Twenty-two samples were collected from two different Chilean regions from which bacterial isolates were obtained: (i) north (Blanca, Chaxa and Céjar lagoons, El Tatío geyser, Death Valley) and (ii) mid-south (Maule lagoon, El Teniente copper mine). A sample from Uyuni Salar, Bolivia, was also included in this study. In turn, isolates MF02 and MF05 were kindly provided by Dr. Juan Carlos Tantaléan, Universidad San Luis Gonzaga, Ica, Perú.

Four different culture media were individually inoculated with each sample and incubated at 25 or 37°C for 24 h: LB (Sambrook and Russell, 2001); R2A (0.5 g/L yeast extract, 0.5 g/L peptone, 0.5 g/L casamino acids, 0.5 g/L glucose, 0.5 g/L soluble starch, 0.3 g/L sodium pyruvate, 0.3 g/L K₂HPO₄, and 0.5 g/L MgSO₄ adjusted to pH 7.2); Acidiphilium (2 g/L (NH₄)₂SO₄, 0.1 g/L KCl, 0.5 g/L K₂HPO₄, 0.5 g/L MgSO₄, 0.3 g/L yeast extract and 1 g/L D-glucose adjusted to pH 3.0); and Aciduric thermophilic Bacillus strains (ATB) (0.2 g/L (NH₄)₂SO₄, 0.5 g/L MgSO₄, 0.25 g/L CaCl₂, 3 g/L KH₂PO₄, 1 g/L yeast extract, 1 g/L tryptone, 1 g/L glucose and 1 g/L starch adjusted to pH 4.3). Pure cultures were obtained after sequential dilutions and successive passages using appropriate solid media.

Purified colonies were seeded on LB-agar plates that contained different concentrations of the following compounds: K₂TeO₃ (0.04–0.4 mM), K₂CrO₄ (1–10 mM), CdCl₂ (3–5 mM), CoCl₂ (3–5 mM), CuSO₄ (3–5 mM), AgNO₃ (1–5 mM), Al(SO₄)₃ (5 mM), FeCl₃ (5–10 mM), PbCl₂ (3 mM), UO₂(CH₃COO)₂ (2–10 mM), ZnSO₄ (2–5 mM), NiSO₄ (2–5 mM), MnSO₄ (10–20 mM), HgCl₂ (0.02–0.1 mM), Na₂SeO₃ (10–50 mM), FeCl₂ (5–10 mM), NaAsO₂ (5–10 mM), Na₂HAsO₄ (10–50 mM), or HAuCl₄ (0.02–1 mM). Selection was for bacterial growth at the highest concentrations and when applicable, also by observing metal(loid) reduction. Strains MF01, MF03, MF04, and MF08 were from Maule lagoon, MF06 from Blanca lagoon (Atacama, Chile), MF07 from Death Valley and MF09 from El Teniente copper mine. Isolate MF10 was from Uyuni Salar, Bolivia.

Cells were grown in LB medium at 37°C; E. coli BL21 carrying the pET 101/gorA vector was grown in ampicillin (0.1 mg/ml)-supplemented LB medium. This gorA-expressing plasmid was constructed as follows: the gorA gene was amplified from the environmental strain MF01 using primers CACCATGACTAAGCATTATGACTACATCGCA (Gforward) and ACGCATGGTGACAAATTCTTCG (Greverse). PCR products were inserted into the vector Champion TM pET101 Directional TOPO Expression (Invitrogen®). Correct gene insertion was checked by PCR using primers pETForward (ATGCGTCCGGCGTAGAGG) and pETReverse (GCTAGTTATTGCTCAGCGGTGG). Their identity was confirmed by DNA sequencing. The expression of the cloned gene was induced with 1 mM IPTG for 16 h. GorA was purified by Ni-affinity chromatography (HisTrap HP, GE Healthcare).

Physiological characterization of the isolates was carried out using standard biochemical assays (API ZYM and API 20E; BioMérieux Inc.), which were conducted according to the manufacturer's instructions. Morphological analyses were initially done by Gram staining and in the case of selected strains, also by scanning electron microscopy (SEM) as described by Arenas et al. (2014a). Briefly, samples were visualized and analyzed by low voltage electron microscopy using a LVEM5 Benchtop equipment (Center for Bioinformatics and Molecular Simulation).

Bacterial identification was accomplished by sequencing the 16S rRNA gene using the universal primer 8F (forward) and 1492R (reverse). GeneBank accession numbers for the 16S rRNA nucleotide sequences are: MG461635 (MF01), MG459178 (MF02), MG461634 (MF03), MG461632 (MF04), MG461629 (MF05), MG461631 (MF06), MG461630 (MF07), MG461633 (MF08), MG461636 (MF09), and MG459165 (MF10).

Phylogenetic trees were constructed using the MEGA 6.06 program applying the neighbor-joining method (Saitou and Nei, 1987), while distances were calculated with bootstrapping (100) applying the Tamura-Nei method (Tamura et al., 2004). Bacterial identity was also assessed by analyzing the fatty acid composition at the DSMZ Institute (Braunschweig, Germany).

Determination of minimal inhibitory concentrations (MIC) was carried out using serial dilutions (1: 2) of sterile solutions of K₂TeO₃, AgNO₃, HAuCl₄, K₂CrO₄, CdCl₂, NaAsO₂, CuSO₄ and Na₂SeO₃ in 1 ml of LB medium in 48-well plates. Subsequently, 10 μl of cultures grown in LB medium up to OD₆₀₀ ~0.6 were added to each well and incubation proceeded with constant shaking at the optimal growth temperature of the particular isolate. MICs were determined after 24 h of incubation. For growth curve construction, overnight cultures were diluted 1:100 with fresh LB medium and incubated with shaking up to OD₆₀₀ ~ 0.6. Then, 10 μL were added to 1 mL of fresh LB medium containing 4 μM K₂TeO₃, 15 μM AgNO₃ or HAuCl₄. Bacterial growth was monitored at 600 nm every 30 min for 14 h in a TECAN Infinite M200 Pro multimode plate reader.

Metal(loid)-reducing activity was determined in a final volume of 200 μl which contained 50 mM Tris-HCl buffer pH 8.0 (or 9.0), 50 mM potassium phosphate pH 7.0, 1 mM NADH (or NADPH), 1 mM β-mercaptoethanol and the crude extract; metal(loid)s tested for reduction were TeO32- (1 mM), SeO32- (1 mM), AgNO₃ (0.2 mM) and HAuCl₄ (1 mM). Reactions were monitored in a TECAN Infinite M200 Pro multimode plate reader at the wavelength that the metal(loid) absorbs in its elemental state: 500 nm (Te; Pugin et al., 2014), 400 nm (Se; Song et al., 2017), 424 nm (Ag; Kumar et al., 2007), and 540 nm (Au; Correa-Llantén et al., 2013). The optimal buffer and best nicotinamide cofactor were determined for each crude extract; an enzyme unit (U) was defined as the amount of enzyme required to increase the absorbance by 0.001 units/min at the respective wavelength. Metal(loid)-reducing activity was also assessed by native polyacrylamide gel electrophoresis (native PAGE). Briefly, proteins in crude extracts were fractionated in 10% gels; after the run, activity was revealed by submerging the gel in the appropriate buffer containing the metal(loid) and a mixture of 1 mM NADH/NADPH as electron donor.

The isolates selected for synthesizing NS in vivo were grown to exponential phase (OD₆₀₀ ~ 0.5) and treated with ¼ of the metal(loid) MIC (TeO32-, SeO32-, Ag⁺ or AuCl4-), incubated for 4 h and then centrifuged for 10 min at 9,000 g. The bacterial pellet was sent for thin sectioning and transmission electron microscopy (TEM) analysis (Advanced Microscopy Unit (AMU), at Pontificia Universidad Católica de Chile).

Formation of NS in vitro was carried out using 150 μg/ml of purified recombinant glutathione reductase (GorA) from E. cloacae (isolate MF01). Additionally, crude extracts that contained GorA overexpressed for 2 h were used with reduction buffer (50 mM potassium phosphate pH 7.0, 1 mM NADPH, 1 mM β-mercaptoethanol and 0.15 mM TeO32- or 1 mM SeO32-). In addition, crude extracts (200 μg/ml of protein per reaction) from strains of interest were used for the production of NS by incubation with 1 mM Ag⁺, AuCl₄, TeO32- or 10 mM SeO32- for 16 h in reduction buffer under optimal pH conditions and the preferred nicotinamide cofactor.

A drop containing in vitro generated NS was placed on a copper grid and NS morphology was analyzed by TEM using a Hitachi Transmission Electron Microscope HT7700. Their chemical composition was determined through X-ray energy dispersion spectroscopy (EDS) using a SEM Zeiss EVO MA 10, EDS Penta FET Precision X-act Oxford Instruments. Both studies were conducted at the Center for the Development of Nanoscience and Nanotechnology–CEDENNA, USACH. In turn, the hydrodynamic diameter of the metallic nanostructures was assessed by dynamic light scattering (DLS) using a Marlvern Zetasizer Nano ZS equipment at room temperature (25°C) as described by Arenas-Salinas et al. (2016). Final values were calculated based on three independent measurements of twenty repetitions. Furthermore, metal(loid)s were quantified using total reflection X-ray fluorescence spectrophotometry (TXRF). Briefly, in vitro-synthesized NS were centrifuged at 13,000 × g for 90 min and the sediment was suspended in 200 μl of 65% HNO₃. Samples were heated at 60°C ~2 h and analyzed in a TXRF Bruker S2 PICOFOX Spectrometer at Bioinformatics and Gene Expression laboratory of the Institute of Nutrition and Food Technology (INTA, University of Chile).

A suspension of in vitro synthesized NS was used to challenge Escherichia coli and Listeria monocytogenes. The NS were dialyzed against 50 mM Tris-HCl buffer (pH 8.0 or 9.0) or 50 mM potassium phosphate (pH 7.0), accordingly to the buffer used to generate them. Dialysis was for 4 h at room temperature with constant shaking. The dialyzed material was maintained at 4°C until further analysis. Concentrations of these metallic NS were calculated by determining the dry weight which was expressed in μg/ml. MIC determinations of the dialyzed NS and growth curves of E. coli BW25113 and L. monocytogenes were carried out as described in section Growth, Identification, and Characterization of the Environmental Strains.

Plots and statistical analyses were carried out using the GraphPad Prism 6.0 (GraphPad Software, Inc.). Analysis of variance (ANOVA) and t-test were used considering p < 0.05. The statistical significance was indicated as follows: ^*, for values where p < 0.05, ^**, for values where p < 0.01, ^***, for values where p < 0.001 and ^****, for values where p < 0.000; ns, no significant difference.

A total of 47 bacterial strains isolated from environmental samples were assessed for metal(loid) resistance. Cells were routinely grown in LB or LB-agar plates at 37°C and exposed to different concentrations of 19 different, defined metal(loid)s. None of them grew in Cd²⁺-, Al³⁺-, or Mn²⁺-containing LB medium nor in R2A, Acidiphilium, or ATB culture media. Table S1 shows the ability to grow in the presence of the indicated metal(loid)s; when reduced, some of them (e.g., Te, Se, Au, and Ag) showed a characteristic color that was monitored spectrophotometrically. In turn, while strain MF47 was the most sensitive to all metal(loid)s tested, strains MF45 and MF46 tolerated only Pb²⁺ and Te⁴⁺, respectively.

Ten strains (MF01-MF10), including cocci, bacilli and short-bacilli, from which 21 and 26 were Gram negative and Gram positive, respectively (Table S2), were selected for further studies because of their ability to grow in the presence of 15, 12, 11, 11, 11, 9, 13, 9, 13, and 9 different metal(loid)s and/or to reduce them (Table S1). They belong to the Enterobacter (MF01 and MF04), Staphylococcus (MF02, MF07, and MF10), Acinetobacter (MF05 and MF09), and Exiguobacterium (MF03, MF06, and MF08) genera, as determined by 16S rRNA gene sequencing (Figure S1, Table S2). Their fine morphology was analyzed by scanning electron microscopy (Figure 1): MF02, MF07, and MF10 were coccoid, MF01, MF03, MF04, MF06, and MF08 showed bacillar forms and MF09 and MF05 were cocobacilli. Their physiological characteristics, as determined by biochemical tests, are listed in Table 1. Excepting MF03, all strains used citrate as carbon source, but none produced indole or hydrogen sulfide. All strains were oxidase negative and catalase positive and while MF05, MF06, MF07, MF08, MF09, and MF10 did not oxidize sugars at all, MF01 oxidized almost all sugars assayed but inositol.

Minimal inhibitory concentrations of K₂TeO₃, AgNO₃, HAuCl₄, K₂CrO₄, CdCl₂, NaAsO₂, CuSO₄, and Na₂SeO₃ were determined under aerobic growth conditions for the strains listed in Table S2. MF02 and MF04 were the most resistant and sensitive to tellurite, respectively. All strains exhibited similar resistance levels to HAuCl₄, CdCl₂, AgNO₃, and CuSO₄, while MF01 and MF03 were the most sensitive to NaAsO₂. Differences up to 500-fold in the MIC of K₂CrO₄ were observed, MF02 and MF08 tolerating the highest concentration (400 mM). Finally, while MF01 and MF02 grew in the presence of Na₂SeO₃ up to 500 mM, MF10 was the most sensitive isolate (3.125 mM). Then, the effect of sublethal concentrations of TeO32-, AuCl4- and Ag⁺ on bacterial growth was assessed (Figure 2). Although an extended lag phase was observed, cells regained growth between ~4 and ~7 h after exposure to the metalloid. Growth of MF01, MF04, and MF07 was completely abolished in the presence of tellurite.

On the other hand, crude cell extracts were assayed to assess tellurium (IV)-, selenium (IV)-, gold (III)-, and silver (I)-reducing activity in the presence of NADH or NADPH and in the pH range 7–9 (Figure 3 and Figure S2). The reduction of three metal(loid)s generated a change of the solution's color: black, purple and yellow-orange in the case of tellurite (Figure S2A), AuCl4- (Figure S2B) and Ag⁺ (Figure S2C), respectively. While MF09 was the best tellurium (IV) reducer (pH 8-9 with NADH as cofactor, Figure 3A), selenium (IV) was better reduced by MF02 at pH 7.0 using NADPH as cofactor (Figure 3B). In turn, MF03 exhibited the best Ag (I)-reducing activity at pH 7.0 in the presence of NADH (Figure 3C). Finally, a robust gold-reducing activity was observed in MF09 at pH 8.0 with NADH as the enzyme cofactor (Figure 3D).

To look for proteins which could be responsible of metal(loid) reduction, tellurite- and gold- reducing activities were assessed in situ by polyacrylamide gel electrophoresis under non-denaturing conditions (Figures S3A,B). Reduction bands generated by MF01 extracts were excised from the gel and fractionated by SDS-PAGE (Figure S3C). Twelve bands were identified by MALDI-TOF (Figure S3D) which included the enzyme glutathione reductase (GorA), belonging to the flavoprotein family. The ability of GorA from the Antarctic strain Pseudomonas sp. BNF22 to reduce TeO32- had been observed previously in our laboratory (Pugin et al., 2014). However, its ability to reduce other metal(loid)s and/or its efficiency has not yet been determined, which prompted us to focus on this phenomenon. The gorA gene was amplified from the E. cloacae genome and cloned into the pET/101 expression vector to generate the pET/gorA recombinant plasmid (Figures S3E,F). A kinetic induction assay showed a ~50 kDa protein band that accumulated over time (Figure S3G). The enzyme was purified and assayed for tellurium (IV), selenium (IV), gold (III), silver (I), and copper (II) reduction, measured in the presence of NADH or NADPH and in the pH range 6–9. GorA reduced optimally TeO32-, SeO32- and Ag⁺ at pH 8.0, 6–7 and 7, respectively (Figure S3H).

Ultrathin sections of selected strains that were exposed to sublethal concentrations (¼ of the MIC) of defined metal(loid)s were analyzed by transmission electron microscopy (Figure 4). Intracellular, evenly distributed long electron-dense rods, representing most probably tellurium nanostructures (TeNS) were observed in tellurite-exposed MF09 (Figure 4A). In turn, SeNS generated by MF02 appeared as spherical particles located near the cell membrane (Figure 4B). In both cases, alterations in size, morphology and cell division were observed (not shown). No silver-containing particles were generated by MF03 under the same conditions (Figure 4C). Finally, small, circular and evenly distributed gold particles were seen when MF09 was treated with sub lethal gold concentrations (Figure 4D).

On the other hand, in vitro generation of NS by bacterial crude extracts lasted ~16 h, irrespective of the strain or metal(loid) tested. NS morphology, size and chemical composition were determined by transmission electron microscopy (TEM), dynamic light scattering (DLS) and total reflection X-ray fluorescence spectrophotometry (TRXF), respectively. TeNS synthetized using MF09 extracts showed heterogeneous morphology with an average diameter ~10 nm (Figure 5A, left panel) and a hydrodynamic diameter ~9.36 nm (polydispersity index PDI 0.446), in agreement with TEM results (Figure 5A, middle panel). The chemical composition was 23.2% tellurium, 55.8% carbon and low amounts of chlorine, potassium, and phosphorus (Figure 5A, right panel). Moreover, TXRF analysis indicated that the tellurium concentration in TeNS was 78.66 ± 7.51 μg/ml. In the case of the SeNS generated with crude extracts from MF02, nanoparticles showed a spherical morphology with regular edges—nanospheres—and a size of ~100 nm (Figure 5B, middle panel); they showed a gaussian distribution (DLS) with a maximum at 105.1 nm (PDI 0.124) (Figure 5B, middle panel). Selenium content in SeNS was 83.9%, while TRXF showed that Se concentration was 158.98 ± 16,4 μg/ml (Figure 5B, right panel). On the other hand, AgNS generated using crude extracts from MF03 showed a size of ~10 nm, spherical-like shape with irregular edges and an electron-dense core (Figure 5C, left panel). These AgNS also showed a gaussian distribution with a maximum centered at 68.46 nm (PDI 0.43) (Figure 5B, middle panel), a size larger than that assessed by TEM; AgNS contained mostly carbon and to a lesser extent silver (9.84 ± 1.05 μg/ml, representing 20.7%) (Figure 5C, right panel). The same MF09 crude extract used to prepare TeNS was utilized to generate AuNS, whose morphology was not easy to determine because of the presence of organic material surrounding the particles (Figure 5D, left panel). Nevertheless, these NS were rather spherical with a size ~50 nm. AuNS showed a size distribution with a maximum at 82.35 nm (Figure 5D, middle panel). These nanoparticles contained 21.2% gold and other elements in lower amounts such as calcium, magnesium, chlorine, sulfur, and phosphorus (Figure 5A, right panel). TXRF analysis indicated that AuNS exhibited a gold concentration of 77.23 ± 12.61 μg/ml.

Finally, crude extracts from GorA-overproducing E. coli as well as purified GorA were used to synthesize in vitro SeNS and TeNS (Figure 6): (i) both, crude extracts and purified protein were able to generate spherical SeNS with a size of ~50 nm and irregular edges (TEM). DLS analysis indicated that GorA-synthesized NS are prone to aggregation with a size distribution of 74.75 nm (PDI 0.159) (Figure 6A); conversely NS obtained using crude extracts did not aggregate and showed a size of 39.09 nm (PDI 0.456) (Figure 6C). SeNS from crude extracts showed a higher amount of selenium (131.65 ± 16.46 μg/ml, representing 29.9%) than GorA-SeNS (164.93 ± 24.7 μg/ml, representing 26.6%) (Figures 6A,C); (ii) in turn, GorA-TeNS showed a rather stick-like morphology (Figure 6B), while nanoparticles synthetized using crude extracts exhibited irregular, non-homogenous morphology, with some of them showing triangular shapes (Figure 6D). Size distribution (DLS) of GorA- and crude extract-TeNS was 35.11 nm (PDI 0.32) and 31 nm (PDI 0.63), respectively (Figures 6B,D). TeNS from crude extracts showed a higher amount of tellurium (127.36 ± 15.26 μg/ml, representing 10%) than GorA-TeNS (157.6 ± 25.3 μg/ml, representing 8.2%) (Figure 6B,D). Both crude extracts and purified protein showed the presence of other elements (more abundant in the GorA solution) such as potassium, sodium, sulfur, and phosphorus, among others.

The minimal inhibitory concentration of in vitro synthesized NS was determined and along with the construction of growth curves, were used to assess their putative antibacterial effects. Aerobically synthesized TeNS using MF09 crude extracts showed MICs of 45- and 66- μg/ml for E. coli and L. monocytogenes, respectively. Similar MIC values (40 and 82 μg/ml, respectively) were observed for TeNS generated using GorA-overproducing E. coli crude extracts. In turn, AuNS MICs for E. coli and L. monocytogenes were 64- and 68- μg/ml, respectively. At the tested concentrations, both SeNS and AgNS did not inhibit E. coli or L. monocytogenes growth. However, Te- and Au-containing NS inhibited completely the growth of E. coli and L. monocytogenes (Figure S4).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.