Engineering Success

At the core of Concave is the idea to create a simple and accessible toolbox that allows scientists to design a protein binder matching their specific needs. To make this development process cheap and time effective, we wanted to establish a system that autonomously creates a vast library of various protein binders and screens that library for the binder with the best affinity to a protein of interest. Repebodies, whose binding affinity is conveyed by leucine-rich repeat modules [3,4], are the optimal choice. They are designed to be easily expressed in bacteria, which makes their production less expensive and they are compatible with our chosen diversification method. We wanted to employ a mutation system based on retrons to continuously mutate the Repebodies in vivo until the selection system recognizes the ideal one. As a first step we decided to test the binding specificity of a mCherry-binding-Repebody [5]. To confirm this interaction, we performed pulldowns, gel filtrations and a Far-Western blot.

We created two versions of the mCherry-binding-Repebody: The unmodified-mCherry-binding-Repebody (BBa_25X0LA8K), with an E. coli codon-optimized version of the original sequence. The ssDNA-optimized-mCherry-binding-Repebody (BBa_25IQCJ3B), which was specifically built to function with the retron mutation system. We decided to use EGFP as a negative control for all our experiments, because it has a very similar size and structure to mCherry, and no interactions between EGFP and the mCherry-binding-Repebody have been reported [5]. The Repebodies were tagged with a 6xHis-tag and mCherry and EGFP with an HA-tag, to facilitate experiments such as pulldowns and Western blot.

Using various experimental setups, such as pulldowns, gel filtration and Far-Western blotting, we were able to show that both Repebodies have an affinity for mCherry, whereas EGFP gets recognized only within the framework of the Far-Western blot. We also worked on testing the retron mutation system, the selection system and the implementation of Repebodies into targeted protein degradation.

During our project we managed to proof that our Repebodies bind specifically to mCherry and not to EGFP. We took the first steps of implementing the retron mutation system into the Concave toolbox. The BACTH system was successfully tested with the Repebody-mCherry combinations and we have moved towards creating the first Repebody-targeting-chimera for targeted protein degradation. Have a look at our results page for more detailed information. We made significant progress towards a functional Concave toolbox and a viable, cheap, easily accessible and ethical antibody alternative. We think that Concave is the future of research.

To investigate the interaction between GBP and EGFP, we designed an experiment using immobilized metal affinity chromatography (IMAC) to capture binding events. Our initial experimental design involved the following constructs (kindly provided by the Kraft Lab at the University Freiburg):

• pIG25-17: EGFP-6xHis
• pIG25-19: GST-mTagBFP2-GBP
• pIG25-20: GST-mTagBFP2-mGBP - a mutated version of GBP
• pIG25-24: GST-mTagBFP2

After expressing the protein in E. coli and preparing the cell lysates, we combined them and performed IMAC to test whether GBP could pull down EGFP through His-tag affinity. The resulting protein mixtures were analyzed by SDS-PAGE and Coomassie staining to observe if GBP specifically binds to EGFP and to distinguish true interaction from background signals.

Proteins were expressed in E. coli BL21 (DE3) cells. After growing and harvesting the cells, we lysed them by sonication to release the proteins. We collected the soluble fraction by centrifugation and loaded them onto Ni-NTA affinity columns, which selectively pull down His-tagged proteins. By combining EGFP-6xHis with the GBP fusion proteins and controls during the binding step, we set up the system to directly test if GBP would co-purify with EGFP through specific binding interactions. The negative controls were used to verify whether a mutated version of GBP could still bind EGFP and if EGFP could bind GST or mTagBFP2.

If GBP bound to EGFP, it would remain in the Ni-NTA elution. SDS-PAGE and Coomassie staining revealed bands corresponding to the expected sizes of 30 kDa for EGFP and approximately 64 kDa for GST-mTagBFP2-(m)GBP fusion proteins, which suggests an interaction. However, the negative controls also showed band retention (Figure 1).

Coomassie staining after SDS-PAGE revealed bands indicating interaction between GBP and EGFP. However, the negative controls revealed background interaction of the non-His tagged GST-mTagBFP2 construct, indicating unspecific binding to the Ni-NTA resin.

Although Coomassie staining confirmed the presence of the proteins, it could not discriminate between specific and non-specific interactions. These limitations highlighted the need for more specific tagging and detection methods.

Building on the last cycle, the next iteration of the experiment was refined to eliminate sources of extraneous interaction. GST and mTagBFP2 were removed, leaving GBP and its mutant each tagged only with a His-tag, while EGFP was fused to an HA-tag for specific detection via Western blotting. This simplified setup aimed to reduce unspecific binding and improve detection specificity.

GBP and mGBP were cloned into the pET303 vector, both being labeled only with a His-tag (pIG25-123 and pIG25-124). EGFP was also cloned into this vector with a HA-tag for detection via western blotting (pIG25-111). These plasmids were then expressed again in E. coli BL21 (DE3) cells, after which the soluble fraction was collected via sonication and centrifugation.

The soluble fractions were used as input for Co-NTA columns. The Western blot revealed that EGFP-HA signal only appeared in the elution fraction when it was co-loaded with GBP-His, which confirmed the binding specificity. The negative controls confirmed that mGBP does not bind EGFP, and EGFP alone is washed away, indicating that it does not interact with the column. (Figure 2)

To detect the presence of a specific protein in a sample, transcriptionally adding a detectable affinity tag like HA and blotting against it serves as a more specific readout than Coomassie staining. While Coomassie can confirm the presence and approximate size of proteins separated by SDS-PAGE, it does not reveal their identity.

Additionally, keeping an experimental design simple, in this case removing unnecessary fusion proteins such as GST or mTagBFP2, can help reduce unexpected issues, such as non-specific interaction with affinity resin.

This refined setup conclusively demonstrated specific binding between GBP and EGFP while eliminating major sources of background interference.

The initial experimental design to test binding included attaching a His-tag to the unmodified-Repebody in tandem to using His-tagged mCherry, purifying them both with Co-NTA columns, and then natively separating them by their hydrodynamic size using a gel filtration column. If they interact, the formation of a dimer would be measurable via the shift in the molecular weight.

The unmodified Repebody was cloned into the pET303 backbone which conveniently includes a His-tag behind its multiple cloning site, allowing for easy expression of His-tagged proteins. The plasmid used for expression of His-tagged mCherry was supplied to us by the Signaling Factory and Robotics (Uni Freiburg, CIBSS BIOSS). Both of them were expressed in E. coli BL21 (DE3) cells, with the Repebody being expressed at the following conditions: 30°C / 3 h and 16°C / 24 h. IPTG concentrations for induction were either 0.25 or 0.5 mM, resulting in a total of 4 expression conditions.

After collecting the soluble fractions of the expressed proteins via sonication and centrifugation, they were purified using Co-NTA columns. Purified proteins were verified via Coomassie staining (Figure 3) and protein concentration was quantified by absorbance at 280 nm.

Although the observed peak shift indicated binding, Coomassie staining showed multiple contaminant bands. Additionally, the Repebody was underrepresented compared to mCherry, which limited the visibility of interaction. These results emphasized the need for purer samples and sufficient concentrations to saturate the binding partners in order to achieve clearer chromatographic outcomes.

To improve the result quality, the next experiment was designed to saturate mCherry with the Repebody. Additionally, we wanted to have large volumes of highly pure samples to obtain clearer images of the shift in molecular weight.

This required extensive purification of large amounts of proteins. To achieve this, the expression of the Repebody in the pET303 backbone was optimized under variable conditions to identify those producing the most soluble protein. Verification was done via semi-quantitative western blotting. The intention was to create large volumes of Repebody to combine with fluorescent proteins mCherry, EGFP, mTagBFP2 in a 2:1 ratio to fully saturate them.

The expression test revealed which conditions were optimal to use for expression of the ssDNA-Repebody (Figure 6), which were then employed for expression of both Repebody variants and mCherry, and also EGFP and mTagBFP2 as negative controls. All these proteins were His-tagged, allowing for both detection via western blot and also purification.

After expression and collecting of the soluble fractions, they were purified with affinity columns and filtered via gel filtration to collect the fractions with the correct molecular weight (Figure 7).

With large volumes of highly pure samples, the Repebody variants could be analyzed for their binding behavior and specificity by co-loading them in a 2:1 ratio with the fluorescent proteins mCherry, EGFP and mTagBFP2, respectively.

The entire mCherry peak shifted toward a higher molecular weight when combined with each of the Repebody variants, whereas neither EGFP nor mTagBFP2 exhibited the same behavior (Figure 8).

With optimized conditions based on the findings from the first iteration, the assay successfully demonstrated the interaction between mCherry and the Repebody variants.

In general, when choosing gel filtration as a binding assay, it is beneficial to have large volumes of pure proteins. In order to achieve this,it is helpful to first evaluate which expression conditions yield the highest of the desired protein, as this will make the results more conclusive.

Using proper negative controls, in this case EGFP and mTagBFP2 also improves the validity of the results and allows for a better estimation of how specific the interaction between our two binding partners is.

To confirm the binding of the Repebody to mCherry, we employed the use of a different binding assay, a pulldown. This assay allows for the analysis of binding affinity between prey and bait proteins by immobilizing one protein on Ni-NTA resin and capturing the interacting partner from the sample. In our case, the prey is the Repebody, which has a His-tag fused to it, thereby allowing Ni-NTA resin to selectively pull on the Repebody. When added,the bait protein, such as mCherry, binds to the Repebody while other proteins are washed away. When detaching the prey protein by outcompeting the affinity of the His-tag to the resin with imidazole, the bait protein is eluted with the prey protein in the corresponding fraction. Detecting the bait protein via methods such as western blot proves the interaction of bait and prey. In our set-up, we used HA-tagged mCherry and EGFP allowing detection via anti-HA blotting and analysis of Repebody specificity by testing whether it can bind EGFP, which is very structurally similar to mCherry.

To express the binding partners recombinantly in E. coli BL21 (DE3), we used the pCDF-Duet1 backbone to express both the unmodified-Repebody and the ssDNA-Repebody with either mCherry-HA or EGFP-HA. This backbone contains two multiple cloning sites, enabling co-expression of two proteins in the same cell, thus allowing for investigation of binding under cytosolic conditions.

The following plasmids were used:

• pIG25-113: pCDFDuet1_ssDNA-Repebody-His / mCherry-HA
• pIG25-114: pCDFDuet1_ssDNA-Repebody-His / EGFP-HA
• pIG25-116: pCDFDuet1_unmodified-Repebody-His / mCherry-HA
• pIG25-117: pCDFDuet1_unmodified-Repebody-His / EGFP-HA

After loading the soluble fractions onto the Ni-NTA resin, washing away unspecific binding and collecting the bait and prey proteins, the samples were analyzed by Western blotting. This revealed mCherry-HA bands when expressed with each Repebody variant (Figure 9), indicating the interaction to not only be specific to mCherry, but also detectable via a pulldown assay.

The pulldown assay verified binding between Repebody and mCherry but introduced new uncertainties: the co-expression setup made it difficult to ensure that both bait proteins were available in equal concentrations. This is a problem, as mCherry and EGFP have to be loaded with the Repebody in the same amounts to compare affinities. Additionally, potential binding of mCherry or EGFP alone to the resin could not be ruled out as they are always co-loaded with the Repebody. These limitations indicated the need for separate protein expression to check unspecific background binding and achieve controlled molar exposures.

While the general principle for the pulldown remains the same, the next pulldown only focused on the ssDNA-Repebody as our best basic part, and did not rely on co-expressed samples anymore. Separate expression of bait proteins enabled measuring of their concentration via fluorescence, which in turn allowed the pulldown to expose the prey protein to known numbers of molecules, therefore giving us better information about how strongly the ssDNA-Repebody interacts with mCherry compared to EGFP.

This setup enabled comparison of binding strengths and specificity, key for evaluating the Repebody as an antibody alternative.

With the plasmid for expression of the ssDNA-Repebody already made, we cloned two new coding sequences into the pET303 vector:

• pIG25-110: pET303_mCherry-HA
• pIG25-111: pET303_EGFP-HA

Additional His-tagged fluorescent proteins from previously established constructs (pIG25-12 and pIG25-01) were purified using Ni-NTA and gel filtration to generate high-purity standards for fluorescence calibration. Using serial dilutions, fluorescence intensity was correlated to protein concentration, enabling the determination of the concentration of mCherry-HA (2.79 mg/mL) and EGFP-HA (2.85 mg/mL) in their soluble fractions. This ensured approximately equal molar exposure during pulldown testing.

We performed a pulldown exposing the ssDNA-Repebody to equal amounts of moles for mCherry and EGFP, analyzing how much EGFP was detected after washing. Western blot of pulldown elution fractions revealed that the affinity towards mCherry was much higher than to EGFP. This is evident from the anti-HA staining of the elution fractions in lane 6 containing Repebody and EGFP compared to lane 8 with the same amount of EGFP alone. The stronger band in lane 6 indicates some interaction between the Repebody and EGFP. However, this interpretation would be more convincing if the negative control (lane 8) showed no EGFP-HA signal, suggesting any signal coming from the lanes with combined EGFP and Repebody results from binding. To account for the background signal in the control, one would need to subtract the band intensity of lane 8 from lane 6 to determine the extent of interaction. This approach is far more error-prone, especially when the difference in band intensity is marginal.

While the pulldown demonstrated qualitatively that the ssDNA-Repebody binds mCherry more strongly than EGFP, the limited quantitative precision of Western blotting and the imperfect negative control behavior made affinity estimation difficult. In order to properly assess whether and to what extent the ssDNA-Repebody interacts with EGFP, the interaction must be analyzed using truly quantitative methods, such as fluorescence polarization or biolayer interferometry (BLI). These methods can accurately measure binding strength and specificity.

Despite these constraints, our results provided valuable insight into the selective, but not entirely exclusive, binding of theRepebod to mCherry compared to similar fluorescent proteins. Encouragingly, we also found that theRepebody`-mCherry interaction is consistently detectable across multiple independent assay types.

To complement the biochemical binding data and advance toward practical applications, we performed a Far Western Blot to evaluate the ability of the Repebody to specifically recognize mCherry immobilized on membranes.

Unlike standard Western blotting that uses antibodies, this experiment explored the use of Repebodies as the primary detection probes. The mCherry-HA target protein was immobilized on a membrane and incubated with the ssDNA-Repebody, after which it was detected using anti-His and HRP-conjugated secondary antibodies. This setup served as a proof-of-concept for applying Repebodies in this fundamental molecular biology technique.

The ssDNA-Repebody containing a 6xHis-tag (pIG25-108 was expressed in E. coli BL21 (DE3), purified using Ni-NTA columns (Figure 13), and prepared as a blotting solution at a concentration of ~1.1 mg/mL. The target proteins, mCherry-HA (pIG25-110) and EGFP-HA (pIG25-111) were expressed similarly in E. coli BL21 (DE3) and total cell lysates were prepared for SDS-PAGE and membrane immobilization.

• Attempt 1:mCherry-HA bands were detected, confirming binding. However, the signals were weak and there was high background noise. Anti-His controls showed no signal in the absence of the Repebody. Background reduction was prioritized.
• Attempt 2:Tween was increased in the blotting and washing steps. A blotting solution control was included. Semi-native (not getting heated) and denatured proteins were tested. The anti-HA and Ponceau S staining were added as controls. Native mCherry and EGFP were detected, but the negative controls behaved unexpectedly, which undermined the validity of the results.
• Attempt 3:The protein transfer time was shortened by half to improve the transfer efficiency of native protein. Native and denatured mCherry and native EGFP were successfully detected with stronger signals and proper control behavior.
• Attempt 4:The transfer time was reduced by a third, diluted samples fourfold and used quantified protein concentrations from previous pulldown assays. Normalized exposure times enabled more quantitative analysis. Even the diluted native samples showed detectable signals. Native EGFP was recognized, and mCherry was detected in both native and denatured forms (Figure 13).

The experiments demonstrated that Repebodies can replace antibodies in Western blotting for certain targets. Iterative refinements, such as increasing detergent concentration, optimizing transfer times, and implementing rigorous controls, significantly reduced background noise and improved the reliability of detection. The assay confirmed the binding of the ssDNA-Repebody to both native and denatured mCherry, and native EGFP. This validates the potential application of Repebodies and highlights the importance of optimization and control design. This process is a proof-of-concept for integrating Repebodies into standard protein detection workflows.

We wanted a measurable report, by which we could determine the rate and efficiency of retron integration into a given gene. Intuitively, it was first proposed to use a fluorescent protein, in conjunction with a loss-of-function point mutation which would cause the fluorescence of the culture to decrease.

In order to select which point mutation to use, three mutations were selected, two proposed by Fu., J. et al. [6] and one which we derived, as it would disrupt the formation of the fluorophore chemically. These mutations were introduced into a T7-lac promoted EGFP gene using site directed mutagenesis. The cloning was performed directly in BL21(DE3) cells, so the fluorescence of each colony could be measured directly.

16 colonies from the cloning plates for each tested mutation were inoculated into LB-Amp in a 96-well plate, and were grown until they had a normalized OD600 of 0.5, at which point they were induced with IPTG at a concentration of 0.5mM. The expression was run overnight at 25˚C, and the fluorescence was measured using a plate reader the following day.

From the fluorescence reading, we could conclude that all mutants had a 10x decrease in fluorescence intensity compared to a EGFP positive control. However, we also realized that this method of testing the mutation rate would be very cumbersome to normalize, and difficult to run for multiple samples from the running mutation mix. Furthermore, the measurable report is very indirect if applied to measuring mutation rates as is planned in the final experiment, as in the final experiment we aim to measure the ratio of mutated genes by the expression of the protein coded by that gene.. Because of this, we decided to move away from fluorescent reporters. Instead, we landed on using qPCR directly with the Repebody as the mutagenic target, since this experiment would also be more representative of our desired system

To test the qPCR measurement of a liquid culture’s cells, we first ran a qPCR using a set of dilutions from a DH10b culture, which had been transformed with pMS366 without any further additions. This was meant to serve as a standard measurement, since the amount of mutations in the mutation target are zero. We wanted to use a liquid culture, since it would be an intuitive way to measure recombination rates in the mutation cultures.

A liquid culture of DH10b + pMS366 cells was grown overnight. Samples from this culture were washed by centrifugation and resuspension of the pellet in HEPES medium. These washed cells were used to create a serial dilution.

Samples from the serial dilution were combined with the primers for amplifying the mutation site and the reference site. The measurement data from this experiment was then used to calculate the CT values for this serial dilution. When a linear regression was performed on these values, the variance in the CT values caused the linear regression’s accuracy to be heavily impacted. We decided to redo this experiment using the purified plasmid pMS366, which yielded a much tighter spread on the CT values.

Due to the higher rate of uncertainty in the qPCR reaction when using cells from a liquid culture directly, we decided to instead prepare our samples for qPCR by performing a miniprep on each liquid culture sample, in order to remove noise from our measurement.

We wanted to analyze a set of basic recombineering attempts, using our pMS366-ret and pMS366-RiboJ-ret. With the miniprepped plasmid mixture from a retron expression culture, we wanted to measure the rate of mutations caused by the retrons. To do this, we wanted to measure the CT values for the mutation culture, and then map this onto a mutation amount using a pMS366 standard curve.

We first generated a standard curve from a purified sample of the pMS366 plasmid. Following this, we prepared liquid cultures of bMS.346 cells, each transformed with pMS366-ret and pMS366-RiboJ-ret, which were then induced with arabinose in order to express the retrons. A set of negative controls was also prepared, from cultures which were not induced.

Samples from the retron expression cultures, as well as the negative controls, were miniprepped and prepared for a qPCR, by adding the qPCR reagents as well as the appropriate primers for amplifying the mutation site and the reference site. From the qPCR, the CT values were obtained, and using the standard curve from pMS366, the concentration of plasmids containing a mutation was determined. From this, the negative control samples appeared to have a non-zero rate of mutations, which was not possible.

The mapping of CT values to plasmid concentration is highly dependent upon which plasmid was used to generate the standard curve. From this, we decided to generate a new standard curve for each plasmid we wanted to test.

In order to measure the rate of mutations and verify the recombination of a retron-generated ssDNA with our primer binding site, we attempted to use both qPCR and colony PCR. For these experiments, we used pMS366-ret and pMS366-RiboJ-ret as both retron source and mutation target. For our PCR reactions, we used two sets of primers, one that binds to the retron target site, and the other binding to an unrelated region of DNA, which we used as a reference signal.

Liquid cultures containing bMS.346 cells, transformed with pMS366-ret or pMS366-RiboJ-ret, were prepared. The retrons were then expressed by inducing the cells with arabinose. Liquid cultures were also prepared which were not induced, these served as our negative controls.

Samples from the expression cultures were analyzed by either qPCR or colony PCR. For qPCR, a sample was miniprepped, and the appropriate reagents and primers were added. The CT values from the qPCR reaction were measured, and translated to plasmid concentrations using previously made standard curves. For the colony PCR, colonies obtained by a dilution streak were added to polymerase master mix and primers, and the resulting PCR products were run through an agarose gel. The thereby generated pattern of bands was analyzed for possible mutations. From these procedures, we were unable to find measurable mutations in our expression cultures.

The publication describing the retron system upon which we have built our project used a mutation in the cell’s RNA polymerase to confer Rifampicin resistance to the bacterium. This can be tested more directly, by counting the number of colonies growing on a Rifampicin plate after induction. Instead of using a PCR based mutation report, it probably would be advantageous to have a more direct report, such as by testing for Rifampicin resistance. Another possibility is a mutation in LacZ, which would allow for separation between positive and negative colonies by blue/white screening.

We wanted to test the retrons we designed to target the Repebody, expressed on pCDFDuetIso-Repebody, both with a single ssDNA donor (plG25-133) and with two (plG25-140). We measured the result through colony PCR, which would allow us to not only determine if a specific colony had a mutation, but also where the mutation occurred based on the length of the amplicon.

Liquid cultures of bMS.346 cells, transformed with pCDFDuetIso-Repebody and either(plG25-133) or (plG25-140), were prepared and grown overnight. The cells of these cultures were then induced by the addition of arabinose, and they were given time to express the retron machinery. Negative controls of each culture were also prepared, these were not induced.

In order to visualize where the retrons had recombined into the Repebody, we performed a colony PCR on samples taken from our arabinose-induced cultures, which we streaked on an agar plate in order to obtain individual colonies. These were then mixed with the necessary PCR reagents, as well as primers which bind to the possible mutated Repebody repeat sequences. The PCR products were separated by gel electrophoresis, and the band patterns were compared to the band pattern generated by a negative control. From these results, we were unable to verify successful retron ssDNA integration into the Repebody.

Since we were unable to observe our desired mutations in the Repebody using our (plG25-133) and (plG25-140) plasmids, we began questioning the retron design of these plasmids. Initially, we had considered using a variety of synthetic retron known as Multitrons, where the ssDNA template is expressed separately from the remainder of the ncRNA scaffold, as described by González-Delgado et al.. While we had initially disregarded them due to their apparent complexity, we now appreciate the work González-Delgado et al. invested into characterizing these variant retron approaches, which would have allowed us to have a larger head start when implementing our recombineering system.

The BACTH system is an already established method to test if two proteins of choice interact with each other. Additionally, it is capable of screening large libraries in vivo. The BACTH system relies on a split adenylate cyclase composed of two subunits called T25 and T18. There are multiple plasmids available on the market, which contain these subunits. We decided on using those included in the BACTH kit from Euromedex. The Euromedex BACTH kit contains two plasmids with the T25 subunit and an N-terminal or C-terminal multiple cloning site (MCS). The same is also true for the T18 subunit. Additionally, the kit includes positive control plasmids with a leucine zipper fused to the subunits, and an adenylate cyclase deficient strain. For information on our plasmids you can visit our plasmid-design page and plasmid list. The Euromedex kit was recommended to us by Dr. Chris van der Does and Dr. Thomas Wallner from the faculty of biology at the university of Freiburg.

We used the plasmids provided in the BACTH kit from Euromedex. We decided to perform the assay with both our unmodified-mCherry-binding-Repebody (BBa_25X0LA8K) and our ssDNA-optimized-mCherry-binding-Repebody (BBa_25IQCJ3B). After some deliberation, we decided to clone these Repebodies only into the plasmids containing the T25 subunits. As the targets of our Repebody that would be cloned into the T18 containing plasmids, we chose mCherry and eGFP. We hoped to prove the functionality of the BACTH system with the Repebodies and mCherry and used eGFP as a negative control. We also decided to label the Repebodies with a 6xHis-tag and mCherry and eGFP with an HA-tag, to make further experiments such as expression verification simpler. We decided to insert a (GGS)3 linker in between the subunits and the proteins of interest. The cloning was done in E. coli DH10b and DH5&alpha. For the assay, we used E. coli BTH101, the adenylate cyclase deficient strain from the Euromedex BACTH kit.

We followed a slightly modified protocol as detailed in Dr. Bouveret’s review [4], Dr. Wallner’s BACTH protocol and the protocol part of Euromedex’s BACTH kit. The only difference from these protocols is that we didn’t let the transformation plates incubate for 48 hours but 24 hours. The results show that the BACTH selection system is capable of detecting an interaction between the Repebodies and mCherry, but not eGFP. To confirm these results, we verified the expression of the constructs in two functional combinations, by performing a dot blot. It showed that the Repebodies, mCherry and eGFP were correctly expressed.

Our results results show the capacity of the BACTH selection system to detect the interaction of both Repebodies and mCherry. It also proves that the Repebodies are selective enough to differentiate between the similar structure of eGFP and mCherry. Surprisingly, some constructs seemed to function with ssDNA-optimized-Repebody, but not with the unoptimized one. The reason for this phenomenon remains unclear. A possible explanation might be that certain fusion proteins were not expressed or that the induction of the lac promoter was ineffective. However, at this point in the project we didn’t have enough time to explore these issues further. We instead focused on testing and characterizing the BACTH system further. The selection system in the Concave toolbox needs to be able to filter out single bacterial cells, with a functional Repebody, from a large colony of nonfunctional ones. We wanted to check if the BACTH system is able to fulfill these requirements.

After confirming the BACTH systems basic functionality with Repebodies, we now could confirm its applicability in the Concave toolbox. We decided to perform a serial dilution experiment with two functional Repebody-mCherry combination to test if it is able to filter out a small amount of bacterial cells containing a functional Repebody-target pair, from a large mixed colony. If the BACTH system was to be applied in the Concave toolkit. In theory the experiment should function by plating the mixed culture onto minimal media plates with maltose and IPTG. Bacteria, which have a functional Repebody-target combination, thereby making them able to use maltose as an energy source, should survive on these plates. Bacteria with a nonfunctional Repebody or interaction should die, since they have no source of energy.

We diluted the functional Repebody-mCherry combinations with their respective eGFP negative controls to simulate a small number of functional Repebodies in a culture of nonfunctional Repebodies. We also diluted the positive control from the BACTH kit with the negative control. On minimal media plates bacteria colonies are invisible, so they are transferred on an LB plate after a few days of incubation. On LB plates the colonies should then grow to visible size. X-Gal and IPTG can also be used to additionally colour them blue.

The experiment was performed as described. The minimal media plates were left to incubate at 30° C for almost 60 hours. The colonies were transferred onto LB plates with IPTG and X-Gal using autoclaved cotton velvet and a custom build rig. Remarkably, a lot of the LB plates showed a mix of blue and white colonies, while some only had white colonies growing on them.

The results from the experiments are inconclusive. Theoretically, all colonies growing on the LB-plates should be blue. Since also white colonies can be seen and blue colonies are also visible on the negative control plates, it is likely the transfer didn’t work properly. A possibility is that the velvet was contaminated. This, however, is unlikely since it would have been contaminated with adenylate cyclase deficient strains resistant against ampicillin and kanamycin and bacteria with a functional adenylate cyclase also resistant against ampicillin and kanamycin. We would need to characterize the bacteria on the plate further and repeat the experiment to find the reason for the results. Due to a lack of time we couldn’t perform these experiments. It remains unknown, if the BACTH system fulfills all requirements needed to be used in the Concave toolbox. The BACTH engineering cycle thereby remains open.