A specific reporter has been constructed in which the arsenic-binding regulatory protein gene, arsR originated from Escherichia coli K12 DNA, and the cadmium-binding regulatory protein gene, cadC from Staphylococcus aureus NCTC50581 plasmid pI258 have been fused to the structural gene for green fluorescent protein. A gene encoding green fluorescent protein from the marine species Aequorea coerulescens (excitation maximum: 475 nm; emission maximum: 505 nm) was excised from pAcGFP1 (Takara, Shiga, Japan), and ligated to an expression vector, pET-3a (Novagen-Merck, Darmstadt, Germany), to construct expression vectors of genes encoding ArsR-GFP or CadC-GFP, as described previously [13]. Cells of Escherichia coli BL21 (DE3) pLysS were transformed with the expression vectors.

Cells of recombinant E. coli were grown in a Sakaguchi flask containing 300 mL of autoclaved Luria-Bertani (LB) medium supplemented with 50 μg/mL ampicillin and 34 μg/mL chloramphenicol at 25 °C for 24 h in an orbital shaker at 140 rpm. The cells were harvested from a culture containing 2 × 10⁹ cells/mL by centrifugation at 4 °C, and washed twice with 50 mM Tris-HCl buffer pH 7.4. The cells were resuspended into 4 mL TG buffer (50 mM Tris-HCl pH 7.4, 15% (v/v) glycerol) and frozen at −80 °C at least 1 h. After thawing, cells in a glass vial placed on ice were disrupted for 5 min with an ultrasonic disruptor equipped with a microtip probe (UD-201, Tomy, Tokyo). Sonication was repeated four times with 5-min interval. After centrifugation at 16,000 ×g for 15 min at 4 °C to remove cell debris, the lysate was divided into small aliquots and stored at −80 °C. The approximate concentration of GFP-tagged trans factor in cell lysate was calculated from fluorescence intensity [13].

Oligonucleotide sequences for P_ars-O_ars-50 [13] and DNA fragments containing O_ars designed in this study were shown in Figure 3(a). The sequence for P_cad-O_cad-50 was indicated previously [13]. Twenty-five picomoles per 100 μL double-stranded DNA fragments modified with biotin at the 5′ or 3′ end in 25 mM Tris-HCl buffer (pH 7.4) were poured and immobilized onto Reacti-bind streptavidin-coated high binding capacity black 96-well microplate wells (Thermo Fisher Scientific, Yokohama, Japan) as described previously [13]. After immobilization, excess unbound DNA was rinsed off three times by 25 mM Tris-HCl buffer pH 7.4.

A solid surface was prepared by contacting ArsR-GFP to O_ars-containing immobilized oligonucleotides. A mixture containing ArsR-GFP was prepared at final concentrations of 50 mM potassium phosphate buffer pH 7.4 (KPB), 50 μg/mL salmon sperm DNA, 40 mM NaCl and approximately 20 μg/mL ArsR-GFP. On the other hand, another solid surface was prepared by contacting CadC-GFP to immobilized P_cad-O_cad-50. A mixture containing CadC-GFP was prepared with a similar composition as that described in case of ArsR-GFP, except that 50 mM Tris-HCl buffer pH 7.4, instead of 50 mM KPB, was used. One hundred microliters of ArsR-GFP or CadC-GFP mixture were poured to each well in which oligonucleotide was immobilized, and incubated for 15 min. Free proteins were once washed off with 200 μL KP-T buffer (10 mM potassium phosphate buffer pH 6.0, 0.05% (w/v) Tween20).

As(III) and Cd(II) solutions were prepared by dissolving 98% NaAsO₂ and CdCl₂·2.5H₂O (both from Sigma-Aldrich) in ultrapure water (Simplicity UV, Milipore-Japan, Tokyo), bottled natural mineral water, or tap water collected at Utsunomiya University. For the As(III) assay, 93.5 volumes of sample was mixed with 5 volumes of 1 M KPB pH 6.7, 0.5 volumes of 10 mg/mL salmon sperm DNA and 1 volume of 4 M NaCl (final concentrations; 50 mM KPB, 50 μg/mL salmon sperm DNA, 40 mM NaCl). For the Cd(II) assay, the sample was mixed with a solution of similar composition with the exception that 1 M Tris-HCl pH 7.9 was used instead of 1 M KPB. Ninety microliters of sample mixture were added to each well in which either of the solid surfaces was prepared, and incubated for 30 min in As(III) assay and for 15 min in Cd(II) assay, respectively, with orbital shaking at 120 rpm to release the GFP-tagged trans factor binding to metal from immobilized cis element. The supernatants were transferred to wells in another black plate when fluorescence intensity was measured with a microplate fluororeader at excitation/emission wavelengths of 490/530 nm (MTP-601, Hitachi High Technologies, Tokyo). The supernatants were transferred to glass vials when fluorescence intensity was measured with a handheld, battery-powered portable fluorometer (GFP-pen GFP 100, Photon Systems Instruments, Brno, Czech Republic). Student’s t-test was used to evaluate probability between two groups including data obtained with ultrapure water.

For lyophilization, solid surfaces on wells after removal of washing buffer were frozen at −80 °C for at least an hour. Then, the solid surfaces were lyophilized with a freeze dryer (VD-500F, Taitec, Saitama, Japan) and stored in a refrigerator for 24 h. The lyophilized solid surfaces were evaluated by directly loading the sample mixtures.

A solid phase biosensor constructed with P_ars-O_ars-50 and ArsR-GFP did not show a marked increase in the fluorescence intensity obtained with 100 μg/L As(III) (Figure 3(b)). Therefore, other solid phase biosensors were constructed using the oligonucleotides containing the different regions of O_ars and their responses to 100 μg/L As(III) were evaluated with a fluororeader. The biosensor comprising ArsR-GFP and O_ars-30-down showed the highest fluorescence intensity, and the difference between the fluorescence intensities obtained with and without As(III) was more marked, compared to the differences in the biosensors comprising O_ars-40 or P_ars-O_ars-50 (Figure 3(b)). On the other hand, only background levels of fluorescence were detected in the biosensor comprising O_ars-30-up, regardless of the presence or absence of As(III). From the result, higher capacity of the O_ars-30-down-immobilized solid surface for ArsR-GFP can be expected. Therefore, O_ars-30-down was selected for the following experiments.

Significant increases of fluorescence in response to 100 μg/L As(III) were observed with both fluororeader (p < 0.001) and fluorometer (p < 0.01) (data not shown). Therefore, a dose-dependent relationship between As(III) concentration and fluorescence intensity was investigated with the fluorometer.. A dose-dependent increase of fluorescence was observed and a detection limit of 5 μg/L for As(III) was achieved (Figure 4(a)). The biosensor comprising ArsR-GFP and O_ars-30-down also responded in the detection of As(III) in drinking water. The responses of the solid phase biosensor to As(III) were analyzed using tap water and two types of bottled natural mineral water. A method detection limit of 10 μg/L As(III) was obtained with fluorometer (p < 0.01) when the fluorescence intensities were compared with the average intensity of all water without addition of As(III) (Figure 4(c)). Significant differences were also observed at 10 μg/L based on comparison within individual water. Thus, the optimized solid phase biosensor could offer a simple protocol for on-site screening of As(III)-containing drinking water

It was examined in the solid phase biosensor whether dissociation of CadC-GFP from P_cad-O_cad-50 is enhanced by Cd(II) within 15 min or not. Sample mixtures containing different concentrations of Cd(II) were poured into the wells, in which the solid phase biosensor was constructed. The fluorescence intensities measured with fluorometer increased with the increases in Cd(II) concentration up to 50 μg/L (Figure 4(b)). A detection limit of 1 μg/L Cd(II) was obtained using ultrapure water. The significant responses of CadC-GFP to Cd(II) in tap water and mineral water brand A were 5 μg/L (p < 0.01) and 1 μg/L (p < 0.001) with comparison to the average of all kinds of water tested (Figure 4(d)). A significance difference was found at 10 μg/L (p < 0.05) using mineral water brand B. Significant differences were observed at 5 μg/L based on comparison within individual water. Therefore, the solid phase biosensor is also available for monitoring of Cd(II) in drinking water.

Kawakami et al. [13] reported a detection limit of 1 μg/L for Cd(II) in the previous assay procedure, which is composed of 15-min pre-incubation of cell lysates containing CadC-GFP with samples, 15-min incubation of the mixtures in wells where P_cad-O_cad-50 was immobilized, and measurement of fluorescence, as shown in Figure 1(a). Therefore, it would be possible to conclude that the solid phase biosensor reduces the assay time and lessens the steps of assay procedure in the previously reported CadC-GFP biosensor while keeping the sensitivity to Cd(II).

The solid phase biosensors were lyophilized to enable preservation in the refrigerator. A microplate modified with lyophilized ArsR-GFP binding to O_ars-30-down or lyophilized CadC-GFP binding to P_cad-O_cad-50 was used to evaluate fluorescence intensities obtained with different concentrations of As(III) or Cd(II), respectively. Detection limits for As(III) and Cd(II) were 5 μg/L with a fluorometer (Figure 5(a,b)). Therefore, the lyophilized solid phase biosensors were preserved stably at 4 °C without lyoprotectants. The dose-dependent increase of fluorescence by the lyophilized biosensors was more marked in Cd(II) than in As(III). Lyophilization increases shelf life, flexibility and stability of the fluorescence protein [14,15]. Therefore, lyophilization is required to produce the stable solid phase biosensors available for practical use.

Escherichia coli ArsR [11] and Staphylococcus aureus CadC [16] lose binding capability to O_ars or O_cad in their metal-binding conformations, respectively [11,16–19]. It is worth to note that in the previously developed biosensors, the association process between trans factor and cis element was evaluated [13] whereas, in the solid phase biosensors, their dissociation process was evaluated. Another important fact demonstrated in this study is that the binding capacity and association/dissociation ratio of ArsR-GFP in response to As(III) are markedly affected by the nucleotide sequences chosen from the promoter–operator region. The response of the solid phase biosensor to As(III) was successively improved by choosing O_ars-30-down.

This study showed that the newly developed solid phase biosensors respond to As(III) and Cd(II). However, besides the specific toxic metals, the biosensors must respond to Sb(III), Pb(II) and Zn(II) because the specificities of ArsR-GFP to Sb(III) and CadC-GFP to Pb(II), Zn(II) and Sb(III) have been demonstrated [13]. The generally recognized advantages of metal-monitoring biosensors over traditional methods such as AAS and ICP are their capability of on-site measurement, low cost and easy manipulation. In this study, solid phase biosensors with easy manipulation were successively developed while maintaining the low detection limits (<10 μg/L) of the previously developed biosensors [13] in comparison with the detection limits of flame AAS (∼1 mg/L) [20,21], ICP-Atomic Emission Spectroscopy (30 μg/L) [22].

When the solid phase biosensors are compared with biosensors that utilize transcriptional switches to respond to analytes and to trigger signal transduction, their superior features can be highlighted. The solid phase biosensors elicit a significant response within 30 min for As(III) and 15 min for Cd(II), which are much shorter than the times required by whole-cell based biosensors whose responses usually take place within 2 to 3 h [22]. The biosensor based on in vitro reconstitution of the transcriptional switch has been reported [23]. In the assay, however, it takes 2 h to bind tetracycline-Renilla luciferase-tagged repressor protein (TetR-Rluc) on its repressor-operator site. In addition to this long assay time requirement, enzyme reactions of luciferase remaining on the wells is needed to obtain a luminescence signal, whereas, in the solid phase biosensors, the fluorescence signal is directly produced by GFP-tagged ArsR or CadC protein with an increase in toxic metal concentrations because dissociation from cis element becomes pronounced in the presence of As(III) or Cd(II). The capability of direct sample addition is another advantage of the solid phase biosensors. In comparison with the previously developed biosensors [13,23], the solid phase biosensors are advantageous in terms of the required assay time and protocol simplicity. Additionally, their ease of handling and storage are worth mention as the whole elements of biosensor can be preserved in a same package without lyoprotectants under normal refrigeration conditions, and can be rehydrated by the direct addition of sample.

Among environmentally found heterogeneous elements, the hardness of water may weaken the fluorescence intensity of CadC-GFP to Cd(II) because the tendency was more marked in very hard water brand B (203 mg/L as CaCO₃) than in soft water, tap water (56.1 mg/L as CaCO₃) and brand A (42.9 mg/L as CaCO₃). Contrary to this tendency, higher fluorescence responses of ArsR-GFP to As(III) were found in mineral water brand B. The opposite effects might be caused due to different behavior of the trans factors (CadC and ArsR) in very hard water and/or different reactivities of the buffers (Tris-HCl and KPB) to minerals. Therefore, in very hard water, dissociation of CadC-GFP from the solid phase might be partially hampered, resulting in the lower fluorescence values. In case of CadC-GFP, by taking an adequate control adjusted to the hardness of water, the solid phase biosensors can be provided as a common tool for monitoring of Cd(II) in drinking water. Thus, it was demonstrated that the developed solid phase biosensors offer a simple operating procedure and have the portability, stability and sensitivity which are necessary for monitoring of toxic metals in drinking water.