Build: DARPin cloning

Two elements are required to detect the IgE responsible for an allergy: strain A displaying the required allergen and strain D displaying the DARPin, a synthetic protein able to bind to the constant part of IgE. If the patient’s blood contains IgE against the tested allergen, the strains aggregate as a means for detection. The first step consisted in building a strain displaying a fusion between OmpA, the DARPin, and a fluorescent reporter (sfGFP)(Fig.2). We wanted to verify the expression and check the subcellular localization of the fusion, expecting the DARPin and sfGFP parts to be displayed on the outer surface of the bacterium.

Figure 2: DARPin-sfGFP construction. The sfGFP served as a marker to prove the correct expression of the protein and to study its subcellular localization.

The OmpA-TEV-DARPin-sfGFP-6xHis-tag encoding gene was successfully assembled in a plasmid under the control of the strong inducible pT7 promoter. The plasmid was then transformed into E. coli Tuner (DE3) to identify suitable conditions for expression. Despite high cytosolic concentrations of the OmpA-TEV-DARPin-sfGFP-6xHis-tag fusion, we could confirm the presence of our chimeric protein in the outer membrane. The sfGFP encoding part was then removed from the plasmid by PCR to use the OmpA-TEV-DARPin for IgE detection, creating strain D.

Test: First aggregation assay

We conducted aggregation assays after constructing the required strains A and D. The idea was to mix strains A and D with a recombinant IgE against the allergen displayed by strains A. Our experiments to check aggregation were designed to test different conditions: the concentration of IgE, the agitation and the incubation temperature. One condition initially showed aggregation in presence of IgE, but we did not manage to repeat the results multiple times independently. In the end, we concluded that there was no difference between the sample containing all the elements required for aggregation (strains A, D and IgE) compared to the negative controls (e.g. strains A and D without any IgE). Furthermore, other results obtained showed no difference between our assays and the negative control. These results led us to assess the correct folding of the construction.

Learn: Structural study of the strain D construction

After our first aggregation attempts failed, we wondered whether strain D behaved as initially thought during the design part. More precisely, we wanted to investigate whether the OmpA-TEV-DARPin(-sfGFP-6xHis-tag) fusion would fold properly to display the DARPin without altering its secondary structure. An AlphaFold 2 simulation was done to build a structural model of the OmpA-TEV-DARPin-sfGFP-6xHis-tag fusion (Figure 3).

Figure 3: AlphaFold 2 prediction of the OmpA-TEV-DARPin-sfGFP-6xHis-tag structure.The construction prediction was recolored with PyMol. The OmpA is shown in orange, in yellow the DARPin, and in green the sfGFP construction.

At first, the complex looked correctly folded. We also checked a previously published DARPin structure obtained by crystallography for comparison (Figure 4).

Figure 4: DARPin structure extracted from the crystal structure 4GRG (Kim et al. 2012).The extraction of the DARPin structure was made on PyMol.

We noticed that the AlphaFold 2 model displayed 6 α-helices in the DARPin while the crystallographic structure presented 8 α-helices. This was confirmed by the analysis of another X-ray crystallographic structure of DARPin (PDB: 7MXI). We immediately checked the amino sequence of the DARPin we used on PDB (https://www.uniprot.org/uniprotkb/L7MTK7/history). We realized that the sequence was incomplete and inappropriately referenced so our genetic construction was possibly wrong. This experience made us realize that the sequences published on databases are not always correct and that we have to double-check every piece of information we use for our project. We replaced the partial DARPin sequence with the full one in our OmpA-TEV-DARPin-sfGFP-6xHis-tag fusion and rebuilt a structural model with AlphaFold 2 (Figure 5).

Figure 5: Reconstruction of the OmpA-TEV-DARPin-sfGFP-6xHis-tag fusion protein with AlphaFold 2. The construction prediction was recolored with PyMol. The OmpA is shown in orange, in yellow the previous DARPin with in clear green the correction of the DARPin sequence, and in green the sfGFP construction.

The AlphaFold 2 model of the corrected construct showed the expected structure and a proper folding of each element, including the DARPin with its 8 α-helices. We therefore decided to build a corrected version of the plasmid, with the full-length DARPin, before trying again the aggregation assays.

Design: DARPin reconstruction

To close the DBTL loop, we had to design a strategy for rebuilding the DARPin. Due to strong time constraints, we chose to use a Polymerase Chain Assembly (PCA) strategy to add the missing sequence. This consists in using three 60 bp primers (1_PCA60, 2_PCA60 and 3_PCA60) to build the missing sequence into two steps. First, 1_PCA60 and 2_PCA60 have to be assembled and elongated by PCR. Then, the hybridization and the elongation of 3_PCA60 should enable us to obtain the complete missing sequence (Figure 6).

Figure 6: Principle of PCA to build the missing part of the DARPin. Three primers 1_PCA60, 2_PCA60 and 3_PCA60 were used to build the complete missing sequence. These three primers were mixed on the same PCR mix, after a first elongation with 1_PCA60 and 2_PCA60, 3_PCA60 binds with the 2_PCA60 reverse to end up the elongation.

Build: Corrected version of the DARPin

The missing part of the gene coding for DARPin was amplified by PCA and inserted at the same time on the OmpA-TEV-DARPin-sfGFP-6xHis-tag and OmpA-TEV-DARPin-6xHis-tag plasmids by In-Fusion. Using the corrected and right plasmids, we were able to repeat aggregation assays less than one week after we realized the initial design flaw.

Test: New aggregation assays

We followed the same protocol as previously described to conduct new aggregation experiments, with the complete sequence of the DARPin this time. Again, different conditions were tested: the IgE concentration, the temperature of incubation and the temperature. Sadly, we could not observe any aggregation in any of the tested conditions.

Learn: Results of new aggregation assays

The aggregation tests did not yield any result corresponding to our expectations but brought a lot of relevant information for further troubleshooting. First, we performed better aggregation experiments allowing us to reduce non-specific aggregation with a cleaner negative control sample, and to have a good dispersion of bacteria (Figures 7 and 8). We acquired some “know-how” about this protocol.

Figure 7: aggregation assay before correction of the DARPin with strains A and D mixed with different concentrations of anti-Ara h 2 IgE: 1:100 (A), 1:250 (B), and 1:1000 (C). This experiment was conducted before the DARPin analysis by AlphaFold 2 and sequence correction.

Figure 8: aggregation assay with the DARPin corrected with strains A and D mixed with different concentrations of anti-Ara h 2 IgE: 1:100 (A), 1:250 (B), and 1:1000 (C). The same mix with strains A and D without IgE was used as negative control (D). The complete DARPin on the strain D was used.

The difference between observations shows the improvement of our aggregation assay design and a better experience of manipulation in the wet lab. This demonstrates that our team can design, build, test, and improve by learning from our errors.

Design : Integrating modeling and surface controls design to improve our idea

Despite the correction of the DARPin DNA sequence, the aggregation experiments did not work as expected. We had to question ourselves one more time to try to understand what could impede the aggregation of the bacteria. In parallel to these experiments, our modeling efforts allowed us to formulate additional hypotheses. The complexation model showed that the ideal cell concentrations and number of proteins exposed on the surface depends intimately on the affinity of the IgE for the corresponding allergen. The complexation model can now be used to guide further experimental designs as shown on Figure 9. Modeling has brought us a great wealth of information to improve aggregation assays by directing the set-up.

Figure 9: Evolution of the aggregation potential for different initial concentrations of A and D free molecules, for the allergen Ara h 2 (KD=1*10-11 M) tested during aggregation assays for 100 000 displays of allergen and DARPin on each strain. The A and D molecule concentrations range from 1e-14 M to 1e-7 M. Knowing the number of DARPin and allergens molecules displayed on the surface of each strain, concentrations of each strain can be estimated for aggregation assays.

Here the complexation model indicates the concentration of DARPin and allergens molecules necessary for aggregation. So to obtain the concentration of strains A and D, the number of DARPin and Allergen molecules displayed on the surface of each strain are needed. That is why membrane proteins quantification experiments have to be performed. These experiments involve checking the presence of DARPin or allergens on the membrane and measuring its concentration by protein extraction. First experiments have pointed out the presence of our OmpA-TEV-DARPin-sfGFP-6xHis-tag fusion protein on both the membrane and the cytoplasm. This result is hopeful and allows to deepen the design to show if the DARPin is correctly displayed on the exterior of the bacterial membrane. Then this strategy can be applied to our allergens strains. Due to the lack of time, we could not continue further with the DBTL cycle despite some promising new ideas for improvement. News tools such as modeling, expression regulation, and the study of proteins on the membrane, will bring together more relevant information to take our project to the proof of functionality.

Design: choice of fluorescent protein

We needed for our FACS experiments to express fluorescence in our strains A and D. Manon Barthe from TWB has provided the mTagBFP for showing blue fluorescence for strain D, and mRFP1 showing red fluorescence for strain A (Figure 10). The sequence coding for both reporters was under the control of the same constitutive ihfB800 promoter. While the blue fluorescence of the mTagBFP was successfully shown with only one DBTL loop, the strain with the red fluorescence has required two loops as explained hereafter.

Figure 10: allergen construction with mScarlet-I, example for Ara h 2 allergen. Promoter used for the fluorescent protein is the constitutive ihfB800.

Build: mRFP cloning

The gene coding for mRFP1 was amplified by PCR and inserted at the same time on three plasmids (OmpA-TEV-Gal d 2-6xHis-tag, OmpA-TEV-Ara h 2-6xHis-tag, and OmpA-TEV-Ana o 3-6xHis-tag) by In-Fusion. Digestions by restriction enzymes confirmed the correct addition of the gene to the three plasmids (Figure 11). They were finally transformed into E. coli Tuner (DE3) cells to form three different strains A, each with a different allergen.

Figure 11: digestion by NotI of Ara h 2 with mRFP1 insertion (A), of Ana o 3 with mRFP1 insertion(B) and of Gal d 2 with mRFP1 insertion (C). The digestion product was checked with 0.6% agarose electrophoresis gel and revealed with EtBr. A theoretical gel is presented with each gel and the NEB 1 kb DNA ladder is used for the experimental gels (note that a different ladder is presented on the theoretical gel). Colonies 1, 2 and 3 present the correct size for Ara h 2; colonies 21 and 23 present the correct size for Ana o 3; colonies 7 and 13 present the correct size for Gal d 2.

Test: Microscope observation of fluorescence

Expression of mRFP1 was assessed on an epifluorescence microscope. There was no fluorescence in either of the strains nor in the negative control (data not shown). In parallel, fluorescence of mTagBFP was successfully observed in strain D. The original plasmid containing the mRFP1 and our plasmids with the mRFP1 were sent to sequencing.

Learn: Conclusions on protein expression and possible cause of the problem

As no fluorescence of mRFP1 was detected, the protein was certainly not expressed in strains A. As fluorescence of mTagBFP in strain D was observed, the ihfB800 promoter used for both protein expressions was surely not involved in the failure of mRFP1 expression. Moreover, sequencing of the mRFP1 gene showed some mutations, so we hypothesized that the initial mRFP1 sequence could be incorrect. We therefore decided to choose another fluorescent protein for our strains A.

Design: Replacement of mRFP1 by mScarlet-I fluorescent protein

We contacted Maxence Holtz (former iGEM Toulouse member and intern at Toulouse White Biotechnology). He advised us to use his mScarlet-I protein, as it was red like the mRFP1 and was supposed to be expressed correctly in E. coli cells. We then decided to switch the mRFP1 gene for the mScarlet-I gene, while keeping in place the constitutive ihfB800 promoter since it was efficient for the mTagBFP. The final construction for Ara h 2 is shown on Figure 12.

Figure 12: allergen construction with mScarlet-I, example for Ara h 2 allergen. Promoter used for the fluorescent protein is the constitutive ihfB800.

Build: mScarlet-I cloning

The gene coding for mScarlet-I was amplified by PCR and inserted instead of mRFP1 in the plasmids containing OmpA-TEV-Gal d 2-6xHis-tag, OmpA-TEV-Ara h 2-6xHis-tag, and OmpA-TEV-Ana o 3-6xHis-tag by In-Fusion, still under the ihfB800 promoter control. Digestions by restriction enzymes confirmed the correct addition of the gene for the three plasmids. They were first transformed into Stellar competent cells in order to amplify them. They were finally transformed into E. coli Tuner (DE3) cells to form three different strains A, each with a different allergen.

Test: macroscopic observation of fluorescence

As shown on Figure 13, red fluorescence was observed with the naked eye as soon as Stellar cells were plated on Petri dishes, giving a strong first idea of the correct expression of the mScarlet-I.

Figure 13: Second plating of pET-21 b (+)_Ara h 2_OmpA_mScarlet-I transformed cells and pET-21 b (+)_Ana o 3_OmpA_mScarlet-I transformed Stellar cells. The colonies were selected on LB-ampicillin plates. Pink coloring indicates the expression of the mScarlet-I.

After transformation into E. coli Tuner cells, expression of mScarlet-I was also assessed on an epifluorescence microscope. Vivid red fluorescence was observed while the negative control was not fluorescent at all (see Figure 14 for the strain A displaying Ana o 3).

Figure 14: fluorescence of pET-21 b (+)_Ana o 3_OmpA_mScarlet-I transformed Tuner cells. Cells transformed with plasmids not containing mScarlet-I were used as negative controls. Observations were made at the microscope parameters for mScarlet-I.

Learn: Conclusion on protein expression

As fluorescence was observed the mScarlet-I protein was expressed successfully in our three strains A. It confirmed that the ihfB800 promoter was indeed functional and that the mRFP1 was surely defective. Having strains with functional fluorescence allowed us to start aggregation experiments with FACS.