Arsenic and the Gastrointestinal Tract Microbiome

Home , Aerotolerant anaerobe, Teichoic acid

Environmental Microbiology Reports (2020) 12(2), 136–159 doi:10.1111/1758-2229.12814

Minireview

Arsenic and the gastrointestinal tract microbiome

Timothy R. McDermott, 1* John F. Stolz2** and Introduction Ronald S. Oremland3 The influence of microbiomes on human health and disease 1Department of Land Resources and Environmental is well established (e.g., the NIH Human Microbiome Project; Sciences, Montana State University, Bozeman, MT, Chatelier et al., 2013), with new studies continuing to demon- 59717, USA. strate the potential for important advancements in our under- 2Department of Biological Sciences and Center for standing of the basis for various human diseases and Environmental Research and Education, Duquesne conditions. One such disease condition concerns arsenic University, Pittsburgh, PA, USA. fi 3United States Geological Survey, Menlo Park, CA, (As), a toxin and carcinogen, ranking rst on the United 94025, USA. States Environmental Protection Agency Priority List of Haz- ardous Substances (Agency for Toxic Substances and Dis- ease Registry, n.d.). Based on hair analyses of Chinchorros Summary culture mummies, excess As exposure dates to prehistoric Arsenic is a toxin, ranking first on the Agency for Toxic populations of present-day Chile (Atacama Desert) (Byrne ’ Substances and Disease Registry and the Environmen- et al., 2010). In today s world more than 200 million people tal Protection Agency Priority List of Hazardous Sub- are exposed to toxic levels of As in drinking water stances. Chronic exposure increases the risk of a broad (Bhattacharjee et al., 2013; Naujokas et al., 2013), with arse- range of human illnesses, most notably cancer; how- nic contamination of drinking water in Bangladesh being ever, there is significant variability in arsenic-induced referred to as the largest mass poisoning event in human his- disease among exposed individuals. Human genetics is tory (Smith et al., 2000). Chronic As exposure has been a known component, but it alone cannot account for the linked to a range of diseases (arsenicosis), including lung, large inter-individual variability in the presentation of skin, bladder and liver cancers (Liu et al., 2008; Faita et al., arsenicosis symptoms. Each part of the gastrointestinal 2013). Interestingly, symptoms vary among individuals shar- tract (GIT) may be considered as a unique environment ing similar exposure (Naujokas et al., 2013; Cubadda et al., with characteristic pH, oxygen concentration, and 2015). Some variability is attributed to host genetics and other microbiome. Given the well-established arsenic redox factors (e.g., diet), but other causes remain to be determined. transformation activities of microorganisms, it is rea- Primary As exposure is via ingestion of either food or, sonable to imagine how the GIT microbiome composi- more commonly, As-contaminated water. In every environ- tion variability among individuals could play a ment thus far studied, microbial activities signifi cantly influ- significant role in determining the fate, mobility and tox- ence As redox speciation (trivalent vs. pentavalent), which icity of arsenic, whether inhaled or ingested. This is a dictates toxicity and mobility. Therefore, it is reasonable to relatively new field of research that would benefitfrom predict that the initial fate of As in the gastrointestinal tract early dialogue aimed at summarizing what is known and (GIT) will likewise be influenced by microbiome As metabo- identifying reasonable research targets and concepts. lism. Since microbiome composition and diversity varies Herein, we strive to initiate this dialogue by reviewing between individuals, there is potential for functional linkages – known aspects of microbe arsenic interactions and between GIT microbiome As metabolism, host exposure, fl placing it in the context of potential for in uencing host relative toxicity and disease likelihood, and thus a factor to fi exposure and health risks. We nish by considering explain symptom variability among individuals. future experimental approaches that might be of value. Much of the basics regarding microbe–As interactions originally derived from work using Escherichia coli or Staphylococcus aureus as models. This was followed by near countless studies using environmental isolates (soil, Received 27 September, 2019; accepted 25 November, 2019. For correspondence. *E-mail [email protected]; **E-mail stolz@ marine, rhizosphere, etc.). Together, these studies have duq.edu been invaluable for understanding what is possible

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd Arsenic and the gastrointestinal tract microbiome 137 elsewhere, including the GIT, and establish a framework in nature. As(V) normally dominates in well-aerated environ- for future work in the GIT microbiome arena. It is worth ments and most likely to be consumed. As(III) is both more noting that the first As-GIT microbiome study was con- toxic and mobile than As(V), making As(III) more problem- ducted by an environmental microbiology group that atic, especially in the context of drinking water from sub-oxic ® developed SHIME (Simulator of the Human Intestinal aquifers. In all environments where arsenic and microbes Microbial Ecosystem), which is a modular bioreactor gut coexist, microbes are the principal drivers of this speciation model that mimics the entire GIT, incorporating the stom- and thus are an integral part of understanding arsenic ach, small intestine and different colon regions as a cycling (Stolz and Oremland, 1999; Inskeep et al., 2001; human intestinal microbial ecosystem (Van de Wiele Oremland and Stolz, 2005; Huang et al., 2014). In simplest et al., 2004; Van de Wiele et al., 2015). It was first used terms, microbial catalysed arsenic transformations can be to examine arsenic metabolism in the context of the GIT summarized as being either oxidation, reduction, and/or (Van de Wiele et al., 2010), exposing a colon microbial (de)methylation. From the perspective of the microbe, all community to either inorganic As (iAs) or As- these reactions are ‘motivated’ by self-interest, such as contaminated soil, recording significant levels of As meth- detoxification of a poison, or in some cases the generation ylation and thiolation. of cellular energy. Table 1 and Figs 2 and 3 are provided as Arsenic research in the GIT microbiome is in a relative reference material for the different transformation functions state of infancy and can benefit from an early exchange discussed below. of ideas, concepts and views. Chi et al. (2018) review how human genetics, lifestyle and diet will influence individual susceptibility to As-related diseases. We offer to Arsenate resistance and transformations continue this dialogue but explicitly from the perspective ars gene-based resistance of the GIT microbiome. There are now growing GIT microbiome databases for healthy or diseased individ- Microbes reduce As(V) as part of a resistance mecha- uals, including the oral cavity (e.g., nasal, buccal, gingiva nism and/or for energy generation (Stolz and Oremland, and tongue), stomach, small intestine and large intestine 1999; Oremland and Stolz, 2005). While genetically dis- (Fig. 1). We consider each as a unique environment with tinct (ars genes encode resistance; arr genes encode respect to pH and oxygen concentration, as well as the respiratory activity), these processes are not necessarily residence time (i.e., how quickly ingested arsenic moves mutually exclusive and can occur in the same organism through the GIT and hence influencing potential host (reviewed by Andres and Bertin, 2016). They are most exposure). often found in the genome, but the literature documents We begin by summarizing how and why microbes react numerous instances where they are also found on plas- to As to establish a context with respect to the range of mids (e.g., San Francisco et al., 1990; Bruhn et al., 1996; microbial biochemistries that may be occurring in the Uhrynowski et al., 2019). GIT. We discuss how these activities may reduce or Arsenic resistance is encoded by the ars genes enhance toxicity and host exposure, and highlight the (Table 1) expressed in response to As exposure. The ars complexities of microbial arsenic transformations in operon is comprised of at least three genes arsRBC regards to sorting out cause and effect, as well as poten- (Rosen et al., 1992), typically expressed as one transcrip- tial influence(s) on the host. This is followed by the tional unit. ArsR is a repressor that exerts on/off control results of our GIT metagenomes surveys and a review of of the ars operon. ArsC is a reductase that converts the GIT microbiome As metabolism. We finish by As(V) to As(III) (Fig. 2A) and ArsB facilitates As(III) extru- suggesting lines of investigation that may provide greater sion (Figs 2A and 3). Acr3 is another extrusion mecha- insight into how the GIT microbiome contributes to nism and its encoding gene is actually found more pathology as well as potential preventative health frequently than arsB in ars operons (Yang et al., 2015). strategies. Just recently, Shi et al. (2018) characterized ArsK, which also facilitates extrusion of As(III) as well as other trivalent metalloids (Fig. 3). Importantly, ars gene/operon Microbe–arsenic interactions expression is independent of redox conditions. Crucial factors affecting As cycling in any environment and When As(V) enters the GIT, a bacterium will take it up human exposure (i.e., toxicity, bioavailability and bio- via a phosphate transporter, with two recognized fates. It accumulation) are directly related to its chemical speciation can be reduced by background levels of ArsC to form (Stolz and Oremland, 1999; Inskeep et al., 2001; Oremland As(III), which is the inducer that triggers up-regulation of and Stolz, 2005). The oxyanions arsenite [As(OH)3,hereaf- the ars genes resulting in considerably enhanced ter referred to as As(III)] and its oxidized counterpart arse- As(V) reductase activity and the As(III) efflux pumping -(3-n) nate [HnAsO4 , As(V)] are the prevalent forms of arsenic mechanism (ArsB, ArsK or Acr3). Thus, somewhat

Fig. 1. The human gastrointestinal tract (GIT) beginning in the oral cavity and ending at the rectum. Major phyla and genera recognized as important members of the GIT microbiome are displayed. GIT microbiome databases can be accessed at https://img.jgi.doe.gov/cgibin/m/main.cgi?sec tion=TaxonList&domain=Metagenome&seq_center=all&page=metaCatList&phylum=Host-associated&ir_class=Human paradoxically, the primary As(V) resistance strategy is to structurally similar enough to glyceraldehyde-3-phosphate transform As(V) into a more toxic form, but which is the to be recognized by ArsJ, facilitating extrusion. This amounts substrate for an active and efficient mechanism that to an As(V) import–export cycle. As(III) can enter the cell via removes it from the cell. We also note that Chauhan an aquaglyceroporin (Sanders et al., 1997) and will subse- et al. (2009) first described ArsN as a putative quently trigger the same set of regulatory and extrusion acetyltransferase and inferred from bioassays that it has responses. ArsC-like activity. Just recently, the acetyltransferase Relevance to the GIT: Ars-based resistance results in activity has been confirmed, but now ArsN is shown to extrusion from the microbial cell (Figs 2 and 3). The host- selectively confer resistance to organoarsenical antibiotic, relevant outcome is to extrude As(V) (via GAPDH/ArsJ), arsinothricin by acetylation of the α-amino group (Nadar which is neutral with respect to influencing toxicity to host et al., 2019). or neighbouring microbiome members. Alternatively, As(V) can also be extruded back out from the cell. Many As(V) is reduced to As(III) and extruded (ArsB, ArsK or ars operons contain genes encoding glyceraldehyde- Acr3), which could increase toxin exposure for the host 3-phosphate dehydrogenase (GAPDH) and arsJ, with both and neighbouring microbiome members. up-regulated in response to As(III). In Pseudomonas aeruginosa (Chen et al., 2016) and Halomonas Arsenic methylation sp. (Wu et al., 2018), GAPDH incorporates As(V) into glyceraldehyde-3-phosphate, generating 1-arseno-3-pho- Arsenic methyltransferase is encoded by arsM and has been sphoglycerate that is a transport substrate for the exporter ArsJ studied to some degree in Rhodopseudomonas (Qin et al., (Figs 2B and 3). Presumably, 1-arseno-3-phosphoglycerate is 2006), Clostridium (Wang et al. 2015a), the archaeon

Table 1. List of microbial genes that have been characterized to varying degrees with respect to their encoded functions, and documented to be involved in some aspect of arsenic biogeochemistry and or mobility.

Function Gene Encoded function References

Resistance arsR Transcriptional repressor involved in Wu and Rosen (1991); Slyemi and regulating genes involved in As Bonnefoy (2012); Kang et al. (2012); resistance, As(III) oxidation and As(V) Zargar et al. (2010) respiration arsB/acr3 Transmembrane carrier pumps (extrusion) Tisa and Rosen (1990); Rosen (2002) arsC Arsenate reductase Mukhopadhyay and Rosen (2002) arsA Anion stimulated ATPase Tisa and Rosen (1990); Rosen et al. (1999) arsD Transcriptional repressor and As(III) Wu and Rosen (1993); Lin et al. (2006) chaperone arsH Oxidizes MMAs(III) to MMAs(V) Chen et al. (2015a) arsJ Extrudes 1-arseno-3-phosphoglycerate Chen and Rosen (2016) arsK Facilitates As(III) extrusion Shi et al. (2018) arsN Confers resistance to the organoarsenical Nadar et al. (2019) arsinothricin Putative acetyltransferase having ArsC Chauhan et al. (2009) activity arsP methylarsenite efﬂux Chen et al. (2015b) arsM Arsenite methyltransferase Qin et al. (2006); Qin et al. (2009) arsI C-As&$$$; lyase Yoshinaga and Rosen (2014) As(III) oxidation aioB Arsenite oxidase small subunit Lett et al. (2012) aioA Arsenite oxidase large subunit Lett et al. (2012) aioC c-type cytochrome Santini et al. (2007); Branco et al. (2009) aioD Molybdenum cofactor Branco et al. (2009) aioX Periplasmic As(III)-binding protein Liu et al. (2012) aioS Two-component signal transduction Kashyap et al. (2006) histidine kinase aioR Response regulator Kashyap et al. (2006) arxA Arsenite oxidase large subunit Zargar et al. (2010) (Photoarsenotrophy) arxB Arsenite oxidase large subunit Zargar et al. (2010) (Photoarsenotrophy) phoR Phosphate stress response two-component Slyemi and Bonnefoy (2012); Kang et al. signal transduction histidine kinase (2012) phoB Phosphate Stress response Response Slyemi and Bonnefoy (2012); Kang et al. regulator (2012); Wang et al. (2018) Anaerobic As(V) Respiration arrA Respiratory arsenate reductase large Saltikov and Newman (2003) subunit arrB Respiratory arsenate reductase small Saltikov and Newman (2003) subunit arrC Membrane subunit involved in anchoring Duval et al. (2008); van Lis et al. (2013) and electron transfer arrD Chaperone van Lis et al. (2013)

Methosarcina (Wang et al., 2014) and fungi (Chen et al., transfer from methylcobalamin [CH3Cob(III)] to 2-mercap- 2017; Li et al., 2018). However, by far the best-studied and toethanesulfonic acid (CoM) (Thomas et al., 2011). understood enzyme was cloned from the acidothermophilic In an aerobic environment where oxygen is available, alga Cyanidioschyzon (Qin et al., 2009) and has been used the function of ArsM in microorganisms appears to be detox- extensively to model this reaction. ArsM generates trivalent ification, where methylated As(III) becomes oxidized to methylated species (Ajees et al., 2012; Ajees and Rosen, either a less toxic pentavalent methylated species or the vol- 2015) (Fig. 2C), not pentavalent species as originally hypoth- atile trimethylarsine (TMAs), which would spontaneously esized (Challenger, 1945). Methylarsenite [MAs(III)] and release from the cell (Qin et al., 2006; Qin et al., 2009). For dimethylarsenite [DMAs(III)] are the most toxic arsenicals the microbe catalysing this reaction, MAs(III) synthesis is known, but in aerobic culture conditions where these methyl- problematic because of its extreme toxicity. This can be ation reactions are most often studied, the pentavalent spe- solved by ArsP, a specificMAs(III)efflux permease or detox- cies MAs(V) and DMAs(V) are the major arsenicals ified via ArsH, which is an organoarsenical oxidase capable observed, likely due to autooxidation of the trivalent species of oxidizing trivalent methylated and aromatic arsenicals (Qin et al., 2006; Qin et al., 2009). In addition to ArsM, some (Chen et al. 2015a) or ArsI, which cleaves the carbon-arsenic methanogens will methylate metalloids (As, Se, Sb, Te and bonds in methylated As(III) derivatives (Figs 2D and 3) Bi) using the methyltransferase MtaA to catalyse methyl (Yoshinaga and Rosen, 2014).

Fig. 2. Potential primary GIT-relevant microbiome arsenic transformations. A. Oxidation or reduction of the principal oxyanions As(III) and As(V). B. Arsenate incorporation into glyceraldehyde 3-phosphate to synthesize the extrusion substrate for ArsJ. C. Arsenic methylation catalysed by ArsM, resulting in three possible products: MAs(III), DMAs(III) and TMAs(III). Also shown are the spontaneous oxidation products for each: MAs(V), DMAs(V) and TMAO, respectively. D. Detoxiﬁcation of methylarsenite by the organoarsenical oxidase ArsH, and ArsI, which cleaves carbon-arsenic bonds.

Relevance to the GIT: Arsenic methylation to the trivalent the host. Microbiome members exposed to extruded species has negative consequences, creating an even more MAs(III) but that lack ArsP, ArsH or ArsI, will be selected toxic situation that potentially interrupts normal host func- against, contributing to shifts in microbiome composition. tions as well as microbiome functions that are beneﬁcial to Indeed, occurrence, abundance and expression of all of the

Fig. 3. Potential microbiome arsenic transformations of relevance to the host. Abbreviations are linked to deﬁnitions in the text. above ars genes anywhere along the GIT represent an microbiome. Unlike the cytoplasmic ArsC, ArrAB is important example of how these microbiomes can serve to located in the periplasm and hence the toxic As(III) decrease or substantially enhance the toxicity of ingested formed is outside the bacterial cytoplasm and does not arsenic.

As(V) reduction for energy generation

Many microorganisms will reduce As(V) as part of anaerobic respiration (Stolz and Oremland, 1999; Oremland and Stolz, 2003). Regulatory control of dissimilatory As(V) reductase genes (arrAB) is more complex than arsC (Glasser et al., 2018), requiring both arsenic exposure and anaerobic conditions (Murphy and Saltikov, 2009). Recently, Switzer Blum et al., 2018 described a Citrobacter sp. isolated from a termite hindgut that displays As(V) dependent growth, but without ArrAB. Initial evidence suggests ArsC may be somehow involved and thus represents somewhat of a hybrid in a physiologic sense. Relevance to the GIT: The lumen is viewed to be a largely anaerobic environment (Fig. 4) (except see below), therefore, the appropriate redox conditions pre- vail for arrAB expression. Thus, As(V) reduction via Fig. 4. Cartoon depiction of the lumen epithelia, illustrating the transi- tion from the anaerobic lumen environment to the inner and outer ArrAB and ArsC could be additive with regards to the net mucus environments that are oxygenated via the villi. Adapted from As transformation capacity of an organism or https://en.wikipedia.org/wiki/Intestinal_epithelium

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 142 T. R. McDermott, J. F. Stolz and R. S. Oremland require extrusion. Therefore, microbiome As(V) respiration arsenic compounds. Offering a chemical perspective, Fan should not be considered a positive outcome in the context et al. (2018) offered a new hypothesis that suggests thio- of host toxicity and exposure. lation is integrated with methylation, involving protein-bound intermediates. Here, we adopt the thioarsenical definition used in a recent review of methylated thioarsenical com- As(III) oxidation pounds (Sun et al., 2016), where glutathione or protein thiol As(III) oxidation can serve as a detoxification mechanism conjugates are not included in this category. Mono-, di-, tri- or can be used to generate energy, using As(III) as an and tetrathioarsenicals are primarily of interest here and are electron donor. The structural genes encoding the As(III) shown in Fig. 5. The degree of sulphur substitution is critical oxidase heterodimer are aioB (small sub-unit, Reiske type) to toxicity and an example where a pentavalent arsenic and now referred to as aioA (large sub-unit). A species can be highly toxic. DMAs(V) is of low relative tox- Centibacterium arsenoxidans aox (aoi) mutant is more icity, but thiolation to dimethylthioarsenate (DMAs(V)S) sensitive to As(III) than the wild-type parental strain (Muller greatly increases its toxicity, approaching that of As(III) and et al., 2003) and thus an example where As(III) oxidation DMAs(III) (Naranmandura et al., 2007). However, further was shown to serve a detoxification purpose. The first sulfidication to form dimethyldithioarsenate (DMAs(V)S2) reproducible report of As(III) chemolithotrophy was in an reduces toxicity significantly (Ochi et al., 2008; Kim et al., Agrobacterium/Rhizobium-like organism (Santini et al., 2016). Consequently, one cannot generalize regarding the 2000). As(III) oxidation is typically found linked to oxygen toxicity of thiolated arsenicals. as an electron acceptor, but anaerobic As(III) oxidation Factors important in determining the extent of forma- has also been documented in support of anoxygenic pho- tion, solubility and stability of the thioarsenicals include tosynthesis (Kulp et al., 2008; Hernandez-Maldonado pH, redox conditions, and prevailing concentrations and −2 et al., 2016) or where nitrate is the alternative electron ratios of As to SO4 and or H2S (Wilkin et al., 2003; acceptor (e.g., Oremland et al., 2002; Rhine et al., 2006). Bostick et al., 2005; Stauder et al., 2005). Abiotic reac- Consequently, As(III) oxidation can proceed under either tions appear to control arsenic thiolation (Planer-Friedrich aerobic or anaerobic conditions and so, like ArsC-based et al., 2007; Stauder et al., 2005; Bostick et al., 2005. As − As(V) reduction, is not tied to prevailing redox conditions. the HS :arsenic ratio increases, so does the degree of The aioBA genes have been described for a range of thiolation (Wilkin et al., 2003; Bostick et al., 2005; microbes (Cai et al., 2009) and are phylogenetically dis- Stauder et al., 2005), although there is some disagree- tinct from the more recently discovered ArxAB that cou- ment whether the thioarsenical formed is trivalent (Wilkin ples As(III) oxidation with photosynthesis (Zargar et al., et al., 2003) or pentavalent (Stauder et al., 2005). Abiotic

2010; Zargar et al., 2012) or nitrate reduction (Hoeft H2S reduction of As(V) is highly pH-dependent, reacting et al., 2007). poorly around pH 7.0, even with H2S:As ratios as high as In the best-studied bacteria, aioBA expression is regu- 100:1, but reaction rates increase markedly at low pH lated by a signal transduction system that includes a peri- (Rochette et al., 2000). As(III) and H2S interactions can plasmic As(III) binding protein, AioX (Liu et al., 2012), the result in poorly soluble arsenic solid phases such as sensor kinase AioS and response regulatory protein AioR amorphous orpiment (As(III)2S3), realgar (As4S4) (Fig. 5) (Kashyap et al., 2006; Koechler et al., 2010) (Table 1). or even elemental As—all of which would be expected to Expression of aioXSR requires As(III) exposure, but in addi- be poorly bioavailable in the GIT (Liu et al., 2008). This too tion, aioSR are ultimately controlled by another signal trans- is highly pH-dependent and oxygen-sensitive. Maximum pH duction system, PhoR-PhoB, which controls the phosphate where arsenical sulphide precipitation might occur is ~8.0, starvation response in bacteria (Kang et al., 2012; Wang whichisstillwithintherangefoundintheGITandcanbe et al., 2018). AioXSR is not a universal requirement, as quantitative at pH 6.1 and increasing with more acidic pH some As(III) oxidizing bacteria lack aioXSR (Koechler et al., (Rodriguez-Freire et al., 2014). Under neutral to alkaline

2015; Li et al., 2013). Wang et al. (2018) suggested that conditions, abiotic As(III)2S3 dissolution is rapid (Hartig and PhoRB is essential for regulating As(III) oxidation in most, if Planer-Friedrich, 2012; Suess and Planer-Friedrich, 2012), not all, microorganisms that do so. although transformation kinetics slow with progression from

Relevance to the GIT: Oxidation of As(III) to As(V) will trithioarsenate As(V)S3 to thioarsenate (As(V)S). Within have a detoxifying effect on both the host as well as the temporal scales of hours (i.e., GIT transit time), As(V)S is microbiome. chemically stable, even in the absence of H2S(Hartiget al., 2014), but can be rapidly decomposed by capable microorganisms leading to As(V) with intermediate accumulation of Thiolated arsenicals As(III) (Hartig and Planer-Friedrich, 2012). Organisms such In terms of potential microbial biogenesis, thiolated as Thermocrinis ruber will use As(V)S as a chemo- arsenicals are the least understood naturally occurring lithotrophic electron donor with O2 as an electron acceptor

Fig. 5. Major sulphur-arsenic compounds discussed in review. Orpiment and realgar are poorly soluble and of low bioavailability in the GIT (Liu et al., 2008), whereas the other thiolated arsenicals shown have been documented for mammalian urine (Raml et al., 2007; Suzuki et al., 2010, blood (Naranmandura et al., 2007b; Naranmandura and Suzuki, 2008) and saliva (Wang et al. 2015b), or in mouse cecum contents incubated anaerobically with As(V) (Pinyayev et al., 2011).

2− (yielding As(V) and SO4 )(HartigandPlaner-Friedrich, and/or As4S4 precipitation or as co-precipitates with ferrous 2012), but this activity has not been examined with GIT- sulphide or pyrite (Altun et al., 2014; Rodriguez-Freire et al., relevant samples. 2016). These mineral phase arsenicals are stable at acidic Evidence of direct microbial enzymatically catalysed pH, but less so with increasing pH, being soluble at high thioarsenical synthesis is lacking. However, by default pH (9.8) (Oremland et al., 2000). SRBs such as 2− SO4 reducing bacteria (SRB) are involved, which are Desulfosporosinus (aka Desulfotomaculum) auripigmentum common in the gut (Gibson et al., 1993; Carbonero and (Newman et al., 1997) and Desulfovibrio sp. str. Ben-RA 2− Gaskins, 2013). A Shewanella putrefaciens isolate may (Macy et al., 2000) will utilize As(V) and SO4 as electron be an exception; it converts MAs(V) to methythioarsenate acceptors, resulting in solid-phase As4S4 or As(III)2S3 (MAs(V)S), and DMAs(V) to dimethythioarsenate (DMAs occurring both intracellularly and extracellularly (Newman (V)S) (Chen and Rosen, 2016). SRB activity provides et al., 1997), occurring as inclusion bodies (Newman et al., opportunity for the strictly anaerobic chemoautotrophic 1997) or as extracellular/intracellular As-S nanotubes in

Deltaproteobacterium (strain MLMS-1) that uses H2Sas Shewanella sp. (Lee et al., 2007; Jiang et al., 2009). its e-donor and As(V) as its e-acceptor (Planar-Friederich Relevance to the GIT: The above are all examples of et al., 2015), and can also result in signiﬁcant As(III)2S3 what is possible in terms of GIT microbiome activities

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 144 T. R. McDermott, J. F. Stolz and R. S. Oremland relevant to arsenic thiolation. When considering the GIT marine microbes have evolved this into an arsenic defence in its entirety, the lowest pH conditions may allow for abi- response; e.g., Phaeodacylum tricornutum will synthesize a otic H2S reduction of As(V) if H2S:As ratios are high phytochelatin in reaction to arsenic exposure. These pep- enough to drive the reaction. Diet (e.g., sulphate levels), tides have the general structure of (γ-Glu–Cys)n-Gly, where pH and SRB activity in the GIT will influence whether As: n =2–6(Grillet al., 1985; Morelli et al., 2005) and in plants S precipitation reactions will occur, and consequently are derived from glutathione (Grill et al., 1989). These influence the relative toxicity of the arsenic passing phytochelatins sequester the trivalent arsenical, avoiding through the GIT. Arsenic precipitation will reduce soluble the inactivation of critical cellular processes. A well- As loads circulating in the GIT, and thus would be characterized arsenic–protein interaction involves ArsR, favourable to the host although it is uncertain if or how which outcompetes As(III) extrusion (via ArsB or Acr3) the microbiome would be influenced. Information regard- (Paez-Espino et al., 2009) so as to activate the ars-based ing host uptake of thiolated arsenicals is quite limited, detoxification system (discussed above). ArsR proteins vary documented only with in vitro studies using cultivated cell in As(III) affinity (Ke et al., 2018). When expressed as lines (Naranmandura et al., 2007; Naranmandura et al., recombinant proteins in Escherichia coli, ArsR can remove 2009; Hinrichsen et al., 2015). Relevance to the in situ is up to 82% of the As(III) in a 10 ppb solution. difficult to judge at this juncture. Relevance to the GIT: Most ars-based activities (discussed above) result in As being extruded (resistance mechanisms) or other activities are employed in energy- Arsenic bioaccumulation generating transformations that take place exterior to the Recent studies have illustrated As(V) uptake and incorpo- cytoplasm (As(III) oxidation and or Arr-based ration into microbial biomass (Takeuchi et al., 2007; Pan- As(V) reduction). These activities maintain similar levels of dey and Bhatt, 2015; Wang et al., 2018), implying that host exposure, with the reduction or methylation activities As(V) is not inert in a biochemical context and that micro- increasing toxicity. Conversely, microbial bioaccumulation bial bioaccumulation can reduce host exposure. Wang of arsenic in the GIT may be an important outcome for et al. (2018) demonstrated As(V) enhanced growth of Pi ingested arsenic and could conceivably be viewed as a pro- stressed Agrobacterium tumefaciens.Thespecificbio- phylactic strategy of therapeutic value to reduce host expo- chemical pathways and enzymes involved are unknown, sure. Indeed, microbial bioaccumulation has been although cellular localization work suggests lipid incorpora- suggested to represent an environmental bioremediation tion (Wang et al. 2015; Wang et al., 2017; Wang et al., strategy (Paez-Espino et al., 2009), and by extension could 2018). Arsenolipids are well documented and naturally be of clinical importance, particularly for developing nations occurring in marine animals (Dembitsky and Levitsky, where filtration systems are unavailable or too expensive. 2004), macroalgae (Francesconi, 2010; Garcia-Salgado In a human trial involving pregnant women in Tanzania, et al., 2012), as well as cyanobacteria (Xue et al., 2017; yogurt treatment (containing Lactobacillus rhamnosus) Xue et al., 2019), providing sufficient precedent for reduced arsenic and mercury bioaccumulation in the test As(V) incorporation into specific cellular structures of vari- subjects (Bisanz et al., 2014), an example of how probiotics ous organisms, including GIT microbiome biomass. This will could function to reduce host exposure. likely involve stable C As bonds as in arsenohydrocarbons or phosphonolipids, or the recently discovered 2-O- Metabolic disruption methylriboside (Glabonjat et al., 2017). Under Pi stress conditions, bacteria may replace up to 40% of cellular Pi by: Arsenic will seriously disrupt bacterial cell metabolism. (i) recycling phospholipids (Geiger et al., 1999); (ii) down- RNASeq-based transcriptomic studies showed the regulating genes encoding synthesis of Pi-rich teichoic effects are global, altering the expression (+/−) of nearly acids (Lang et al., 1982; Salzberg et al., 2015); and 500 genes encoding a broad range of cellular functions (iii) synthesizing and utilizing sulfolipids (Souza et al., 2008; in A. tumefaciens (Rawle et al., 2019). Prominent exam- Zavaleta-Pastor et al., 2010). It is possible that Pi sparing ples include elements of the phosphorus stress by As(V) may be a fate for As(V) in the GIT microbiomes, response, carbohydrate metabolism, oxidoreductases/ but not yet examined at any level. As currently understood, electron transport, as well as copper tolerance. Interest- this would require low phosphorus conditions (e.g., <5 μM) ingly, As(III) exposure was associated with an organized and As(V):P ratios of at least 1:1 in the GIT environment. down-regulation of >40 genes linked with iron acquisition, Another mechanism to explain As bioaccumulation in indicating some bacteria may close down iron uptake as microorganisms involves arsenic binding to proteins (Shen part of an As(III) response, and an example of collateral et al., 2013) due to the strong attraction of trivalent arseni- effects of arsenic ingestion. cals for sulfhydryl groups (Shen et al., 2013) and is the pri- Such broad sweeping transcriptional changes would be mary basis for As(III) being more toxic than As(V). Some expected to result in metabolism changes, which is exactly

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 Arsenic and the gastrointestinal tract microbiome 145 what has been observed. Based on mass spectrometry and including the dental plaque forming Staphylococcus and a nuclear magnetic resonance analyses of the same A. variety of Campylobacter species. The nasal cavity would tumefaciens strain mentioned directly above for RNASeq be the route for inhaling volatile arsenicals or arsenic-laced work, As(III) perturbation of metabolism is likewise global in dust (e.g., chicken litter and feed). The oral and posterior nature (Tokmina-Lukaszewska et al., 2017), involving Æ nasal flora differ, with the latter dominated by Actinobacteria changes in >300 polar metabolites and ~600 non-polar and Firmicutes (Bassis et al., 2014; Koskinen et al., 2018). metabolites. Primary influences were documented as inter- The stomach microbiome has five major phyla: Firmicutes, rupted glycolysis, and disruption of two key reactions asso- Bacteroidetes, Actinobacteria, Fusobacteria and Prote- ciated with carbon flow into and within the tricarboxylic acid obacteria,withspeciesofPrevotella, Streptococcus, cycle; i.e., pyruvate dehydrogenase and α-ketoglutarate Veillonella, Rothia and Haemophilus, and the occasional dehydrogenase, as well as branched-chain α-ketoacid Helicobacter pylori (Nardone and Compare, 2015) The dehydrogenases. These enzymes contain a dihydro- large intestine has the greatest species richness (Qin et al., lipoamide dehydrogenase subunit that is sensitive to As(III) 2010). Major phyla are Firmicutes, Bacteroidetes, inactivation (Bergquist et al., 2009; Afzal et al., 2012), Actinobacteria, Fusobacteria and Proteobacteria, however, resulting in carbon flow being diverted to amino acids the number of genera represented is far greater and (e.g., glutamate, glutamine, aspartate and alanine) includes a wide range of physiological and biochemical (Tokmina-Lukaszewska et al., 2017). diversity (Agioutantisa and Koumandou, 2018). Initial high Relevance to the GIT: A recent Pubmed search identi- throughput 16S rRNA gene sequence analysis provided a fied 347 reviews discussing the importance of micro- glimpse of the overall species diversity of the human microbiome metabolism to human health. Consequently, it is biome, but more recent metagenomic studies have revealed easy to propose that arsenic disruption of microbiome the full genetic diversity and rich gene pool. This is particu- metabolism will have direct, significant effects on host larly true for arsenic resistance genes. metabolism, with a significant likelihood of leading to dis- As described above, both As(III) oxidation and ease. Below, we now review these effects in more detail. As(V) reduction can and will occur independent of redox conditions and thus can occur anywhere along the GIT. Respiratory As(V) reduction requires anaerobic condi- Arsenic and the GIT microbiome tions and thus should be viewed as more relevant to the The above sections summarize known microbial interac- lumen interior. Oxygen is secreted from the gut epithe- tions and responses to arsenic in pure cultures or in various lium, especially the small intestine (Fig. 4), generating environments. However, it is reasonable to predict that if steep oxygen gradients extending out roughly 1 mm these genes are present in the GIT, their regulation and (Espey, 2013), providing a niche for O2-linked As(III) oxi- encoded transformations would be similar. Owing to the com- dation and O2-linked metabolisms (Donaldson et al., plexity of the GIT microbiome (NIH HMP Working Group et al., 2016). Thus it is reasonable to imagine how O2 availabil- 2009), the genetic and functional potential for all of the above ity at this precise location could be of significant value to transformations may be present and occurring, perhaps simul- the host, allowing conversion of more toxic trivalent taneously. This could result in altered community composition arsenicals [e.g., As(III) or MAs(III)] to the less toxic penta- and potentially lead to microbiome dysfunction, depending on valent species [e.g., As(V) or MAs(V)] (Figs 2C and 4). the extent of the change(s). Furthermore, the relative abun- As(III) oxidation is clearly possible in the GIT, not only by dance and expression of specific functional genes will dictate O2 but also via alternative electron acceptors such as net community arsenical output and thus ultimately whether the nitrate, which is easily possible and illuminates the role of microbiome activities will be of positive or negative influence for diet when attempting to fully understand causes and the host. The discussion below relates to transformations gene penetrance of arsenicosis. Whether diet can ulti- shown in Fig. 2 and potential microbiome activities summarized mately alter the redox balance towards As(III) or As(V) is in Fig. 3. Documented functions and genes in GIT-specificset- a topic worthy of examination. tings, the controlling factors, and the pros and cons of each with respect to host exposure and well-being are assessed. Arsenic perturbation of the microbiome The GIT microbiome will encounter ingested arsenic in the oral cavity, stomach, small intestine and large intestine Functional and phylogenetic change. Several studies have (i.e., colon) (Fig. 1). The oral cavity offers both oxic and examined how and to what extent dietary arsenic impacts anoxic environments and a complex community dominated the gut microbiome, documenting change in at least two perspectives. In the functional context, a pioneering rumen- by species belonging to six major phyla, Firmicutes, Actino- based study found fermentation rates were inhibited ~30% bacteria, Proteobacteria, Fusobacteria, Bacteroidetes and by the addition of ~30 μM As(III) to rumen microflora enrich- Spirochetes (Verma et al., 2018). The Human Oral Micro- ments; levels of As(III) required to inhibit fermentation were biome Database currently lists over 770 bacterial species, less than that toxic to the animal itself (Forsberg, 1978). In a

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 146 T. R. McDermott, J. F. Stolz and R. S. Oremland phylogenetic context, studies examining insects, fish, and so caution must be exercised when attempting to relate rodents and ruminants (Schramm et al., 2003; Pass et al., community changes to function following arsenic exposure. 2015; Dahan et al., 2018; Gaulke et al., 2018; Liu et al., Associating cause and effect can be elusive and potentially 2018; Xue et al., 2019b) are congruent in illustrating that near meaningless, particularly if the analyses are at higher arsenic alters the microbiome community composition. taxonomic levels. Changes corresponded to increases in Bacteroidetes and Arsenic-associated microbiome compositional changes decreases of Firmicutes. Although, when assessed at the may also relate to modified host metabolism that then sub-phylum taxonomic level the classes Bacilli and Clos- imposes selection effects on the microbiome, but that do not tridia actually increase (Lu et al., 2014), illustrating how involve direct arsenic toxicity effects on the microbial cell. It reporting community responses at only the phylum level is is reasonable to predict that microbiome community changes misleading and fails to recognize potentially important derive from host selection can, in turn, result in altered arse- changes in key populations as defined at the genus level or nic transformations and net microbiome arsenic species out- even as OTUs. put. Arsenic-induced mouse microbiome perturbations have Arsenic effects on human gut microbiomes show similar been related to significant changes in host serum profiles of trends, and in some cases begin to shed light on how such fatty acids, phospholipids, sphingolipids, cholesterols and changes may relate to the manifestation of human tryptophan (Xue et al., 2019b). Other mouse gut arsenicosis symptoms. The New England region of the microbiome-related changes included downregulation of host United States has elevated arsenic due principally to natural genes encoding one-carbon and glutathione metabolisms, geologic inputs. Analysis of stool microbiome composition of significantly decreased liver S-adenosylmethionine levels, 204 New Hampshire infants found several genera within the and numerous genes encoding various functions associated Firmicutes to be suppressed; 15 genera were negatively cor- with hepatocellular carcinoma were significantly altered (Chi related with urine arsenic levels, including Bacteroides and et al., 2019). Arsenic exposed adult mice display higher Bifidobacterium (Hoen et al., 2018). A comparison of stool levels of CC chemokines, and pro-inflammatory and anti- microbiomes of Bangladesh children exposed to 10 ppb ver- inflammatory cytokine secretion in the intestine accompanied sus >50 ppb arsenic in drinking water (Dong et al., 2017) by significant changes in the gut microbiome (Gokulan et al., revealed microbiome compositional shifts, with Prote- 2018). It is also reasonable to assume microbiome change/ obacteria and arsB and arsC copy number increasing at the differences relate to host genotype as shown in significant higher arsenic exposure (Dong et al., 2017). To our knowl- microbiome compositional differences between wild-type and edge, this is the only example where relevant microbial func- interleukin-10 knockout mice that in turn was tied to altered tional genes were examined or recorded as shifting in arsenic metabolism (Lu et al., 2014b), supporting the link response to host exposure to arsenic and provides the first between microbiome composition and pre-systemic arseni- hint of the microbiome physiologic responses to be expected cal metabolism. When the murine pathogen Helicobacter or predicted. In another Bangladesh study (250 human vol- trogontum was used to ‘perturb’ a mouse microbiome, sub- unteers), arsenic-associated microbiome perturbations rev- sequent arsenic exposure resulted in generally higher levels ealed that the relative abundances of the family of As(V) and lower levels of MAs(III) and DMAs(V) relative to Aeromonadaceae and genus Citrobacter were significantly ‘uninfected’ mice (Lu et al., 2013). This is an excellent exam- associated with intima-media thickness, a surrogate marker ple where microbiome composition influences arsenic bio- for the cardio-vascular disease atherosclerosis (Wu et al., transformation potential, particularly methylation, and in this 2019). Obviously, correlative data are only suggestive but case resulted in reduced host exposure to the more toxic nevertheless may provide important clues for future work arsenicals. that links the microbiome with host health.

Basis for change. In the context of arsenic exposure, Genetic potential for microbiome As transformations in changes in microbiome community composition can derive the GIT from several factors. Variable As tolerance and detoxification capacity among genera and species are well documented Microbiome characterization studies use high throughput and could result in selective sweeps of the microbiome, per- sequencing approaches that generate high-level taxon haps becoming permanent if selection pressure is constant. descriptors. Taxonomic resolution can often be limited to The ability to utilize As(III) or As(V) for energy generation uncharacterized OTUs within a phylum. This provides lit- would provide a distinct advantage. Furthermore, production fi of a toxin such as MAs(III) by even a low abundance organ- tle information about the organism identi ed and con- ism could have devastating effects on microbiome neigh- strains any understanding of whether the represented bours (Li et al., 2016) that lack the ability to eliminate it from organism has any As metabolic/transformation potential. the cell (ArsP) or detoxify it (ArsI or ArsH) (Chen et al., By contrast, functional gene analysis will provide informa- 2018). Production of arsenic-based antibiotics such as tion regarding transformation potentials, although inter- arsinothricin (Kuramata et al., 2016) could have similar pretations or predictions become difficult for microbiomes effects, though moderated by ArsN that inactivates it (Nadar containing multiple functionalities. For example, co- et al., 2019) and arsenic resistance acquired from spontaneous mutations in vivo will also account for some changes. existence of As(III) oxidizers and As(V) reducers in the These properties can vary at the species level or even same environment can favour either As(III) oxidation or between strains of the same species (Macur et al., 2004), As(V) reduction (Macur et al., 2001; Macur et al., 2004;

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 Arsenic and the gastrointestinal tract microbiome 147 Hoeft et al., 2010). However, unless utilizing radiolabeled sequences of hits from across the search results spectrum As (Hoeft et al., 2010), quantifying each activity is difficult (i.e., best to worst) were examined for conserved amino because one or the other will dominate the HPLC-ICP- acids or amino acid motifs. The searched metagenomes MS output that measures the net/dominant redox activity, varied with respect to the total number of identified genes but does not account for As(III) $ As(V) cycling. Further- (i.e., ~36 000–300 000; Table S1), which contributed to the more, being able to define specific genera is not neces- variation of average hits observed per metagenome. sarily reliable for predicting which type of activity is present. These properties can vary at the species level or Arsenate reduction. ArsC was most abundant and found at the greatest frequency (Table 2). Escherichia. coli is not a even between strains of the same species (Macur et al., dominant GIT organism, which was reflected by the lower 2004). Further complicating matters, the net arsenic frequency of its ArsC in contrast to the more abundant transformation phenotype of an organism may be As(III) Bacteroides and Faecalibacterium type ArsC. The respira- oxidation if it also carries the aioBA genes (Kashyap tory As(V) reductase, ArrAB, was far less prevalent, but nev- et al., 2006). In fact, all organisms isolated and character- ertheless represented in two separate individuals and ized as As(III) oxidizers also carry the ars genetic deter- verified by amino acid alignments, phylogenetic analysis and fi minates for resistance-based As(V) reduction. motif veri cation. Finding both ArsC and ArrA is an example of how this transformation activity can be redundant and The above caveats and complexities notwithstanding, potentially additive. However, an important caveat concerns a functional biomarker approach to assess the presence relative expression levels of each. For example, transcrip- of key genes (Dietert and Silbergeld, 2015) should tional analysis of arsC and arrA genes in Shewanella ANA-3 become the standard for examining GIT microbiome As showed that the expression of arrA was induced by 100 nM metabolic potential. Gene selection ideally would encom- As(III), while arsC was not induced until the As(III) levels pass as many functions as possible (see Figs 2 and 3). were increased by three orders of magnitude (Saltikov et al., Very few human microbiome relevant microorganisms 2005). The level of As(III) required in the GIT for arsC to be activated is currently unknown. Another factor contributing to carry a full suite of ars genes (arsA, arsB, arsC, arsD, this regulatory quagmire is that arrA expression is repressed arsR, aqp and usp) (Isokpehi et al., 2014), but in reality by other electron acceptors such as nitrate and fumarate. an organism would only need an arsC and arsB/acr3 and perhaps a means of regulation (typically arsR) in order to Arsenite oxidation. As(III) oxidase belongs to the large contribute to pre-systemic host As metabolism. dimethylsulfoxide (DMSO) reductase family of molyb- We conducted BLASTp searches of stool microbiome doenzymes (McEwan et al., 2002). There were no hits for metagenomes reported for 25 randomly selected healthy AioA, but searches did identify other DMSO family members (e.g., formate dehydrogenase), illustrating the sensitivity of individuals (Table 2), focusing on functions depicted in Figs 2 the search process but also illustrates that one must be and 3. For ArsC, we considered the two major ArsC func- aware of how similar protein sequences that turn up in such tional classifications; grx type and trx type based on their searches actually represent very different functions. The activity linked to glutaredoxin or thioredoxin, and that repre- gene encoding the small subunit of the arsenite oxidase sent Gram-negatives and positives that are well docu- enzyme (aioB) has been reported in stool microbiomes of mented to occur and/or particularly abundant in the GIT alcoholic cirrhosis patients (Chen et al., 2014). It is not (E. coli grx type and Bacteroides trx type) and also a good immediately clear why aioB would be abundant enough to detect in the microbiomes of such individuals, although wine representative Gram-positive (Faecalibacterium trx type). has drawn considerable attention in this context because of Similarly for ArsB/Acr3, the search queries represented well the arsenic content (Wilson, 2015; Vacchina et al., 2018). known and abundant GIT community genera: e.g., Higher abundances of As-relevant genes might be expected Escherichia, Bacteroides and Faecalibacterium. Using if wines were a beverage of choice for these particular test both query sequences from Gram+ and Gram– organisms individuals. No ArxA proteins were found. allowed us to capture some degree of phylogenetic diversity. However, we note that there is a fair amount of Other functions. ArsM was found in the stool microbiome of seven individuals and annotated as either Bacteroides or phylogenetic diversity within each well-known arsenic trans- Parabacteroides proteins. In one instance the encoding formation activity (e.g., Dunivin et al., 2019) and thus a com- arsM was found adjacent to an arsC, suggesting that this prehensive search would require many different query particular ArsM is part of an organized arsenic resistance sequences. For other functions for which less is known, response for a (Para)Bacteroides organism. ArsM functional queries used amino acid sequences of known and charac- variability would contribute to inter-individual microbiome ’ terized proteins. Hits were screened by the expected proba- uniqueness (Eckburg et al., 2005; O Toole, 2012; Ukhanova − bility of a random hit (E-value >10 20) and identity match et al., 2012; Pérez-Cobas et al., 2013), which is supported by a SHIME-based study that concluded arsenic methylation scores. The encoding genes were then examined to deter- varies between individuals and particularly when comparing mine if they are physically associated with other ars genes, adults to children (Yin et al., 2017). Again, however, there which would indicate they are part of an organized arsenic are caveats. This study and others (Juhasz et al., 2011; response. When available for specific proteins, amino acid Smith et al., 2014) inoculated the system with soil as a

Table 2. Frequency of various ars, arr and aio genes in the metagenomes of 25 healthy individuals

09SceyfrApidMcoilg n onWly&Sn Ltd, Sons & Wiley John and Microbiology Applied for Society 2019 © Genome Protein and or occurrence Average hits per E-value ID Score Adjacent to ars Category source (query) frequency (of 25) metagenome range range (%) or aio genes? Comments

Redox/energy ArrA 2 0.08 Æ 0.06 −77 to 0 47–53 Not expected Several hits based on e-value and identity, but did not have the required GSPNNISHSSVCAEAHKMG Mo-binding domain ArxA 0 0 NA NA NA AioA 0 0 NA NA NA Queries found only formate dehydrogenase Ars − ArsH 1 0.04 6 × 10 39 63% Yes ArsI 0 0 NA NA NA ArsJ 0 0 NA NA NA ArsJ from Anoxybacillus and Pseudomonas used as queries ArsN 0 0 NA NA ArsP 25 5 Æ 2.9 −20 to −51 24–42 Yes Veriﬁed ‘TPFCSCSXXP’ as deﬁned by Chen et al. (2015b) ArsB/Acr3 (As(III) extrusion) E. coli ArsB 2 0.08 Æ 0.27 −63 to −65 97–98 Yes Bacteroides Acr3-2 25 4.4 Æ 2.4 −20 to 0 60–100 Yes All hits ≥60% ID to query Faecalibacterium Acr3 25 5.0 Æ 2.6 −20 tp 0 60–100 Yes All hits ≥60% ID to query

niomna irbooyReports Microbiology Environmental ArsC (As(V) reductase) E. coli 1 0.04 −22 to −24 40 Yes Bacteroides 25 13.8 Æ 7.8 −20 to −81 51–99 No Adjacent to mismatch repair DNA glycosylase Faecalibacterium 24 8.8 Æ 6.1 −20 to 0 51–100 No Adjacent to ice-structuring protein ArsM (As(III) methyltransferase) Bacillus sp. CX-1 00 00NA Clostridium 7 1.1 Æ 0.35 −39 to −42 37–46 One hit adjacent All queries generated the same Rhodopseudomonas to an arsC hits and all Blastp identified Halobacterium as Bacteroides or Methanocella Parabacteroides arsenite methyltransferase , 12, Human microbial communities from the National Institute of Health, USA. 136 – 159 Arsenic and the gastrointestinal tract microbiome 149 source of arsenic to simulate soil-borne exposure routes. Considering the lower regions of the GIT, microbiome Consequently, a fair degree of caution must be used in inter- methanogens (Miller et al., 1982, 1986) should be capable preting such results; i.e. did the ArsM activity originate from of reductive methylation of arsenate (McBride and Wolfe, the faecal inoculants or the soils? 1971), attributable to a minor methyl transferase side reac- ArsP was found in all microbiomes. Very few ArsP pro- tion occurring between a methyl-cobalamin donor with teins have been characterized and so besides E-values and identity scores, our searches relied heavily on finding the As(III) [and other metal(oid)s like Sb, Se, Bi, Te] (Thomas ‘TPFCSCSXXP’ motif (Chen et al. 2015b) in amino acid et al., 2001). Bovine rumen studies documented quantitative alignments. At present, it is not clear if this motif alone would reduction of millimolar levels of As(V) to As(III), and that was fi suf ce to enable MAs(III) extrusion, but Chen et al. (2015b) faster and more complete with the provision of H2 as an showed that it is an essential property of a functional ArsP. exogenous electron donor as opposed to incubation under Genes encoding Acr3 were abundant for the dominant gen- only N2 (Herbel et al., 2002). Similarly, As(V) reduction was era Bacteroides and Faecalibacterium (found in all observed in hamster faecal pellets from As(V)-watered and 25 metagenomes), but much less so for E. coli and again consistent with the lower prevalence of the latter (Table 2). control animals; however, samples from As(V)-watered ani- Of the other Ars functions, ArsH was observed in a single mals were obviously primed with respect to gene expression individual, but no hits were found for ArsI, ArsJ and ArsN, or selection for As(V) tolerance/utilization, resulting in signifi- even when using less selective E-values (Table 2). cantly faster As(V) reduction than controls. Use of 73/74As We also surveyed nares, buccal and gingiva. There was a (V) radiotracers in similar experiments but at low concentra- limited degree of overlap as only a few individuals had all com- tion As(V) incubations (20 μM) achieved greater temporal ponents of their GIT sampled. Thus, searches of the upper sensitivity, demonstrating immediate As(V) reduction to GIT had to be expanded to include additional individuals. Of As(III). In this instance, arsenic was mostly recovered in the these, 21 buccal, eight nares and one gingiva metagenomes 73/74 yielded positive hits for arsenic genes. BLASTp searches were solid phase, likely as an As(III)2S3-precipitate facilitated negative for ArrA, ArxA, AioA and ArsM, however as with the by concurrent sulphate-reduction as the source of sulphide lower GIT, apparent ArsP homologues were present. Again, (Herbel et al., 2002). These experiments are important in ArsC and ArsB were identified in nares, buccal and gingiva illustrating how SRB activity can ameliorate toxicity; metagenomes. Three individuals had arsenic resistance i.e. precipitating the arsenic as a poorly soluble solid phase genes (i.e. arsRBC) in at least one component of the upper (likely orpiment), which is excreted and thus reducing host GIT and lower GIT, and one individual had resistance genes exposure. throughout (e.g., buccal, nares and lower GIT). Although the − metagenomes used were believed to be from healthy individ- In a comparison of wild type and As3mt knockout (lacks uals (based on the annotation), it is noteworthy that in addition host arsenic methyltransferase) mouse genotypes fed to E. coli and Actinobacillus sp., the ArsC homologues identi- As(V) (Naranmandura et al., 2013), As(III) was clearly the fied included those from Neisseria, Moraxella, Klebsiella and dominant arsenical found in the mutant mouse liver and Haemophilus species. urine. These mice were not germ-free and so the relative contribution of the GIT microbiomes to As(V) reduction is unknown but is a reasonable expectation that GIT microorganisms could contribute in this manner to pre-systemic Evidence of microbiome As interactions metabolism. This is supported by a SHIME-based study, In theory, almost any known microbial As transformation wherein the colon component registered near-complete should be at least possible in the GIT microbiome envi- As(V) reduction (Yin et al., 2015), implying the capacity to do ronment (an exception presumably being phototrophy- so in the GIT. linked As(III) oxidation as mentioned above). Residence Aerobic incubation of colon material collected from a time in the oral cavity is relatively short, but evidence SHIME system resulted in extensive As(III) oxidation (Alava suggests that it is still relevant for assessing the overall et al., 2011), illustrating that the genetic potential (presum- budget of arsenic ingestion, absorption and excretion. ably aio genes) was present and could be activated. How- Methylarsines are garlicky in odour and garlic odours in ever, under anaerobic conditions no As(III) oxidation was breath are often associated with acute arsenic exposure, apparent relative to initial levels of As(V) and thus the capac- and thus suggests arsenic metabolism in the oral cavity. ity to oxidize As(III) anaerobically was not detectable. Known Simulated digestions of certified reference seafoods with alternate electron acceptors that would support anaerobic − artificial saliva resulted in substantial release of arsenic As(III) oxidation (i.e. NO3 ) would not be expected to be pre- (Leufroy et al., 2012). In a modified SHIME experiment sent in a typical SHIME system (Molly et al., 1993) unless that included a simulated oral chamber containing sali- specificallytestedbyaddingthemintothemix. vary bacteria, partial digestion of arsenic-containing rice, At present, it is uncertain whether microbiome As(III) mussels and seaweed resulted in significant arsenic methylation activity is of positive or negative value to the release (primarily arsenosugars) from the mussels and host, but this is an example of a GIT microbiome As transfor- seaweed (Calatayud et al., 2018). mation activity that requires significant scrutiny in order to

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 150 T. R. McDermott, J. F. Stolz and R. S. Oremland understand the consequences for host arsenic exposure. homogenates from different human donors clearly demon- SHIME-based human microbiome studies have demonstrated how the GIT microorganisms provide a protective strated As methylation [DMAs(V) and MMAs(V)] in simu- measure by significantly reducing arsenic-induced mortality, lated transverse and descending colon sections (Yin et al., although varying between human stool donors (Coryell 2015; Yu et al., 2016), showing the genetic and physiologic et al., 2018). Illumina 16S metagenome analysis of these capacity to contribute this type of activity. The preponder- arsenic-exposed humanized mouse microbiomes suggested ance of pentavalent methylated species in various studies Faecalibacterium plays an important role, and indeed mice may be an artefact of abiotic auto-oxidation of trivalent spe- co-colonized with F. prausnitzii and E. coli (E. coli helps sta- cies under aerobic cultivation conditions, in aerobic purified bilize F. prausnitzii) mice were afforded significantly enzyme reaction conditions, and or handling of metabolite increased protection against arsenic relative to mice mono- extracts prior to analysis. Discussion and interpretation of associated with E. coli (Coryell et al., 2018). these and future experiments involving microbiome ArsM Evidence of potential microbiome arsenic thiolation activ- activity is not trivial nor simply an interesting biochemical ity stems from studies of rat caecal contents. Anaerobic puzzle. Depending on whether microbiome methylation in vitro incubations of samples fed As(V) resulted in the gen- activity generates trivalent or pentavalent arsenicals in large eration of As(V)S, dithioarsenate (As(V)S2), and As(V)S3, part dictates the relative toxicity of this reaction series and MAs(V), methyldithioarsenate (MAs(V)S2), DMAs(V)S2, outcomes, and potentially account for variability in arsenic- and methyltrithioarsenate (MAs(V)S3)(Pinyayevet al., related disease (Valenzuela et al., 2009; Engström et al., 2011). Similar experiments documented transformation of

2011, 2013). In the GIT environment, with the exception of DMAs(V) to DMAs(V)S, DMAs(V)S2, and trimethylarsine lumen epithelial (Fig. 4), the predominance of anaerobic sulphide (TMAs(V)S) (Kubachka et al., 2009). Presumably, conditions would be predicted to not favour the spontaneous SRB were present and responsible for generating the H2S oxidation of trivalent methylated arsenicals, and thus the (discussed above) and indeed evidence of SRB involve- question: Is microbiome As(III) methylation of any value to ment is strong. SHIME-based studies with human faecal the host or is it a source of significant toxicity? By contrast, and colon microbiome inoculants fed MAs(V) resulted in would the availability of oxygen near the lumen epithelium extensive MAs(V)S synthesis (Rubin et al., 2014). Produc- oxidize these trivalent species prior to uptake (Fig. 4)? tion of H2S and MAs(V)S were greatly enhanced by the Detection of arsH (one individual) and arsP (all addition of sulphate to the incubations and correlated with a metagenomes) in our survey (Table 2) further complicates substantial increase in Desulfovibrio desulfuricans (piger). this question. While we did not detect ArsI in the survey By contrast, H2S production and MAs(V)S synthesis were (Table 2), an in vitro dynamic gut simulator study (Rubin abolished by the addition of sodium molybdate, a well- et al., 2014) involving human faecal inoculant documented known SRB inhibitor (Kiene et al., 1986). Other interesting V III demethylation of MMA to As , indicating ArsI type activity. observations derived from this study include (i) H2S genera- Extrusion of MAs(III) from the bacterial cell (via ArsP), oxidation in these experiments varied significantly between indi- tion of MAs(III) to MAs(V) (via ArsH), or conversion of triva- vidual humans (healthy individuals range from 120 to lent organo-arsenicals back to As(III) (via ArsI) (Yoshinaga 450 μM) (Pitcher et al., 2000); and (ii) MAs(V)S was and Rosen, 2014) could be of critical importance. The occur- undetectable when H2S production was <~75 μM. Based on rence of such enzymes in the microbial world argues that these data and the arsenic thiolation literature reviewed microbial encounters with trivalent methylated species are above, one might conclude that the H2S:As ratio will likely not a rare event. ArsP, ArsH and ArsI detoxification activities need to be high (~100 H2S:1 As) to facilitate the formation of offer clear advantages to the bacterial cell. But what about these thiolated arsenicals. Relative differences in micro- the host? As is always the case, the occurrence, relative biome SRB abundance and activity, As exposure, and die- abundances and expression of these microbiome genes are tary sulphate would be expected to greatly influence arsenic the key considerations. thiolation.

A recent study by Coryell et al. (2018) used a combination Mineralization of soluble arsenic as As(III)2S3 and As4S4 of antibiotic-disturbed mouse microbiomes and germ-free serves to constrain As mobility, solubility (Cavalca et al., mice to directly address the question of whether a gut micro- 2013; Altun et al., 2014; Le Pape et al., 2017) and toxicity, biome can be of value to the host. In this study and similarly reducing overall host exposure and toxicity. Once formed, in a just recently reported study by Chi et al. (2019), the metal(loid) precipitates or solid forms can be quite stable. mouse gut microbiomes performed a bioaccumulation func- For instance, in clinical trials where human volunteers were tion, resulting in increased arsenic concentrations in mouse administered colloidal bismuth, it was estimated that ~99% stool while simultaneously reducing arsenic uptake and was excreted in faeces and only traces (pg/ml) were found accumulation in the mouse liver, spleen, heart and lung as CH3BiH2 in blood (Boertz et al., 2009). Long-term − (Coryell et al., 2018). Further, humanizing germ-free As3mt (2 months) administration of realgar (As4S4) to rats resulted knockout mice (hypersensitive to arsenic) with stool in the accumulation of DMAs(V) in livers, although there was

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 Arsenic and the gastrointestinal tract microbiome 151 no overt liver pathology (Zhou et al., 2019). In a case involv- sorption capacity of the iron solid phase (ferrihydrite) was ing a reported attempted suicide presented in an emergency exceeded; if the system was spiked with As:ferrihydrite sam- room (Buchanan et al., 2012), the individual ingested 84 g of ples where the As sorption capacity was not exceeded, then

As2S3 (‘chicken egg-size rock of orpiment’), but the patient’s bio-accessible As was below the detection limit of their ‘physical examination was unremarkable, and diagnostic assay. While this was basically a risk assessment study, it is tests included a normal electrolyte panel, a normal serum also informative in the context of host exposure if ingested lactate and a normal electrocardiogram.’ A urine sample As undergoes sorption reactions while transiting the GIT. taken 12 h post-admission registered 1490 ppb total arsenic, which is clearly well outside normal (<100 ppb As) Future directions and concluding remarks (Agency for Toxic Substances and Disease Registry, n.d.), but given the large bolus of As consumed it should be Looking forward, there is much to be done to understand viewed as exceptionally low. Clearly, a small portion of the the response of our microbiomes to the presence of As As became bioavailable, however, if the ingested As had oxyanions or any environmental toxicant. SHIME studies been a readily bioavailable form (e.g., As(III)), the patient have been valuable in demonstrating the potential for dif- would not have survived (acute dose ~70–180 mg As(III), ferent microbial As transformations, but this approach depending on body mass) (Agency for Toxic Substances essentially involves As enrichments that select for micro- and Disease Registry, n.d.). This clinical case is consistent biome members capable of metabolisms of interest. The with the view that mineralization of ingested As results in time lag typically observed suggests significant selection reduced bioavailability during GIT transit and thus reduced is occurring, implying that the particular functions/trans- toxicity. Depending on the form ingested, As bioavailability formations being observed are not dominant and do not will vary along the different phases of the GIT (Yin et al., represent equilibrium conditions of the sampled material. 2− 2016), and levels and rates of SO4 and As(V) reductions We suggest future work should shift to in situ oriented may have varying biochemical endpoints that vary in studies. toxicity. Much can be learned from work with germ-free mice Arsenic adsorption to solid phases will also limit As bio- wherein microbiomes can be customized to focus on spe- accessibility, again with the expectation being reduced over- cific arsenic transformations. The use of germ-free mice all exposure and toxicity in the GIT. Various studies suggest will also avoid the vendor-based variability in microbiome this may be important, although with caveats. As(V) binds composition (Velazquez et al., 2019) that will likely strongly to iron, manganese and aluminium oxides become very important as efforts begin to focus on spe- (Hingston et al., 1971; Gupta and Chen, 1978; Manning cific members of the GIT microbiome and for systemati- et al., 2002; Dixit and Hering, 2003). As(III) binding occurs cally comparing results between studies. Mouse with iron (hydr)oxides (Gupta and Chen, 1978), but this genotype can also be manipulated in the germ-free set- sorption to iron solid phases is relatively weak (Zobrist et al., ting, allowing for gene-for-gene strategies. To be sure, 2000), readily desorbing at circumneutral pH that occurs such approaches are reductionist, but nevertheless allow throughout most of the human colon. Environmental condi- for genetically clean comparisons that will provide impor- tions that favour microbial reduction of either tant baseline answers to modest questions such as: As(V) (particularly dissimilatory) and or Fe3+ do not favour ‘What microbial activities are harmful to the host?’ At this As(V) adsorption (Ahmann et al., 1994; Zobrist et al., 2000; juncture, the answer to even this simple query involves Tufano et al., 2008) suggesting the anaerobic GIT environ- significant conjecture. To the extent possible, future ment might not be conducive to this type of (biogeo)chemis- community-wide efforts would benefit from experimental try. In a SHIME study where FeCl3 was variously added mice having standardized GIT microbiomes, perhaps (0-3 mg/l) with 600 μg/l As(III), all added arsenic remained in involving a sponsored centralized service lab(s). solution in the simulated stomach (pH 2.0), but then The omics-level work that dominates the recent litera- decreased significantly in the small intestine (pH ~ 7), ture has been very important for demonstrating significant ascending colon (5.5 < pH < 5.9), transverse colon (dramatic in some cases) microbiome community shifts (6.0 < pH < 6.4) and descending colon (6.6 < pH < 6.9) upon As exposure. However, there is much to be done at chambers (Yu et al., 2016). Relative to zero Fe controls, iron more refined taxonomic levels. We predict that autecolog- amendments significantly reduced soluble As in the small ical strategies and approaches will be important, if not intestine but not in the other chambers. A human lower GIT essential, complements to the community-level efforts, experimental flow-through system examined the fate of and that some of the most important work will take place ferrihydrite-adsorbed As(V) with transit through the system at the strain level. Much effort is required to better under- in order to model bio-accessible As(V) as it is released stand how ingested arsenic is perceived by the microbes (by desorption) from the Fe(III) (Beak et al., 2006). Bio- along the different phases of the GIT. Clearly, as has accessibility of the As was governed by whether the arsenic been demonstrated in the above-reviewed studies, a

© 2019 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports, 12, 136–159 152 T. R. McDermott, J. F. Stolz and R. S. Oremland significant bolus brings about changes in the microbiome. flavor compound 3-methylbutanal from leucine catabolism But as an example of the fundamental work that is by Carnobacterium maltaromaticum LMA 28. Int J Food – needed, we are unaware of any studies that have Microbiol 157: 332 339. assessed the minimal As(V) or As(III) levels required to Agency for Toxic Substances and Disease Registry n.d. URL https://www.atsdr.cdc.gov/. elicit induction of the ars genes in vivo. What proportion Agioutantisa, P., and Koumandou, V.L. (2018) Bioenergetic of the ingested arsenic actually makes its way through diversity of the human gut microbiome. Meta Gene 16: the entire GIT? There is a broad bandwidth of possible 10–14. interactions predicted or possible (Figs 2 and 3), but Ahmann, D., Roberts, A.L., Krumholz, L.R., and Morel, F.M. characterizing them in reasonable detail will require con- M. (1994) Microbe grows by reducing arsenic. Nature siderable effort. What type of microbe–microbe interac- 371: 750. tions occur or are important, both in a positive as well as Ajees, A.A., Marapakala, K., Packianathan, C., Sankaran, B., and Rosen, B.P. (2012) Structure of an a negative sense? Microbiome responses and microbial As(III) S-adenosylmethionine methyltransferase: insights transformation notwithstanding, abiotic chemical reac- into the mechanism of arsenic biotransformation. Bio- tions occurring in the GIT are also an important topic chemistry 51: 5476–5485. (particularly thiolation reactions). Ajees, A.A., and Rosen, B.P. (2015) As(III) S- One final point bears mentioning. Microbiome investi- adenosylmethionine methyltransferases and other arsenic gations often emphasize gross population shifts binding proteins. Geomicrobiol J 32: 570–576. (e.g., 16 S rRNA genes) with respect to changing envi- Alava, P., Tack, F., Laing, G.D., and van de Wiele, T. (2011) ronmental parameters. But steadily introduced low micro- HPLC-ICP-MS method development to monitor arsenic speciation changes by human gut microbiota. Biomed molar quantities of As(V) into the GIT are less likely to Chromatogr 26: 524–533. cause notable shifts in sub-populations that can exploit Altun, M., Sahinkaya, E., Durukan, I., Bektas, S., and this oxyanion for growth. Thus, it is important that such Komnitsas, K. (2014) Arsenic removal in a sulfidogenic efforts focus on functional arsenic genes (e.g., arsC, fixed-bed column bioreactor. J Hazard Mater 269:31–37. arrA, aioA, arxA) in addition to phylogenetic composi- Andres, J., and Bertin, P.N. (2016) The microbial genomics tional parameters, and include transcriptomic approaches of arsenic. FEMS Microbiol Rev 40: 299–322. that would delineate whether such genes are expressed Bassis, C.M., Tang, A.L., Young, V.B., and Pynnonen, M.A. when the resident sub-population is exposed to arsenic (2014) The nasal cavity microbiota of healthy adults. Microbiome 2: 27. https://doi.org/10.1186/2049-2618- oxyanions. Moreover, because not all such genes have 2-27. fi been identi ed in enteric genomes despite the capacity Beak, D.G., Basta, N.T., Sheckel, K.G., and Traina, S.J. of some species to respire arsenate (e.g., Citrobacter (2006) Bioaccessibility of arsenic(V) bound to ferrihydrite strain TSA-1), there is ample room for discovery of ‘novel’ using a simulated gastrointestinal system. Environ Sci biochemical mechanisms and their encoding genes to Technol 40: 1364–1370. carry out such reactions (Switzer Blum et al., 2018). Bergquist, E.R., Fischer, R.J., Sugden, K.D., and Martin, B. While this type of research will be conducted in the con- D. (2009) Inhibition by methylated organoarsenicals of the respiratory 2-oxo-acid dehydrogenases. J Organomet text of human health and welfare, undoubtedly our under- Chem 694: 973–980. standing of how and why microbes react to arsenic will Bhattacharjee, P., Chatterjee, D., Singh, K.K., and Giri, A.K. greatly expand. (2013) Systems biology approaches to evaluate arsenic toxicity and carcinogenicity: an overview. Int J Hyg Envi- ron Health 216: 574–586. Acknowledgements Bisanz, J.E., Enos, M.K., Mwanga, J.R., Changalucha, J., T.R.M. acknowledges primary support for this review from Burton, J.P., Gloor, G.B., and Reid, G. (2014) Random- fl the National Institutes of General Medical Sciences and the ized open-label pilot study of the in uence of probiotics National Cancer Institute (R01CA215784), and additional and the gut microbiome on toxic metal levels in Tanzanian MBio support from the National Science Foundation Systems and pregnant women and school children. 5: e01580. Synthetic Biology Program (MCB-1714556), and Montana https://doi.org/10.1128/mBio.01580-14. Agricultural Experiment Station (Project 911310). J.F.S. and Boertz, J., Harmann, L.M., Sulkowski, M., Hippler, J., R.S.O. acknowledge support from NASA NNX09AW41G. Mosel, F., Diaz-Bone, R.A., et al. (2009) Determination of We also thank Alyssa Veliz in assisting with the upper GIT trimethylbismuth in the human body after ingestion of col- metagenome searches. loidal bismuth subcitrate. Drug Metab Dispos 37: 352–358. Bostick, B.C., Fendorf, S., and Brown, G.E., Jr. (2005) In situ analysis of thioarsenite complexes in neutral to alkaline References arsenic sulphide solutions. Mineral Mag 69: 781–795. Afzal, M.I., Delaunay, S., Paris, C., Borges, F., Revol- Branco, R., Francisco, R., Chung, A.P., and Morais, P.V. Junelles, A.M., and Cailliez-Grimal, C. (2012) Identification (2009) Identification of an aox system that requires cyto- of metabolic pathways involved in the biosynthesis of chrome c in the highly arsenic-resistant bacterium

Ochrobactrum tritici SCII24. Appl Environ Microbiol 75: Chen, J., Yoshinaga, M., and Rosen, B.P. (2018) The antibi- 5141–5147. otic action of methylarsenite is an emergent property of Bruhn, D.F., Li, J., Silver, S., Roberto, F., and Rosen, B.P. microbial communities. Mol Microbiol 111: 487–494. (1996) The arsenical resistance operon of IncN plasmid Chen, Y., Qin, N., Guo, J., Qian, G., Fang, D., Shi, D., et al. R46. FEMS Microbiol Lett 139: 149–153. (2014) Functional gene arrays-based analysis of fecal Buchanan, J.A., Eberhardt, A., Tebb, Z.D., Heard, K., microbiomes in patients with liver cirrhosis. BMC Geno- Wendlandt, R.F., and Kosnett, M.J. (2012) Massive mics 15: 753. human ingestion of orpiment (arsenic trisulfide). J Emerg Chi, L., Gao, B., Tu, P., Liu, C.W., Xue, J., Lai, Y., et al. Med 44: 367–372. (2018) Individual susceptibility to arsenic-induced dis- Byrne, S., Amarasiriwardena, B., Bandak, L., Bartkus, J., eases: the role of host genetics, nutritional status, and the – Kane, J., Jones, J., et al. (2010) Were Chinchorros gut microbiome. Mamm Genome 29:63 79. exposed to arsenic? Arsenic determination in Chinchorro Chi, L., Xue, J., Tu, J., Lai, Y., Ru, H., and Lu, K. (2019) Gut mummies’ hair by laser ablation inductively coupled microbiome disruption altered the biotransformation and plasma-mass spectrometry (LA-ICP-MS). Microchem J liver toxicity of arsenic in mice. Arch Toxicol 93:25–35. 94:28–35. Coryell, M., McAlpine, M., Pinkham, N.V., McDermott, T.R., Cai, L., Rensing, C., Li, X., and Wang, G. (2009) Novel gene and Walk, S.T. (2018) The gut microbiome is required for clusters involved in arsenite oxidation and resistance in full protection against acute arsenic toxicity in mouse two arsenite oxidizers: Achromobacter sp. SY8 and Pseu- models. Nat Commun 9: 5424. ’ domonas sp. TS44. Appl Microbiol Biotechnol 83: Cubadda, F., D Amato, M., Mancini, F.R., Aureli, F., 715–725. Raggi, A., Busani, L., and Mantovani, A. (2015) Assessing Calatayud, M., Xiong, C., Laing, D., Raber, G., human exposure to inorganic arsenic in high-arsenic Francesconi, K., and van de Wiele, T. (2018) Salivary and areas of Latium: a biomonitoring study integrated with indi- – gut microbiomes play a significant role in in vitro oral bio- cators of dietary intake. Ann Ig 27:39 51. accessibility, biotransformation, and intestinal absorption Dahan, D., Jude, B.A., Lamendella, R., Keesing, F., and of arsenic from food. Environ Sci Technol 52: Perron, G.G. (2018) Exposure to arsenic alters the micro- fi 14422–14435. biome of larval zebra sh. Front Microbiol 9: 1323. https:// doi.org/10.3389/fmicb.2018.01323. Carbonero, F., and Gaskins, H.R. (2013) Sulfate-reducing Dembitsky, V.M., and Levitsky, D.O. (2004) Arsenolipids. bacteria in the human gut microbiome. In Encyclopedia of Prog Lipid Res 43: 403–448. Metagenomics. Nelson, K. (ed). New York, NY: Springer. Dietert, R.R., and Silbergeld, E.K. (2015) Biomarkers for the Cavalca, L., Corsini, A., Zaccheo, P., Andreoni, V., and 21st century: listening to the microbiome. Toxicol Sci 144: Muyzer, G. (2013) Microbial transformations of arsenic: 208–216. perspectives for biological removal of arsenic from water. Dixit, S., and Hering, J.G. (2003) Comparison of arsenic(V) Future Microbiol 8: 753–768. and arsenic(III) sorption onto iron oxide minerals: implica- Challenger, F. (1945) Biological methylation. Chem Rev 36: tions for arsenic mobility. Environ Sci Technol 37: 315–361. 4182–4189. Chauhan, N.S., Ranjan, R., Purohit, H.J., Kalia, V.C., and Donaldson, G.P., Lee, S.M., and Mazmanian, S.K. (2016) fi Sharma, R. (2009) Identi cation of genes conferring arse- Gut biogeography of the bacterial microbiota. Nat Rev fl nic resistance to Escherichia coli from an ef uent treat- Microbiol 14:20–32. ment plant sludge metagenomic library. FEMS Microbiol Dong, X., Shulzhenko, N., Lemaitre, J., Greer, R.L., – Ecol 67: 130 139. Peremyslova, K., Quamruzzaman, Q., et al. (2017) Arse- Chen, J., Bhattacharjee, H., and Rosen, B.P. (2015a) ArsH nic exposure and intestinal microbiota in children from is an organoarsenical oxidase that confers resistance to Sirajdikhan, Bangladesh. PLoS One 12: e0188487. trivalent forms of the herbicide monosodium Dunivin, T.K., Yeh, S.Y., and Shade, A. (2019) A global sur- methylarsenate and the poultry growth promoter rox- vey of arsenic-related genes in soil microbiomes. BMC – arsone. Mol Microbiol 96: 1042 1052. Biol 17: 45. Chen, J., Li, J., Jiang, X., and Rosen, B.P. (2017) Conserved Duval, S., Ducluzeau, A.L., Nitschke, W., and Schoepp- fi cysteine residues determine substrate speci city in a Cothenet, B. (2008) Enzyme phylogenies as markers for novel As(III) S-adenosylmethionine methyltransferase the oxidation state of the environment: the case of respira- – from Aspergillus fumigatus. Mol Microbiol 104: 250 259. tory arsenate reductase and related enzymes. BMC Evol Chen, J., Madegowda, M., Bhattacharjee, H., and Rosen, B. Biol 8: 206. P. (2015b) ArsP: a methylarsenite efflux permease. Mol Eckburg, P.B., Bik, E.M., Bernstein, C.N., Purdom, E., Microbiol 98: 625–635. Dethlefsen, L., Sargent, M., et al. (2005) Diversity of the Chen, J., and Rosen, B.P. (2016) Organoarsenical biotrans- human intestinal microbial flora. Science 308: 1635–1638. formations by Shewanella putrefaciens. Environ Sci Engström, K., Vahter, M., Mlakar, S.J., Concha, G., Technol 50: 7956–7963. Nermell, B., Raqib, R., et al. (2011) Polymorphisms in Chen, J., Yoshinaga, M., Garbinski, L.D., and Rosen, B.P. arsenic(+III oxidation state) methyltransferase (AS3MT) (2016) Synergistic interaction of glyceraldehydes- predict gene expression of AS3MT as well as arsenic 3-phosphate dehydrogenase and ArsJ, a novel metabolism. Environ Health Perspect 119: 182–188. organoarsenical efflux permease, confers arsenate resis- Engström, K.S., Hossain, M.B., Lauss, M., Ahmed, S., tance. Mol Microbiol 100: 945–953. Raqib, R., Vahter, M., and Broberg, K. (2013) Efficient

arsenic metabolism—the AS3MT haplotype is associated Gupta, S.K., and Chen, S.Y. (1978) Arsenic removal by with DNA methylation and expression of multiple genes adsorption. J Water Pollut Control Fed 50: 493–506. around AS3MT. PLoS One 8: e53732. Hartig, C., Lohmayer, R., Kolb, S., Horn, M.A., Inskeep, W. Espey, M.G. (2013) Role of oxygen gradients in shaping P., and Planer-Friedrich, B. (2014) Chemolithotrophic redox relationships between the human intestine and its growth of the aerobic hyperthermophilic bacterium microbiota. Free Radical Biol Med 55: 130–140. Thermocrinis ruber OC14/7/2 on monothioarsenate and Faita, F., Cori, L., Bianchi, F., and Andreassi, M.G. (2013) arsenite. FEMS Microbiol Ecol 90: 747–760. Arsenic-induced genotoxicity and genetic susceptibility to Hartig, C., and Planer-Friedrich, B. (2012) Thioarsenate arsenic-related pathologies. Int J Environ Res Public transformation by filamentous microbial mats thriving in an Health 10: 1527–1546. alkaline, sulfidic hot spring. Environ Sci Technol 46: Fan, C., Liu, G., Long, Y., Rosen, B., and Cai, Y. (2018) 4348–4356. Thiolation in arsenic metabolism: a chemical perspective. Herbel, M.J., Switzer Blum, J., Hoeft, S.E., Cohen, S.M., Metallomics 10: 1368–1382. Arnold, L.L., Lisak, J., et al. (2002) Dissimilatory arsenate Forsberg, C.W. (1978) Some effects of arsenic on the rumen reductase activity and arsenate-respiring bacteria in microflora: an in vitro study. Can J Microbiol 24:36–44. bovine rumen fluid, hamster feces, and the termite hind- Francesconi, K.A. (2010) Arsenic species in seafood: origin gut. FEMS Microbiol Ecol 41:59–67. and human health implications. Pure Appl Chem 82: Hernandez-Maldonado, J., Sanchez-Sedillo, B., 373–381. Stoneburner, B., Boren, A., Miller, L., McCann, S., et al. Garcia-Salgado, S., Raber, G., Raml, R., Magnes, C., and (2016) The genetic basis of anoxygenic photosynthetic Francesconi, K.A. (2012) Arsenosugar phospholipids and arsenite oxidation. Environ Microbiol 19: 130–141. arsenic hydrocarbons in two species of brown macro- Hingston, F.J., Posner, A.M., and Quirk, J.P. (1971) Compet- algae. Environ Chem 9:63–66. itive adsorption of negatively charged ligands on oxide Gaulke, C.A., Rolshoven, J., Wong, C.P., Hudson, L.G., surfaces. Discuss Faraday Soc 52: 334–342. Ho, E., and Sharpton, T.J. (2018) Marginal zinc deficiency Hinrichsen, S., Geist, F., and Planer-Friedrich, B. (2015) and environmentally relevant concentrations of arsenic Inorganic and methylated thioarsenates pass the gastroin- elicit combined effects on the gut microbiome. mSphere 3: testinal barrier. Chem Res Toxicol 28: 1678–1680. e00521–e00518. https://doi.org/10.1128/mSphere. Hoeft, S.E., Kulp, T.R., Han, S., Lanoil, B., and Oremland, R. 00521-18. S. (2010) Coupled arsenotrophy in a photosynthetic hot Geiger, O., Röhrs, V., Weissenmayer, B., Finan, T.M., and spring biofilm from Mono Lake, California. Appl Environ Thomas-Oates, J.E. (1999) The regulator gene phoB Microbiol. 76: 4633–4639. mediates phosphate stress controlled synthesis of the Hoeft, S.E., Switzer Blum, J., Stolz, J.F., Tabita, F.R., membrane lipid diacylglyceryl-N,N,N-trimethylhomoserine Witte, B., King, G.M., et al. (2007) Alkalilimnicola ehlichii, in Rhizobium (Sinorhizobium) meliloti. Mol Microbiol 32: sp. nov., a novel, arsenite-oxidizing haloalkalipihilic 63–73. γ-Proteobacterium capable of chemoautotrophic or hetero- Gibson, G.R., Macfarlane, G.T., and Cummings, J.H. (1993) trophic growth with nitrate or oxygen as the electron Sulfate reducing bacteria and hydrogen metabolism in the acceptor. Int J Syst Evol Microbiol 57: 504–512. human large-intestine. Gut 34: 437–439. Hoen, A.G., Madan, J.C., Li, Z., Coker, M., Lundgren, S.N., Glabonjat, R.A., Raber, G., Jensen, K.B., Guttenberger, N., Morrison, H.G., et al. (2018) Sex-specific associations of Zangger, K., and Francesconi, K.A. (2017) A 2-O- infants’ gut microbiome with arsenic exposure in a US methylriboside unknown outside the RNA world contains population. Sci Rep 8: 12627. arsenic. Angew Chem Int Ed Engl 56: 11963–11965. Huang, S., Liu, C., Wang, Y., and Zhan, H. (2014) Multivari- Glasser, N., Oyala, P., Osborne, T., Santini, J., and ate analysis of the heterogeneous geochemical processes Newman, D.K. (2018) Structural and mechanistic analysis controlling arsenic enrichment in a shallow groundwater of the arsenate respiratory reductase provides insight into system. J Environ Sci Health Part A 49: 478–489. environmental arsenic transformations. Proc Natl Acad Inskeep, W.P., McDermott, T.R., and Fendorf, S.E. (2001) Sci U S A 115: E8614–E8623. Arsenic (V)/(III) cycling in soils and natural waters: chemi- Gokulan, K., Arnold, M.G., Jensen, J., Vanlandingham, M., cal and microbiological processes. In Environmental Twaddle, N.C., Doerge, D.R., et al. (2018) Exposure to Chemistry of Arsenic. Frankenberger, W.F., Jr., and arsenite in CD-1 mice during juvenile and adult stages: Macy, J.M. (eds). NY: Marcell Dekker, pp. 183–215. effects on intestinal microbiota and gut-associated Isokpehi, R.D., Udensi, U.K., Simmons, S.S., Hollman, A.L., immune status. MBio 9: e01418. Cain, A.E., Olofinae, S.A., et al. (2014) Evaluative profiling Grill, E., Loffler, S., Winnacker, E.L., and Zenk, M.H. (1989) of arsenic sensing and regulatory systems in the human Phytochelatins, the heavy-metal-binding-peptides of microbiome project genomes. Microbiol Insights 7:25–34. plants, are synthesized from glutathione by a specific Jiang, S., Lee, J.-H., Kim, M.-G., Myung, N.V., gamma-glutamylcysteine dipeptidyl transpeptidase (phyto- Fredrickson, J.K., Sadowsky, M.J., and Hur, H.-G. (2009) chelatin synthase). Proc Natl Acad Sci U S A 86: Biogenic formation of As-S nanotubes by diverse 6838–6842. Shewanella strains. Appl Environ Microbiol 5: 6896–6899. Grill, E., Winnacker, E.L., and Zenk, M.H. (1985) Juhasz, A.L., Weber, J., and Smith, E. (2011) Influence of Phytochelatins: the principal heavy-metal complexing pep- saliva, gastric and intestinal phases on the prediction of tides of higher plants. Science 230: 674–676. As relative bioavailability using the Unified Bioaccessibility

Research Group of Europe Method (UBM). J Hazard Complete removal of arsenic and zinc from a heavily con- Mater 197: 161–168. taminated acid mine drainage via an indigenous SRB con- Kang, Y.-S., Heinemann, J., Bothner, B., Rensing, C., and sortium. J Hazard Mater 321: 764–772. McDermott, T.R. (2012) Integrated co-regulation of bacte- Lee, J.H., Kim, M.G., Yoo, B., Myung, N.V., Maeng, J., rial arsenic and phosphorus metabolisms. Environ Lee, T., et al. (2007) Biogenic formation of photoactive Microbiol 14: 3097–3109. arsenic-sulfide nanotubes by Shewanella sp. strain HN- Kashyap, D.R., Botero, L.M., Franck, W.L., Hassett, D.J., 41. Proc Natl Acad Sci U S A 104: 20410–20415. and McDermott, T.R. (2006) Complex regulation of arse- Lett, M.C., Muller, D., Lièvremont, D., Silver, S., and nite oxidation in Agrobacterium tumefaciens. J Bacteriol Santini, J. (2012) Unified nomenclature for genes involved 188: 1081–1088. in prokaryotic aerobic arsenite oxidation. J Bacteriol 194: Ke, C., Zhao, C., Rensing, C., Yang, S., and Zhang, Y. 207–208. (2018) Characterization of recombinant E. coli expressing Leufroy, A., Noel, L., Beauchemin, D., and Guerin, T. (2012) arsR from Rhodopseudomonas palustris CGA009 that Bioaccessibility of total arsenic and arsenic species in displays highly selective arsenic adsorption. Appl seafood as determined by a continuous online leaching Microbiol Biotechnol 102: 6247–6255. method. Anal Bioanal Chem 402: 2849–2859. Kiene, R.P., Oremland, R.S., Catena, A., Miller, L.G., and Li, H., Li, M., Huang, Y., Rensing, C., and Wang, G. (2013) Capone, D.G. (1986) Metabolism of reduced methylated In silico analysis of bacterial arsenic islands reveals sulfur compounds in anaerobic sediments and by a pure remarkable synteny and functional relatedness between culture of an estuarine methanogen. Appl Environ arsenate and phosphate. Front Microbiol 4: 347. – Microbiol 52: 1037 1045. Li, J., Pawitwar, S.S., and Rosen, B.P. (2016) The Kim, Y.-T., Lee, H., Yoon, H.-O., and Woo, N.C. (2016) organoarsenical biocycle and the primordial antibiotic Kinetics of dimethylated thioarsenicals and the formation methylarsenite. Metallomics 8: 1047–1055. of highly toxic dimethylmonothioarsinic acid in environ- Li, J., Sun, Y., Zhang, X., Hu, Y., Li, T., Zhang, X., et al. ment. Environ Sci Technol 50: 11637–11645. (2018) A methyltransferase gene from arbuscular mycor- Koechler, S., Arsene-Ploetze, F., Brochier-Armanet, C., rhizal fungi involved in arsenic methylation and volatiliza- Goulhen-Chollet, F., Heinrich-Salmeron, A., Jost, B., et al. tion. Chemosphere 209: 392–400. (2015) Constitutive arsenite oxidase expression detected Lin, Y.F., Walmsley, A.R., and Rosen, B.P. (2006) An arse- in arsenic-hypertolerant Pseudomonas xanthomarina S11. nic metallochaperone for an arsenic detoxification pump. Res Microbiol 166: 205–214. Proc Natl Acad Sci U S A 103: 15617–15622. Koechler, S., Cleiss-Arnold, J., Proux, C., Sismeiro, O., Liu, C.-W., Chi, L., Tu, P., Xue, J., Ru, H., and Lu, K. (2018) Dillies, M.A., Goulhen-Chollet, F., et al. (2010) Multiple Isobaric labeling quantitative metaproteomics for the study controls affect arsenite oxidase gene expression in Her- of gut microbiome response to arsenic. J Proteome Res miniimonas arsenicoxydans. BMC Microbiol 10: 53. 18: 970–981. Koskinen, K., Reichert, J.L., Hoier, S., Schachenreiter, J., Duller, S., Moissl-Eichinger, C., and Schoept, V. (2018) Liu, G., Liu, M., Kim, E.H., Maaty, W.S., Bothner, B., Lei, B., The nasal microbiome mirrors and potentially shapes et al. (2012) A periplasmic arsenite-binding protein olfactory function. Sci Rep 8: 1296. involved in regulating arsenite oxidation. Environ Microbiol 14: 1624–1634. Kubachka, K.M., Kohan, M.C., Conklin, S.D., Herbin- Davis, K., Creed, J.T., and Thomas, D.J. (2009) In vitro Liu, J., Lu, Y., Wu, Q., Goyer, R.A., and Waalkes, M.P. biotransformation of dimethylarsinic acid and (2008) Mineral arsenicals in traditional medicines: orpi- trimethylarsine oxide by anaerobic microflora of mouse ment, realgar, and arsenolite. J Pharmacol Exp Ther 326: – cecum analyzed by HPLC–ICP-MS and HPLC–ESI-MS. 363 368. J Anal At Spectrom 24: 1062–1068. Lu, K., Abo, R.P., Schlieper, K.A., Graffam, M.E., Levine, S., Kulp, T.R., Hoeft, S.E., Asao, M., Madigan, M.T., Wishnok, J.S., et al. (2014) Arsenic exposure perturbs the fi Hollibaugh, J.T., Fisher, J.C., et al. (2008) Arsenic(III) gut microbiome and its metabolic pro le in mice: an inte- fuels anoxygenic photosynthesis in hot spring biofilms grated metagenomics and metabolomics analysis. Environ – from Mono Lake, California. Science 321: 967–970. Health Perspect 122: 284 291. Kuramata, M., Sakakibara, F., Kataoka, R., Yamazaki, K., Lu, K., Cable, P.H., Abu, R.P., Ru, H., Graffam, M.E., Baba, K., Ishizaka, M., et al. (2016) Arsinothricin, a novel Schlieper, K.A., et al. (2013) Gut microbiome perturbations organoarsenic species produced by a rice rhizosphere induced by bacterial infection affect arsenic biotransforma- bacterium. Environ Chem 13: 723–731. tion. Chem Res Toxicol 26: 1893–1903. Lang, W.K., Glassey, K., and Archibald, A.R. (1982) Influ- Lu, K., Mahbub, R., Cable, P.H., Ru, H., Parry, N.A., ence of phosphate supply on teichoic acid and teichuronic Bodnar, W.M., et al. (2014b) Gut microbiome phenotypes acid content of Bacillus subtilis cell walls. J Bacteriol 151: driven by host genetics affect arsenic metabolism. Chem 367–375. Res Toxicol 27: 172–174. Le Chatelier, E., Nielsen, T., Qin, J., Prifti, E., Hildebrand, F., Macur, R.E., Jackson, C.R., McDermott, T.R., and Falony, G., et al. (2013) Richness of human gut micro- Inskeep, W.P. (2004) Bacterial populations associated biome correlates with metabolic markers. Nature 500: with the oxidation and reduction of arsenic in an unsatu- 541–546. rated soil. Environ Sci Technol 38: 104–111. Le Pape, P., Battaglia-Brunet, F., Parmentier, M., Macur, R.E., Wheeler, J.T., McDermott, T.R., and Joulian, C., Gassaud, C., Fernandez-Rojo, L., et al. (2017) Inskeep, W.P. (2001) Microbial populations associated

with the reduction and enhanced mobilization of arsenic in methyltransferase knockout mice exposed to arsenate. mine tailings. Environ Sci Technol 35: 3676–3682. Anal Bioanal Chem 405: 1885–1891. Macy, J.M., Santini, J.M., Pauling, B.V., O’Neill, A.H., and Naranmandura, H., and Suzuki, K.T. (2008) Formation of Sly, L.I. (2000) Two new arsenate/sulfate-reducing bacte- dimethylthioarsenicals in red blood cells. Toxicol Appl ria: mechanisms of arsenate reduction. Arch Microbiol Pharmacol 227: 390–399. 173:49–57. Naranmandura, H., Suzuki, N., Iwata, K., Hirano, S., and Manning, B.A., Fendorf, S.E., Bostick, B., and Suarez, D.L. Suzuki, T. (2007b) Arsenic metabolism and thioarsenicals (2002) Arsenic(III) oxidation and arsenic(V) adsorption in hamsters and rats. Chem Res Toxicol 20: 616–624. reactions on synthetic birnessite. Environ Sci Technol 3: Nardone, G., and Compare, D. (2015) The human gastric 976–981. microbiota: is it time to rethink the pathogenesis of stom- McBride, B.C., and Wolfe, R.S. (1971) Biosynthesis of ach diseases? United Eur Gastroenterol J 3: 255–260. dimethylarsine by Methanobacterium. Biochemistry 10: Naujokas, M.F., Anderson, B., Ahsan, H., Aposhian, H.V., – 4312 4317. Graziano, J.H., Thompson, C., and Suk, W.A. (2013) The McEwan, A.G., Ridge, J.P., McDevitt, C.A., and broad scope of health effects from chronic arsenic expo- Hugenholtz, P. (2002) The DMSO reductase family of sure: update on a worldwide public health problem. Envi- microbial molybdenum enzymes; molecular properties and ron Health Perspect 121: 295–302. role in the dissimilatory reduction of toxic elements. Geo- Newman, D.K., Beveridge, T.J., and Morel, F.M.M. (1997) – microbiol J 19:3 21. Precipitation of arsenic trisulﬁde by Desulfotomaculum Miller, T.L., Wolin, M.J., de Marcario, C., and Marcario, A.J. auripigmentum. Appl Environ Microbiol 63: 2022–2028. Methanobrevibacter smithii (1982) Isolation of from NIH HMP Working Group, Peterson, J., Garges, S., human feces. Appl Environ Microbiol 43: 227–232. Giovanni, M., McInnes, P., Wang, L., et al. (2009) The Miller, T.L., Wolin, M.J., and Kusel, E.S. (1986) Isolation and NIH Human Microbiome Project. Genome Res 19: characterization of methanogens from animal feces. Syst 2317–2323. Appl Microbiol 8: 234–238. O’ Toole, P.W. (2012) Changes in the intestinal microbiota Molly, K., Vande Woestyne, M., and Verstraete, W. (1993) from adulthood through to old age. Clin Microbiol Infect 4: Development of a S-step multi-chamber reactor as a simu- 44–46. lation of the human intestinal microbial ecosystem. Appl Ochi, T., Kita, K., Suzuki, T., Rumpler, A., Goessler, W., and Microbiol Biotechnol 39: 254–258. Francescon, K.A. (2008) Cytotoxic, genotoxic and cell- Morelli, E., Mascherpa, M.C., and Scarano, G. (2005) Bio- cycle disruptive effects of thio-dimethylarsinate in cultured synthesis of phytochelatins and arsenic accumulation in human cells and the role of glutathione. Toxicol Appl the marine microalga Phaeodactylum tricornutum in Pharmacol 228:59–67. response to arsenate exposure. Biometals 18: 587–593. Oremland, R.S., Dowdle, P.R., Hoeft, S., Sharp, J.O., Mukhopadhyay, R., and Rosen, B.P. (2002) Arsenate reduc- Schaefer, J.K., Miller, L.G., et al. (2000) Bacterial dissimi- tases in prokaryotes and eukaryotes. Environ Health Per- latory reduction of arsenate and sulfate in meromictic spect 110: 745–748. Mono Lake, California. Geochim Cosmochim Acta 64: Muller, D., Lievremont, D., Simeonova, D.D., Hubert, J.-C., 3073–3084. and Lett, M.-C. (2003) Arsenite oxidase aox genes from a Oremland, R.S., Hoeft, S.E., Bano, N., Hollibaugh, R.A., and metal-resistant betaproteobacterium. J Bacteriol 185: 135–141. Hollibaugh, J.T. (2002) Anaerobic oxidation of arsenite in Mono Lake water and by a facultative, arsenite-oxidizing Murphy, J., and Saltikov, C.W. (2009) The ArsR repressor chemoautotroph, strain MLHE-1. Appl Environ Microbiol mediates arsenite-dependent regulation of arsenate respi- 68: 4795–4802. ration and detoxiﬁ cation operons of Shewanella sp. strain ANA-3. J Bacteriol 191: 6722–6731. Oremland, R.S., and Stolz, J.F. (2005) Arsenic, microbes – Nadar, V.S., Chen, J., Dheeman, D.S., Galvin, A.E., and contaminated aquifers. Trends Microbiol 13:45 49. Yoshinaga-Sakurai, K., Kandavelu, P., et al. (2019) Paez-Espino, D., Tamames, J., de Lorenzo, V., and Arsinothricin, an arsenic-containing non-proteinogenic Canovas, D. (2009) Microbial responses to environmental – amino acid analog of glutamate, is a broad-spectrum anti- arsenic. Biometals 22: 117 130. biotic. Commun Biol 2: 131. https://doi.org/10.1038/ Pandey, N., and Bhatt, R. (2015) Arsenic resistance and s42003-019-0365-y. accumulation by two bacteria isolated from a natural arse- Naranmandura, H., Ibata, K., and Suzuki, K.T. (2007) Toxic- nic contaminated site. J Basic Microbiol 55: 1275–1286. ity of dimethylmonothioarsinic acid toward human epider- Pass, D.A., Morgan, A.J., Read, D.S., Field, D., moid carcinoma A431 cells. Chem Res Toxicol 20: Weightman, A.J., and Kille, P. (2015) The effect of anthro- 1120–1125. pogenic arsenic contamination on the earthworm micro- Naranmandura, H., Ogra, Y., Iwata, K., Lee, J., Suzuki, K.T., biome. Environ Microbiol 17: 1884–1896. Weinfeld, M., and Le, X.C. (2009) Evidence for toxicity dif- Pérez-Cobas, A.E., Artacho, A., Knecht, H., Ferrús, M.L., ferences between inorganic arsenite and thioarsenicals in Friedrichs, A., Ott, S.J., et al. (2013) Differential effects of human bladder cancer cells. Toxicol Appl Pharmacol 238: antibiotic therapy on the structure and function of human 133–140. gut microbiota. PLoS One 8: e80201. Naranmandura, H., Rehman, K., Le, X.C., and Thomas, D.J. Pinyayev, T.S., Kohan, M.J., Herbin-Davis, K., Creed, J.T., (2013) Formation of methylated oxyarsenicals and and Thomas, D.J. (2011) Preabsorptive metabolism of thioarsenicals in wild-type and arsenic (+3 oxidation state) sodium arsenate by anaerobic microbiota of mouse cecum

forms a variety of methylated and thiolated arsenicals. Rosen, B.P., Dey, S., Dou, D., Ji, G., Kaur, P., Chem Res Toxicol 24: 475–477. Ksenzenko, M.Y., et al. (1992) Evolution of an ion- Pitcher, M.L.C., Beatty, E.R., and Cummings, J.H. (2000) translocating ATPase. Ann N Y Acad Sci 671: 257–272. The contribution of sulphate reducing bacteria and Rubin, S.S.C.D.C., Alava, P., Zekker, I., Du Laing, G., and 5-aminosalicylic acid to faecal sulphide in patients with Van de Wiele, T. (2014) Arsenic thiolation and the role of ulcerative colitis. Gut 46:64–72. sulfate-reducing bacteria from the human intestinal tract. Planar-Friederich, B., Härtig, C., Lohmayer, R., Suess, E., Environ Health Perspect 122: 817–822. McCann, S.H., and Oremland, R.S. (2015) Anaerobic Saltikov, C., and Newman, D.K. (2003) Expression dynamics chemolithotrophic growth of the halophilic bacterium strain of arsenic respiration and detoxification in Shewanella MLMS-1 by disproportionation of monothioarsenate. Envi- sp. strain ANA-3. Proc Natl Acad Sci U S A 100: ron Sci Technol 49: 6554–6563. 10983–10988. Planer-Friedrich, B., London, J., McCleskey, R.B., Saltikov, C.W., Wildman, R., Jr., and Newman, D.K. (2005) Nordstrom, D.K., and Wallschläger, D. (2007) Expression dynamics of arsenic respiration and detoxifica- Thioarsenates in geothermal waters of Yellowstone tion in Shewanella sp. strain ANA-3. J Bacteriol 187: National Park: determination, preservation, and geochemi- 7390–7396. Environ Sci Technol – cal importance. 41: 5245 5251. Salzberg, L.I., Botella, E., Hokamp, K., Antelmann, H., Qin, J., Lehr, C.R., Yuan, C., Le, X.C., McDermott, T.R., and Maaß, S., Becher, D., et al. (2015) Genome-wide analysis Rosen, B.P. (2009) Biotransformation of arsenic by a Yel- of phosphorylated PhoP binding to chromosomal DNA lowstone thermoacidophilic eukaryotic alga. Proc Natl reveals several novel features of the PhoPR-mediated – Acad Sci U S A 106: 5213 5217. phosphate limitation response in Bacillus subtilis. Qin, J., Li, R., Raes, J., Arumugam, M., Burgdorf, K.S., J Bacteriol 197: 1492–1506. Manichanh, C., et al. (2010) A human gut microbial gene San Francisco, M.J., Hope, C.L., Owolabi, J.B., Tisa, L.S., catalogue established by metagenomic sequencing. and Rosen, B.P. (1990) Identification of the met- – Nature 464:59 65. alloregulatory element of the plasmid-encoded arsenical Qin, J., Rosen, B.P., Zhang, Y., Wang, G., Franke, S., and resistance operon. Nucleic Acids Res 18: 619–624. Rensing, C. (2006) Arsenic detoxification and evolution of Sanders, O.I., Rensing, C., Kuroda, M., Mitra, B., and trimethylarsine gas by a microbial arsenite S- Rosen, B.P. (1997) Antimonite is accumulated by the glyc- adenosylmethionine methyltransferase. Proc Natl Acad erol facilitator GlpF in Escherichia coli. J Bacteriol 179: Sci U S A 103: 2075–2080. 3365–3367. Raml, R., Rumpler, A., Goessler, W., Vahter, M., Li, L., Santini, J.E., Sly, L.I., Schnagl, R.D., and Macy, J.M. (2000) Ochi, T., and Francesconi, K.A. (2007) Thio- A new chemolithotrophic arsenite-oxidizing bacterium iso- dimethylarsinate is a common metabolite in urine samples lated from a gold mine: phylogenetic, physiological, and from arsenic-exposed women in Bangladesh. Toxicol Appl preliminary biochemical studies. Appl Environ Microbiol Pharmacol 222: 374–380. 66:92–97. Rawle, R.S., Kang, Y.-S., Bothner, B., Wang, G., and Santini, J.M., Kappler, U., Ward, S.A., Honeychurch, M.J., McDermott, T.R. (2019) Transcriptomics analysis defines vanden Hoven, R.N., and Bernhardt, P.V. (2007) The NT- global cellular response of Agrobacterium tumefaciens 5A 26 cytochrome c552 and its role in arsenite oxidation. Bio- to arsenite exposure regulated through the histidine chim Biophys Acta 1767: 189–196. kinases PhoR and AioS. Environ Microbiol 21: 2659–2676. https://doi.org/10.1111/1462-2920.14577. Schramm, A., Davidson, S.K., Dodsworth, J.A., Drake, H.L., Rhine, E.D., Phelps, C.D., and Young, L.Y. (2006) Anaerobic Stahl, D.A., and Dubilier, N. (2003) Acidovorax-like symbi- arsenite oxidation by novel denitrifying isolates. Environ onts in the nephridia of earthworms. Environ Microbiol 5: – Microbiol 8: 899–908. 804 809. Rochette, E.A., Bostick, B.C., Li, G., and Fendorf, S. (2000) Shen, S., Li, X.-F., Cullen, W.R., Weinfeld, M., and Le, X.C. Kinetics of arsenate reduction by dissolved sulfide. Envi- (2013) Arsenic binding to proteins. Chem Rev 113: – ron Sci Technol 34: 4714–4720. 7769 7792. Rodriguez-Freire, L., Moore, S.E., Sierra-Alvarez, R., Shi, K., Li, C., Rensing, C., Dai, X., Fan, X., and Wang, G. Root, R.A., Chorover, J., and Field, J.A. (2016) Arsenic (2018) Efflux transporter ArsK is responsible for bacterial remediation by formation of arsenic sulfide minerals in a resistance to arsenite, antimonite, trivalent roxarsone and continuous anaerobic bioreactor. Biotechnol Bioeng 113: methylarsenite. Appl Environ Microbiol 84: e0184218. 522–530. Slyemi, D., and Bonnefoy, V. (2012) How prokaryotes deal Rodriguez-Freire, L., Sierra-Alvarez, R., Root, R., with arsenic. Environ Microbiol Rep 4: 571–586. Chorover, J., and Field, J.A. (2014) Biomineralization of Smith, A.H., Lingas, E.O., and Rahman, M. (2000) Contami- arsenate to arsenic sulfides is greatly enhanced at mildly nation of drinking-water by arsenic in Bangladesh: a public acidic conditions. Water Res 66: 242–253. health emergency. Bull World Health Organ 78: Rosen, B.P. (2002) Biochemistry of arsenic detoxification. 1093–1103. FEBS Lett 529:86–92. Smith, E., Scheckel, K., Miller, B.W., Weber, J., and Rosen, B.P., Bhattacharjee, H., Zhou, T., and Walmsley, A. Juhasz, A.L. (2014) Influence of in vitro assay pH and R. (1999) Mechanism of the ArsA ATPase. Biochim extractant composition on As bioaccessibility in contami- Biophys Acta 1461: 207–215. nated soils. Sci Tot Environ 473–474: 171–177.

Souza, V., Eguiarte, L.E., Siefert, J., and Elser, J.J. (2008) speciation from worldwide collected red wine samples. Microbial endemism: does phosphorus limitation enhance Food Addit Contam Part B Surveill 11: 286–292. speciation? Nat Rev Microbiol 6: 559–564. Valenzuela, O.L., Drobna, Z., Hernandez-Castellanos, E., Stauder, S., Raue, B., and Sacher, F. (2005) Thioarsenates Sanchez-Pena, L.C., Garcia-Vargas, G.G., Borja- in sulfidic waters. Environ Sci Technol 39: 5933–5939. Aburto, V.H., et al. (2009) Association of AS3MT polymor- Stolz, J.F., and Oremland, R.S. (1999) Bacterial respiration phisms and the risk of premalignant arsenic skin lesions. of arsenic and selenium. FEMS Microbiol Rev 23: Toxicol Appl Pharmacol 239: 200–207. 615–627. Van de Wiele, T., Gallawa, C.M., Kubachka, K.M., Creed, J. Suess, E., and Planer-Friedrich, B. (2012) Thioarsenate for- T., Basta, N., Dayton, E.A., et al. (2010) Arsenic metabo- mation upon dissolution of orpiment and arsenopyrite. lism by human gut microbiota upon in vitro digestion of Chemosphere 89: 1390–1398. contaminated soils. Environ Health Perspect 118: Sun, Y., Liu, G., and Cai, Y. (2016) Thiolated arsenicals in 1004–1009. arsenic metabolism: occurrence, formation, and biological Van de Wiele, T., Van den Abbeele, P., Ossieur, W., implications. J Environ Sci 49:59–73. Possemiers, S., and Marzorati, M. (2015) The simulator of ® Suzuki, S., Arnold, L.L., Pennington, K.L., Chen, B., the human intestinal microbial ecosystem (SHIME ). In Naranmandura, H., Le, X.C., and Cohen, S.M. (2010) Die- Verhoeckx, K., Cotter, P., López-Expósito, I., tary administration of sodium arsenite to rats: relations Kleiveland, C., Lea, T., Mackie, A., et al. (eds) The Impact between dose and urinary concentrations of methylated of Food Bio actives on Health. Cham: Springer. and thio-metabolites and effects on the rat urinary bladder Van de Wiele, T.R., Peru, K.M., Verstraete, W., Siciliano, S. epithelium. Toxicol Appl Pharmacol 244:99–105. D., and Headley, J.V. (2004) Liquid chromatography-mass Switzer Blum, J., Hernandez-Maldonado, J., Redford, K., spectrometry analysis of hydroxylated polycyclic aromatic Sing, C., Bennett, S.C., Saltikov, C.W., and Oremland, R. hydrocarbons, formed in a simulator of the human gastro- S. (2018) Arsenate-dependent growth is independent of intestinal tract. J Chromatogr B Analyt Technol Biomed an ArrA mechanism of arsenate respiration in the termite Life Sci 806: 245–253. hindgut isolate Citrobacter sp. strain TSA-1. Can J van Lis, R., Nitschke, W., Duval, S., and Schoepp- Microbiol 64: 619–627. Cothenet, B. (2013) Arsenics as bioenergetic substrates. Takeuchi, M., Kawahata, H., Gupta, L.P., Kita, N., Biochim Biophys Acta 1827: 176–188. Morishita, Y., Ono, Y., and Komai, T. (2007) Arsenic resis- Velazquez, E.M., Nguyen, H., Heasley, K.T., Saechao, C.H., tance and removal by marine and non-marine bacteria. Gil, L.M., Rogers, A.W.L., et al. (2019) Endogenous J Biotechnol 127: 434–442. Enterobacteriaceae underlie variation in susceptibility to Thomas, D.J., Styblo, M., and Lin, S. (2001) The cellular Salmonella infection. Nat Microbiol 4: 1057–1064. metabolism and systemic toxicity of arsenic. Toxicol Appl Verma, D., Garg, P.K., and Dubey, A.K. (2018) Insights into Pharmacol 176: 127–144. the human oral microbiome. Arch Microbiol 200: Thomas, F., Diaz-Bone, R.A., Wuerfel, O., Huber, B., 525–540. Weidenbach, K., Schmitz, R.A., and Hensel, R. (2011) Wang, P.P., Bao, P., and Sun, G.X. (2015a) Identification Connection between multimetal(loid) methylation in Meth- and catalytic residues of the arsenite methyltransferase anoarchaea and central intermediates of methanogenesis. from a sulfate-reducing bacterium, Clostridium sp. BXM. Appl Environ Microbiol 77: 8669–8675. FEMS Microbiol Lett 362:1–8. Tisa, L.S., and Rosen, B.P. (1990) Molecular characteriza- Wang, D., Shimoda, Y., Kurosawa, H., Liu, J., Xu, X., tion of an anion pump. The ArsB protein is the membrane Liu, X., et al. (2015b) Excretion patterns of arsenic and its anchor for the ArsA protein. J Biol Chem 265: 190–194. metabolites in human saliva and urine after ingestion of Tokmina-Lukaszewska, M., Shi, Z., Tripet, B., McDermott, T. Chinese seaweed. Int J Environ Anal Chem 95: 379–389. R., Copié, V., Bothner, B., and Wang, G. (2017) Metabolic Wang, P.P., Sun, G.X., and Zhu, Y.G. (2014) Identification response of Agrobacterium tumefaciens 5A to arsenite. and characterization of arsenite methyltransferase from an Environ Microbiol 19: 710–721. archaeon, Methanosarcina acetivorans C2A. Environ Sci Tufano, K.J., Reyes, C.A., Saltikov, C.W., and Fendorf, S. Technol 48: 12706–12713. (2008) Reductive processes controlling arsenic retention: Wang, Q., Kang, Y.S., Alowaifeer, A., Shi, K., Fan, X., revealing the relative importance of iron and arsenic Wang, L., et al. (2018) Phosphate starvation response reduction. Environ Sci Technol 42: 8283–8289. controls genes required to synthesize the phosphate ana- Uhrynowski, W., Radlinska, M., and Drewniak, L. (2019) log arsenate. Environ Microbiol 20: 1782–1793. Genomic analysis of Shewanella sp. O23S-the natural Wang, Q., Qin, D., Zhang, S., Wang, L., Li, J., Rensing, C., host of the pSheB plasmid carrying genes for arsenic et al. (2017) Fate of arsenate following arsenite oxidation resistance and dissimilatory reduction. Int J Mol Sci 20: in Agrobacterium tumefaciens GW4. Environ Microbiol E1018. https://doi.org/10.3390/ijms20051018. 17: 1926–1940. Ukhanova, M., Culpepper, T., Baer, D., Gordon, D., Wilkin, R.T., Wallschläger, D., and Ford, R.G. (2003) Specia- Kanahori, S., Valentine, J., et al. (2012) Gut microbiota tion of arsenic in sulfidic waters. Geochem Trans 4:1–7. correlates with energy gain from dietary fibre and appears Wilson, D. (2015) Arsenic content in American wine. to be associated with acute and chronic intestinal dis- J Environ Health 78:16–22. eases. Clin Microbiol Infect 18:62–66. Wu, F., Yang, L., Islam, M.T., Jasmine, F., Kibriya, M.G., Vacchina, V., Epova, E.N., Bérail, S., Médina, B., Nahar, J., et al. (2019) The role of gut microbiome and its Donard, O.F.X., and Séby, F. (2018) Total As and As interaction with arsenic exposure in carotid intima-media

thickness in a Bangladesh population. Environ Int 123: Yoshinaga, M., and Rosen, B.P. (2014) A CÁAs lyase for 104–113. degradation of environmental organoarsenical herbicides Wu, J., and Rosen, B.P. (1991) The ArsR protein is a trans- and animal husbandry growth promoters. Proc Natl Acad acting regulatory protein. Mol Microbiol 5: 1331–1336. Sci U S A 111: 7701–7706. Wu, J., and Rosen, B.P. (1993) The arsD gene encodes a Yu, H., Wu, B., Zhang, X.-X., Liu, S., Yu, J., Cheng, S., et al. second trans-acting regulatory protein of the plasmid- (2016) Arsenic metabolism and toxicity influenced by ferric encoded arsenical resistance operon. Mol Microbiol 8: iron in simulated gastrointestinal tract and the roles of gut 615–623. microbiota. Environ Sci Technol 50: 7189–7197. Wu, S., Wang, L., Gan, R., Tong, T., Bian, H., Li, Z., et al. Zargar, K., Conrad, A., Bernick, D.L., Lowe, T.M., Stolc, V., (2018) Signature arsenic detoxification pathways in Hal- Hoeft, S., et al. (2012) ArxA, a new clade of arsenite oxi- omonas sp. strain GFAJ-1. MBio 9: e00515–e00518. dase within the DMSO reductase family of molybdenum Environ Microbiol – Xue, J., Lai, Y., Chi, L., Tu, P., Leng, J., Liu, C.-W., et al. oxidoreductases. 14: 1635 1645. (2019b) Serum metabolomics reveals that gut microbiome Zargar, K., Hoeft, S., Oremland, R., and Saltikov, C.W. fi perturbation mediates metabolic disruption induced by (2010) Identi cation of a novel arsenite oxidase gene, arxA arsenic exposure in mice. J Proteome Res 18: , in the haloalkaliphilic, arsenite-oxidizing bacterium Alkalilimnicola ehrlichii J Bacteriol 1006–1018. strain MLHE-1. 192: 3755–3762. Xue, X.M., Ye, J., Raber, G., Francesconi, K.A., Li, G., Zavaleta-Pastor, M., Sohlenkamp, C., Gao, J.L., Guan, Z., Gao, H., et al. (2017) Arsenic methyltransferase is Zaheer, R., Finan, T.M., et al. (2010) Sinorhizobium involved in arsenosugar biosynthesis by providing DMA. meliloti phospholipase C required for lipid remodeling dur- Environ Sci Technol 51: 1224–1230. ing phosphorus limitation. Proc Natl Acad Sci U S A 107: Xue, X.M., Ye, J., Raber, G., Rosen, B.P., Francesconi, K. 302–307. et al fi A., Xiong, C., . (2019) Identi cation of steps in the Zhou, J., Ma, H., Wu, Y., Lv, X., Wang, J., Liu, S., et al. Environ Sci Technol pathway of arsenosugar biosynthesis. (2019) Lipidomic profiling of subchronic As S exposure – 4 4 53: 634 641. identifies inflammatory mediators as sensitive biomarkers Yang, Y., Wu, S., Lilley, R.M., and Zhang, R. (2015) The in rats. Metallomics 11: 576–585. diversity of membrane transporters encoded in bacterial Zobrist, J., Dowdle, P.R., Davis, J.A., and Oremland, R.S. arsenic-resistance operons. PeerJ 3: e943. (2000) Mobilization of arsenite by dissimilatory reduction Yin, N., Du, H., Wang, P., Cai, X., Chen, P., Sun, G., and of adsorbed arsenate. Environ Sci Technol 34: Cui, Y. (2017) Interindividual variability of soil arsenic 4747–4753. metabolism by human gut microbiota using SHIME model. Chemosphere 184: 460–466. Yin, N., Du, H., Zhang, Z., Cai, X., Li, Z., Sun, G., and Supporting Information Cui, Y. (2016) Variability of arsenic bioaccessibility and Additional Supporting Information may be found in the online metabolism in soils by human gut microbiota using differ- ’ ent in vitro methods combined with SHIME. Sci Tot Envi- version of this article at the publisher s web-site: ron 566–567: 1670–1677. Table S1. List of metagenomes searched via Blastp at the Yin, N., Zhang, Z., Cai, X., Du, H., Sun, G., and Cui, Y. Joint Genome Institute Metagenome Host-Associated data- (2015) In vitro method to assess soil arsenic metabolism base. All metagenomes are from the same study: Human by human gut microbiota: arsenic speciation and distribu- microbial communities from the National Institute of Health, tion. Environ Sci Technol 49: 10675–10681. USA, HMP