The procedure performed for PRIA was based on his-tagged repressor proteins immobilized on Ni-NTA agarose. Herefore, we constructed a device for the expression of his-tagged LacI. After purification of the protein we immobilized it on Ni-NTA agarose in a reaction tube. Plasmid DNA containing the operator site for specific binding of the repressor was added to the Ni-NTA agarose, unbound plasmid was washed out and the remaining plasmid could be eluted by addition of the analyte to the wash buffer. The DNA amount eluted upon addition of the analyte depends strongly on the salt concentration in the buffer used for washing and elution (See figure 1). We analyzed the released DNA in the supernatant with agarose gel electrophoresis. The DNA amount eluted in the first elution step compared to the total DNA amount bound to the agarose after the forth washing step, is much higher (See figure 2). Another explanation for an increased DNA amount in the elution steps could have been dissociation of the protein and the bound DNA from the Ni-NTA agarose. The last step of the assay is elution of the protein from the Ni-NTA agarose with an imidazole buffer to confirm the presence of the protein at the end of the procedure. Our analysis of the samples via SDS-Page shows clearly, that there is no significant loss of protein during the assay and most of the protein can be recovered upon elution with imidazole (See figure 3). On a future test strip the immobilized DNA-protein complex would be provided. So the user just needs to add the potentially contaminated water plus sodium chloride and buffer solution which we would provide in the kit.

Furthermore, we tested many different conditions to further evaluate the robustness of PRIA. Tap water with the analyte alone could not disrupt the binding between the plasmid and the protein, the major part of plasmid remains bound to the protein. In the agarose gel DNA can be detected in the steps where the protein (and the DNA bound to it) is eluted and in the fraction with the rest of the Ni-NTA agarose, which also still contains protein-DNA complexes (See figure 4). Thus, the user would be required to add a certain amount of salt to the sample. In our assay we mainly applied pottasium chloride, but common sodium chloride is suitable for that purpose, as we tested (See figure 5). Nevertheless, the solution should be slightly buffered with TRIS-HCl or sodium/potassium phosphate.

To further reduce the time for the assay, we also tested whether the complex could be formed prior to addition of the complex to the Ni-NTA agarose. The results were similar to the performance with water instead of buffer. In this case the bond between repressor and plasmid could not be disrupted at all. In the agarose gel DNA was visible after the elution of the protein with imidazole that was performed to verify the presence of the protein at the end of the assay. Furthermore, it could still be detected on the Ni-NTA agarose (See figure 6).

We aimed at the development of a paper-based system and optimized the procedure for immobilized protein. So we focused on the approach which was based on repressors fused to a cellulose binding domain (BBa_1321340). We cloned and expressed fusion proteins for our LacI-lacO model system (BBa_K1758202), the repressor ArsR (BBa_K1758203) for the detection of arsenic and for the protein BlcR, the repressor of the blc-operon (BBa_K1758204), which we were working with in order to detect date rape drugs. When we added the proteins on various types of paper, their presence could be confirmed by staining the paper with Coomassie brilliant blue and destaining with destaining solution used for SDS-PAGEs. The binding was unspecific, since the spots with proteins containing the CBD and the spots containing proteins fused with sfGFP, which is not supposed to bind to paper specifically, still showed the same intensity of blue dye after overnight washing (See figure 7). Besides, most CBDs bind to microcrystalline cellulose or cotton (Lethiö et al.). We were not able to find a hint about the binding of CBDs to common paper. Additionally, we came across certain difficulties concerning the purification of these proteins. That is why we focused on the second approach with immobilized DNA.

Based on a method proposed by Araújo et al., DNA was immobilized on filter paper previously activated with p-phenylene-diisothiocyanate (PDITC). To accomplish this, amino-labeled DNA is required. We made some adaptations to the original protocol in order to immobilize dsDNA instead of ssDNA. We omitted a 30 min washing step with 4x SSC buffer, since this serves for the denaturing of DNA. Furthermore, we hybridized the amino-labeled operator strand with the complementary strand that was Cy3 labeled to be able to quantify the DNA immobilized on the paper. This was performed via a simple annealing of these two oligonucleotides. So by detecting the Cy3 label via the Typhoon fluorescence scanner, we could be sure that the immobilized DNA was the correct dsDNA.

We dissolved PDITC, the substance for activation of paper for DNA immobilization, in pure ethanol and DMSO. We compared which solvent is better for the activation. Herefore, we washed the immobilized DNA on paper, which was activated differently, and compared the remaining intensities of the Cy3 label on paper after washing. The immobilization works even better if ethanol is used as a solvent for the PDITC. The loss of signal can be mainly attributed to the washing out of DNA, since a control showed that the decrease of fluorescence of the Cy3 dye due to repeated scanning is minimal. We applied three different liquids for washing:

1. an antibody stripping buffer that is normally used to disrupt all protein-protein interactions in a western blot. This resulted in strong blurring of the DNA spots.
2. water
3. the binding buffer that was normally used in PRIA

Compared to the remaining fluorescence signal on the activated papers the signal on the unactivated paper decreases strongly (See figures 9 and 10). This indicated, that the immobilization of DNA on activated paper was successful. The fluorescence signals were quantified with ImageJ.