Results

The BLISS biosensor utilizes the bacterial two-component system VbrK/VbrR which originates from the gram-negative bacterium Vibrio parahaemolyticus. This system has been shown in literature to grant bacteria broad-spectrum resistance against beta-lactam antibiotics (Li et al. 2016). In our case, we wanted to use the sensing properties and functionalize them for a biosensor by replacing the endogenous expression cassette with reporter proteins.

Since Vibrio parahaemolyticus is a organism of the biosafety level 2, we decided to use the biosafety level 1 gram-negative bacterium Escherischia coli for our biosensor. Therefore, our first step was to design the codon-optimized coding sequences for gene expression in E. coli (Parts BBa_K4660000 - BBa_K4660005).

Initially, we planned to use Golden Gate cloning for the assembly of our biosensor (Figure 4).

Figure 4: Plasmid design for the Golden Gate assembly of the VbrK/VbrR biosensor. The upper panel shows the level 1 plasmids, the pOdd1 vector containing the transcriptional unit (TU) for VbrK, pOdd2 containing the TU for VbrR and pOdd3 containing the promotor region of VbrR with the β-galactosidase gene. The pOdd4-spacer plasmid also needed for assembly is not shown here.

But we ran into some problems with this cloning approach and had to eventually change our cloning set-up to build our biosensor.

Therefore we changed our cloning approach and used the fragment exchange (FX)-cloning method established by Geerstma et al. in 2011 to generate our protein constructs. FX-cloning allows the simple cloning of protein coding sequences into a variety of different protein expression vectors. In our case, we used the pBxC3H vector (Figure 5), which utilizes an L-arabinose inducible promotor (araBAD) for protein expression.

Figure 5: Plasmid map of the pBxC3H backbone we used for FX-cloning.

To implement our coding sequences into pBxC3H vector, we first had to make our basic parts compatible for FX-cloning using specially designed primer pairs (Table 1) and performing a Polymerase Chain Reaction (PCR) to amplify the FX-cloning compatible gene fragment. Afterwards, the PCR product was digested using the DpnI restriction enzyme, which specifically cleaves the template DNA used in the PCR reaction. To proceed with the restriction cloning more efficiently, the PCR product was purified using the NucleoSpin Gel and PCR Clean-up kit by Macherey-Nagel and applied to a 1% agarose gel and gel electrophoresis was performed (Figure 6).

Figure 6: DNA gel electrophoresis of the FX-cloning compatible PCR products in a 1% agarose gel. The ladder used is the GeneRuler 1 kb DNA Ladder by ThermoScientific.

In the agarose-gel it is clearly visible that the PCR amplification of the FX-cloning compatible coding sequences for VbrK (~1470 bp), VbrR (~ 700 bp), BlaR1 (~ 1800 bp) and BlaI (~400 bp). Some remains of the template plasmids that were used are also visible, but with noticably smaller intensitiy.

Lastly, 250 ng of the respective PCR fragments (”inserts”) were used for the restriction digest with SapI, together with 50 ng of the vector backbone pBxC3H. After ligation with the T4 DNA ligase, the reaction products were transformed in E. coli TOP10 cells to amplify and subsequently isolate the plasmid.

To verify the sequece of the coding sequence and check succesful incooperation of the insert into the pBxC3H backbone, we used the sequencing service of Microsynth Seqlab. Our samples were sequenced with the standard primer pBAD-for (3'-ATGCCATAGCATTTTTATCC-5'). The sequencing result confirmed the correct sequence of our plasmid constructs (an exemplary cutout of the sequencing result for pBxC3H-VbrR as well as the sequence alignment performed is shown in Figure 7 and 8).

Figure 7: Cutout of the sequencing chromatogram from pBxC3H-VbrR.

Figure 8: Cutout of the sequence alignment of the sequencing result of pBxC3H-VbrR.

In conclusion, we successfully cloned our designed coding sequence for VbrK and VbrR into the pBxC3H expression vector (Figure 9). Since we also wanted to try the two-component system BlaR1/BlaI from Staphylococcus aureus for a biosensor, we cloned these coding sequences into the pBxC3H vector as well.

Figure 9: Finished plasmid map for protein production and purification: A. pBxC3H-VbrK. B. pBxC3H-VbrR.

After cloning the sequences of VbrK, VbrR, BlaR1 and BlaI into pBxC3H, succesfully transformed E. coli BL21 colonies were picked for each of the protein constructs to innoculate a 50 ml LB bacteria culture. The culture was incubated at 37°C, while shaking at 180 rpm, until the cells reached an OD600 of 0.6 - 0.8. To check the protein yield of the expression test, a 1 ml sample was taken before induction with L-arabinose and the OD600 was measured. After induction, further samples were taken at 30 min intervals and the respective OD600 was measured in order to bring all smaples to an OD600 of 10. The cell samples were pelleted, resuspended in 2x laemmli buffer and loaded an an SDS-PAGE, to then perform a Western Blot. Figure 10 shows the results of VbrK overexpression, induced with 0.002% (w/v) L-arabinose. It clearly shows that no protein could be detected before the induction (0 h), however only after 30 min two protein bands are noticeable, of which the band at 55 kDa is highly suspected to be our 6xHis-tagged Vbrk. The amount of protein does not increase much after 3 h.

Figure 10: Expression test of pBxC3H-VbrK, induced with 0.002% (w/v) L-arabinose. Cell samples were extracted every 30 min during the expression test, harvested and respectively dilluted in 2x laemmli buffer to an OD600 of 10. The samples were loaded a Mini-PROTEAN TGX Stain-Free Gel (BioRad). The marker used is the PageRuler™ Prestained Protein Ladder (Thermo Fisher Scientific). The gel was run at 90 V for 90 min and tranfered to a nitrocellulose membrane using the Trans-Blot Turbo™ Transfer System (BioRad). Primary antibody: Monoclonal Anti-polyHistidine antibody produced in mouse (Sigma Aldrich), 1:2000 dillution in 3% BSA (TBST). Secondary antibody: Anti-Mouse-IgG produced in goat (Sigma Aldrich), 1:10000 dillution in 3% BSA (TBST). Amersham ECL Prime Western Blotting Detection Reagent (Cytiva) was used for antibody detection.

Figure 11 shows the results of VbrR expression test which was performed using different L-arabinose concentrations. In both cases, there is only a very light band in the sample that was taken before induction at a height of 40 kDa, which is also present after induction. Four more bands were detected in the remaining samples after induction, with a size of ~60 kDa, 35 kDa, ~27 kDa and 24 kDa. The band at 27 kDa is suspected to be VbrR (~25 kDa) with the additional 6xHis-tag. The higher concentration of 0.02 % (w/v) on the left side shows a higher protein yield, in comparison to the right side where protein expression was induced using 0.002 % (w/v) L-arabinose. The overall protein amount decreases with time in both cases, suggesting that a longer expression time has a negative impact on the protein yield.

Figure 11: Expression test of pBxC3H-VbrR, induced with 0.02% (left) and 0.002% (w/v) L-arabinose (right). Cell samples were extracted every 30 min during the expression test, harvested and respectively dilluted in 2x laemmli buffer to an OD600 of 10. The samples were loaded a Mini-PROTEAN TGX Stain-Free Gel (BioRad). The marker used is the PageRuler™ Prestained Protein Ladder (Thermo Fisher Scientific). The gel was run at 90 V for 90 min and tranfered to a nitrocellulose membrane using the Trans-Blot Turbo™ Transfer System (BioRad). Primary antibody: Monoclonal Anti-polyHistidine antibody produced in mouse (Sigma Aldrich), 1:2000 dillution in 3% BSA (TBST). Secondary antibody: Anti-Mouse-IgG produced in goat (Sigma Aldrich), 1:10000 dillution in 3% BSA (TBST). Amersham ECL Prime Western Blotting Detection Reagent (Cytiva) was used for antibody detection.

The expression test for BlaR1 with the according western blot (Figure 12) shows very low signal strength in all of the collected samples. It only shows a single protein band with a size of 55 kDa, which corresponds to the size of BlaR1. However, the band can also be seen in the sample before protein induction, which indicates either a leaky promotor or that a different protein was detected during western blot analysis.

Figure 12: Expression test of pBxC3H-BlaR1, induced with 0.02% (w/v) L-arabinose. Cell samples were extracted every 30 min during the expression test, harvested and respectively dilluted in 2x laemmli buffer to an OD600 of 10. The samples were loaded a Mini-PROTEAN TGX Stain-Free Gel (BioRad). The marker used is the PageRuler™ Prestained Protein Ladder (Thermo Fisher Scientific). The gel was run at 90 V for 90 min and tranfered to a nitrocellulose membrane using the Trans-Blot Turbo™ Transfer System (BioRad). Primary antibody: Monoclonal Anti-polyHistidine antibody produced in mouse (Sigma Aldrich), 1:2000 dillution in 3% BSA (TBST). Secondary antibody: Anti-Mouse-IgG produced in goat (Sigma Aldrich), 1:10000 dillution in 3% BSA (TBST). Amersham ECL Prime Western Blotting Detection Reagent (Cytiva) was used for antibody detection.

These results confirmed that we were also able to express the proteins needed for our biosensor, and the next step in building our biosensor was accomplished.

While whole-cell microbial biosensors have many advantages, their real-life application is restricted due to various regulations limiting the use of genetically modified organims. That is why we wanted to test the purification of VbrK as well as BlaR1, to see whether we could use them in an in vitro approach for direct detection of β-lactam antibiotics.

Therefore, we tried to establish the purification of both membrane proteins. We adapted protocols published in literature and adapted them for protein purification in E. coli.

To analyze if the purification worked, we performed an SDS-PAGE with a subsequent Coomassie staining to visualize protein bands in the gel. The purification of both proteins, VbrK (Figure 13) and BlaR1 (Figure 14) was successful!

Figure 13: Coomassie-stained SDS-PAGE (12.5 % acrylamide) with samples of the purification of VbrK. Left to right: PageRuler™ Prestained Protein Ladder (ThermoFisher Scientific Inc.; Waltham, USA), Cell lysate, Cytosol, Unsolubilized fraction (SolP), Solubilized fraction (SolSN), IMAC Flowthrough (FT), IMAC Wash, IMAC elution fractions 1-5. The IMAC elution fractions E2 and E3 show protein bands for VbrK at 55 kDa.

Figure 14: Coomassie-stained SDS-PAGE (12.5 % acrylamide) with samples of the purification of BlaR1. Left to right: PageRuler™ Prestained Protein Ladder (ThermoFisher Scientific Inc.; Waltham, USA), Cell lysate, Cytosol, Unsolubilized fraction (SolP), Solubilized fraction (SolSN), IMAC Flowthrough (FT), IMAC Wash, IMAC elution fractions 1-5. The IMAC elution fractions E2-E4 all show protein bands for BlaR1 at 70 kDa.

While we did not finish the final prototype of our biosensor yet, our FX-cloning strategy was sucessful and we were able to express VbrK, VbrR as well as BlaR1 in E. coli cells. Additionally, first purification attempts of VbrK and BlaR1 were successful. These results are the perfect basis to continue with building our biosensor!

For the positive control we tranformed the plasmid backbone pOdd3 into E. coli BL21 cells. The plasmid contains a β-galactosidase under control of a lac promotor and can be tested by induction with IPTG. On the premarked spot, 10 µl LB Medium, 10 µl IPTG and 10 µl of 1 mg/ml X-Gal in water were added. After incubation time, the spot where the samples were applied turned blue (Figure 15) while a negative control stayed colorless (not shown here). As the lac promotor itself is leaky, a color change without the addition of IPTG can be observed.

Figure 15: Paperstrip test with different IPTG concentrations and 1 mg/ml X-Gal in water differnt X-gal concentrations.

We were able to establish a method to functionalize our paperstrip approach.

Figure 16: In-gel fluorescence of the GFP-positive control used to test the boba approach. Samples were measured for 1-3 hours and after induction with either 250 µM or 500 µM IPTG.

Figure 17: In-gel fluorescence of the GFP-positive control used to test the boba approach. Samples were measured for 1-3 hours and after induction with 1000 µM IPTG.

These cells were then encapsulated in bobas, which were then incubated with 250 µM, 500 µM and 1000 µM IPTG for 3 hours, and being imaged under a fluorescence microscope every hour during this process. Again, an IPTG concentration dependent change in fluorescence intensity was observed, gradually increasing over time.

The bobas containing fluorescent bacteria were imaged using an epi-fluorescence inverse microscope (Zeiss). Bobas were subemerged in solutions with 1mM, 0.5mM and 0.25mM IPTG, respectively, and images where taken at time-points 0h, 1h and 2h.

Images were taken with a monochromatic camera and the mean pixel intensity of each boba was analyzed. The images were segmented using ImageJ. Mean pixel intensity was calculated using ImageJ, as well.

The analysis shows a fast increase in fluorescence from time point 0 to time point 2 proving a fast read-out and a drop slight drop after 2 hours. The highest increase was observed in 1000 µM IPTG solution. Still, lower concentrations are sufficient for a reliable read-out and are less harmful to the cells. As a matter of fact, IPTG has cytotoxic effect when applied in high concentrations. This may the reason for the drop of fluorescence (Dvorak et al. 2015).

Figure 18: Graph showing the mean pixel intensity per boba (normalized) in three different condition over time. Bobas were submerged in 250µM IPTG solution, 500µM IPTG solution or 1000µM IPTG solution. 8-bit images were used and segmentation was performed free-handedly in ImageJ.