V. fischeri strain NRRL-11177 (Dr Lange S.A.) was grown in 20 mL DSMZ No 6904 broth [4] for 20 h in an orbital incubator at 190 rpm at 24 °C [3]. The thus developed stock culture was mixed with 30% glycerin, dispensed in 2 mL vials and stored at −70 °C. Twenty μL of the stock culture were inoculated in 50 mL of growth medium and grown overnight. Then the culture was centrifuged and resuspended in artificial sea water, preserving their function but stopping their growth. This suspension was used with the FI system.

All chemicals used in the system were of analytical reagent grade purchased from Merck (Darmstadt, Germany). 0.0200 M Hg²⁺, Cu²⁺ and Pb²⁺ stock solutions were prepared by dissolving HgCl₂, Cu(SO₄)·5H₂O, Pb(NO₃)₂·H₂O. Acetate and phosphate buffer solutions were prepared for the pH ranges 3–6 and 7–11, respectively. pH was adjusted with NaOH and HCl.

Light detection device: Photomultipliers (PMTs) are the most sensitive devices for the detection of low level optical signals. An integrated PMT tube, incorporating the high voltage source and divider circuit along with an RS 232 signal output was selected (Hamamatsu HC-135 01) resulting in a compact biosensor device.

Flow cell: Due to the use of a PMT detector, a wall-jet flow cell configuration was implemented. Figure 1 depicts the design of the prototype based on a previous study by Divritsioti et al. [7]. The assembly was clamped between two stainless steel plates featuring appropriate openings for light detection and passage of the In (entrance of carrier solution) and W (exit of carrier solution to the wastes) polytrafluoroethylene (PTFE) tubes of 0.8 mm internal diameter.

Temperature control of the flow cell: To maintain immobilized bacteria at their optimal temperature, the flow cell is enclosed in a thermostated aluminium frame using a water bath at 20 °C.

The single line Flow Injection system that was developed, incorporating the above described detector unit, is depicted in Figure 3.

The analyzer design is based on the continuous flow of a carrier solution. The sample loop of the injection valve is loaded with the Vibrio fischeri cell suspension that is subsequently automatically injected in the carrier solution flow. Vibrio fischeri cells are mixed with the carrier solution in the mixing coil and then driven to the detector unit for bioluminescence assessment. % Inhibition of bioluminescence is assessed with a two step experimental protocol. The first step is the assessment of V. fischeri bioluminescence by injection in a non toxic carrier solution. The second step is the assessment of V. fischeri bioluminescence by injection in a toxic carrier solution (sample). %Bioluminescence inhibition is calculated using the readouts (peaks) from the two injections: %Bioluminescence inhibition = (Peak 1–Peak 2) × 100/Peak 1

The thus calculated bioluminescence inhibition correlates with the toxic effect of the sample to V. fischeri cells. For the evaluation of the peak readouts, both heights and areas have been assessed. V. fischeri wash out from the analyzer was a slow procedure, resulting in extensive peak tailing. This was probably due to the shape of the flow cell that does not facilitate the wash out of bacteria. To overcome this problem and shorten the analysis time, 2 s after recording the peak maximum, the analyzer pump was set to maximum flow rate (∼10 mL/min). In this way, the turnover time for a single peak was kept to just 40 s. Precision of peak height measurements assessed through 10 injections of the same V. fischeri culture was found to be 0.7% RSD. Peak area measurements resulted in lower precision presumably due to the high flow rate used for the wash out step.

Three different volumes of V. fischeri suspension were tried: 80, 100 and 200 μL. Signal increased with increasing the injected volume. Although the signal is maximized when injecting 200 μL of V. fischeri culture, we choose 100 μL in order to minimize consumption of the culture. In this way, 12 injections were feasible using the same culture suspension. Although just 100 μL are injected, another portion of around 400 μL is consumed while filling the sample loop of the injection valve. This is actually a general disadvantage of the procedure of filling a sample loop through aspiration. Although this disadvantage could be overcome, when using conventional reagents, by replacing aspiration through the use of a syringe, this is not an option when using bacterial suspensions as reagents. Bacterial membranes could be partialy disrupted resulting in loss of signal. When using 8 mL culture suspension 16 injections are possible. It should be noted that V. fischeri culture cannot be used the way reagents are used: upon standing, due to gravity, bacteria tend to accumulate in the bottom of the container. This results in a gradual signal decrease that was up to 30% for one hour operation and diminished precision. To overcome this problem, a small volume culture (8 mL) that was continuously stirred with a magnetic stirrer, was used.

Flow rate was optimized in the range of 0.5 to 2.0 mL/min. At flow rates lower than 1.5 mL/min peaks were not sharp and the whole procedure was slow resulting in high turnover time. Flow rates higher than 1.5 mL/min resulted in diminished precision (RSD’s up to 5%). The flow rate of 1.5 mL/min selected as a compromise between fast washout of V. fischeri cells from the analyzer and precision. Mixing coils of 50, 100 and 150 cm have been tested. The 100 cm mixing coil was selected as a compromise between adequate mixing and analysis time.

The software package developed in Labview language provides functions for data acquisition from the biosensor unit, control of the peristaltic pump and the injection valve, bubble detection and subsequent correction of FI peaks, quantitative measurements, drifting baseline correction and export of data to text files. The data acquisition and control program flowchart is depicted in Figure 4:

The developed Flow Injection system was used for the assessment of toxicity due to different heavy metals. A typical FI toxicity assessment output is shown in Figure 5, while the dose response curves for Cu²⁺, Pb²⁺ and Hg²⁺ are depicted in Figure 6. It is clear that the developed system is able to assess toxicity due to heavy metals present in water samples. Effective concentrations were in the range of 1.0 × 10⁻² M–1.0 × 10⁻⁵ M.

It should be noted that the low limit of the useful concentration range is not much lower than that presented in Figure 6. The work presented here is showing the principle of using cells as a “reagent” in flow systems in order to use the system in water analysis. Further work is needed to lower the detection limits. If needed detection limits can be modulated by stopping the flow – after stopping the flow for 15 minutes detection can go down to 10⁻⁷ M, depending on the toxic metal compound.

Hydrolysis of metal salts modulates the pH of aqueous solutions. To assure that the systems’ response was due to heavy metal content and not to pH changes during dissolution of metal salts, we checked the response to carrier solutions of different pHs in the range of 3–10 using as blank distilled water (pH = 5.50). Results depicted in Figure 7 show that V. fischeri’s bioluminescence increases along the pH range from pH 5 to pH 10. It must be noted that the pH of the suspension of V. fischeri cells of in the reconstitution solution (pH 5.21) was 6.22.

The pH range of 1.0 × 10⁻²–1.0 × 10⁻⁵ M Hg²⁺ was 3.75–4.47. For 1.0 × 10⁻²–1.0 × 10⁻⁴ M Cu²⁺ and Pb²⁺ pH was in the range of 4.24–4.93 and 4.23–5.00, respectively. According to Figure 7 the signal is this pH range is completely lost. At this point it should be noted that results presented in Figure 7 refer to solutions of phosphate and acetate buffers prepared as described in the reagents section. However, as shown in Figure 6, when using metal solutions inhibition is differentiated.

To measure dispersion we first recorded the signal (D_s) of the bioluminescent suspension using it as a carrier stream and then the signal from 100 μL the same suspension injected in a carrier stream of deionised water (D_o). The D_s signal divided by D_o gives the dispersion which in the developed FIA system was calculated and found to be equal to 4.52. That means that the pH in the flow cell is between 5.50 (pH of deionised water) and 6.22 (pH of the reconstituted suspension of V. fischeri cells).