DNA shuffling to improve crude-water interfacial activity in biosurfactants with OmpA protein of Escherichia coli

Strains and growth conditions

E. coli K-12 W3110/pCA24N OmpA⁺ was obtained from the ASKA library (Kitagawa et al., 2005). It was grown overnight at 37 °C on Lysogeny Broth (LB) agar plates containing 50 µg/mL chloramphenicol. The strain was grown from single colonies in 50 mL of LB medium supplemented with 50 µg/mL chloramphenicol and incubated overnight at 37 °C at 250 rpm.

DNA shuffling

The ompA gene was amplified using external primers (Table 1) that annealed 150 base pairs upstream and downstream of the target gene (Fig. S1). The reaction contained: 1 µl pCA24N OmpA⁺ template (102 ng), primers EF and ER (0.3 µM each), 5 µl 10X Buffer (Mg²⁺ Plus), 4 µl dNTP Mixture (2.5 mM each), and 5 U of Takara Taq (Takara Bio., Shiga, Japan), for a total volume of 50 µl. PCR using iCycler (Bio-Rad, Hercules, CA, USA) was performed at 94 °C for 4 min, with 35 cycles of 94 °C for 45 s; 55 °C for 30 s; 72 °C for 90 s, and a final extension of 72 °C for ten minutes. 2 µg of the resultant band (1,439 bp) was fragmented randomly with DNase I. The fragmentation was performed in a 50 µl reaction mixture containing 5 µl of 10X buffer (1 M Tris-HCl pH 7.4 and 200 mM MnCl₂) and 0.05 U of DNase I. The reaction was incubated for 30 s at 80 °C before the incubation was terminated after 12 min. Following electrophoresis on 1% agarose, fragments between 400 bp and 1,000 bp were purified according to the protocol by the PureLink^® Quick Gel Extraction Kit (Invitrogen, Carlsbad, CA, USA). A mixture with 40 µl of purified fragments, 5 ul of 10X Buffer (Mg⁺² Plus), 2 µl of dNTPs, and 5 U of Takara Taq (Takara Bio., Shiga, Japan) was prepared. PCR was performed at 96 °C for 90 s, with 35 cycles of 94 °C for 30 s; 65 °C for 90 s; 62 °C for 90 s; 59 °C for 90 s; 56 °C for 90 s; 53 °C for 90 s; 50 °C for 90 s; 47 °C for 90 s; 72 °C for 4 min, and a final extension of 72 °C for 7 min.

Using 200 ng from the resulting PCR product, 5 µl 10X Buffer (Mg²⁺ Plus), 4 µl dNTP Mixture (2.5 mM each), 0.15 µl of each primer IF and IR (Table 1), 5 U of Takara Taq (Takara Bio., Shiga, Japan) and 3 µl of DMSO in a total volume of 50 µl, to achieve punctual mutations by mismatched base-pairing. Final PCR with inner primers (IF and IR) is conducted at 94 °C for 30 s; 25 cycles of 98 °C for 10 s; 65 °C for 30 s; 72 °C for 150 s, and final elongation at 72 °C for 7 min. The two randomized resulting bands (1,100 bp and 300 bp) were cloned into pCR4-TOPO^® and transferred to competent E. coli DH5α. 50 µg/ml ampicillin was used to select clones for library construction.

Molecular techniques

Plasmid DNA was extracted utilizing the Spin Miniprep Kit (QIAGEN, Inc., Valencia, CA, USA). DNA fragments were isolated from agarose gels using PureLink^® Quick Gel Extraction Kit (Invitrogen, Waltham, MA, USA). The recombination reaction was carried out at 22 °C using TOPO TA^® Cloning^® Kit (Invitrogen, Waltham, MA, USA). In this study, the NanoDrop^® ND-1000 Thermo Scientific (Waltham, MA, USA) was used to quantify DNA.

Screening of biosurfactant activity using 96-well plates

210 µl of LB medium with ampicillin (50 µg/ml) were added to 96-microwell plates, inoculating 10 µl of previously grown bacterial culture for each clone. The plates were incubated at 37 °C overnight. After 20 min of centrifugation at 4,500 rpm, the plate was analyzed using a paper backing sheet with a grid. According to Walter, Syldatk & Hausmann (2010), the clones presenting distorted-looking wells were selected as biosurfactant producers. Tween 20 and P. aeruginosa rhamnolipid were utilized as positive controls (Sajna et al., 2015).

Interfacial tension measurement

For each clone and the purified proteins, 5 ml of LB medium was inoculated with single colonies previously grown in LB agar plates with ampicillin (50 µg/ml). When OD₆₀₀ reached 0.6, IPTG (isopropyl-b-D-thiogalactopyranoside) was added to a final concentration of 0.5 mM, and incubation at 250 rpm was continued for 4 h at 37 °C for induction of protein expression. Cells were harvested by centrifugation at 4 °C and 3,500 rpm for 15 min. Cells were lysed in a Sonicator VC-750 (SONICS, Newtons, CT, USA) as follows: amplitude 20% and with the pulse set to 1 s × 1 s for 2 min and then kept on ice for 10 min (three times). The supernatant was recovered after centrifugation at 4 °C and 13,000 rpm for 15 min. Each protein extract was added to 2 ml of diluted crude oil to a final concentration of 0.055% (v/v) and homogenized via sonication at 19% amplitude and pulse set to 1 s × 1 s for 2 min. The light phase was prepared by diluting crude oil (12° API) with dodecane to a final concentration of 75% (v/v) and mixed using Dissolver DISPERMAT^® LC (VMA-GETZMANN GMBH, Hilden, Germany) at 2,500 rpm for 45 min. Interfacial tension was quantified using the Attension Theta Tensiometer (Biolin Scientific, Gothenburg, Sweden). The technique selected was the optical method of the inverted pendant drop, with the drop profile fitting method of Young-Laplace (Berry et al., 2015). The image scan was obtained at a rate of 20 frames/s for 5 min. Table 2 lists the properties of the crude oil used in this work.

DNA sequencing

Inserts were amplified by combining 2.5 µl of 10X Buffer, 1.5 µl of MgCl, 4 µl of dNTPs, 1 µl of each M13 F and M13 R primers (Table 1), and 5 U of Takara Taq (Takara Bio., Shiga, Japan), to a total reaction volume of 25 µl. The reaction took place using the thermocycler iCycler (Bio-Rad, Hercules, CA, USA) at 95 °C for 4 min, 25 cycles of 95 °C for 30 s; 54 °C for 30 s; 72 °C for 90 s; and final elongation of 72 °C for 5 min. Fragments were sequenced at the Universidad de los Andes by Sanger sequencing method. Alignment of all the sequences against OmpA was done using Clustal Omega (McWilliam et al., 2013).

3D structure prediction and protein analysis

The 3D structures of each protein were acquired from ColabFold v1.5.3: AlphaFold2 using MMseqs2 (Mirdita et al., 2022). The top-ranked model structure underwent relaxation using Amber. msa_mode was configured as mmseqs2_uniref_env, and the pair_mode was set to unpaired_paired. num_recycles setting was adjusted to 24, while the remaining options were kept at their default settings. The stability index (Guruprasad, Reddy & Pandit, 1990) and solubility values for the nine proteins were predicted using the Protparam tool (Gasteiger et al., 2005) and Protein-Sol (Hebditch et al., 2017) servers, respectively.

Protein purification by immobilized metal affinity chromatography

First, clone 6 and 12 sequences were synthesized individually in plasmid puc57 in Shanghai Shine Gene^R Molecular Biotech, then they were digested and ligated in pET6xHN-N (Fig. S2). For each selected clone, 50 ml of LB medium was inoculated with 100 µg/ml of ampicillin from single colonies previously grown on LB agar plates, and allowed to incubate overnight. With 100 mL of LB containing an initial OD of 0.2, a new culture was started for every mutation. When the OD600 reached 0.6, induction with isopropyl-β-D-thiogalactopyranoside (IPTG) at a concentration of 1.5 mM was initiated, and incubation was continued for 5 h at 37 °C. Cells were collected by centrifugation at 4 °C and 4,000 rpm for 20 min. Cell lysis was performed with a tip sonicator as follows: amplitude 37%, for 15 min with pulses of 20 s on and 40 s off. The supernatant was collected after centrifuging at 4 °C and 4,000 rpm for 20 min. The supernatant was purified with cobalt columns (Cobalt chelating resin, 1 mL spin Column, G-Biosciences, St. Louis, MO, USA). The column was equilibrated with Cobalt equilibration buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole pH 7.4), and then the supernatant of the mutations was added to separate columns. Cobalt buffer was added once more, and three washes were collected. To detach the target protein from the resin, heparin wash buffer (50 mM Na₂HPO₄, 300 mM NaCl, 250 mM imidazole, pH 8.0) was utilized. For each mutation, four elusions were collected. The purification of mutations was confirmed by tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis (tricine-SDS-PAGE) (Bio-Rad Labs, 2022). The collected fractions were mixed with sample buffer and denatured for 5 min at 70 °C. Electrophoresis was run for 65 min at a voltage of 140V. The gel was left in a fixing solution (glutaraldehyde 5%), stained with Coomassie blue, and the protein bands were visualized by adding a staining solution.

DNA Shuffling for ompA

To accomplish DNA shuffling (23), the strain E. coli K-12 W3110/pCA24N OmpA⁺ was grown on LB agar plates with 50 µg/ml chloramphenicol, and purification of the plasmid pCA24N was carried out (Winstanley & Rapley, 2000). Amplification of the ompA gene was successfully achieved using EF and ER as external primers (Table 1, Fig. S4). Random fragmentation was conducted by evaluating six different time reactions 2 min, 30 s, 60 s, 90 s, 10 s, 5 s to analyze the fragments’ sizes after different times of incubation. Furthermore, the reassembly reaction without primers was carried out with fragments between 400 bp and 1,000 bp. As expected, the resulting fragments obtained are larger than the original gene (Figs. S5 and S6). Final amplification was carried out with dimethylsulfoxide (DMSO) and external primers IF and IR primers (Table 1). Results showed two different fragments, one with a size similar to the original gene, and the other with approximately 400 bp. We previously showed that the biosurfactant performance is possibly associated with its size, mostly affecting the formation of micelles (Segura et al., 2014a). Cultures with noticeable growth were twice plated on LB agar to maintain the concentration of antibiotics in the medium and prevent the formation of satellite colonies. The continued growth of 25 clones was detected, and the library was constructed in glycerol stocks and stored at −80 °C.

Initial screening of the IFT ability of the engineered OmpA proteins

The 96-well plates with the supernatants of each clone were observed through a grid. The main goal of this test is to identify the grid distortion when viewed through an aqueous medium containing a surfactant. The results of these tests are mainly qualitative and a distortion of the grid for all clones is observed (Fig. S7), showing that all clones still conserved IFT activity. Then, IFT was also evaluated in the Attension^® Theta Tensiometer.

Alignment of sequences and comparison with OmpA

Among the clones studied, nine of them were selected according to the obtained IFT values and consequently sequenced based on the Sanger method on the fact that they exhibited three behaviors: lower, equal, and higher IFT compared to the parental OmpA protein. To identify the protein mutations, we used Clustal Omega EMBOSS Needle (Huang et al., 2014). The pairwise sequence alignments show different parts of the original sequence that are conserved in the fragments obtained and some other parts that are deleted around the positions 350 to 420 (of the original sequence) are lost possibly due to the presence of DMSO, conditions of amplification and long time exposure to DNAse. These results show that random mutations are present throughout the original gene (Supporting Material- Pair-wise Sequence Alignments). Departing from the Kyte-Dolittle plots and molecular dynamics to calculate the free energy change when inserting molecules in water-dodecane interfaces, our group previously designed surfactants derived from OmpA (Aguilera-Segura et al., 2016). However, we did not find similarities between those sequences and the ones derived from the DNA shuffling so both approaches could provide new structures with enhanced functionality.

Analysis of isoelectric point and hydropathic properties of the shuffled ompA proteins

Here we found that clones 6 and 12 significantly cloaser and lower the IFT compared to OmpA (the negative control (water/crude oil) displayed a value of 25 mN/m). Protein-IFT correlation can be influenced by several factors that permit to explain protein functionality (Singh et al., 2020), like repulsive interactions (steric or electrostatic) that can play a big role at the interface. Kyte-Doolittle plots have been constantly used to determine hydrophobicity regions (Huang et al., 2014) therefore these plots were obtained for all peptides (Fig. 2).

On one hand clone 12 shows defined hydrophilic-hydrophobic regions with neutral and positively charged zones (Fig. 3), typical for a surfactant when facilitating the interaction at the interface. Moreover, a clear charged-like behavior of the protein is elucidated based on isoelectric point calculation assuming a neutral pH of the sample (Tables 3 and 4). On the other hand, clone 6 does not display a well-defined protein structure (Fig. 4). Nevertheless, the isoelectric point could justify that the protein is charged at neutral pH conditions, possibly explaining the property of the protein to reach a lower IFT value compared to OmpA.

Molecular size and weight have been reported to be indicators for surfactant performance because they can explain how easily a molecule could migrate to the interface, without taking into account specific non-bonding (Bak & Podgorska, 2016). Our results indicate that size and weight play a minor role in improving IFT activity as there is no direct correlation between this variable an the reached IFT. Moreover the isoelectric point for other proteins such as those derived from clones 5, 7, 13, and 16 present approximately neutral values and higher IFT values than OmpA (Table 3), probably corroborating the charge as one of the main factors regarding interface activity in protein design and indicating that proteins with an isoelectric point closer to the solution’s pH would take longer to travel to the interface (Table 4).

Functionality-protein structure relation

The tertiary structure of each protein was predicted using Alphafold2 (Fig. 2). Moreover we obtained the distribution of charges for clones (Fig. 4). It can be noticed that the protein produced by Clone 12 has positive charged residues, while the protein produced by Clone 13 has both (positive and negative) charges in its structure. Studies conducted by Nurdin (2014), show that surfactants with both charges can interact with aromatic compounds naturally present in crude oil, obstructing the adequate organization of the surfactant at the oil-water interface. The behavior in the migration of ionic proteins found in this research, can be attributed to these electrostatic interactions with components present in the oil. Possibly indicating that proteins that are able to reduce the interfacial tension rapidly migrate to the interface, are mainly positively charged.

A deeper examination of the 3D structures shows that alpha-helixes were predicted for clone 12 (Fig. 4). Dexter, Malcolm & Middelberg (2006) designed a peptide with such a secondary structure and the capacity to stabilize emulsions and foams based on forming cohesive interfacial films. Therefore, the role of alpha-helix secondary structures is documented and could explain the mechanism of the protein expressed by clone 12 for decreasing IFT. On the other hand clone 6 displayed a non-typical structure, possibly indicating that conformational changes could take place during the protein migration to the interface which structure prediction algorithms are not able to explain (Fillery-Travis, Mills & Wilde, 2000).

We calculated the stability index (Gasteiger et al., 2005) and predicted solubility (Hebditch et al., 2017) values for the nine clones (Table 5). A protein with a stability index less than 40 is predicted to be stable, while a value above 40 suggests potential instability. These results indicate that clones 6 and 12 are unstable proteins and also exhibit higher solubility compared to the other clones. Plotting the IFT values and stability index (Fig. 5) reveals an apparent inversely proportional relationship, indicating that a more unstable protein has a higher surfactant capacity. When a protein is more unstable, it is more likely to undergo structural changes. Based on our findings, it is conceivable that a more unstable protein may harbor more flexible regions, enabling it to better accommodate interfaces. This, in turn, facilitates its capacity to diminish surface tension and function as a surfactant. The optical density at 280 nm was measured to assess the stability of the clones and OmpA, finding a constant absorbance value. This method is crucial for determining the concentration of protein in the samples, as the absorbance at this specific wavelength is indicative of the presence of aromatic amino acids, primarily tryptophan and tyrosine, which absorb UV light. The constant absorbance reading suggests that the protein concentration remained stable over the course of the experiments, indicating no significant degradation or loss of the clones and OmpA, thus providing a reliable measure of their stability.

IFT evaluation for purified 6 and 12 clones proteins

For the measurement of the IFT at the dodecane-water interface, first, the samples obtained in the purification were quantified, obtaining 0.588 mg/mL for elution with mutant 6 and 0.451 mg/mL for mutant 12. Then, we found that the IFT obtained for the purified clones was lower compared to their correspondent protein extract and lower than OmpA corroborating the individual effect of the individual proteins (Table 3).