ATTOSTAT (NLC Laboratories, Salt Lake City, UT) was used as the nano-Ag source, with NP of a reported size 10 nm and a concentration of 30 mg Ag/L. The bulk Ag source was from Alfa Aesar, Ward Hill, MA, with a reported particle size of 44,000 nm. Bulk and NP of CuO and ZnO were purchased from Sigma-Aldrich, St. Louis, MO. The reported "as manufactured" sizes were: nano-CuO, 33 nm; nano-ZnO, 50–70 nm; bulk CuO, 8000–9000 nm; and ZnO, less than 1000 nm. Exposure to ions was from solutions of CuCl₂, Zn(NO₃)₂ and AgNO₃. All solutions were prepared in distilled, sterile water.

The biosensor was constructed in strain P. putida KT2440 to harbor a plasmid with a luxAB fusion to a Cu-responsive promoter [Pettee et al., unpublished]. Oligonucleotide primers were designed to amplify approximately 500 bps 5' to 100 bps 3' downstream of the translational start site at locus PP_0588 in wild type P. putida KT2440. The primers were: For, CGATGCGGTATTTGTTGATCT and Rev, AATCGCAGTGAGGATCTGCT. PCR products containing the PP_0588 promoter region were ligated to the promoterless luxAB::npt cassette in plasmid pCR2.1 5' bearing resistance genes for kanamycin and ampicillin (Invitrogen.com) in E. coli. Determination of the promoter orientation in the clones was achieved by PCR analysis using a primer to the 5' end of the luxA gene in the reverse orientation and identifying PCR products when used with the 5' promoter primer of PP_0588, 5'-CGATGCGGTATTTGTTGATCT-3'. The luxA primer sequence was 5'-CAACCAAATTTTCCCCAAGA-3'. Positive clones were ultimately confirmed by the presence of Lux activity and ability to grow on kanamycin at 20 μg/ml. The PP_0588 lux fusion was removed from the pCR2.1 vector and inserted into the stable plasmid pCPP45, bearing a resistance gene for tetracycline, for triparental mating into P. putida KT2440.

The PP_0588 cells were stored in 15% glycerol at -80°C. Logarithmic phase cells were generated by reculturing from an overnight culture grown in minimal medium (MM) with shaking at 25°C to OD600_nm = 0.1. MM contained in 1 L: 10.5 g K₂HPO₄, 4.5 g KH₂PO₄, 0.5 g sodium citrate (2H₂O), 1.0 g (NH₄)₂ SO₄, 0.25 g MgSO₄.7H₂O, and 2.0 g sucrose. The culture (200 ml) was centrifuged at 10,000 g for 10 min and the cells were resuspended in 200 ml sterile distilled water and used immediately in the Lux assay. After dividing into 50 ml aliquots in 125 ml flasks, the suspensions were treated with NP, bulk material or ions at defined final concentrations or were left without treatment as a control. Initially the cells were treated with 0.1, 1 and 10 mg metal (M)/L to determine the sensitivity range. Subsequently doses were adjusted to determine the level at which toxicity was observed. Flasks were shaken at 200 rpm and 25°C during the study. At defined times, 200 μl of the suspensions were transferred in triplicate into well plates for Lux readings. The luciferase substrate, 1% decanal in ethanol, 10 μl, was added automatically in the L MAXII Luminometer (Molecular Devices Corporation, Sunnyvale CA). Light output was recorded with a 10 sec. exposure. Generally samples were assayed every 10 minutes up to 1 h. At each sampling time, the Lux activity from three aliquots of the cell suspension was measured. Each treatment was replicated in three or more separate studies.

Cells, after a 60 minute treatment with or without metal exposure, were assayed for culturability by dilution plating on salt-free Luria Broth (Difco, Sparks, MD) agar medium. Colonies were counted after 24 h incubation at 28°C and the colony forming units (Cfu)/ml determined.

An aqueous suspension of 10,000 mg Cu/L of nano-CuO or nano-ZnO in sterile distilled water was filtered sequentially through sterile filters with pore sizes of 450 and 200 nm (Whatman Inc., Florham Park, NJ, USA). The filtrates were collected and diluted 5, 10, 100 or 1000-fold into cultures of KT2440 to determine effect on light output as described above. After 60 minutes of exposure cells were plated to determine culturability.

Suspensions of "as manufactured" nano-CuO and ZnO were fractionated according to size using asymmetric FlFFF (AF4) (Postnova Analytics, Landsberg, Germany). The operational procedure for FlFFF followed published procedures 383940. In FlFFF, carrier fluid (and introduced sample) flowed down the length of a channel bounded along its length by a membrane. NP size separation occurred in the presence of a cross-flow field perpendicular to the flow axis, in which the particles migrated differentially across the channel and aligned themselves within different streamlines in the laminar parabolic flow field, resulting in different-sized particles being carried toward the channel exit at different velocities. Under the FlFFF conditions used as described in Table 1, the elution time of a NP was proportional to its size; two experimental conditions were used to best tune the instrument to particle size of interest. Operating condition I (Table 1) was used for fractionation in the size range between 10 to 250 nm, and was calibrated using colloidal gold and fluorescent latex beads with known sizes (10, 98 and 200 nm, respectively). Operating condition II (Table 1) was used for fractionating particles smaller than 10 nm, and was calibrated using colloidal silver and colloidal gold with known sizes (5 nm and 10 nm). NPs with high diffusivity (small size) relative to the applied cross flow were eluted in the so-called "void peak". Particles with high fluid drag (large size) relative to diffusivity were held against the membrane until elution in the so-called "rinse peak" upon relaxation of the cross flow.

The dimensions of the asymmetric FlFFF channel used were 27.3 cm in length, 224 μm in thickness. The channel volume was 0.71 ml, calculated according to described methods 41. The membrane used was made from regenerated cellulose (10 K Dalton pore size). Milli-Q water (Millipore System) with 2% v/v FL-70 was used as carrier. The pH of the carrier solution was 8.93. Three detectors (UV absorbance, fluorescence and inductively coupled plasma mass spectrometry [ICP-MS]) were used in series downstream of the FlFFF to analyze the fractionated sample. FlFFF coupled with ICP-MS allowed online simultaneous determination of particle size distribution and elemental distribution of the nanomaterials 40. The ICP-MS Agilent 7500 ce (Agilent Technologies Inc., Santa Claram, CA, USA) was used in continuous mode. The general operating parameters are summarized in Table 2.

As shown in Fig. 1, nano-Ag was toxic to the biosensor. Treatments above 0.2 mg Ag/L caused immediate loss in light output (Fig. 1A). When treated with bulk Ag, no loss in Lux output was observed even for concentrations of 10 mg Ag/L (Fig. 1B). Cells also exhibited loss in light output with treatment from Ag ions with 1 and 10 mg/L doses (Fig. 1C). Treatment with a range of lower doses showed that 0.2 Ag ion/L also caused rapid loss in light output (data not shown). Consequently, there was a sharp threshold between 0.1 and 0.2 mg/L for toxicity with the Ag ions.

The tables adjacent to the RLU graphs report the changes in culturability of the cells transferred to plating medium after 60 minutes of treatment compared with unchallenged controls. For treatment with nano-Ag and Ag ions, loss of Lux activity correlated with loss in culturability. No loss in Lux output or culturability was observed with exposure to bulk Ag. At 0.25 mg/L nano-Ag no culturable cells were obtained; with Ag ion a culturability threshold near 0.2 mg Ag ion/L (data not shown) was determined.

The biosensor also showed loss in Lux activity when treated with nano-CuO and Cu ions but not with bulk CuO. Fig. 2A demonstrates that treatment with 10 mg Cu/L from the nano-CuO caused a time-dependent loss in light output whereas bulk CuO was inactive (Fig. 2B). Treatment with 1.0 and 0.1 mg Cu/L nano-CuO caused no effect (Fig. 2A). Ten mg Cu/L nano-CuO rapidly reduced RLU, and a toxicity threshold showing rapid RLU reduction was observed between 5 mg and 7 mg Cu/L (data not shown). Toxicity of the Cu ion, from CuCl₂, towards the biosensor was apparent at 1.0 mg Cu/L with 0.1 mg/L having little effect (Fig. 2C). A rapid RLU reduction was observed for 0.5 mg/L (data not shown). Thus, nano-CuO was about ten-fold less active than the free ions for the biosensor response. To confirm that toxicity with the Cu²⁺ was due to the metal ions rather than the Cl^-, the biosensor was exposed to Cl^- from NaCl. At a dose level where Cl^- was at the same concentration as that from CuCl₂ when Cu was present at 10 mg/L there was no observed toxicity (data not shown).

Loss in Lux correlated with loss in culturability upon exposure to Cu ions and reduced culturability upon exposure to nano-CuO at 10 mg/L treatment. Cell culturability did not decline with bulk CuO exposure.

Fig. 3A illustrates that treatment with 7 and 10 mg Zn/L of nano-ZnO rapidly eliminated light output from the biosensor. Interestingly, lower nano-ZnO concentrations of 1.0 (Fig. 3A) and 5.0 (data not shown) increased light output above the control. With bulk ZnO no toxicity was observed up to 10 mg Zn/L, rather increased light output was observed (Fig. 3B). Treatment with 10 (not shown) and 1 mg Zn/L as the free ion (Fig. 3C) caused rapid loss in light output. The threshold value for reduction of Lux activity of the free ion was between 0.1 and 0.5 mg/L. As with the nano- and bulk ZnO, increased light output also was observed for the free ion, with activating concentrations observed between 0.1 and 0.05 mg/L of ionic Zn.

None of the treatments with Zn caused complete loss in culturability, rather they were bacteriostatic. Cells grew from the Zn-exposed samples at a delayed rate. Whereas colonies from the control cells could be counted in 2 days, those from cells exposed to zinc required at least 5 days. The culturability data shown in Fig. 3 are after 1 week of growth.

Fig. 4A demonstrates that co-treatment of the biosensor with 0.1 mg nano-Ag/L plus 1.0 mg/L of nano-CuO caused a time-dependent loss in light output whereas the individual NP treatments had no effect. The adjacent culturability data are after 1 week of growth on the plate medium – the longer time was required as the combined treatment caused bacteriostasis. Cell culturability declined but was not reduced to zero upon exposure to this NP mixture. NP interaction promoting loss in Lux output was not observed in cotreatments with 1.0 mg/L nano-CuO plus 1.0 mg/L nano-ZnO or 0.1 mg nano-Ag/L plus 1.0 mg/L nano-ZnO (Fig. 4B and 4C).

FlFFF of the CuO and ZnO NP filtrates obtained using filters with a 450 nm cut off showed the presence of structures smaller than 5 nm (Fig. 5A and 5B). The material remaining on the filter surface for both nano-CuO and -ZnO (Fig 6A and 6B) contained poly-dispersed particulates in a broad size range from 70 nm to larger than 300 nm, with the bulk being larger than 300 nm, as demonstrated by the high ICP-MS counts during the rinse peak.

Treatment of the biosensor cells with the filtrates passing through both 450 nm and 200 nm filters from a suspension of nano-CuO at 10,000 mg Cu/L caused dose dependent loss in light output. No toxicity was observed with a 100-fold dilution of the filtrate but no light was emitted when cells were treated with the × 10 diluted filtrate. The loss in Lux activity with the 10-fold diluted filtrate correlated with loss in culturability. Similarly, treatment with the filtrate prepared from 10,000 mg Zn/L from nano-ZnO at × 5 and × 10, but not × 100 dilution, caused partial loss in light output. In contrast to the filtrate from the nano-CuO, no change in culturability was observed for nano-ZnO (data not shown).