under
95 mesophilic conditions (Hedderich et al. 1998; Findlay 2016a). These S8 rings are readily dissolved in the presence of sulfide in aqueous solutions, producing the multivalent intermediate polysulfide, which consists of a single sulfide atom and a number of elemental sulfur atoms.
Selenium biochemistry has not considered the possibility that multivalent selenium intermediates like polyselenide could serve biological functions. Polyselenides, however, are stable in strictly anoxic aqueous solutions and comprise the predominant reduced selenium species in alkaline solutions (pH > 9.0) (Iida et al. 2010), and have been shown to form spontaneously in water when Se(-II) is exposed to O2 (Nuttall and Allen 1984). The preference of B. selenitireducens for biogenic Se(0) then could easily be explained by the fact that some allotropes of Se(0) are more amenable to dissolution by Se(-II) and subsequent polyselenide formation. Thus, Se(0) respiration necessitates sustained biochemical and physiological investigation. The putative respiratory Se(0) reductase of B. selenitireducens remains to be determined, and purification and biochemical characterization of this Se(0) reductase could determine whether Se(0) or polyselenide is the physiological substrate.
It should also be determined whether other Se(IV) respiring bacteria can grow anaerobically when supplied with Se(0). Se(VI) respiring prokaryotes typically produce a red allotrope of Se(0) after reduction of Se(VI) to Se(IV) and detoxification to Se(0). Se(IV) respiring prokaryotes, in contrast, initially produce a similar red Se(0) allotrope that transitions to the black Se(0) allotrope after Se(IV) has been depleted in the culture medium and Se(-II) begins to accumulate (Switzer Blum et al. 1998; Huber et al. 2000; Rauschenbach, Narasingarao and
Häggblom 2011). The transition of Se(0) from a red to a black allotrope could reflect changes in the organization of Se(0) driven by partial dissolution of Se(0) by Se(-II). If polyselenide is the physiological terminal electron acceptor during respiratory growth on Se(0), then polyselenides
96 could have served as a crucial reservoir of bioavailable selenium in the Archean if Se(0) and Se(-
II) were indeed the only species of selenium available (Stüeken et al. 2015a; Stüeken 2017).
Evolutionary analysis of Se(0) or polyselenide reductases could test this hypothesis.
Oxidation of reduced selenium species during chemolithoautotrophic growth
It is curious that the only known specific biological interactions with selenium comprise reductive transformations. This is despite the fact that the first indications of selenium specific metabolism consist of reports of bacteria utilizing Se(-II) (Brenner 1916) and Se(0) (Lipman and
Waksman 1923) as electron donors in aerobic chemolithoauthotrophic growth. As was noted above, physiological evidence that these bacteria were indeed capable of chemolithoautotrophic selenium oxidation were never provided. Regardless, aerobic growth using Se(-II) as an electron donor is not bioenergetically viable, given how rapidly Se(-II) is oxidized by O2 (Nuttall and
Allen 1984). Subsequently, it was demonstrated that bacteria grown in pure cultures
(Sarathchandra and Watkinson 1981; Dowdle and Oremland 1998) and enrichment cultures
(Dowdle and Oremland 1998; Losi and Frankenberger 1998) could oxidize Se(0) to Se(IV), though the rates of Se(0) oxidation were slow and it was never clear whether this oxidation was coupled to the generation of proton motive force. It is intriguing to note that the authors of two of these studies found that biogenic Se(0) was oxidized more readily than abiogenic Se(0)
(Sarathchandra and Watkinson 1981; Losi and Frankenberger 1998), similar to observations of
Se(0) respiration in B. selenitireducens.
To our knowledge, no studies have been conducted to assess for the chemolithoautotrophic selenium oxidation of Se(-II) and Se(0) in anoxic cultures. Both species, and polyselenide, could serve readily as electron donors in anaerobic respiration, or even photoauthotrophic growth, if the oxidative transformations of sulfur species and multivalent
97 intermediates is taken as a reasonable model for the selenium biogeochemical cycle (Ghosh and
Dam 2009; Findlay 2016a). The principle obstacle confronting such studies is the biochemical and analytical difficulties of working with reduced species of selenium. Bacterial cultures could presumably use the reagent sodium selenide as an electron donor, but this reagent can only be purchased in quantities up to 5 g and is quite expensive. H2Se gas cylinders could be exploited as a source Se(-II), but the gas is highly toxic and thus a fume hood would need to be modified to accommodate both the H2Se gas cylinder and a degassing station. Similarly, the remarkable oxidation rate of Se(-II) in the presence of O2 (Nuttall and Allen 1984) also means that analytical techniques to quantify Se(-II) concentrations (such as ion chromatography) would almost certainly have to be done under anoxic conditions. Abiogenic Se(0) can be generated by bubbling an aqueous solution with H2Se and O2 simultaneously (Herbel et al. 2003). Biogenic
Se(0) can either be generated by incubating cell suspensions of organisms capable of Se(IV) detoxification with Se(IV) (Dowdle and Oremland 1998; Herbel et al. 2003) or alternatively by incubating selenite with glutathione (Tarze et al. 2007). Despite these technical and analytic challenges, addressing this significant gap in our knowledge of the selenium biogeochemical cycle should be a priority for future research on selenium respiratory metabolism.
Conclusions
As was noted at the beginning of this essay, research into selenium metabolism has proceeded steadily for over a century now. The field of selenium biochemistry has a number of truly impressive achievements to its credit. The physiology of Sec synthesis and incorporation in bacteria has been meticulously elucidated. For the archaeal domain, the only substantial gap in the physiology of Sec incorporation is how the SelB translation factor interacts with the SECIS element in the 3’ or 5’ untranslated regions whilst still binding to the ribosome to ensure Sec is
98 incorporated into the selenopolypeptide. Similarly, the mechanism of mnm5Se2U synthesis has uncovered the protein mediating the selenium replacement reaction, as well as the basic biochemistry of how the sulfur atom of mnm5S2U is replaced with selenium. The only outstanding questions concern how SelU in bacteria and the SelU N terminus-like and SelU C- terminus like proteins in archaea specifically interacts with mnm5S2U-containing tRNA molecules, and how the two domains of SelU in bacteria and the two proteins in archaea interact together to bind selenophosphate, the tRNA molecule, produce the S-geranyl intermediate, and ultimately replace the S-geranyl intermediate with selenium. Moreover, while the question of why Sec and mnm5Se2U utilization emerged early in the evolution of life and has persisted for approximately 3.7 billion years in some lineages and been lost in many others hasn’t been definitively answered, a recent hypothesis linking the evolution of selenium assimilation to the evolution of antioxidant defense mechanisms is promising for guiding future research.
Nonetheless, fundamental questions in selenium biochemistry remain to be addressed.
The physiological and biochemical function of selenoproteins, particularly those involved in antioxidant defense and redox homeostasis, still need to be elucidated. In the bacteria, this will require researchers to work with oxygen sensitive organisms from deeply branched bacterial lineages to determine under what redox regimes selenoproteins are expressed, and what oxidants serve as substrates for these oxidoreductases. If the ancestral function of selenoproteins in antioxidant defense prior to the GOE has been conserved in bacterial lineages, many of these selenoproteins will function specifically to protect cells from oxidative stress induced by milder oxidants than O2 or reactive oxygen species. These organisms can serve as models for the function of similar selenoproteins in uncultured lineages of Asgard archaea.
99 The third selenium utilization trait, the incorporation of selenium into molybdenum hydroxylase family enzymes as an inorganic Se(-II) cofactor, still requires a great deal of basic physiological and biochemical research. It seems likely that the location of the selenium atom within the first Mo coordination sphere varies between different molybdenum hydroxylase family members. The proteins involved in inserting the Se(-II) cofactor remain unknown, and the only proteins putatively involved in the process remain uncharacterized biochemically. It must also be noted that the only organisms definitely known to incorporate Se(-II) as a cofactor are purinolytic clostridia during fermentative growth on nicontinic acid, hypoxanthine, and xanthine. Yet bioinformatics approaches have suggested that this third utilization trait has a fairly wide phylogenetic distribution, encompassing members of Clostridiales,
Enterobacteriales, and even the Halobacteriales orders in both bacteria and archaea. This suggests that purine and nicotinic acid fermentation could be more widespread among prokaryotes than is currently appreciated, the enzyme homologs in non-purinolytic organisms may have different biochemical and physiological functions, or even that different molybdenum hydroxylase members may use Se(-II) as a cofactor. Identifying specific proteins essential to the insertion of Se(-II) into the molybdopterin cytosine dinucleotide cofactor and determining the distribution of selenium-incorporating molybdenum hydroxylase members is an essential prerequisite to evolutionary study of this selenium utilization trait.
In striking contrast to the impressive accomplishments in our understanding of the physiology and evolution of selenium assimilatory metabolism, research on selenium respiratory metabolism is still very much in its infancy. The fundamental physiological distinction between selenium oxyanion reduction as a detoxification strategy and as a means of generating a proton motive force sufficient to drive ATP synthesis and cell growth still is not widely appreciated.
100 Selenium oxyanions, much less Se(0), are still not routinely tested in physiological characterization of new isolates as potential terminal electron acceptors for anaerobic respiration.
Research into the physiology of Se(VI) respiration has established that respiratory Se(VI) reduction is a comparatively recent biogeochemical adaptation, but the concomitant need to identify Se(VI) reductases in organisms, particularly archaea, that lack Ser or Srd operon homologs isn’t widely appreciated. Similarly, it remains to be determined whether Se(IV) respiration is mediated by multiple different mechanisms, or if respiratory Se(IV) reductases are conserved in bacteria and archaea. The possibility that prokaryotes could utilize reduced selenium species (Se(-II) and Se(0)) as electron donors to fuel autotrophic growth has not yet been considered in the literature and will require basic analytic and biochemical techniques for
Se(-II) and Se(0) production and Se(-II) quantification to be developed and optimized for large- scale cultivation of organisms before it can be considered a viable avenue for physiological and biochemical exploration.
Finally, the single greatest obstacle confronting future research on selenium biochemistry is the lack of a single coherent broadly shared model for how selenium assimilatory and respiratory metabolism is integrated within the selenium biogeochemical cycle. A model is needed that integrates physiological, biochemical, and evolutionary study of model organisms and relevant proteins with the broader biogeochemical evolution of the biosphere over geological time. The interrelationship between the evolution of specific metabolic adaptations (e.g., Sec utilization and Se(IV) respiration) and the availability of particular selenium species in specific geologic eras and environments is required to fully understand the emergence and persistence of selenium metabolism. The principle aim of this essay has been to provide such a paradigm that
101 can ensure the next century of research into the physiology and evolution of selenium metabolism is as productive as the previous one.
102 CHAPTER 3
MATERIALS AND METHODS
Biochemical characterization of the Bacillus selenitireducens Se(IV) reductase
Cultivation and fractionation of MLS10 cells and determination of protein content- Cells of MLS10 were cultivated in the medium described by Switzer Blum et al. (Switzer Blum et al.
-1 1998). The medium consisted of (in g L ): (NH4)2SO4 (0.1), MgSO4 * 7H2O (0.025), K2HPO4
(0.15), KH2PO4 (0.08), NaCl (40), Na2CO3 (10.6), NaHCO3 (4.2), yeast extract (0.2), and 5mL
L-1 of SL-10 trace elements solution. The pH of the medium was adjusted to 9.8 and was made anaerobic by degassing under a 100% N2 atmosphere. The medium was then amended with sodium lactate (10mM) and sodium selenite (10mM) as electron donor and acceptor, respectively, and with cysteine-HCl (0.025%) as a reducing agent. Cells were cultivated in 2L
Pyrex bottles with lids modified to accommodate black rubber septa to prevent O2 contamination. The cells were harvested via centrifugation at 9,000 rpm for 20 min at 4C as described in (Afkar et al. 2003). Periplasmic protein fractions were obtained by suspending the harvested cells in a Tris sucrose buffer containing 50mM Tris-HCl (pH 8.0), 5mM EDTA, and
500mM sucrose. The cells were then treated with lysozyme (0.6 mg mL-1), and incubated at
37C for 15 minutes, and subsequently at 4C for 30 minutes. The cells were then centrifuged at
7232 x g for 15 minutes using an Eppendorf 5810R refrigerated centrifuge (Eppendorf AG,
Hamburg, Germany). The supernatant was the periplasmic fraction. The spheroplasts were then suspended in a Tris buffer containing 20mM Tris-HCl (pH 8.0) and 1mM EDTA and sonicated with a Microson XL-2000 sonicator (Misonix Incorporated, Farmingdale, NY) three times at
12W. The spheroplasts were then centrifuged at 7232 x g for 15 minutes. The sonicated cells were then treated with DNase (0.1mg mL-1) for approximately 15 minutes and centrifuged at
103 7232 x g for 15 minutes. The resulting cell lysate was then ultracentrifuged (Optima XE-90
Beckman Coulter Inc., Brea, CA) at 100,000 x g for 1 hour to obtain the cytoplasmic (the supernatant) and particulate fractions. The particulate fractions were then suspended in the
20mM Tris-HCl (pH 8.0) and 1mM EDTA Tris buffer. The protein concentration of the MLS10 cell lysates, crude fractions, and Se(IV) reductase active fractions after anion exchange chromatography was determined using the Lowry method (Lowry et al. 1951).
In-gel enzyme assays for Se(IV) reductase activity- Non-denaturing gels were made according to standard protocols (Sambrook and Russell 2006), with the exception that CHAPS (3-(3- cholamidopropyl) dimethylammonio)-1-propanesulfonate) was used in lieu of SDS (sodium dodecyl sulfate). Non-denaturing gels consisted of an 8% acrylamide running gel and a 4% acrylamide stacking gel. Approximately 125µg each of periplasmic, particulate, and cytoplasmic fractions were incubated overnight in a non-denaturing 4X Laemmli buffer that omitted both reducing agents and SDS and used 1% CHAPS. The non-denaturing gels were run for 6 hours at
4C and 60V in a reservoir that contained a non-denaturing Tris-glycine buffer with 0.05%
CHAPS substituted for 0.1% SDS.
The gel was then transferred to a Bactron IV anaerobic chamber (Sheldon Manufacturing
Inc. Cornelius, OR) and the in-gel enzyme assay was performed under anoxic conditions. The non-denaturing gel was incubated in a solution containing 10mM methyl viologen reduced with
10mM sodium dithionite, 20mM Tris-HCl (pH 8.0), and 1mM EDTA for several minutes. The gel was then incubated in a solution containing 20mM sodium selenite, 20mM Tris-HCl (pH
8.0), and 1mM EDTA until a band of Se(IV) reductase activity was observed. The non- denaturing gel was then stained overnight in a 0.1% Coomassie R-250, 40% methanol, and 10% glacial acetic acid solution, and destained the following day in a 50% methanol and 10% glacial
104 acetic acid solution. The destained non-denaturing gels were visualized using a GS-800 densitometer (Bio-Rad, Hercules CA). Se(IV) reductase active bands were excised from native gels, and soaked overnight in a 4X Laemmli buffer. The Laemmli buffer supernatant was then used in SDS-PAGE to evaluate the protein composition and estimate the molecular weight of the proteins. SDS-PAGE gels were made following standard protocols (Sambrook and Russell
2006), and the gels were run at 160V for approximately 1.5 hours.
Phylogenetic analysis of the MLS10 SrrA- Homologs of the MLS10 SrrA protein subunit were obtained using the DELTA-BLAST tool (Altschul et al. 1997), and the formate dehydrogenase N alpha subunit of E. coli was selected as an outgroup (Jormakka et al. 2002).
The genomic context of each putative MLS10 SrrA homolog was then evaluated using the
Integrated Microbial Genomes tool (Markowitz et al. 2006). The homologs were aligned using the MUSCLE tool (Edgar 2004) available on the MEGA 7 program (Kumar, Stecher and Tamura
2016). The alignments were then trimmed using the trimAl tool (Capella-Gutiérrez, Silla-
Martínez and Gabaldón 2009). The sequence alignments were trimmed such that all columns with gaps in more than 20% of the SrrA homolog sequences, or columns with a similarity score below 0.001 were omitted, with the caveat that 60% of the columns were to be conserved. This resulted in a trimmed sequence alignment of 722 amino acids for phylogenetic analysis. The alignment was then manually inspected to ensure that the putative [4Fe-4S] sequence motif at the
N terminus of the SrrA homologs and the E. coli formate dehydrogenase N alpha subunit was aligned, as was the cysteine residue predicted to coordinate the molybdopterin guanine dinucleotide cofactor.
Bayesian phylogenies were constructed using the MrBayes 3.2.6 phylogenetic program
(Ronquist et al. 2012). The phylogenetic analysis consisted of 2 parallel Metropolis-coupled
105 Markov chain Monte Carlo (MCMC) analyses (Altekar et al. 2004) with 4 chains each. Both the
ProtTest software program (Darriba et al. 2011) and MrBayes (Ronquist et al. 2012) were used to evaluate probable amino acid substitution models and rate variation models. The four most probable combinations of amino acid substitution and rate variation models were chosen from the ProtTest program, and the most probable amino acid substitution model was chosen by
MrBayes by specifying a mixed amino acid substitution model prior to the analysis when analyzing the data using gamma and gamma with a proportion of invariant sites distributed rate variation models. All tested models were run in triplicates and yielded identical topologies, with consistent levels of posterior probability support at each node. Convergence of the MCMC runs was tested using the Tracer 1.6.0 software program (Rambaut et al. 2018). The resulting trees were visualized with the iTOL program (Letunic and Bork 2016).
Enrichment of MLS10 SrrA- Se(IV) reductase active fractions highly enriched in SrrA were obtained using a Toyopearl GigaCap® DEAE 650M anion exchange resin (Tosoh
Bioscience, Tokyo Japan) for anion exchange chromatography. Chromatographic separation of proteins was performed using the Bio-Rad BioLogic LP chromatography system (Hercules CA).
Periplasmic protein fractions were loaded onto the resin for the first anion exchange chromatography run, and the proteins were eluted in a 20mM Tris-HCl (pH 8.0) buffer using a
NaCl gradient (0-500mM NaCl). Elution of proteins off the column was measured using a UV monitor at 280nm. Se(IV) reductase activity was assessed in a 100mM phosphate
(KH2PO4/NaOH) buffer (pH 7.5) with 500µM methyl viologen and 500µM sodium selenite.
Assays were performed in 2mL volumes with approximately 50µg total protein, and the solution was made anaerobic by degassing in a 100% N2 atmosphere. The reaction was then initiated by the addition of 50µM sodium dithionite. Oxidation of methyl viologen was monitored at 600nm
106 using a T3 single beam spectrophotometer (Persee, Auburn CA). After screening of the first
Se(IV) reductase active fractions, two fractions were selected to be run down a smaller anion exchange column using the identical anion exchange resin. The fractions were diluted in an equal volume of 20mM Tris-HCl 1mM EDTA (pH 8.0) buffer to reduce the salt concentration.
The NaCl gradient employed was identical to the first run, as was the procedure for screening
Se(IV) reductase fractions after elution from the second column. The purity of the Se(IV) reductase active fractions was assessed using SDS-PAGE. SDS-PAGE gels were made according to standard protocols (Sambrook and Russell 2006). Running gels consisted of 8% acrylamide and stacking gels consisted of 4% acrylamide. Approximately 20µg total protein was suspended in a 4X Laemmli sample buffer and then incubated at 90C for 5 minutes. Gels were then run at 160V for approximately 1.5 hours. The gels were stained overnight, destained, and visualized as described above for non-denaturing gels.
Kinetic analysis of SrrA- The Km of the MLS10 SrrA was obtained using the Se(IV) reductase assay described above, with the exception that the concentration of Se(IV) was adjusted to include the following concentrations: 10µM, 50µM, 100µM, 250µM, 500µM, and
750µM. Each Se(IV) concentration was assessed in triplicate. All Se(IV) reductase assays contained approximately 150g total protein from enriched fractions and were initiated with the addition of 50µM sodium dithionite and monitored at 600nm using a T3 single beam spectrophotometer (Persee, Auburn CA). Se(IV) reductase activity was determined by calculating the amount of methyl viologen oxidized (µM min-1) using the extinction coefficient
-1 -1 for methyl viologen of 13 mM cm (52). The Km, Vmax, and kcat of SrrA were calculated with the Prism 7 (GraphPad Software, La Jolla, CA) statistical analysis software package using the
Michaelis-Menten equation with a non-linear regression fit. The kcat was calculated using the
107 concentration of SrrA in the 2mL cuvette as a constant. The substrate specificity of SrrA was assessed using the same assay, but with 500µM of the following alternative electron acceptors: arsenate, selenate, and thiosulfate.
UV-Vis analysis of enriched fractions of SrrA- Fractions enriched with SrrA were analyzed with a Shimadzu UV-1800 dual beam spectrophotometer (Shimadzu Scientific
Instruments, Columbia MD). 1mL of the enriched fraction was analyzed in a cuvette over a wavelength range of 800-300nm to obtain an oxidized spectrum. A reduced spectrum was obtained by the addition of crystals of sodium dithionite. In both cases, 1mL of blank Tris-HCl
(pH 8.0) containing 20mM Tris-HCl, 1mM EDTA, and 500mM NaCl was used as a control in the reference cuvette.
Metals analysis of SrrA- Approximately 3.6mg total protein (~800L volume) from active fractions after elution down the 1st anion exchange column was filtered through a 0.22 μm
PES filter (VWR, Bridgeport, NJ) for particle removal and diluted 1:100 in a 2% nitric acid solution. The samples were subsequently analyzed on a NexION X300 ICP-MS with S10
Autosampler and the NexION 300x ICP-MS software (Perkin Elmer, San Jose CA). CPI international single element standards were used to create the multi-element standards for ICP-
MS. All solutions were stabilized in 2% nitric acid (sub-boil distilled trace metal grade). All blanks, standards, and samples were spiked in-line with a beryllium, germanium and thallium mix solution (Internal Standard). The beryllium, germanium and thallium mix solution was made from single element standards (CPI International, Santa Rosa CA). 2% nitric acid was used as the rinse solution.
LC MS/MS analysis of protein bands- Bands displaying Se(IV) reductase activity in non- denaturing gels, and protein bands in SDS-PAGE gels of fractions enriched in SrrA were excised
108 and diced into approximately 1mm long pieces and the Coomassie R-250 stain removed by several washes in a 50% acetonitrile 25mM ammonium bicarbonate buffer. The gel bands were then stored in a 25mM ammonium bicarbonate buffer. Samples were digested with trypsin and loaded onto a 100 µm x 2 cm Acclaim PepMap 100 C18 nano trap column (5 µm, 100 Å) with an Ultimate 3000 liquid chromatograph (Thermo Fisher, Pittsburgh PA) at 3 µl/min. The peptides were separated on a silica capillary column that was custom packed with C18 reverse phase material (Halo peptide, 2.7 µM, 160 Å). The samples were run at 0.3 µl/min starting with
96% solvent A (100% water, 0.1% formic acid) and 4% solvent B (100% acetonitrile, 0.1% formic acid) over 25 minutes followed by a 45-minute gradient from 10% to 90% solvent B. The column effluent was directly coupled to the Q-exactive orbitrap mass spectrometer using a Flex ion source (Thermo Fisher, Pittsburgh PA). The mass spectrometer was controlled by Xcalibur
2.2 software and operated in data-dependent acquisition mode. Spectra of peptide mass were acquired from an m/z range of 300-2000 at a high mass resolving power. The top 25 most abundant charged ions were subjected to higher-energy collisional dissociation (HCD) with an m/z range of 300-2000 with intensity threshold of 3.3e4, charge states less than 7 and unassigned charge states excluded.
Data analysis. The higher-energy collisional dissociated data were extracted using Proteome
Discoverer using Sequest version 1.4.1.14 (Thermo Fisher, Pittsburgh PA). The files were searched against the Uniprot B. selenitireducans MLS10 database which contains 3231 proteins.
Sequest was searched with a fragment ion tolerance of 0.020 Da and a parent ion tolerance of 10 pm. Carbamidomethylation of cysteine was specified as a fixed modification. The enzyme specificity was set to trypsin with missed cleavage of 3. A target decoy data base search was performed. Peptide identifications were accepted with a target false discovery rate (FDR)
109 between 0.05 (relaxed, moderate confidence) and 0.01 (strict, high confidence). The validation was based on the q-value which is the minimal FDR at which the id is considered correct. A post database search filter was applied with an Xcorr value greater than 2 for high peptide confidence.
Identification of Se(IV) reductases in bacteria with homologs to the MLS10 respiratory
Se(IV) reductase
Cultivation and fractionation of Thermus scotoductus SA-01, Desulfitobacterium hafniense PCP-1, and Bacillus beveridgei MLTeJB and determination of protein content- Two media were used for cultivation of Thermus scotoductus SA-01 cells. A complex Castenholtz
TYE medium was prepared as recommended by the American Type Culture Collection and
-1 contained (in g L ): nitrilotriacetic acid (0.1), CaSO4 * 2H2O (0.06), MgSO4 * 7H2O (0.1), NaCl
(0.008), KNO3 (0.105), NaNO3 (0.7), Na2HPO4 (0.11), tryptone (1.0), yeast extract (1.0), and
-1 additionally 1 mL L of a 0.3% FeCl3 solution and Nitsch’s trace elements. Nitsch’s trace
-1 elements consisted of the following (in g L ): MnSO4 * H2O (2.2), ZnSO4 * 7H2O (0.5), H3BO3
-1 (0.5), CuSO4 * 5H2O (0.016), Na2MoO4 * 2H2O (0.025), CoCl2 * 6H2O (0.046), and 0.5 mL L
H2SO4. The defined medium used to cultivate T. scotoductus SA-01 was prepared as previously described (Kieft et al. 1999) during the isolation of this organism. The medium consisted of (in
-1 g L ): KH2PO4 (0.42), K2HPO4 (0.22), NH4Cl (0.2), KCl (0.38), NaCl (0.36), CaCl2 * 2H2O
-1 (0.04), MgSO4 * 7H2O (0.1), NaHCO3 (1.8), Na2CO3 (0.5), and 10 mL L of 10X Wolfe’s trace elements solution and 15 mL L-1 of 10X Wolfe’s vitamins solution. The 10X Wolfe’s trace elements and 10X Wolfe’s vitamins solutions were prepared as described by Kieft et al. The pH of the complex medium was adjusted to 8.2, and the pH of the defined medium was adjusted to
6.8. The pH of both media was adjusted prior to degassing.
110 Desulfitobacterium hafniense PCP-1 was cultivated in the medium recommended by the
ATCC, though the trace elements and vitamins solutions utilized for cultivation of PCP-1 were the solutions recommended by the German Collection of Microorganisms and Cell Cultures
-1 (DSMZ). The medium contained (in g L ): NH4Cl (1.0), MgCl2 * 6H2O (0.1), FeCl2 (0.002),
CaCl2 * 2H2O (0.05), KH2PO4 (0.27), K2HPO4 (0.35), NaHCO3 (2.6), yeast extract (1.0), and 1 mL L-1 SL-10 trace elements solution and 10 mL L-1 of a vitamin solution. Note that the addition of NaHCO3 was recommended by the DSMZ to prevent the lowering of the medium’s pH due to addition of CO2 during degassing. The vitamin solution was prepared as described previously for cultivation of Sulfurospirillum barnesii SeS-3 (Stolz et al. 1997). The pH of the medium was adjusted to 7.2 and subsequently degassed with an 80% N2 / 20% CO2 atmosphere using a heated copper column to remove traces of O2. The medium used for the cultivation of
Bacillus beveridgei MLTeJB is the same medium previously utilized for the isolation of
-1 MLTeJB (Baesman et al. 2009). The medium consisted of (in g L ): (NH4)SO4 (0.1), MgSO4 *
7H2O (0.025), K2HPO4 (0.15), KH2PO4 (0.08), NaCl (50), Na2CO3 (10.6), NaHCO3 (4.2), yeast extract (0.2), and 5mL L-1 of SL-10 trace elements solution. The pH of the medium was adjusted to 9.8 and was made anaerobic by degassing under a 100% N2 atmosphere.
All media were amended with sodium lactate (10mM) and sodium selenite (10mM) as electron donor and acceptor, respectively, and with cysteine-HCl (0.025%) as a reducing agent.
Cells were cultivated in 2L Pyrex bottles with lids modified to accommodate black rubber septa to prevent O2 contamination. T. scotoductus SA-01 was cultivated at 65C. D. hafniense PCP-1 and B. beveridgei MLTeJB were cultivated at 37C. The procedures for harvesting the cells and obtained crude fractions were identical to those described in (Wells et al. 2019). The cells were harvested via centrifugation at 9,000 rpm for 20 min at 4C. Periplasmic protein fractions were
111 obtained by suspending the harvested cells in a Tris sucrose buffer containing 50mM Tris-HCl
(pH 8.0), 5mM EDTA, and 500mM sucrose. The cells were then treated with lysozyme (0.6 mg mL-1), and incubated at 37C for 15 minutes, and subsequently at 4C for 30 minutes. The cells were then centrifuged at 7232 x g for 15 minutes using an Eppendorf 5810R refrigerated centrifuge (Eppendorf AG, Hamburg, Germany). The supernatant was the periplasmic fraction.
The spheroplasts were then suspended in a Tris buffer containing 20mM Tris-HCl (pH 8.0) and
1mM EDTA and sonicated with a Microson XL-2000 sonicator (Misonix Incorporated,
Farmingdale, NY) three times at 12W. The spheroplasts were then centrifuged at 7232 x g for
15 minutes. The sonicated cells were then treated with DNase (0.1mg mL-1) for approximately
15 minutes and centrifuged at 7232 x g for 15 minutes. The resulting cell lysate was then ultracentrifuged (Optima XE-90 Beckman Coulter Inc., Brea, CA) at 100,000 x g for 1 hour to obtain the cytoplasmic (the supernatant) and particulate fractions. The particulate fractions were then suspended in the 20mM Tris-HCl (pH 8.0) and 1mM EDTA Tris buffer. The protein concentration of the MLS10 cell lysates, crude fractions, and Se(IV) reductase active fractions after anion exchange chromatography was determined using the Lowry method (Lowry et al.
1951).
In-gel enzyme assays for Se(IV) reductase activity- Non-denaturing gels were made according to standard protocols (Sambrook and Russell 2006), with the exception that CHAPS
(3-(3-cholamidopropyl) dimethylammonio)-1-propanesulfonate) was used in lieu of SDS
(sodium dodecyl sulfate). Non-denaturing gels consisted of an 10% acrylamide running gel and a 5% acrylamide stacking gel. Approximately 300-400 µg each of periplasmic, particulate, and cytoplasmic fractions for PCP-1 and MLTeJB and approximately 150 µg for SA-01 crude fractions were incubated overnight in a non-denaturing 4X Laemmli buffer that omitted both
112 reducing agents and SDS and used 1% CHAPS. For PCP-1, an enriched fraction of 35% ammonium sulfate (approximately 300 µg) was also subjected to electrophoretic separation. The non-denaturing gels were run for 6 hours at 4C and 60V in a reservoir that contained a non- denaturing Tris-glycine buffer with 0.05% CHAPS substituted for 0.1% SDS.
The gels were then transferred to a Bactron IV anaerobic chamber (Sheldon
Manufacturing Inc. Cornelius, OR) and the in-gel enzyme assay was performed under anoxic conditions. The non-denaturing gel was incubated in a solution containing 10mM methyl viologen reduced with 10mM sodium dithionite in a 100 mM phosphate (KH2PO4/NaOH) buffer
(pH 7.5) for several minutes. The gel was then incubated in a solution containing 20mM sodium selenite dissolved in a 100 mM phosphate (KH2PO4/NaOH) buffer (pH 7.5) until a band of
Se(IV) reductase activity was observed. The non-denaturing gel was then stained overnight in a
0.1% Coomassie R-250, 40% methanol, and 10% glacial acetic acid solution, and destained the following day in a 50% methanol and 10% glacial acetic acid solution. The destained non- denaturing gels were visualized using a GS-800 densitometer (Bio-Rad, Hercules CA). Se(IV) reductase active bands were excised from native gels, and soaked overnight in a 4X Laemmli buffer. The Laemmli buffer supernatant was then used in SDS-PAGE to evaluate the protein composition and estimate the molecular weight of the proteins. SDS-PAGE gels were made following standard protocols (Sambrook and Russell 2006) and the gels were run at 160V for approximately 1.5 hours.
UV-Vis analysis and cytochrome content- Crude fractions of T. scotoductus SA-01, D. hafniense PCP-1, and B. beveridgei MLTeJB were analyzed with the pyridine hemochromagen assay to look for cytochromes that could be expressed during Se(IV) or Se(VI) respiration.
Solutions for the pyridine hemochrome assay were prepared according to standard protocols
113 (Barr and Guo 2015). Solutions consisted of approximately 500 µg protein, 0.1 M NaOH, 20% pyridine, and 250µM potassium ferricyanide(III) in a 50 mM phosphate (KH2PO4/NaOH) buffer
(pH 7.5). The addition of potassium ferricyanide assured that the protein sample was fully oxidized. Spectra were recorded over a range of 700 – 300 nm using a Shimadzu UV-1800 dual beam spectrophotometer (Shimadzu Scientific Instruments, Columbia MD). After an oxidized spectrum was obtained, a reduced spectrum was obtained after the addition of several crystals of sodium dithionite. A solution of 0.1 M NaOH, 20% pyridine, and 250µM potassium ferricyanide(III) in a 50 mM phosphate (KH2PO4/NaOH) buffer (pH 7.5) was used as a control in the reference cuvette.
LC MS/MS analysis of protein bands- Protein bands from SDS-PAGE gels containing protein eluted from Se(IV) reducing bands were excised and diced into approximately 1mm long pieces and the Coomassie R-250 stain removed by several washes in a 50% acetonitrile 25mM ammonium bicarbonate buffer. The gel bands were then stored in a 25mM ammonium bicarbonate buffer. Samples were digested with trypsin and loaded onto a 100 µm x 2 cm
Acclaim PepMap 100 C18 nano trap column (5 µm, 100 Å) with an Ultimate 3000 liquid chromatograph (Thermo Fisher, Pittsburgh PA) at 3 µl/min. The peptides were separated on a silica capillary column that was custom packed with C18 reverse phase material (Halo peptide,
2.7 µM, 160 Å). The samples were run at 0.3 µl/min starting with 96% solvent A (100% water,
0.1% formic acid) and 4% solvent B (100% acetonitrile, 0.1% formic acid) over 25 minutes followed by a 45-minute gradient from 10% to 90% solvent B. The column effluent was directly coupled to the Q-exactive orbitrap mass spectrometer using a Flex ion source (Thermo Fisher,
Pittsburgh PA). The mass spectrometer was controlled by Xcalibur 2.2 software and operated in data-dependent acquisition mode. Spectra of peptide mass were acquired from an m/z range of
114 300-2000 at a high mass resolving power. The top 25 most abundant charged ions were subjected to higher-energy collisional dissociation (HCD) with an m/z range of 300-2000 with intensity threshold of 3.3e4, charge states less than 7 and unassigned charge states excluded.
Data analysis. The higher-energy collisional dissociated data were extracted using
Proteome Discoverer using Sequest version 1.4.1.14 (Thermo Fisher, Pittsburgh PA). The files were searched against the Uniprot Bacillus beverdigei MLTeJB and the NCBI
Desulfitobacterium hafniense PCP-1 database which contains 3245 and 5074 proteins, respectively. Sequest was searched with a fragment ion tolerance of 0.020 Da and a parent ion tolerance of 10 pm. Carbamidomethylation of cysteine was specified as a fixed modification.
The enzyme specificity was set to trypsin with missed cleavage of 3. A target decoy data base search was performed. Peptide identifications were accepted with a target false discovery rate
(FDR) between 0.05 (relaxed, moderate confidence) and 0.01 (strict, high confidence). The validation was based on the q-value which is the minimal FDR at which the id is considered correct. A post database search filter was applied with an Xcorr value greater than 2 for high peptide confidence.
Phylogenetic analysis of the respiratory Se(IV) reductase of B. selenitireducens
Sequence selection- Protein representatives from DMSOR family lineages where the function of the enzyme was previously elucidated by biochemical, genetic, or structural methods were selected to be BLAST queries to construct a comprehensive library of sequences. For the families we analyzed, these included PsrA from W. succinogenes (Krafft et al. 1992), PhsA from
S. enterica serovar Typhimurium (Heinzinger et al. 1995), SrrA from B. selenitireducens (Wells et al. 2019), ArrA from Chrysiogenes arsenatis (Krafft and Macy 1998) and B. selenitireducens
(Afkar et al. 2003), ArxA from A. ehrlichii (Zargar et al. 2010), TtrA from S. enterica serovar
115 Typhimurium (Hensel et al. 1999), SrdA from B. selenatarsenatis (Kuroda et al. 2011b), the arsenate reductase of P. aerophilum (Cozen et al. 2009), AioA from A. faecalis (Ellis et al. 2001) and Rhizobium sp. NT-26 (Warelow et al. 2017), FdhG from Escherichia coli (Jormakka et al.
2002), D. gigas (Raaijmakers et al. 2002), and D. desulfuricans (Raaijmakers et al. 2002), FdhH from E. coli (Boyington et al. 1997), NAD-dependent formate dehydrogenases from Moorella thermoacetica (Yamamoto et al. 1983) and Peptoclostridium acidaminophilum (Graentzdoerffer et al. 2003), and the F420-dependent formate dehydrogenases of Methanococcus maripaludis
(Wood, Haydock and Leigh 2003) and M. vannielli (Jones and Stadtman 1981). These queries were blasted against the bacterial and archaeal phyla described by Wu et al. (Wu et al. 2009).
Homologous proteins across the bacterial and archaeal domains were identified using the
DELTA-BLAST tool from the National Center for Biotechnology Information (NCBI) web server (Boratyn et al. 2012).
Candidates were selected if the query aligned over at least 95% of the homologous sequence with an amino acid identify of at least 30%, if the sequence length was consistent with the sequence length of the query, and if the primary sequence contained motifs considered characteristic of the enzyme family (e.g., a twin-arginine translocation motif, a [4Fe-4S] or [3Fe-
4S] cluster binding motif, and a Mo/W-bisPGD binding motif). Candidates were additionally screened using the Integrated Microbial Genomics (IMG) (Markowitz et al. 2012) platform to view the genomic context of the putative homolog. Sequences were retained only if the primary sequences between the NCBI database and IMG database were conserved, and if the operon contained other subunits consistent with the operon structure described in model organisms previously (e.g., a four [4Fe-4S] cluster containing protein, a [2Fe-2S] Rieske protein, a membrane anchor).
116 Additional well-characterized representatives from other DMSOR family lineages were included in the analysis to provide a fuller context of where the analyzed families fit in the larger superfamily. These proteins included NapA from D. desulfuricans (Dias et al. 1999), E. coli
(Thomas, Potter and Cole 1999), and Rhodobacter sphaeroides (Pignol et al. 2001), NarG from
E. coli (Bertero et al. 2003), P.aerophilium (Afshar et al. 2001), and Thermus thermophilus
(Ramírez-Arcos, Fernández-Herrero and Berenguer 1998), DmsA from E. coli (Weiner et al.
1988), chlorate reductase (ClrA) from Ideonella dechloratans (Thorell et al. 2003), SerA from
Thauera selenatis (Schröder et al. 1997), dimethyl sulfide dehydrogenase (DdhA) from
Rhodovulum sulfidophilum (McDevitt et al. 2002), TorA from E. coli (Méjean et al. 1994) and
Shewanella massilia (Czjzek et al. 1998), and BisC from E. coli (Pierson and Campbell 1990).
Well-characterized members of the aldehyde:ferredoxin reductase family were included as outgroups. These proteins included the aldehyde ferredoxin oxidoreductases (AORs) of M. thermoacetica (White et al. 1989) and Pyrococcus furiosus (Mukund and Adams 1991), the formaldehyde ferredoxin oxidoreductases (FORs) of Pyr. furiosus (Hu et al. 1999) and
Thermococcus litoralis (Mukund and Adams 1993), and the glyceraldehyde-3-phosphate ferredoxin oxidoreductases (GAPORs) of Pyr. furiosus (Mukund and Adams 1995), M. maripaludis (Park, Mizutani and Jones 2007), and P. aerophilum (Reher, Gebhard and Schönheit
2007).
Phylogenetic analysis- A total of 1,077 sequences were included for the phylogenetic analysis. The sequences were aligned using the online platform of MAFFT (Katoh, Rozewicki and Yamada 2017). Alignments were made using the G-INS-1 method. The amino acid selection model that best fit our data was chosen using the ModelFinder program
(Kalyaanamoorthy et al. 2017), which compared 546 different amino acid models and
117 determined whether the rate heterogeneity in the amino acid selection model was either gamma or free rate distributed. The best fitting model for our data was found to be the LG model (Le and Gascuel 2008) with free rate distributed rate heterogeneity with 10 rate categories.
Phylogenies were constructed with maximum likelihood in RAxML Version 8 (Stamatakis 2014) using the LG model with gamma distributed rate heterogeneity to find the best scoring tree. The best tree topology was obtained from 10 independent maximum likelihood tree searches starting on parsimony trees and support was determined from 100 parametric bootstrap replicates.
Phylogenies were also constructed in IQ-TREE (Nguyen et al. 2015) version 1.6.10 using the LG model with free rate distributed rate variation with 10 rate categories. The maximum likelihood analysis was initiated from 100 initial parsimony trees, and the 20 best scoring trees were refined through a series of nearest neighbor interchanges, and the support for the most likely consensus tree was determined from 1,000 and 10,000 parametric UF bootstrap (Hoang et al. 2018) replicates. The ancestral sequence reconstruction analysis was performed using RAxML. The relative evolutionary rate inference analysis was performed using LEISR (Spielman and
Kosakovsky Pond 2018), a tool that is modelled on the program Rate4Site (Pupko et al. 2002), available in the HyPhy package (Pond, Frost and Muse 2005). Model selection and the phylogenetic and ancestral sequence reconstruction analyses were performed using the CIPRES gateway portal (Miller, Pfeiffer and Schwartz 2010). The relative evolutionary rate inference analyses were done using a desktop computer.
118 CHAPTER 4
THE RESPIRATORY SELENITE REDUCTASE FROM BACILLUS SELENITIREDUCENS
STRAIN MLS10
Abstract
The putative respiratory Se(IV) reductase (Srr) from Bacillus selenitireducens MLS10 has been identified through a polyphasic approach involving genomics, proteomics, and enzymology. In gel assays were used to identify Srr in cell fractions and the active band was shown to contain a single protein of 80 kDa. The protein was identified through ICP-MS to be a homolog of the catalytic subunit of polysulfide reductase (PsrA). Further, the operon structure was found to contain six genes, srrE, srrA, srrB, srrC, srrD, srrF, with SrrA the catalytic subunit (80.09 kDa) that contained a TAT leader sequence indicative of a periplasmic protein and one putative
[4Fe-4S] binding site, SrrB a small subunit (17.74 kDa) with four putative 4Fe4S binding sites,
SrrC (43.14 kDa) an anchoring subunit, and SrrD (24.2 kDa) a chaperon protein. Both SrrE
(38.44 kDa) and SrrF (45.56 kdA) where annotated as rhodanese-domain containing proteins.
Phylogenetic analysis revealed that SrrA belonged to the Psr/Phs clade, but that it did not form a distinct cluster, based on the homologs that were identified from other known selenite respiring bacteria (e.g., Desulfurispirillum indicum, Pyrobaculum aerophilum). The enzyme appeared to be specific for Se(IV), showing no activity with selenate, arsenate, or thiosulfate, with a Km of
142 M.
Importance
This report identifies for the first time the dissimilatory Se(IV) reductase (Srr) from the haloalkalophilic bacterium Bacillus selenitireducens strain MLS10. Selenium is an essential element for life and Se(IV) reduction a key step in its biogeochemical cycle. Thus identification
119 of Srr has further elucidated the biochemical pathway involved. The work extends the versatility of the DMSO reductase family of mononuclear molybdenum enzymes in electron transfer involving chalcogen substrates with different redox potentials. Further, it underscores the importance of biochemical and enzymological approaches in establishing the functionality of these enzymes that may or may not have been correctly annotated.
Introduction
The selenium biogeochemical cycle sustains life primarily through the assimilatory and dissimilatory reduction of the selenium oxyanions selenate (Se(VI)) and selenite (Se(IV)). The assimilatory reduction of the selenium oxyanion Se(IV) to selenide is integral to the synthesis of the 21st amino acid selenocysteine (Kryukov et al. 2003; Carlson et al. 2004; Kryukov and
Gladyshev 2004; Stolz et al. 2006) and the tRNA nucleoside 2-selenouridine (Veres et al. 1992b;
Wolfe et al. 2004c). The assimilatory reduction of Se(IV) for selenocysteine synthesis occurs in all three domains of life, but the assimilatory reduction of Se(IV) for 2-selenouridine synthesis occurs exclusively in bacteria and archaea. The second major component of the selenium biogeochemical cycle is the dissimilatory reduction of the selenium oxyanions Se(VI) and Se(IV) during anaerobic respiration (Stolz et al. 2006; Nancharaiah and Lens 2015). Se(VI) and Se(IV) respiring microorganisms have been isolated from both the bacterial and archaeal domains, including the Proteobacteria (Macy et al. 1993; Oremland et al. 1994; Nakagawa et al. 2006;
Narasingarao and Häggblom 2006, 2007b), the Firmicutes (Switzer Blum et al. 1998; Blum et al.
2001; Yamamura et al. 2007; Abin and Hollibaugh 2017), the Chrysiogenetes (Rauschenbach,
Narasingarao and Häggblom 2011), and the Crenarchaeota (Huber et al. 2000; Slobodkina et al.
2015) phyla. The phylogenetic breadth represented by selenium respiring prokaryotes suggests that selenium respiration is widespread in the Bacteria and Archaea.
120 A stunning array of bacteria are capable of rapidly and efficiently detoxifying up to millimolar quantities of both Se(VI) and Se(IV) aerobically and anaerobically via reduction to red elemental selenium (Se(0)) via several different mechanisms (Nancharaiah and Lens 2015).
The phenomenon has also been observed in archaea (Güven et al. 2013) and eukaryotes
(Nickerson and Falcone 1963; Rosenfeld et al. 2017). This poses a significant challenge in determining whether an anaerobically grown organism reducing Se(VI) or Se(IV) is coupling that reduction to energy conservation, or is merely a detoxification strategy. An additional possibility, heretofore not considered in the literature, is that Se(VI) and Se(IV) could also serve as electron sinks during fermentative growth. There are examples of organisms known to exploit elemental sulfur and polysulfides as electron sinks during fermentation (Schönheit and Schäfer
1995; Hedderich et al. 1998), and selenium is a more voracious electrophile than sulfur (Reich and Hondal 2016). Thus, this possibility requires more experimental evidence to confirm if an organism is capable of Se(VI) or Se(IV) respiration (e.g., demonstrating that Se(VI) or Se(IV) reduction is being stoichiometrically coupled to the oxidation of an electron donor).
Of the two respiratory processes, Se(VI) respiration has been studied most extensively.
During anaerobic respiration on Se(VI), Se(VI) is reduced to Se(IV) in a two electron transfer reaction (Macy, Michel and Kirsch 1989; Switzer Blum et al. 1998), with the Se(IV) subsequently reduced to red Se(0), presumably as a detoxification mechanism. Multiple biochemical pathways and enzymes are utilized during Se(VI) respiration, however, all known respiratory Se(VI) reductases are members of the dimethyl sulfoxide reductase (DSMOR) family
(Rothery, Workun and Weiner 2008), suggesting that Se(VI) respiration has independently evolved multiple times in bacteria. The periplasmic Se(VI) reductase (Ser) of T. selenatis has been purified, and its operon sequenced (Schröder et al. 1997; Krafft et al. 2000). The active
121 enzyme was purified to homogeneity and consisted of three subunits: SerA, SerB, and SerC, with an apparent molecular mass of 180 kDa (28). The operon was found to contain the genes encoding the large catalytic subunit, serA, with close amino acid sequence homology to both nitrate and chlorate reductases in the DMSOR family (Stolz et al. 2006; Harel et al. 2016), the
[4Fe-4S] electron transfer subunit, serB, and the cytochrome b containing subunit, serC.
Additionally, a fourth open reading frame, serD, that overlaps with both ser B and serC, has homology to the chaperone protein, NarJ from E. coli nitrate reductase (29). The Se(VI) reductase has been structurally characterized using X-ray crystallography and X-ray spectroscopy, and the redox potential of the metal clusters of all 3 subunits of the Se(VI) reductase has been determined via EPR spectroscopy (Maher and Macy 2002; Maher et al. 2004;
Dridge et al. 2007). Another bacterium, Bacillus selenatarsenatis, has been shown to stoichiometrically couple reduction of Se(VI) to Se(IV) with oxidation of lactate during anaerobic respiration (Fujita et al. 1997). Genetic knock-outs have been used to identify the respiratory Se(VI) reductase of B. selenatarsenatis (Kuroda et al. 2011b). Curiously, the Se(VI) reductase catalytic subunit of B. selenatarsenatis shows significant sequence homology with tetrathionate reductases within the DMSOR family, making it only distantly related to the Se(VI) reductase catalytic subunit of T. selenatis (Grimaldi et al. 2013; Harel et al. 2016).
Se(VI) reductases have also been identified in Enterobacter cloacae SLD1a-1 and
Escherichia coli K-12. Previous work has shown that E. cloacae SLD1a-1 expresses a Se(VI) reductase under both aerobic and anaerobic conditions, though expression of the complex is greatest under suboxic conditions, and that the enzyme complex is distinct from the nitrate reductase complex of E. cloacae SLD1a-1 (Watts et al. 2003). The Se(VI) reductase was subsequently purified and shown to be a heterotrimer. Further, metals analysis of the Se(VI)
122 reductase complex found evidence for a molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD) cofactor and several iron-sulfur clusters, though the primary amino acid sequence of the reductase remains unknown (Ridley et al. 2006). The function of the
Se(VI) reductase in E. cloacae SLD1a-1 appears to be detoxification, rather than respiration, given that the complex is expressed under aerobic and anaerobic conditions. Additionally, attempts to cultivate E. cloacae SLD1a-1 anaerobically with Se(VI) as the sole terminal electron acceptor were unsuccessful (Losi and Frankenberger 1997; Ridley et al. 2006). Genetic knockouts in E. coli K-12 have also implicated a DSMOR enzyme as a Se(VI) reductase (Bébien et al. 2002). The catalytic subunit is encoded by the ygfN gene. As with E. cloacae SLD1a-1, this Se(VI) reductase appears to be involved in detoxification, rather than respiration, given that
E. coli K-12 reduces Se(VI) under both aerobic and anaerobic conditions. The evolutionary relationship between the YgfN protein, the Se(VI) reductase of E. cloacae SLD1a-1, and other
DMSOR memberss, is unknown. The YgfN protein has also been implicated in Se(VI) reduction in Citrobacter freundii strain RLS1 (Theisen and Yee 2014).
Far less is known about the molecular mechanisms of Se(IV) respiration, and it is uncertain whether any of the identified biochemical mechanisms of Se(IV) reduction in the literature are involved in Se(IV) respiration, or other physiological processes. Genetic knock- outs identified the fumarate reductase (encoded by the fccA gene) as a putative Se(IV) reductase in Shewanella oneidensis MR-1 (Li et al. 2014). Curiously, Se(IV) reductase activity in S. oneidensis was significantly reduced in fccA knockout mutants (~60%), but S. oneidensis mutant cells were still capable of Se(IV) reduction. This suggests either that the fumarate reductase of S. oneidensis is involved in mediating Se(IV) reduction, but is not actually the Se(IV) reductase, or that the Se(IV) reduction phenotype can be partially rescued by other unknown enzymes.
123 Regardless, the reduction of Se(IV) via the fumarate reductase does not appear to be involved in
Se(IV) respiration, given that the same mechanism has been implicated aerobically in
Enterobacter cloacae Z0206 (Song et al. 2017). An unknown periplasmic protein possessing
Se(IV) reductase activity has been purified in Shewanella frigidimarina strain ER-Te-48
(Maltman, Donald and Yurkov 2017). The enzyme has been proposed to be involved in Se(IV) respiration in this organism, but its low affinity for Se(IV) with an apparent Km of 12.1mM make this doubtful.
Current efforts to identify a respiratory Se(IV) reductase have been limited to organisms that have not been shown to be capable of using Se(IV) as a terminal electron acceptor in anaerobic respiration (Klonowska, Heulin and Vermeglio 2005; Maltman, Walter and Yurkov
2016). Thus far, only four prokaryotes have been shown to stoichiometrically couple oxidation of an electron donor to reduction of Se(IV) to Se(0) during a four electron transfer reaction:
Bacillus selenitireducens MLS10 (Switzer Blum et al. 1998), Pyrobaculum aerophilum IM2
(Huber et al. 2000), Desulfurispirillum indicum S5 (Rauschenbach, Narasingarao and Häggblom
2011), and Bacillus beveridgei MLTeJB (Baesman et al. 2009). Curiously, Se(IV) respiration in these organisms, in contrast to Se(VI) respiration and Se(IV) detoxification, results in the formation of black Se(0), with red Se(0) appearing as an intermediate that transitions to black
Se(0) as growth approaches the stationary phase. In MLS10, this black Se(0) has been shown to form as a result of respiratory growth on red Se(0) (Herbel et al. 2003), producing selenide. It is unknown if the black Se(0) in the other three taxa are also byproducts of Se(0) respiration.
As a model for a Se(IV) respiring organism, B. selenitireducens MLS10 is a well-suited candidate. MLS10 is a Gram-positive haloalkaliphilic Firmicute isolated from Mono Lake
(Switzer Blum et al. 1998). MLS10 is the first prokaryote shown to be capable of Se(IV)
124 respiration, grows readily on Se(IV) in large batch cultures, and the organism has previously been used to provide insights into the physiology and biochemistry of both selenium and arsenic respiration (Herbel et al. 2000, 2003; Afkar et al. 2003; Oremland et al. 2004; Baesman et al.
2007). We report here on the identification and initial characterization of a putative respiratory
Se(IV) reductase (Srr) from MLS10 through a polyphasic approach involving genomics, proteomics, and enzymology. This work provides our first insights into the biochemistry of
Se(IV) respiration from an organism definitively shown to utilize selenite as a terminal electron acceptor, and extends the catalytic versatility of a well-known family of enzymes within the larger DMSOR family.
Results
Identification of a putative Srr using in-gel enzyme assays- In order to identify candidates for the putative Srr, we performed in-gel enzyme assays using reduced methyl viologen as an artificial electron donor and Se(IV) as the electron acceptor. Periplasmic, particulate (e.g. membrane), and cytoplasmic protein fractions from MLS10 cells grown with Se(IV) were tested.
All three fractions showed activity at the same location in the gel, but the bulk of the activity was in the periplasmic fraction (as determined by protein quantification and gel staining). The Se(IV) reducing active band from the periplasmic fraction was excised and run both on SDS-PAGE and liquid chromatography tandem mass spectrometry (LC-MS/MS) to determine the protein composition. SDS-PAGE gels indicated that the Se(IV) reducing active band consisted primarily of an approximately 80kDa protein (Fig. 4.1). The LC-MS/MS results revealed that the active band eluted directly from the native gel contained 3 proteins identified with high confidence as a trimethylamine-N-oxide (TMAO) reductase (cytochrome c), a [4Fe-4S] ferredoxin iron-sulfur protein, and a rhodanese domain-containing protein corresponding to proteins encoded by the
125 genes Bsel_1476, Bsel_1477, and Bsel_1475 respectively, from the MLS10 annotated genome
(Table 4.1).
Figure 4.1: Identification of the respiratory Se(IV) reductase (Srr) in the periplasmic fraction of Bacillus selenitireducens strain MLS10. A) non-denaturing gel developed for
Se(IV) reducing activity with reduced methyl viologen. The clearing indicates Srr activity
(arrow). B) the lane after staining with Commassie Brilliant Blue showing a protein band at the same location as the Srr activity. C) SDS-PAGE gel of the excised band of Srr activity from (A) visualized by staining with Commassie Brilliant Blue.
126 Table 4.1. LC-MS/MS results for proteins eluted from non-denaturing and SDS-PAGE gels.
Sample Locus tag Protein Annotation Score Number of unique peptides
Se(IV) Bsel_1476 Trimethylamine-N-oxide 256.27 16 reducing active reductase band eluted Bsel_1477 [4Fe-4S] ferredoxin iron-sulfur 30.70 5 from non- binding domain protein denaturing gel Bsel_1475 Rhodanese domain protein 19.27 3 shown in Figure 4.1A.
80 kDa protein Bsel_1476 Trimethylamine-N-oxide 191.68 21 from SDS- reductase PAGE shown in Figure 4.1B.
80kDa band Bsel_1476 Trimethylamine-N-oxide 139.75 15 from the reductase second anion exchange column shown in Figure 4.2.
The kinetics of Srr- Efforts to measure the kinetics of Srr were complicated by several factors. We found that Se(IV) by itself could oxidize methyl viologen, benzyl viologen, flavin adenine dinucleotide (FAD), flavin mononucleotide (FMN), and NADPH in controls at pH 7.0 and below. Methyl viologen, however was stable at pH 7.5 and above. Thus we were able to test a pH range between 7.5 and 10. The enzyme, however showed no activity above pH 8.0. Further, any trace of Se(0) would also result in spurious methyl viologen oxidation, as was the case for the whole cell and cell lysate fractions. Hence, a purification table was not possible.
Nonetheless, we were able to obtain Se(IV) reducing active fractions from the periplasmic fraction after two sequential runs of anion exchange chromatography that were highly enriched in Srr (Fig. 4.2), as determined by LC MS/MS analysis of the 80kDa protein (Table 4.1). SrrA was found to have a high affinity for Se(IV) with an apparent Km of 142µM (Fig. 4.3). SrrA also
127 appeared to be specific to Se(IV), as no activity was observed using arsenate, selenate, or thiosulfate as electron acceptors in the assay. ICP-MS found evidence for both Mo and Fe, with a ratio of about 1 mol Mo to 10 mol Fe, suggesting to us that metal loss has likely occurred.
Figure 4.2: SDS-PAGE gels of MLS10 periplasmic fraction and Srr-active fractions from ion exchange columns. A) periplasmic fraction. B) Srr-active fraction eluted from the first anion exchange column loaded with the periplasmic fraction. C) Srr-active fraction obtained from a second anion exchange column loaded with the pooled Se(IV) reductase active fractions from the previous ion exchange column. The 80k Da band (arrow) corresponds to the putative SrrA in A and B
128
Figure 4.3: Kinetics of the MLS10 SrrA. The kinetics of the enzyme are expressed in terms of methyl viologen oxidized (M min-1).
129 Sequence analysis of Srr- The LC-MS/MS data was used to locate the putative srr operon in the genome of B. selenitireducens. Annotated as Bsel_1475 to Bsel_1480, it was found to contain six genes, srrE, srrA, srrB, srrC, srrD, and srrF, with SrrA the catalytic subunit (80.09 kDa), SrrB a small subunit (17.74 kDa), SrrC (43.14 kDa) an anchoring subunit, and SrrD (24.2 kDa) a chaperon protein (Fig. 4.4). Both SrrE (38.44 kDa) and SrrF (45.56 kdA) where annotated as rhodanese-domain containing proteins. Features of SrrA include a twin arginine translocation
(TAT) motif for export to the periplasm (confirming the original observation that Srr is periplasmic), a motif associated with a [4Fe-4S] cluster, and a molybdopterin guanine dinucleotide coordination motif coordinated by a cysteine residue (Figure S4.1). The annotation of SrrA as a TMAO reductase, however, is most likely incorrect for several reasons. Firstly, the
Mo/W-bisPGD cofactor of TMAO reductases is coordinated by a Ser residue, not a Cys residue.
Secondly, the TMAO reductase catalytic subunit lacks a [4Fe-4S] cluster. The primary sequence of the annotated [4Fe-4S] iron-sulfur binding domain protein (SrrB) suggests this protein contains four [4Fe-4S] clusters. SrrC is a homolog of the polysulfide reductase NrfD membrane anchoring subunit and has no apparent binding motifs for either a c or b type heme. SrrD. encoded by Bsel_1479, is most likely a homolog of the NarJ protein, which functions to facilitate incorporation of the Mo/W-bisPGD cofactor into the active site of the respiratory nitrate reductase of E. coli (Blasco et al. 1998). The PROSITE tool (de Castro et al. 2006) identified two structural motifs in the rhodanese domain proteins annotated by Bsel_1475 (SrrE) and
Bsel_1480 (SrrF). Both SrrE and SrrF contain a lipoprotein motif for attachment to the cytoplasmic membrane at the N terminus of the proteins, and two rhodanese domains, with the rhodanese domain closest to the N terminus containing an active site cysteine residue.
130
Figure 4.4: Schematic of the operon organization of the Wolinella succinogenes (WS) polysulfide reductase, Salmonella enterica serovar Typhimurium (STM) thiosulfate reductase, and putative Bacillus selenitireducens MLS10 (B_sel) selenite reductase operons. psrA, phsA, and srrA encode the catalytic subunits of the Psr/Phs/Srr reductase complex. psrB, phsB, and srrB encode the electron transfer subunits. psrC, phsC, and srrC encode the membrane anchor subunits. srrD encodes a protein with strong sequence homology to the NarJ protein involved in facilitating the incorporation of the Mo/W-bisPGD co-factor into the catalytic subunit NarG. srrE and srrF encode rhodanese domain containing proteins. The arrows refer to the direction of transcription.
131 Phylogenetic analysis of SrrA- In order to place SrrA more accurately within the
DMSOR family, Bayesian phylogenies using SrrA homologs were constructed, with the formate dehydrogenase N catalytic subunit of E. coli as an outgroup (Fig. 4.5). The ProtTest tool
(Darriba et al. 2011) and MrBayes (Ronquist et al. 2012) were used to determine both appropriate amino acid substitution models and rate variation models. The ProTest tool suggested that the LG amino acid substitution model (Le and Gascuel 2008) and the WAG amino acid substitution model (Whelan and Goldman 2001), with a gamma distributed (Yang
1994) or gamma distributed with a portion of invariant sites rate variation model, best described the observed differences in the amino acid sequences of SrrA homologs. MrBayes found that the
WAG amino acid substitution model best fit our data with a confidence of 1.0.
132
Figure 4.5: Bayesian phylogenetic analysis of protein homologs of the MLS10 SrrA. The phylogenetic analysis was done using MrBayes, using the LG amino acid substitution model with a gamma distribution. The numbers at each clade refer to the posterior probability support for that clade. The protein annotation after each species name is the annotation assigned by the
Integrated Microbial Genomes database. The tree scale bar refers to 0.1 amino acid substitutions per site.
133 All four models, LG with a gamma distribution, LG with a gamma distribution with a portion of invariant sties, WAG with a gamma distribution, and WAG with a gamma distribution with a portion of invariant sites, strongly supported the positioning of SrrA within the polysulfide/thiosulfate reductase (Psr/Phs) family of CISM molybdoenzymes. All four models yielded identical topologies with each node showing robust support (posterior probabilities greater than or equal to 0.9). The Tracer program (Rambaut et al. 2018) was used to assess convergence in the log-likelihood of MCMC runs, and to compare the likelihood of the four evolutionary models. The Tracer program suggested that the LG amino acid substitution model best fit our data, with either a gamma or a gamma with a proportion of invariant sites being equally likely (Fig. 4.5). Other tree topologies, as well as the parameters estimated from all evolutionary models tested can be found in the supplemental data.
The Psr/Phs family mediates polysulfide and thiosulfate respiration in bacteria, both crucial components of the sulfur biogeochemical cycle (53). The involvement of Psr in polysulfide respiration has been most extensively studied in Wolinella succinogenes (Schröder,
Kröger and Macy 1988; Krafft et al. 1992; Dietrich and Klimmek 2002), whereas Salmonella enterica serovar Typhimurium has been used to elucidate the involvement of Phs in thiosulfate respiration (Heinzinger et al. 1995; Stoffels et al. 2012). The Psr operon of W. succinogenes encodes a catalytic subunit (PsrA), a four [4Fe-4S] electron transfer subunit (PsrB), and a membrane anchor subunit that lacks heme groups (PsrC). The Phs operon of S. enterica serovar
Typhimurium has an identical organization, with the exception that the membrane anchor subunit
(PhsC) harbors two heme b groups. PsrA/PhsA catalytic subunits and the PsrB/PhsB electron transfer subunits are closely related, while PsrC is closely related to the membrane anchor of
134 nitrite reductases and PhsC is closely related to the membrane anchor of formate dehydrogenases
(Rothery, Workun and Weiner 2008).
Examining the genomic context of SrrA homologs, three operon types were apparent.
Several SrrA homologs were found in putative Psr operons (Fig. 4.4). Other SrrA homologs were found in putative Phs operons (Fig. 4.5). The third operon organization contained a catalytic subunit, a four [4Fe-4S] electron transfer subunit, a membrane anchor subunit that lacks heme groups, and one or two additional rhodanese domain proteins. We refer to this operon organization as a putative Srr operon. These operon organizations were mapped onto the phylogenies. Curiously, Psr, Phs, and Srr operons did not form monophyletic clades, suggesting that the differences in the physiological functions of these enzymes do not stem from evolutionarily conserved differences in the primary sequences of the putative Psr/Phs/SrrA homologs.
Discussion
A polysulfide/thiosulfate reductase homolog as the MLS10 putative selenite reductase-
Our work substantially advances our knowledge of the biology of selenium by providing the first report of a putative respiratory Se(IV) reductase in a bona fide Se(IV) respiring organism. It is especially significant that we have strong evidence that SrrA is a member of the Psr/Phs clade of the DMSOR family. The Psr/Phs clade of the this family has representatives from both bacteria and archaea and is deeply branched in phylogenies of CISM family enzymes, suggesting that this clade is ancient (Rothery, Workun and Weiner 2008). These enzymes mediate crucial reactions of transient sulfur species in the sulfur biogeochemical cycle. While little is known about the environmental distribution and availability of polysulfide and thiosulfate, a number of reports suggest that polysulfide respiration is an important component of the sulfur biogeochemical
135 cycle in a number of environments, including soda lakes and oceans (Wright et al. 2014; Findlay
2016b; Vavourakis et al. 2016). Our identification of SrrA as a member of the Psr/Phs clade thus not only expands our knowledge of the catalytic diversity of this clade, but also implicates these enzymes in a crucial component of the selenium biogeochemical cycle.
Phylogenetic evidence suggests that the DMSOR oxidoreductases for several toxic oxyanions, including the T. selenatis selenate reductase (Ser), Ideonella dechloratans chlorate reductase (Clr), and the anaerobic arsenite oxidase (Arx) of Alkalilimnicola ehrlichii diverged from more deeply branched clades within the family, namely the respiratory nitrate reductase
(NarG) and arsenate reductase (ArrA) clades (Stolz et al. 2006; Zargar et al. 2012). This current work finds no evidence that SrrA, nor any other putative Srr operons, form a monophyletic clade within the Psr/Phs family, nor is there any compelling evidence that the MLS10 SrrA and closely related SrrA homologs constitute a discrete clade. This is very surprising, given that the Psr/Phs catalytic subunit clade must mediate electron transfer using substrates with remarkably different redox potentials. Both polysulfide and thiosulfate are highly reduced substrates, with redox potentials of -260mV and -402mV, respectively (Thauer, Jungermann and Decker 1977;
Hedderich et al. 1998). Se(IV), however, has a redox potential of 216mV (Nancharaiah and
Lens 2015). It might seem logical to propose that the rhodanese domain-containing proteins in the MLS10 Srr operon directly confers the ability to reduce Se(IV) to SrrA, but this hypothesis is unlikely given that active fractions were found without the detection of SrrE or SrrF (Table 4.1).
This suggests that the ability of the MLS10 SrrA to reduce Se(IV) stems from features unique to the catalytic or binding sites.
Kinetics of SrrA- The affinity of SrrA for Se(IV) is much higher than that reported for other selenium oxidoreductases. As mentioned, previously reported Km values have ranged from
136 high micromolar to millimolar (44). A curious shared feature of this enzyme family is that the electron transfer subunit does not appear to be necessary for enzyme activity. This contrasts starkly with previous work on DMSOR enzymes, where the electron transport subunit is essential for activity (Schröder et al. 1997; Krafft and Macy 1998; Afkar et al. 2003). The observed pH of the enzyme was also unexpected. As a haloalkaliphile, MLS10 grows at a pH of
9.8, and the previously published purification of the MLS10 respiratory arsenate reductase found that the enzyme had an optimal pH of 9.5 (Afkar et al. 2003). We failed to observe any enzyme activity above pH 8.0, despite the fact that Srr from MLS10, like its Arr, is oriented towards the periplasm. More puzzling still was the apparent specificity of SrrA for Se(IV). Mono Lake, where MLS10 was isolated, is particularly rich in arsenic (Oremland et al. 2000), with lake water containing 200µM arsenic. This is decidedly not the case for selenium, as previous work estimates the concentration of selenium in Mono Lake water at 38nM (Baesman et al. 2009).
Further, the arsenate reductase (Arr) from MLS10 can use Se(VI) and Se(IV) to oxidize reduced methyl viologen in vitro (52). It seems that MLS10 cells express SrrA specifically for Se(IV) respiration, despite the fact that Se(IV) is not readily available as an energy source in its environment. Regardless, both Se(IV) and Se(VI) are a rich source of energy compared to many other terminal electron acceptors (Newman, Ahmann and Morel 1998), and this may offer an important selective advantage to prokaryotes that can respire selenium oxyanions, even in environments not particularly rich in selenium.
Involvement of rhodanese domain-containing proteins in selenite respiration- Neither
SrrE nor SrrF appear to be essential for Se(IV) respiration in vitro, yet our non-denaturing in-gel enzyme assays suggest that SrrE at least functions in vivo with SrrA and SrrB to reduce Se(IV) to
Se(0). The use of rhodanese domain-containing proteins during anaerobic respiration has
137 previously been documented in W. succinogenes during polysulfide respiration (Klimmek et al.
1998). The Sud protein, while not part of the Psr operon, is a single active site rhodanese domain containing protein that increases the affinity of PsrA in W. succinogenes by nearly an order of magnitude when cells of W. succinogenes are grown on polysulfide. Curiously, there appears to be no relation between the Sud protein and SrrE and SrrF, as both Srr rhodanese proteins contain two active sites, rather than one, and BLAST searches for the Sud protein in the
MLS10 genome did not elicit any Sud homologs in MLS10. Nonetheless, it is possible that SrrE and SrrF perform an analogous role in MLS10 in increasing the affinity of Srr for Se(IV). It should be noted that a putative Srr operon has previously been reported in the literature in
Desulfitobacterium hafniense TCE1 (Prat et al. 2012). The authors found that the protein homologous to the MLS10 SrrF was likely involved in sulfide detoxification in TCE1.
Curiously, the authors could not stimulate expression of any other components of the putative Srr operon on any growth substrates tested. While the authors did not test for expression of this operon when growing TCE1 on Se(IV), previous attempts to cultivate D. hafniense strains on
Se(IV) have not been successful (Niggemyer et al. 2001).
Implications for future research- This report is part of a growing body of literature that emphasizes the extraordinary catalytic diversity of DMSOR enzymes (Grimaldi et al. 2013).
This report extends these findings by offering evidence that a well-known family of molybdoenzymes can mediate electron transfer on a wide variety of chalcogen substrates with different redox potentials. The work also demonstrates that research into selenium respiration benefits substantially from considering the physiology of selenium reducing organisms, and whether that reduction of selenium is being coupled to energy conservation. Definitive proof that
SrrA is the respiratory Se(IV) reductase in MLS10 will require gene knockouts. Additionally,
138 while few Se(IV) respiring prokaryotes have been documented in the literature, the phylogenetic breadth of Se(IV) respiring prokaryotes is substantial. The dearth of known Se(IV) respiring bacteria thus likely stems from a failure to include Se(IV) as a possible electron acceptor when characterizing new isolates. For instance, new reports have identified a bacterium and an archaeon that reduce Se(IV) to black Se(0) when grown anaerobically in the presence of Se(IV)
(Slobodkina et al. 2015; Abin and Hollibaugh 2017), suggesting more Se(IV) respiring prokaryotes remain to be discovered. It remains to be seen if Srr mediated Se(IV) respiration is conserved in other Se(IV) respiring organisms, or if multiple pathways for Se(IV) respiration have evolved in prokaryotes.
139 Supplemental Materials
Figure S1
A) Bsel_1475 Rhodanese domain protein MKHKWTGLLMGAALLTATACSSDEASVLADEENDNEEQAAVETNVFNDLMANSAAY MEDPNFNVITGEALHNKTVFENPEDYFILDVRDTTTFVSGHIPGAVNVPYRVSGLEQMY LELPEDKTIYVVCFSGHTASHTVGMLNALGYDAEALQFGMGGYASGTDLGSNIPGGPA ELPVVTTGFELTETHDLPEIMSDESDIRTIAHEQSQNFLDQEPPGVMGAPDLNEMINEGN LDGHQLIDIRLGEHYEQGHIEGAANLPYNELFNEENLSLLDPEKMTVVIGYNGYDASQV TRLLGQLEYSAVPLAYGMSIWSGDESVVGEHMFDFSNVYSLPVRELKFDMDAGDIEAG CR
B) Bsel_1476 Trimethylamine-N-oxide reductase MPKLKRRSFLKASALTAAAAAVPFKMTMGDFREASAEAEEKVIPSTCNGCASMCGIYA HVKNDRLWYVEGHPVHLKAGGRLCARGHGMAADIYGKGRVQGPMKKVAEGEFEPIS WEQAFKEIGEKMGNLRDQYGGNSFLWLEHGVRGKRYADPLLDRMGSSNYITHYSTCFT SKTNAWQHMVGSMPAGDHENAKYMIFEGRNFAGAIIPNGMKKILKAKDNGAKIVVIDP RYSEIAKVADEWIPIRPGTDLALRLGMAHTLISENLYDSAFVKKYVTDFDEFWSLNKDK DADWAAEITGIDADTIRRVAREFAEHAPEAFMEPGWHGLHCHYFNSTQTAQMGIILNAL VGNFFKRGGLMPSANVEFGEYMHTDVEAVEKGPRADGAGVEGEHMTVEPGRGIAQNV PDMIDKGRIKSVFIYHFNPLRTAPDPEYQKKIANAELVVSIPVDWNETSVYAADYILPEN YYLERTEVPQAVSGHISHDWPQISIRQQVTDPLHDTLPLLDIMRGITKEMGYDNLYDFTV DDEIAAMLEPTGVTPEELKEKGTVELRTNKVEPKFPVNLSGEPNLGTFSGKIQFSAEIFKI DGKRGVPTWIPTMVQPDLNNPEEFRLIHGKQPYHSHSVTSTNASLLRITEKYNGEAMWI NTKRAKDLGIEDGDTVSVKSSIANKTVPVHVTQLIHPECVWIPSAYGAFSNKIKEGYQLG INFNDFIPMMIEPYSGSTMSQEVVVNVQKGGEA
C) Bsel_1477 [4Fe-4S] ferredoxin iron-sulfur binding domain protein MANYGLLIDTRKCTGCHACSIACSTYNELEPEVSYNRLEFIESGTYPNVKMDIVPVQCM HCDDAPCVKVCSTQATFKVEENGIVAFNPDKCTGCKACMAACPYDARALNEARLSEKC RWCPEMLTQGKQPSCASTCMNEVRLFGDLDDPDSEINKKMAQFNTVQLVPEAGTKPRI WYIK
D) Bsel_1478 Polysulfide reductase NrfD MLNRNLWYGITGIMIIFGIIGIGNIFIHGEHVMGTSNYVPWGSLIGAYVFFVAISTGLTFLS SMVHVFKMKQFEFLTKRLTLASIATLLMGFVMIGVELGNPLAMVYILVTPNVMAPIFW MGAFYGLYLVLHIIEFFFQIKDNHKIVNTISPFVLVVGIAAQSTLGAVFGLSVARGIWNSA YLSIFFLIMAFVSGLAVAMIMAFFLSKGNVMKQEDREKLPGLYPFMSKLTMGLLAVGIIF VTWNWIYGLHSGNPNRMASMELMLNGPLAVPYWVLEVGFVFLIPLLILGLVKAKKAAT MLTTGVVLLIGLFAMRIILTFAGQMVPLEVVTGSLTMNELRDVSILWSEWATMIFGVGG SILIYMLGERFLNLDVTEHGSHGTHDKKATQAS
E) Bsel_1479 Chaperon protein NarJ MNKTDRKETYTELIRLLGELYKFPDDEVEQAIAEGVLDQEIDDYLSLFPEIRIENKQSFSS
140 LRDKELTAKKQYMTAYSGITTPFHPPVESLYKPWSVDPKDQTGLYNKTGYYMGDAAL HMKHLLQHYEIEVPKDYEMMPDHLAISLEFYAMLLERDTEAAEAFHQDHFDWLKAFH KKHQEIDDVPFYDYLLRVLIAVTAVSPRELT
F) Bsel_1480 Rhodanese domain protein MKKWAIGLLSLAGMITLSACGTSQASETEPTSFLPMEPDVSPYIVEEEAEEEGQDHVPYH ENWNYIDTVQLTRLMDGLPEISQDRESYDQVPPEWGDVALIDSRPPGVYAAGHINGAINI PDSEFDDYKHLLPEDKDTQLIFYCGGLHCALSGNSSEKAMDMGYENSYVYQEGTPAWK SAGNYFTVTPEYVEEQILESNVARDDTDPVMIIDTRTYAGYFAEHIPTAVFWDDTQYGT KYQGFAPENKDAEIIIYCGGFFCHKSPALADDLLGDGYTNVKVLSGGMPAWKQAGLPT FGMETADADFDVSAGKVDRSVSAEDFEDLIASGATVVDVRGDGEVANGMIDGALH VPDGDIHANDPSVEETLPDDKSTTLLIHCASGARASGVVEKIADLGYEEVYYWNGGISIS DDGSYTLN
Figure S4.1. Structural motifs predicted in proteins of the MLS10 Srr operon srrEABCDF
(Bsel 1475 – Bsel_1480).
A) SrrE, Rhodanese domain protein. The putative lipoprotein motif anchoring the protein to the membrane is highlighted in grey. The putative rhodanese domains are highlighted in purple.
One rhodanese domain has a predicted catalytically active cysteine residue, highlighted in teal.
B) SrrA, Trimethylamine-N-oxide reductase. The putative twin arginine translocation motif is highlighted in teal. The putative [4Fe-4S] cluster is highlighted in yellow. The cysteine residue predicted to coordinate the Mo/W-bisPGD cofactor is highlighted in red.
C) SrrB, [4Fe-4S] ferredoxin iron-sulfur binding domain protein. The four predicted [4Fe-4S] iron clusters are highlighted. The first [4Fe-4S] cluster is highlighted in red. The second [4Fe-
4S] cluster is highlighted in purple. The third [4Fe-4S] cluster is highlighted in teal. The fourth
[4Fe-4S] cluster is highlighted in yellow.
D) SrrC, Polysulfide reductase NrfD.
E) SrrD, Chaperon protein NarJ.
141 F). SrrF, Rhodanese domain protein. The putative lipoprotein motif anchoring the protein to the membrane is highlighted in grey. The putative rhodanese domains are highlighted in purple. All three rhodanese domains have predicted catalytically active cysteine residues, highlighted in teal.
142 CHAPTER 5
SELENITE METABOLISM IN THERMUS SCOTODUCTUS SA-01,
DESULFITOBACTERIUM HAFNIENSE PCP-1, AND BACILLUS BEVERIDGEI MLTeJB
Introduction
Selenium metabolism is widespread in the bacterial domain. Approximately 1/5th of all bacterial genomes, including representatives from nearly every bacterial phylum, contain genes for the synthesis of the 21st amino acid selenocysteine (Sec) and the tRNA nucleoside 5- methylaminomethyl-2-selenouridine (mnm5Se2U) (Romero et al. 2005; Zhang et al. 2006; Zhang and Gladyshev 2010; Lin et al. 2015; Mariotti et al. 2015; Peng et al. 2016). In contrast, genes for Sec and mnm5Se2U synthesis in archaeal genomes were thought to be confined to the
Methanococcales and Methanopyrales orders of the Euryarchaeota phylum (Zhang and
Gladyshev 2010; Su et al. 2012; Lin et al. 2015; Mariotti et al. 2015). The recent discovery of
Sec synthesis genes in members of the Lokiarchaeota (Mariotti et al. 2016) and Thorarchaeota
(Liu et al. 2018) lineages of the Asgard archaea indicate that selenium utilization in this domain may be more widespread than is currently appreciated. The distribution of selenium utilization genes in bacteria and archaea indicates that selenium assimilation could have been part of the metabolic repertoire of the last universal common ancestor (LUCA) of all three domains of life.
Phylogenetic analyses of the selenophosphate synthetase, which is indispensable for both Sec and mnm5Se2U synthesis, confirms that selenium assimilation most likely an adaptation present in LUCA (Mariotti et al. 2015; Weiss et al. 2016).
The selenium oxyanions selenate (Se(VI)) and selenite (Se(IV)) can also serve as terminal electron acceptors in anaerobic respiration. Se(VI) respiring organisms have been isolated from the Betaproteobacteria (Macy et al. 1993), Gammaproteobacteria (Nakagawa et
143 al. 2006; Narasingarao and Häggblom 2006), Deltaproteobacteria (Narasingarao and Häggblom
2007b), Epsilonproteobacteria (Stolz et al. 1999), Actinobacteria (von Wintzingerode et al.
2001), Chrysiogenetes (Rauschenbach, Narasingarao and Häggblom 2011), Deferribacteres
(Rauschenbach et al. 2013), and the Firmicutes (Fujita et al. 1997; Switzer Blum et al. 1998;
Blum et al. 2001; Niggemyer et al. 2001; Baesman et al. 2009; Abin and Hollibaugh 2017) phyla in bacteria. Se(IV) respiration has been observed in bacteria from the Chyrsiogenetes
(Rauschenbach, Narasingarao and Häggblom 2011) and Firmicutes (Switzer Blum et al. 1998;
Baesman et al. 2009; Abin and Hollibaugh 2017) phyla. Se(VI) and Se(IV) respiration in archaea appears confined to hyperthermophiles from the Pyrobaculum genus in the
Crenarchaeota phylum (Huber et al. 2000; Slobodkina et al. 2015). The phylogenetic and ecological diversity of Se(VI) and Se(IV) respiring prokaryotes demonstrates that these reductive transformations are essential and ubiquitous components of the global selenium biogeochemical cycle.
Respiratory Se(VI) reductases from the betaproteobacterium Thauera selenatis (Schröder et al. 1997; Krafft et al. 2000) and from the firmicute Bacillus selenatarsenatis (Kuroda et al.
2011b) have been identified. Both enzymes consist of multi subunit complexes with catalytic subunits from the mononuclear molybdoenzyme dimethyl sulfoxide reductase (DMSOR) family.
Phylogenetic analysis of DMSOR members indicates that these two Se(VI) reductases are members of distantly related lineages within this family (Kuroda et al. 2011b). Se(VI) respiration therefore represents a relatively recent biogeochemical adaptation, with different lineages of Se(VI) respiring bacteria independently evolving this trait. It is currently unknown when Se(IV) respiration evolved. Addressing this question is particularly relevant for the evolution of the selenium biogeochemical cycle. Se(IV) is the principle source of inorganic
144 selenium for assimilatory metabolism in organisms from all three domains of life (Tobe and
Mihara 2018). Se(IV) may have been an essential reservoir of bioavailable selenium for the emergence of selenium assimilatory metabolism early in the Archean Eon (~4.0 – 2.5 billion years ago (Gya)), as the other more reduced inorganic species of selenium, elemental selenium and selenide, are presumed to have limited bioavailability (Masscheleyn, Delaune and Patrick
1990, 1991). This has been the subject of debate in the geochemical literature, as some researchers contend that patterns of selenium isotope fractionation observed in the sedimentary rock record precludes the availability of Se(IV) for most of this geological eon (Stüeken et al.
2015a; Stüeken 2017). Other researchers argue that widespread and efficient reduction of Se(IV) for assimilation into macromolecules would mute selenium isotopic signatures indicative of active Se(IV) reduction (Mitchell et al. 2016).
Evolutionary analysis of respiratory Se(IV) reductases would be an attractive candidate for independently testing these competing hypotheses. Respiratory reduction of Se(IV) is associated with the generation of significant free energy, offering superior energetic yields compared to many widely utilized terminal electron acceptors such as sulfate and arsenate
(Newman, Ahmann and Morel 1998). The presence of Se(IV) would therefore exert strong selective pressure on organisms to exploit this energy source, making the enzyme(s) or catalytic subunit(s) mediating Se(IV) respiration useful proxies for the emergence of Se(IV) in the geological record. The recent identification of the first respiratory Se(IV) reductase (Srr) from
Bacillus selenitireducens MLS10 (Wells et al. 2019) has made such an analysis possible. Srr is a multi-subunit complex consisting of a catalytic subunit with a molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD) cofactor (SrrA), a four [4Fe-4S] electron transfer subunit (SrrB), a membrane anchor (without heme groups) (SrrC), a NarJ-like
145 chaperone subunit (SrrD), and two rhodanese domain-containing proteins each composed of two rhodanese domains with single active site Cys residues (SrrE and SrrF). The catalytic subunit is a DMSOR member containing both the characteristic Mo/W-bisPGD and a [4Fe-4S] cofactor and specifically is a member of the polysulfide reductase catalytic subunit (PsrA) and thiosulfate reductase catalytic subunit (PhsA) lineage within the family. This lineage functions physiologically to allow bacteria and archaea to exploit these low redox potential sulfur intermediates as terminal electron acceptors in anaerobic respiration (Grimaldi et al. 2013; Hille,
Hall and Basu 2014). The electron transfer, membrane anchor, and the NarJ-like chaperone that apparently mediates the insertion of the Mo/W-bisPGD cofactor (Blasco et al. 1998) subunits that are associated with SrrA are typical partner subunits for DMSOR members involved in anaerobic respiration (Rothery, Workun and Weiner 2008).
The two distinctive rhodanese domain-containing proteins have never before been reported in association with DMSOR members and appear to distinguish srr operons from psr and phs operons. In our phylogenetic analysis, we observed that putative srr operons did not form a single monophyletic clade, but were distributed through several deeply branching bacterial lineages (Wells et al. 2019). If Se(IV) respiration is truly an ancient trait, predating the diversification of bacteria into separate phyla, then this bioenergetic adaptation should have evolved a single time. This would make the srr operon a reliable signature for the ability of an organism respire Se(IV). If Se(IV) respiration is a relatively recent innovation, emerging after bacteria began diversifying into separate phyla, then multiple mechanisms of Se(IV) respiration should exist, as has been observed for respiratory Se(VI) reduction. In this report, we determined whether two organisms, Thermus scotoductus SA-01 and Desulfitobacterium hafniense PCP-1, not known to respire Se(IV) but whose genomes contain putative Srr
146 homologs, could utilize Se(IV) as a terminal electron acceptor, and specifically expressed Srr when grown in the presence of Se(IV). Similarly, our previous report noted that all known
Se(IV) respiring bacteria contain homologs for the Srr operon. Thus, we also determined whether Bacillus beveridgei MLTeJB, which has previously been shown to be capable of Se(IV) respiration (Baesman et al. 2009), expresses Srr as the Se(IV) reductase when grown in the presence of Se(IV). We use a combination of in-gel enzyme assays to detect enzymes with
Se(IV) reducing activity in crude fractions and liquid chromatography tandem mass spectrometry
(LC MS/MS) to identify these putative Se(IV) reductases.
Our work shows that neither T. scotoductus SA-01 nor D. hafniense PCP-1 can respire
Se(IV), though both can reduce the oxyanion to elemental selenium. In the case of SA-01, we definitively establish that the putative srr operon, which only contains a single rhodanese domain-containing protein with one rhodanese domain and a single active site Cys residue, is not associated with Se(IV) reduction in vitro. Rather, Se(IV) reduction seems to be mediated by components of the sulfur oxidation (Sox) pathway. Low levels of protein expression preclude definitive identification of a Se(IV) reductase in Se(VI)-grown D. hafniense PCP-1 cells. In contrast, we determine that both Se(VI)- and Se(IV)-grown cells of B. beveridgei MLTeJB express Srr, and that Srr functions in vitro as the Se(IV) reductase. Both D. hafniense PCP-1 and
B. beveridgei MLTeJB contain putative srr operons that are exact homologs of the MLS10 Srr operon, suggesting that the two rhodanese domain-containing proteins with two rhodanese domains may be essential for the ability of Srr to reduce Se(IV).
147 Results
Selenite reduction in Thermus scotoductus SA-01
We identified a putative srr operon in the genome of T. scotoductus SA-01 using the
DELTA-BLAST search tool (Boratyn et al. 2012) with the B. selenitireducens SrrA as query.
The operon contained (from the start site of transcription), a gene encoding for a putative NarJ- like chaperone for the insertion of the Mo/W-bisPGD cofactor, a gene encoding for the [4Fe-4S] and Mo/W-bisPGD-containing catalytic subunit, a gene encoding for a four [4Fe-4S] cluster protein, a gene encoding for a membrane anchor subunit with no heme groups, and a gene encoding for a protein with a single rhodanese domain featuring an active site Cys residue for transferase activity. These genes are encoded for by the TSC_c06960, TSC_c06950.
TSC_c06940, and TSC_c06930 loci respectively. This putative srr operon differs from the B. selenitireducens operon in two respects. First, the operon has a single gene for a rhodanese domain-containing protein with a single active rhodanese domain, rather than two genes encoding two rhodanese domain-containing proteins each with two active rhodanese domains.
Secondly, the operon organization of T. scotoductus (srrDABCE) differs from the operon organization of B. selenitireducens (srrEABCDF) (Wells et al. 2019). The relationship between rhodanese domain-containing proteins and Se(IV) respiration is unclear. It is possible that the presence of rhodanese domain-containing proteins alone is sufficient to differentiate an srr operon from a psr or phs operon. It is also possible that rhodanese domain-containing proteins with two rhodanese domains are the relevant determinant for srr operons. Thus, we were interested in assessing the ability of T. scotoductus SA-01 to exploit Se(IV) as a terminal electron acceptor.
148 We initially cultured T. scotoductus SA-01 in the Castenholtz TYE complex medium recommended by the American Type Culture Collection with 10 mM lactate as the electron donor and 10 mM Se(IV) as the terminal electron acceptor. We were able to grow SA-01 repeatedly at 65C in large batch cultures. We used non-denaturing gels to electrophoretically separate periplasmic, cytoplasmic, and insoluble crude fractions. In-gel enzyme assays for
Se(IV) reducing activity determined that the periplasmic fraction was the only fraction with
Se(IV) reducing activity, and this activity was confined to a single band of activity (Fig. 5.1).
LC MS/MS analysis of this band found no Srr components but identified a number of proteins in the sulfur oxidation (Sox) pathway (Ghosh and Dam 2009). The analysis determined SoxA,
SoxC, and SoxX were present in this Se(IV) reducing band with high confidence (Table 5.1).
During chemolithoautotrophic growth on the sulfur intermediate thiosulfate, the diheme SoxA and the monoheme SoxX function together as a triheme cytochrome c complex that simultaneously oxidizes the sulfane sulfur (sulfide) of thiosulfate in a two electron transfer reaction that links the molecule to the SoxYZ complex, entering it into the Sox pathway for the ultimate oxidation of thiosulfate to produce two sulfate molecules (Ghosh and Dam 2009). SoxC is a molybdopterin-containing catalytic subunit similar to sulfite dehydrogenases that functions with the diheme SoxD protein to fully oxidize the sulfane sulfur to sulfate, releasing the sulfur atom from SoxYZ and regenerating SoxYZ (Ghosh and Dam 2009; Zander et al. 2011).
149 Figure 5.1: Identification of a Se(IV) reducing band in Thermus scotoductus SA-01 periplasmic fractions electrophoretically separated on a non-denaturing gel. The lane was stained with
Coomassie brilliant blue. The band with Se(IV) reducing activity is highlighted by an arrow.
150 Locus tag/accession Protein annotation Score Number of number unique peptides TSC_c22660, L-glutamate gamma- 88.34 16 WP_015718128.1 semialdehyde dehydrogenase TSC_c04450, Aldo/keto reductase 81.05 11 WP_148228680.1 TSC_c10230, Acetyl ornithine aminotransferase 66.19 10 WP_015716922.1 family protein TSC_c21050, Sulfur oxidation c-type 52.84 5 WP_015717973.1 cytochrome SoxA TSC_c21060, Sulfur oxidation c-type 47.07 4 WP_038031180.1 cytochrome SoxX TSC_c21070, SoxB (pseudogene) 26.68 7 NC_014974.1 TSC_c21010, SoxC sulfite dehydrogenase 25.54 7 WP_015717969.1 TSC_c22050, Glu/Leu/Phe/Val dehydrogenase 23.33 8 WP_015718072.1 TSC_c01740, Elongation factor Tu 19.41 4 WP_015716104.1 TSC_c20650, Isocitrate/isopropylmalate 14.60 7 WP_015716104.1 dehydrogenase family protein TSC_c24060, Phosphate/phosphite/phosphonate 13.84 4 WP_028493746.1 ABC transporter substrate- binding protein Table 5.1: LC MS/MS results for proteins eluted from a Se(IV) reducing band from Thermus scotoductus SA-01 periplasmic crude fraction. Sulfur oxidation (Sox) pathway proteins are highlighted in bold.
151 These results suggested that the Sox pathway could function in SA-01 to allow the organism to exploit Se(IV) as a terminal electron acceptor in anaerobic respiration. Thus, an attempt was made to cultivate T. scotoductus SA-01 on a defined medium with 10 mM lactate and 10 mM Se(IV) so that the stoichiometric oxidation of lactate coupled to Se(IV) reduction to elemental selenium could be established. However, SA-01 could not grow in the medium with
Se(IV) through three successive transfers. Thus, the Se(IV) reduction we observed in the complex medium was clearly not related to Se(IV) respiration. It should be noted that the
Castenholtz TYE medium contains 5 mM nitrate. This suggests that T. scotoductus SA-01 was likely reducing Se(IV) as a detoxification mechanism and utilizing nitrate as the terminal electron acceptor. Consistent with this, UV-Vis spectroscopic characterization of oxidized membrane fractions of SA-01 displayed evidence of an enzyme containing a blue copper cofactor with a peak observed at 600 nm in spectra (data not shown). This 600 nm peak is similar to a peak observed in UV-Vis spectra of the blue copper-containing nitrite reductase
NirK of a halophilic archaeon (Ichiki et al. 2001).
Selenite reduction in Desulfitobacterium hafniense PCP-1
We identified a putative srr operon in the genome of D. hafniense PCP-1 as we described above for T. scotoductus SA-01. The operon was identical to the B. selenitireducens MLS10 srr operon, with the genes arranged as srrEABCDF in both genomes. These genes are encoded by the A37YDRAFT_00019 (srrE) , A37YDRAFT_00020 (srrA) , A37YDRAFT_00021 (srrB),
A37YDRAFT_00022 (srrC), A37YDRAFT_00023 (srrD), and A37YDRAFT_00024 (srrF) loci on the Integrated Microbial Genomics (IMG) platform (Markowitz et al. 2006). Previous attempts to cultivate D. hafniense PCP-1 on Se(IV) were unsuccessful, though the organism was found to be capable of Se(VI) respiration, reducing Se(VI) to Se(IV) in a two electron transfer
152 reaction (Niggemyer et al. 2001). Similarly, we were unable to cultivate PCP-1 on medium with
10mM lactate and 10mM Se(IV), but were able to repeatedly cultivate PCP-1 in large batch cultures with 10mM lactate and 10mM Se(VI). As PCP-1 grew on Se(VI), Se(IV) was concomitantly reduced to red elemental selenium. This suggests that Srr could still mediate
Se(IV) reduction in this organism, even if that reduction was not linked to energy conservation.
We were unable to obtain periplasmic crude fractions from D. hafniense PCP-1, so in-gel enzyme assays for Se(IV) reducing activity were confined to soluble and insoluble crude fractions. We found two bands with Se(IV) reducing activity in the soluble fraction. The soluble fraction contained too many proteins for distinct protein bands to be observed in destained gels, so we utilized ammonium sulfate precipitation to obtain enriched fractions of SrrA. A 35% ammonium sulfate fraction was found to contain substantial Se(IV) reducing activity, and this fraction was used for additional in-gel enzyme assays. We found that the 35% ammonium sulfate enriched fraction also had the same two bands of Se(IV) reducing activity (Fig. 5.2).
Unfortunately, the bands with Se(IV) reducing activity were extremely faint, making determintation of the protein composition of the Se(IV) reducing bands infeasible by LC
MS/MS. SDS-PAGE gel cuts of a number of these bands indicated that they contained many other additional proteins besides a putative Se(IV) reductase. Thus, we are unable to ascertain if
Srr is expressed by D. hafniense PCP-1 when cells are grown on Se(VI).
153 Figure 5.2: Identification of Se(IV) reducing bands in soluble fractions of Desulfitobacterium hafniense PCP-1 with 35% ammonium sulfate. Lane A, nondenaturing gel developed for Se(IV) reducing activity with reduced methyl viologen. The clearing indicates Srr activity. Lane B, soluble fraction before ammonium sulfate precipitation protocol after staining with Coomassie
Brilliant Blue. Lane C, lane A, after staining with Coomassie Brilliant Blue. Lane D is an SDS- polyacrylamide gel of the excised upper band with Se(IV) reducing activity from lane A after staining with Coomassie Brilliant Blue. Lane E is an SDS-polyacrylamide gel of the excised lower band with Se(IV) reducing activity from lane A after staining with Coomassie Brilliant
Blue.
154 Selenite reduction in Bacillus beveridgei MLTeJB
The haloalkaliphilic bacterium B. beveridgei MLTeJB is closely related to B. selenitireducens MLS10 and has previously been shown to be capable of both Se(VI) and Se(IV) respiration (Baesman et al. 2009). Consistent with the hypothesis that srr is a reliable marker for the ability to respire Se(IV), we found an identical srrEABCDF operon in MLTeJB as the
MLS10 operon. These genes are encoded for in the B. beveridgei MLTeJB genome by the
Ga0309649_112450 (srrE), Ga0309649_112449 (srrA), Ga0309649_112448 (srrB),
Ga0309649_112447 (srrC), Ga0309649_112446 (srrD), and Ga0309649_112445 (srrF) loci on the IMG platform. We were able to repeatedly culture B. beveridgei MLTeJB in large batch cultures containing 10 mM lactate and 10 mM Se(VI) and 10 mM lactate and 10 mM Se(IV). In- gel enzyme assays of Se(VI)-grown cells found that Se(IV) reducing activity was confined to periplasmic crude fractions, and SDS-PAGE gel cuts of the Se(IV) reducing band confirmed that the Se(IV) reductase appeared to consist of a single 80 kDa protein (Fig. 5.3). Similarly, in-gel enzyme assays using periplasmic, cytoplasmic, and insoluble fractions from Se(IV)-grown B. beveridgei MLTeJB cells also found a single band of Se(IV) reducing activity (Fig. 5.4). SDS-
PAGE gel cuts found that the Se(IV) reducing band was apparently composed of a single 80 kDa protein. LC MS/MS analysis of the 80 kDa band seen in Se(IV) reducing bands from both
Se(VI)-grown and Se(IV)-grown cells confirmed that this protein was SrrA (Table 5.2). Thus,
Srr is expressed in MLTeJB when the organism is grown in the presence of Se(IV), and that SrrA moreover has Se(IV) reducing activity in vitro.
155 Figure 5.3: Identification of a Se(IV) reducing band from periplasmic fractions of Se(VI)-grown
Bacillus beveridgei MLTeJB. Lane A, nondenaturing gel developed for Se(IV) reducing activity with reduced methyl viologen. The clearing indicates Srr activity. Lane B, lane A after staining with Coomassie brilliant blue, showing a protein band at the same location as the Srr activity.
Lane C, SDS-polyacrylamide gel of the excised band Srr activity from lane A visualized by staining with Coomassie brilliant blue.
156
Figure 5.4: Identification of a Se(IV) reducing band from periplasmic fractions of Se(IV)-grown
Bacillus beveridgei MLTeJB. Lane A, nondenaturing gel developed for Se(IV) reducing activity with reduced methyl viologen. The clearing indicates Srr activity. Lane B, lane A after staining with Coomassie brilliant blue, showing a protein band at the same location as the Srr activity.
Lane C, SDS-polyacrylamide gel of the excised band Srr activity from lane A visualized by staining with Coomassie brilliant blue.
157 Sample Locus tag/ Protein Score Number of accession annotation unique peptides number 80 kDa protein BBEV_1018, Molybdopterin 52.77 17 from SDS- AOM82387.1 oxidoreductase, PAGE gel cut of thiosulfate Se(IV) reducing reductase-like band from Se(VI)-grown MLTeJB periplasmic fraction
80 kDa protein BBEV_1018, Molybdopterin 13.97 11 from SDS- AOM82387.1 oxidoreductase, PAGE gel cut of thiosulfate Se(IV) reducing reductase-like band from Se(IV)-grown MLTeJB periplasmic fraction
Table 5.2: LC MS/MS results for proteins eluted from SDS-PAGE gel cuts of Se(IV) reducing bands from Bacillus beveridgei MLTeJB Se(VI)- and Se(IV)-grown periplasmic fractions.
158 Discussion, conclusion, and future directions
Our investigation of Se(IV) metabolism in T. scotoductus SA-01 did not identify any components of Srr in Se(IV) reducing bands. It is likely, then, that the putative srr operon is actually a psr operon. Polysulfide respiration in the epsilonproteobacterium Wolinella succinogenes requires a sulfide dehydrogenase (Sud) protein that is a 149 amino acid protein composed of a single rhodanese domain with an active site Cys residue (Klimmek et al. 1998).
The Sud protein functions in W. succinogenes functions in polysulfide respiration to increase the affinity of Psr for polysulfide by an order of magnitude by transporting polysulfide to Psr for reduction, though the gene encoding this dehydrogenase is not found in the psr operon.
Consistent with the possibility that the putative srr operon in T. scotoductus is actually a psr operon, SA-01 has been shown to be capable of sulfur reduction (Kieft et al. 1999). W. succinogenes was initially thought to be respiring sulfur when the organism was grown in the presence of sulfur (Schröder, Kröger and Macy 1988), but it was later determined that the organism was utilizing polysulfide as a terminal electron acceptor that was formed when sulfide
(added as a reducing agent to the medium) dissolved the sulfur supplied as the electron acceptor
(Klimmek et al. 1991). It has previously been noted that “sulfur respiration” in non- hyperthermophilic organisms is likely polysulfide respiration, given the extremely low solubility of sulfur (Hedderich et al. 1998). Thus, we contend that investigation of Se(IV) metabolism in
T. scotoductus SA-01 has determined conclusively that this operon is a psr operon, and constitutes the first report of the gene encoding the Sud protein being part of a psr operon.
The proteomic identification of components of the Sox pathway in a Se(IV) reducing band in SA-01 periplasmic fractions is an extremely interesting result. It should be noted straightaway that only purification of the Se(IV) reducing enzyme(s) or genetic knockouts can
159 definitely conclude that these Sox pathway enzymes participate in Se(IV) reduction either in vitro or in vivo. Thus, concluding that Sox proteins mediate Se(IV) reduction in this organism remains speculative. This merits further investigation, however. The Sox pathway mediates the oxidation of sulfur species, including sulfide, elemental sulfur, sulfite, and thiosulfate during chemolithoautotrophic growth (Ghosh and Dam 2009). To our knowledge, there is no report in the literature of Sox proteins functioning in vivo to reduce substrates.
Of course, it is possible that Sox pathway proteins were not reducing Se(IV) as a detoxification mechanism. Several reviews have noted that no organisms are known to oxidize selenium species as electron donors for chemolithoautotrophic growth (Stolz et al. 2006;
Nancharaiah and Lens 2015). This constitutes a glaring imbalance in the selenium biogeochemical cycle, as all known selenium specific transformations are reductive, not oxidative. The reduced selenium species elemental selenium and selenide could function as electron donors to fuel chemolithoautotrophic growth, though the extreme sensitivity of selenide to O2 (Nuttall and Allen 1984). Testing this hypothesis would require overcoming numerous practical obstacles. Purchasing large quantities of selenide (either as sodium selenide salt crystals or hydrogen selenide gas cylidners) is financially impractical, and methods for generating large quantities of elemental selenium are laborious (e.g., see (Dowdle and Oremland
1998; Herbel et al. 2003; Tarze et al. 2007) . Furthermore, basic analytic techniques for quantifying selenide and elemental selenium concentrations to establish a mass balance between the oxidation of these selenium species and reduction of a terminal electron acceptor, as well as biochemical assays for detecting their oxidation, would need to be developed.
Our investigation of Se(IV) metabolism in D. hafniense PCP-1 demonstrates that the organism is clearly able to efficiently reduce large quantities of Se(IV), but, in line with previous
160 results, this reduction is not linked to energy conservation. The very low levels of expression of
Se(IV) reducing protein, however, precluded identification of a specific Se(IV) reductase. The low yields of Se(IV) reductase in PCP-1 suggests that gene knock-outs would be the only practical way of assessing whether Srr mediates Se(IV) reduction in this organism. Even if Srr is the physiological Se(IV) reductase of D. hafniense PCP-1, the inability of this enzyme to mediate
Se(IV) respiration would seem to undercut the hypothesis that Srr is a reliable marker for the ability to exploit Se(IV) as a terminal electron acceptor.
However, strains of D. hafniense possess an astonishing degree of metabolic flexibility, utilizing a wide array of halogenated compounds, metals, and metalloids as substrates for growth, and part of this versatility is undergirded by an extraordinary array of over 50 genes encoding putative molybdopterin-containing proteins and subunits (Nonaka et al. 2006). Many of these genes appear to be the result of gene duplications. Gene duplication events are known mechanisms by which enzymes acquire novel catalytic function in protein evolution (Jensen
1976; Jacob 1977). Several lines of evidence indicate the potential of Srr in PCP-1 to have acquired novel functions in this atypical lineage of clostridia. First, the genome of D. hafniense
PCP-1 contains genes for two psrA homologs, with one situated in a putative srr operon and the other situated in a putative psr operon. Secondly, an SrrF homolog has been detected in D. hafniense TCE1 when the organism has been cultivated on tetrachloroethene as a terminal electron acceptor (Prat et al. 2011) and in the presence of large quantities of sulfide, indicating that SrrF could function to detoxify sulfide (Prat et al. 2012). Thus, Srr could have acquired novel catalytic functions as an adaptation for growth on a staggering array of substrates.
Unlike the two organisms discussed above, investigation of Se(IV) metabolism in B. beveridgei MLTeJB was straightforward. MLTeJB expresses Srr, and the enzyme functions to
161 reduce Se(IV) during respiratory growth when supplied with either Se(VI) or Se(IV). As with its closely related sister taxon B. selenitireducens MLS10, Srr is expressed in the periplasm, and can essentially be purified in non-denaturing gels via electrophoretic separation. Unlike the MLS10
Srr, however, the MLTeJB Srr fractionated cleanly with the periplasmic fraction. The results of these proteomics experiments in B. beveridgei MLTeJB are consistent with the hypothesis that
Srr evolved in bacteria as an adaptation to utilize Se(IV) as a source of energy during anaerobic respiration. However, the close phylogenetic relationship between MLS10 and MLTeJB
(Baesman et al. 2009) implies that the hypothesis that Srr functions as a respiratory Se(IV) reductase as a derived adaptation in a lineage of bacteria cannot be ruled out. Thus, while the inability of the D. hafniense PCP-1 Srr to mediate Se(IV) respiration could be rationalized as a byproduct of the unusual metabolic versatility D. hafniense strains have acquired over the course of evolution, study of Se(IV) respiration in bacteria outside of the Firmicutes phylum is essential to ascertain whether Se(IV) had a single origin in bacteria, or if multiple independent respiratory
Se(IV) reductases have evolved in this domain.
162 CHAPTER 6
PHYLOGENETIC ANALYSIS OF SrrA AND OTHER DIMETHYL SULFOXIDE
REDUCTASE (DMSOR) FAMILY MEMBER CATALYTIC SUBUNITS
Abstract
Mononuclear molybdoenzymes of the dimethyl sulfoxide reductase (DMSOR) family catalyze a number of reactions essential to the carbon, nitrogen, sulfur, arsenic, and selenium biogeochemical cycles. These enzymes are also ancient, likely predating the divergence of the last universal common ancestor (LUCA) into the Bacteria and Archaea domains. We have constructed rooted phylogenies for over 1,000 representatives of the DMSOR family using maximum likelihood methods, focusing on enzymes that mediate crucial reactions in the biogeochemical cycles of chalcophiles. Our results provide the first robust relative ordination of the diversification of these enzymes over deep time. The phylogenetic analysis provides compelling evidence that formate dehydrogenases involved in acetogenesis and methanogenesis, as well as formate dehydrogenase N and polysulfide and thiosulfate reductases were likely part of LUCA’s metabolic repertoire. Anaerobic arsenite oxidases and respiratory arsenate and selenite reductases arose later in the Archean Eon. Tetrathionate reductases and aerobic arsenite oxidases represent the most recently derived lineages. Our results argue persuasively that chalcophilic substrates, including arsenic and selenium oxyanions and multivalent sulfur intermediates that were present on the anoxic Archean Earth, were key to the diversification of this family. Additionally, ancestral reconstruction and evolutionary rate analyses identified a number of amino acid residues in the iron-sulfur cluster, Mo/W, and distal pyranopterin binding sites that form the catalytic core of these enzymes and were conserved across multiple diversification events. Most significantly, our work provides a robust framework for future
163 evolutionary study of an assemblage of enzymes that have sustained anaerobic life on Earth for over 3.7 billion years.
Introduction
Ubiquitous in Archaea and Bacteria, mononuclear molybdoenzymes of the dimethyl sulfoxide reductase (DMSOR) family are believed to have been core components of the first anaerobic respiratory chains, and thus present at life’s origins (Schoepp-Cothenet et al. 2012,
2013; Nitschke and Russell 2013; Sousa et al. 2013). Reactions catalyzed by these oxidoreductases are integral components of the carbon, nitrogen, and sulfur biogeochemical cycles, as well as the biogeochemical cycles of arsenic and selenium (Grimaldi et al. 2013). The family, which has been defined by the presence of a mononuclear molybdopterin or tungstopterin bis(pyranopterin guanine dinucleotide) (Mo/W-bisPGD) co-factor (Grimaldi et al. 2013), was named after DMSO reductases as these were among the first members to be well-characterized
(Bilous et al. 1988; Weiner et al. 1988; Cammack and Weiner 1990; Schindelin et al. 1996;
Schneider et al. 1996). As more representatives of the family were discovered and characterized, these molybdoenzymes were frequently found to be members of a heterotrimeric complex, consisting of the catalytic subunit (the Mo/W-bisPGD-harboring subunit), an electron transfer subunit with as many as four Fe-S clusters, and a membrane anchoring subunit that tethers the complex to the membrane and also functions to transfer electrons to or from the membrane quinone pool. Thus, members of the family have also been referred to as Complex Iron-Sulfur
Molybdoenzymes (CISMs) (Rothery, Workun and Weiner 2008). The catalytic subunit may additionally have a twin-arginine translocation motif for export to the periplasm, and a [4Fe-4S] or [3Fe-4S] iron-sulfur cluster (Rothery, Workun and Weiner 2008; Grimaldi et al. 2013). It has been noted, however, that members of this family may be part of a complex in which one or
164 more of these subunits are missing, or form a complex with different subunits altogether
(McEwan et al. 2002; Rothery, Workun and Weiner 2008; Grimaldi et al. 2013). One such example is the periplasmic nitrate reductase, which can be as simple as two peptides (e.g.,
NapAB) or have additional subunits in a complex (e.g., NapABCGH) (Sparacino-Watkins, Stolz and Basu 2014).
Arsenic metabolism has previously been documented in 2.72 billion year old stromatolites, providing conclusive evidence that arsenic cycling was an active feature of
Neoarchean (2.8 – 2.5 billion years ago (Gya)) environments (Sforna et al. 2014). This resulted in a vigorous debate concerning the evolution of arsenic oxyanion utilization in respiration.
Arsenite can serve as an electron donor to stimulate chemolithoauthotrophic growth during aerobic (Stolz et al. 2006) and anaerobic respiration (Zargar et al. 2010, 2012). Arsenite can also be utilized as an electron donor to fuel photolithoautotrophic growth during anoxygenic photosynthesis (Kulp et al. 2008; Stolz 2017). The ability to use arsenate as a terminal electron acceptor during anaerobic respiration is a trait found among phylogenetically diverse Bacteria and Archaea (Oremland et al. 2009). Arsenate respiration is catalyzed by the respiratory arsenate reductase (Arr), while the anaerobic arsenite oxidase (Arx) mediates the oxidation of arsenite during anaerobic respiration. A different enzyme, aerobic arsenite oxidase (Aio), catalyzes the oxidation of arsenite during aerobic respiration. Both Aio and Arx can function in photosynthetic organisms to exploit arsenite as an electron donor during anoxygenic photosynthesis.
The catalytic subunits of Arr and Arx, ArrA and ArxA, appear to be closely related members of the DMSOR family (Zargar et al. 2012), though both seem to be only distantly related to the catalytic subunit of Aio, AioA (Stolz et al. 2006). Phylogenies constructed using
165 the neighbor-joining method have been used previously to support the idea that AioA was the ancestral arsenite oxidase catalytic subunit (Lebrun et al. 2003, 2006; Duval et al. 2008), consistent with an origin of arsenite oxidation from around the Great Oxygenation Event (GOE)
~2.5 Gya. Other researchers, utilizing geochemical and physiological arguments, have argued persuasively that anaerobic photo- and chemolithoautotrophic arsenite oxidation mediated by
ArxA functions efficiently in modern environments (Kulp et al. 2008; Oremland et al. 2009), demonstrating that molecular oxygen is not necessary for arsenite oxidation. This raises the possibility that an active arsenic biogeochemical cycle could have been a salient feature of environments throughout much of the Archean Eon (~4.0 – 2.5 Gya).
It has recently been established that the ability to synthesize and incorporate the 21st proteinogenic amino acid selenocysteine (Sec) was likely a feature of the metabolism of the last universal common ancestor (LUCA) (Mariotti et al. 2015; Weiss et al. 2016). While the evolution of the mechanisms of Sec utilization has received some attention in the literature
(Zhang et al. 2006; Peng et al. 2016), the evolutionary history of proteins that harbor Sec residues (termed selenoproteins) has yet to be addressed. A number of DMSOR family enzymes are known to be selenoproteins, including NAD+-dependent formate dehydrogenases involved in acetogenesis (Andreesen and Ljungdahl 1973, 1974; Graentzdoerffer et al. 2003) and F420- dependent formate dehydrogenases involved in hydrogenotrophic methanogenesis (Jones,
Dilworth and Stadtman 1979b; Jones and Stadtman 1981; Wood, Haydock and Leigh 2003).
The Sec residue functions in these selenoproteins to coordinate the Mo/W-bisPGD co-factor
(Boyington et al. 1997; Jormakka et al. 2002; Raaijmakers et al. 2002). Indeed, formate dehydrogenases are among the most common selenoproteins in organisms capable of utilizing
Sec (Zhang et al. 2006; Rother and Krzycki 2010; Peng et al. 2016). Assessing whether Sec use
166 is ancestral in these enzymes then would provide important insights into the evolution of Sec utilization in prokaryotes.
Additional interest in the evolution of the selenium biogeochemical cycle has been driven by the possibility of using selenium isotopes to probe the onset and dynamics of the growing oxidation state of Earth environments throughout the Archean Eon, the GOE, and the
Neoproterozoic Oxygenation Event (1.0 – 0.541 Gya) (Pogge von Strandmann et al. 2015;
Stüeken, Buick and Anbar 2015; Kipp et al. 2017). A number of stable isotopes of selenium exist in nature and include 74Se, 76Se, 77Se, 78Se, 80Se, and 82Se (Stüeken 2017). These isotopes have abundances of 0.87%, 9.36%, 7.63%, 23.78%, 49.61%, and 8.73% respectively. These isotopes display a pattern of mass dependent fractionation, with lighter isotopes tending to fractionate with the reduced species selenide and elemental selenium during reduction of the more oxidized oxyanions selenate and selenite (Stüeken, Buick and Anbar 2015). Tracking the patterns of selenium isotope fractionation in organic and pyritic shales over deep time could thus potentially serve as a highly sensitive barometer for the presence of O2, given the element’s many oxidation states and exquisite sensitivity to fluctuations in environmental redox states
(Stüeken 2017). This has led to a detailed model of how the selenium biogeochemical cycle evolved in which the most prevalent species of selenium in the Archean Eon were selenide and elemental selenium, with a small reservoir of selenite present only in the Neoarchean Eon
(Stüeken, Buick and Anbar 2015). This model seems at odds with the molecular evidence for the deep antiquity of Sec utilization, as both selenide and elemental selenium have extremely low bioavailability (Sharma et al. 2014). Selenate and selenite, however, are bioavailable, and selenite is widely metabolized by organisms in all three domains of life to synthesize Sec (Tobe and Mihara 2018). While selenate would lack thermodynamic stability without the presence of a
167 very strong oxidant, selenite would be thermodynamically stable even in the presence of weak oxidants (Stüeken 2017). It is possible that hydrothermal vent systems, for example, could have been an early reservoir of selenite for ancient life (Nitschke and Russell 2009). The recent discovery of a respiratory selenite reductase (Srr) (Wells et al. 2019) offers a way to independently assess the antiquity of selenite respiration, and hence the availability of selenite as a source of energy over geologic time.
Enzymes that utilize multivalent sulfur intermediates (e.g., polysulfide, thiosulfate, and tetrathionate) as terminal electron acceptors are also of great evolutionary interest. There is a growing body of literature on how these transient sulfur species shape modern microbial communities (Jørgensen 1990; Jørgensen and Bak 1991; Findlay 2016a), but the role of sulfur intermediates in sustaining life over geologic time has yet to receive consideration. Despite this, the only comprehensive phylogenetic analysis found that the catalytic subunits of polysulfide reductases (PsrA), thiosulfate reductases (PhsA), and tetrathionate reductases (TtrA) were likely ancient representatives of DMSOR family enzymes (Rothery, Workun and Weiner 2008), having homologs in both Bacteria and Archaea. Of even greater interest is the stunning promiscuity of these catalytic subunits. The biochemical function of PsrA homologs in vivo was initially shown to be to reduce polysulfide during polysulfide respiration in W. succinogenes (Krafft et al. 1992).
However, subsequent investigation has implicated the PsrA homolog PhsA as the terminal reductase of thiosulfate during thiosulfate respiration in Salmonella enterica serovar
Typhimurium (Heinzinger et al. 1995) and the PsrA homolog SrrA as the terminal reductase of selenite during selenite respiration in Bacillus selenitireducens (Wells et al. 2019). Similarly, the biochemical function of TtrA was initially identified as the terminal reductase of tetrathionate during tetrathionate respiration in Salmonella enterica serovar Typhimurium (Hensel et al. 1999).
168 Later studies have implicated TtrA homologs, however, as the terminal reductase of selenate
(SrdA) during selenate respiration in B. selenatarsenatis (Kuroda et al. 2011b) and as the terminal reductase of arsenate during arsenate respiration in the archaeon Pyrobaculum aerophilum (Cozen et al. 2009).
Mapping enzyme promiscuity (both in vivo and in vitro) onto tree topologies can provide powerful insights into the evolution of functional diversity within large enzyme families and superfamilies, functional connectivity through time between ancestral lineages of promiscuous enzymes and derived lineages that possess specific catalytic activities that may only be present in vitro in more ancestral lineages, as well as revealing patterns of protein “moonlighting” (i.e., proteins possessing multiple functions in vivo) (Copley 2012; Baier and Tokuriki 2014; Jeffery
2014; Baier, Copp and Tokuriki 2016). Incorporating such an approach into a phylogenetic analysis of DMSOR family enzymes would therefore be part of a broader effort to understand the evolutionary mechanisms by which enzymes generally gain novel functions (Jensen 1976;
Jacob 1977). Furthermore, recent phylogenetic analyses of other large enzyme families (Huang et al. 2012; Ngaki et al. 2012; Voordeckers et al. 2012; Mohamed and Hollfelder 2013; Baier and Tokuriki 2014; Pabis, Duarte and Kamerlin 2016) suggests that incorporating data on the in vivo catalytic functions of DMSOR family members into a phylogenetic analysis would yield novel insights into to how the arsenic, selenium, and sulfur biogeochemical cycles have evolved over deep time, given the functional promiscuity of these catalytic subunits.
Despite the diverse substrates utilized by these molybdoenzymes and their significance in core biogeochemical cycles, the evolutionary relationships among members of the DMSOR family remain poorly resolved. The only previous phylogenetic analysis employed the neighbor- joining method and lacked a root that would allow for a relative ordination of when these
169 enzymes, and their associated biochemical functions, diversified from the broader family through geologic time (Rothery, Workun and Weiner 2008). The neighbor-joining method has been superseded by likelihood-driven methods (e.g., Bayesian (Huelsenbeck et al. 2001) and maximum likelihood (Huelsenbeck and Crandall 1997) analysis) that offer more robust statistical evaluation of nodes within tree topologies as well as superior phylogenetic resolution. Even so, recent phylogenetic analyses of DMSOR family enzymes with maximum likelihood methods used a small subset of enzymes (Harel et al. 2016; Edwardson and Hollibaugh 2017). Further, these phylogenies either lacked a root altogether or used enzymes within the family to root the tree.
We were interested in exploring the evolutionary relationships among DMSOR catalytic subunits, particularly focusing on those that utilize arsenic and selenium oxyanions and multivalent sulfur intermediates. Focusing on these members is warranted, as the biogeochemical cycles of the elements these catalytic subunits exploit as substrates share a common geochemical affinity for sulfide-bearing minerals, uniting them as crucial and relatively abundant chalcophilic elements (Goldschmidt 1937). Exploiting recent advances in our knowledge of the probable physiology of LUCA (Weiss et al. 2016), we constructed maximum likelihood phylogenies with
1,077 members of this family using the aldehyde:ferredoxin family as a root. The aldehyde:ferredoxin family was also likely a part of LUCA’s physiology (Weiss et al. 2016) and harbors a similar tungstopterin co-factor (Hille, Hall and Basu 2014). We were also able to take advantage of a wealth of genomic data that has been generated since the previous comprehensive phylogenetic analysis of this family (Wu et al. 2009; Mukherjee et al. 2017) to enable substantially more extensive taxon sampling. Our results illustrate a catalytic core for this staggeringly diverse family of enzymes. Extensive taxon sampling allowed us to identify
170 lineages that were likely present in LUCA. Ancestral sequence reconstruction and an evolutionary rate analysis identified a number of amino acid residues that have been strongly conserved through multiple diversification events over the course of 3.7 billion years of evolution. These sites were all binding sites associated with the [4Fe-4S] cluster present in many of these enzymes and the distal pyranopterin of the Mo/W-bisPGD co-factor. Most importantly, our work demonstrates that adapting the paradigms and tools available in the broader literature of protein evolution will be enormously productive for further study of ancient bioenergetic enzymes.
Results and Discussion
We sampled a total of thirteen physiologically distinct enzymes from the DMSOR family
(Table 6.1) to reconstruct a maximum likelihood analysis using RAxML (Stamatakis 2014) (Fig.
6.1) and IQ-TREE (Nguyen et al. 2015) (Figs. S6.1 and S6.2). Both trees yielded similar topologies. The trees provide robust support for grouping the enzymes into two major clades, with robust bootstrap support of 100. One clade consists of formate dehydrogenases (nitrate- inducible formate dehydrogenase N (FdhN) catalytic subunits (FdhG), formate hydrogen lyases
(FdhH), and NAD- and F420-dependent formate dehydrogenases), as well as AioA. We refer to this clade as the FDH clade. The other clade consists exclusively of the other chalcophilic oxidoreductases (ChalcORs), and includes PsrA, PhsA, SrrA, ArxA, ArrA, TtrA, and SrdA.
Other representatives of the family (e.g., periplasmic nitrate reductase catalytic subunits (NapA), respiratory nitrate reductase catalytic subunits (NarG), dimethyl sulfoxide reductase catalytic subunits (DmsA)) were included to provide insights as to the evolutionary relationships between the enzyme lineages we were interested in and the broader family. The NapA representatives fit within the FDH clade. The other representatives were enzymes that use a Ser or an Asp residue
171 as a ligand for the Mo/W atom in the Mo/W-bisPGD co-factor. These representatives formed a third monophyletic clade in our trees with robust bootstrap support of 100.
Mo/W-bisPGD DMSOR family Substrate Function Localization Mo catalytic lineage ligand subunit 2- Polysulfide PsrA/PhsA/SrrA Sn Reduces Periplasm Cys reductase polysulfide in (PsrA) anaerobic respiration 2- Thiosulfate PsrA/PhsA/SrrA S2O3 Reduces Periplasm Cys reductase thiosulfate in (PhsA) anaerobic respiration 2- Selenite PsrA/PhsA/SrrA SeO3 Reduces selenite Periplasm Cys reductase in anaerobic (SrrA) respiration 3- Arsenite ArxA/ArrA AsO3 Oxidizes arsenite Periplasm Cys oxidase (ArxA) as an electron donor in anaerobic respiration or anoxygenic photosynthesis 3- Arsenate ArxA/ArrA AsO4 Reduces arsenate Periplasm Cys reductase in anaerobic (ArrA) respiration 2- Tetrathionate TtrA/SrdA S4O6 Reduces Periplasm Cys reductase tetrathionate in (TtrA) anaerobic respiration 2- Selenate TtrA/SrdA SeO4 Reduces selenate Periplasm Cys reductase in anaerobic (SrdA) respiration 3 Archaeal TtrA/SrdA AsO4 Reduces arsenate Periplasm Cys arsenate in anaerobic reductase respiration in some archaea Formate FdhG HCOO-1 Oxidizes formate Periplasm Sec/Cys dehydrogenase as an electron N (FdhG) donor in anaerobic respiration
172 NAD- ? CO2 Reduces carbon Cytoplasm Sec/Cys dependent dioxide to formate formate during dehydrogenase acetogenesis F420-dependent ? CO2 Reduces carbon Cytoplasm Sec/Cys formate dioxide to formate dehydrogenase during methanogenesis Formate ? HCOO-1 Oxidizes excess Cytoplasm Sec/Cys hydrogen lyase formate to carbon (FdhH) dioxide during fermentative growth 3- Arsenite AioA AsO3 Oxidizes arsenite Periplasm No oxidase (AioA) as an electron ligand donor in aerobic respiration and anoxygenic photosynthesis Table 6.1
List of all DMSOR family lineages analyzed in the phylogenetic analysis, as well as their function and localization in bacterial and archaeal cells.
173
Fig. 6.1
Maximum likelihood phylogeny of 1,070 DMSOR family protein sequences. All sequences came from cultured organisms with sequenced genomes. Each subfamily we analyzed formed a monophyletic clade, and these clades are each given a distinct color. The subfamily associated with each color is indicated in the figure. The bootstrap support for crucial nodes in our phylogeny is provided in text. The scale bar refers to the number of amino acid substitutions per site.
174 This work is the first robust phylogenetic study of formate dehydrogenases and reveals that the current practice of organizing these DMSOR members by their biochemical and physiological function is a misrepresentation of their evolutionary history. Formate dehydrogenases have been traditionally classified within three distinct groups (Hille, Hall and
Basu 2014). The first group comprises a collection of periplasmic formate dehydrogenases that are canonically expressed during respiratory growth on nitrate and function to oxidize formate to
CO2. These include the FdhG catalytic subunits of FdhN and another formate dehydrogenase expressed in E. coli during respiratory growth on nitrate and O2 in low O2 concentrations
(Sawers 1994). The FdhN complex has been studied extensively (e.g., (Jormakka et al. 2002;
Raaijmakers et al. 2002)), but the evolutionary relationship between the FdhG catalytic subunits of FdhO and FdhN is unclear. The similar operon organization of the FdhN and FdhO complexes(Hille, Hall and Basu 2014) and the identical molecular weights (Sawers et al. 1991) and extremely high sequence identity (~75%) of the catalytic subunits (Hille, Hall and Basu
2014) strongly suggest that these enzymes are closely related. We found that putative FdhG did indeed form a coherent monophyletic clade with a robust bootstrap support of 99, as expected under the current paradigm of formate dehydrogenase taxonomy.
The conventional formate dehydrogenase taxonomy classifies remaining formate dehydrogenases into two distinct groups, with one group comprising formate hydrogen lyase
(FdhH) that functions to reduce excess formate to CO2 and H2 during fermentative growth
(Sawers 1994). FdhH from E. coli has been extensively studied (Boyington et al. 1997;
Khangulov et al. 1998). The other group comprises a disparate collection of NAD+-dependent formate dehydrogenases (FdsA) found in aerobic bacteria that function to oxidize formate to
CO2, and like FdhH does not appear to be linked to energy conservation (Oh and Bowien 1998;
175 Niks et al. 2016). FdsA of Cupriavidus necator (formerly Ralstonia eutropha) has been extensively characterized (Oh and Bowien 1998; Niks et al. 2016), and appears to be able to catalyze the reverse reaction (CO2 reduction to formate) in vitro (Yu et al. 2017). Curiously, this taxonomy neglects the formate dehydrogenases that function in acetogenesis and hydrogenotrophic methanogenesis (Stock and Rother 2009). Regardless, our results offer powerful evidence that this scheme is not tenable from an evolutionary perspective. NAD+- dependent formate dehydrogenases involved in acetogenesis, F420-dependent formate dehydrogenases involved in methanogenesis, FdH, and FdsA all form a coherent monophyletic clade in our tree topology (Figs. 6.1 and 6.2) with robust bootstrap support of 97. Given that the most basal lineages within this clade are involved in acetogenesis and methanogenesis (e.g., the formate dehydrogenases of Moorella thermoacetica (Andreesen and Ljungdahl 1973, 1974),
Eubacterium barkeri (Graentzdoerffer et al. 2003), Methanococcus vanielli (Jones, Dilworth and
Stadtman 1979b; Jones and Stadtman 1981), and M. maripaludis (Wood, Haydock and Leigh
2003)), and the acetogenic formate dehydrogenases function to reduce CO2 to formate (Stock and Rother 2009), it seems to us that referring to this assemblage of enzymes as “formate dehydrogenases” is misleading. We propose referring to these enzymes as formate/CO2 oxidoreductases (FormCO2ORs) would better reflect their physiological function.
Both monophyletic clades of FdhG and FormCO2OR form the basal lineages of the broader FDH clade. All of these homologs are predicted to coordinate the Mo/W atom of the
Mo/W-bisPGD co-factor using either Sec or Cys. Two lineages of the DMSO family are more recently derived within FDH. One monophyletic clade includes NapA. NapA uses a Cys residue to position the Mo/W atom, and these catalytic subunits function in vivo in different cells for dissimilatory and assimilatory reduction of nitrate, as well as redox homoeostasis (Sparacino-
176 Watkins, Stolz and Basu 2014). The other monophyletic clade comprises AioA. AioA is highly unusual in that there is no amino acid ligand coordinated to the Mo/W atom (Grimaldi et al.
2013). There was robust support for grouping NapA and AioA into two distinct monophyletic clades, with bootstrap support of 96. The monophyly of AioA was also robustly supported, with bootstrap support of 100.
The tree topology then suggests two substantial clades diverged from the broader family after FDH. These clades comprise the ChalcOR clade and a clade of enzymes that use Ser or
Asp ligands for coordinating the Mo/W atom (Fig. 6.1), with a bootstrap support of 80. In contrast to the FDH clade, the ChalcOR clade shares a common physiological function to reduce or oxidize chalcophilic substrates, as the clade includes PsrA, PhsA, SrrA, ArxA, ArrA, TtrA, and SrdA homologs (Figs. 6.1-6.3). Homologs from this clade also share a common amino acid ligand for the Mo/W atom, namely a Cys residue. Within the ChalcOR clade, PsrA/PhsA/SrrA are the most ancestral members, and the monophyly of this clade is robustly supported with bootstrap support of 97. ArxA/ArrA and TtrA/SrdA diverged after and comprise separate monophyletic lineages, which is well-supported with bootstrap support of 100. The previous phylogenetic analysis suggested that these chalcophile reducing and oxidizing enzymes were likely more closely related to each other than to other members of this family (Rothery, Workun and Weiner 2008), but we provide the most robust support yet to date that this is indeed the case.
The placement of the ChalcOR clade within the tree topology offers strong evidence that exploiting chalcophilic substrates during anaerobic respiration was a powerful selective force in the evolution of these enzymes.
Our phylogenies cannot assess when in the evolutionary record lineages of DMSOR family enzymes diversified absent geochemical evidence for these metabolisms in dated
177 lithologies of the fossil record. The lack of data prevents traditional divergence time analysis using the relaxed molecular clock. Nonetheless, the phylogeny can be used to determine whether lineages within the DMSOR family were likely to have diverged prior to the divergence of
LUCA into the Bacteria and Archaea domains. This can be assessed by determining whether homologs from the Bacteria form coherent monophyletic clades from homologs from the
Archaea within lineages of these enzymes. Three lineages clearly meet this criterion (Fig. 6.2):
FormCO2ORs, and the FdhG and PsrA/PhsA/SrrA catalytic subunits. This provides evidence that the basal divergence of these proteins occurred over 3.7 Gya, making them among the most ancient enzymes within the family. The physiological function of these well-characterized members in anaerobic respiration provide compelling evidence that LUCA likely possessed the capacity for energy conservation via acetogenesis, using the Wood-Ljungdahl pathway, and/or hydrogenotrophic methanogenesis. This is consistent with previous work indicating LUCA was capable of chemolithoautotrophic growth using H2 as an electron donor and CO2 as an electron acceptor (Weiss et al. 2016). LUCA also likely possessed a simple electron transport chain to exploit formate as an electron donor and polysulfide and possibly thiosulfate (see below) as a terminal electron acceptor.
178
Fig. 6.2
This tree is identical to the phylogeny in Fig. 1, except that the organism whose genome the putative CISM protein is encoded in is provided in a shorthand code in the text at each branch, along with the family we predicted the protein homolog was part of. The key to each organism’s code is found in the Supplemental Information. The name used for each protein family is consistent with the rest of the text. The phylum each organism is affiliated with is indicated by the color of the text of each code name as described in the figure legend. Archaeal phyla in the legend is indicated by an asterisk. Additionally, protein representatives from archaea are represented in the phylogeny by dotted lines. The DMSOR family each protein is affiliated with is indicated by the branch color as described in the figure legend. The scale bar refers to the number of amino acid substitutions per site.
179 The phylogeny suggests that the ArxA/ArrA, TtrA/SrdA, and AioA lineages diversified sometime later after Bacteria and Archaea diverged into independent domains. The most recently derived branches in the tree topology are AioA and TtrA/SrdA. Our finding that
TtrA/SrdA represents a highly derived lineage within the DMSOR family contradicts previous work (Rothery, Workun and Weiner 2008), and demonstrates the benefits of expanded taxon sampling in separating genuine vertical descent from horizontal gene transfer events. The
ArxA/ArrA lineage likely predates the diversification of the AioA and TtrA/SrdA lineages. A direct comparison of the bacterial and archaeal phyla represented in the ArxA/ArrA and AioA lineges (Fig. 6.3) is also consistent with this scenario, as the ArxA/ArrA lineage contains homologs from several deep-branching bacterial phyla (e.g., Firmicutes, Chrysiogenetes, and
Spirochetes) not represented in the AioA lineage. The only homolog from the Firmicutes phylum in the AioA lineage comes from Bacillus oryziterrae, which clustered with the
Haloarchaea. This suggests possible lateral gene transfer into a recently derived lineage of
Bacillales. Additional lines of evidence argue for the greater antiquity of ArxA/ArrA compared to AioA. First is the robust support (Fig. 6.3), with a bootstrap support of 100 in the tree topology, for considering ArxA homologs a distinct clade from ArrA homologs, as has been proposed previously (Zargar et al. 2012). Moreover, this clade of ArxA homologs is ancestral to the ArrA clade. The greater antiquity of ArxA homologs, suggested by its basal position within the lineage, is consistent with an anaerobic origin for arsenite oxidation. This would introduce arsenate in anoxic environments, creating an ecological niche for subsequent evolution of arsenate respiration.
180
Fig. 6.3
These are subtrees of the phylogeny shown in Figs. 1 and 2 that focuses exclusively on the AioA family (left) and the ArxA/ArrA family (right). The organism whose genome the putative CISM protein is encoded in is provided in shorthand code in the text at each branch, along with the family we predicted the protein homolog was part of. The key to each organism’s code is found in the Supplemental Information. The name used for each protein family is consistent with the rest of the text. The phylum each organism is affiliated with is indicated by the color of the text of each code name as described in the figure legend. Putative ArxA homologs are differentiated from putative ArrA homologs by the color of the branches. Light red branches indicate a protein is an ArxA homolog; black branches indicates a protein is an ArrA homolog. The scale bar in each subtree refers to the number of amino acid substitutions per site.
181 Finally, a number of geochemical and physiological lines of evidence support the greater antiquity of the ArxA/ArrA lineage compared to the AioA lineage. The 2.7 Gya Archean stromatolites displaying evidence of arsenic metabolism were formed under anoxic hypersaline and alkaline conditions analogous to conditions prevalent in many modern soda lake environments (Sforna et al. 2014). Anaerobic arsenite oxidation has been shown to be mediated by ArxA homologs in these environments, rather than AioA (Zargar et al. 2012; Hernandez-
Maldonado et al. 2017). Furthermore, AioA functions predominantly to oxidize arsenite during aerobic respiration, even in arsenite oxidizing representatives occupying deep branches in the family (Fig. 3), including representatives from the Haloarchaea (Ordoñez et al. 2018) and the
Aquificales (Härtig et al. 2014). The only demonstration of AioA catalyzing anaerobic arsenite oxidation comes from Chloroflexus aurantiacus, which can exploit arsenite as a source of reducing equivalents during anoxygenic phototrophy (Stolz 2017). In contrast, no ArxA is known to be expressed under oxic conditions. These diverse lines of evidence all argue for the greater antiquity of anaerobic arsenite oxidation mediated by ArxA rather than aerobic arsenite oxidation mediated by AioA.
Since our phylogeny robustly supported the greater antiquity of the ArxA clade within the ArxA/ArrA lineage, we were interested to see if the phylogeny could provide an ordination for when the different catalytic functions of the PsrA/PhsA/SrrA and TtrA/SrdA lineages evolved. We have previously noted that putative PsrA, PhsA, and SrrA can seemingly be differentiated by operon organization (Wells et al. 2019). PsrA homologs are observed in operons that contain a PsrA catalytic subunit, a four [4Fe-4S] cluster electron transfer subunit, a membrane anchor subunit with no heme groups, and, in some cases, a rhodanese domain- containing protein with a single rhodanese domain containing a catalytically active Cys residue.
182 Putative PhsA homologs are found in operons with identical catalytic and electron transfer subunits, but the membrane anchor subunit has two heme b groups, similar to FdhN membrane anchor subunits (Rothery, Workun and Weiner 2008). No rhodanese domain-containing proteins have been reported to be associated with Phs operons. Putative Srr operons have identical catalytic, electron transfer, and membrane anchor subunits as Psr operons, but additionally contain one or more distinct rhodanese domain-containing proteins that have two rhodanese domains, both with catalytically active Cys residues (Wells et al. 2019).
When these operon structures are mapped onto the PsrA/PhsA/SrrA lineage and the clade of putative haloarchaeal PsrA/PhsA/SrrA (Figs. S6.3 and S6.4), it is apparent that the three different operon organizations do not form monophyletic clades, consistent with our previous results (Wells et al. 2019). Nonetheless, putative PsrA and PhsA are found in both Bacteria and
Archaea (Fig. S6.3), suggesting that both polysulfide and thiosulfate respiration may pre-date the divergence of LUCA into the Bacteria and Archaea domains. The ability of LUCA to respire thiosulfate is also supported by recent results demonstrating that the P. aerophilum putative PsrA homolog actually mediates thiosulfate reduction in vivo (Haja et al. 2019). Selenite respiration appears to have evolved after the divergence of LUCA, as no archaeal homologs seem to have
Srr operons. This suggests selenite respiration evolved early in the diversification of bacterial lineages, given the presence of Srr operons in the deepest branches of bacterial PsrA/PhsA/SrrA.
The lack of Srr operons in archaea may be misleading, however. The rhodanese domain- containing Sud protein in W. succinogenes, which participates in the polysulfide reduction reaction (Prisner et al. 2003), is not part of the Psr operon. Only physiological characterization of selenite respiring hyperthermophilic archaea, such as P. aerophilum (Huber et al. 2000), can definitively determine whether SrrA is confined to bacteria.
183 We also investigated the operon organization of the Ttr operon of S. enterica serovar
Typhimurium, the Srd operon of B. selenatarsenatis, and the putative arsenate reductase operon of P. aerophilum. Both the tetrathionate reductase and the putative archaeal arsenate reductase were associated with identical operons, with a TtrA-like catalytic subunit, a four [4Fe-4S] electron transfer subunit, and a membrane anchor subunit with no heme groups. The operon organization of the Srd operon of B. selenatarsenatis, however, contained a TtrA-like catalytic subunit, a PsrA-like catalytic subunit, two four [4Fe-4S] cluster electron transfer subunits, and a single membrane anchor subunit with no heme groups. When the two operon organizations were mapped onto the TtrA/SrdA lineage (Fig. S6.5), the phylogeny reveals that the putative Srd operon is found in only three organisms. They are all members of the Firmicutes phylum and have been shown previously to utilize selenate as a terminal electron acceptor (Switzer Blum et al. 1998; Kuroda et al. 2011b; Abin and Hollibaugh 2017). This suggests that the ability of
TtrA-like proteins to function as selenate reductases evolved late in the diversification of this family when a Psr-type operon merged with a Ttr-type operon in a specific lineage of TtrA homologs in the Firmicutes.
Manual inspection of our alignment revealed two regions with substantial conservation of sequence length and amino acid residues across the enzymes we analyzed (Figs. S6.6 and S6.7).
In our alignment, these two regions were sites 563-779 and 1120-1383. These sites corresponded to the [3Fe-4S] cluster binding motif in the AioA lineage and the [4Fe-4S] binding motif in the other lineages, as well as part of the Mo/W-bisPGD co-factor binding site. In order to identify residues that could potentially be universally important for the catalytic activity, an ancestral sequence reconstruction analysis was performed. Additionally, a relative evolutionary rate inference analysis at each site of the alignment was performed. We hypothesized that
184 residues that have been conserved across multiple diversification events and were subject to strong negative selection could be important for catalysis.
A number of residues were identified in the iron-sulfur cluster and Mo/W-bisPGD binding sites that were conserved across the FDH and ChalcOR clades. A number of residues conserved in multiple lineages within each clade were also found. Amino acid residues that were recognized as the ancestral state at a particular site with a probability ≥ 0.95 (see Supplementary
Information for the probabilities at each site and the estimated evolutionary rate) were used to construct putative iron-sulfur cluster and Mo/W-bisPGD binding sites at each major node of our phylogeny (Fig. 6.4). The four Cys residues at sites 565, 568, 572, and 762 that coordinate a
[4Fe-4S] cluster present in nearly all of the enzymes we analyzed were predicted to be ancestral and subject to strong purifying selection as expected. The only exception to this was the AioA lineage, in which the ancestral state at site 762 was inferred to be a Ser residue. The Ser residue at site 762 is present in all AioA homologs in our alignment. This suggests that the distinctive high redox potential [3Fe-4S] cluster of AioA (Hoke et al. 2004; van Lis et al. 2012) was ancestral in this lineage.
185
Fig. 6.4
The phylogeny shown in this figure is identical to the phylogeny in Figs. 1 and 2. Each major
DMSOR family we analyzed is collapsed. The color identifying each family is identical to the color scheme used in Fig. 1. Each major node in the phylogeny is indicated by a number. The putative ancestral sequence of each iron-sulfur cluster and distal pyranopterin binding site for each node is provided, with the numbers of the alignment corresponding to the number of the node on the tree.
186 Our ancestral state reconstruction analysis suggests that the ancestral amino acid ligand for Mo/W atom coordination was a Cys residue for the FdhG, FormCO2OR, PsrA/PhsA/SrrA,
ArxA/ArrA, and TtrA/SrdA lineages. The Ala residue in the Mo/W binding site of all AioA homologs in our alignment was found to be the ancestral state. This Ala residue has previously been shown to occupy an analogous position to the Mo/W coordinating Cys and Sec residues of other enzymes in the family (Hoke et al. 2004). This provides substantial evidence that the unusual absence of an amino acid ligand to coordinate the Mo/W atom during catalysis is the ancestral state of the AioA lineage. It has been noted that the lack of an amino acid ligand in
AioA confers on the enzyme active site a geometry that more resembles the active sites of enzyme representatives from the aldehyde:ferredoxin family (Conrads et al. 2002; Hoke et al.
2004) . The derived position of the AioA lineage within our tree topology offers convincing evidence that this is a function of convergent evolution, rather than reflecting an ancestral state for DMSOR family members.
The ancestral sequence reconstruction strongly supports a scenario where the earliest enzymes utilized a [4Fe-4S] cluster to facilitate electron transfer and coordinated the Mo/W atom of the Mo/W-bisPGD co-factor with a Cys residue. The presence of Sec residues throughout the
FdhG and FormCO2OR lineages, however, warrants caution in this interpretation. Currently available bioinformatic programs treat Sec residues as unknown residues. In order to assess whether a Cys residue was indeed the likely ancestral state of these enzymes, we mapped the distribution of Sec- and Cys-containing homologs in the FdhG (Fig. S6.8) and the FormCO2OR
(Fig. S6.9) lineages onto each clade. The distribution of Sec residues in the FdhG lineage is consistent with the ancestral sequence reconstruction analysis, as no archaeal FdhG homologs harbor a Sec residue. This suggests that Sec use in FdhG evolved early in bacteria, and
187 subsequently a number of bacterial lineages have independently lost the ability to incorporate
Sec at this site.
The distribution of Sec use in the FormCO2OR lineage, however, argues persuasively that
Sec was the most probable ancestral ligand for Mo/W coordination in this lineage. The archaeal
FormCO2OR homologs are F420-dependent formate dehydrogenases involved in hydrogenotrophic methanogenesis. The most deeply branched of these exploit a Sec residue for
Mo/W coordination. In bacterial homologs, the use of a Sec residue predominates in the deeper branches of the family. Consistent with the possibility that Sec was the ancestral Mo/W coordinating ligand in the FormCO2OR lineage is the fact that the most ancient bacterial Sec- containing FormCO2OR homologs in the clade are also involved in energy conservation (i.e., acetogenesis). This indicates that as FormCO2OR homologs in bacteria became disconnected from energy conservation (e.g., FdsA catalytic subunits) this lineage experienced strong selective pressure to dispense with Sec. This is consistent with previous work on the evolution of Sec within bacterial lineages, as the evolution of Sec use in proteins is characterized by frequent, independent loss of Sec use across lineages (Zhang et al. 2006). These results indicate that an early selective pressure for the evolution of Sec utilization in LUCA, or even proto-cellular lineages, could have been the utility of selenium in catalyzing the reduction of CO2 to formate during carbon fixation.
Our analysis suggested that a number of residues in the iron-sulfur cluster and Mo/W- bisPGD binding sites not involved in metal coordination were also strongly conserved across multiple diversification events in both major clades and were subject to strong purifying selection. In order to validate our hypothesis that these residues were indeed important for catalysis, we looked through the X-ray crystallography literature. This survey validated our
188 hypothesis, as several of these conserved amino acid residues were identified as forming H bonds with the Q (distal) pyranopterin subunit of the Mo/W-bisPGD co-factor (Table 6.2). This provides compelling evidence that an evolutionary analysis of ancient enzymes can identify conserved regions that have experienced strong negative selective pressure over geologic time due to their fundamental role in catalysis.
Residue Site Number of H bonds Proteins to distal pyranopterin Gly 1225 1 FdhH of Escherichia coli, FdhG of Desulfovibrio gigas Asn 1227 1 FdhH of E. coli, FdhG of D. gigas, NapA of D. desulfuricans AioA of Alcaligenes faecalis Asp 1291 2 FdhH of E. coli*, FdhG of D. gigas, NapA of D. desulfuricans, AioA of A. faecalis Arg 1293 2 FdhH of E. coli FdhG of D. gigas, NapA of D. desulfuricans, AioA of A. faecalis Gly 1341 1 FdhH of E. coli, FdhG of D. gigas, NapA of D. desulfuricans, AioA of A. faecalis Asp 1343 2 FdhH of E. coli*, FdhG of D. gigas, NapA of D. desulfuricans, AioA of A. faecalis, DorA of Rhodobacter capsulatus Table 6.2
List of all amino acid residues identified as being the ancestral state in multiple DMSOR family lineages that have been found to form hydrogen bonds with the distal pyranopterin of one or more representatives in the X-ray crystallography literature. Each residue is provided, along with the site of that residue in our alignment. The number of hydrogen bonds formed with the distal pyranopterin is also provided. The proteins are listed, along with the organism in which the protein is expressed. The numbers in superscript indicate the reference for the crystallography study in the text. * indicates that the residue at that site of the alignment had been mutated to an Asn residue in that particular enzyme
189 This is particularly exciting given recent re-appraisals of the role of pyranopterins in facilitating electron transfer in molybdo- and tungstoenzymes (Rothery et al. 2012; Adamson et al. 2015). Computation chemistry approaches have suggested that the DMSOR family has attained its catalytic versatility through the use of two distinct pyranopterins in the Mo/W- bisPGD co-factor (Rothery et al. 2012). The distal pyranopterin has been predicted to function during catalysis primarily to redox tune the Mo/W center of the enzyme to ensure optimal activity. The P (proximal) pyranopterin is predicted to actually facilitate electron transfer between the Mo/W atom, the substrate, and iron-sulfur cluster (if present). The nomenclature of the two pyranopterin groups refers to their proximity to the iron-sulfur cluster of the catalytic subunits- proximal pyranopterins are the pyranopterin groups closest to the iron-sulfur cluster in the tertiary structure. In agreement with this hypothesis, structural predictions of pyranopterin conformation have suggested that the proximal pyranopterin of the Mo/W-bisPGD co-factor displays significant structural variation compared to the distal pyranopterin (Rothery et al. 2012).
Our bioinformatic approach has independently supported this. The striking conservation of a number of critical amino acid residues in the distal pyranopterin binding site is consistent with the function of this pyranopterin in redox tuning of the Mo/W atom. The proximal pyranopterin binding site cannot be easily discerned by visual inspection. The sequence length varies between families and strongly conserved residues between families are not apparent in the alignment. This is consistent with the need to alter the conformation of the proximal pyranopterin to facilitate oxidation or reduction of diverse substrates with a wide range of midpoint potentials.
The validation of our hypothesis in the X-ray crystallography literature more importantly suggests that the evolutionary analysis of these enzymes can be used in conjunction with
190 structural data and site-directed mutagenesis experiments to identify previously unknown residues that are important for distal and proximal pyranopterin coordination, substrate binding,
Mo/W coordination, and iron-sulfur clustering binding. Two Pro residues at sites 1292 and 1339 and an Asp residue at site 1209 were found to be ancestral, and strongly conserved, in many lineages in our alignment in the distal pyranopterin binding site. However, previous work has not established a putative biochemical function for these residues. This evolutionary perspective can also suggest how particular lineages have modulated pyranopterin, Mo/W, or iron-sulfur cluster coordination over time as they diversified from the family and acquired novel catalytic function. For example, a Glu residue at site 1230 of our alignment is strongly conserved in several lineages in the FDH clade. This residue has previously been found to form a single H bond with the distal pyranopterin of the FdhG protein of Desulfovibrio gigas (Raaijmakers et al.
2002) and the AioA protein of Alcaligenes faecalis(Ellis et al. 2001). However, in the DorA protein of R. capsulatus (Schneider et al. 1996), the analogous residue at this site is a Lys residue, which also forms a single H bond with the distal pyranopterin. Given that Glu and Lys have opposite charges, it appears that the function of these residues in distal pyranopterin coordination would differ between these enzymes.
Our work has presented the first comprehensive phylogenetic analysis of members of the
DMSOR family using maximum likelihood methods and appropriate outgroups to produce rooted phylogenies. Our focus on families that catalyze reactions involving chalcophilic elements have provided new insights into the evolution of the sulfur, arsenic, and selenium biogeochemical cycles. The evidence that LUCA likely possessed the capacity for polysulfide, and probably thiosulfate, respiration provides strong evidence that sulfur intermediates were crucial components of the sulfur biogeochemical cycle early in the evolution of life. We also
191 demonstrate that robust phylogenetic methods compliment geochemical, structural, and physiological data arguing that anaerobic arsenite oxidation mediated by ArxA likely emerged later in the Archean Eon as the first step in the evolution of an active arsenic biogeochemical cycle. The presence of putative Srr operons throughout bacterial lineages provides evidence that selenite respiration could have been an important component of the selenium biogeochemical cycle during the Archean Eon. The lack of Srr operons in archaeal genomes is broadly consistent with current models for the evolution of the selenium biogeochemical cycle (Stüeken
2017). Sec utilization as the amino acid ligand for Mo/W coordination throughout the FdhG and
FormCO2OR lineages also attests to the deep antiquity of the selenium biogeochemical cycle.
Our tree topology indicates that Sec use was likely ancestral within the FormCO2OR lineage, consistent with previous phylogenetic and genomic evidence arguing that Sec use predates the diversification of life into three domains.
Conclusions and future directions
Most importantly, our work establishes that study of the evolution of this family would benefit substantially from insights found in the broader literature on enzyme superfamily evolution. The radiation of the different lineages is characterized by extensive functional promiscuity, as our phylogenies show. There is ample evidence that identical biochemical functions in disparate enzymes are byproducts of convergent evolution, rather than reflecting inheritance of ancestral functions. Our phylogeny confirms that NapA and NarG, which both catalyze nitrate reduction, AioA and ArxA, which both catalyze arsenite oxidation, and the selenate reductase (SerA) of Thauera selenatis and SrdA of Bacillus selenatarsenatis, which both catalyze selenate reduction, are only distantly related (Figs. 6.1 and 6.2). Furthermore, the evolution of these enzymes demonstrates that novel domain acquisitions and the loss of the iron-
192 sulfur cluster in the enzymes and catalytic subunits are quite common, with only the Mo/W- bisPGD co-factor reliably conserved. FsdA homologs, for example, contain N-terminal sequence insertions giving these enzymes seven [4Fe-4S] clusters and a distinctive FMN co- factor binding site (Niks et al. 2016) that bears substantial sequence homology to the Nqo3 subunit of the Thermus thermophilus respiratory complex I (Oh and Bowien 1998). FdhG and
NarG homologs contain novel domains associated with the substrate binding funnel, indicating that these novel domains were important for driving the evolution of their biochemical function
(Rothery, Workun and Weiner 2008). Dimethyl sulfoxide reductases (DorA), trimethylamine N- oxide reductases (TorA), and biotin sulfoxide reductases (BisC) have all lost the [4Fe-4S] cluster.
There are many examples where biochemical study of enzyme activity has illustrated that novel catalytic functions in derived representatives of enzyme super families frequently emerge from the in vitro promiscuity of more ancestral lineages (Huang et al. 2012; Ngaki et al. 2012;
Voordeckers et al. 2012; Mohamed and Hollfelder 2013; Baier and Tokuriki 2014; Pabis, Duarte and Kamerlin 2016). A crucial mechanism by which this arises is gene duplication events (Jacob
1977), and our phylogeny illustrates numerous examples (particularly in the PsrA/PhsA/SrrA lineage) of an organism having multiple enzyme homologs encoded in its genome. This phenomenon has been noted previously in the literature as well, with Alkalilimniola ehrlichii possessing two ArrA/ArxA homologs (Zargar et al. 2010). An additional example is the remarkably versatile bacterium Desulfitobacterium hafniense, which can grow on a wide variety of respiratory substrates, and has a genome that encodes for more than 50 DMSOR family representatives (Nonaka et al. 2006). It underscores the importance of protein purification and characterization studies to augment those using solely bioinformatics or gene knockouts to
193 identify function (Krafft et al. 1992; Heinzinger et al. 1995; Hensel et al. 1999; Zargar et al.
2010).
194 Supplemental figures
Fig. S6.1
Maximum likelihood phylogeny of 1,070 DMSOR family protein sequences reconstructed using the IQ-TREE program. The support for the phylogeny shown was determined from 1000
UltraFast bootstraps. The family associated with the color of each color is indicated in the figure. The scale bar refers to the number of amino acid substitutions per site.
195
Fig. S6.2
Maximum likelihood phylogeny of 1,070 DMSOR family protein sequences reconstructed using the IQ-TREE program. The support for the phylogeny shown was determined from 10000
UltraFast bootstraps. The family associated with the color of each color is indicated in the figure. The scale bar refers to the number of amino acid substitutions per site.
196
Fig. S6.3
Sub-pruned portion of the maximum likelihood phylogeny shown in Figs. 1 and 2 reconstructed using the RAxML program. The sub-prune shows the PsrA/PhsA/SrrA lineage. The text labels for each protein representative in the family is color coded by the operon organization in which the protein is encoded. The operon organizations were determined using the IMG platform. The scale bar refers to the number of amino acid substitutions per site.
197
Fig. S6.4
Sub-pruned portion of the maximum likelihood phylogeny shown in Figs. 1 and 2 reconstructed using the RAxML program. The sub-prune shows the haloarchaeal PsrA and PhsA reductases.
The text labels for each protein representative in the family is color coded by the operon organization in which the protein is encoded. The operon organizations were determined using the IMG platform. The scale bar refers to the number of amino acid substitutions per site.
198
Fig. S6.5
Sub-pruned portion of the maximum likelihood phylogeny shown in Figs. 1 and 2 reconstructed using the RAxML program. The sub-prune shows the TtrA/SrdA/archaeal arsenate reductase lineage. The text labels for each protein representative in the family is color coded by the operon organization in which the protein is encoded. The operon organizations were determined using the IMG platform. The scale bar refers to the number of amino acid substitutions per site.
199
Fig. S6.6
Sub-pruned portion of the maximum likelihood phylogeny shown in Figs. 1 and 2 reconstructed using the RAxML program. The sub-prune shows the FdhG lineage. The text labels for each protein representative in the family is color coded by whether the Mo or W atom is coordinated by a Cys or a Sec residue. The scale bar refers to the number of amino acid substitutions per site.
200
Fig. S6.7
Sub-pruned portion of the maximum likelihood phylogeny shown in Figs. 1 and 2 reconstructed using the RAxML program. The sub-prune shows the FormCO2OR lineage. The text labels for each protein representative in the family is color coded by whether the Mo or W atom is coordinated by a Cys or a Sec residue. The scale bar refers to the number of amino acid substitutions per site.
201
Fig. S6.8
MAFFT alignment of the iron-sulfur cluster binding motif of select representatives from each
CISM family analyzed in this study. Consensus sequences are highlighted in colors.
202
Fig. S6.9
MAFFT alignment of the distal pyranopterin binding motif of select representatives from each
CISM family analyzed in this study. Consensus sequences are highlighted in colors
203 APPENDIX 1
MLS10 AND MLTeJB MEDIA AND TRACE ELEMENTS SOLUTIONS
The recipes for the media used to cultivate Bacillus selenitireducens MLS10 (Switzer Blum et al.
1998) and B. beveridgei MLTeJB (Baesman et al. 2009) are described in the respective publications reporting their isolation. The medium for both MLS10 and MLTeJB were prepared as follows:
-1 (NH4)2SO4 0.1 g L
-1 MgSO4 * 7H2O 0.025 g L
-1 K2HPO4 0.15 g L
-1 KH2PO4 0.08 g L
NaCl 40 g L-1*
-1 Na2CO3 10.6 g L
-1 NaHCO3 4.2 g L
Yeast extract 0.2 g L-1
SL-10 trace elements solution 5 mL L-1
* The amount of NaCl added to the medium for the cultivation of MLTeJB was 50 g L -1
The pH of the medium was adjusted to 9.8 and was then degassed with 100% N2.
The SL-10 trace element solution was prepared as recommended by the American Type Culture
Collection (ATCC) for the 2100 Bacillus haloalkaliphile medium, with the exception of the
-1 addition of 9 mg L Na2WO4 * 2H2O. The SL-10 trace elements solution was prepared as follows:
204 HCl (37.0 %) 10 mL L-1
-1 FeCl2 1.5 g L
-1 ZnCl2 70 mg L
-1 MnCl2 * 4H2O 100 mg L
-1 H3BO3 6.0 mg L
-1 CoCl2 * 6H2O 190 mg L
-1 CuCl2 * 2H2O 2.0 mg L
-1 NiCl2 * 6H2O 24 mg L
-1 Na2MoO4 * 2H2O 36 mg L
-1 Na2WO4 * 2H2O 9.0 mg L
The FeCl2 was first dissolved in 10 mL of 37% HCl before being added to the trace elements solution.
205 APPENDIX 2
PCP-1 MEDIUM AND TRACE ELEMENTS AND VITAMINS SOLUTIONS
The medium used to cultivate Desulfitobacterium hafniense PCP-1 was the medium recommended by ATCC, though the trace elements and vitamins solutions utitlized were recommended by the German Collection of Microorganisms and Cell Cultures (DSMZ). It was prepared as follows:
-1 NH4Cl 1.0 g L
-1 MgCl2 * 6H2O 0.1 g L
-1 FeCl2 2.0 mg L
-1 CaCl2 * 2H2O 73 mg L
-1 KH2PO4 0.27 g L
-1 K2HPO4 0.35 g L
* -1 NaHCO3 2.6 g L
Yeast extract 1.0 g L-1
SL-10 trace elements solution 1.0 mL L-1
Vitamin solution 10 mL L-1
The FeCl2 was dissolved in 100 L 37% HCl prior to being added to the medium. The pH of the medium was adjusted to 7.2 and subsequently degassed with an 80% N2 / 20% CO2 atmosphere using a heated copper column to remove traces of O2.
The SL-10 trace elements solution was prepared as described in Appendix 1.
The vitamins solution was prepared as described for Sulfurosprillum barnesii SeS-3 (formerly
Geospirillum barnesii) (Stolz et al. 1997) and contained:
206 Biotin 2.0 mg L-1
Folic acid 2.0 mg L-1
Pyridoxine-HCl 10.0 mg L-1
Thiamine-HCl 5.0 mg L-1
Riboflavin 5.0 mg L-1
Nicotinic acid 5.0 mg L-1
D-Ca-pantothenate 5.0 mg L-1 p-Aminobenzoic acid 5.0 mg L-1
Thioctic (lipoic) acid 5.0 mg L-1
-1 Vitamin B12 (cyanocobalamin) 0.1 mg L
207 APPENDIX 3
SA-01 MEDIA AND TRACE ELEMENTS AND VITAMINS SOLUTIONS
Thermus scotoductus SA-01 was cultivated on two media. The medium is the Castenholtz TYE medium recommended for cultivation of SA-01 by ATCC and contains:
Nitrilotriacetic acid 0.1 g L-1
-1 CaSO4 * 2H2O 0.06 g L
-1 MgSO4 * 7H2O 0.1 g L
NaCl 0.008 g L-1
-1 KNO3 0.105 g L
-1 NaNO3 0.7 g L
-1 Na2HPO4 0.11 g L
Tryptone 1.0 g L-1
Yeast extract 1.0 g L-1
-1 0.3% FeCl3 solution 1 mL L
Nitsch’s trace elements solution 1 mL L-1
The pH of the medium was adjusted to 7.2 and subsequently degassed with an 80% N2 / 20%
CO2 atmosphere using a heated copper column to remove traces of O2.
Nitsch’s trace elements solution was prepared as recommended by ATCC and contained:
208 -1 H2SO4 0.5 mL L
-1 MnSO4 * H2O 2.2 g L
-1 ZnSO4 * 7H2O 0.5 g L
-1 H3BO3 0.5 g L
-1 CuSO4 * 5H2O 0.016 g L
-1 Na2MoO4 * 2H2O 0.025 g L
-1 CoCl2 * 6H2O 0.046 g L
The defined medium used to isolate SA-01 (Kieft et al. 1999) was prepared as follows:
-1 KH2PO4 0.42 g L
-1 K2PO4 0.22 g L
-1 NH4Cl 0.2 g L
KCl 0.38 g L-1
NaCl 0.36 g L-1
-1 CaCl2 * 2H2O 0.04 g L
-1 MgSO4 * 7H2O 0.1 g L
-1 NaHCO3 1.8 g L
-1 Na2CO3 0.5 g L
10X Wolfe’s trace elements solution 10 mL L-1
10X Wolfe’s vitamins solution 15 mL L-1
The pH of the medium was adjusted to 6.8, and was degassed with an 80% N2 / 20% CO2 atmosphere using a heated copper column to remove traces of O2. Both the 10X Wolfe’s trace elements and vitamins solutions were prepared as described previously (Kieft et al. 1999). The
10X Wolfe’s trace elements solution was prepared as follows:
209 Nitrilotriacetic acid 2.14 g L-1
-1 MnCl2 * 4H2O 0.1 g L
-1 FeSO4 * 7H2O 0.3 g L
-1 CoCl2 * 6H2O 0.17 g L
-1 ZnSO4 * 7H2O 0.2 g L
-1 CuCl2 * 2H2O 0.03 g L
-1 KAl(SO4)2 * 12H2O 0.005 g L
-1 H3BO3 0.005 g L
-1 Na2MoO4 * 2H2O 0.09 g L
-1 NiSO4 * 6H2O 0.11 g L
-1 Na2WO4 * 2H2O 0.02 g L
The 10X Wolfe’s vitamins solution was prepared as described below:
Biotin 2.0 mg L-1
Folic acid 2.0 mg L-1
Pyridoxine HCl 10.0 mg L-1
Riboflavin 5.0 mg L-1
Thiamine 5.0 mg L-1
Nicotinic acid 5.0 mg L-1
Pantothenic acid 5.0 mg L-1
-1 Vitamin B12 (cyanocobalamin) 0.1 mg L p-Aminobenzoic acid 5.0 mg L-1
Thioctic (lipoic) acid 5.0 mg
210 REFERENCES
Abin CA, Hollibaugh JT. Desulfuribacillus stibiiarsenatis sp. nov., an obligately anaerobic, dissimilatory antimonate- and arsenate-reducing bacterium isolated from anoxic sediments, and emended description of the genus Desulfuribacillus. Int J Syst Evol Microbiol 2017;67:1011–7.
Adamson H, Simonov AN, Kierzek M et al. Electrochemical evidence that pyranopterin redox chemistry controls the catalysis of YedY, a mononuclear Mo enzyme. Proc Natl Acad Sci U S A 2015;112:14506–11.
Afkar E, Lisak J, Saltikov C et al. The respiratory arsenate reductase from Bacillus selenitireducens strain MLS10. FEMS Microbiol Lett 2003;226:107–12.
Afshar S, Johnson E, de Vries S et al. Properties of a thermostable nitrate reductase from the hyperthermophilic archaeon Pyrobaculum aerophilum. J Bacteriol 2001;183:5491–5.
Altekar G, Dwarkadas S, Huelsenbeck JP et al. Parallel Metropolis coupled Markov chain Monte Carlo for Bayesian phylogenetic inference. Bioinforma Oxf Engl 2004;20:407–15.
Altschul SF, Madden TL, Schäffer AA et al. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 1997;25:3389–402.
Amouroux D, Liss PS, Tessier E et al. Role of oceans as biogenic sources of selenium. Earth Planet Sci Lett 2001;189:277–83.
Andreesen JR, Ljungdahl LG. Formate dehydrogenase of Clostridium thermoaceticum: incorporation of selenium-75, and the effects of selenite, molybdate, and tungstate on the enzyme. J Bacteriol 1973;116:867–73.
Andreesen JR, Ljungdahl LG. Nicotinamide adenine dinucleotide phosphate-dependent formate dehydrogenase from Clostridium thermoaceticum: purification and properties. J Bacteriol 1974;120:6–14.
Andreesen JR, Wagner M, Sonntag D et al. Various functions of selenols and thiols in anaerobic gram-positive, amino acids-utilizing bacteria. BioFactors Oxf Engl 1999a;10:263–70.
Andren AW, Klein DH, Talmi Y. Selenium in coal-fired steam plant emissions. Environ Sci Technol 1975;9:856–8.
Andres J, Bertin PN. The microbial genomics of arsenic. FEMS Microbiol Rev 2016;40:299– 322.
Araiso Y, Palioura S, Ishitani R et al. Structural insights into RNA-dependent eukaryal and archaeal selenocysteine formation. Nucleic Acids Res 2008;36:1187–99.
Araiso Y, Sherrer RL, Ishitani R et al. Structure of a tRNA-dependent kinase essential for selenocysteine decoding. Proc Natl Acad Sci U S A 2009;106:16215–20.
211 Baesman SM, Bullen TD, Dewald J et al. Formation of tellurium nanocrystals during anaerobic growth of bacteria that use Te oxyanions as respiratory electron acceptors. Appl Environ Microbiol 2007;73:2135–43.
Baesman SM, Stolz JF, Kulp TR et al. Enrichment and isolation of Bacillus beveridgei sp. nov., a facultative anaerobic haloalkaliphile from Mono Lake, California, that respires oxyanions of tellurium, selenium, and arsenic. Extrem Life Extreme Cond 2009;13:695– 705.
Baier F, Copp JN, Tokuriki N. Evolution of enzyme superfamilies: Comprehensive exploration of sequence-function relationships. Biochemistry 2016;55:6375–88.
Baier F, Tokuriki N. Connectivity between catalytic landscapes of the metallo-β-lactamase superfamily. J Mol Biol 2014;426:2442–56.
Baines SB, Fisher NS, Doblin MA et al. Uptake of dissolved organic selenides by marine phytoplankton. Limnol Oceanogr 2001;46:1936–44.
Baron C, Böck A. The length of the aminoacyl-acceptor stem of the selenocysteine-specific tRNA(Sec) of Escherichia coli is the determinant for binding to elongation factors SELB or Tu. J Biol Chem 1991;266:20375–9.
Baron C, Heider J, Böck A. Interaction of translation factor SELB with the formate dehydrogenase H selenopolypeptide mRNA. Proc Natl Acad Sci U S A 1993;90:4181–5.
Baron C, Westhof E, Böck A et al. Solution structure of selenocysteine-inserting tRNA(Sec) from Escherichia coli. Comparison with canonical tRNA(Ser). J Mol Biol 1993;231:274– 92.
Barr I, Guo F. Pyridine Hemochromagen assay for determining the concentration of heme in purified protein solutions. Bio-Protoc 2015;5.
Bartos P, Maciaszek A, Rosinska A et al. Transformation of a wobble 2-thiouridine to 2- selenouridine via S-geranyl-2-thiouridine as a possible cellular pathway. Bioorganic Chem 2014;56:49–53.
Bébien M, Kirsch J, Méjean V et al. Involvement of a putative molybdenum enzyme in the reduction of selenate by Escherichia coli. Microbiol Read Engl 2002;148:3865–72.
Berg BL, Baron C, Stewart V. Nitrate-inducible formate dehydrogenase in Escherichia coli K- 12. II. Evidence that a mRNA stem-loop structure is essential for decoding opal (UGA) as selenocysteine. J Biol Chem 1991;266:22386–91.
Beribisky AV, Tavares TJ, Amborski AN et al. The three-dimensional structure of the Moorella thermoacetica selenocysteine insertion sequence RNA hairpin and its interaction with the elongation factor SelB. RNA N Y N 2007;13:1948–56.
212 Berry MJ, Banu L, Chen YY et al. Recognition of UGA as a selenocysteine codon in type I deiodinase requires sequences in the 3’ untranslated region. Nature 1991;353:273–6.
Berry MJ, Banu L, Larsen PR. Type I iodothyronine deiodinase is a selenocysteine-containing enzyme. Nature 1991;349:438–40.
Bertero MG, Rothery RA, Palak M et al. Insights into the respiratory electron transfer pathway from the structure of nitrate reductase A. Nat Struct Biol 2003;10:681–7.
Bilous PT, Cole ST, Anderson WF et al. Nucleotide sequence of the dmsABC operon encoding the anaerobic dimethylsulphoxide reductase of Escherichia coli. Mol Microbiol 1988;2:785–95.
Blasco F, Dos Santos JP, Magalon A et al. NarJ is a specific chaperone required for molybdenum cofactor assembly in nitrate reductase A of Escherichia coli. Mol Microbiol 1998;28:435–47.
Blum JS, Stolz JF, Oren A et al. Selenihalanaerobacter shriftii gen. nov., sp. nov., a halophilic anaerobe from Dead Sea sediments that respires selenate. Arch Microbiol 2001;175:208– 19.
Boratyn GM, Schäffer AA, Agarwala R et al. Domain enhanced lookup time accelerated BLAST. Biol Direct 2012;7:12.
Börner G, Karrasch M, Thauer RK. Molybdopterin adenine dinucleotide and molybdopterin hypoxanthine dinucleotide in formylmethanofuran dehydrogenase from Methanobacterium thermoautotrophicum (Marburg). FEBS Lett 1991;290:31–4.
Boschi-Muller S, Muller S, Van Dorsselaer A et al. Substituting selenocysteine for active site cysteine 149 of phosphorylating glyceraldehyde 3-phosphate dehydrogenase reveals a peroxidase activity. FEBS Lett 1998;439:241–5.
Boyington JC, Gladyshev VN, Khangulov SV et al. Crystal structure of formate dehydrogenase H: catalysis involving Mo, molybdopterin, selenocysteine, and an Fe4S4 cluster. Science 1997;275:1305–8.
Brenner W. Zuchtungsversuche einiger im Schlamm lebenden Bakterien Auf Salzhaltigen Nährboden. Jahrb Für Wiss Bot 1916;57:95.
Bruggeman C, Maes A, Vancluysen J et al. Selenite reduction in Boom clay: Effect of FeS2, clay minerals and dissolved organic matter. Environ Pollut 2005;137:209–21.
Bruggeman C, Maes A, Vancluysen J. The interaction of dissolved Boom Clay and Gorleben humic substances with selenium oxyanions (selenite and selenate). Appl Geochem 2007;22:1371–9.
Bujdoš M, Muľová A, Kubová J et al. Selenium fractionation and speciation in rocks, soils, waters and plants in polluted surface mine environment. Environ Geol 2005;47:353–60.
213 Bult CJ, White O, Olsen GJ et al. Complete genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science 1996;273:1058–73.
Burton GA, Giddings TH, DeBrine P et al. High incidence of selenite-resistant bacteria from a site polluted with selenium. Appl Environ Microbiol 1987;53:185–8.
Cammack R, Weiner JH. Electron paramagnetic resonance spectroscopic characterization of dimethyl sulfoxide reductase of Escherichia coli. Biochemistry 1990;29:8410–6.
Capella-Gutiérrez S, Silla-Martínez JM, Gabaldón T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinforma Oxf Engl 2009;25:1972–3.
Carlson BA, Xu X-M, Kryukov GV et al. Identification and characterization of phosphoseryl- tRNA[Ser]Sec kinase. Proc Natl Acad Sci 2004;101:12848–53.
Casi G, Roelfes G, Hilvert D. Selenoglutaredoxin as a glutathione peroxidase mimic. Chembiochem Eur J Chem Biol 2008;9:1623–31.
Castellano S. On the unique function of selenocysteine - insights from the evolution of selenoproteins. Biochim Biophys Acta 2009;1790:1463–70.
Castellano S, Andrés AM, Bosch E et al. Low exchangeability of selenocysteine, the 21st amino acid, in vertebrate proteins. Mol Biol Evol 2009;26:2031–40.
Castellano S, Gladyshev VN, Guigó R et al. SelenoDB 1.0 : a database of selenoprotein genes, proteins and SECIS elements. Nucleic Acids Res 2008;36:D332-338.
Castellano S, Lobanov AV, Chapple C et al. Diversity and functional plasticity of eukaryotic selenoproteins: identification and characterization of the SelJ family. Proc Natl Acad Sci U S A 2005;102:16188–93. de Castro E, Sigrist CJA, Gattiker A et al. ScanProsite: detection of PROSITE signature matches and ProRule-associated functional and structural residues in proteins. Nucleic Acids Res 2006;34:W362-365.
Chambers I, Frampton J, Goldfarb P et al. The structure of the mouse glutathione peroxidase gene: the selenocysteine in the active site is encoded by the “termination” codon, TGA. EMBO J 1986;5:1221–7.
Chang Y, Zhang J, Qu J et al. The behavior of dissolved inorganic selenium in the Changjiang Estuary. J Mar Syst 2016;154:110–21.
Chen CS, Stadtman TC. Selenium-containing tRNAs from Clostridium sticklandii: cochromatography of one species with L-prolyl-tRNA. Proc Natl Acad Sci U S A 1980;77:1403–7.
Chen P, Crain PF, Näsvall SJ et al. A “gain of function” mutation in a protein mediates production of novel modified nucleosides. EMBO J 2005;24:1842–51.
214 Chiba S, Itoh Y, Sekine S et al. Structural basis for the major role of O-phosphoseryl-tRNA kinase in the UGA-specific encoding of selenocysteine. Mol Cell 2010;39:410–20.
Ching WM. Characterization of selenium-containing tRNAGlu from Clostridium sticklandii. Arch Biochem Biophys 1986;244:137–46.
Ching WM, Alzner-DeWeerd B, Stadtman TC. A selenium-containing nucleoside at the first position of the anticodon in seleno-tRNAGlu from Clostridium sticklandii. Proc Natl Acad Sci 1985;82:347–50.
Ching WM, Stadtman TC. Selenium-containing tRNAGlu from Clostridium sticklandii: correlation of aminoacylation with selenium content. Proc Natl Acad Sci U S A 1982;79:374–7.
Ching WM, Wittwer AJ, Tsai L et al. Distribution of two selenonucleosides among the selenium-containing tRNAs from Methanococcus vannielii. Proc Natl Acad Sci U S A 1984;81:57–60.
Chocat P, Esaki N, Tanizawa K et al. Purification and characterization of selenocysteine beta- lyase from Citrobacter freundii. J Bacteriol 1985;163:669–76.
Clark DR. Selenium accumulation in mammals exposed to contaminated California irrigation drainwater. Sci Total Environ 1987;66:147–68.
Cnossen I, Sanz‐Forcada J, Favata F et al. Habitat of early life: Solar X-ray and UV radiation at Earth’s surface 4–3.5 billion years ago. J Geophys Res Planets 2007;112, DOI: 10.1029/2006JE002784.
Collins MD, Lawson PA, Willems A et al. The phylogeny of the genus Clostridium: proposal of five new genera and eleven new species combinations. Int J Syst Bacteriol 1994;44:812– 26.
Cone JE, Del Río RM, Davis JN et al. Chemical characterization of the selenoprotein component of clostridial glycine reductase: identification of selenocysteine as the organoselenium moiety. Proc Natl Acad Sci U S A 1976;73:2659–63.
Conrads T, Hemann C, George GN et al. The active site of arsenite oxidase from Alcaligenes faecalis. J Am Chem Soc 2002;124:11276–7.
Copeland PR, Fletcher JE, Carlson BA et al. A novel RNA binding protein, SBP2, is required for the translation of mammalian selenoprotein mRNAs. EMBO J 2000;19:306–14.
Copley SD. Moonlighting is mainstream: paradigm adjustment required. BioEssays News Rev Mol Cell Dev Biol 2012;34:578–88.
Costa KC, Lie TJ, Xia Q et al. VhuD facilitates electron flow from H2 or formate to heterodisulfide reductase in Methanococcus maripaludis. J Bacteriol 2013;195:5160–5.
215 Cozen AE, Weirauch MT, Pollard KS et al. Transcriptional map of respiratory versatility in the hyperthermophilic crenarchaeon Pyrobaculum aerophilum. J Bacteriol 2009;191:782–94.
Cupp-Sutton KA, Ashby MT. Biological chemistry of hydrogen selenide. Antioxid Basel Switz 2016;5, DOI: 10.3390/antiox5040042.
Cutter GA. Selenium in reducing waters. Science 1982;217:829–31.
Cutter GA. The estuarine behaviour of selenium in San Francisco Bay. Estuar Coast Shelf Sci 1989a;28:13–34.
Cutter GA. Freshwater systems. Occurrence and Distribution of Selenium. CRC Press Boca Raton, 1989b.
Cutter GA. Kinetic controls on metalloid speciation in seawater. Mar Chem 1992;40:65–80.
Cutter GA, Bruland KW. The marine biogeochemistry of selenium: A re‐evaluation. Limnol Oceanogr 1984;29:1179–92.
Cutter GA, Church TM. Selenium in western Atlantic precipitation. Nature 1986;322:720.
Cutter GA, Cutter LS. Behavior of dissolved antimony, arsenic, and selenium in the Atlantic Ocean. Mar Chem 1995;49:295–306.
Cutter GA, Cutter LS. Sources and cycling of selenium in the western and equatorial Atlantic Ocean. Deep Sea Res Part II Top Stud Oceanogr 2001;48:2917–31.
Cutter GA, Cutter LS. Selenium biogeochemistry in the San Francisco Bay estuary: changes in water column behavior. Estuar Coast Shelf Sci 2004;61:463–76.
Cutter GA, San Diego-McGlone MLC. Temporal variability of selenium fluxes in San Francisco Bay. Sci Total Environ 1990;97–98:235–50.
Czjzek M, Dos Santos JP, Pommier J et al. Crystal structure of oxidized trimethylamine N-oxide reductase from Shewanella massilia at 2.5 Å resolution. J Mol Biol 1998;284:435–47.
Darriba D, Taboada GL, Doallo R et al. ProtTest 3: fast selection of best-fit models of protein evolution. Bioinforma Oxf Engl 2011;27:1164–5.
Dias JM, Than ME, Humm A et al. Crystal structure of the first dissimilatory nitrate reductase at 1.9 Å solved by MAD methods. Struct Lond Engl 1993 1999;7:65–79.
Dietrich W, Klimmek O. The function of methyl-menaquinone-6 and polysulfide reductase membrane anchor (PsrC) in polysulfide respiration of Wolinella succinogenes. Eur J Biochem 2002;269:1086–95.
Dilworth GL. Properties of the selenium-containing moiety of nicotinic acid hydroxylase from Clostridium barkeri. Arch Biochem Biophys 1982;219:30–8.
216 Diplock AT, Caygill CPJ, Jeffery EH et al. The nature of the acid-volatile selenium in the liver of the male rat. Biochem J 1973;134:283–93.
Dobbek H. Structural aspects of mononuclear Mo/W-enzymes. Coord Chem Rev 2011;255:1104–16.
Dobbek H, Gremer L, Kiefersauer R et al. Catalysis at a dinuclear [CuSMo(O)OH] cluster in a CO dehydrogenase resolved at 1.1-Å resolution. Proc Natl Acad Sci 2002;99:15971–6.
Dobbek H, Gremer L, Meyer O et al. Crystal structure and mechanism of CO dehydrogenase, a molybdo iron-sulfur flavoprotein containing S-selanylcysteine. Proc Natl Acad Sci U S A 1999;96:8884–9.
Doblin MA, Baines SB, Cutter LS et al. Sources and biogeochemical cycling of particulate selenium in the San Francisco Bay estuary. Estuar Coast Shelf Sci 2006;67:681–94.
Dowdle PR, Oremland RS. Microbial oxidation of elemental selenium in soil slurries and bacterial cultures. Environ Sci Technol 1998;32:3749–55.
Dridge EJ, Watts CA, Jepson BJN et al. Investigation of the redox centres of periplasmic selenate reductase from Thauera selenatis by EPR spectroscopy. Biochem J 2007;408:19–28.
Dumelin CE, Chen Y, Leconte AM et al. Discovery and biological characterization of geranylated RNA in bacteria. Nat Chem Biol 2012;8:913–9.
Dürre P, Andreesen JR. Pathway of carbon dioxide reduction to acetate without a net energy requirement in Clostridium purinolyticum. FEMS Microbiol Lett 1982;15:51–6.
Dürre P, Andreesen JR. Purine and glycine metabolism by purinolytic clostridia. J Bacteriol 1983;154:192–9.
Duval S, Ducluzeau A-L, Nitschke W et al. Enzyme phylogenies as markers for the oxidation state of the environment: the case of respiratory arsenate reductase and related enzymes. BMC Evol Biol 2008;8:206.
Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 2004;32:1792–7.
Edwardson CF, Hollibaugh JT. Metatranscriptomic analysis of prokaryotic communities active in sulfur and arsenic cycling in Mono Lake, California, USA. ISME J 2017;11:2195–208.
Ehrenreich A, Forchhammer K, Tormay P et al. Selenoprotein synthesis in E. coli. Eur J Biochem 1992;206:767–73.
Ellis PJ, Conrads T, Hille R et al. Crystal structure of the 100 kDa arsenite oxidase from Alcaligenes faecalis in two crystal forms at 1.64 Å and 2.03 Å. Struct Lond Engl 1993 2001;9:125–32.
217 Engelhardt H, Forchhammer K, Müller S et al. Structure of selenocysteine synthase from Escherichia coli and location of tRNA in the seryl-tRNA(Sec)-enzyme complex. Mol Microbiol 1992;6:3461–7.
Eswayah AS, Smith TJ, Gardiner PHE. Microbial transformations of selenium species of relevance to bioremediation. Appl Environ Microbiol 2016;82:4848–59.
Fahey RC. Glutathione analogs in prokaryotes. Biochim Biophys Acta 2013;1830:3182–98.
Fernández-Martínez A, Charlet L. Selenium environmental cycling and bioavailability: a structural chemist point of view. Rev Environ Sci Biotechnol 2009;8:81–110.
Ferry JG. Formate dehydrogenase. FEMS Microbiol Rev 1990;7:377–82.
Findlay AJ. Microbial impact on polysulfide dynamics in the environment. FEMS Microbiol Lett 2016a;363, DOI: 10.1093/femsle/fnw103.
Fischer N, Neumann P, Bock LV et al. The pathway to GTPase activation of elongation factor SelB on the ribosome. Nature 2016;540:80–5.
Fischer WW, Hemp J, Valentine JS. How did life survive Earth’s great oxygenation? Curr Opin Chem Biol 2016;31:166–78.
Fisher JC, Hollibaugh JT. Selenate-dependent anaerobic arsenite oxidation by a bacterium from Mono Lake, California. Appl Environ Microbiol 2008;74:2588–94.
Flohe L, Günzler WA, Schock HH. Glutathione peroxidase: a selenoenzyme. FEBS Lett 1973;32:132–4.
Floor GH, Román-Ross G. Selenium in volcanic environments: A review. Appl Geochem 2012;27:517–31.
Forchhammer K, Böck A. Selenocysteine synthase from Escherichia coli. Analysis of the reaction sequence. J Biol Chem 1991;266:6324–8.
Forchhammer K, Leinfelder W, Böck A. Identification of a novel translation factor necessary for the incorporation of selenocysteine into protein. Nature 1989;342:453–6.
Forchhammer K, Leinfelder W, Boesmiller K et al. Selenocysteine synthase from Escherichia coli. Nucleotide sequence of the gene (selA) and purification of the protein. J Biol Chem 1991;266:6318–23.
Forchhammer K, Rücknagel KP, Böck A. Purification and biochemical characterization of SELB, a translation factor involved in selenoprotein synthesis. J Biol Chem 1990;265:9346–50.
Förster C, Ott G, Forchhammer K et al. Interaction of a selenocysteine-incorporating tRNA with elongation factor Tu from E.coli. Nucleic Acids Res 1990;18:487–91.
218 Fourmy D, Guittet E, Yoshizawa S. Structure of prokaryotic SECIS mRNA hairpin and its interaction with elongation factor SelB. J Mol Biol 2002;324:137–50.
Franke KW, Rice TD, Johnson AG et al. Report on a preliminary field survey of the so-called “alkali disease” of livestock. Rep Prelim Field Surv -Call Alkali Dis Livest 1934.
Fujita M, Ike M, Nishimoto S et al. Isolation and characterization of a novel selenate-reducing bacterium, Bacillus sp. SF-1. J Ferment Bioeng 1997;83:517–22.
Ghosh W, Dam B. Biochemistry and molecular biology of lithotrophic sulfur oxidation by taxonomically and ecologically diverse bacteria and archaea. FEMS Microbiol Rev 2009;33:999–1043.
Gladyshev VN, Jeang KT, Stadtman TC. Selenocysteine, identified as the penultimate C- terminal residue in human T-cell thioredoxin reductase, corresponds to TGA in the human placental gene. Proc Natl Acad Sci U S A 1996;93:6146–51.
Gladyshev VN, Khangulov SV, Stadtman TC. Nicotinic acid hydroxylase from Clostridium barkeri: electron paramagnetic resonance studies show that selenium is coordinated with molybdenum in the catalytically active selenium-dependent enzyme. Proc Natl Acad Sci U S A 1994;91:232–6.
Gladyshev VN, Khangulov SV, Stadtman TC. Properties of the selenium- and molybdenum- containing nicotinic acid hydroxylase from Clostridium barkeri. Biochemistry 1996;35:212–23.
Glass RS, Singh WP, Jung W et al. Monoselenophosphate: synthesis, characterization, and identity with the prokaryotic biological selenium donor, compound SePX. Biochemistry 1993;32:12555–9.
Goldschmidt VM. The principles of distribution of chemical elements in minerals and rocks. The seventh Hugo Müller Lecture, delivered before the Chemical Society on March 17th, 1937. J Chem Soc Resumed 1937:655–73.
Graentzdoerffer A, Rauh D, Pich A et al. Molecular and biochemical characterization of two tungsten- and selenium-containing formate dehydrogenases from Eubacterium acidaminophilum that are associated with components of an iron-only hydrogenase. Arch Microbiol 2003;179:116–30.
Graham DE, Kyrpides N, Anderson IJ et al. Genome of Methanocaldococcus (Methanococcus) jannaschii. Methods Enzymol 2001;330:40–123.
Grimaldi S, Schoepp-Cothenet B, Ceccaldi P et al. The prokaryotic Mo/W-bisPGD enzymes family: a catalytic workhorse in bioenergetic. Biochim Biophys Acta 2013;1827:1048–85.
Gromer S, Johansson L, Bauer H et al. Active sites of thioredoxin reductases: why selenoproteins? Proc Natl Acad Sci U S A 2003;100:12618–23.
219 Gromer S, Schirmer RH, Becker K. The 58 kDa mouse selenoprotein is a BCNU-sensitive thioredoxin reductase. FEBS Lett 1997;412:318–20.
Guimarães MJ, Peterson D, Vicari A et al. Identification of a novel selD homolog from eukaryotes, bacteria, and archaea: Is there an autoregulatory mechanism in selenocysteine metabolism? Proc Natl Acad Sci 1996;93:15086–91.
Gursinsky T, Jäger J, Andreesen JR et al. A selDABC cluster for selenocysteine incorporation in Eubacterium acidaminophilum. Arch Microbiol 2000;174:200–12.
Gustafsson JP, Johnsson L. The association between selenium and humic substances in forested ecosystems—laboratory evidence. Appl Organomet Chem 1994;8:141–7.
Güven K, Mutlu MB, Çırpan C et al. Isolation and identification of selenite reducing archaea from Tuz (salt) Lake In Turkey. J Basic Microbiol 2013;53:397–401.
Haft DH, Self WT. Orphan SelD proteins and selenium-dependent molybdenum hydroxylases. Biol Direct 2008;3:4.
Haja DK, Wu C-H, Poole FL et al. Characterization of thiosulfate reductase from Pyrobaculum aerophilum heterologously produced in Pyrococcus furiosus. Extrem Life Extreme Cond 2019, DOI: 10.1007/s00792-019-01112-9.
Halboth S, Klein A. Methanococcus voltae harbors four gene clusters potentially encoding two [NiFe] and two [NiFeSe] hydrogenases, each of the cofactor F420-reducing or F420-non- reducing types. Mol Gen Genet MGG 1992;233:217–24.
Harel A, Häggblom MM, Falkowski PG et al. Evolution of prokaryotic respiratory molybdoenzymes and the frequency of their genomic co-occurrence. FEMS Microbiol Ecol 2016;92, DOI: 10.1093/femsec/fiw187.
Härtig C, Lohmayer R, Kolb S et al. Chemolithotrophic growth of the aerobic hyperthermophilic bacterium Thermocrinis ruber OC 14/7/2 on monothioarsenate and arsenite. FEMS Microbiol Ecol 2014;90:747–60.
Hedderich R, Klimmek O, Kröger A et al. Anaerobic respiration with elemental sulfur and with disulfides. FEMS Microbiol Rev 1998;22:353–81.
Heider J, Baron C, Böck A. Coding from a distance: dissection of the mRNA determinants required for the incorporation of selenocysteine into protein. EMBO J 1992;11:3759–66.
Heinzinger NK, Fujimoto SY, Clark MA et al. Sequence analysis of the phs operon in Salmonella typhimurium and the contribution of thiosulfate reduction to anaerobic energy metabolism. J Bacteriol 1995;177:2813–20.
Hensel M, Hinsley AP, Nikolaus T et al. The genetic basis of tetrathionate respiration in Salmonella typhimurium. Mol Microbiol 1999;32:275–87.
220 Herbel MJ, Blum JS, Oremland RS et al. Reduction of elemental selenium to selenide: Experiments with anoxic sediments and bacteria that respire Se-oxyanions. Geomicrobiol J 2003;20:587–602.
Herbel MJ, Johnson TM, Oremland RS et al. Fractionation of selenium isotopes during bacterial respiratory reduction of selenium oxyanions. Geochim Cosmochim Acta 2000;64:3701–9.
Hernandez-Maldonado J, Sanchez-Sedillo B, Stoneburner B et al. The genetic basis of anoxygenic photosynthetic arsenite oxidation. Environ Microbiol 2017;19:130–41.
Hilgenfeld R, Böck A, Wilting R. Structural model for the selenocysteine-specific elongation factor SelB. Biochimie 1996;78:971–8.
Hill KE, Lloyd RS, Burk RF. Conserved nucleotide sequences in the open reading frame and 3’ untranslated region of selenoprotein P mRNA. Proc Natl Acad Sci U S A 1993;90:537– 41.
Hille R, Hall J, Basu P. The mononuclear molybdenum enzymes. Chem Rev 2014;114:3963– 4038.
Hoang DT, Chernomor O, von Haeseler A et al. UFBoot2: Improving the Ultrafast Bootstrap Approximation. Mol Biol Evol 2018;35:518–22.
Hoffman JL, McConnell KP. The presence of 4-selenouridine in Escherichia coli tRNA. Biochim Biophys Acta 1974;366:109–13.
Hohn MJ, Palioura S, Su D et al. Genetic analysis of selenocysteine biosynthesis in the archaeon Methanococcus maripaludis. Mol Microbiol 2011;81:249–58.
Hoke KR, Cobb N, Armstrong FA et al. Electrochemical studies of arsenite oxidase: an unusual example of a highly cooperative two-electron molybdenum center. Biochemistry 2004;43:1667–74.
Hormann K, Andreesen JR. Reductive cleavage of sarcosine and betaine by Eubacterium acidaminophilum via enzyme systems different from glycine reductase. Arch Microbiol 1989;153:50–9.
Hu M, Yang Y, Martin J-M et al. Preferential uptake of Se(IV) over Se(VI) and the production of dissolved organic Se by marine phytoplankton. Mar Environ Res 1997;44:225–31.
Hu Y, Faham S, Roy R et al. Formaldehyde ferredoxin oxidoreductase from Pyrococcus furiosus: the 1.85 Å resolution crystal structure and its mechanistic implications. J Mol Biol 1999;286:899–914.
Huang R, Hippauf F, Rohrbeck D et al. Enzyme functional evolution through improved catalysis of ancestrally nonpreferred substrates. Proc Natl Acad Sci U S A 2012;109:2966–71.
221 Huber R, Sacher M, Vollmann A et al. Respiration of arsenate and selenate by hyperthermophilic archaea. Syst Appl Microbiol 2000;23:305–14.
Huelsenbeck JP, Crandall KA. Phylogeny estimation and hypothesis testing using maximum likelihood. Annu Rev Ecol Syst 1997;28:437–66.
Huelsenbeck JP, Ronquist F, Nielsen R et al. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 2001;294:2310–4.
Hüttenhofer A, Heider J, Böck A. Interaction of the Escherichia coli fdhF mRNA hairpin promoting selenocysteine incorporation with the ribosome. Nucleic Acids Res 1996;24:3903–10.
Hüttenhofer A, Westhof E, Böck A. Solution structure of mRNA hairpins promoting selenocysteine incorporation in Escherichia coli and their base-specific interaction with special elongation factor SELB. RNA N Y N 1996;2:354–66.
Ichiki H, Tanaka Y, Mochizuki K et al. Purification, characterization, and genetic analysis of Cu-containing dissimilatory nitrite reductase from a denitrifying halophilic archaeon, Haloarcula marismortui. J Bacteriol 2001;183:4149–56.
Iida Y, Yamaguchi T, Tanaka T et al. Solubility of selenium at high ionic strength under anoxic conditions. J Nucl Sci Technol Tokyo 2010;47:431–8.
Imhoff D, Andreesen JR. Nicotinic acid hydroxylase from Clostridium barkeri: Selenium- dependent formation of active enzyme. FEMS Microbiol Lett 1979;5:155–8.
Ioudovitch A, Steinberg SV. Structural compensation in an archaeal selenocysteine transfer RNA. J Mol Biol 1999;290:365–71.
Itoh Y, Bröcker MJ, Sekine S et al. Decameric SelA•tRNA(Sec) ring structure reveals mechanism of bacterial selenocysteine formation. Science 2013;340:75–8.
Itoh Y, Bröcker MJ, Sekine S et al. Dimer-dimer interaction of the bacterial selenocysteine synthase SelA promotes functional active-site formation and catalytic specificity. J Mol Biol 2014;426:1723–35.
Itoh Y, Sekine S, Matsumoto E et al. Structure of selenophosphate synthetase essential for selenium incorporation into proteins and RNAs. J Mol Biol 2009;385:1456–69.
Itoh Y, Sekine S-I, Yokoyama S. Crystal structure of the full-length bacterial selenocysteine- specific elongation factor SelB. Nucleic Acids Res 2015;43:9028–38.
Jacob F. Evolution and tinkering. Science 1977;196:1161–6.
Jäger G, Chen P, Björk GR. Transfer RNA bound to MnmH protein Is enriched with geranylated tRNA--A possible intermediate in its selenation? PloS One 2016;11:e0153488.
222 Jeffery CJ. An introduction to protein moonlighting. Biochem Soc Trans 2014;42:1679–83.
Jensen RA. Enzyme recruitment in evolution of new function. Annu Rev Microbiol 1976;30:409– 25.
Jiang L, Ni J, Liu Q. Evolution of selenoproteins in the metazoan. BMC Genomics 2012;13:446.
Jones JB, Dilworth GL, Stadtman TC. Occurrence of selenocysteine in the selenium-dependent formate dehydrogenase of Methanococcus vannielii. Arch Biochem Biophys 1979a;195:255–60.
Jones JB, Stadtman TC. Selenium-dependent and selenium-independent formate dehydrogenases of Methanococcus vannielii. Separation of the two forms and characterization of the purified selenium-independent form. J Biol Chem 1981;256:656–63.
Jørgensen BB. A thiosulfate shunt in the sulfur cycle of marine sediments. Science 1990;249:152–4.
Jørgensen BB, Bak F. Pathways and microbiology of thiosulfate transformations and sulfate reduction in a marine sediment (Kattegat, Denmark). Appl Environ Microbiol 1991;57:847–56.
Jormakka M, Törnroth S, Byrne B et al. Molecular basis of proton motive force generation: structure of formate dehydrogenase-N. Science 2002;295:1863–8.
Kabisch UC, Gräntzdörffer A, Schierhorn A et al. Identification of d-proline reductase from Clostridium sticklandii as a selenoenzyme and indications for a catalytically active pyruvoyl group derived from a cysteine residue by cleavage of a proprotein. J Biol Chem 1999;274:8445–54.
Kaiser JT, Gromadski K, Rother M et al. Structural and functional investigation of a putative archaeal selenocysteine synthase. Biochemistry 2005;44:13315–27.
Kalyaanamoorthy S, Minh BQ, Wong TKF et al. ModelFinder: fast model selection for accurate phylogenetic estimates. Nat Methods 2017;14:587–9.
Kaster A-K, Moll J, Parey K et al. Coupling of ferredoxin and heterodisulfide reduction via electron bifurcation in hydrogenotrophic methanogenic archaea. Proc Natl Acad Sci U S A 2011;108:2981–6.
Katoh K, Rozewicki J, Yamada KD. MAFFT online service: multiple sequence alignment, interactive sequence choice and visualization. Brief Bioinform 2017, DOI: 10.1093/bib/bbx108.
Keeling PJ, Fast NM, McFadden GI. Evolutionary relationship between translation initiation factor eIF-2 and selenocysteine-specific elongation factor SELB: change of function in translation factors. J Mol Evol 1998;47:649–55.
223 Kessi J. Enzymic systems proposed to be involved in the dissimilatory reduction of selenite in the purple non-sulfur bacteria Rhodospirillum rubrum and Rhodobacter capsulatus. Microbiol Read Engl 2006;152:731–43.
Kessi J, Hanselmann KW. Similarities between the abiotic reduction of selenite with glutathione and the dissimilatory reaction mediated by Rhodospirillum rubrum and Escherichia coli. J Biol Chem 2004;279:50662–9.
Khangulov SV, Gladyshev VN, Dismukes GC et al. Selenium-containing formate dehydrogenase H from Escherichia coli: a molybdopterin enzyme that catalyzes formate oxidation without oxygen transfer. Biochemistry 1998;37:3518–28.
Kieft TL, Fredrickson JK, Onstott TC et al. Dissimilatory reduction of Fe(III) and other electron acceptors by a Thermus isolate. Appl Environ Microbiol 1999;65:1214–21.
Kim H-Y. The methionine sulfoxide reduction system: selenium utilization and methionine sulfoxide reductase enzymes and their functions. Antioxid Redox Signal 2013;19:958–69.
Kim H-Y, Fomenko DE, Yoon Y-E et al. Catalytic advantages provided by selenocysteine in methionine-S-sulfoxide reductases. Biochemistry 2006;45:13697–704.
Kim H-Y, Zhang Y, Lee BC et al. The selenoproteome of Clostridium sp. OhILAs: characterization of anaerobic bacterial selenoprotein methionine sulfoxide reductase A. Proteins 2009;74:1008–17.
Kim IY, Stadtman TC. Selenophosphate synthetase: detection in extracts of rat tissues by immunoblot assay and partial purification of the enzyme from the archaean Methanococcus vannielii. Proc Natl Acad Sci U S A 1995;92:7710–3.
Kim IY, Veres Z, Stadtman TC. Escherichia coli mutant SELD enzymes. The cysteine 17 residue is essential for selenophosphate formation from ATP and selenide. J Biol Chem 1992;267:19650–4.
Kim IY, Veres Z, Stadtman TC. Biochemical analysis of Escherichia coli selenophosphate synthetase mutants. Lysine 20 is essential for catalytic activity and cysteine 17/19 for 8- azido-ATP derivatization. J Biol Chem 1993;268:27020–5.
Kim M-J, Lee BC, Hwang KY et al. Selenium utilization in thioredoxin and catalytic advantage provided by selenocysteine. Biochem Biophys Res Commun 2015;461:648–52.
Kipp MA, Stüeken EE, Bekker A et al. Selenium isotopes record extensive marine suboxia during the Great Oxidation Event. Proc Natl Acad Sci 2017:201615867.
Klimmek O, Kreis V, Klein C et al. The function of the periplasmic Sud protein in polysulfide respiration of Wolinella succinogenes. Eur J Biochem 1998;253:263–9.
224 Klimmek O, Kröger A, Steudel R et al. Growth of Wolinella succinogenes with polysulphide as terminal acceptor of phosphorylative electron transport. Arch Microbiol 1991;155:177– 82.
Klonowska A, Heulin T, Vermeglio A. Selenite and tellurite reduction by Shewanella oneidensis. Appl Environ Microbiol 2005;71:5607–9.
Klug SJ, Hüttenhofer A, Kromayer M et al. In vitro and in vivo characterization of novel mRNA motifs that bind special elongation factor SelB. Proc Natl Acad Sci U S A 1997;94:6676– 81.
Knoll AH, Bergmann KD, Strauss JV. Life: the first two billion years. Philos Trans R Soc Lond B Biol Sci 2016;371, DOI: 10.1098/rstb.2015.0493.
Koch T, Dahl C. A novel bacterial sulfur oxidation pathway provides a new link between the cycles of organic and inorganic sulfur compounds. ISME J 2018;12:2479–91.
Krafft T, Bokranz M, Klimmek O et al. Cloning and nucleotide sequence of the psrA gene of Wolinella succinogenes polysulphide reductase. Eur J Biochem 1992;206:503–10.
Krafft T, Bowen A, Theis F et al. Cloning and sequencing of the genes encoding the periplasmic-cytochrome B-containing selenate reductase of Thauera selenatis. DNA Seq J DNA Seq Mapp 2000;10:365–77.
Krafft T, Macy JM. Purification and characterization of the respiratory arsenate reductase of Chrysiogenes arsenatis. Eur J Biochem 1998;255:647–53.
Kramer GF, Ames BN. Isolation and characterization of a selenium metabolism mutant of Salmonella typhimurium. J Bacteriol 1988;170:736–43.
Krol A. Evolutionarily different RNA motifs and RNA-protein complexes to achieve selenoprotein synthesis. Biochimie 2002;84:765–74.
Kromayer M, Neuhierl B, Friebel A et al. Genetic probing of the interaction between the translation factor SelB and its mRNA binding element in Escherichia coli. Mol Gen Genet MGG 1999;262:800–6.
Kromayer M, Wilting R, Tormay P et al. Domain structure of the prokaryotic selenocysteine- specific elongation factor SelB. J Mol Biol 1996;262:413–20.
Kryukov GV, Castellano S, Novoselov SV et al. Characterization of mammalian selenoproteomes. Science 2003;300:1439–43.
Kryukov GV, Gladyshev VN. The prokaryotic selenoproteome. EMBO Rep 2004;5:538–43.
Kryukov GV, Kryukov VM, Gladyshev VN. New mammalian selenocysteine-containing proteins identified with an algorithm that searches for selenocysteine insertion sequence elements. J Biol Chem 1999;274:33888–97.
225 Kryukov GV, Kumar RA, Koc A et al. Selenoprotein R is a zinc-containing stereo-specific methionine sulfoxide reductase. Proc Natl Acad Sci U S A 2002;99:4245–50.
Kulp TR, Hoeft SE, Asao M et al. Arsenic(III) fuels anoxygenic photosynthesis in hot spring biofilms from Mono Lake, California. Science 2008;321:967–70.
Kumar S, Stecher G, Tamura K. MEGA7: Molecular Evolutionary Genetics Analysis Version 7.0 for bigger datasets. Mol Biol Evol 2016;33:1870–4.
Kuroda M, Notaguchi E, Sato A et al. Characterization of Pseudomonas stutzeri NT-I capable of removing soluble selenium from the aqueous phase under aerobic conditions. J Biosci Bioeng 2011a;112:259–64.
Kuroda M, Yamashita M, Miwa E et al. Molecular cloning and characterization of the srdBCA operon, encoding the respiratory selenate reductase complex, from the selenate-reducing bacterium Bacillus selenatarsenatis SF-1. J Bacteriol 2011b;193:2141–8.
Labunskyy VM, Hatfield DL, Gladyshev VN. Selenoproteins: molecular pathways and physiological roles. Physiol Rev 2014;94:739–77.
Lacourciere GM, Mihara H, Kurihara T et al. Escherichia coli NifS-like proteins provide selenium in the pathway for the biosynthesis of selenophosphate. J Biol Chem 2000;275:23769–73.
Lacourciere GM, Stadtman TC. The NIFS protein can function as a selenide delivery protein in the biosynthesis of selenophosphate. J Biol Chem 1998;273:30921–6.
Lacourciere GM, Stadtman TC. Catalytic properties of selenophosphate synthetases: Comparison of the selenocysteine-containing enzyme from Haemophilus influenzae with the corresponding cysteine-containing enzyme from Escherichia coli. Proc Natl Acad Sci 1999;96:44–8.
Lane N, Allen JF, Martin W. How did LUCA make a living? Chemiosmosis in the origin of life. BioEssays 2010;32:271–80.
Le SQ, Gascuel O. An improved general amino acid replacement matrix. Mol Biol Evol 2008;25:1307–20.
Leaver JT, Richardson DJ, Butler CS. Enterobacter cloacae SLD1a-1 gains a selective advantage from selenate reduction when growing in nitrate-depleted anaerobic environments. J Ind Microbiol Biotechnol 2008;35:867–73.
Lebrun E, Brugna M, Baymann F et al. Arsenite oxidase, an ancient bioenergetic enzyme. Mol Biol Evol 2003;20:686–93.
Lebrun E, Santini JM, Brugna M et al. The Rieske protein: a case study on the pitfalls of multiple sequence alignments and phylogenetic reconstruction. Mol Biol Evol 2006;23:1180–91.
226 Lee BC, Lobanov AV, Marino SM et al. A 4-selenocysteine, 2-selenocysteine insertion sequence (SECIS) element methionine sulfoxide reductase from Metridium senile reveals a non- catalytic function of selenocysteines. J Biol Chem 2011;286:18747–55.
Leibundgut M, Frick C, Thanbichler M et al. Selenocysteine tRNA-specific elongation factor SelB is a structural chimaera of elongation and initiation factors. EMBO J 2005;24:11– 22.
Leinfelder W, Forchhammer K, Veprek B et al. In vitro synthesis of selenocysteinyl- tRNA(UCA) from seryl-tRNA(UCA): involvement and characterization of the selD gene product. Proc Natl Acad Sci U S A 1990;87:543–7.
Leinfelder W, Forchhammer K, Zinoni F et al. Escherichia coli genes whose products are involved in selenium metabolism. J Bacteriol 1988a;170:540–6.
Leinfelder W, Stadtman TC, Böck A. Occurrence in vivo of selenocysteyl-tRNA(SERUCA) in Escherichia coli. Effect of sel mutations. J Biol Chem 1989;264:9720–3.
Leinfelder W, Zehelein E, Mandrand-Berthelot MA et al. Gene for a novel tRNA species that accepts L-serine and cotranslationally inserts selenocysteine. Nature 1988b;331:723–5.
Lemly AD. Agriculture and wildlife: ecological implications of subsurface irrigation drainage. J Arid Environ 1994;28:85–94.
Lemly AD. Aquatic selenium pollution is a global environmental safety issue. Ecotoxicol Environ Saf 2004;59:44–56.
Lemly AD. Damage cost of the Dan River coal ash spill. Environ Pollut Barking Essex 1987 2015;197:55–61.
Lemly AD, Finger SE, Nelson MK. Sources and impacts of irrigation drainwater contaminants in arid wetlands. Environ Toxicol Chem 1993;12:2265–79.
Lescure A, Gautheret D, Carbon P et al. Novel selenoproteins identified in silico and in vivo by using a conserved RNA structural motif. J Biol Chem 1999;274:38147–54.
Letunic I, Bork P. Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Res 2016;44:W242-245.
Levine VE. Biochemical studies of selenium. Ann N Y Acad Sci 1915;26:385–94.
Levine VE. The reducing properties of microorganisms with special reference to selenium compounds. J Bacteriol 1925;10:217–63.
Li C, Reches M, Engelberg-Kulka H. The bulged nucleotide in the Escherichia coli minimal selenocysteine insertion sequence participates in interaction with SelB: a genetic approach. J Bacteriol 2000;182:6302–7.
227 Li D-B, Cheng Y-Y, Wu C et al. Selenite reduction by Shewanella oneidensis MR-1 is mediated by fumarate reductase in periplasm. Sci Rep 2014;4:3735.
Li Z, Liang D, Peng Q et al. Interaction between selenium and soil organic matter and its impact on soil selenium bioavailability: A review. Geoderma 2017;295:69–79.
Lidman F, Mörth C-M, Björkvald L et al. Selenium dynamics in boreal streams: The role of wetlands and changing groundwater tables. Environ Sci Technol 2011;45:2677–83.
Lin J, Peng T, Jiang L et al. Comparative genomics reveals new candidate genes involved in selenium metabolism in prokaryotes. Genome Biol Evol 2015;7:664–76.
Lipman JG, Waksman SA. The oxidation of selenium by a new group of autotrophic microorganisms. Science 1923;57:60. van Lis R, Nitschke W, Warelow TP et al. Heterologously expressed arsenite oxidase: a system to study biogenesis and structure/function relationships of the enzyme family. Biochim Biophys Acta 2012;1817:1701–8.
Liu Y, Zhou Z, Pan J et al. Comparative genomic inference suggests mixotrophic lifestyle for Thorarchaeota. ISME J 2018;12:1021–31.
Liu Z, Reches M, Engelberg-Kulka H. A sequence in the Escherichia coli fdhF “selenocysteine insertion sequence” (SECIS) operates in the absence of selenium. J Mol Biol 1999;294:1073–86.
Liu Z, Reches M, Groisman I et al. The nature of the minimal ‘selenocysteine insertion sequence’ (SECIS) in Escherichia coli. Nucleic Acids Res 1998;26:896–902.
Lobanov AV, Fomenko DE, Zhang Y et al. Evolutionary dynamics of eukaryotic selenoproteomes: large selenoproteomes may associate with aquatic life and small with terrestrial life. Genome Biol 2007;8:R198.
Lobanov AV, Hatfield DL, Gladyshev VN. Eukaryotic selenoproteins and selenoproteomes. Biochim Biophys Acta 2009;1790:1424–8.
Long JA, Large RR, Lee MSY et al. Severe selenium depletion in the Phanerozoic oceans as a factor in three global mass extinction events. Gondwana Res 2016;36:209–18.
Losi ME, Frankenberger WT. Reduction of selenium oxyanions by Enterobacter cloacae SLD1a-1: Isolation and growth of the bacterium and its expulsion of selenium particles. Appl Environ Microbiol 1997;63:3079–84.
Losi ME, Frankenberger WT. Microbial oxidation and solubilization of precipitated elemental selenium in soil. J Environ Qual Madison 1998;27:836.
228 Lowe EC, Bydder S, Hartshorne RS et al. Quinol-cytochrome c oxidoreductase and cytochrome c4 mediate electron transfer during selenate respiration in Thauera selenatis. J Biol Chem 2010;285:18433–42.
Lowry OH, Rosebrough NJ, Farr AL et al. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265–75.
Lucy JA, Diplock AT, Caygill CP. The intracellular distribution of acid-labile selenium in rat liver and the possible biological function of selenium and vitamin E. Biochem J 1970;119:40P.
Luoma SN, Johns C, Fisher NS et al. Determination of selenium bioavailability to a benthic bivalve from particulate and solute pathways. Environ Sci Technol 1992;26:485–91.
Macy JM, Michel TA, Kirsch DG. Selenate reduction by a Pseudomonas species: a new mode of anaerobic respiration. FEMS Microbiol Lett 1989;52:195–8.
Macy JM, Rech S, Auling G et al. Thauera selenatis gen. nov., sp. nov., a member of the beta subclass of Proteobacteria with a novel type of anaerobic respiration. Int J Syst Bacteriol 1993;43:135–42.
Maher MJ, Macy JM. Crystallization and preliminary X-ray analysis of the selenate reductase from Thauera selenatis. Acta Crystallogr D Biol Crystallogr 2002;58:706–8.
Maher MJ, Santini J, Pickering IJ et al. X-ray absorption spectroscopy of selenate reductase. Inorg Chem 2004;43:402–4.
Maiers DT, Wichlacz PL, Thompson DL et al. Selenate reduction by bacteria from a selenium- rich environment. Appl Environ Microbiol 1988;54:2591–3.
Maltman C, Donald LJ, Yurkov V. Two distinct periplasmic enzymes are responsible for tellurite/tellurate and selenite reduction by strain ER-Te-48 associated with the deep sea hydrothermal vent tube worms at the Juan de Fuca Ridge black smokers. Arch Microbiol 2017, DOI: 10.1007/s00203-017-1382-1.
Maltman C, Walter G, Yurkov V. A diverse community of metal(loid) oxide respiring bacteria is associated with tube worms in the vicinity of the Juan de Fuca Ridge black smoker field. PloS One 2016;11:e0149812.
Mander GJ, Pierik AJ, Huber H et al. Two distinct heterodisulfide reductase-like enzymes in the sulfate-reducing archaeon Archaeoglobus profundus. Eur J Biochem 2004;271:1106–16.
Manville IA. The selenium problem and its relationship to public health. Am J Public Health Nations Health 1939;29:709–19.
Manzine LR, Serrão VHB, da Rocha e Lima LMT et al. Assembly stoichiometry of bacterial selenocysteine synthase and SelC (tRNAsec). FEBS Lett 2013;587:906–11.
229 Mariotti M, Lobanov AV, Guigo R et al. SECISearch3 and Seblastian: new tools for prediction of SECIS elements and selenoproteins. Nucleic Acids Res 2013;41:e149.
Mariotti M, Lobanov AV, Manta B et al. Lokiarchaeota marks the transition between the archaeal and eukaryotic selenocysteine encoding systems. Mol Biol Evol 2016;33:2441– 53.
Mariotti M, Salinas G, Gabaldón T et al. Utilization of selenocysteine in early-branching fungal phyla. Nat Microbiol 2019;4:759–65.
Mariotti M, Santesmasses D, Capella-Gutierrez S et al. Evolution of selenophosphate synthetases: emergence and relocation of function through independent duplications and recurrent subfunctionalization. Genome Res 2015;25:1256–67.
Markowitz VM, Chen I-MA, Chu K et al. IMG/M: the integrated metagenome data management and comparative analysis system. Nucleic Acids Res 2012;40:D123–9.
Markowitz VM, Korzeniewski F, Palaniappan K et al. The integrated microbial genomes (IMG) system. Nucleic Acids Res 2006;34:D344–8.
Martens D a., Suarez D l. Selenium speciation of marine shales, alluvial soils, and evaporation basin soils of California. J Environ Qual 1997;26:424–32.
Martin W, Baross J, Kelley D et al. Hydrothermal vents and the origin of life. Nat Rev Microbiol 2008;6:805–14.
Mason RP, Soerensen AL, DiMento BP et al. The global marine selenium cycle: Insights from measurements and modeling. Glob Biogeochem Cycles 2018;32:1720–37.
Masscheleyn PH, Delaune RD, Patrick WH Jr. Transformations of selenium as affected by sediment oxidation–reduction potential and pH. Environ Sci Technol 1990;24:91–6.
Masscheleyn PH, Delaune RD, Patrick WH Jr. Biogeochernical behavior of selenium in anoxic soils and sediments: An equilibrium thermodynamics approach. J Environ Sci Health Part Environ Sci Eng Toxicol 1991;26:555–73.
McDevitt CA, Hugenholtz P, Hanson GR et al. Molecular analysis of dimethyl sulphide dehydrogenase from Rhodovulum sulfidophilum: its place in the dimethyl sulphoxide reductase family of microbial molybdopterin-containing enzymes. Mol Microbiol 2002;44:1575–87.
McEwan AG, Ridge JP, McDevitt CA et al. The DMSO reductase family of microbial molybdenum enzymes: Molecular properties and role in the dissimilatory reduction of toxic elements. Geomicrobiol J 2002;19:3–21.
Measures CI, Burton JD. Behaviour and speciation of dissolved selenium in estuarine waters. Nature 1978;273:293.
230 Measures CI, Burton JD. The vertical distribution and oxidation states of dissolved selenium in the northeast Atlantic Ocean and their relationship to biological processes. Earth Planet Sci Lett 1980;46:385–96.
Méjean V, Iobbi-Nivol C, Lepelletier M et al. TMAO anaerobic respiration in Escherichia coli: involvement of the tor operon. Mol Microbiol 1994;11:1169–79.
Meyer M, Granderath K, Andreesen JR. Purification and characterization of protein PB of betaine reductase and its relationship to the corresponding proteins glycine reductase and sarcosine reductase from Eubacterium acidaminophilum. Eur J Biochem 1995;234:184– 91.
Mihara H, Maeda M, Fujii T et al. A nifS-like gene, csdB, encodes an Escherichia coli counterpart of mammalian selenocysteine lyase. Gene cloning, purification, characterization and preliminary x-ray crystallographic studies. J Biol Chem 1999;274:14768–72.
Miller MA, Pfeiffer W, Schwartz T. Creating the CIPRES Science Gateway for inference of large phylogenetic trees. 2010 Gateway Computing Environments Workshop (GCE). 2010, 1–8.
Mitchell K, Mansoor SZ, Mason PRD et al. Geological evolution of the marine selenium cycle: Insights from the bulk shale δ82/76Se record and isotope mass balance modeling. Earth Planet Sci Lett 2016;441:178–87.
Mohamed MF, Hollfelder F. Efficient, crosswise catalytic promiscuity among enzymes that catalyze phosphoryl transfer. Biochim Biophys Acta 2013;1834:417–24.
Monahan-Pendergast M, Przybylek M, Lindblad M et al. Theoretical predictions of arsenic and selenium species under atmospheric conditions. Atmos Environ 2008;42:2349–57.
Mosher BW, Duce RA. A global atmospheric selenium budget. J Geophys Res Atmospheres 1987;92:13289–98.
Mosher BW, Duce RA, Prospero JM et al. Atmospheric selenium: Geographical distribution and ocean to atmosphere flux in the Pacific. J Geophys Res Atmospheres 1987;92:13277–87.
Mukherjee S, Seshadri R, Varghese NJ et al. 1,003 reference genomes of bacterial and archaeal isolates expand coverage of the tree of life. Nat Biotechnol 2017;35:676–83.
Mukund S, Adams MW. The novel tungsten-iron-sulfur protein of the hyperthermophilic archaebacterium, Pyrococcus furiosus, is an aldehyde ferredoxin oxidoreductase. Evidence for its participation in a unique glycolytic pathway. J Biol Chem 1991;266:14208–16.
Mukund S, Adams MW. Characterization of a novel tungsten-containing formaldehyde ferredoxin oxidoreductase from the hyperthermophilic archaeon, Thermococcus litoralis. A role for tungsten in peptide catabolism. J Biol Chem 1993;268:13592–600.
231 Mukund S, Adams MW. Glyceraldehyde-3-phosphate ferredoxin oxidoreductase, a novel tungsten-containing enzyme with a potential glycolytic role in the hyperthermophilic archaeon Pyrococcus furiosus. J Biol Chem 1995;270:8389–92.
Muth OH, Oldfield JE, Remmert LF et al. Effects of selenium and vitamin E on white muscle disease. Science 1958;128:1090.
Nakagawa T, Iino T, Suzuki K et al. Ferrimonas futtsuensis sp. nov. and Ferrimonas kyonanensis sp. nov., selenate-reducing bacteria belonging to the Gammaproteobacteria isolated from Tokyo Bay. Int J Syst Evol Microbiol 2006;56:2639–45.
Nancharaiah YV, Lens PNL. Ecology and biotechnology of selenium-respiring bacteria. Microbiol Mol Biol Rev MMBR 2015;79:61–80.
Narasingarao P, Häggblom MM. Sedimenticola selenatireducens, gen. nov., sp. nov., an anaerobic selenate-respiring bacterium isolated from estuarine sediment. Syst Appl Microbiol 2006;29:382–8.
Narasingarao P, Häggblom MM. Identification of anaerobic selenate-respiring bacteria from aquatic sediments. Appl Environ Microbiol 2007a;73:3519–27.
Narasingarao P, Häggblom MM. Pelobacter seleniigenes sp. nov., a selenate-respiring bacterium. Int J Syst Evol Microbiol 2007b;57:1937–42.
Newman DK, Ahmann D, Morel FMM. A brief review of microbial arsenate respiration. Geomicrobiol J 1998;15:255–68.
Ngaki MN, Louie GV, Philippe RN et al. Evolution of the chalcone-isomerase fold from fatty- acid binding to stereospecific catalysis. Nature 2012;485:530–3.
Nguyen L-T, Schmidt HA, von Haeseler A et al. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol 2015;32:268– 74.
Nickerson WJ, Falcone G. Enzymatic reduction of selenite. J Bacteriol 1963;85:763–71.
Niggemyer A, Spring S, Stackebrandt E et al. Isolation and characterization of a novel As(V)- reducing bacterium: implications for arsenic mobilization and the genus Desulfitobacterium. Appl Environ Microbiol 2001;67:5568–80.
Niks D, Duvvuru J, Escalona M et al. Spectroscopic and kinetic properties of the molybdenum- containing, NAD+-dependent formate dehydrogenase from Ralstonia eutropha. J Biol Chem 2016;291:1162–74.
Nitschke W, Russell MJ. Hydrothermal focusing of chemical and chemiosmotic energy, supported by delivery of catalytic Fe, Ni, Mo/W, Co, S and Se, forced life to emerge. J Mol Evol 2009;69:481–96.
232 Nitschke W, Russell MJ. Beating the acetyl coenzyme A-pathway to the origin of life. Philos Trans R Soc Lond B Biol Sci 2013;368:20120258.
Noinaj N, Wattanasak R, Lee D-Y et al. Structural insights into the catalytic mechanism of Escherichia coli selenophosphate synthetase. J Bacteriol 2012;194:499–508.
Nonaka H, Keresztes G, Shinoda Y et al. Complete genome sequence of the dehalorespiring bacterium Desulfitobacterium hafniense Y51 and comparison with Dehalococcoides ethenogenes 195. J Bacteriol 2006;188:2262–74.
Nriagu JO. A global assessment of natural sources of atmospheric trace metals. Nature 1989;338:47–9.
Nriagu JO, Pacyna JM. Quantitative assessment of worldwide contamination of air, water and soils by trace metals. Nature 1988;333:134–9.
Nuttall KL, Allen FS. Kinetics of the reaction between hydrogen selenide ion and oxygen. Inorganica Chim Acta 1984;91:243–6.
Ogasawara Y, Lacourciere G, Stadtman TC. Formation of a selenium-substituted rhodanese by reaction with selenite and glutathione: possible role of a protein perselenide in a selenium delivery system. Proc Natl Acad Sci U S A 2001;98:9494–8.
Ogasawara Y, Lacourciere GM, Ishii K et al. Characterization of potential selenium-binding proteins in the selenophosphate synthetase system. Proc Natl Acad Sci U S A 2005;102:1012–6.
Oh JI, Bowien B. Structural analysis of the fds operon encoding the NAD+-linked formate dehydrogenase of Ralstonia eutropha. J Biol Chem 1998;273:26349–60.
Ohlendorf HM, Kilness AW, Simmons JL et al. Selenium toxicosis in wild aquatic birds. J Toxicol Environ Health 1988;24:67–92.
Oram LL, Strawn DG, Marcus MA et al. Macro- and microscale investigation of selenium speciation in Blackfoot river, Idaho sediments. Environ Sci Technol 2008;42:6830–6.
Oram LL, Strawn DG, Morra MJ et al. Selenium biogeochemical cycling and fluxes in the hyporheic zone of a mining-impacted stream. Environ Sci Technol 2010;44:4176–83.
Ordoñez OF, Rasuk MC, Soria MN et al. Haloarchaea from the Andean Puna: Biological role in the energy metabolism of arsenic. Microb Ecol 2018;76:695–705.
Oremland RS, Dowdle PR, Hoeft S et al. Bacterial dissimilatory reduction of arsenate and sulfate in meromictic Mono Lake, California. Geochim Cosmochim Acta 2000;64:3073– 84.
233 Oremland RS, Blum JS, Culbertson CW et al. Isolation, growth, and metabolism of an obligately anaerobic, selenate-respiring bacterium, Strain SES-3. Appl Environ Microbiol 1994;60:3011–9.
Oremland RS, Herbel MJ, Blum JS et al. Structural and spectral features of selenium nanospheres produced by Se-respiring bacteria. Appl Environ Microbiol 2004;70:52.
Oremland RS, Hollibaugh JT, Maest AS et al. Selenate reduction to elemental selenium by anaerobic bacteria in sediments and culture: biogeochemical significance of a novel, sulfate-independent respiration. Appl Environ Microbiol 1989;55:2333–43.
Oremland RS, Saltikov CW, Wolfe-Simon F et al. Arsenic in the evolution of Earth and extraterrestrial ecosystems. Geomicrobiol J 2009;26:522–36.
Oremland RS, Steinberg NA, Presser TS et al. In situ bacterial selenate reduction in the agricultural drainage systems of western Nevada. Appl Environ Microbiol 1991;57:615– 7.
Ose T, Soler N, Rasubala L et al. Structural basis for dynamic interdomain movement and RNA recognition of the selenocysteine-specific elongation factor SelB. Struct Lond Engl 1993 2007;15:577–86.
Pabis A, Duarte F, Kamerlin SCL. Promiscuity in the enzymatic catalysis of phosphate and sulfate transfer. Biochemistry 2016;55:3061–81.
Paleskava A, Konevega AL, Rodnina MV. Thermodynamics of the GTP-GDP-operated conformational switch of selenocysteine-specific translation factor SelB. J Biol Chem 2012;287:27906–12.
Park M-O, Mizutani T, Jones PR. Glyceraldehyde-3-phosphate ferredoxin oxidoreductase from Methanococcus maripaludis. J Bacteriol 2007;189:7281–9.
Parkin A, Goldet G, Cavazza C et al. The difference a Se makes? Oxygen-tolerant hydrogen production by the [NiFeSe]-hydrogenase from Desulfomicrobium baculatum. J Am Chem Soc 2008;130:13410–6.
Payne NC, Geissler A, Button A et al. Comparison of the redox chemistry of sulfur- and selenium-containing analogs of uracil. Free Radic Biol Med 2017;104:249–61.
Peng T, Lin J, Xu Y-Z et al. Comparative genomics reveals new evolutionary and ecological patterns of selenium utilization in bacteria. ISME J 2016;10:2048–59.
Pierson DE, Campbell A. Cloning and nucleotide sequence of bisC, the structural gene for biotin sulfoxide reductase in Escherichia coli. J Bacteriol 1990;172:2194–8.
Pignol D, Adriano JM, Fontecilla-Camps JC et al. Crystallization and preliminary X-ray analysis of the periplasmic nitrate reductase (NapA-NapB complex) from Rhodobacter sphaeroides f. sp. denitrificans. Acta Crystallogr D Biol Crystallogr 2001;57:1900–2.
234 Pinsent J. The need for selenite and molybdate in the formation of formic dehydrogenase by members of the coli-aerogenes group of bacteria. Biochem J 1954;57:10–6.
Poehlein A, Bengelsdorf FR, Schiel-Bengelsdorf B et al. Draft genome sequence of purine- degrading Gottschalkia purinilyticum (formerly Clostridium purinilyticum) WA1 (DSM 1384). Genome Announc 2015;3, DOI: 10.1128/genomeA.01088-15.
Poehlein A, Yutin N, Daniel R et al. Erratum: Proposal for the reclassification of obligately purine-fermenting bacteria Clostridium acidurici (Barker 1938) and Clostridium purinilyticum (Dürre et al. 1981) as Gottschalkia acidurici gen. nov. comb. nov. and Gottschalkia purinilytica comb. nov. and of Eubacterium angustum (Beuscher and Andreesen 1985) as Andreesenia angusta gen. nov. comb. nov. in the family Gottschalkiaceae fam. nov. Int J Syst Evol Microbiol 2017;67:4287–8.
Pogge von Strandmann PAE, Stüeken EE, Elliott T et al. Selenium isotope evidence for progressive oxidation of the Neoproterozoic biosphere. Nat Commun 2015;6:10157.
Politino M, Tsai L, Veres Z et al. Biosynthesis of selenium-modified tRNAs in Methanococcus vannielii. Proc Natl Acad Sci U S A 1990;87:6345–8.
Pond SLK, Frost SDW, Muse SV. HyPhy: hypothesis testing using phylogenies. Bioinforma Oxf Engl 2005;21:676–9.
Prasada Rao YS, Cherayil JD. Number and proportion of selenonucleosides in the transfer RNA of Escherichia coli. Life Sci 1974;14:2051–9.
Prat L, Maillard J, Grimaud R et al. Physiological adaptation of Desulfitobacterium hafniense strain TCE1 to tetrachloroethene respiration. Appl Environ Microbiol 2011;77:3853–9.
Prat L, Maillard J, Rohrbach-Brandt E et al. An unusual tandem-domain rhodanese harbouring two active sites identified in Desulfitobacterium hafniense. FEBS J 2012;279:2754–67.
Presser TS, Ohlendorf HM. Biogeochemical cycling of selenium in the San Joaquin Valley, California, USA. Environ Manage 1987;11:805–21.
Prisner T, Lyubenova S, Atabay Y et al. Multifrequency cw-EPR investigation of the catalytic molybdenum cofactor of polysulfide reductase from Wolinella succinogenes. J Biol Inorg Chem JBIC Publ Soc Biol Inorg Chem 2003;8:419–26.
Pupko T, Bell RE, Mayrose I et al. Rate4Site: an algorithmic tool for the identification of functional regions in proteins by surface mapping of evolutionary determinants within their homologues. Bioinforma Oxf Engl 2002;18 Suppl 1:S71-77.
Raaijmakers H, Macieira S, Dias JM et al. Gene sequence and the 1.8 Å crystal structure of the tungsten-containing formate dehydrogenase from Desulfovibrio gigas. Struct Lond Engl 1993 2002;10:1261–72.
235 Rambaut A, Drummond AJ, Xie D et al. Posterior summarisation in Bayesian phylogenetics using Tracer 1.7. Syst Biol 2018, DOI: 10.1093/sysbio/syy032.
Ramírez-Arcos S, Fernández-Herrero LA, Berenguer J. A thermophilic nitrate reductase is responsible for the strain specific anaerobic growth of Thermus thermophilus HB8. Biochim Biophys Acta 1998;1396:215–27.
Ramos AR, Grein F, Oliveira GP et al. The FlxABCD-HdrABC proteins correspond to a novel NADH dehydrogenase/heterodisulfide reductase widespread in anaerobic bacteria and involved in ethanol metabolism in Desulfovibrio vulgaris Hildenborough. Environ Microbiol 2015;17:2288–305.
Rauschenbach I, Narasingarao P, Häggblom MM. Desulfurispirillum indicum sp. nov., a selenate- and selenite-respiring bacterium isolated from an estuarine canal. Int J Syst Evol Microbiol 2011;61:654–8.
Rauschenbach I, Posternak V, Cantarella P et al. Seleniivibrio woodruffii gen. nov., sp. nov., a selenate- and arsenate-respiring bacterium in the Deferribacteraceae. Int J Syst Evol Microbiol 2013;63:3659–65.
Reher M, Gebhard S, Schönheit P. Glyceraldehyde-3-phosphate ferredoxin oxidoreductase (GAPOR) and nonphosphorylating glyceraldehyde-3-phosphate dehydrogenase (GAPN), key enzymes of the respective modified Embden-Meyerhof pathways in the hyperthermophilic crenarchaeota Pyrobaculum aerophilum and Aeropyrum pernix. FEMS Microbiol Lett 2007;273:196–205.
Reich HJ, Hondal RJ. Why nature chose selenium. ACS Chem Biol 2016;11:821–41.
Ridley H, Watts CA, Richardson DJ et al. Resolution of distinct membrane-bound enzymes from Enterobacter cloacae SLD1a-1 that are responsible for selective reduction of nitrate and selenate oxyanions. Appl Environ Microbiol 2006;72:5173–80.
Ringquist S, Schneider D, Gibson T et al. Recognition of the mRNA selenocysteine insertion sequence by the specialized translational elongation factor SELB. Genes Dev 1994;8:376–85.
Rocher C, Lalanne JL, Chaudière J. Purification and properties of a recombinant sulfur analog of murine selenium-glutathione peroxidase. Eur J Biochem 1992;205:955–60.
Romagné F, Santesmasses D, White L et al. SelenoDB 2.0: annotation of selenoprotein genes in animals and their genetic diversity in humans. Nucleic Acids Res 2014;42:D437-443.
Romero H, Zhang Y, Gladyshev VN et al. Evolution of selenium utilization traits. Genome Biol 2005;6:R66.
Ronquist F, Teslenko M, van der Mark P et al. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol 2012;61:539–42.
236 Rosenfeld CE, Kenyon JA, James BR et al. Selenium (IV,VI) reduction and tolerance by fungi in an oxic environment. Geobiology 2017;15:441–52.
Ross HB. An atmospheric selenium budget for the region 30° N to 90° N. Tellus B 1985;37 B:78–90.
Rother M, Krzycki JA. Selenocysteine, pyrrolysine, and the unique energy metabolism of methanogenic Archaea. Archaea 2010;2010, DOI: 10.1155/2010/453642.
Rother M, Mathes I, Lottspeich F et al. Inactivation of the selB gene in Methanococcus maripaludis: effect on synthesis of selenoproteins and their sulfur-containing homologs. J Bacteriol 2003;185:107–14.
Rother M, Quitzke V. Selenoprotein synthesis and regulation in Archaea. Biochim Biophys Acta Gen Subj 2018, DOI: 10.1016/j.bbagen.2018.04.008.
Rother M, Resch A, Gardner WL et al. Heterologous expression of archaeal selenoprotein genes directed by the SECIS element located in the 3’ non-translated region. Mol Microbiol 2001;40:900–8.
Rother M, Wilting R, Commans S et al. Identification and characterisation of the selenocysteine- specific translation factor SelB from the archaeon Methanococcus jannaschii. J Mol Biol 2000;299:351–8.
Rothery RA, Stein B, Solomonson M et al. Pyranopterin conformation defines the function of molybdenum and tungsten enzymes. Proc Natl Acad Sci U S A 2012;109:14773–8.
Rothery RA, Workun GJ, Weiner JH. The prokaryotic complex iron-sulfur molybdoenzyme family. Biochim Biophys Acta 2008;1778:1897–929.
Rotruck JT, Pope AL, Ganther HE et al. Selenium: biochemical role as a component of glutathione peroxidase. Science 1973;179:588–90.
Rouxel O, Fouquet Y, Ludden JN. Subsurface processes at the lucky strike hydrothermal field, Mid-Atlantic ridge: evidence from sulfur, selenium, and iron isotopes Geochim Cosmochim Acta 2004;68:2295–311.
Saltikov CW, Newman DK. Genetic identification of a respiratory arsenate reductase. Proc Natl Acad Sci 2003;100:10983–8.
Sambrook J, Russell DW. SDS-polyacrylamide gel electrophoresis of proteins. Cold Spring Harb Protoc 2006;2006:pdb.prot4540.
Santesmasses D, Mariotti M, Guigó R. Computational identification of the selenocysteine tRNA (tRNASec) in genomes. PLOS Comput Biol 2017;13:e1005383.
Sarathchandra SU, Watkinson JH. Oxidation of elemental selenium to selenite by Bacillus megaterium. Science 1981;211:600–1.
237 Sawers G. The hydrogenases and formate dehydrogenases of Escherichia coli. Antonie Van Leeuwenhoek 1994;66:57–88.
Sawers G, Heider J, Zehelein E et al. Expression and operon structure of the sel genes of Escherichia coli and identification of a third selenium-containing formate dehydrogenase isoenzyme. J Bacteriol 1991;173:4983–93.
Schindelin H, Kisker C, Hilton J et al. Crystal structure of DMSO reductase: Redox-linked changes in molybdopterin coordination. Science 1996;272:1615–21.
Schlekat CE, Dowdle PR, Lee B-G et al. Bioavailability of particle-associated Se to the bivalve Potamocorbula amurensis. Environ Sci Technol 2000;34:4504–10.
Schneider F, Löwe J, Huber R et al. Crystal structure of dimethyl sulfoxide reductase from Rhodobacter capsulatus at 1.88 Å resolution. J Mol Biol 1996;263:53–69.
Schneider HA. Selenium in nutrition. Science 1936:32.
Schoepp-Cothenet B, van Lis R, Atteia A et al. On the universal core of bioenergetics. Biochim Biophys Acta 2013;1827:79–93.
Schoepp-Cothenet B, van Lis R, Philippot P et al. The ineluctable requirement for the trans-iron elements molybdenum and/or tungsten in the origin of life. Sci Rep 2012;2, DOI: 10.1038/srep00263.
Schönheit P, Schäfer T. Metabolism of hyperthermophiles. World J Microbiol Biotechnol 1995;11:26–57.
Schräder T, Rienhöfer A, Andreesen JR. Selenium-containing xanthine dehydrogenase from Eubacterium barkeri. Eur J Biochem FEBS 1999;264:862–71.
Schröder I, Johnson E, de Vries S. Microbial ferric iron reductases. FEMS Microbiol Rev 2003;27:427–47.
Schröder I, Kröger A, Macy JM. Isolation of the sulphur reductase and reconstitution of the sulphur respiration of Wolinella succinogenes. Arch Microbiol 1988;149:572–9.
Schröder I, Rech S, Krafft T et al. Purification and characterization of the selenate reductase from Thauera selenatis. J Biol Chem 1997;272:23765–8.
Schwarz K, Foltz CM. Selenium as an integral part of factor 3 against dietary necrotic liver degeneration. J Am Chem Soc 1957;79:3292–3.
Schwarz K, Foltz CM. Factor 3 activity of selenium compounds. J Biol Chem 1958;233:245–51.
Seifter J, Ehrich WE. Thyroid adenomas in rats receiving selenium. Science 1946;103:762.
238 Self WT, Stadtman TC. Selenium-dependent metabolism of purines: A selenium-dependent purine hydroxylase and xanthine dehydrogenase were purified from Clostridium purinolyticum and characterized. Proc Natl Acad Sci U S A 2000;97:7208–13.
Self WT, Wolfe MD, Stadtman TC. Cofactor determination and spectroscopic characterization of the selenium-dependent purine hydroxylase from Clostridium purinolyticum. Biochemistry 2003;42:11382–90.
Selmer M, Su X-D. Crystal structure of an mRNA-binding fragment of Moorella thermoacetica elongation factor SelB. EMBO J 2002;21:4145–53.
Sforna MC, Philippot P, Somogyi A et al. Evidence for arsenic metabolism and cycling by microorganisms 2.7 billion years ago. Nat Geosci 2014;7:811–5.
Sharma VK, McDonald TJ, Sohn M et al. Biogeochemistry of selenium. A review. Environ Chem Lett 2014;13:49–58.
Shchedrina VA, Novoselov SV, Malinouski MY et al. Identification and characterization of a selenoprotein family containing a diselenide bond in a redox motif. Proc Natl Acad Sci U S A 2007;104:13919–24.
Sherrard JC, Hunter KA, Boyd PW. Selenium speciation in subantarctic and subtropical waters east of New Zealand: trends and temporal variations. Deep Sea Res Part Oceanogr Res Pap 2004;51:491–506.
Sherrer RL, Araiso Y, Aldag C et al. C-terminal domain of archaeal O-phosphoseryl-tRNA kinase displays large-scale motion to bind the 7-bp D-stem of archaeal tRNA(Sec). Nucleic Acids Res 2011;39:1034–41.
Sherrer RL, Ho JML, Söll D. Divergence of selenocysteine tRNA recognition by archaeal and eukaryotic O-phosphoseryl-tRNASec kinase. Nucleic Acids Res 2008;36:1871–80.
Sherrer RL, O’Donoghue P, Söll D. Characterization and evolutionary history of an archaeal kinase involved in selenocysteinyl-tRNA formation. Nucleic Acids Res 2008;36:1247–59.
Shrift A. A selenium cycle in nature? Nature 1964;201:1304–5.
Shrift A, Virupaksha TK. Biosynthesis of Se-methyl-selenocysteine from selenite in selenium- accumulating plants. Biochim Biophys Acta 1963;71:483–5.
Sierant M, Leszczynska G, Sadowska K et al. Escherichia coli tRNA 2-selenouridine synthase (SelU) converts S2U-RNA to Se2U-RNA via S-geranylated-intermediate. FEBS Lett 2018;592:2248–58.
Silva IR, Serrão VHB, Manzine LR et al. Formation of a ternary complex for selenocysteine biosynthesis in bacteria. J Biol Chem 2015;290:29178–88.
239 Sliwkowski MX, Stadtman TC. Selenoprotein A of the clostridial glycine reductase complex: purification and amino acid sequence of the selenocysteine-containing peptide. Proc Natl Acad Sci U S A 1988;85:368–71.
Slobodkina GB, Lebedinsky AV, Chernyh NA et al. Pyrobaculum ferrireducens sp. nov., a hyperthermophilic Fe(III)-, selenate- and arsenate-reducing crenarchaeon isolated from a hot spring. Int J Syst Evol Microbiol 2015;65:851–6.
Snider GW, Ruggles E, Khan N et al. Selenocysteine confers resistance to inactivation by oxidation in thioredoxin reductase: comparison of selenium and sulfur enzymes. Biochemistry 2013;52:5472–81.
Sochacka E, Bartos P, Kraszewska K et al. Desulfuration of 2-thiouridine with hydrogen peroxide in the physiological pH range 6.6-7.6 is pH-dependent and results in two distinct products. Bioorg Med Chem Lett 2013;23:5803–5.
Sochacka E, Kraszewska K, Sochacki M et al. The 2-thiouridine unit in the RNA strand is desulfured predominantly to 4-pyrimidinone nucleoside under in vitro oxidative stress conditions. Chem Commun Camb Engl 2011;47:4914–6.
Söhling B, Parther T, Rücknagel KP et al. A selenocysteine-containing peroxiredoxin from the strictly anaerobic organism Eubacterium acidaminophilum. Biol Chem 2001;382:979–86.
Song D, Li X, Cheng Y et al. Aerobic biogenesis of selenium nanoparticles by Enterobacter cloacae Z0206 as a consequence of fumarate reductase mediated selenite reduction. Sci Rep 2017;7:3239.
Sorgenfrei O, Linder D, Karas M et al. A novel very small subunit of a selenium containing [NiFe] hydrogenase of Methanococcus voltae is postranslationally processed by cleavage at a defined position. Eur J Biochem 1993;213:1355–8.
Sorgenfrei O, Müller S, Pfeiffer M et al. The [NiFe] hydrogenases of Methanococcus voltae: genes, enzymes and regulation. Arch Microbiol 1997;167:189–95.
Sousa FL, Thiergart T, Landan G et al. Early bioenergetic evolution. Philos Trans R Soc Lond B Biol Sci 2013;368:20130088.
Sparacino-Watkins C, Stolz JF, Basu P. Nitrate and periplasmic nitrate reductases. Chem Soc Rev 2014;43:676–706.
Spielman SJ, Kosakovsky Pond SL. Relative evolutionary rate inference in HyPhy with LEISR. PeerJ 2018;6, DOI: 10.7717/peerj.4339.
Stadtman T. Methanococcus vannielii selenium metabolism: purification and N-terminal amino acid sequences of a novel selenium-binding protein and selenocysteine lyase. IUBMB Life 2004;56:427–31.
240 Stadtman TC. Specific occurrence of selenium in enzymes and amino acid tRNAs. FASEB J Off Publ Fed Am Soc Exp Biol 1987;1:375–9.
Stadtman TC, Davis JN, Zehelein E et al. Biochemical and genetic analysis of Salmonella typhimurium and Escherichia coli mutants defective in specific incorporation of selenium into formate dehydrogenase and tRNAs. BioFactors Oxf Engl 1989;2:35–44.
Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinforma Oxf Engl 2014;30:1312–3.
Stock T, Rother M. Selenoproteins in Archaea and Gram-positive bacteria. Biochim Biophys Acta 2009;1790:1520–32.
Stock T, Selzer M, Rother M. In vivo requirement of selenophosphate for selenoprotein synthesis in archaea. Mol Microbiol 2010;75:149–60.
Stoffels L, Krehenbrink M, Berks BC et al. Thiosulfate reduction in Salmonella enterica is driven by the proton motive force. J Bacteriol 2012;194:475–85.
Stolz JF. Gaia and her microbiome. FEMS Microbiol Ecol 2017;93, DOI: 10.1093/femsec/fiw247.
Stolz JF, Basu P, Santini JM et al. Arsenic and selenium in microbial metabolism. Annu Rev Microbiol 2006;60:107–30.
Stolz JF, Ellis DJ, Blum JS et al. Sulfurospirillum barnesii sp. nov. and Sulfurospirillum arsenophilum sp. nov., new members of the Sulfurospirillum clade of the epsilon Proteobacteria. Int J Syst Bacteriol 1999;49 Pt 3:1177–80.
Stolz JF, Gugliuzza T, Switzer Blum J et al. Differential cytochrome content and reductase activity in Geospirillum barnesii strain SeS3. Arch Microbiol 1997;167:1–5.
Stolz JF, Oremland RS. Bacterial respiration of arsenic and selenium. FEMS Microbiol Rev 1999;23:615–27.
Stüeken EE. Selenium isotopes as a biogeochemical proxy in deep time. Rev Mineral Geochem 2017;82:657–82.
Stüeken EE, Buick R, Anbar AD. Selenium isotopes support free O2 in the latest Archean. Geology 2015;43:259–62.
Stüeken EE, Buick R, Bekker A et al. The evolution of the global selenium cycle: Secular trends in Se isotopes and abundances. Geochim Cosmochim Acta 2015a;162:109–25.
Stüeken EE, Foriel J, Buick R et al. Selenium isotope ratios, redox changes and biological productivity across the end-Permian mass extinction. Chem Geol 2015b;410:28–39.
241 Su D, Ojo TT, Söll D et al. Selenomodification of tRNA in Archaea requires a bipartite rhodanese enzyme. Febs Lett 2012;586:717.
Sun Y, Li T, Chen H et al. Selenium-containing 15-mer peptides with high glutathione peroxidase-like activity. J Biol Chem 2004;279:37235–40.
Suppmann S, Persson BC, Böck A. Dynamics and efficiency in vivo of UGA-directed selenocysteine insertion at the ribosome. EMBO J 1999;18:2284–93.
Switzer Blum J, Burns Bindi A, Buzzelli J et al. Bacillus arsenicoselenatis, sp. nov., and Bacillus selenitireducens, sp. nov.: two haloalkaliphiles from Mono Lake, California that respire oxyanions of selenium and arsenic. Arch Microbiol 1998;171:19–30.
Takahata M, Tamura T, Abe K et al. Selenite assimilation into formate dehydrogenase H depends on thioredoxin reductase in Escherichia coli. J Biochem (Tokyo) 2008;143:467– 73.
Takayanagi K, Wong GTF. Total selenium and selenium (IV) in the James River estuary and southern Chesapeake Bay. Estuar Coast Shelf Sci 1984;18:113–9.
Tanji K, Läuchli A, Meyer J. Selenium in the San Joaquin Valley. Environment 1986;28:6–39.
Tarze A, Dauplais M, Grigoras I et al. Extracellular production of hydrogen selenide accounts for thiol-assisted toxicity of selenite against Saccharomyces cerevisiae. J Biol Chem 2007;282:8759–67.
Teixeira M, Fauque G, Moura I et al. Nickel-[iron-sulfur]-selenium-containing hydrogenases from Desulfovibrio baculatus (DSM 1743). Redox centers and catalytic properties. Eur J Biochem 1987;167:47–58.
Thanbichler M, Böck A. The function of SECIS RNA in translational control of gene expression in Escherichia coli. EMBO J 2002;21:6925–34.
Thanbichler M, Bock A, Goody RS. Kinetics of the interaction of translation factor SelB from Escherichia coli with guanosine nucleotides and selenocysteine insertion sequence RNA. J Biol Chem 2000;275:20458–66.
Thauer RK, Jungermann K, Decker K. Energy conservation in chemotrophic anaerobic bacteria. Bacteriol Rev 1977;41:100–80.
Theisen J, Yee N. The molecular basis for selenate reduction in Citrobacter freundii. Geomicrobiol J 2014;31:875–83.
Thomas G, Potter L, Cole JA. The periplasmic nitrate reductase from Escherichia coli: a heterodimeric molybdoprotein with a double-arginine signal sequence and an unusual leader peptide cleavage site. FEMS Microbiol Lett 1999;174:167–71.
242 Thorell HD, Stenklo K, Karlsson J et al. A gene cluster for chlorate metabolism in Ideonella dechloratans. Appl Environ Microbiol 2003;69:5585–92.
Tobe R, Mihara H. Delivery of selenium to selenophosphate synthetase for selenoprotein biosynthesis. Biochim Biophys Acta BBA - Gen Subj 2018;1862:2433–40.
Tomei FA, Barton LL, Lemanski CL et al. Transformation of selenate and selenite to elemental selenium by Desulfovibrio desulfuricans. J Ind Microbiol 1995;14:329–36.
Tormay P, Sawers A, Böck A. Role of stoichiometry between mRNA, translation factor SelB and selenocysteyl-tRNA in selenoprotein synthesis. Mol Microbiol 1996;21:1253–9.
Tormay P, Wilting R, Lottspeich F et al. Bacterial selenocysteine synthase--structural and functional properties. Eur J Biochem 1998;254:655–61.
Trelease SF, Trelease HM. Selenium as a stimulating and possibly essential element for certain plants. Science 1938;87:70–1.
Turner DC, Stadtman TC. Purification of protein components of the clostridial glycine reductase system and characterization of protein A as a selenoprotein. Arch Biochem Biophys 1973;154:366–81.
Vavourakis CD, Ghai R, Rodriguez-Valera F et al. Metagenomic insights into the uncultured diversity and physiology of microbes in four hypersaline soda lake brines. Front Microbiol 2016;7:211.
Vendeland SC, Beilstein MA, Chen CL et al. Purification and properties of selenoprotein W from rat muscle. J Biol Chem 1993;268:17103–7.
Veres Z, Kim IY, Scholz TD et al. Selenophosphate synthetase. Enzyme properties and catalytic reaction. J Biol Chem 1994;269:10597–603.
Veres Z, Stadtman TC. A purified selenophosphate-dependent enzyme from Salmonella typhimurium catalyzes the replacement of sulfur in 2-thiouridine residues in tRNAs with selenium. Proc Natl Acad Sci U S A 1994;91:8092–6.
Veres Z, Tsai L, Politino M et al. In vitro incorporation of selenium into tRNAs of Salmonella typhimurium. Proc Natl Acad Sci U S A 1990;87:6341–4.
Veres Z, Tsai L, Scholz TD et al. Synthesis of 5-methylaminomethyl-2-selenouridine in tRNAs: 31P NMR studies show the labile selenium donor synthesized by the selD gene product contains selenium bonded to phosphorus. Proc Natl Acad Sci 1992a;89:2975–9.
Voordeckers K, Brown CA, Vanneste K et al. Reconstruction of ancestral metabolic enzymes reveals molecular mechanisms underlying evolutionary innovation through gene duplication. PLoS Biol 2012;10:e1001446.
243 Vorholt JA, Vaupel M, Thauer RK. A selenium-dependent and a selenium-independent formylmethanofuran dehydrogenase and their transcriptional regulation in the hyperthermophilic Methanopyrus kandleri. Mol Microbiol 1997;23:1033–42.
Wagener N, Pierik AJ, Ibdah A et al. The Mo-Se active site of nicotinate dehydrogenase. Proc Natl Acad Sci U S A 2009;106:11055–60.
Wagner M, Sonntag D, Grimm R et al. Substrate-specific selenoprotein B of glycine reductase from Eubacterium acidaminophilum. Biochemical and molecular analysis. Eur J Biochem 1999;260:38–49.
Wagner R, Andreesen JR. Selenium requirement for active xanthine dehydrogenase from Clostridium acidiurici and Clostridium cylindrosporum. Arch Microbiol 1979;121:255– 60.
Wagner R, Cammack R, Andreesen JR. Purification and characterization of xanthine dehydrogenase from Clostridium acidiurici grown in the presence of selenium. Biochim Biophys Acta BBA - Protein Struct Mol Enzymol 1984;791:63–74.
Wagner T, Koch J, Ermler U et al. Methanogenic heterodisulfide reductase (HdrABC-MvhAGD) uses two noncubane [4Fe-4S] clusters for reduction. Science 2017;357:699–703.
Warelow TP, Pushie MJ, Cotelesage JJH et al. The active site structure and catalytic mechanism of arsenite oxidase. Sci Rep 2017;7:1757.
Watts CA, Ridley H, Condie KL et al. Selenate reduction by Enterobacter cloacae SLD1a-1 is catalysed by a molybdenum-dependent membrane-bound enzyme that is distinct from the membrane-bound nitrate reductase. FEMS Microbiol Lett 2003;228:273–9.
Weiner JH, MacIsaac DP, Bishop RE et al. Purification and properties of Escherichia coli dimethyl sulfoxide reductase, an iron-sulfur molybdoenzyme with broad substrate specificity. J Bacteriol 1988;170:1505–10.
Weiss MC, Sousa FL, Mrnjavac N et al. The physiology and habitat of the last universal common ancestor. Nat Microbiol 2016;1:16116.
Wells M, McGarry J, Gaye MM et al. Respiratory selenite reductase from Bacillus selenitireducens strain MLS10. J Bacteriol 2019;201:e00614-18.
Wen H, Carignan J. Reviews on atmospheric selenium: Emissions, speciation and fate. Atmos Environ 2007;41:7151–65.
Whelan S, Goldman N. A general empirical model of protein evolution derived from multiple protein families using a maximum-likelihood approach. Mol Biol Evol 2001;18:691–9.
White H, Strobl G, Feicht R et al. Carboxylic acid reductase: a new tungsten enzyme catalyses the reduction of non-activated carboxylic acids to aldehydes. Eur J Biochem 1989;184:89–96.
244 Wilting R, Schorling S, Persson BC et al. Selenoprotein synthesis in archaea: identification of an mRNA element of Methanococcus jannaschii probably directing selenocysteine insertion. J Mol Biol 1997;266:637–41.
Winkel LHE, Johnson CA, Lenz M et al. Environmental selenium research: From microscopic processes to global understanding. Environ Sci Technol 2012;46:571–9. von Wintzingerode F, Gobel UB, Siddiqui RA et al. Salana multivorans gen. nov., sp. nov., a novel actinobacterium isolated from an anaerobic bioreactor and capable of selenate reduction. Int J Syst Evol Microbiol 2001;51:1653–61.
Wittwer AJ. Specific incorporation of selenium into lysine- and glutamate- accepting tRNAs from Escherichia coli. J Biol Chem 1983;258:8637–41.
Wittwer AJ, Ching WM. Selenium-containing tRNA(Glu) and tRNA(Lys) from Escherichia coli: purification, codon specificity and translational activity. BioFactors Oxf Engl 1989;2:27–34.
Wittwer AJ, Stadtman TC. Biosynthesis of 5-methylaminomethyl-2-selenouridine, a naturally occurring nucleoside in Escherichia coli tRNA. Arch Biochem Biophys 1986;248:540– 50.
Wittwer AJ, Tsai L, Ching WM et al. Identification and synthesis of a naturally occurring selenonucleoside in bacterial tRNAs: 5-[(methylamino)methyl]-2-selenouridine. Biochemistry 1984;23:4650–5.
Wolfe MD, Ahmed F, Lacourciere GM et al. Functional diversity of the rhodanese homology domain: the Escherichia coli ybbB gene encodes a selenophosphate-dependent tRNA 2- selenouridine synthase. J Biol Chem 2004a;279:1801–9.
Wood GE, Haydock AK, Leigh JA. Function and regulation of the formate dehydrogenase genes of the methanogenic archaeon Methanococcus maripaludis. J Bacteriol 2003;185:2548– 54.
Wright JJ, Mewis K, Hanson NW et al. Genomic properties of Marine Group A bacteria indicate a role in the marine sulfur cycle. ISME J 2014;8:455–68.
Wu D, Hugenholtz P, Mavromatis K et al. A phylogeny-driven genomic encyclopaedia of Bacteria and Archaea. Nature 2009;462:1056–60.
Wu M, Ren Q, Durkin AS et al. Life in hot carbon monoxide: The complete genome sequence of Carboxydothermus hydrogenoformans Z-2901. PLoS Genet 2005;1, DOI: 10.1371/journal.pgen.0010065.
Yamamoto I, Saiki T, Liu SM et al. Purification and properties of NADP-dependent formate dehydrogenase from Clostridium thermoaceticum, a tungsten-selenium-iron protein. J Biol Chem 1983;258:1826–32.
245 Yamamura S, Yamashita M, Fujimoto N et al. Bacillus selenatarsenatis sp. nov., a selenate- and arsenate-reducing bacterium isolated from the effluent drain of a glass-manufacturing plant. Int J Syst Evol Microbiol 2007;57:1060–4.
Yamazaki S. A selenium-containing hydrogenase from Methanococcus vannielii. Identification of the selenium moiety as a selenocysteine residue. J Biol Chem 1982;257:7926–9.
Yan Z, Wang M, Ferry JG. A Ferredoxin- and F420H2-dependent, electron-bifurcating, heterodisulfide reductase with homologs in the domains Bacteria and Archaea. mBio 2017;8, DOI: 10.1128/mBio.02285-16.
Yang Z. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J Mol Evol 1994;39:306–14.
Yee HS., Measures CI., Edmond JM. Selenium in the tributaries of the Orinoco in Venezuela. Nature 1987;326:686–9.
Yoshizawa S, Rasubala L, Ose T et al. Structural basis for mRNA recognition by elongation factor SelB. Nat Struct Mol Biol 2005;12:198–203.
Yu X, Niks D, Mulchandani A et al. Efficient reduction of CO2 by the molybdenum-containing formate dehydrogenase from Cupriavidus necator (Ralstonia eutropha). J Biol Chem 2017;292:16872–9.
Yuan J, Palioura S, Salazar JC et al. RNA-dependent conversion of phosphoserine forms selenocysteine in eukaryotes and archaea. Proc Natl Acad Sci U S A 2006;103:18923–7.
Zander U, Faust A, Klink BU et al. Structural basis for the oxidation of protein-bound sulfur by the sulfur cycle molybdohemo-enzyme sulfane dehydrogenase SoxCD. J Biol Chem 2011;286:8349–60.
Zargar K, Conrad A, Bernick DL et al. ArxA, a new clade of arsenite oxidase within the DMSO reductase family of molybdenum oxidoreductases. Environ Microbiol 2012;14:1635–45.
Zargar K, Hoeft S, Oremland R et al. Identification of a novel arsenite oxidase gene, arxA, in the haloalkaliphilic, arsenite-oxidizing bacterium Alkalilimnicola ehrlichii strain MLHE-1. J Bacteriol 2010;192:3755–62.
Zehr JP, Oremland RS. Reduction of selenate to selenide by sulfate-respiring bacteria: experiments with cell suspensions and estuarine sediments. Appl Environ Microbiol 1987;53:1365–9.
Zhang H, Feng X, Larssen T. Selenium speciation, distribution, and transport in a river catchment affected by mercury mining and smelting in Wanshan, China. Appl Geochem 2014;40:1–10.
Zhang Y, Fomenko DE, Gladyshev VN. The microbial selenoproteome of the Sargasso Sea. Genome Biol 2005;6:R37.
246 Zhang Y, Gladyshev VN. An algorithm for identification of bacterial selenocysteine insertion sequence elements and selenoprotein genes. Bioinforma Oxf Engl 2005;21:2580–9.
Zhang Y, Gladyshev VN. Trends in selenium utilization in marine microbial world revealed through the analysis of the global ocean sampling (GOS) project. PLoS Genet 2008;4:e1000095.
Zhang Y, Gladyshev VN. General trends in trace element utilization revealed by comparative genomic analyses of Co, Cu, Mo, Ni, and Se. J Biol Chem 2010;285:3393–405.
Zhang Y, Moore JN. Selenium fractionation and speciation in a wetland system. Environ Sci Technol 1996;30:2613–9.
Zhang Y, Romero H, Salinas G et al. Dynamic evolution of selenocysteine utilization in bacteria: a balance between selenoprotein loss and evolution of selenocysteine from redox active cysteine residues. Genome Biol 2006;7:R94.
Zhang Y, Turanov AA, Hatfield DL et al. In silico identification of genes involved in selenium metabolism: evidence for a third selenium utilization trait. BMC Genomics 2008;9:251.
Zhang Y, Zahir ZA, Frankenberger WT. Fate of colloidal-particulate elemental selenium in aquatic systems. J Environ Qual Madison 2004;33:559–64.
Zinoni F, Birkmann A, Leinfelder W et al. Cotranslational insertion of selenocysteine into formate dehydrogenase from Escherichia coli directed by a UGA codon. Proc Natl Acad Sci U S A 1987;84:3156–60.
Zinoni F, Birkmann A, Stadtman TC et al. Nucleotide sequence and expression of the selenocysteine-containing polypeptide of formate dehydrogenase (formate-hydrogen- lyase-linked) from Escherichia coli. Proc Natl Acad Sci U S A 1986;83:4650–4.
Zinoni F, Heider J, Böck A. Features of the formate dehydrogenase mRNA necessary for decoding of the UGA codon as selenocysteine. Proc Natl Acad Sci U S A 1990;87:4660– 4.
247