Enhanced Glutathione Content Allows the In Vivo Synthesis of Fluorescent CdTe Nanoparticles by Escherichia coli

The vast application of fluorescent semiconductor nanoparticles (NPs) or quantum dots (QDs) has prompted the development of new, cheap and safer methods that allow generating QDs with improved biocompatibility. In this context, green or biological QDs production represents a still unexplored area. This work reports the intracellular CdTe QDs biosynthesis in bacteria. Escherichia coli overexpressing the gshA gene, involved in glutathione (GSH) biosynthesis, was used to produce CdTe QDs. Cells exhibited higher reduced thiols, GSH and Cd/Te contents that allow generating fluorescent intracellular NP-like structures when exposed to CdCl2 and K2TeO3. Fluorescence microscopy revealed that QDs-producing cells accumulate defined structures of various colors, suggesting the production of differently-sized NPs. Purified fluorescent NPs exhibited structural and spectroscopic properties characteristic of CdTe QDs, as size and absorption/emission spectra. Elemental analysis confirmed that biosynthesized QDs were formed by Cd and Te with Cd/Te ratios expected for CdTe QDs. Finally, fluorescent properties of QDs-producing cells, such as color and intensity, were improved by temperature control and the use of reducing buffers.


Introduction
Semiconductor nanoparticles (NPs) or quantum dots (QDs) are bimetallic structures of elements like Cd, S, Se or Te that given their particular physicochemical and optoelectronic properties exhibit great technological potential. These semiconductor NPs have the ability of absorbing light and emitting the exceeding energy in the form of fluorescence at high yields. The absorption and emission spectra of QDs depend on size and shape of the nanoparticle due to a quantum confinement effect. The idea of slightly changing the shape of these NPs and hence their optical properties has made them very popular in optoelectronics. In particular, CdTe QDs are used in electronic and optoelectronic devices and during the last decade, as an important tool for new solar cell technology (photovoltaic panels) and in biomedicine [1,2].
Currently, CdTe QDs synthesis involves chemical procedures requiring high temperature, anaerobic conditions, toxic substrates/residues and displays unfavorable capital/energy ratios. As a consequence, QDs produced by these methods involve organic solvents and the resulting nanoparticles are not capped with soluble, more compatible agents. This renders chemically synthesized Quantum dots difficult to work with because of its insolubility and elevated citotoxicity due to cadmium, thus decreasing their potential applications. Development of new, simpler, less toxic and environmentally friendly synthesis procedures is then a subject of growing interest.
Metal-bacteria interactions play a crucial role in a number of biotechnological applications [3]. Microorganisms in general and bacteria in particular, are able to produce different metallic nanoparticles as gold, silver or iron particles, among others [4][5][6]. During the last decade, some advances were made regarding the mechanism underlying the synthesis of iron NPs by magnetotactic bacteria [7,8]. These bacteria synthesize intracellular magnetic nanocrystals consisting of Fe 3 O 4 or Fe 3 S 4 (35-120 nm) that are surrounded by a membranous structure known as magnetosome [7]. Although some work has been dedicated to the control of NP shape and size in biologically synthesized silver and gold nanoparticles [9,10], the molecular functions that govern this processes are still unknown.
A few reports regarding the use of microorganisms for synthesizing fluorescent semiconductor NPs have been published to date. Some experiments related to CdS QDs synthesis using E. coli and the yeast Schizosaccharomyces pombe have been reported [11,12]. Also, extracellular CdSe biosynthesis using Fusarium oxysporum or Saccharomyces cerevisiae extracts has been recently communicated [13,14]. The importance of metal-binding peptides in the biosynthesis process was established in all these communications. For example, the Arabidopsis thaliana protein pythochelatin was used for QDs synthesis in E. coli and S. cerevisiae [11,14]. Recently, the use of E. coli overexpressing a specific histidine-rich peptide with affinity for CdS that facilitates the intracellular synthesis of CdS QDs was reported [15].
Recently, Bao et al. [16,17] reported the extracellular synthesis of CdTe QDs by E. coli and S. cerevisiae. CdTe synthesis occurred in LB rich media and the presence of cells, cadmium and tellurium salts (CdCl 2 , Na 2 TeO 3 ) and the strong chemical reducer NaBH 4 was required. Interestingly, when compared to chemically-produced CdTe QDs, S. cerevisiae-biosynthesized QDs exhibited increased biocompatibility in HeLa cells [16,17]. Also, during 2010 Park et al. [18] reported the intracellular synthesis of CdTe and other semiconductor NPs by overexpressing heterologous metal-binding proteins such as metallothionein or phytochelatin in E. coli.
Cellular events underlying the molecular mechanism(s) involved in bacterial production of QDs are still unknown and optimal conditions favoring CdTe QDs biosynthesis have not been determined to date. These considerations, in conjunction with unveiling biosynthetic mechanisms should allow the design of NPs with defined properties. In this context and to establish the basis for the biological synthesis of CdTe QDs, a procedure mimicking biological conditions was developed for the chemical synthesis of CdTe NPs [19]. The method requires just CdCl 2 , K 2 TeO 3 and GSH as substrates and capping/reducing agent. Highly fluorescent QDs were obtained at temperatures, pH values and oxygen conditions similar to those found in microbial cells.
Taking into consideration GSH i) redox properties, ii) its abundance in cells and iii) interaction with K 2 TeO 3 and CdCl 2 [20][21][22][23][24], this work describes, for the first time, using this important biological thiol for intracellular CdTe QDs production by bacteria.

E. coli GSH Content Assessment
Since GSH is both a stabilizing and reducing agent during the chemical synthesis of CdTe QDs [19], we hypothesized that a bacterial strain containing increased levels of this tripeptide should be a good candidate for biosynthesizing this kind of NPs. To enhance GSH concentration, genes encoding the principal enzymes involved in glutathione biosynthesis, gshA and gshB [25,26], were overexpressed in E. coli. Total RSH content was first determined in AG1, AG1/pCA24NgshA and AG1/pCA24NgshB strains (Fig. 1A). Despite a maximum expression level of GshA and GshB was observed after 4 h IPTG induction (Fig. S1), increased levels of cellular RSH were only observed in the gshA-expressing strain (Fig. 1A). To assess if it was paralleled by GSH overproduction, the intracellular GSH content was determined. As expected, GSH concentration increased in E. coli AG1/pCA24Ng-shA, but not in wild type or gshB-expressing cells (Fig. 1B).
Given that increased nanocrystal production has been observed in some microorganisms during stationary growth phase [27,28], the GSH content was assessed during late stationary phase (after 16 h of IPTG induction). After this time, the GSH content in the gshA-expressing strain decreased as compared to that observed after 4 h induction (Fig. 1B). In the light of these results, the 4 hinduced AG1/pCA24NgshA strain represents the best choice to carry out QDs biosynthesis.

Nanoparticle Biosynthesis
Biosynthesis of fluorescent, semiconductor NPs was monitored in E. coli AG1/pCA24NgshA and AG1/pCA24NgshB strains as described in Methods. Given the intrinsic fluorescence properties of CdTe QDs, cell-based NP biosynthesis was assessed by exposing cell pellets, after metal treatment, to UV light (Fig. 2). Fluorescence was only observed in CdCl 2 -and CdCl 2 /K 2 TeO 3 -exposed E. coli AG1/pCA24NgshA pellets ( Fig. 2A, 2 and 4, respectively). No fluorescence was observed in untreated E. coli AG1/ pCA24NgshA or in cells exposed only to K 2 TeO 3 ( Fig. 2A, 1 and 3, respectively). As expected, AG1/pCA24NgshB cells did not produce fluorescent structures under any treatment (Fig. 2B). No fluorescence was observed in cell supernatants, suggesting that produced fluorescent structures are either inside cells or associated with the cell surface. Fluorescence of Cd +2 -exposed cells most probably reflects CdS NPs, as has been previously described in cells expressing thiol-rich metal-binding peptides [11,14].

Metal Content Assessment of the Biosynthesized NPs
To evaluate metal content in fluorescent cells, ICP-AES elemental analysis of cell pellets was carried out. Results showed that in most conditions tested the AG1/pCA24NgshA strain displayed higher Cd and Te content than AG1/pCA24NgshB cells (Table 1). Interestingly, despite Te per cent increase in AG1/ pCA24NgshA cells exposed to CdCl 2 /K 2 TeO 3 , no differences of intracellular Cd content were observed. Given that under these conditions cells do not display black deposits (Te 0 ) this result suggests that fluorescence is probably a consequence of CdTe production. The ICP metal determinations clearly indicates that when E. coli wt, gshA or gshB strains were exposed to metals, only the fluorescent cells (E. coli gshA) display higher intracellular levels of Cd and Te as compared to none fluorescent cells (E. coli wt or gshB).

Fluorescence Microscopy of AG1/pCA24NgshA cells
This technique was used to further characterize bacterial cell fluorescence. Morphology of the AG1/pCA24NgshA strain exposed to the metals under study indicates that cells were highly stressed since filamentous and fragmented cells are observed after 24 h exposure. Epifluorescence microscopy revealed that fluorescence was not evenly distributed in the bacterial population and just some cells were fluorescent ( Fig. 3A and B). Interestingly, fluorescence seems to accumulate in well defined structures which are found both inside cells and in the extracellular milieu (Fig. 3B). When exposed to UV light, cells displayed different fluorescence colors (green, yellow, red and blue, Fig. 3C), suggesting that CdTe QDs of different sizes and/or shapes are being produced and also that the biosynthesis process is not uniform under these experimental conditions.

Purification and Characterization of Biosynthesized NPs
To demonstrate that the observed cell fluorescence is the consequence of CdTe QDs biosynthesis, these structures were purified and characterized. As shown in Figure 4A, concentrated cell suspensions of CdCl 2 -or CdCl 2 /K 2 TeO 3 -treated bacteria displayed fluorescence when exposed to UV light. Untreated cells did not display fluorescence and were used as control. After cell disruption, fluorescent structures were purified as described in Methods. When exposed to UV light, fractions from metalexposed cells became fluorescent, as contrasted to those of control, untreated cells (Fig. 4B).
A Dynamic Light Scattering (DLS) analysis was carried out to confirm the presence of nano-sized material in fluorescent fractions. Fluorescent material purified from CdCl 2 -or CdCl 2 / K 2 TeO 3 -treated E. coli AG1/pCA24NgshA exhibited nanostructured material whose dimensions averaged 5.98 and 4.8 nm, respectively ( Fig. 4C and D), which are in agreement with those reported for CdS and CdTe QDs, respectively [11,28,19,29]. This size could also represent thiol-or GSH-capped QDs, since it falls in the size category of previously synthesized CdTe-GSH NPs [19,29]. The DLS analysis also shows a wide size distribution of NPs ( Fig. 4C and D), a result that is in agreement with the different colors of cells as determined by fluorescence microscopy (Fig. 3C). Additional size information for the CdTe sample was gathered through atomic force microscopy (Fig. S2); particle size was shown to be ,2-3 nm, a result that is consistent with nanocrystal sizes reported by other authors [16,17,19,29].
To confirm that these fluorescent NPs contain the expected metallic elements, an elemental analysis of purified samples was carried out. ICP-AES analysis of purified samples from cells exposed to either only CdCl 2 or CdCl 2 /K 2 TeO 3 indicated that fluorescence was related to the presence of Cd or Cd/Te, respectively ( Table 2). In addition, the Cd/Te ratio (,7) is close to that expected for CdTe NPs and is in agreement with those previously reported for CdTe-GSH QDs [19,29].
On the other hand and since they are well defined characteristics of QDs, absorption and emission spectra of purified samples were determined. A noticeable difference in the absorbance spectra (300-450 nm) was observed between cell treatments and untreated controls (Fig. 5A). Absorption spectra were those expected for biosynthesized QDs and are in agreement with those previously reported for CdS and CdTe QDs [11,16,17,28].
Samples from CdCl 2 -exposed cells displayed a fluorescence maximum at 350-400 nm (Fig. 5B), characteristic of CdS QDs [11]. On the other hand, a peak above 450 nm was determined in the sample purified from cells exposed to CdCl 2 /K 2 TeO 3 (Fig. 5C). This fluorescence peak corresponds to that expected for CdTe QDs [16,17]. These spectroscopic properties indicate that CdS and CdTe NPs are being bacterially produced under these experimental conditions. XRD analyses were performed to further confirm that the produced NPs are crystal-shaped. The XRD peaks were located between positions for a pure cubic CdTe crystal (Fig. 6) and for CdS greenockite crystals (not shown). The strong peak at 27,5u observed in biosynthesized CdTe NPs corresponds to the (111) planes of the standard pattern for cubic CdTe [30]. Altogether, XRD results indicate that biosynthesized QDs have a crystalline structure similar to those produced by chemical methods and particularly to biosynthesized CdTe QDs [30,31,32,33].

Conditions Favoring QDs Biosynthesis
Conditions such as increased incubation temperature, microaerophilic environments and the use of reducing buffers (citrate), known to favor the chemical synthesis of CdTe QDs [19], were tested for their effect upon bacterial NPs biosynthesis. Fig. 8 shows that with increasing incubation temperature (42uC), the fluorescence of cell pellets is enhanced when exposed to UV light (compare 1 and 2). Interestingly, incubation of bacteria with citrate buffer changes fluorescence color, suggesting that NPs with different spectroscopic properties are produced (compare lane 1 and 3). This could be the consequence of size, shape and/or composition changes mediated by a specific cellular status or factors. On the other hand, microaerophilic conditions did not seem to favor or alter the cell fluorescence, as compared to controls (compare lane 1 and 4). Other environmental factors such as increased pH values seem to prevent NP formation and no fluorescence was observed when cells were incubated at basic pH (Fig. S3).

Discussion
As a consequence of the multiple applications of nano-sized compounds, interest in manufacturing nanomaterials has grown exponentially during the last decade. Unfortunately, most NPsynthesizing procedures described to date are expensive, require dangerous compounds and produce NPs with elevated toxicity [34,35]. In this context, an alternative procedure to increase NPs biocompatibility has been the use of non-toxic stabilizing agents that reduce the oxidant properties of NPs and increase their solubility in aqueous systems [36,37]. Based on these considerations, a number of thiols as mercapto succinic acid (MSA), thioglycolic acid (TGA) and mercapto propionic acid (MPA), among others, have been successfully bound to QDs [31,[38][39][40]. Also, the use of biological thiols like GSH or cysteine as QDsstabilizing agents has been reported [32,33].
Recently, and to determine the minimal conditions for synthesizing CdTe QDs using biocompatible agents, we developed a biomimetic chemical synthesis procedure that requires only GSH as a reducing and stabilizing agent and Cd and Te salts as substrates [19]. This protocol functions under mild conditions of temperature, pH and requires substrates that are normally found in microorganisms such as GSH as well as compounds like CdCl 2 and K 2 TeO 3 whose effects in cells have been documented [19,23,41]. GSH plays a crucial role acting as Te 4+ to Te 22 reductant and, at the same time, stabilizing NPs by interacting with Cd 2+ to form a cap of oxidized thiol [19].
Most procedures for the biological synthesis of cadmium QDs such as CdS, CdSe or CdTe described to date use Cd 2+ salts as . Particle purification and size determination. A, UV-exposed cell suspensions of E. coli AG1/pCA24NgshA, untreated or exposed to CdCl 2 or CdCl 2 /K 2 TeO 3 (from left to right). B, purified fractions exposed to UV light. C and D, DLS particle size determination of fluorescent samples from cells exposed to CdCl 2 or CdCl 2 /K 2 TeO 3 , respectively. doi:10.1371/journal.pone.0048657.g004 cadmium donors [11][12][13][14][15]. On the other hand, we believe that Te 22 , required for CdTe biosynthesis, could be generated by GSH-mediated intracellular tellurite reduction. Moreover, cells exposed to tellurite become dark as consequence of Te 4+ -Te 0 reduction, a phenomenon that is mediated -at least in part-by cellular thiols such as GSH [22] and enzymes as nitrate reductase, catalase and pyruvate dehydrogenase [42,43,44]. Although reduction of Te 4+ to Te 22 is an energetically unfavorable process, we propose that this chemical reduction can be improved by modifying culture conditions or manipulating the expression of defined genes. Evidence of this is the intracellular Te 22 generation observed during the evolution of methylated Te compounds such as (CH 3 ) 2 Te in E. coli overexpressing the Geobacillus stearothermophilus V ubiE gene [45]. Based on the experimental conditions of the biomimetic protocol [19] and the known interactions of GSH with CdCl 2 and K 2 TeO 3 , an E. coli strain with increased levels of GSH was used to biosynthesize CdTe QDs in the presence of CdCl 2 and K 2 TeO 3 . The E. coli AG1/pCA24NgshA strain displayed higher RSH and GSH levels than the wt or AG1/pCA24NgshB strains ( Fig. 1 A and B). Despite the fact that gshA and gshB genes are both involved in GSH biosynthesis and also because GshA catalyzes the final and limiting step in GSH synthesis, it was not surprising to find higher glutathione content in the gshA-expressing strain [25]. Thus, this strain represented an excellent candidate to test the role of GSH as reducing and stabilizing agent in the biological synthesis of CdTe QDs. As expected, after exposure to sub-lethal Cd and Te concentrations, cells overexpressing gshA displayed fluorescence under UV excitation ( Fig. 2A). In contrast, no fluorescence was observed in wt or gshB-expressing cells; GSH is thus favoring the synthesis of fluorescent structures. Since fluorescence is an intrinsic QDs feature, this result represents the first evidence suggesting that QDs are being produced by bacteria (Fig. 2). Fluorescent cells displayed higher Cd and Te amounts (Table 1), a result that correlates fluorescence with the incorporation of metal substrates for QDs biosynthesis. This Figure 5. Absorbance and fluorescence spectra of purified NPs. A, absorption spectra of NPs from unexposed (solid black line), exposed to CdCl 2 (dashed grey line) or CdCl 2 /K 2 TeO 3 (dashed black line) cells. B and C, emission spectra of NPs from cells exposed to CdCl 2 or CdCl 2 /K 2 TeO 3, respectively. doi:10.1371/journal.pone.0048657.g005  Biological CdTe NPs where purified as described in Methods and chemical, biocompatible CdTe QDs where synthesized as described [19]. FTIR spectroscopy spectra were recorded and compared with a GSH standard. doi:10.1371/journal.pone.0048657.g007 observation could be the basis for future applications in heavy metal bioremediation associated with the production of QDs for biotechnological purposes. In this context, preliminary experiments in our laboratory indicate that upon CdCl 2/ K 2 TeO 3 exposure, other microorganisms like Saccharomyces cerevisiae and Aeromonas caviae ST produce fluorescent structures whose spectroscopic properties resemble that of CdTe QDs (not shown). This data also suggests that QDs biosynthesis seems to not be restricted to a particular species, thus facilitating putative biotechnological applications.
Recently, the extracellular synthesis of CdTe QDs using E. coli and S. cerevisiae was reported [16,17]. CdTe NPs were produced in LB growth medium in the presence of MSA, CdCl 2 and Na 2 TeO 3 , and require adding the strong chemical reducing agent NaBH 4 . Those authors proposed a mechanism in which microorganisms secrete some unidentified protein factor that allows the chemical synthesis of CdTe QDs in the extracellular milieu. Interestingly, we were able to synthesize our QDs under conditions similar to those described by Bao et al. (2010) but in the absence of bacterial cells, i.e. using only CdCl 2 , K 2 TeO 3 , NaBH 4 , LB medium and MSA. Highly fluorescent QDs were obtained in LB media (pH 7.0) in the presence of 4 mM CdCl 2 , 1 mM K 2 TeO 3 , 15 mM MSA and 10 mM NaBH 4 after 2 and 3 h synthesis (Fig. S4). Nevertheless, extracellular CdTe QDs synthesis is still puzzling, and the mechanism(s) involved in the biosynthesis are unknown.
Fluorescence microscopy was the first approach to characterize the bacterially-produced fluorescent structures. Even though not all cells were fluorescent, those displaying fluorescence exhibited a common morphology with vesicles-like structures which can be found inside or outside cells (Fig. 4). When samples were excited at 330 or 488 nm a highly fluorescent spot was observed in these structures. Since a wide range of excitation wavelengths is another intrinsic property of CdTe QDs, these results suggest that fluorescent semiconductor nanostructures are being produced. A possibility is that these vesicle-like structures actually correspond to inclusion bodies where NPs are accumulated, which would be consequence of GshA overproduction or the metal-mediated stress (evidenced by cell damage and filamentation). In this context, the fluorescent structures described here resemble outer membrane vesicles that some Gram negative bacteria release in response to envelope stress [46]. This idea is supported by microarray studies performed by our group in E. coli exposed to nonlethal-tellurite concentrations (unpublished results). The induction of the stress envelope regulators RseA and YaeL, the murein-related yiiX, ddl, lpxC, and some genes related with plasma membrane proteins such as nadH, narI narJ and ndH, among others, suggests the generation of an envelope stress in these conditions. Experiments to unveil the nature of these vesicle-like structures are being carried out in our laboratory.
NPs were obtained by lysozyme treatment and sonic disruption of cells. However, fluorescent structures could not be recovered in the supernatants as has been reported for other QDs [11,16,17,28], and SDS treatment was required to purify them. This suggests that NPs are being produced through different cellular processes in the AG1/pCA24NgshA strain, probably involving the association with the cell membrane or membranederived vesicles.
Elemental analysis confirmed that fractions enriched in fluorescent structures contain the expected metals ( Table 2). Cells exposed to CdCl 2 or CdCl 2 /K 2 TeO 3 displayed a high content of Cd and Cd+Te, respectively, indicating that produced QDs correspond to CdS and CdTe. Particularly in the case of the CdTe fraction, the Cd/Te ratio was slightly higher than those reported for CdTe QDs produced by chemical methods [29]. Higher Cd content is probably the consequence of binding to capping agents which in this case would correspond to GSH (or cysteine), as has been previously described for CdTe-GSH QDs [19].
DLS analysis indicated that purified samples contained nanosized material averaging 5.98 and 4.8 nm for CdS and CdTe QDs, respectively. The ,5 nm size observed for CdTe QDs is within the dimensions expected for a population of QDs conformed mainly by red-colored NPs [19,30]. However, size distribution in the case of both samples reflects that a variable population of NPs with sizes ranging from 4-8 and 3-6 nm for CdS and CdTe, respectively, are produced by the bacterial population. This observation is in agreement with fluorescence microscopy results, which showed cells of different colors when exposed to UV light (Fig. 3). Visualization of purified samples by AFM led to the conclusion that biosynthesized particles with uniform characteristics displayed 2-3 nm sizes in the CdTe sample. All size determination data and XRD analyss of purified samples indicate that CdTe NPs show characteristics of a pure cubic CdTe nanocrystal.
Besides nanometric size, one of the most noticeable QDs features is their special spectroscopic properties. CdS QDs absorb and emit light in the blue zone while CdTe QDs shift their absorption to the red zone of the spectrum [11,16,17,28]. Biosynthesized NPs displayed increased absorption between 300 and 500 nm, similar to that previously reported for biosynthesized cadmium QDs [11,16,17,28]. As expected, a slight absorbance red shift was determined for CdTe QDs (Fig. 5A). Differences observed between biosynthesized and chemically-produced QDs are probably the consequence of light-absorbing biomolecules.
Fluorescence spectra of CdS NPs exhibited a peak near 370 nm, which is in agreement with that reported for CdS QDs produced by cells or chemical protocols [11,34]. In addition, the peak width was ,80 nm, indicating the presence of different-sized NPs. On the other hand, emission spectra of purified NPs from CdCl 2 / K 2 TeO 3 -exposed cells showed a peak at 450 nm, characteristic of CdTe QDs [16,17]. In this case the peak width was ,110 nm, again confirming the presence of different-sized NPs.
Once the production of QDs by E. coli was confirmed, the effect of microaerophilia, culture temperature or the use of reducing buffers on cell fluorescence was investigated. Despite the fact that it has been reported that cellular reduction of Te 4+ is favored by O 2 [43], no effect on cell fluorescence was observed under oxygen deprivation (Fig. 8). On the other hand, slightly higher temperatures improved cell fluorescence upon metal exposure, Figure 8. Effect of temperature, microaerophilic conditions and citrate on E. coli fluorescence. UV-exposed E. coli AG1/pCA24NgshA cells previously incubated under NP biosynthesis conditions with some modifications: 1, control (37uC); 2, increased incubation temperature (42uC); 3, reducing agent (citrate buffer pH 7.0); 4, microaerophilic conditions. doi:10.1371/journal.pone.0048657.g008 confirming the importance of this parameter in CdTe QDs production as shown in the biomimetic method [19]. Fluorescence color was shifted to red when citrate buffer was used, indicating that different NPs are being produced. These observations suggest that little changes in culture conditions can alter NPs properties, opening the opportunity to evaluate new conditions or microorganisms able to produce QDs with desired properties, as color, size and bound molecules. For instance, the effect of temperature offers the chance of using thermophiles for improved biological production of semiconductor NPs. Despite all these observations, optimal conditions and factors to enhance NP biosynthesis by microorganisms is far from being understood.
Finally, results from this work indicate that thiol content and redox state can promote the formation of bimetallic CdTe QDs after treating E. coli with cadmium and tellurium salts. Microorganisms displaying high-reducing environments and/or increased antioxidant defenses as GSH or RSH, represent excellent candidates to develop cell nano-factories for synthesizing different QDs.

Bacterial Strains and Culture Conditions
The E. coli strains AG1 (wild type) and those over expressing the homologous gshA (encoding L-glutamate cysteine ligase) or gshB (encoding glutathione synthetase) genes were from the ASKA collection [47]. E. coli AG1 is the parental strain of those over expressing gshA and gshB and all of them have the same genetic background.
Cells were routinely grown at 37uC in Luria-Bertani (LB) medium with shaking. When required, chloramphenicol (25 mg/ ml) was added to the medium. Growth in liquid medium was normally started with 1/100 dilutions of overnight cultures. Induction of gshA and gshB genes was carried out in the presence of 0.5 mM isopropil-b-D-1-thiogalactopyranoside (IPTG).

Quantification of Intracellular GSH and RSH
Cells were grown to OD 600 ,0.5 and IPTG was added. After inducing for 4 or 16 h, 1 ml aliquots were collected and cells were washed twice with PBS buffer and suspended in 100 ml of 5% sulfosalicylic acid. Cells were disrupted by repeated cycles of freeze/thawing and after centrifuging at 10,0006g for 15 min supernatants were saved and used for RSH determination.

Nanoparticle Biosynthesis
E. coli AG1 harboring plasmid pCA24NgshA was grown to OD 600 ,0.5, 500 mM IPTG was added and after inducing for 4 h, cells were exposed to 54 mM CdCl 2 and 2 mM K 2 TeO 3 for 24 h. After washing twice with 50 mM potassium phosphate pH 7.4 buffer, cells were centrifuged for 15 min at 13.0006g and stored at 280uC until use.
Microaerophilic conditions were obtained by filling culture tubes completely with sterile mineral oil at the top to avoid oxygenation. For high temperature or reducing buffer conditions, cultures were incubated at 42uC or suspended in 50 mM citrate buffer pH 7.0, respectively, immediately after metal addition.

Fluorescence Microscopy
An aliquot of the fluorescent AG1/pCA24NgshA culture was diluted and mounted on a glass slide with 2.5% 1,4-diazobiciclo [2,2,2] octane (DABCO) and glycerol. Samples were assessed for individual cell fluorescence using an Olympus BX51 fluorescence microscope following excitation at 350 nm.
Monochromatic Epifluorescence microscopy (Olympus, model CKX41) was used to identify cell structures and morphology in NP-producing bacteria. Cells were fixed with PBS-4% paraformaldehyde for 30 min at 25uC and washed three times with PBS. Samples were mounted onto slides with 10% Mowiol22.5% DABCO and visualized using an UV-DAPI/FITC filter (Ex 400-420 nm, Em 448-465 nm/Ex 475-500 nm, Em 510-560). Images were processed with MIS Viewer F 3.0 software.

Purification of Biosynthesized Nanoparticles
Cells were suspended in 1 ml of PBS buffer containing 1 mg/ml lysozyme and incubated at room temperature for 30 min. After centrifuging at 5,0006g for 5 min cells were disrupted by sonication (3 min with 30 s intervals). The fluorescent debris was treated with 4% SDS at 90uC for 30 min and incubated overnight at 37uC to solubilize membranes. Afterwards, 1 ml of the solution was ultracentrifuged over a 40-60% sucrose cushion at 300,0006g for 2 h. The fluorescent fraction was detected by exposure to UV light (312 nm) and concentrated using a 30 kDa cut-off membrane. The resulting solution was heated at 90uC for 20 min and centrifuged to remove contaminating proteins. This solution, highly enriched in NPs, was used in NPs characterization experiments.

Nanoparticle Characterization
i) Absorption and fluorescence spectroscopy: to get absorbance and fluorescence spectra, purified NPs were diluted 1/10 with ultrapure water. Absorbance spectra were recorded with a Perkin-Elmer Lambda 11 UV-vis spectrophotometer using MiliQ water as blank. Fluorescence spectra were obtained with an ISS-PC photocounting spectrofluorimeter. An excitation spectra was constructed recording fluorescence emission at 515 nm while exciting the sample at different wavelengths (300-500 nm); ii) Dynamic light scattering (DLS): dynamic light scattering of purified fluorescent fractions was used to evaluate NP size as described before [19]. Data were acquired using a Zetasizer nano S90 (Malvern Instruments Limited, UK) instrument with a refraction index of 2.6 and 4-optical sides disposable cuvettes; iii) Atomic force microscopy (AFM): AFM measurements were conducted in a NT-MDT NTEGRA prima equipment using silica-etched tips. Typical samples were investigated at scan rates of 1-3 Hz with 2566256 pixel resolution during image capturing. Samples were dropped onto a mica sheet. Size determination was performed individually recording the z-axis height of each particle and subtracting the surrounding z-axis mica sheet height; iv)Elemental analysis by inductive coupled plasma atomic emission spectrometry: determining and quantifying Cd and Te metal species in both cell pellets and purified NPs was carried out using Spectro CIROS Vision ICP-AES. Strains pCA24NgshA and pCA24NgshB were grown as described and treated with CdCl 2 , K 2 TeO 3 or both for 24 h at 37uC.
After centrifuging at 10,000 g for 5 min, samples were washed twice with sterile Millipure water, heated and taken to dryness with concentrated HNO 3 and then dissolved in 10 mL of 10% nitric acid. Cd and Te content was determined using calibration curves specifically constructed for each element and normalized to the dry bacterial mass. The Te analytical line was 214.281 nm and for Cd, 228.802 nm; v) X-ray diffraction analysis: Power X-ray diffraction (XRD) measurements were performed using a Diffractometer Bruker D8 Advance using Cu Ka radiation. For XRD characterization, an aqueous QDs solution was diluted with 1 volume of ethanol and centrifuged at 4,000 rpm for 10 min. The precipitate was collected, dropped onto a glass slide and dried at room temperature. vi) IR studies: Infrared transmission spectra of purified NPs were recorded in KBr pellets from 2000 to 400 cm 21 on a Nicolet Impact 410 IR spectrophotometer.

Statistical Analysis
Statistical analysis was carried out using the GraphPad Prism v5.0 software. One-way ANOVA was used for establishing significant differences between more than two groups using a p,0.05. Dunnet post-test was used to compare differences between all the groups (GSH quantification) and Bonferroni posttest to compare different samples with the control (RSH determination). Figure S1 GshA and GshB induction kinetics. SDS-PAGE analysis of total proteins from E. coli AG1 and E. coli gshA after IPTG (0.5 mM) treatment for the indicated times. (TIF) Figure S2 AFM of biosynthesized CdTe QDs. Biologically synthesized CdTe nanoparticles were purified as described in Methods and the nanometric size was evaluated by AFM. (TIF) Figure S3 Effect of pH on nanoparticle biosynthesis. E. coli gshA were grown to stationary phase and suspended in water or phosphate buffer adjusted to different pH values. Cells were exposed to both CdCl 2