A Hyperthermoactive-Cas9 Editing Tool Reveals the Role of a Unique Arsenite Methyltransferase in the Arsenic Resistance System of Thermus thermophilus HB27

ABSTRACT Arsenic detoxification systems can be found in a wide range of organisms, from bacteria to humans. In a previous study, we discovered an arsenic-responsive transcriptional regulator in the thermophilic bacterium Thermus thermophilus HB27 (TtSmtB). Here, we characterize the arsenic resistance system of T. thermophilus in more detail. We employed TtSmtB-based pulldown assays with protein extracts from cultures treated with arsenate and arsenite to obtain an S-adenosyl-l-methionine (SAM)-dependent arsenite methyltransferase (TtArsM). In vivo and in vitro analyses were performed to shed light on this new component of the arsenic resistance network and its peculiar catalytic mechanism. Heterologous expression of TtarsM in Escherichia coli resulted in arsenite detoxification at mesophilic temperatures. Although TtArsM does not contain a canonical arsenite binding site, the purified protein does catalyze SAM-dependent arsenite methylation with formation of monomethylarsenites (MMAs) and dimethylarsenites (DMAs). In addition, in vitro analyses confirmed the unique interaction between TtArsM and TtSmtB. Next, a highly efficient ThermoCas9-based genome-editing tool was developed to delete the TtArsM-encoding gene on the T. thermophilus genome and to confirm its involvement in the arsenite detoxification system. Finally, the TtarsX efflux pump gene in the T. thermophilus ΔTtarsM genome was substituted by a gene encoding a stabilized yellow fluorescent protein (sYFP) to create a sensitive genome-based bioreporter system for the detection of arsenic ions.

ABSTRACT Arsenic detoxification systems can be found in a wide range of organisms, from bacteria to humans. In a previous study, we discovered an arsenic-responsive transcriptional regulator in the thermophilic bacterium Thermus thermophilus HB27 (TtSmtB). Here, we characterize the arsenic resistance system of T. thermophilus in more detail. We employed TtSmtB-based pulldown assays with protein extracts from cultures treated with arsenate and arsenite to obtain an S-adenosyl-L-methionine (SAM)-dependent arsenite methyltransferase (TtArsM). In vivo and in vitro analyses were performed to shed light on this new component of the arsenic resistance network and its peculiar catalytic mechanism. Heterologous expression of TtarsM in Escherichia coli resulted in arsenite detoxification at mesophilic temperatures. Although TtArsM does not contain a canonical arsenite binding site, the purified protein does catalyze SAM-dependent arsenite methylation with formation of monomethylarsenites (MMAs) and dimethylarsenites (DMAs). In addition, in vitro analyses confirmed the unique interaction between TtArsM and TtSmtB. Next, a highly efficient ThermoCas9-based genome-editing tool was developed to delete the TtArsM-encoding gene on the T. thermophilus genome and to confirm its involvement in the arsenite detoxification system. Finally, the TtarsX efflux pump gene in the T. thermophilus DTtarsM genome was substituted by a gene encoding a stabilized yellow fluorescent protein (sYFP) to create a sensitive genomebased bioreporter system for the detection of arsenic ions. IMPORTANCE We here describe the discovery of an unknown protein by using a proteomics approach with a transcriptional regulator as bait. Remarkably, we successfully obtained a novel type of enzyme through the interaction with a transcriptional regulator controlling the expression of this enzyme. Employing this strategy, we isolated TtArsM, the first thermophilic prokaryotic arsenite methyltransferase, as a new enzyme of the arsenic resistance mechanism in T. thermophilus HB27. The atypical arsenite binding site of TtArsM categorizes the enzyme as the first member of a new arsenite methyltransferase type, exclusively present in the Thermus genus. The enzyme methylates arsenite-producing MMAs and DMAs. Furthermore, we developed an hyperthermophilic Cas9-based genome-editing tool, active up to 65°C. The tool allowed us to perform highly efficient, marker-free modifications (either gene deletion or insertion) in the T. thermophilus genome. With these modifications, we confirmed the critical role of TtArsM in the arsenite detoxification system and developed a sensitive whole-cell bioreporter for arsenic ions. We anticipate that the developed tool can be easily adapted for editing the genomes of other thermophilic bacteria, significantly boosting fundamental and metabolic engineering in hyperthermophilic microorganisms.
KEYWORDS arsenic resistance, thermoresistance, extremophiles, Thermus thermophilus, CRISPR-Cas9 genome editing, genetic tool, bioreporter A rsenic is the most abundant environmental toxic element which enters the biosphere mainly from geochemical and (to a lesser extent) anthropogenic sources such as herbicides, growth promoters for livestock, and industrial activities (1). Arsenic has two relevant oxidation states, trivalent arsenite [As(III)] and pentavalent arsenate [As(V)]. Methylated arsenicals include mono-(MA), di-(DMA), and tri-(TMA) methylated forms. In general, trivalent states are more toxic than pentavalent ones, and TMAs are more toxic than inorganic arsenite. Although arsenic is not beneficial for life, it can enter cells through transporters such as aquaglyceroporins. Hence, arsenic detoxification systems can be found in a wide range of organisms, from bacteria to humans. Arsenic resistance genes (ars) include genes encoding efflux transporters, redox enzymes, methyltransferases, transcriptional repressors, and biosynthetic pathways for arsenosugars and arsenolipids (2,3). The identification and characterization of these pathways have attracted the attention of fundamental, evolutionary, and biotechnological research (4)(5)(6).
Microorganisms have been exposed to arsenic since the origin of life and consequently have evolved arsenic resistance systems, encoded by genes generally clustered in operons (7). The organization and number of the operons of arsenic resistance genes are highly variable between different species (8,9), reflecting differences in the level of arsenic resistance. The key players of arsenic detoxification are (i) arsenate reductases (ArsC) that reduce intracellular arsenate to arsenite (10), (ii) efflux permeases responsible for arsenite transport outside the cell (11), and (iii) transcriptional repressors that are generally metalloregulatory proteins of the ArsR/SmtB family (12). In addition, arsenite can be methylated by arsenite S-adenosylmethionine methyltransferases (ArsM) into MMAs, DMAs, and TMAs (13), after which they can passively leave the cell or be extruded by methylarsenite-specific efflux permease (ArsP) (14).
Based on recent molecular clock analyses, it has been concluded that arsenite efflux and arsenite methylation represented the core of microbial arsenic resistance systems before the rise of atmospheric oxygen (15). In such primordial anoxic environments, methyl-arsenicals could also function as antibiotics against competitor microbes. After the rise of atmospheric oxygen, the ArsM enzymes did become primary components of the arsenic detoxification machinery; nevertheless, in some microorganisms, they maintained their antibiotic activity (16,17).
Hydrothermal hot springs, which can be considered environments with conditions similar to niches of primordial Earth, may contain large amounts of arsenic. In these cases, these hot springs are niches for arsenic-tolerant microorganisms, which play a critical role in the global arsenic biogeochemical cycle (18). Although the resistance mechanisms to inorganic arsenic have been studied in many microorganisms (19), the contribution of organo-arsenical biotransformation in extreme environments is still at a stage of infancy. In this regard, only two algal thermoactive ArsM enzymes have been characterized to date (20).
The thermophilic bacterium Thermus thermophilus HB27, originally isolated from a volcanic hot spring in Japan (21), has an unusual genetic organization of its machinery to cope with arsenic toxicity. The currently identified arsenic resistance genes are randomly scattered in its genome (22), complicating the identification of all the genes involved. In previous studies, we elucidated some components of the arsenic resistance system of T. thermophilus HB27. We identified and characterized TtSmtB, the metalloregulatory transcriptional repressor that is responsible for the regulation of the arsenic detoxification system. TtSmtB recognizes and firmly binds to operator sequences in the promoter regions of the arsenite efflux gene (TtarsX) (23) and the arsenate reductase gene (Tarec) (24,25), efficiently repressing their transcription in the absence of arsenic ions. TtSmtB and TtArsX are also involved in cadmium sensing and export, respectively (23). In two T. thermophilus HB27 deletion mutant strains (DTtarsX and DTtsmtB), tolerance to arsenate, arsenite, and cadmium was significantly reduced compared to the wild-type strain. Although these analyses confirmed the involvement of TtArsX and TtSmtB in the promiscuous resistance mechanism, the mutant strains could still grow at concentrations of arsenic up to 3 mM (22,23). Notably, the genome of T. thermophilus HB27 is not predicted to express arsenite methyltransferases or arsenite oxidases, suggesting the existence of unidentified component(s) of the T. thermophilus HB27 arsenic resistance system that cannot be predicted by in silico approaches, highlighting the need to employ an alternative experimental identification method.
Since members of the ArsR/SmtB family are a group of homodimeric proteins with a common HTH-winged helix DNA binding domain and heterogeneous metal binding domain architectures and interaction modes (26), we hypothesized that TtSmtB could even form protein interactions with unknown, functionally related protein partners.
In this study, using an integrated proteomic, biochemical, and genetic approach, we provide a gain of insight into the arsenic resistance system of T. thermophilus HB27. We report the discovery of the first T. thermophilus HB27 arsenite methyltransferase, TtArsM. Moreover, we describe the development of a highly efficient, markerless Cas9based genome-editing tool at temperatures up to 65°C. Using this ThermoCas9 system, we demonstrated the in vivo involvement of TtArsM in arsenite detoxification. The newly developed genome editing tool was further validated by constructing a very sensitive whole-cell bioreporter system in which the TtarsX efflux transporter gene was substituted by a gene encoding a thermo-adapted superfolder yellow fluorescent protein (syfp) (27).

RESULTS
Exploring the protein-protein interactions of TtSmtB. A combined comparative and functional proteomic approach was employed to identify putative TtSmtB-interacting proteins with a role in arsenite metabolism/detoxification. Purified recombinant His-tagged TtSmtB was bound to a Ni 21 -nitrilotriacetic acid (NTA) resin for protein pulldown assays using T. thermophilus HB27 cell extracts (CFE) from cultures exposed either to arsenite or arsenate or untreated CFE cultures used as control. SDS-PAGE separation of the pulled-down proteins eluted with 0.5 M imidazole, followed by liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) (28), and comparative analysis of the acquired data resulted in the identification of 51 cytosolic proteins that interact with TtSmtB (Table S1 in the supplemental material). Only five of these proteins are simultaneously present in CFE from cultures exposed to arsenite and arsenate, but not in control CFE from the nonexposed cultures. Among these proteins, TTC0109 (GenPept accession no. AAS80457; UniProt code Q72LF0) was predicted to be involved in the arsenic detoxification system (based on homology to annotated ArsR family transcriptional regulators) and to contain a C-terminal S-adenosyl-L-methionine (SAM)-dependent methyltransferase domain (based on homology to annotated methyltransferase domain-containing proteins), suggesting a role in arsenic methylation. To date, there are no annotated arsenite SAM-dependent methyltransferases in the genome of T. thermophilus HB27 or the genomes of other thermophilic bacteria. Hence, we selected TTC0109 for further investigation as a potential novel arsenite methyltransferase.
Bioinformatics analysis of TTC0109. BLASTP analysis of TTC0109 translated sequence with sequenced microbial genomes and evolutionary analysis conducted with MEGA X demonstrated that TTC0109 is highly conserved among the members of the Thermus genus (Fig. 1A). Moreover, multiple-sequence alignment of TTC0109 with characterized prokaryotic arsenite methyltransferases (Fig. 1B) showed that all the aligned proteins contain a typical Rossman fold (29). This fold contains a GxGxG motif in a loop region, which presumably interacts with the carboxypropyl moiety of SAM, and a highly conserved aspartic (Continued on next page) Gallo et al. ® acid residue at the end of the b2 strand, which forms hydrogen bonds with the ribose hydroxyls of the cofactor (30). In the case of TTC0109, the predicted GxGxG motif is composed of G114, T115, G116, T117, and G118 residues, and the conserved aspartic residue is D135 (29,30).
On the other hand, the alignment shows that TTC0109 greatly differs from characterized ArsM proteins in the remaining sequence (Fig. 1B); nonetheless, if TTC0109 is an arsenite methyltransferase, it would be evolutionarily distant from other archaeal and bacterial arsenite methyltransferases, as shown in the phylogenetic tree (Fig. 1A), and therefore, it could belong to a new type of arsenite methyltransferases. Additionally, all the known arsenite methyltransferases, including those in the alignment, possess at least two, usually three, cysteines that are responsible for the binding of arsenite and its subsequent methylation (16,31). TTC0109 contains a single cysteine residue at position 77, which is perfectly conserved in all the sequences analyzed (Fig. 1B); hence, TTC0109 could be an arsenite methyltransferase with a distinct reaction mechanism.
Reasoning that the TTC0109 structure could provide more information regarding the function of TTC0109, we generated a structural model of the protein and performed molecular docking with arsenite and SAM (Fig. S1). The obtained model predicts that TTC0109 forms homodimers via its N-terminal moiety; molecular docking highlighted that arsenite could be coordinated by two histidines, H40 and H179, while C77 interacts with the methyl group of SAM (Fig. S1A).
Although H40 and H179 residues are not conserved in characterized ArsM proteins, they are maintained at an identical position in the translated genomes of all Thermus species (five of them are shown in Fig. 1B); moreover, H40 is encompassed in a sequence motif (34-YRVFPTHSE-42) that shares 45% identity (underlined) with a sequence motif (101-YRLADRHVE-109) at the C terminus of the TtSmtB metal binding site (32), strengthening the hypothesis that H40 could be involved in arsenite binding and suggesting an evolutionary connection between the two proteins.
We proceeded with the generation of structural models of mutant proteins in which the amino acids H40 or H179 were replaced with alanine residues producing TTC0109 H40A and TTC0109 H179A and C77 replaced with a serine residue generating TTC0109 C77S ( Fig. S1B to D). The predicted models showed that the substitution of either H40 or H179 with an alanine residue altered the three-dimensional structure of TTC0109, whereas the effect of the cysteine-to-serine substitution had only a minimal effect ( Fig. S1C and D), supporting the hypothesis that this residue could have a functional role.
TTC0109 is a novel arsenite methyltransferase. A recombinant His-tagged version of TTC0109 was produced and purified to homogeneity from Escherichia coli BL21-CodonPlus (DE3)-RIL cells transformed with pET30b(1)/TtarsM (predicted mass, 29.1 kDa) (Fig. S2A). Gel filtration chromatography analysis agreed with the in silicopredicted dimeric configuration of the protein, showing that the homodimer has a mass of approximately 64.5 kDa (Fig. S2B). To determine whether TTC0109 had arsenite methyltransferase activity, a coupled spectrophotometric enzymatic assay based on the formation of S-adenosylhomocysteine (SAH) from SAM after transfer of methyl group(s) on the substrate was employed (33,34). In this case, the acceptor of methyl groups was As(III). SAH is degraded by SAH nucleosidase into S-ribosylhomocysteine New Players in Thermophilic Arsenic Resistance System ® and adenine; adenine deaminase acts on adenine producing hypoxanthine, which is converted into urate and hydrogen peroxide (H 2 O 2 ) by xanthine oxidase. The rate of production of H 2 O 2 is measured by an increase in absorbance at 510 nm with the help of the colorimetric reagent 3,5-dichloro-2-hydroxybenzensulfonic acid (DHBS). Then, arsenite methyltransferase activity was assayed following the increase in absorbance at 510 nm (35). Preliminary assays were set up to assess the thermal stability of the different components, and, consequently, the optimal assay temperature; afterward, the saturating concentrations of SAM and arsenite were determined. Therefore, the optimal assay conditions resulted as follows: 50°C, 200 mM arsenite, 800 mM SAM, and 3.1 mM TTC0109. Under these conditions, the specific arsenite methyltransferase activity of TTC0109 was 4.5 mU/mg ( Fig. 2A). For this reason, here, the TTC0109 protein will be denoted as TtArsM.
In order to determine which products are formed upon As(III) methylation by TtArsM, we incubated 10 mM TtArsM with As(III), GSH, and SAM at 65°C for 24 h; the mixture was then solubilized and analyzed by gas chromatography-mass spectrometry (GC-MS) (36). Two sharp peaks, at 2.92 min and 2.96 min, were present in the chromatogram of the product of the TtArsM reaction ( Fig. 2B) but not in the chromatogram of the product of the control reaction (Fig. 2C). These peaks were tentatively attributed to MMAs and DMAs, respectively. Unambiguous identification of each compound was accomplished by comparison of their retention times and fragmentation spectra with those of MMA and DMA standards. Butterfly plots showing the strong correspondence of experimental versus theoretical fragmentation spectra for MMAs and DMAs are depicted in Fig. 2D and E, respectively. The assignment was further confirmed by matching the experimental fragmentation spectra with those publicly available in the NIST 05 mass spectral library. According to NIST guidelines, score values higher than 700 indicate an effective identification; the scores determined for MMA and DMA were 790 and 820, respectively (37). MMAs and DMAs were manually integrated, and the results are summarized in Fig. S2C.
The results of in vitro assays using purified TtArsM protein confirmed the ability of the protein to methylate As(III)-producing mono-and dimethylated arsenic, the latter being the primary product. The oxidation state of the products could not be determined because the reactions were terminated with H 2 O 2 , which oxidized all arsenicals to pentavalent states.
Since the in silico predictions of TtArsM led to the hypothesis that C77, H40, and H179 were catalytic amino acids, three mutated versions of TtarsM, namely, TtarsM C77S, TtarsM H40A, and TtarsM H179A, were constructed and expressed in E. coli BL21-CodonPlus(DE3)-RIL cells, and the corresponding TtArsM mutants were purified (Fig. S3). Although the expression levels of the three mutant proteins are comparable (Fig. S3), it was not possible to perform in vitro characterization of purified TtArsM H40A and TtArsM H179A, which precipitated in solution after purification; this phenomenon is probably due to protein instability, thus indicating the importance of these amino acids for TtArsM structure. On the other hand, the purification of soluble TtArsM C77S protein was possible, albeit with a lower yield than the wild-type TtArsM. Nonetheless, using the previously mentioned coupled assay, this mutant enzyme did not show any in vitro arsenite methyltransferase activity, confirming the C77 residue plays a role in TtArsM activity ( Fig. 2A). These in vitro results demonstrated that TtArsM has arsenite methyltransferase activity, and its distinct active site suggests a novel reaction mechanism compared to other characterized arsenite methyltransferases (13,38,39).
TtSmtB interacts with TtArsM and binds to its promoter. This is the first study to report the protein-protein interaction of an ArsR/SmtB transcriptional regulator with a member of the arsenic detoxification system as identified by pulldown and mass spectrometry. For this reason, we decided to confirm the physical interaction between TtSmtB and TtArsM and to investigate the effect of different metals on the TtSmtB-TtArsM interaction. A coimmunoprecipitation (co-IP) assay was carried out upon incubation of purified TtArsM and TtSmtB, either in the presence or in the absence of arsenite, arsenate, cadmium, and antimony. The first three ions are TtSmtB effectors, as their interaction weakened the binding to target promoters, while antimony had no effect on DNA recognition (32). Immunoprecipitation with anti-TtSmtB antibodies, followed by detection of the His-tagged TtArsM by anti-His-tag antibodies, showed that the two proteins interact and form a complex in the absence of arsenite and arsenate, confirming the existence of physical interaction between them (Fig. 3). No band was detected when the immunoprecipitation was carried out with the unrelated control protein, TtGalA (40). Increasing arsenate and arsenite concentrations negatively affected the stability of TtSmtB-TtArsM complex; in fact, densitometric analysis of the Western blot revealed up to a 3-fold decrease in the intensity of the band corresponding to the complex (at 1:100 protein/arsenic ratio) in the presence of both arsenate (Fig. 3A) and arsenite (Fig. 3B). Interestingly, the presence of cadmium had the opposite effect, enhancing the band intensity by up to 2-fold (Fig. 3C), suggesting that the interaction of this metal with the complex occurs with a different mechanism. Finally, the presence of antimony had a negligible effect on complex stability, in agreement with previous data showing that this metal ion is not an effector for TtSmtB (32) (Fig. 3D). Since TtSmtB is the transcriptional repressor of the genes involved in arsenic and cadmium resistance in T. thermophilus HB27 (22,23), we hypothesized that it could also regulate TtarsM transcription. Sequence analysis of TtarsM promoter (p arsM ), a 108bp-long region upstream of TtarsM and encompassing the translation start codon revealed the presence of an inverted repeat region [GAAC(N14)CTTG] between positions 26 and 227 upstream of the start codon. The sequence overlaps 210 and 235 putative basal promoter region, is 100% identical to the TtarsX operator recognized by TtSmtB, and matches the consensus binding sites of ArsR/SmtB proteins (41). Hence, we performed electrophoretic mobility shift assay (EMSA) to investigate the capacity of purified TtSmtB to bind to the promoter region of TtarsM. TtSmtB binds to p arsM in a concentration-dependent manner, as shown by the gradual formation of lower-mobility complexes and the gradual decrease of residual unbound DNA (Fig. 4A, lanes 2 to 5); at 10 mM protein, the complex hardly enters the gel, suggesting the formation of multiple dimers associated with target DNA (Fig. 4A, lane 6). This observation suggests that by interacting with the regulatory region, TtSmtB controls TtarsM transcription in a way comparable to that already reported for other arsenic resistance genes, i.e., the arsenate reductase, the metal ion transporter, and itself (22,23).
Since the existence of a physical interaction between TtSmtB and TtArsM was established, we asked whether TtArsM influenced TtSmtB interaction with p arsM. Therefore, we preincubated 3 mM of both proteins before performing an EMSA under the same conditions described above. Interestingly, when the two proteins are coincubated, shifted bands can be observed (Fig. 4B, lane 3, complex 3) corresponding to complexes of higher molecular weight in comparison to those generated or not by TtSmtB or TtArsM, respectively (Fig. 4B, lane 2, complex 2, and lane 4); this analysis indicates that TtSmtB-TtArsM multimeric complexes bind to the promoter and suggests that TtSmtB-TtArsM protein-protein interaction may function in either transcriptional and posttranscriptional control. Notably, very few studies in bacteria report protein-protein interactions of transcriptional regulators with the product of the genes they regulate (42,43).
In vivo activity of TtArsM and its mutants in E. coli. Aiming to explore the role of TtArsM in arsenite resistance in vivo, we challenged E. coli BL21-CodonPlus (DE3)-RIL strains transformed with plasmids expressing TtArsM and its catalytic mutants (TtArsM C77S, TtArsM H40A, and TtArsM H179A) to grow in the presence of arsenite. Each recombinant strain was grown in the presence of different arsenite concentrations for 24 h to determine the MIC toward the metal ion. The TtArsM-expressing strain appeared to be more resistant to arsenite than the control strain (MICs of 6 mM and New Players in Thermophilic Arsenic Resistance System ® 4.5 mM, respectively). Additionally, the strains expressing mutated TtArsM were inhibited by the presence of arsenite to the same extent as the control strain (Fig. 5). This shows that the heterologous expression of TtArsM in E. coli increases arsenite resistance even at mesophilic temperatures, indicating the role of TtArsM in arsenite detoxification. Moreover, the result obtained with the mutant strains demonstrates the role of C77, H40, and H179 in the catalytic function of TtArsM.
Developing a hyperthermoactive-Cas9 editing tool. We further aimed to investigate in vivo the contribution of TtArsM to the arsenite detoxification mechanism via the deletion of the TtarsM gene from the T. thermophilus HB27 genome. Nonetheless, the currently available genome editing tool for T. thermophilus is time-consuming, not marker-free, and not always efficient (44). For this purpose, we reasoned to develop a marker-free, plasmid-based, homologous recombination (HR) Cas9 counterselection (CS) genome-editing tool for T. thermophilus employing ThermoCas9, a thermotolerant and thermoactive Cas9 orthologue (45).
We initially evaluated the targeting efficiency of ThermoCas9 in T. thermophilus HB27. Therefore, a set of 3 vectors was constructed, namely, pMK-ThermoCas9-NT, pMK-ThermoCas9-sp1, and pMK-ThermoCas9-sp2 (Fig. 6A) by cloning into the pMK18 vector (46) (i) the codon-harmonized version of the thermocas9 gene under the transcriptional control of the constitutive nqo promoter (44), and (ii) the single guide RNA (sgRNA)expressing module under the transcriptional control of the constitutive 16S rRNA promoter, either with a nontargeting/control spacer (NT, 59-CTAGATCCGCAGTAACCCCATGG-39) or with spacers that target the TtarsM gene (sp1, 59-GGGCGTTGGTGATGTGGGCCCTC-39, and sp2, 59-CCACCTCCTCCTCCCGGTAAGGC-39). The 3 vectors were used to transform T. thermophilus HB27, along with pMK-Pnqo-syfp vector (27), as transformation control. The cells were allowed to recover at 70°C before being plated on selective agar plates and incubated overnight at 60°C due to the sensitivity of pMK18 at temperatures above 65°C. The transformation efficiencies of pMK-ThermoCas9-sp1 and pMK-ThermoCas9-sp2 targeting vectors were significantly reduced compared to the transformation efficiency with the pMK-ThermoCas9-NT nontargeting vector (Fig. 6B). Moreover, the transformation efficiency with the pMK-Pnqo-syfp vector was only slightly higher than the transformation efficiency with the pMK-ThermoCas9-NT vector (Fig. 6B), which could be attributed to the significant size difference between the two vectors (7,552 bp and 13,554 bp, respectively). This result indicates that ThermoCas9 is expressed in T. thermophilus HB27 cells in an active and nottoxic form, motivating the development of a ThermoCas9-based genome-editing tool.
We set out to develop and test the efficiency of an HR ThermoCas9-based CS genomeediting tool in T. thermophilus HB27. For this purpose, we introduced an HR template for the deletion of the TtarsM gene into the 3 previously described ThermoCas9 vectors. The HR template was composed of the fused 1-kb upstream and downstream flanking regions of the TtarsM gene (Fig. 6C). The three resulting editing vectors, namely, pMK-ThermoCas9-HR-NT, pMK-ThermoCas9-HR-sp1, and pMK-ThermoCas9-HR-sp2, were transformed into T. thermophilus HB27 cells, recovered at 70°C, and grown on selective agar plates overnight at 60°C. Colony PCR with genome-specific primers was subsequently employed to screen several colonies for each transformation (Fig. S4; Table S2). None of the colonies from the pMK-ThermoCas9-NT transformation were clean DTtarsM mutants (0/10 colonies), and only a small number of colonies were mixed wild-type/DTtarsM mutants (2/10 colonies) ( Fig. S4A; Table S2). On the other hand, almost all the screened colonies from the pMK-ThermoCas9-sp1 transformation were clean DTtarsM mutants (19/19 colonies) ( Fig. S4B; Table S2); most of the screened colonies from the pMK-ThermoCas9-sp2 transformation were clean DTtarsM mutants (13/18 colonies), and the remaining were mixed wild-type/DTtarsM mutants (5/18 colonies) (Fig. S4C; Table S2); the latter result suggests that less efficient ThermoCas9 targeting is resulting products from genome-specific colony PCRs on T. thermophilus colonies formed from the ThermoCas9-based TtarsM deletion process. A wild-type colony was subjected to the same PCR, and the related product is shown here as negative control for TtarsM deletion. The expected sizes of the PCR amplification products that correspond to the wild-type and DTtarsM genotypes are indicated with black arrows.
New Players in Thermophilic Arsenic Resistance System obtained when employing spacer 2. Subsequently, DNA sequencing on randomly selected clean DTtarsM mutant colonies was performed to verify the correctness of the genome editing ( Fig. S4D; Table S2).
Aiming to test the temperature limit of the developed ThermoCas9-based genomeediting tool, we repeated the editing experiment, increasing the plating temperature to 65°C, corresponding to the temperature limit of the pMK18 backbone for propagation. Under these conditions, the numbers of colonies formed upon transformation with the pMK-ThermoCas9-sp1 and pMK-ThermoCas9-sp2 were 3 and 5 vectors, respectively, much lower than the corresponding numbers when the plating temperature was 60°C (Table S2). This can be ascribed to the high ThermoCas9 targeting activity at 65°C (45) and decreased vector stability at 65°C. Nonetheless, the DNA sequence of all the screened colonies confirmed that they were clean DTtarsM mutants, demonstrating the high efficiency of the developed tool at 65°C (Fig. 6D).
Finally, assuming that the curing of the editing plasmid from a DTtarsM mutant strain would facilitate additional editing steps, we randomly selected a DTtarsM mutant colony for inoculation in liquid antibiotic-free TM medium for two culturing rounds at 65°C and then plated the cultures on TM agar plates with and without antibiotic. Multiple colonies were found on the antibiotic-free plate and no colonies on the plate supplemented with the antibiotic, demonstrating that the cells were cured from the edited vector. Seven of these colonies were randomly selected and the absence of the plasmid confirmed by colony PCR using thermoCas9-specific primers (Fig. S4E).
Therefore, a markerless HR-ThermoCas9-based CS genome-editing tool was developed for T. thermophilus HB27, highly efficient at temperatures up to 65°C. Using this tool, a T. thermophilus DTtarsM strain was constructed in less than 10 days (including the plasmidcuring process), expanding the repertoire of available genetic tools for this microorganism and considerably accelerating the required time for editing its genome.
TtArsM mutant is more sensitive to arsenite. To compare the arsenic resistance of DTtarsM to that of wild-type T. thermophilus HB27, both strains were grown in TM liquid medium with different arsenite and arsenate concentrations for 24 h (Fig. 7). As expected, the arsenite resistance of DTtarsM was significantly lower than that of the wild-type strain, with the corresponding MIC values being 18 mM and 40 mM, respectively (Fig. 7A). Moreover, the resistance of the DTtarsM strain to arsenate was comparable to the wild-type strain (42 mM and 44 mM, respectively) (Fig. 7B), in agreement with its role in arsenite resistance. This result confirmed that the thermoactive arsenite methyltransferase TtArsM is involved in arsenite detoxification and is a novel component of the arsenic resistance machinery.
Developing a sensitive arsenic bioreporter. In a previous study, we demonstrated that TtArsX is the arsenic efflux membrane protein of T. thermophilus HB27 and reported that a DTtarsX mutant strain is more sensitive to arsenate and arsenite (23). In this study, we wanted to ascertain whether a strain lacking both TtarsM and TtarsX would be even more sensitive to arsenic ions than the single-mutant DTtarsX and DTtarsM strains and therefore could represent an even better bioreporter strain for arsenic detection.
For this purpose, the HR ThermoCas9-based CS-editing tool was employed to exchange the TtarsX gene on the genome of the DTtarsM strain with the syfp reporter gene (27), setting the expression of the encoded thermotolerant yellow fluorescence protein (sYFP) under the control of the arsenic-responsive TtarsX promoter (p arsX ). The employed editing vector, denoted as pMK-ThermoCas9-HR-syfp, contained a spacer that targets the TtarsX gene (59-TTTCGACGGAGGAGGCCTTGGCC-39) and an HR template composed of the 1-kb upstream flanking genomic region of TtarsX followed by syfp and the 1-kb downstream flanking genomic region of TtarsX. Ten colonies grown after transformation of pMK-ThermoCas9-HR-syfp vector into T. thermophilus DTtarsM cells were screened by colony PCR with genome-specific primers and sequenced; eight of them were clean T. thermophilus HB27 DTtarsM-DTtarsX (syfp) knock-in mutants (Fig. S5), also proving that the developed tool was highly efficient for gene insertions and substitutions.
To evaluate the sensitivity to arsenate and arsenite of the whole-cell bioreporter system, exponentially growing cultures of T. thermophilus HB27 DTtarsM-DTtarsX (syfp) were treated with increasing concentrations of arsenite and arsenate, and the intensity of the emitted fluorescence was compared (Fig. 8). The background fluorescence of the DTtarsM-DTtarsX (syfp) strain was low, indicating that the system is repressed in the absence of metal ions. Moreover, the developed bioreporter system was able to detect arsenite and arsenate concentrations as low as 0.5 mM (Fig. 8). This performance substantially overtakes the detection limit of the previously developed arsenite and arsenate bioreporter system, which was based on the T. thermophilus DTtarsX strain and plasmid-based expression of the b-galactosidase (23).

DISCUSSION
In this study, we aimed to identify novel proteins involved in the arsenic resistance system of T. thermophilus HB27 and employed TtSmtB as a starting point looking beyond its transcriptional regulation activity. As TtSmtB contains a protein interaction domain (32), we set out to identify putative TtSmtB-interacting proteins with a role in arsenic metabolism/detoxification, following an immunoprecipitation and comparative proteomics approach. This strategy led to the discovery of TtArsM, the first prokaryotic thermoactive arsenite SAM-dependent methyltransferase, evolutionarily distant from other known arsenite methyltransferases. New Players in Thermophilic Arsenic Resistance System ® The original structure and activity mechanism of TtArsM were explored in silico. Like other arsenite methyltransferases known to date, TtArsM contains a C-terminal SAMdependent methyltransferase domain and an N-terminal domain comprised of only one (instead of the usual three) conserved cysteine; the other two catalytic cysteines required for arsenite coordination might be replaced by two histidines identified by docking analysis. Interestingly, H40 is part of an arsenite binding domain typical of many ArsR metalloregulatory proteins (47), and H179 is located close to the cysteine (C177) of the arsenite methyltransferase of the extremophilic alga Cyanidioschyzon merolae (48). Moreover, the conservation of these three amino acids among the putative arsenite methyltransferases in the Thermus genus supports their possible role in catalysis and suggests an adaptation in this group of microorganisms.
Indeed, the role of the cysteine and histidine residues in the structure-function of TtArsM predicted in silico was demonstrated by site-directed mutagenesis, as the heterologous expression in E. coli of TtArsM mutants conferred lower arsenite resistance than TtArsM.
To date, only for a few arsenite methyltransferases, the reaction mechanism has been reported (13,20,38,39,49). Those described possess three, or at least two, cysteine residues present in their catalytic site that can methylate the arsenic sequentially in its trivalent form through alternating reduction and oxidative methylation reactions; of note, different enzymes produce mono-, di-, and trimethylated forms of arsenic in diverse amounts, highlighting a biochemical diversity in the arsenite methylation mechanism (38). The newly discovered TtArsM, highly conserved within the Thermus genus, possesses only one cysteine and can methylate As(III) mainly into DMAs and a smaller amount of MMAs as determined by GC-MS analysis of the products of the in vitro assay. To the best of our knowledge, this is the first arsenite methyltransferase functioning with a single cysteine in the active site.
The discovered transcriptional and posttranslational interaction of TtArsM with the transcriptional regulator TtSmtB was investigated in more detail. It was demonstrated that TtSmtB binds to the promoter region of TtarsM and that this binding is stabilized by the TtSmtB-TtArsM complex. Moreover, co-IP experiments confirmed the interaction of TtSmtB with TtArsM and showed a reverse correlation between the stability of the complex and arsenic concentration. Presumably, the complex enhances the repression of TtarsM transcription in the absence of arsenic ions through a novel mechanism. Hence, through this analysis, we shed light on a novel kind of interaction, rarely described for bacteria, in which the transcriptional repressor of a gene interacts with the protein product of the gene that it regulates (43,50). An example of an enzyme that can modulate the transcriptional activity of regulators by protein-protein interaction is reported in the cysteine metabolism of Bacillus subtilis where the stable complex formed by CymR (the master regulator of the system) and CysK (O-acetyl-L-serine-thiol-lyase) represses the transcription of the genes involved in the cysteine pathway (including the cysK gene itself) when cysteine concentration is low. The advantage of this regulatory mechanism is that it employs enzymes that can specifically recognize their substrates or allosteric effectors; thus, enzymes and/or transcriptional regulators can act simultaneously as intracellular molecular sensors and participate in their own transcriptional regulation (42).
The role of TtArsM in T. thermophilus HB27 arsenic detoxification was also demonstrated via the construction and characterization of a DTtarsM mutant strain. For this purpose, a marker-free homologous recombination and ThermoCas9-based counterselection genome-editing tool was developed which was highly efficient and active at temperatures up to 65°C. Our tool equals the highest reported temperature for a Cas9based editing tool to date (51,52). The characterization of the T. thermophilus DTtarsM strain confirmed its expected higher sensitivity to arsenite, but not arsenate, than the wild-type strain. To better define the role of TtArsM in the context of the already characterized components of the arsenic resistance system, a double mutant was constructed upon exchanging the TtarsX efflux pump gene in the T. thermophilus DTtarsM genome with the gene encoding the yellow fluorescent protein; the double mutant resulted as much more sensitive to arsenite and arsenate treatment. Hence, it was demonstrated that TtArsM and TtArsX are critical players of the arsenite detoxification system. This is the first example of a successful insertion of a heterologous gene on the T. thermophilus genome by genome editing. The double-mutant strain was also considered a sensitive bioreporter for the development of a whole-cell biosensor system. Indeed, it was able to detect arsenite and arsenate concentrations as low as 0.5 mM, showing 40 times higher sensitivity than the previously developed T. thermophilus HB27 DTtarsXplasmid-based biosensor (23).
In conclusion, this study explores a unique strategy to identify novel enzymes and/ or regulative networks in nonmodel bacteria and expands the repertoire of genetic systems for hyperthermophiles. In addition, this work has resulted in a gain of insight into the arsenite/arsenate detoxification mechanism, particularly that of T. thermophilus. On top of that, this has allowed us to develop a highly robust and sensitive biosensor.

MATERIALS AND METHODS
T. thermophilus HB27 cell extract preparation. T. thermophilus HB27 cultures were grown aerobically at 70°C in TM medium, as previously described (24). Once the cultures reached an optical density at 600 nm (OD 600 ) of 0.5, they were treated either with 8 mM NaAsO 2 or with 12 mM NaH 2 AsO 4 (Sigma) (the used concentrations were below the previously reported MIC values for arsenate and arsenite) (22,23) or they remained untreated. Samples were harvested from each culture, either immediately after treatment or 60 min posttreatment. The samples were centrifuged, the precipitates were resuspended in phosphate buffer (20 mM Na 3 PO 4 , pH 7.5) supplemented with protease inhibitor cocktail (Thermo Scientific), and the resuspended cells were lysed by sonication (10 cycles of 30 s on/30 s off, 40% power; Misonix Sonicator Ultrasonic Processor XL). The lysates were centrifuged and the cell extracts (CFE) used for pulldown assays.
Purification of recombinant TtSmtB, immobilized metal affinity chromatography, and pulldown. C-terminal His-tagged TtSmtB was purified from E. coli BL21-CodonPlus (DE3)-RIL cells transformed with the pET28/TtsmtB vector, as previously described (22). Purified C-terminal His-tagged TtSmtB (2 mg) was incubated with 200 ml of Ni 21 -NTA resin (Sigma-Aldrich) equilibrated in 20 mM Na 3 PO 4 , 0.5 M NaCl, and 20 mM imidazole, pH 7.5, for 16 h at 4°C and then washed three times with the same buffer to remove unbound proteins. T. thermophilus HB27 CFE, treated with arsenite, treated with arsenate, or not treated were incubated with the functionalized resin (Ni 21 -NTA/TtSmtB) for 16 h at 4°C under stirring conditions. Subsequently, the resin was extensively washed, and the interacting proteins were eluted with 20 mM Na 3 PO 4 , 0.5 M NaCl, and 0.5 M imidazole, pH 7.5. As negative controls, samples of Ni 21 -NTA resin not functionalized with TtSmtB were incubated with the same T. thermophilus HB27 CFE.
In situ hydrolysis and LC-MS/MS analysis. The fractions eluted from the pulldown process were analyzed by 15% SDS-PAGE and in situ hydrolyzed for mass spectrometry analysis. Specifically, monodimensional SDS-PAGE gel was colored with Coomassie brilliant blue; the revealed bands were cut and destained with 100 ml of 0.1 M ammonium bicarbonate (AMBIC) and 130 ml of acetonitrile (ACN). Each band was hydrolyzed in situ with 0.1 mg/ml trypsin in 10 mM AMBIC and incubated at first for 1.5 h at 4°C and then for an additional 16 h at 37°C. The hydrolysis reactions were stopped by adding acetonitrile and 0.1% formic acid; then, the samples were filtered and dried in a Savant vacuum centrifuge before New Players in Thermophilic Arsenic Resistance System being analyzed by LC-MS/MS mass spectrometry. In detail, before analysis, the samples were dissolved in 10 ml of 0.1% formic acid, and 5 ml were directly loaded into the instrument. Reverse-phase capillary liquid chromatography (HPLC 1200 system experiments), followed by MS analysis, was performed using a binary pump system connected to a nanospray source of the mass spectrometer (28,53). The latter is represented by a hybrid quadrupole time of flight (Q-TOF) spectrometer (MS Chip 6520 QTOF) equipped with a chip (Agilent Technologies).
The alignment of TtArsM to its templates was based on a multiple-sequence alignment, performed with the program Clustal Omega (57); the amino acidic sequences used for the construction of the alignment of functionally characterized archaeal and bacterial arsenite methyltransferases are TtArsM from T. thermophilus HB27, 5 members of the Thermus genus (T. islandicus, T. caldilimi, T. antranikianii, T. oshimai, and T. brockianus), Clostridium sp. BMX, R. palustris, M. acetivorans, H. salinarum, P. alcaligenes, and Cyanidioschyzon merolae (48).
Models of TtArsM were generated through I-TASSER (58) (https://zhanglab.ccmb.med.umich.edu/I-TASSER/) using as input the complete sequence of TtArsM (C score, 22.5). The dimeric structure was predicted using the GalaxyWEB tool (http://galaxy.seoklab.org/index.html) (59). The molecular dockings of TtArsM with arsenite and SAM were generated using the Hex protein docking server (60). One hundred rigid-body docking solutions were generated per case, and the best 10 were refined by energy minimization. The proposed model for the metal ion docked into TtArsM is the structure with the smallest distance between arsenite-histidine and cysteine-SAM (4.33 Å from H40 and 5.77 Å from H179 in TtArsM model and 4.40 Å from C77).
Cloning, expression, and purification of recombinant TtArsM and TtArsM mutants. The pET30b (1)/TtarsM vector was constructed for the expression and subsequent purification of the C-terminal Histagged version of TtArsM. For the construction of the pET30b(1)/TtarsM, TtarsM gene was PCR amplified from T. thermophilus HB27 genome, using Taq DNA polymerase (Thermo Fisher Scientific) and primers containing the NdeI (arsMfw; Table S3 in the supplemental material) and HindIII (arsMrv ; Table S3) restriction sites at their 59 ends. The PCR product was purified, digested with the NdeI and HindIII restriction enzymes (NEB), and ligated (T4 ligase; NEB) into NdeI/HindIII-digested pET30b(1) vector (Novagen). The ligase mixture was transformed into E. coli TOP10F9 cells, which were plated on LB agar plates supplemented with 50 mg/ml kanamycin (Sigma-Aldrich). Single colonies were selected and inoculated in LB liquid medium supplemented with 50 mg/ml kanamycin. Plasmid isolation and sequencing were subsequently performed before transforming E. coli BL21-CodonPlus (DE3)-RIL cells with pET30b(1)/TtarsM vector.
For protein expression of the His-tagged versions of TtArsM C77S, TtArsM H40A, and TtArsM H179A catalytic mutants, the recombinant E. coli BL21-CodonPlus (DE3)-RIL strains were cultured in LB medium supplemented with kanamycin (50 mg/ml) and chloramphenicol (33 mg/ml). Protein expression was induced via the addition of 1 mM isopropyl-1-thio-b-D-galactopyranoside (IPTG) when the cultures reached OD 600 of 0.7. The cultures were further incubated with vigorous shaking at 37°C for 16 h, then centrifuged, resuspended in lysis buffer (20 mM NaP, pH 7.4, 50 mM NaCl, and 20 mM imidazole) supplemented with protease inhibitor cocktail (Thermo Scientific), and lysed by sonication (10 cycles of 30 s on/30 s off, 40% power; Misonix Sonicator Ultrasonic Processor XL). The lysates were centrifuged and the supernatants used for the purification on HisTrap HP columns (1 ml; GE Healthcare) connected to an AKTA Explorer system (GE Healthcare). The fractions containing His-tagged TtArsM proteins were eluted from the columns using a linear gradient of the elution buffer (20 mM NaP, pH 7.4, 50 mM NaCl, and 500 mM imidazole). The eluted protein fractions were subjected to SDS-PAGE analysis, and the fractions containing purified TtArsM were pooled and dialyzed for 16 h at 4°C in 20 mM NaP, pH 7.4, buffer supplemented with protease inhibitor cocktail (Thermo Scientific). The identity of the purified His-tagged TtArsM protein was confirmed by mass spectrometry, and protein aliquots were stored at 220°C.
TtArsM quaternary structure assessment. The native molecular mass of TtArsM was determined by loading 500 mg of the purified protein onto an analytical Superdex PC75 column (3.2 by 30 cm) connected to an AKTA Pure system in 50-mM Tris-HCl, pH 7.5, and 0.2-M KCl buffer. The column was calibrated using a set of gel filtration markers (low range; GE Healthcare), including ovalbumin (43.0 kDa), carbonic anhydrase (29.0 kDa), RNase A (13.7 kDa), and aprotinin (6.5 kDa) as previously described (24).
Methyltransferase activity assay. According to the manufacturer's protocol, the TtArsM arsenite methyltransferase activity was measured using the SAM510 SAM methyltransferase assay kit (G-Biosciences) with modifications regarding the temperature, SAM concentration, and reaction time (33)(34)(35). The assay relies on the degradation of S-adenosylhomocysteine (SAH) into urate and hydrogen peroxide by a mixture of enzymes (adenosylhomocysteine nucleosidase, adenine deaminase, and xanthine oxidase). Then, the reaction of hydrogen peroxide with 4-aminoantipyrine produces 5-dichloro-2hydroxybenzene sulfonic acid (DHBS) with « mM of 15.0 at 510 nm. A typical reaction mixture containing 200 mM As(III), 800 mM SAM, 3.1 mM the enzyme, SAM enzyme mixture, and SAM colorimetric mix in a final reaction volume of 115 ml was incubated for 1 h at 50°C in a Synergy HTX multimode microplate reader (BioTek). The same reaction mixture was tested with 10 mg of TtArsC or 10 mg of TtSmtB as negative controls. One unit of arsenite methyltransferase produces 1.0 mmol of DHBS per minute at 50°C under the conditions described above. Preliminary assays were performed to define substrate saturating concentrations, varying the As(III) and SAM concentrations from 50 mM to 300 mM and from 200 mM to 1.2 mM, respectively.
In vitro arsenite methylation. As(III) methylation by TtArsM was determined in an assay solution containing 10 mM TtArsM, 250 mM As(III), 6 mM glutathione (GSH), and 1 mM SAM, in Na-phosphate 50 mM, pH 7.4, at 65°C for 24 h; the same reaction without TtArsM was used as the negative control (control sample). The reactions were terminated by the addition of 10% (vol/vol) H 2 O 2 . Samples were then filtered through 0.22-mm mixed cellulose ester (MCE) syringe filters and used for GC-MS analysis performed in an Agilent GC 6890, coupled with a 5973 MS detector. A modification of the Huang protocol (36) was employed for the detection of the methylated products, and the arsenic compounds were analyzed by using hyphenated mass spectrometry techniques without derivatization. Unambiguous identification of arsenic compounds was accomplished using authentic standards, including disodium methyl arsonate hexahydrate (MMAs) (ChemService) and dimethylarsinic acids (DMAs) (Merck Life Science S.r.l.).
In more detail, 200 ml of each sample was treated with 800 ml of methanol (Sigma-Aldrich). The supernatants were recovered, vacuum-dried, and solubilized in 10 ml of methanol. We analyzed 1 ml of each sample by GC-MS, using the HP5 capillary column (30 m Â 0.25 mm, 0.25 mm; Agilent). Helium was employed as carrier gas at a rate of 1.0 ml min 21 . The temperature of the GC injector was maintained at 230°C; while the oven temperature was initially set at 40°C for 5 min, it was subsequently increased to 280°C for 5 min with an increment of 20°C/min, resulting in a total separation time of 20 min. The temperature of the analyzer was kept at 250°C. The collision energy was set to 70 eV, and the generated fragment ions were analyzed in a mass range of 20 to 450 m/z. The presence of arsenic compounds in the analyzed samples was confirmed by comparing the fragmentation spectra of the samples and the corresponding retention times with those of MMA and DMA standard solutions. Finally, the presence of MMAs and DMAs in the samples was confirmed by matching the experimental fragmentation spectra with those published in the NIST 05 mass spectral library. According to the NIST guidelines, the identification was reliable when the matching values were higher than 700 (37). The analyses of the samples were performed in technical triplicates.
Co-IP assay. The protein-protein interaction between TtSmtB and TtArsM was in vitro verified via coimmunoprecipitation assays that employed recombinant TtSmtB, recombinant His-tagged-TtArsM, anti-TtSmtB antibodies (GeneCust), and His tag antibodies (Sigma-Aldrich). His tag removal from recombinant TtSmtB was performed as previously described (22).
ImageJ software (https://imagej.net/) was used for densitometric analysis of the formed bands, setting as maximum value the intensity of the band of the sample containing TtArsM and TtSmtB in the absence of any metal. Each experiment was performed in three technical and two biological replicates; statistical analysis was performed using one-way analysis of variance (ANOVA), and significant differences are indicated as *, P , 0.05; **, P , 0.01; and ***, P , 0.001.
EMSAs. Electrophoretic mobility shift assays (EMSAs) were performed to determine the in vitro binding of TtSmtB to the promoter region upstream of TtarsM. The 108-bp chromosomal region that encompasses the start codon of the TtarsM gene and the 105-bp-long region upstream, denoted as p arsM , was PCR amplified with Taq DNA polymerase (Thermo Fisher Scientific) using the p arsM Fw and p arsM Rv primers (Table S3). EMSA reactions were set up as previously described (23,62) in the presence of 1 mg of poly(dI-dC), 20 ng p arsM , and increasing concentrations of TtSmtB (1, 2, 3, 5, and 10 mM, considering TtSmtB as a dimer) using SYBR Gold nucleic acid gel stain for band detection. The EMSA reactions that simultaneously employed TtSmtB and TtArsM were set up in identical conditions using 3 mM of each protein alone or in combination.
coli BL21-CodonPlus (DE3)-RIL, pET30/TtarsM H40A; (iv) E. coli BL21-CodonPlus (DE3)-RIL, pET30/TtarsM H179A; and (v) E. coli BL21-CodonPlus (DE3)-RIL, pET30 (control). The precultures were incubated at 37°C for 16 h at 180 rpm. Subsequently, 50-ml LB cultures, supplemented with antibiotics, were inoculated with the precultures to initial OD 600 of 0.08 and incubated at 37°C and 180 rpm until an OD 600 of 0.6 was reached (exponential growth). At that point, protein expression was induced with 1 mM IPTG, and the cultures were incubated at 37°C and 180 rpm for 3 additional hours. From these growing cells, fresh LB cultures were inoculated to OD 600 of 0.05 and distributed to 24-well plates (1 ml per well) containing LB medium supplemented with 1 mM IPTG, kanamycin (50 mg/ml), chloramphenicol (33 mg/ml), and increasing concentrations of arsenate and arsenite (from 2.5 mM to 7.0 mM). The MICs endpoint for each strain were determined as the lowest concentration of arsenite at which there was the difference between grown and start culture lower than 0.01 OD 600 after 16 h of incubation at 37°C (22). All the cells up to the MIC value were able to grow if reinoculated in an arsenic-free medium. The reported values are the average of three biological replicates.
ThermoCas9 editing and targeting constructs. The plasmids used for the ThermoCas9-based targeting and editing experiments are listed in Table S4. The vector pMK18 was used as the template for the construction of the ThermoCas9-based targeting and editing plasmids and the employed primers, the DNA templates, and the DNA fragments, which are listed in Table S4. The thermoCas9 gene was codon harmonized according to T. thermophilus HB27 codon usage using the Galaxy/Codon Harmonizer online tool (63), and it was synthesized (Twist Bioscience) ( Table S5). The DNA fragments were designed with appropriate overhangs for NEBuilder HiFi DNA assembly (NEB), and they were obtained through PCR with Q5 polymerase (NEB). The PCR products were subjected to 1% agarose gel electrophoresis, and they were purified using a Zymogen gel DNA recovery kit (Zymo Research). The assembly reactions were transformed to chemically competent E. coli DH5a cells (NEB), and the cells were plated on LB agar plates supplemented with kanamycin (50 mg/ml). Single colonies were inoculated in LB medium supplemented with kanamycin (50 mg/ml) for overnight incubation at 37°C. Plasmid material was isolated using the GeneJet plasmid miniprep kit (Thermo Fisher Scientific) and sequence verified (GATC Biotech), and 300 ng of each plasmid (pMK-ThermoCas9-NT/sp1/sp2 and pMK-ThermoCas9-HR-NT/sp1/sp2) was transformed to either T. thermophilus HB27 cells (22), as indicated per experimental process.
The obtained plasmid, pMK-ThermoCas9-NT, was used as the backbone to construct a new plasmid to obtain the deletion of the TtarsX gene (Table S4) and the insertion of the gene coding sYFP (27). The obtained plasmid (pMK-ThermoCas9-HR-syfp) was used to transform T. thermophilus HB27 DTtarsM cells as already described to obtain the strain denoted DTtarsM-DTtarsX (syfp).
Arsenic tolerance of T. thermophilus HB27 wild-type and mutant strains. Exponentially growing precultures of T. thermophilus HB27 (control), T. thermophilus HB27 DTtarsM, and T. thermophilus HB27 DTtarsM-DTtarsX (syfp) were diluted to an OD 600 of 0.08 in 10-ml TM cultures containing increasing concentrations of arsenite and arsenate (from 0.1 mM to 50 mM). The cultures were incubated aerobically at 70°C for 18 h, and the MIC values were determined as the lowest concentrations of arsenite and arsenate that completely inhibited the growth of a strain (22). The reported values are the average of three biological replicates.
Bioreporter activity measurement. Overnight cultures of T. thermophilus HB27 DTtarsM-DTtarsX (syfp) strain were diluted to an OD 600 of 0.08 in TM medium and then grown aerobically at 70°C until an OD 600 of 0.5 was reached. The cultures were divided into samples of 5 ml each and subsequently supplemented with increasing concentrations of arsenite and arsenate (0.5 mM to 4 mM). After 1 h of incubation at 70°C, 200 ml of each cell sample were removed and centrifuged for 5 min at 6,000 rpm. The pellets were washed twice with equal volumes of phosphate-buffered saline (PBS) 1Â and resuspended with equal volumes of PBS 1Â before being distributed into a 96-well plate. sYFP fluorescence intensity of each sample was measured employing a Synergy HTX multimode microplate reader (BioTek), using excitation and emission wavelengths of 458 nm and 540 nm, respectively. The measured fluorescence intensities were normalized for the optical density of each sample at 600 nm. The measured fluorescence was reported as fluorescence relative expression, assuming that the fluorescence value of nottreated cells (control) was 1.
Each experiment was performed in three technical and biological replicates. Statistical analysis was performed using one-way ANOVA; significant differences are indicated as *, P , 0.05; **, P , 0.01; and ***, P , 0.001.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.