Created with BioRender.com.
We first wanted to test the paper strip approach with a positive control. As the pOdd plasmids that we used for molecular cloning already contain a β-galactosidase under control of a lac promotor, we chose them as a positive control. Here, the protein expression should be induced by the addition of Isopropyl-β-D-thiogalactopyranoside (IPTG), a substance that binds and inhibits the repressor. However, the lac promotor is considered to be a leaky promotor and thus there will be some background protein expression prior to induction. As a negative control we chose the plasmid backbone JUMP29-1A which contains a GFP under control of a lac promotor, but no β-galactosidase. Our first test of the paper strips was an approach published in literauture by Stocker et al. in 2003.
We transformed pOdd3 into the E. coli expression strain BL21(DE3). We subsequently inoculated a 5 ml LB preculture supplemented with 100 µg/ml ampicillin and let the culture grow at 37°C and 180 rpm. The next day, we used this culture and followed the approach from Stocker et al 2003. Here, the overnight culture is harvested by centrifugation and then resuspended in 1 ml of drying protectant solution (0.5% (w/v) peptone, 0.3% meat extract, 10% gelatine, 1% sodium ascorbate, 5% raffinose and 5% sodium glutamate). After applying the samples to Whatman filter paper, the paperstrips are freeze-dried using lyophilization.
Before testing the paperstrips, we first wanted to see if we can induce and see the expression of the β-galactosidase in the cell suspension we applied to our paperstrips. This allows us to actually verify if our samples work and is an important control to check wether the paperstrip approach worked. For the test we took 20 µl of cell suspension, added 4 µl of 100 µM IPTG and 12 µl 20 mg/ml X-Gal in DMF and incubated them at 37°C. The tested cultures of the BL21 cells containing pOdd3 all turned blue and thus, protein expression of the β-galactosidase was verified.
For testing the paper strips, we followed the procedure from Stocker et al 2003. We placed the paper strips in 1M, 100 µM and 10 µM IPTG solution and incubated them at 30°C for 30 min. To prevent them from drying out, we placed them in a plastic bag. 5 µl of 20 mg/ml X-Gal in DMF solution were then pipetted on the premarked spot. 2 h later, there was a color change to a greenish color.
Since the positive control for β-galactosidase expression in the cell suspension turned blue as expected, we were irritated by the greenish colour of the paperstrip. After some reading and evaluation, we concluded that the green color is due to the reaction of X-Gal with light and not due to cleavage by the β-galactosidase. Another possible error source was the treatment process of the paperstrips: we were worried, that by placing the paper strips in a solution, the cells might diffuse from the paper strip. After all, we concluded three possible error sources: either the signal intensity was too low because the cell density was too low, the cells diffused from the paperstrip into the testing solution or the reactivation process of the bacteria did not work.
To enhance the signal, we decided to place 2x 5µl of bacteria suspension on the paper strip instead of only 1 x 5µl. This time, we also did not place the whole paper strip into the testing solution to prevent possible diffusion of the cells into the testing solution.
In this cycle, we only adapted the amount of bacteria which were placed on the paper strip but still followed the same procedure as in the first cycle. Additionally we also established a proper negative control by only placing LB medium or drying protectant solution on the paper strip. This way we can directly compare paperstrips witha and without X-Gal and confirm that a change in color is specific due to the enzymatic reaction of the β-galactosidase and not just a reaction of X-Gal with light.
Instead of placing the paper strip into the testing solution, we pipetted LB medium, the test solution and X-Gal directly onto the paper strip. After incubation at 37°C in the dark for about 30 minutes the color change could be observed. We also tested to reactivate the bacteria with water. The blue spot which appeared was less vibrant than with LB Medium. The negative controls with only medium or drying protectant solution stayed colorless. We also noticed that using 20 mg/ml X-Gal in DMF causes only a color change at the outer circle of the bacterial spot. We diluted this X-Gal stock in water to test different X-Gal concentrations and noted that 1 mg/ml X-Gal is sufficient to observe the color change.
Figure: Paper strip test with different IPTG concentrations and 20 mg/ml X-Gal in DMF.
Figure: Paper strip test with differnt X-Gal concentrations.
In order to use paper strips as a platform for any kind of whole-cell biosensor with a β-galactosidase as reporter gene, the cells need to be reactivated by pipetting LB medium directly onto the dried spot. Additionally, lower concentrations of X-Gal are more effective to get an vibrant and round colour change.
To test the bobas with GFP as a reportergene, we first wanted to test a plasmid containing gfp under the control of a lac promotor.
We adapted the protocol form the 2020 Vilnius iGEM team. A 2,5% alginate solution is mixed with an overnight culture in the ratio of 4:1. Using a syringe the bobas then are released in droplets into a 0,1 M CaCl2 solution.
We observed the bobas with a fluorescence microscope with a GFP Filter. The smallest magnification is 10x and for the first try, the bobas were too big to be seen as a sphere. So we tried using a canula to obtain smaller bobas. After creation of the bobas, they were incubated in LB Medium, water and 1M IPTG. Fluorescence could only be observed in the 1M IPTG solution.
The bobas need to be made as small as possible to be observed with a fluorescence microscope but can be used as a platform for biosensors with gfp as reportergene.
To assemble our whole-cell biosensor for detection of β-lactam antibiotics, we first had to clone the separate parts into expression vectors to test the optimal conditions and sensitivity of the promotors. Two different systems were chosen to be promising. The VbrR/VbrK system and the BlaI/BlaR1 system both contain two components and a respective reporter gene, which is either the expression of a β-galactosidase or the expression of GFP. VbrR/VbrK and BlaI/BlaR1 were all tagged with a 6xHistidin-tag, to purify the proteins and to detect the protein in western blot analysis. The GFP fluorescence should be detected through in-gel fluorescence and the presence of the β-galactosidase can easily be confirmed by adding IPTG and X-Gal and observing a blue color.
Each gene was inserted into a plasmid vector containing a promotor under the control of L-arabinose. The plasmid was cloned via FX-cloning and transformed into E. coli C43 cells. A single clone was picked and grown, to purify the inserted plasmid. The plasmid was prepped and sequenced to confirm the right insert and to exclude mutations.
To test the best conditions for protein expression in order to determine which conditions are suitable for our biosensor and how fast it responds, we inoculated a preculture with a single positive clone. This was used to inoculate a 50 ml main culture, which was grown at 37°C to an OD600 of 0.6-0,8. The protein overexpression was induced by adding 0.02% (w/v) or 0.002% (w/v) L-arabinose. Samples of the bacteria culture were taken before induction and in steps of 30 mins after induction, up to 3 h. All samples were diluted respectively in SDS loading buffer to an OD600 of 10 and loaded on an SDS-PAGE, to perform a western blot. The overexpressed protein was detected using a primary murine antibody against the polyhistidin-tag (1:3000 dilution) and a secondary antibody against mouse (1:10000 dilution).
The western blots of BlaR1 overexpression showed very low signal, so we decided to increase the concentration of the primary antibody against the polyhistidin-tag to a 1:1000 dilution and also increase the incubation time. The membranes of the VbrK overexpression showed that 0.02% and 0.002% (w/v) L-arabinose induced protein expression was very high. Using the higher concentration resulted in such high protein amounts that there was no visible difference of VbrK between the timepoints 30 min and 3 h. This showed that a very low L-arabinose concentration is sufficient to express VbrK and that an incubation time of 30 min is enough to expect results from the biosensor.
Figure: Western blot of VbrK expression test. Induction with 0.02% (w/v) L-Arabinose.
Figure: Western blot of VbrK expression test. Induction with 0.02% (w/v) L-Arabinose.
We conducted a two-step analysis for predicting the VbrR binding site. For the promoter prediction, we opted for SAPPHIRE, which is tailored for σ70 promoter prediction in Pseudomonas and Salmonella, two closely related genera to Vibrio haemolyticus. With our insights into potential promoter sites for σ-factors, we were well-equipped to ascertain the binding site for VbrR. Assuming similarity to a known 49 bp binding site in the exsC promoter of type 3 secretion systems, we performed local alignments against our 300 bp sequence containing the putative binding site. Additionally, we decided to map two DNA half-sites identified by Hong et al. onto our sequence to identify potential binding sites for the VbrR DNA-binding domains.
Besides using SAPPHIRE for promoter prediction, we further investigated the region from the BlaA gene start to 300 bp upstream to identify the VbrR binding site. Employing local alignments, we clustered the top 1000 sequences by score, discarding those exceeding a limit of 20 bp on one side of the prior sequence. For DNA half-site identification, we also employed alignments, focusing solely on evaluating the gap size between the two half-sites and the number of mismatches.
We were able to enhance our understanding of the genetic regulation of our reporter genes. After generating our local alignments, we found about 90% of all local alignments for the binding site prediction to be potential binding sites. We further found no perfect alignment for the mapping of the two DNA half-site motifs to our sequence, but selected two sequences as promising binding sites, favoring fewer mismatches and accepting larger gap sizes. The exact interaction between VbrR and the σ70 factor remains unknown. However, by integrating several studies connected to VbrR's superfamily, and predicting potential promoter sequences using SAPPHIRE, we were able to determine VbrR's binding sites and their potential function. We predicted that the promoter region is located within sequence positions 477560 ← 477668 (108 bp), and the enhancing region is positioned at 477707 ← 477761 (54 bp).
We were able to predict the promoter region of BlaA and the binding site of VbrR. However, experimental work is required to validate the predicted binding sites, promoter regions, and their function. Nevertheless, we are confident in the reliability of our results.
For in silico mutation of our receptor, we employed the Rosetta interface design protocol to automatically design the protein interface interacting with our antibiotics. The antibiotics, in their closed β-lactam ring state, were placed in the binding pocket. A rotamer library was created and 1000 mutated BlaR1 receptors were generated. Afterward, the top 3 mutants determined by Rosetta's total score were evaluated using MOE and PLIP. Thereby, the interaction between the antibiotics and the receptor was assessed after docking them covalently back into the optimized receptors. The approach was driven by the intricate process of the antibiotic's interaction with the protein. Specifically, the antibiotic must initially reach the serine residue while still in its active state, before subsequently undergoing a reaction and forming a covalent connection with it.
Molecular dockings were conducted in the Molecular Operating Environment (MOE). The binding site for β-lactam antibiotics was chosen based on Wilke et al.'s publication. For noncovalent docking, we identified the general binding pocket, for covalent docking, we further specified the preferred Serin as the reactive site, employing an alkylester reaction with ring opening as the reaction mechanism. However, MOE does not complete covalent docking autonomously. Hence, a manual intervention was necessary to establish a truly covalent bond between the macromolecule and its ligand.
The approach with Penicillin G failed to yield a receptor with significantly increased binding affinity. However, for Ampicillin, one mutant displayed higher binding affinity, indicated by more hydrogen bonds and hydrophobic interactions, than the covalent Penicillin G binding state. Compared to the naive receptor, the mutant receptor exhibited a noticeable reduction of critical interactions, notably the high-energy ionic bond of the carboxylate group and specific hydrogen bonds. However, it did manage to establish a distinctive T-shaped π-stacking interaction between the benzyl ring and a phenylalanine residue.
Although we found a BlaR1 mutant for Ampicillin, we did not manage to find major differences in the interactions between BlaR1, BlaR1 mutants and the considered antibiotics. In the non-covalent approach, we optimized the protein's interface considering potential conformations of the antibiotics. This approach could lead to changes in amino acids that do not interact with the antibiotic in its inactive state (open β-lactam ring) and therefore make it difficult for Rosetta to generate suitable mutants of BlaR1. Hence, we chose a second approach for generating the BlaR1 mutants.
To address the limitations of our non-covalent approach, we considered incorporating covalent dockings into the Rosetta workflow. However, the protocol proposed by Rosetta was not suitable for dealing with covalent binding. As an alternative, we pursued optimization of the protein interface using the inactive state of the antibiotics without actually forming a covalent bond with the protein. The rest of the pipeline remained intact, constituting what we term the pseudo-covalent approach.
Instead of handing over noncovalent dockings to Rosetta, we generated covalent dockings in MOE without manually linking the ligand to the receptor at the end. We then passed these pseudo-covalent dockings to Rosetta. The key difference lies in the fact, that Rosetta could now optimize the receptor for the inactivated antibiotics with the opened β-lactam ring.
The pseudo-covalent approach, in a similar vein, did not yield a structure with markedly increased estimated binding affinities for Penicillin G. Surprisingly, the optimization consistently leaned towards Ampicillin, regardless of the intended antibiotic in the interface design. However, upon closer inspection of the top three results for Penicillin G, we identified mutants that exhibited notably exceptional estimated binding affinities for Ampicillin.
In total, we successfully developed two mutants of BlaR1 with significantly increased estimated binding affinities for Ampicillin. However, the preferable BlaR1 mutant structure for Ampicillin was found during optimization for Penicillin G. This suggests the potential benefit of employing similar antibiotics in the optimization process to attain globally optimal results for the desired antibiotic, though it also stresses the difficulty to make the receptor discriminate between very similar molecule structures.