Comparative modeling of DszC, an enzyme in biodesulfurization, and performing in silico point mutation for increasing tendency to oil.

Desulfurization protein named DszC from Rhodococcus erythropolis is the key enzyme for biodesulforization of dibenzothiophene (DBT) in 4S pathway, which is a pathway with four enzymes. DszC enzyme biodesulfurizes DBT and its derivatives in oil components and biphasic systems. It functions well at the oil- water interface. In this study point mutation performed in DszC enzyme regarding to increase protein hydrophobicity and stability for application in immobilized form. 3D model of DszC predicted using Phyre2, SAM-T08 and M4t servers. I-Mutant 2 server used to determine potential spots for point mutation, and Molegro Virtual Docker (MVD) used for performing point mutation on 3D model. Hydrophobicity plots generated by Bioedit version 7.0.8.0 in Kyte-Doolittle scale indicated that protein hydrophobicity is increased after mutation. Also protein stability increased 26.11 units in scale of DDC2.

Background

Sulfur concentrations in crude oil supplies are increasing, resulting in high sulfur concentrations in finished petroleum products. Hydrodesulfurization (HDS) used to achieve lower sulfur concentrations in refiners that need to high temperatures and pressure, requiring high operating costs [1]. Benzothiophene (BT), DBT, and alkilated derivatives are recalcitrant organic sulfur and more resistant to HDS treatment than other sulfur compounds such as mercaptans and sulfides [2]. Application of HDS also elicits undesirable effects on fuel quality; other chemical compounds in the oil reduced due to high temperature and pressure needed to achieve low sulfur levels. Biodesulfurization is used as a more selective method for lowering the sulfur content of petroleum products. DBT has been used as a model aromatic sulfur compound for the isolation and characterization of bacteria capable of biodesulfurize organosulfur compounds found in a variety of fossil fuels. The 4S pathway is a specific desulfurization pathway in which DBT is desulfurized and converted to 2-HBP. Through this pathway the carbon skeleton of DBT is released intact and thus the calorific value of the fuel is not lost. Consequently, development of biocatalytic desulfurization for the selective removal of aromatic sulfur compound from petroleum products has focused on the 4S pathway. Several bacterial species have been identified, including Rhodococcus erythropolis, Gordonia, Arthrobacter, Brevibacterium, and Pseudomonas, which are capable of either biotransforming DBT or growing with it as a sole sulfur source. The DBT desulfurization (dsz) operon from Rhodococcus erythropolis IGTS8 encodes three proteins, DszC, DszA, and DszB. These proteins have been isolated, cloned, mutated, and overexpressed [3, 4]. DBT is oxidized by DszC, first to DBT-5-oxide (DBTO) and then to DBT-5, 5'-oxide (DBTO2). DszA catalyzes transformation of DBTO2 to 2-(2'-hydroxyphenyl) benzene sulfinate (HPBS), which opens the thiophenic ring. HPBS is then desulfinated by DszB to produce 2-hydroxybiphenyl (2-HBP) (Figure 1) [3].

Figure 1

Proposed Dsz (4S) pathway for the biodesulfurization of DBT by R. erythropolis IGTS8. DBTO, DBT-5-oxide; DBTO2, DBT-5, 5'-dioxide; HPBS, 2-(2'-hydroxyphenyl)-benzene sulfinate; HBP, 2-hydroxybiphenyl; DBT-MO, DBT monooxygenase; DBTO2-MO, DBTO2 mono-oxygenase; FMN, flavin mononucleotide

The transport of substrates and products might also contribute to desulfurization activity, as demonstrated by the fact that cellfree lysates of desulfurization cultures can exhibit a broader substrate range than the intact cell culture [5]. There is no evidence that Dsz enzymes are excreted from IGTS8 cells, but the size of substrates metabolized show that desulfurization does not occur intracellular; desulfurization occurs in association with the external surface of cells [6]. Immobilizedcell or free-cell biocatalyst (Dsz enzymes) was used to remove sulfur from oil in two-liquid phase systems in researches [7]. To increase the rate of biodesulfurization by cultured or immobilized cells and Dsz enzymes, more researches are needed in areas such as increasing specific desulfurization activity, enzymes stability and hydrophobicity. The purpose of this study was the changes in the structure of DszC enzyme to increase stability and improve its performance in the oil phase in biphasic system. The sequence of key desulfurization pathway enzyme DszC, retrieved from NCBI database with accession number 32363475, was used to optimization of the protein stability and increasing hydrophobicity.

Methodology

Prediction of 3D model of DszC:

Phyre2 [8], SAM-T08 [9] and m4t [10] servers used for predicting 3D structure of DszC. M4t model choosed best based on prosa score [11].

Optimization of model:

Modeller v9.8 [12] used for optimization of predicted model. Also molegro virtual docker 2011.4.3.0 (MVD) used for final check of model. For this purpose charge calculated by MVD and added to model. Explicit hydrogen bonds created and explicit bonds assigned. Energy minimization performed using UCSF chimera candidate version1.5.3. During minimization step update interval was 10 and step size was 0.02. Minimization performed in 100 steps.

Finding ligand binding sites:

3DLigand Site server [11] used for prediction of potentially binding sites of model. This server searches query structure against Uniprot [13] library using matching molecular models obtained from theory (MAMMOTH) [14] method. In the server output 129 ASN, 131 SER, 132 SER, 137 HIS, 138 VAL, 160 HIS, 161 PHE, 162 CYS, 163 SER, 215 SER, 391 HIS, 394 VAL, 396TYR predicted as present in binding site. Also as an alternative approach MVD used for finding cavities of model. For this purpose probe size was 1.2, max number of ray checks was 16, minimum number of ray hits was 12 and Grid Resolution was 0.8. Five cavities found by MVD (Figure 2).

Figure 2

Cavities of DszC predicted model. MVD used for cavity detection. Detecting parameters: probe size 1.2, max number of ray checks was 16, minimum number of ray hits 12 and Grid Resolution 0.8

Point mutation:

Point mutation performed in positions that had not any significant effect in 3D configuration of DszC model. To do this we selected amino acids which were not present in cavities and their side chain did not affect 3D structure of binding sites, as candidate for point mutation. Selected positions submitted to IMutant2 server [15] and in output data energy level of protein for changing a specific position to any other 19 possible amino acids received. This was done for all positions that had potential to perform mutation. Based on increasing stability of the protein, hydrophobic amino acids according to the energy business were selected. The best amino acid substitution was not selected in terms of energy but the selection only was taken in hydrophobic amino acids.

Based on the output of the server, best spots for performing point mutation in a manner that the protein become more hydrophobic and enhance its stability were in positions including:9,11,12,13,14,51,119,122,178,179,180,203,204,221,222,27 8,295 and 338. Since conditions of DszC enzyme activity are at 30°C and neutral pH, thus protein stability upon point mutation was measured regarding to the same temperature and pH.

Mutation in three-dimensional model:

MVD software was used for doing mutations in 3D model. In Selected locations on three-dimensional model, first, native amino acid was replaced with a new amino acid. Replaced residues in new position may stand with it's beside residues at a smaller distance than the threshold. In this case residue sidechain charges are in interference. To avoid interference of charges, optimization of neighborhood residues performed. During this step the charge of replaced amino acid root with lateral-roots of neighbor amino acids became in optimal state, because of the minimum force of mutated amino acid to its neighbors. Mutation and optimization of all selected positions were done to hydrophobic amino acids Table 1 (see supplementary material). The final model for analysis, again submitted to the prosa server. The Z-score for this model was - 5.08

Analysis of mutated model in aspect of affinity to water and oil:

BioEdit version 7.0.8.0 used for generating hydrophobicity plot of native and mutant sequence of query (Figure 3-A). Kyte- Doolittle scale used for depiction of hydrophobic plot.

Figure 3

(A): Hydrophobicity plot of mutated (red) and native (blue) sequence of DsZc; (B): Mutation in positions 9, 11, 12, 13, 14; (C): Mutation in position 51; (D): Mutation in positions 119 and 122; (E): Mutation in positions 178, 179 and 180; (F): Mutation in positions 203 and 204; (G): hydrophobicity based coloring of native and mutated model, (a: after mutation, b: before mutation), Blue color shows hydrophilic area.

Discussion

Predicting 3D structure of query:

Phyre2 server uses comparative modeling and any twenty identical amino acids are hits for this server. In phyre2, 3D model of query generated based on template of chain A from acyl CoA dehydrogenase (C2Z1QA). 394 residues (94% coverage) with 100% confidence have been modeled by this server. M4t and SAM-T08 servers used as alternative for predicting 3D structure of query. M4t uses comparative modeling for predicting query model. In this server five known structures used as template. (2rfq_12_367_A, 2vig_10_374_E, 2vig_10_368_C, 3mpi_5_378_C and 1ukw_4_379_A). SAM-T08 uses Hidden Markov Model algorithm to predict 3D model of Protein. In SAM-T08 serve misaligned residues in query, that have not any identical structure in database, are modeled by HMM algorithm and 3D model of misaligned residues conjectures by this server. Generated models checked in prosa server and Z-score of phyre2 was-7.53, SAM-T08 reached score- 6.83and score of m4t model was -7.66. M4t output use for further study.

Prediction of protein stability:

I-Mutant 2 which used in this research is a Support Vector Machine –based server that is trained based on a dataset from ProTherm. Trained SVM in this server can predict stability of protein upon point mutation based on both protein sequence and structure. When protein sequence is used, overall accuracy of prediction is 0.77 Table 2 (see supplementary material). Output of server is energy level of protein in DDG value. DDG is defined as unfolding Gibbs free energy of mutated protein minus unfolding Gibbs free energy of wild type protein. DDG unit is Kcal/mol.

Mutation in positions 9, 11, 12, 13, 14 changes structure of first loop of protein (Figure 3-B). This fold has a loop. Because amino acids that present in this loop were not detected as present in binding site by 3DLigandSite and MVD, it is probable that structure changing which caused by mutating 5 amino acids has not significant effect in structure of protein pockets. Mutation in position 51 (Figure 3-C) changes structure of a separate and surface exposed fold of protein. This fold has not any effect in configuration of detected binding sites of enzyme. Positions 119 and 122 are located in an alpha helix and mutating these positions do not change helix form (Figure 3-D). Positions 142, 278, 295 and 338 are in surface of protein and mutation in these positions does not affect structure of fold and cavities of protein significantly. Mutation in 178, 179 and 180 do not change 3D configuration of protein (Figure 3-E), because neighbor tern is made by prolin 182. Also as we see in (Figure 3-F), mutating positions 203 and 204 do not have any significant effect on protein folding. His 221 and Asn 222 are two surface amino acids that are located away from cavities and turns. After performing mutation, protein stability increased 6.24 units in DDG scale.

Generating hydrophobicity plot of query:

BioEdit version 7.0.8.0 was used for generating hydrophobicity plot and hydrophobicity scale was Kyte-Doolittle [16]. This scale is widely used for detecting hydrophobic regions in proteins. In this scale, regions which reach positive value are hydrophob. Depending on the window size, this scale can be used for surface exposed and transmembrane regions.Window sizes of 5-7 is used for surface-exposed regions and window sizes of 19-21 is used for finding transmembrane domains if the values calculated are above 1.6 . Because DszC is not transmembrane protein, default window size of 13 used for detecting surface exposed regions of query.

Conclusion

In native DszC enzyme affinity to water is more than oil, because of the side-chain root of amino acids present in surface of protein. Causing the mutation in surface amino acids of DszC to hydrophobe residues changed hydrophobicity properties of this protein. After performing mutation, the DszC enzyme had more tendencies to oil surface in a manner that its stability increased. This mutated enzyme can work better in biphasic system by increasing hydrophobicity, and can maintain in active structure in longer time by increasing stability. Also differences in coloring of native model and mutant model (Figure 3-G) showed the more water escaping in mutant model. Since the stability of this enzyme increased with mutation, so it is suggesting that mutated enzyme can be used by high performance for biodesulfurization in biphasic system as immobilized enzyme.