In Silico Analysis of the Dextransucrase Obtained From Leuconostoc mesenteroides Strain IBUN 91.2.98

Abstract

The DSR-IBUN dextransucrase produced by Leuconostoc mesenteroides strain IBUN 91.2.98 has a short production time (4.5 hours), an enzymatic activity of 24.8 U/mL, and a specific activity of purified enzyme 2 times higher (331.6 U/mg) than that reported for similar enzymes. The aim of this study was to generate a structural model that, from an in silico approach, allows a better understanding, from the structural point of view, of the activity obtained by the enzyme of interest, which is key to continue with its study and industry application. For this, we translated the nucleotide sequence of the dsr_IBUN gene. With the primary structure of DSR-IBUN, the in silico prediction of physicochemical parameters, the possible subcellular localization, the presence of signal peptide, and the location of domains and functional and structural motifs of the protein were established. Subsequently, its secondary and tertiary structure were predicted and a homology model of the dextransucrase under study was constructed using Swiss-Model, performing careful template selection. The values obtained for the model, Global Model Quality Estimation (0.63), Quality Mean (−1.49), and root-mean-square deviation (0.09), allow us to affirm that the model for the enzyme dextransucrase DSR-IBUN is of adequate quality and can be used as a source of information for this protein.

Introduction

The dextransucrase enzyme (DSR, EC 2.4.1.5) catalyzes the transfer of α-d-glucose units from sucrose to low-molecular-weight acceptors such as other sucrose molecules, primary alcohols, or a growing polymer, called glucan,^1,2 to synthesize oligo and polysaccharides while releasing fructose to the medium.^3,4 The main product of this reaction is a biopolymer of great economic importance called dextran, with a wide range of applications (1) in the food industry as an additive, viscosifier, and reducer of caloric content; (2) in the pharmaceutical industry as a volume expander of the blood plasma and promoter of blood flow; and (3) in the area of medicine due to the promotion of intestinal health due to its anti-inflammatory activity.^5-7

The biopolymers and biofunctionals research group obtained a high-molecular-weight dextran-type biopolymer from the native strain Leuconostoc mesenteroides IBUN 91.2.98 culture, with promising properties for its use as a prebiotic and soluble fiber source in animals and humans. An enzyme responsible of dextran synthesis was then obtained by Flórez-Guzmán. ⁸ This enzyme, called DSR-IBUN, was purified and characterized. It was found that, under used experimental conditions, DSR-IBUN had a short production time (4.5 hours) and an enzymatic activity of 24.8 U/mL, which is higher than those reported for L mesenteroides NRRL B-512F, ⁹ L mesenteroides B512 FMC, ¹⁰ L mesenteroides NRRL B-1299, ¹¹ and the alternansucrase of L citreum NRRL B-1355, ¹² with values of 5.85, 6.33, 0.58, and 0.66 U/mL, respectively, as well as a specific activity up to 2 times higher (331.6 U/mg) than that observed to date for other strains such as L mesenteroides B-512FMC, ¹³ L mesenteroides 0326, ¹⁴ and L mesenteroides B-51 ¹³ with values of 183, 72.54, and 53 U/mg, respectively. Although several research groups have researched about DSR-IBUN and the potential uses of the polymer, ⁸ the sequencing of the gene encoding dextransucrase had not been done nor had proposed a structural model that, from an in silico approach, allows a better understanding of protein activity.

Due to high activity of this enzyme, it is important to know the activity dynamics and the enzymatic kinetics from a structural point of view. In particular, dextransucrase DSR-IBUN and other large membrane protein structures are difficult to solve even with methods such as x-ray crystallography, nuclear magnetic resonance (NMR), or Cryo-EM. ¹⁵ However, the understanding of protein function generally requires understanding its 3-dimensional (3D) structure, to uncover biological process involved, and also to perform important modification that may lead to important activity dynamic changes.^15,16 A viable alternative approach is to predict the 3D structure in silico using homology modeling or comparative modeling, ¹⁷ in which the 3D model of a protein of interest (target) can be built from one or several related proteins of experimentally determined structure (template) that share a statistically global similarity in its sequence.^18-20

The aim of this work was to generate a 3D model for the dextransucrase DSR-IBUN of L mesenteroides IBUN 91.2.98 based on gene sequence that encodes it, to study possible modifications that could imply changes that allow the production of polymers with different characteristics that could be used for different applications.

Methodology

Sequencing of dsr_IBUN gene

Applying Sambrook and Russell ²¹ criteria, the following primers were designed for amplification of the interest gene: dsr_IBUN_dir_HindIII: 5′-AAGCTTATGCCATTTACAGAAAAAAGT-3′ and dsr_IBUN_Rev_XbaI: 5′-TCTAGAAGAAAACTTATGCTGACACA-3′. A polymerase chain reaction (PCR) was performed using high-fidelity enzyme Phusion DNA polymerase (New England biolabs) (0.5 U), buffer (1X), MgCl₂ (1.5mM), dNTP mix (200 μM), direct primer (0.5 μM), reverse primer (0.5 μM), template DNA (150 ng), and dimethyl sulfoxide (3%). The PCR conditions were 98 °C for 30 seconds and 30 cycles of 98 °C for 5 seconds, 54 °C for 30 seconds, 72 °C for 2.5 minutes, and a final extension at 72 °C for 8 minutes. The amplified PCR product was analyzed by electrophoresis on a 1% agarose gel at 80 V. The product was purified, cloned in the vector pTOP TA (Takara) and sequenced by Macrogen. Gene sequence is available in GenBank under the accession number OQ731663.1.

DSR-IBUN sequence analysis

To obtain primary protein structure of DSR-IBUN dextransucrase, translation of the nucleotide sequence of dsr_IBUN gene was performed using Biomodel application. ²² Features such as protein molecular weight, theoretical Ip, amino acid composition, atomic composition, estimated half-life, instability index (II), aliphatic index, and average hydrophobicity index (GRAVY for Grand Average of Hydropathy) were estimated using the Expasy ProtParam tool. ²³ Subcellular localization prediction was performed with the CELLO v2.5 server.^24,25 Signal peptide sequence was established using SignalP 5.0 server (www.cbs.dtu.dk/services/SignalP). ²⁶ Subsequently, the amino acid sequence obtained was compared with those reported in the NCBI using the BLAST algorithm (https://www.ncbi.nlm.nih.gov/), and phylogenetic tree was constructed with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). ²⁷

To establish the presence of conserved functional domains and motifs in deduced DSR-IBUN sequence, Motif Scan program was used (https://myhits.sib.swiss/). Structural motifs were determined by sequence alignment with several GH70 family proteins using the algorithm for multiple sequence alignments MAFF (https://www.ebi.ac.uk/Tools/msa/clustalo/). To visualize these alignments, Jalview software was used (https://www.jalview.org/). ²⁸

Modeling of the DSR-IBUN

Primary DSR-IBUN structure was used to predict its secondary and tertiary structure to build a model. Homology with similar proteins, for which its 3D structure had been determined experimentally, was determined using 4 steps methodology proposed by Quinn ²⁹ : (1) sequence preparation, (2) modeling, (3) validation, and (4) application.

Initial prediction of DSR-IBUN secondary structure was performed with SOPMA server (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_sopma.htm). ³⁰ The amino acid sequence of DSR-IBUN was compared with those available in the Protein Data Bank (PDB) using Swiss-Model (https://swissmodel.expasy.org/), ³¹ Phyre² (http://www.sbg.bio.ic.ac.uk/phyre2/), ³² I-TASSER (https://zhanglab.ccmb.med.umich.edu/I-TASSER/), ³³ and the NCBI servers (https://www.ncbi.nlm.nih.gov/) allowing comparison between them to select those structures with a higher identity percentage and greater coverage as possible candidates to be used as templates for structure modeling.

To select template structure, quality of preselected crystal structures was addressed, in terms of their resolution and R values, as these indicate the quality obtained from crystallographic data. Metrics structure summary graphs validation reports were also considered, in which the overall quality of each structure was compared with the ranges of structures previously reported in the PDB.

Once template structure was selected, Swiss-Model, Phyre, ² and I-TASSER servers were used to corroborate secondary structure prediction and structural domains distribution, as well as to validate the different models proposed for DSR-IBUN. Protein structure models were compared using Ramachandran plots data, concerning to energetically allowed regions for the dihedral angles of models’ skeletons, as well as the obtained GMQE (Global Model Quality Estimation) and QMEANDisCo Global (Quality Mean) values. With these data, a 3D in silico model structure for DSR-IBUN was selected.

Interactions in the catalytic domain of DSR-IBUN proposed model

Active site catalytic triad (Asp, Glu, Asp) previously reported by other authors was identified using Pymol.^34-37 The validated in silico structure was used to establish the location of the active site and the catalytic triad involved in the function of the enzyme by superimposing template structure used above and structural DSR-IBUN protein model using the Pymol software. ³⁸ Interacting amino acids of DSR-IBUN validated model and sucrose molecule were analyzed via Discovery Studio Visualizer software. ³⁹

Molecular docking between isomaltose and DSR-IBUN was directed to glucan-binding pockets as the interaction sites using the HADDOCK 2.4 platform (https://wenmr.science.uu.nl/haddock2.4/). ⁴⁰

Results and Discussion

DSR-IBUN primary and secondary sequence analysis

Translation of obtained nucleotide sequence for dsr_IBUN gene resulted in a 1527 residues protein. Its estimated molecular weight and isoelectric point (Ip) were 169.8 kDa and 4.29, respectively. Data are consistent with those reported experimentally by Flórez-Guzmán et al, ⁸ in an electrophoretic analysis performed to DSR-IBUN, which showed a molecular mass of 170.1 kDa and a Ip of 4.2. On the contrary, the predicted molecular formula for this dextransucrase was C₇₄₈₇H₁₁₂₄₂N₁₉₆₆O₂₅₁₉S₂₁.

Most abundant amino acid is aspartate (10% of the residues found), whereas the least abundant is cysteine with only 0.1% (1 residue). Total number of negatively charged residues (Asp+Glu: 217) was greater than those observed for positively charged residues (Arg+Lys: 101). Although amino acids such as Asp and Gly do not appear preferentially in extracellular or intracellular proteins, it has been observed that polar amino acids such as Thr, Asn, Gln, and Ser are characteristic of extracellular enzymes, ⁴¹ which coincides with observed for DSR-IBUN, in which these 4 amino acids are abundant, representing 30.6% of total residues. Even though extracellular eukaryotic proteins frequently have a high percentage of cysteines, prokaryotic proteins rarely present disulfide bridges, even in the case of extracellular ones. ⁴¹ These results agree with the prediction of the subcellular localization done, which indicates that DSR-IBUN is an extracellular protein which presents a fraction anchored to cell wall and membrane, a characteristic that is confirmed with subsequent bioinformatic analysis.

Estimated DSR-IBUN half-life values in different biological media such as mammals, yeast, and Escherichia coli (in vivo and in vitro) for all cases were greater than 10 hours, indicating that it is a stable protein (Table 1). This stability over time was corroborated by the II, which estimates the protein stability under laboratory conditions ⁴² and by the aliphatic index, which refers to the relative volume occupied by aliphatic chains (alanine, valine, isoleucine, and leucine). ⁴³ Any protein that, like DSR-IBUN, presents an II value <40 ⁴² and an aliphatic index >50 is considered stable, ⁴³ as aliphatic chains provide heat resistance and therefore stability in different conditions. ⁴⁴ As the value of GRAVY determined for the studied dextransucrase was negative (−0.599), which indicates that it is a hydrophilic protein, ⁴⁵ this is consistent with the proportion of polar groups present in the side chains located on the protein surface, which promotes solubility. ⁴³

Table 1.

Predicted physicochemical properties for the DSR-IBUN protein.

Parameter	Worth
Half-life (mammalian reticulocytes in vitro)	30 pm
Half-life in yeast in vivo	>20 hours
Half-life in Escherichia coli in vivo	>10 am.
Instability index (II)	22.16
Aliphatic index	68.22
GRAVY (Grand Average of Hydropathy)	−0.599

Source: Protparam-obtained data.

Using SignalP 5.0 server, a typical secretion signal peptide sequence in the N-terminal region was predicted with a peptidase cleavage site between amino acids at positions 42 and 43 (G-D), which is consistent with the reported data for other DSR such as DexYG ³³ and DsrD ⁴⁶ and supports the already observed fact of dextransucrase secretion of from L mesenteroides IBUN 91.2.98 to growth media. ⁴⁷

DSR-IBUN showed 100% coverage and an identity percentage greater than 98% compared with other L mesenteroides strain sequences, such as DsrN from KIBGE-IB-22, ⁴⁸ , ⁴⁹ DexYG from L mesenteroides 0326, ⁵⁰ DEX protein from strain NRRL B-512F, ⁵¹ and DsrD from the Lcc4 strain. ⁴⁶ Compared with other proteins, dsrV (AHC31982.1) from L citreum M-3 ⁵² displayed 100% coverage and 94.31% identity percentage, whereas sequences such as dsrb742 (AAG38021.1) from L mesenteroides B-742CB ⁵³ and dsrB from L mesenteroides NRRL B-1299 (AAB95453.1) ⁵⁴ showed a 99% coverage and 66.25% and 66.45% identity percentages, respectively. Lower identity percentages were observed comparing with strains of other species such as Lactococcus lactis (around 95% coverage and 60% of identity percentage). Phylogenetic tree constructed using 14 sequences of dextransucrases deposited in NCBI database showed that DsrD is the closest to DSR-IBUN protein as Flórez-Guzmán et al ⁸ reported. Phylogenetic analysis also showed that glucansucrases (GS) of the same species grouped together; as Martínez et al ⁵⁵ mentioned, in its simplest form, the genetic diversity that drives adaptation is based on the accumulation of point mutations that are established in the same species.

By comparing the deduced amino acid sequence for DSR-IBUN with the databases available information, it was possible to corroborate the presence of a functional domain conserved in GH70 family, as well as to observe the functional domain of α-amylase super family, to which dextransucrase belongs, which is organized under a structure (α/β).^{8,14,34,54,56-60} In addition, repeated motifs were observed both at the start and at the end of the sequence related to the binding site to cell wall binding (cell wall motifs), which agrees with that mentioned by Vuillemin et al, ⁶¹ who point out that it has been determined that most of the active GH70 enzymes are strongly bound to the cell wall, making their purification difficult.

Likewise, repeated motifs related to sugar-binding sites were found at the N and C terminus of the protein. That is why domain V is known as the glucan-binding domain (GBD), as Claverie et al ⁵⁹ reported that these motifs are made up of “pockets” in which the glucoses can interact. Importance of this domain has been suggested to be related to the polymer size and, therefore, the enzyme activity, which varies between dextransucrases. ¹⁴

The 7 structural motifs (I to VII) of the GH70 indicated by Yan et al, ⁶² were identified in the DSR-IBUN primary sequence that has 22 strictly conserved residues in this family. Vuillemin et al ⁶¹ and Venkatachalam et al ⁶³ have shown that GH70 enzymes can be grouped into 2 subfamilies, (1) GS and (2) branching enzymes or branching sucrase enzymes, sharing the enzyme called Gys as an evolutionary intermediate between them. In the alignment (Figure 1), it can be observed that DSR-IBUN is part of the GS subfamily, which indicates that the synthesized dextran has mainly α1,6 bonds, accompanied by α1,3 bonds. ⁶⁰

Figure 1.

Structural motifs conserved in DSR-IBUN.

DSR-IBUN secondary structure prediction showed 39.03% random coils, 28.55% β-sheets, 20.89% α-helices, and 11.53% β-turns, which is comparable with what was obtained by Du et al ⁶⁴ for L mesenteroides DRP105 dextransucrase.

DSR-IBUN prediction structure

When comparing DSR-IBUN amino acid sequence with the experimental data of solved structures reported in the PDB using 4 different platforms, it was found that, in all of them, the first 5 suggested templates correspond with the same ones (5NGY, 5LFC, 6HVG, 6SYQ, and 3HZ3), but the order proposed on each platform for the templates was different. Both NCBI and Swiss-Model match in suggesting 5NGY as the highest value for GMQE, followed by 5LFC, while Phyre² and I-TASSER suggested 6HVG as the first option. This is explained by the fact that the first 2 platforms use modeling algorithms in which the sequence prevails over the structure, whereas Phyre² and I-TASSER use threading algorithms, in which structural templates of PDB are first identified and later the models are built by iterative simulations that assemble or weave the fragments. ⁶⁵

Models that showed the highest quality indices on each platform were evaluated to compare them and select the most appropriated one. Comparison was made using Ramachandran plots (Figure 2). Swiss-Model (2A) showed a Ramachandran favorability of 94.14%, followed by the model generated using Phyre² (2B) with a favorability of 93.35%, while in the model suggested by I-TASSER (2 C), this parameter had only 71.90%. This is observed in the plots where it is clear that the outliers for the evaluated structure in Figure 2C had the highest number of outliers (11.37%), whereas in the proposed models by Phyre² and Swiss-Model, these values did not exceed 2% (1.17% and 1.44%, respectively). Taking these results into account, the Swiss-Model platform was used to create the final model.

Figure 2.

Ramachandran plots for DSR-IBUN models, obtained by (A) Swiss-Model, (B) Phyre², and (C) I-TASSER platforms.

Coverage and the identity percentage values were taken as referents for the template structure selection. In all cases, the highest coverage (89%) and identity percentage (51%) were turned up by the resolved structures for the dextransucrase DSR-M from Leuconostoc citreum and NRRL B-1299 with PDB codes 5LFC and 5NGY, respectively. ⁵⁸ The DSR sequences reported for L mesenteroides such as DSR-E and its mutants had a coverage of 79% and an identity percentage of 44%. Finally, alternansucrase corresponding to the structure 6HVG showed coverage of 73% and presented identity percentage of 51%.

To select the template, structure quality of 5LFC and 5NGY structures was checked. Both structures had adequate resolution values (around 4Å), but R values, which indicate the measure of the atomic model quality obtained from the crystallographic data, were better for 5NGY, since the closer to 0 the value of R is, the greater the fit of the structure to the model, with a typical value of 0.2. ¹⁹ This, together with the summary graphs metrics in which the global quality of each structure (black bars) is compared, with respect to the ranges of similar structures previously deposited in the PDB (white bars), allows us to observe that the quality of 5NGY is superior as it does not present outliers in the Ramachandran plots and fewer outliers in the resulting side chains, which is why it was used as a template for the construction of the model (Figure 3).

Figure 3.

Validation report DSR-M dextransucrase from Leuconostoc citreum NRRL B-1299 solved structures. (A) 5LFC structure and (B) 5NGY structure. The values presented in the blue range are considered good, whereas those observed toward the red range are considered of low quality. PDB indicates Protein Data Bank.

Source: Taken from PDB. ¹⁹

Once 5NGY was selected as a template, SOPMA was used to obtain secondary structure prediction, while using Swiss-Model, DSR-IBUN model was constructed and validated. As expected, it corresponds to a U-shaped monomer (Figure 4A), according to that reported for the glycosylhydrolases of the GH-H Clan, which corresponds to the GH70 family.^34,66

Figure 4.

(A) Model of the dextransucrase of Leuconostoc mesenteroides IBUN 91.2.98 tertiary structure built and validated with Swiss-Model. (B) Scheme of predicted organizational domains for DSR-IBUN. Residue number at which each domain in the primary structure begins and ends is shown and the initial and final amino acids present in the model are highlighted.

Global Model Quality Estimation value which estimates quality and reflects the expected precision of a model created from the alignment between query sequence and template sequence displayed a value of 0.56, which is considered positive as higher values at 0.5 indicate acceptable overall quality (PDB, nd). The Mean Quality indicator (QMEANDisCo Global) score, based on different geometric properties of the model that provides both global (ie, for the whole structure) and local (ie, by residue) quality estimates and distance evaluation pairwise in a model and a set of constraints derived from experimentally determined protein structures that are homologous to the model being tested, was 0.72 ± 0.005. As scores between 0 and 1 indicate that the model is comparable to what would be expected from experimental structures of similar size, the calculated value is considered adequate.

DSR-IBUN structure model displays a typical organization of dextransucrase consisting of 5 structural domains called A, B, C, IV, and V, with domain C being the only one continuous throughout the structure (Figure 5). The U-shaped folding of DSR-IBUN tallies with reported structures of other GH70 enzymes. ³⁴ Figure 5 shows the organization of the 5 DSR domains: domain A comprising a barrel (α/β)8; domain B (from ß3 to α3); and domains C, IV, and V which form the bottom of the “U.”

Figure 5.

Typical domains of dextransucrase A, B, C, IV, V, C-terminal, and N-terminal. The identified glucan-binding sites for each of the domains are shown.

Figure 5 shows a domain A which consists of 8 α-helices and 8 β-strands alternated with each other, linked by irregular loops of variable sizes that engage the C-terminal ends of the β-strands with the N-terminal ends of the α-helices and the C-terminal ends of the α-helices joined with the N-terminal ends of the β-chains.^34,58 The beta barrel structure (β/α)8 is organized with parallel β strands organized almost parallel to each other, as if they were around a cylinder and located toward within the protein. On the contrary, the 8 α-helices are located toward the outside of the β strands, on the protein surface. ⁶⁷

Domain B, located close to the catalytic domain (A), is formed by 5 β-sheets (Figure 5). It has been suggested that this domain is essential for the enzyme function as it provides residues that are part of the protein donor and acceptor sites and some of this domain loops contribute to give the groove shape near the catalytic site. ⁶⁸ Domain C, also located close to the catalytic domain, has 8 β-sheets and, as seen in the model, joins sections of domains A and B so that they can perform their function.

Domain IV structure reveals a novel folding that is characteristic of GH70 enzymes. Domain IV connects domain B to V, but other functions remain unknown, although it has been proposed that it may provide a “hinge” that would lead domain V to attached glycans near or far from the catalytic site. In fact, the connection between IV and V domains consists of 2 relatively long polypeptide stretches with disordered structure, and the relative position of different dextransucrases V domain with respect to the rest of the protein indicates some flexibility in these stretches. ⁶⁹

Finally, domain V is made up of the N-terminal and C-terminal segments. ⁶⁹ This agrees with that reported by Vujicic-Zagar et al, ³⁴ who point out that in family 70 glycosylhydrolases, except for the C domain that forms the base of the U-shaped fold, the domains are formed by 2 noncontiguous segments of amino acids from the N-terminal and C-terminal ends, where the amino and carboxy termini meet to form the V domain of the framework. Therefore, it has been proposed that the GH70 enzymes originated from the GH13 enzymes, ⁷⁰ by an evolutionary pathway based on duplication permutation, as the GH70 enzymes share the domains with the GH13 enzymes (A, B, and C), but in GH70 enzymes, 2 of these domains, as previously discussed, are made up of noncontiguous sections in the structure. Despite these rearrangements, there are several conserved motifs (eg, motifs I-VII) in members of the 2 families, and these conserved motifs provide crucial information for analyzing the origin and evolution of GH70 enzymes. ⁶²

Domain V, known as GBD, has YG repeat motifs ⁷¹ and some pockets that can bind glycan residues which were identified in the model (Figure 5). Several studies suggest that domain V, located at the N-terminus and C-terminus of enzymes, determines (1) the size of the polymer, (2) enzymatic activity, (3) specificity, and (4) processivity. In addition, it has been shown that in some Streptococcus and Leuconostoc species these amino acids allow the binding of the enzyme to the cell wall.^{34,54,59,72,73} However, its precise role is not yet fully understood. ³⁵ Unfortunately, the structures reported for GH70 enzymes are truncated and the complete structure is yet unknown.

Some structures showing interacting residues with ligands such as sucrase, glucose, isomaltose, and isomaltriose provide structural evidence of common topology of glucan-binding sites in V-domain repeats. ³⁷ The importance of these repeats in the V domain lies in the fact that deleting some of them produces a progressive loss of activity and a decrease in the size of the polymer. ⁷² In Figure 6, the hydrogen bond type interactions between isomaltose and some residues of the glucan-binding pockets of the IV, V, and N-terminal domains are shown (Figure 6).

Figure 6.

Hydrogen bonds present in glucan-binding pockets with isomaltose: (A) N-terminal domain binding pocket, (B) glucan-binding pocket domain V, and (C) glucan-binding pocket domain IV.

Identification of active site and calcium ion pocket

Taking into account that the data quality obtained for DSR-IBUN model and its validation supported its reliability to be used for subsequent in silico studies, deeper analyses were done using modeled structure.

Localization of the active site and the catalytic triad involved in enzyme function was determined in the structural model, by making an alignment and superimposition between the model and the template. This allowed to determine the root-mean-square deviation (RMSD) value of the atomic positions (RMSD), which is a measure of the average distance between the carbon skeleton atoms of 2 superimposed proteins ²⁹ and the value of which allows us to affirm whether the prediction made for the structural model is adequate with respect to the available information available on the template. For this case, the RMSD was 0.09, which is considered positive, since the closer the value is to 0, it is inferred that the positions in the 3D structure are similar. After reviewing the interaction interface using sucrose as ligand, it was observed that the substrate in its most stable conformation shows interactions with the catalytic triad Asp509, Glu547, and Asp620 established mainly by the formation of hydrogen bonds. It is also possible to find other interactions between substrate and amino acids close to the catalytic site, such as Arg 507, His 619, and Asp 936 (Figure 7).

Figure 7.

Surface interaction between sucrase (brown) and the active site of the catalytic domain Asp509, Glu547, and Asp620 of dextransucrase (lavender).

To complement it was found that the model suggests the presence of a conserved site for the binding of Ca⁺² ions, interacting with the residues Glu 483, Asp 469, Asn 513, and Asp 986 (Figure 8), it was found that interaction occurs near to nucleophilic aspartate of the active site, which has been shown to be essential for the enzyme catalytic activity. ³⁶

Figure 8.

Surface interaction between calcium (green) and the residues Glu 483, Asp 469, Asn 513, Asp 986 of the DSR-IBUN (lavender).

Calcium dependence has been reported for some GS,^34,37,59 showing a 14% decrease in enzymatic activity in the presence of chelating agents such as EDTA, ³⁷ as well as increases of up to 48% of the activity in presence of these ions. ⁷³ However, this aspect must be carefully considered, as according to what was reported by Flórez-Guzmán ⁴⁷ for the dextransucrase of L mesenteroides IBUN 91.2.98 strain, the calcium ion had a slight inhibitory effect of 8% of this enzyme activity.

With the structural model obtained, it is possible to generate a methodological approach that allows to propose strategies to carry out mutations in the gene that codes for DSR-IBUN, first by establishing that in silico mutations do not have destabilizing effects on the structure of mutant proteins, seeking to guarantee that they are functional in vitro.

Taking into account the benefits of homology modeling applied in the methodology carried out in this study, it is emphasized that the template used had 52.06% identity, which is considered optimum according to what has been proposed by other authors. Comparative modeling of the 3D structure, if the target presents at least 30% to 40% identity with an empirically determined structure, can be applied with reasonable accuracy to 10 times as many protein sequences, relative to the number of experimentally determined protein structures, ⁷⁴ result in a model with a precision close to that of a structure obtained by low-resolution x-rays or one studied by medium-resolution NMR. ¹⁶

Within the limitations of homology modeling, it is found that similarity values less than 30% decrease the accuracy of comparative models and therefore their applications, mainly because of the intrinsic errors of the alignment algorithms. ⁷⁵ This can be overcome if the quality parameters of the template and the model are considered, as was done in this work. Errors in the packaging of the side chains and in the conformation of the central segments and loops have also been observed in the comparative models, ¹⁶ as well as the transmembrane sequences of the proteins, ⁷⁶ , ⁷⁷ as these are the ones that tend to present the greatest experimental error, which, in turn, is reflected in the models. These drawbacks have been minimized by development of new algorithms for both sequence comparison and structure comparison, ³² and because other techniques such as single-particle cryoelectron microscopy (Cryo-EM) and small-angle x-ray scattering have complemented existing x-ray information and provided new information, obtaining more and better experimentally determined protein structures. ⁷⁸ For this, the number of applications in which comparative modeling has been shown to be useful has grown rapidly.

It is important to emphasize that, as is known, knowledge of the 3D structure of a protein is crucial to answer many biological questions; however, obtaining protein structures resolved by experimental methods is up to 7 times smaller, compared with the number of sequences of genes and genomes reported, ⁷⁹ a difference that increases day by day, thanks to the next-generation sequencing (NGS) systems. Finally, comparative modeling methods, such as the one presented in this work, reduce the gap in knowledge of a specific structure, by providing more reliable and accurate protein structural models, compared with the experimental data provided. That is the reason why the model obtained is going to be used to study possible modifications that could imply changes that allow the production of polymers with different characteristics that could be used for different applications. Molecular dynamics tests will also be carried out to establish more precisely what has been determined with the bioinformatics analyses carried out to date.

Conclusions

The values obtained for DSR-IBUN structural model for the GMQE and the QMEAN (Quality Mean) of 0.63 and −1.49, respectively, as well as the value obtained for the mean square deviation of the atomic positions (RMSD), corresponding to 0.09, allow us to affirm that the model obtained for the enzyme dextransucrase DSR_IBUN has adequate quality to be used as a source of information to perform different analyses directed to answer biological questions related to this enzyme.