Usage and Biology

Barbie1 design

Starting from the BaCBM2 structure model generated by the AlphaFold2 software, we performed docking assays with six types of plastic: polypropylene (PP), polyethylene (PE), polyethylene terephthalate (PET), nylon (NY), polyvinyl chloride (PVC) and polystirene (PS). We made the docking using Gnina software with relaxed parameters to screen many proteins and features for plastic affinity.

Thereafter, the produced overlaps were removed by the docking assays using the ChimeraX software, as well as used for visualization and sequence manipulation. A reverse folding was then performed with the protein output from the docking using the LigandMPNN tool. The original protein set generated from the docking was filtered to maintain just unique positions, considering the associated score , without overlap between them.

By doing that, 6.000 sequences were generated for each ligand, totalizing 6 plastics x 6.000 sequences = 36.000 sequences, as illustrated in Figure 1. The consensus sequence from the 36000 sequence originated our most optimized protein sequence sensitive for several plastics types was named as Barbie1!

Figure 1. Protein-ligand docking representation of the Barbie1 protein docked with PP, PE, PET, NY, PVC.

The BARBIE1 protein was redocked with the same plastics as before, once more using Gnina software. The result comparing its affinity with BaCBM2 in silico assays performed are described in the barplot of Figure 5. Comparing the predicted affinity between the original and the modified protein, it is notable a substantial increase for all plastics, in particular for PE, PP, and PS, highlighting the effectiveness of the processing pipeline.

Besides the monomers tests, we also wanted to test the affinity using different sizes of plastic in order to guarantee that this could be a valuable parameter to future analysis and experiments. Therefore, the tested ligands were PE and PET plastics with 50 and 25 repeating units, respectively. As a result, the previous behavior at maintaining a higher KD for Barbie1 when compared to BaCBM2 was preserved.

Figure 2. AlphaFold 3 3D simulation of Barbie1 protein with miRFP and three Mad10 tags.

miRFP670

miRFP670 is a monomeric near-infrared fluorescent protein (NIR FP) derived from the bacterial photoreceptor Rhodopseudomonas palustris. It exhibits excitation and emission maxima at 643 nm and 670 nm, respectively, making it suitable for applications in multicolor microscopy. With a molecular weight of approximately 34.5 kDa, miRFP670 is characterized by its high brightness and photostability in various cells, outperforming many other NIR FPs in terms of fluorescence intensity. The protein utilizes biliverdin as its chromophore, which enhances its effective brightness and allows for efficient labeling in live-cell imaging. The miRFP670 demonstrates low cytotoxicity and stability across a range of pH levels. Its structural properties enable the development of biosensors and facilitate protein-protein interaction studies.

Mad10 peptide

Mad10 (Magnetosome-associated deep-branching 10) is a protein derived from the magnetotactic bacterium Desulfamplus magnetovallimortis BW-1, known for its ability to synthesize magnetite crystals, which are essential for its navigation in aquatic environments. This protein is important for the formation and stabilization of magnetosomes—organelles that contain magnetic particles.

The Mad10 tag refers to a peptide derived from Mad10, designed to include the most conserved amino acids within the protein sequence. The Mad10-derived peptide exhibits a strong affinity for magnetite, allowing it to effectively bind to magnetic columns during purification processes.

Protein Expression and Purification

The sequence for this part was chemically synthesized by Genscript and arrived pre-cloned in the pET-15b(+) vector. This vector includes a T7 promoter, lacI, a ribosome binding site (RBS), a 6xHis tag, and a T7 terminator. After the arrival, we proceeded with transforming the plasmid into the Escherichia coli BL21(DE3) strain.

Optimization of protein expression

We inoculated E. coli cells expressing Barbie1-RFP-3xMad10 into LB medium with ampicillin, incubating them at 37°C until an OD600 of 0.5 was reached. At this point, 1 mM of IPTG was added to each culture to induce protein expression, and the cultures were left shaking at 37°C for different durations: 3, 4, and 5 hours.

After harvesting the pellets, we extracted the proteins using a protein extraction buffer and sonication, followed by protein quantification using the BSA assay. The samples were then prepared for SDS-PAGE by denaturing them at 95°C and loaded into the gel for electrophoresis.

The electrophoresis results showed a clear and significant difference in protein expression when comparing the IPTG-induced samples (3, 4, and 5 hours) to the control at 0 hours, which had no visible expression bands. However, when comparing the 3, 4, and 5-hour induction times to each other, there was no substantial visual difference in the intensity of the bands, indicating that protein expression was already strong at the 3-hour mark and remained consistent through the later induction times.

Figura 3. Analysis of protein expression of Barbie1-RFP-3xMad10 over different induction times (0h, 3h, 4h, 5h) via SDS-PAGE. Lanes: 1- Ladder (molecular weight marker), 2- Barbie1-RFP-3xMad10 (0h), 3- Barbie1-RFP-3xMad10 (3h), 4- Barbie1-RFP-3xMad10 (4h), 5- Barbie1-RFP-3xMad10 (5h), 6- BLS1 (untransformed control).

Final expression and purification

During the sample preparation, we observed a high level of viscosity, which complicated the process. To address this, we centrifuged the sample multiple times and filtered it twice before loading it onto the equipment. Unfortunately, a significant portion of the sample was lost during this additional processing. Despite these challenges, the chromatogram still showed a clear peak corresponding to Barbie1-RFP-Mad10.

After purification, we sought to understand the cause of the excessive viscosity in our sample. To investigate, we performed a computational analysis using Aggrescan4D (A4D) to predict the aggregation propensities of the protein fold state. The results indicate that Barbie1-RFP-3xMad10 exhibits a higher aggregation propensity compared to BaCBM2-RFP-3xMad10. Figure 4 shows the scores for each residue in both proteins, where residues with negative values are considered "soluble residues", and those with positive values are classified as "aggregation-prone residues". The difference between the two proteins is particularly noticeable in the initial residues, corresponding to BaCBM2 and Barbie1.

Figure 4. Scores for each residue in both proteins (BaCBM2-RFP-3xMad10 and Barbie1-RFP-3xMad10). Residues with negative values are identified as "soluble residues", while those with positive values are considered "aggregation-prone residues”.

Based on this finding, we decided to perform new rounds of expression and purification, this time incorporating detergents to improve protein solubilization. In subsequent cycles, we adapted our strategy to account for the aggregation tendency of Barbie1-RFP-Mad10, aiming to minimize this characteristic and improve the purification process.

Figure 5. Chromatogram of Barbie1-RFP-Mad10 purification using IMAC (Immobilized Metal Affinity Chromatography) on a Ni-column. The peak corresponding to the BARBIE1-RFP-Mad10 protein is highlighted, indicating successful, although partial, purification.

Characterization

SEC-MALS

After purifying Barbie1-RFP-3xMad10, SDS-PAGE analysis revealed that the sample was eluted along with other E. coli proteins. To better characterize the sample and improve its purity, a 30 kDa concentrator was used to separate the larger proteins from the smaller ones. With the sample more concentrated, we proceeded with SEC-MALS analysis. Unlike the results observed for BaCBM2-RFP-3xMad10, the SEC-MALS outcome for Barbie1-RFP-3xMad10 displayed three distinct peaks in the curve (Figure X). However, none of the three peaks had an estimated molecular mass consistent with the expected mass of Barbie1-RFP-3xMad10 (46.47 kDa).

As noted earlier, during sample preparation for purification, we detected the formation of large aggregates. Additionally, the aggregation score for Barbie1-RFP-3xMad10, calculated by Aggrescan4D, is significantly higher than that of BaCBM2-RFP-3xMad10. Considering the SEC-MALS results along with the SDS-PAGE findings, which confirmed the presence of Barbie1-RFP-3xMad10 in the collected sample, we believe that the peak with a molecular mass of 157.7 kDa corresponds to a protein aggregate.

Figure 6. SEC-MALS result of 6His-BARBIE1-RFP-3xMAD10.

Circular Dichroism (CD)

CD can provide crucial insights due to the presence of beta sheets in their structures. Typically, β-sheets exhibit a characteristic negative band around 218 nanometers (nm) and a positive band near 195 nm, while α-helices show a negative band around 222 nm and a positive band around 190 nm. However, a limitation of the equipment used restricts the analyzed wavelength range to between 200 and 250 nm.

The pipeline acquisition of the spectrum was followed by a 2 seconds of integration for each point in each 6 spectrum calculated at room temperature (20 ºC). From the collected data, it is calculated its average, subtracted from the water absorbance, normalized, and smoothed using a Savitzky-Golay filter.

Figura 7. CD acquisition of Barbie1-RFP-3xMad10 protein at 20 ºC.

Barbie1-RFP-3xMad10 Heating Ramp

An essential evaluation for the protein is the heating ramp, in which we can evaluate how the protein acts in different temperatures and their melting point. For doing so, the same procedure was followed, in which each protein was heated from 20 to 95ºC, as shown in Figure 8. The data acquisition was made in steps of 5ºC.

The CD absorbance is shown as the figure’s color map. It is evident that the protein show absorbance in the secondary structure regions, indicating a thermal resistivity.

Figura 8. Heating ramp for Barbie1-RFP-3xMad10.

Barbie1-RFP-3xMad10 Pontual Spectra

With this result, it is interesting to observe that, even after the protein engineering, the protein could still maintain great results for tests like the described. To further explore it, we also ran a protein pontual spectra, in which we measured CD absorbance for only 215 nanometers, but each measure made at 1ºC increase.

Figure 9 shows the result of the experiment, which emphasizes the decrease in the CD absorbance for Barbie1-RFP-3xMad10 as already expected from Figure 8. Although Barbie1-RFP-3xMad10 loses part of its absorbance with the increase of temperature, it is not possible to confirm its denaturation. A viable hypothesis is some reduction in the beta-sheets structures.

Figure 9. Pontual spectra for Barbie1-RFP-3xMad10 at 215 nm.

In the first instant, this behavior might show a bad characteristic of Barbie1-RFP-3xMad10. However, the loss of function at high temperatures can be used to properly design the water filter. Once saturated with plastics, the system can be heated to high temperatures making Barbie1-RFP-3xMad10 lose part of its structure. Thus, knowing the function is correlated to the form, the protein might lose its plastic affinity, allowing the removal of microplastics and further processing for reutilizing it.

Heating and Cooling Barbie1-RFP-3xMad10 Ramp

With this possibility in our sights, the following experiment is heating and cooling the protein, in a way we can observe how its structures will behave. Illustrated in Figure 10, the plot was rotated for better visualizing the CD absorbance in function of temperature variation. From left to right, it is possible to firstly see how Barbie1-RFP-3xMad10behaves when heated and then cooled.

As a result, protein’s CD absorbance in the cooling ramp still presented a signal, indicating it maintained its structure and a consequently renovelation of the possible denoveled parts.

Figure 10. Heating and cooling ramp CD absorbance of Barbie1-RFP-3xMad10.

Specific Wavelength Analysis

If we deepen our analysis into 210 and 222 nm, which are in the region of secondary structure absorbance, we can better understand the variancy. In Figure 5, the above graphs are related to the 210 nm wavelength, whereas the graphs from below to the 222 nm.

In a dashed gray line, an average initial absorbance for Barbie1-RFP-3xMad10 was added to highlight the loss of the original absorbance. Additionally, both wavelengths show an interesting behavior, that is an important trend change between 55 and 60 ºC, which may be a critical point for the protein. From the cooling graphs, it is notable how the protein loses part of the original signal, indicating a conformational alteration from the temperature change.

For future adaptation of Barbie1-RFP-3xMad10, it is fundamental to assess and design the protein in a way it can be further reused for multiple filtration.

Figure 11. Two specific wavelengths, 210 nm above and the 222 nm below, plots for the heating and cooling ramp for Barbie1-RFP-3xMad10.

Barbie1-RFP-3xMad10 Melting Temperature

From the previous result of analyzing a specific wavelength, it is possible to note the curve’s behavior, which is usual for a protein denaturation. Thus, a viable way to uncover the protein’s melting temperature is by fitting the Boltzmann curve in the collected data, which is given by:

$$A2 + (A1 - A2) / (1 + np.exp((x - x0) / dx)),$$

Where:

A1 is the protein’s native state;
A2 is the completely unfolded protein’s state;
Tm is the protein’s melting temperature;
k is a constant that represents the curve inclination.

After fitting the curve in Figure 12, the melting point found for Barbie1-RFP-3xMad10 is approximately 58.81ºC. This information is fundamental for building the filter, since it represents stability in a higher temperature. In particular, it is known that in water purifiers the temperatures do not exceed 30 to 35ºC, making the protein appropriate to the system.

Figure 12. Boltzmann fit in the absorbance values of Barbie1-RFP-3xMad10 in a heating ramp.

Plastic Interaction Analysis

Through secondary structures absorbance, it was possible to observe that Barbie1-RFP-3xMad10 structure do not change in the presence of plastic nanoparticles

Finally, the last experimental evaluation at CD was with the polystyrene nanoplastics. We used a 100 nanometer particle which was inserted into a solution with Barbie1-RFP-3xMad10.

Nevertheless, as reported from the computational simulations, we do not expect any conformational change. Since CD is only evaluating the absorbance in secondary structures, there may not happen major differences in the experiment.

Consequently, Figure 13 shows as projected. When the Barbie1-RFP-3xMad10a spectra is plotted with the nanoparticle solution spectra, minor changes can be observed. Specifically, this contrast might be a result of the polystyrene nanoparticle scattering, creating noise in the measurement.

Figure 13. Plastic binding proteins Barbie1-RFP-3xMad10and BaCBM2-RFP-3xMad10 spectra acquisition with 100 nm polystyrene nanoparticles.

Considering we were not able to assess the plastic-protein binding, further studies might be done. One important form to evaluate this is the infrared spectroscopy, which allows studying the vibrational frequencies of molecules.

Sequence and Features

Assembly Compatibility:

10
COMPATIBLE WITH RFC[10]
12
COMPATIBLE WITH RFC[12]
21
INCOMPATIBLE WITH RFC[21]
Illegal BglII site found at 594
23
COMPATIBLE WITH RFC[23]
25
INCOMPATIBLE WITH RFC[25]
Illegal AgeI site found at 88
1000
COMPATIBLE WITH RFC[1000]

Part:BBa_K5396001

Contents