RESULTS

Introduction

With SPYKE our goal is to develop a bioelectric sensor for the detection of the rape drug GHB. With our device, we try to protect the user from consuming their spiked drink. Upon GHB presence, the sensor turns on a strong light for the user to know that the drink has been spiked. The light is switched by a capacitance change from the electrode. This change comes from a transcription factor (BlcR)-DNA dissociation caused by GHB presence in the sample. More information about the biosensor design and mechanisms of action can be found here .

The goal of our Wet Lab experiments was to 1) test the binding and dissociation efficiency of WT BlcR to its cognate operator sequence, 2) engineer BlcR protein and/or its DNA binding sequence to obtain better protein-DNA binding and dissociation upon GHB presence, and 3) efficiently convert the protein association and dissociation into a reliable electrical signal. To achieve these three goals, we divided our project into four different modules:

• Module 1: we aim to produce, purify and characterize BlcR.
• Module 2: we try to engineer BlcR to get a better binding affinity to its DNA sequence.
• Module 3: we aim to modify the blc operator sequence to get a higher binding affinity of BlcR to its binding sequence.
• Module 4: we aim to optimize the immobilization technique of DNA to the gold surface of the electrode and convert the dissociation of BlcR to an electric output.

TU Delft is not in possession of an Opium act license therefore it was not possible to test with the opium drug GHB. We then decided to perform most of our Wet Lab experiments with a GHB analog: succinic semialdehyde (SSA). From literature, we know that binding of SSA to BlcR also causes dissociation of BlcR from the blc operator sequence [1][2] . Nonetheless, we found our way to validate the efficacy of our biosensor with GHB. We made an appointment with the Trimbos institute, a drug and addiction center in the Netherlands [3] where we were able to test the dissociation of BlcR after the addition of GHB.

Module 1: Production, Purification and Characterization of WT BlcR

In Module 1, we focused on optimization of the production and purification of BlcR. BlcR was utilized in all the experiments over the different modules therefore highly concentrated and pure BlcR samples were needed. Read in our engineering cycle about the protein purification to explore how we optimized the protein production and purification cycle.

EMSA study for the characterization of BlcR

To characterize the binding of BlcR to its cognate DNA binding sequence (blc operator), we used electrophoresis mobility assay (EMSA). The concentration of the DNA was maintained constant while the concentration of BlcR was increased gradually from 0 to 1 µM (Figure 1).

We successfully validated binding of BlcR to the blc operator sequence. From the results of EMSA we determined the binding affinity of BlcR to the blc operator sequence, and established the degree of cooperative binding, expressed in the hill coefficient. We found a binding affinity of 390 nM, together with a hill coefficient of 1.79. These values were not far from what is currently reported in literature. Earlier research of Pan et al. reported a binding affinity of 490 nM [1] . The 1.79 hill coefficient was further on used in modeling experiments. Read more about how we calculated the Kd and the hill coefficient in our model section.

EMSA study of BlcR with SSA

With our next experiment we wanted to characterize the dissociation of BlcR from the blc operator in presence of SSA. We performed an EMSA study where we kept the concentration of BlcR and DNA constant at 1.6 µM and 25 nM respectively, and titrated the concentration of SSA in a range from 0 to 1 mM (Figure 2). With this study we successfully established the full dissociation of BlcR in presence of > 40 µM SSA.

We are creating a bioelectric sensor for the detection of GHB in drinks, aiming to fastly spot a spiked drink during the nightlife. Since many drinks contain alcohol, we wanted to make sure that association of BlcR to the blc operator is still possible when alcohol is present.

EMSA study of BlcR with ethanol

In our next EMSA study we looked at the association of BlcR in the presence of different ethanol concentrations. We found that BlcR still binds to the blc operator sequence in presence of ethanol concentrations of up to 20 % (Figure 3).

Module 2: Engineering of BlcR

To engineer BlcR, we used a rational design approach. Based on reported biochemical and structural data [1][2] , we created a mutant library with amino acid substitutions that could possibly result in higher binding affinity of BlcR to the blc operator sequence [TU Delft, section library ]. We successfully created 20 mutants of BlcR.

To screen the mutants, we utilized an already existing superfolded GFP expressing plasmid that was created by the Bielefeld iGEM team 2015: BBa_K4361115. This plasmid contains the BlcR binding sequence upstream the CDS of the superfolded GFP. Upon BlcR binding, GFP expression is repressed. Our working hypothesis for mutant screening was: the lower the fluorescence signal, the higher the percentage of BlcR bound to its operator site, the higher the binding affinity of BlcR to the blc operator.

Since BlcR WT can be actively produced in E.coli, we decided to utilize an E. coli based cell-free system to screen all 20 mutants. With this strategy, we aimed to have a robust screening platform that could bypass the conventional - often cumbersome - protein production and purification of our mutant protein candidates. We used the PURE (Protein synthesis Using Recombinant Elements) system, here PUREfrex2.0, [4] .

As a first step to validate our screening strategy, we checked the repression of GFP expression with our purified WT BlcR protein. We prepared a cell-free reaction with 2.8 nM of GFP plasmid supplemented with 0.725 μM of purified BlcR protein. The reaction was then incubated for 6 hours at 37°C.

Furthermore, we verified the specificity of BlcR to the DNA binding sequence. From literature, we found that one single nucleotide deletion in the DNA binding cassette impairs BlcR DNA-binding (Figure 4) [1][2] . We built a new GFP plasmid (GFPdel plasmid BBa_K4361117.) harboring the corresponding nucleotide deletion. We conducted another cell-free reaction but this time with 2.8 nM GFPdel plasmid supplemented with 0.725 μM of purified BlcR protein.

End-point measurements of GFP fluorescence showed that BlcR leads to a 55% decrease of GFP signal with the reporter plasmid containing the correct blc operator sequence (Figure 5a). This result suggests that BlcR can bind to the blc operator and partially represses GFP expression. In contrast, the use of GFPdel plasmid did not result in a decrease of fluorescence, demonstrating that BlcR-induced transcription inhibition is specific to the presence of the cognate blc operator site (Figure 5b).

From the EMSA results, we saw that BlcR dissociates from the operator sequence in presence of SSA (Figure 2). We then decided to perform a dissociation experiment in the cell-free PURE system by analyzing the end-point fluorescence of the blc-containing GFP reporter plasmid in the presence or absence of SSA.

While GFP intensity consistently decreased (about 63% lower) when purified BlcR was added to the reaction, no significant increase (only 12%) was observed upon addition of SSA (Figure 6a). We had expected to recover a similar GFP intensity as in the condition without BlcR. We performed another experiment with a higher concentration of SSA (7 mM vs 0.7 mM), but this did not result in an increased GFP signal (Figure 6b).

We hypothesized that SSA could be directly interfering with the expression of GFP in PURE reactions. However, we performed GFP expression experiments with varying SSA concentrations and did not see a strong inhibition. We then decided to continue with this cell-free PURE expression strategy to screen for BlcR-binding DNA variants and resorted to other experimental approaches for assaying SSA-mediated dissociation of BlcR-DNA complexes.

Prior to screening our library of BlcR-binding DNA sequences, we decided to validate the production of WT BlcR in PURE reactions. To this end, we expressed 4 nM plasmid encoding for BlcR and supplemented PURE with GreenLys, a reagent used for co-translational fluorescent labeling of the synthesized products. The reaction solution was analyzed by SDS PAGE and the fluorescently labeled proteins were visualized with the Typhoon instrument. Compared to the control reaction that did not contain DNA, two protein bands corresponding to the expected size of the BlcR dimer (~70 kDa) and the BlcR monomer (~35 kDa) could be detected (Figure 7). These results demonstrate that BlcR can efficiently be expressed in the PURE system, where it can oligomerize in dimers.

In light of these results with the WT BlcR we decided to investigate if our designed BlcR mutants could be expressed in the PURE system. All 20 plasmids coding for different mutants were individually expressed in PURE reactions supplemented with GreenLys. We found that not all the mutants could be produced in the PURE system. For the mutants D37R, A62V, L66V, and A67H, no protein bands corresponding to the monomeric BlcR were visible. Expression of all the other mutants led to a protein band that is not detectable in the negative control, corresponding to the monomer of BlcR. An upper (fainter) band that does not appear in the negative control and that may correspond to the dimer of BlcR can also be seen in most of these mutants. In conclusion, most of the selected mutants can effectively be produced in an E. coli based cell-free system (Figure 8).

We finally examined whether the cell-free expressed BlcR was able to repress transcription of the GFP reporter gene containing a blc site. Unfortunately, protein activity could not unambiguously be demonstrated, most probably because the buffer composition is not compatible for functional BlcR folding. Therefore, we decided to not use the PURE system as a platform to test the activity of the different BlcR mutants. For activity assays, we recommend to use purified BlcR in combination with the blc-containing GFP reporter plasmid.

Module 3: Engineering of the BlcR operator

To complement the protein engineering approach taken in module 2, this module focuses on the design of novel blc operator sequences to which BlcR binds. To improve the affinity between BlcR and DNA, we designed and screened variations of the known operator sequence and evaluated their ability to bind to BlcR. You can find an overview of all the different DNA sequences in Table 2. We used automated gel electrophoresis to determine the fraction of BlcR bound to the engineered oligos.

Moreover, we performed DNA binding measurements with isothermal titration calorimetry (ITC). However, we did not obtain conclusive results. ITC is highly sensitive to contaminants and we could not reach the required purity and protein concentration needed to avoid background noise.

To validate our screening method by automated gel electrophoresis, a control experiment was first conducted with samples of the WT operator sequence and a random DNA sequence, both with or without addition of BlcR. Here, we aimed to determine the positions at which the free DNA and DNA bound by BlcR migrate. As Figure 9a shows, DNA is significantly retained in the samples containing BlcR and appears as a band in the region of 250 base pairs, while free DNA appears clearly as a band in the region of around 100 base pairs.

Using this same method, a library of 20 operator sequence variants was screened to evaluate their affinity to BlcR. Read why we chose those operator sequences on the library page.

As shown in Figure 10, the DNA variants appear to have a wide range of affinities for BlcR. Some variants (04, 11, 12, 13, 14) show as little binding as the negative control, suggesting no specific affinity at all. Interestingly, the apparent affinity of other variants (07, 08, 09, 21, 22) matches or even exceeds the reference value of the wildtype sequence.

The 5 variants exhibiting high affinity were selected for a second round of screening, both for confirmation of the previous result and to evaluate the BlcR-DNA unbinding effect in the presence of SSA. Figure 11 shows that all of the selected variants have similarly increased binding to BlcR with a bound fraction of around 0.83 compared to the wildtype at 0.73. Moreover, the apparent binding affinity of each variant drops when SSA is present, as expected for a functional drug-responsive system. This is evidenced as a decrease in the fraction of DNA bound from an average value of 0.83 to 0.64. This drop is similar amongst the 5 variants, but slightly less than the drop from the wild type DNA sequence. These results suggest that all DNA variants have a higher binding affinity to BlcR compared to the wildtype sequence, without a significant loss of the SSA-dependent dissociation mechanism.

Since all 5 variants show a similar increase in affinity, we investigated what their common DNA features are and what the differences are compared to the sequences with low binding affinity. Figure 12 shows a multiple sequence alignment of the 5 selected variants with the highest apparent affinity (upper five) and 5 variants with the lowest apparent affinity (lower 5). Interestingly, we found that the IR1 region described by Pan et al. as the possible binding site of BlcR [1][2] , is not conserved among the high-affinity variants (Figure 11).

The most striking difference between high- and low-affinity variants is that an 8-bp sequence is almost entirely conserved in all high-affinity variants and lacking in almost all low-affinity variants. This suggests that this 8-bp sequence is essential for BlcR-DNA binding. Remarkably, the exact reverse complement of this sequence occurs at the other end of the site, forming a set of inverted repeats (pink outlines in Figure 13). Two 5-bp inverted repeats (blue outlines in Figure 13) can also be recognized in between the larger ones, although there is no suggestion that these are important for BlcR-DNA binding, as one of them is not conserved throughout the high-affinity variants.

This set of inverted repeats does not coincide with the inverted repeats reported in literature (Figure 12). Nonetheless, our partner, the DTU iGEM team, compared the analogous operator sequences from different strains of A. tumefaciens, and reported the existence of these other sets of inverted repeats. More of it can be found on our partnership page. If we highlight these inverted repeats on a model of the wildtype operator, it shows that the regions are not only symmetrical in sequence, but also in structure (see Figure 14). This might suggest that BlcR binds to this site through symmetrical tetramerization.

From screening different operator sequences we found that five operator sequences (7, 8, 9, 21, 22) bind BlcR stronger than the original WT sequence. We also found that the similarities between those sequences are in a set of inverted repeats found by the DTU iGEM team, suggesting a different BlcR binding site than previously described in the literature [1][2] .

Module 4: Immobilization and Electrical measurements

The first goal of module 4 was to verify the binding of the thiol-modified BlcR DNA binding sequence on the gold surface of our electrode. We used the engineering cycle to establish the best immobilization conditions. With Atomic Force Microscopy (AFM), surfaces can be analyzed and displayed in a height distribution pattern. We used AFM to confirm the immobilization of DNA and BlcR on a gold surface. In addition, we conducted a DNA-BlcR complex dissociation experiment to see the difference in height distribution after adding SSA. To reach these goals, we first used AFM to scan an empty gold plate (Figure 15a), a gold plate with immobilized blc operator (Figure 15b), a gold plate with immobilized blc operator and added BlcR (Figure 15c), and a gold plate with immobilized blc operator, added BlcR plus SSA, the GHB analog (Figure 15d). To make the pictures comparable across the different conditions, all the height distribution patterns of Figure 15 are scaled from 0 to 35 nm.

By calculating the number of peaks in the images (using ImageJ) for all resulting AFM scans (Table 3), we confirmed that 1) the DNA could evenly bind to the gold plate surface (Figure 14b), 2) BlcR is able to bind an already immobilized cognate binding sequence and 3) SSA addition causes some BlcR to unbind the immobilized DNA.

Table 3. The amount of peaks in the height distribution patterns visualized with AFM after image analysis.

	Empty (dots)	DNA (dots)	BlcR (dots)	SSA (dots)
With smoothing	x	9	366	197

With AFM we successfully confirmed the immobilization of thiol-modified blc operator onto a gold surface. Moreover, we showed that the already immobilized DNA is accessible for BlcR binding and unbinding. Our next step was to convert BlcR-DNA complex dissociation events into a capacitance change. More information about this can be found in our design page. To test whether this was possible, we immobilized the blc operator sequence (electrode A), and a scrambled DNA sequence (electrode B) to the gold surface of an interdigitated electrode. We measured the capacitance change with vector impedance at 1 kHz after addition of BlcR and SSA.

In Figure 16, one can see that with electrode A, the capacitance decreases after addition of BlcR and increases again after incubation with SSA. This specific response is not visible with electrode B, where the capacitance stays in the same range after addition of BlcR and SSA. These results support our hypothesis: when BlcR is added it binds to the blc operator sequence replacing the water molecules between the electrodes. Upon the removal of the solvent, the permittivity decreases, which leads to a drop in the capacitance. When SSA is added, BlcR dissociates from the DNA and water molecules can diffuse back between the plates, which in turn increases the permittivity and results in a capacitance increase.

From vector impedance, we moved on to the universal transducer interface (UTI) to measure capacitance. More information about UTI can be found here. The results of the UTI and the vector impedance were comparable to our previous results: we saw a decrease in capacitance when BlcR is added and a subsequent increase after SSA addition. Experiments with scrambled DNA did not show capacitance changes after BlcR and SSA addition, as expected (Figure 17).

To make our device more accessible, we created an Arduino circuit to measure capacitance. All the information about our circuit can be found in the hardware page. With Arduino, we aimed to turn on a light when a difference of 1.6*10^-2 pF is measured; read our hardware section to understand why we chose this specific threshold. We used Arduino to measure the capacitance after incubating the electrode with 1 µM DNA (blc operator: electrode A, scrambled DNA: electrode B) and 4 µM of BlcR. We set the measured capacitance value as a baseline to measure the difference in capacitance following the addition of 150 µM SSA (Figure 18). After addition of SSA we measured a capacitance difference 5.4*10^-2 pF with electrode A and 0.0 pF with electrode B (Figure 18). The results confirm a possible change of >1.6*10^-2 pF in the presence 150 µM SSA.

Up until here, all experiments were performed with the GHB analog SSA. We managed to visit the Trimbos Institute, where we could perform capacitance measurement using GHB. For this assay, we decided to use a UTI device due to its accuracy: it can measure the capacitance differences with a standard deviation of 3*10^-4 [5] . We incubated different electrodes with 1 µM DNA; blc operator: electrode A, scrambled DNA; electrode B and 4 µM BlcR. We aimed to measure the capacitance difference after addition of GHB to a 55 mM final concentration (Figure 19).

Results showed an increase in capacitance of 2.5*10^-2 pF after GHB addition on an electrode with immobilized blc operator sequence (Figure 19a). This increase in capacitance is not shown when GHB was added to the electrode coated with the scrambled DNA (Figure 19c). With these final results, we successfully validated that our sensor can measure an > 1.6*10^-2 pF increase in capacitance when GHB is present at a concentration of 55 mM, which corresponds to the typical GHB concentration used to spike drinks [6] . This specific change in capacitance could turn on the light of our Arduino circuit. This light is our main biosensor output to warn people that their drink has been spiked.

Conclusion

The goal of our Wet Lab experiments was to 1) test the binding and dissociation efficiency of WT BlcR to its cognate operator sequence, 2) engineer BlcR protein and/or its DNA binding sequence to obtain improved protein-DNA binding and dissociation upon GHB presence, and 3) efficiently convert the protein association and dissociation into a reliable electrical signal.

With EMSA studies we successfully characterized the binding between BlcR and the blc operator sequence and established that protein-DNA interaction occurs in an environment up to 20% of ethanol. Twenty mutants of BlcR were created and production of 16 mutants in an E. coli-based cell-free system was verified. By screening the different operator sequences for BlcR binding, we found five oligos that bind BlcR stronger. After sequence analysis we discovered together with the DTU iGEM team another possible binding site for BlcR that has not been described in literature.

Our overall project's objective was to create a functional device that could accurately detect GHB in beverages by turning on a light. From electrical experiments we could conclude that the capacitance signal generated by the dissociation of BlcR to its DNA binding sequence is strong enough to turn on a light to warn the user.

Future Prospects

Engineering of BlcR

We found out that the PURE system is not the optimal environment to screen the cell-free expressed mutants on their activity. Future experiments could either focus on optimizing the PURE system for protein mutant screening, or on a purified-protein screening strategy. The mutants that could successfully be expressed in the PURE system can be produced in E. coli following the optimized protein production protocol we developed through different iterations of the engineering cycle . The activity of the mutants can then be determined electrically with Arduino or UTI measurements.

Engineering of the BlcR operator

From the results of screening of the different oligos and the sequence analysis performed by DTU we concluded that BlcR could possibly bind to different inverted repeats than the ones proposed by Pan et al [1]. The crystal structure of dimeric BlcR has been reported [1], but a crystal structure of tetrameric BlcR bound to the blc operator sequence is still missing. For future experiments we propose crystallizing BlcR bound to DNA to better visualize the tetrameric-DNA binding structure of BlcR. This information could help to optimize the bioelectrical sensor:

• When the position where BlcR binds on the DNA is known, amino acids involved in the interaction with DNA can be targeted for site-directed mutagenesis. Mutants can therefore be created more specifically to increase the binding affinity of BlcR to its DNA sequence.
• Once the binding site of the DNA is revealed, we could engineer longer oligos with multiple binding sites to implement in our sensor. From literature we know that increasing the amount of transcription factors bound to one oligo could improve the sensitivity of electrical biosensors [7].

Immobilization and Electrical measurements

From the EMSA results we learnt that BlcR is still active with ethanol concentrations up to 20% (Figure 3). When the bio-electrical system was tested with 10% ethanol during the UTI measurements we saw that the signal became very unstable ( Notebook 4 ). Because of this instability in the electrical output we were not able to detect an increase of capacitance after addition of GHB. Our sensor is based on drinks, often alcoholic drinks, therefore it is of great importance to make it work in an environment with alcohol up to 20%. Further experiments are needed to determine the cause of the system’s instability, and subsequently tackle the issue to ensure the correct functioning of our product.