Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2023 Dec 1;32(12):e4837. doi: 10.1002/pro.4837

Exploring the molecular mechanism of cold‐adaption of an alkaline protease mutant by molecular dynamics simulations and residue interaction network

Ailan Huang 1, Fuping Lu 1,2, Fufeng Liu 1,2,✉
PMCID: PMC10682693  PMID: 37984374

Abstract

Psychrophilic proteases have attracted enormous attention in past decades, due to their high catalytic activity at low temperatures in a wide range of industrial processes, especially in the detergent and leather industries. Among them, H5 is an alkaline protease mutant, which featuring psychrophilic‐like behavior, but the reasons that H5 with higher activity at low temperatures are still poorly understood. Herein, the molecular dynamics (MD) simulations combined with residue interaction network (RIN) were utilized to investigate the mechanisms of the cold‐adaption of mutant H5. The results demonstrated that two loops involved in the substrate binding G100‐S104 and S125‐S129 in H5 had higher mobility, and the distance enlargement between the two loops modulated the substrate's accessibility compared with wild type counterpart. Besides, H5 enhanced conformational flexibility by weakening salt bridges and increasing interaction with the solvent. In particular, the absence of Lys251–Asp197–Arg247 salt bridge network may contribute to the structural mobility. Based on the free energy landscape and molecular mechanics Poisson−Boltzmann surface area of the wild type and H5, it was elucidated that H5 possessed a large population of interconvertible conformations, resulting in the weaker substrate binding free energy. The calculated RIN topology parameters such as the average degree, average cluster coefficient, and average path length further verified that the mutant H5 attenuated residue‐to‐residue interactions. The investigation of the mechanisms by which how the residue mutation affects the stability and activity of enzymes provides a theoretical basis for the development of cold‐adapted protease.

Keywords: alkaline proteases, cold adaptation, free energy landscapes, molecular dynamics simulations, residue interaction network

1. INTRODUCTION

Cold‐adapted enzymes have attracted extensive attention due to their high catalytic activity and stability at low temperatures (Collins et al., 2008; Feller, 2013; Feller & Gerday, 2003; Siddiqui & Cavicchioli, 2006). They have shown broad biotechnology applications in versatile industries. For instance, alkaline proteases catalyze proteins into peptides at high pH, which are widely used in detergent, leather, textile, food, and pharmaceutical industries (Gupta, Beg, Khan, & Chauhan, 2002; Pathak & Rathod, 2018; Sharma et al., 2019). Especially as a detergent additive, cold‐active alkaline proteases digest the blood, milk, and food scraps efficiently at low temperatures, thus have little damage to the fabric and can prolong the longevity of the textile (Barzkar, 2020; Ellaiah et al., 2002; Gupta, Beg, & Lorenz, 2002). In addition, because of the superior cold wash performance and no extra heating process required, the cold‐adapted alkaline proteases are eco‐friendly and energy‐saving.

Extensive efforts have been done to explore cold‐active proteases, and several cold‐adapted proteases have been isolated from nature (Almog et al., 2009; Arnorsdottir et al., 2005; Chen et al., 2018). In addition to the identification of natural cold‐active enzymes, the modification of existing mesophilic enzymes to generate cold‐adapted protease is also carried out. Through substitutions of amino acids, mesophilic proteins show psychrophilic‐like characteristics (Liu et al., 2014; Tindbaek et al., 2004; Tsigos et al., 2001; Wintrode et al., 2000; Zhao & Feng, 2018). Among them, an engineered variant H5 was constructed by transferring a flexible loop region (S39) from psychrophilic enzyme to the mesophilic enzyme savinase, which showed high catalytic activity at all measured temperatures (Tindbaek et al., 2004).

When modifying of existing mesophilic enzymes to generate cold‐adapted protease, the crystal structure of proteins can provide guidance for the rational design in protein engineering. For example, the resolved structure of savinase exhibits a hemispherical architecture with a diameter of approximately 40 Å, wherein the active center is partially buried on the surface of the protein. The catalytic triad, a highly conserved motif in proteases, is arranged in a precise orientation in the 3D structure (Hedstrom, 2002; Siezen & Leunissen, 1997). Specifically, in wild type (WT) and mutant H5, the catalytic triad consists of Asp32, His64, and Ser221, where Ser221 acts as a nucleophile to attack the peptide bond of the substrate and a proton transfer from Ser221 to His64. As for Asp32, it helps to assist the correct tautomer of His62. The substrate binding regions of WT savinase is residues S99‐V104 and S125‐P131. This region has a relatively high B factor, indicating that this is a flexible region (Betzel et al., 1992). The rationale behind that these residues were targeted for replacement stems from the understanding that cold‐adapted enzymes exhibit improved activity at low temperatures by modificating local flexibility. Therefore, the transplantation of highly flexible regions onto the mesophilic enzyme was employed to improve the activity at low temperatures.

Nevertheless, the cold‐adaption mechanism of the engineered variant H5 was not well understood. As a computational technique, molecular dynamics (MD) simulations can provide detailed protein structural information over time from the atomic level. It has been utilized for investigating the effect of substrate or ligand on the protein conformational stability (Feng et al., 2023; Tao et al., 2010), the tuning function of different oligomeric states and different regions of protein on enzymatic activity (East et al., 2022; Lilina et al., 2022), the mechanism of the enhanced activity, and thermostability of proteins (Hu et al., 2022; Huang et al., 2021; Wu et al., 2022; Xu et al., 2021). Therefore, the MD simulations was applied to identify the key features responsible for the cold adaption of mutant H5.

Complex networks, derived from graph theory, has been widely used in diverse fields of science and engineering, such as social network, traffic network, and complex biological systems (Di Paola et al., 2013; Sethi et al., 2009; Yan et al., 2011). The protein structure can be represented by a residue interaction network (RIN), where each amino acid residue is regarded as a node and an edge exists between two residues forming covalent and non‐covalent interactions (Doncheva et al., 2012; Grewal & Roy, 2015) RIN has been utilized in protein folding (Dill et al., 2008; Gromiha, 2009), protein structure prediction (Yan et al., 2011), allosteric communication in proteins (Bhattacharyya & Vishveshwara, 2011; Ghosh & Vishveshwara, 2008; Guo et al., 2015), functional residues, and active site prediction (Chea & Livesay, 2007), and protein mutation effects (Medeiros Almeida et al., 2021). Hubs are those residues that have more contact with other residues (edge >4), which play a crucial role in maintaining the tertiary structure of the proteins, therefore mutations in some crucial hub residues may cause protein instability (Brinda & Vishveshwara, 2005). For example, nodes with a high degree of centrality can be identified as evolutionarily conserved residues (central) and functional residues by studying “network centrality” (Tang et al., 2008). A centrality residue was identified in the HisF protein from Thermotoga maritima, and RIN was used to explore the mutation effects on the protein's thermal stability (Medeiros Almeida et al., 2021). A comparative analysis between 10 thermophile and mesophile protein counterparts revealed that network parameters can account for the thermal stability of thermophilic proteins (Brinda & Vishveshwara, 2005). Due to the consideration of all kinds of interactions, the RIN could better present the global topology of the protein structure, and thus would be more rational to explore the structural effects induced by mutations.

In the present study, MD simulations and RIN were used to further underlying the mechanisms of increased low‐temperature catalytic activity of H5. The dynamic structural parameters from the MD simulations and seven network topology parameters of WT and H5 were computed. The analysis of salt bridges and hydrogen bonds between inter‐protein and water molecules revealed the structural factors contributing to H5 flexibility. Furthermore, the essential dynamics (ED) unraveled the correlated motions of proteins. And the application of RIN enhanced the understanding of the effects of mutation on protein structure and function.

2. MATERIALS AND METHODS

2.1. Preparation of protein structures

The WT protein structure was retrieved from the Protein Data Bank (PDB ID: 1SVN), while the mutant H5 structure was generated utilizing the SWISS‐MODEL tool (Waterhouse et al., 2018) (https://swissmodel.expasy.org/), with 1SVN serving as the template. The three‐dimensional structure of Suc‐Ala‐Ala‐Pro‐Phe‐pNA (AAPF) was obtained from the PubChem database, and the structure was shown in Figure 1. The 3D structure of savinase and the modification of selected residues were shown in Figure 2. Molecular docking was performed using Autodock Vina (Trott & Olson, 2010). The docking box was defined with precise coordinates of 11.191, 23.799, and 10.726 for center_x, center_y, and center_z, respectively. The cubic box had a length of 40 Å. Additionally, an exhaustiveness value of 32 was set to ensure thorough exploration of the docking space. The complex structure with the lowest binding energy and reasonable conformation was selected, and the AAPF PDB file was submitted to the ATB website (Stroet et al., 2018) (http://atb.uq.edu.au/) to obtain the topology file for MD simulations.

FIGURE 1.

FIGURE 1

Open in a new tab

The structure of substrate Suc‐Ala‐Ala‐Pro‐Phe‐pNA.

FIGURE 2.

FIGURE 2

Open in a new tab

The 3D structure of savinase (PDB ID 1SVN), and the mutated residues are represented by magenta sticks, the catalytic triad is represented by cyan sticks, and the protein is shown in a cartoon.

2.2. Molecular dynamics simulations

In MD simulations, 300 K is a relatively suitable simulation temperature to determine the unique characteristics of cold‐adaption of psychrophilic proteins by comparing homologous mesophilic and thermophilic proteins (Papaleo et al., 2011; Tiberti & Papaleo, 2011). At 300 K, WT and H5 have relatively higher enzymatic activity. Meanwhile, from the computational perspective, many force fields are parameterized at 300 K, so the conformational sampling of MD simulations at this temperature is more reasonable and can better reproduce the experimental results. Therefore, 300 K was chosen as the simulated temperature, and MD simulations was performed using the software GROMACS 5.1.4, with a force field of GROMOS96 54A7. The single point charge (SPC) model was used to describe water molecules. The protein was placed in a cube box, and the minimum distance of all residues of the protease from the box boundary was set to 10 Å to ensure that the protein does not interact with itself. Periodic boundary conditions (PBC) was used in the MD simulations.

Energy minimization was performed to eliminate irrational atom–atom contact in the system, and this process was accomplished using a gradient descent of 50,000 steps. The interaction between the short‐distance van der Waals (VDW) force and the electrostatic force was truncated at 10 Å, and the time step was set to 2 fs. After NVT and NPT equilibrium, a 500 ns MD simulation was performed, and the trajectory file was saved every 0.1 ps. The LINCS algorithm was used to constrain all covalent bonds involving hydrogen atoms (Hess et al., 1997). Long‐range electrostatic interactions were calculated using the Particle Mesh Ewald (PME) method. The MD simulations were performed three times with different velocities to ensure reliable results. The analysis was performed with the auxiliary program of Gromacs 5.1.4. The salt bridges were computed with the VMD console. Furthermore, the persistence of the salt bridges was defined as the number of frames at a distance less than 4 Å divided by the total number of frames (Tiberti & Papaleo, 2011). The equilibrium trajectories of the three simulations were concatenated for the analysis of binding free energy between substrate and protein via the g_mmpbsa tool (Kumari et al., 2014).

2.3. Essential dynamics and analysis tools

The principal component analysis (PCA) of MD simulations, also known as essential dynamics (ED), aims to provide detailed information about the conformational changes and large‐scale collective movements of the protein throughout the MD simulations. The covariance matrix was constructed after removing the translation and rotation of the equilibrated portions of concatenated trajectory in three MD simulations. Then the matrix was diagonalized to generate eigenvectors and eigenvalues, where each eigenvector describes the collective motion of the atoms in the conformational space, and the eigenvalue represents the corresponding amplitude of atomic motion along the eigenvector direction. The covariance matrix, eigenvector, and eigenvalues were obtained using gmx_covar and gmx_anaeig, and the Cα atoms were considered when generating the covariance matrix. The free energy landscape (FEL) was also modeled using the gmx_sham module, and the projections of first (PC1) and second (PC2) eigenvectors to show protein conformational space. ED and FEL analysis provide more structural information on WT and H5 mutant under the same conditions.

2.4. Residue interaction network

The residue interaction network was generated using the RING server (Piovesan et al., 2016) by uploading the protein structure, where the residue interaction distance cutoff adopted the default parameters, and the output node and edge file were used for subsequent topological parameter analysis.

The network topology parameters were computed by Gephi software, an open‐source software for the visualization and analysis of networks (Bastian et al., 2009). Seven topological parameters were calculated for WT and H5: network average degree, average weighted degree, graph density, average clustering coefficient, and average path length, respectively.

The average degree of the network is the average of all node degrees, and it is computed according to the following equation:

<k>=∑ikiN

where k i is the degree of node i and N is the number of nodes in a network (i.e., the number of residues in enzymes). The average weighted degree considers the distance between a node with other nodes. The graph density measures the connectivity of the network, which implies the graph density of a complete graph (each pair of graph vertices is connected by an edge) is 1. For node i in a network, the clustering coefficient is defined as the ratio of the number of edges that exist between nodes in the neighborhood to their maximum number of edges. The average clustering coefficient for the network is the average of the clustering coefficients for all nodes. The shortest path length between node i and node j is the distance connecting two nodes, and the average path length of the network is the average distance between any two nodes.

3. RESULTS AND DISCUSSION

3.1. Structural and conformational flexibility evaluation

Since only the apo protein structure was resolved, the complex structure was obtained using molecular docking. Then the complex structure with the best docking score was selected for three MD simulations. The average root mean square deviation (RMSD) of WT and H5 protease were calculated to reveal the protein deviation along simulation time compared with the initial structure. As shown in Figure 3a, the WT was relatively stable throughout the simulation period, while H5 had a large fluctuation, and then remained reasonably stable. The results indicated that H5 experienced a larger structural fluctuation compared with WT. And the respective RMSD values for three MD simulations are shown in Figure S1.

FIGURE 3.

FIGURE 3

Open in a new tab

The structural flexibility of WT and H5. The average RMSD (a) and RMSF (b) fluctuations of WT and H5 in three 500 ns MD simulations. The representative 3D structures of WT (c) and H5 (d) were extracted according to the average backbone RMSF. The color ranges from blue to red, and represents the RMSF varies from the lowest to the highest values, respectively. RMSD, root mean square deviation; RMSF, root mean square fluctuation; WT, wild type.

The average root mean square fluctuation (RMSF) in three MD runs was calculated to uncover the atomic volatility of each residue (Figure 3b), and the higher RMSF values implied higher mobility and lower RMSF values indicated restricted mobility. The average RMSF comparison of the catalytic triad of WT and H5 showed little difference (Figure 4). This observation is expected, as catalytic residues remain rigid to maintain specific conformation for the catalytic role. It implied that the local flexibility of mutant H5 increased, which was mainly located in the segments of S101‐S106, S128‐S132, R186‐F189, and N204‐T208. The two loop regions S101‐S106 and S128‐S132 are involved in modulating the substrate binding. The increased loop flexibility around active sites allows the protein to better adapt to substrate binding and recognition, which can be an important factor for the H5 with high activity at low temperatures. Similarly, the research of Liu et al. indicated that the larger displacement of substrate binding sites may be related to the cold‐adaption of serine proteinase K53. Several cold‐active proteins such as uracil‐DNA glycosylase (Olufsen et al., 2005) and phosphoglycerate kinase (Bentahir et al., 2000) also appear psychrophilic character accompanied by increased flexibility of the active site loop region. In phenylalanine hydroxylase from Colwellia psychrerythraea 34H and isocitrate dehydrogenase from Desulfotalea psychrophila enzymes, the strategy that promotion of the local flexibility around active sites was also adopted to enhance the low‐temperature catalytic activity. Instead, other segments such as residues T71‐V81, and S153‐G157 become more rigid in H5 than in WT. This suggested that H5 employs a local rigidity/flexibility mechanism to regulate temperature adaptation, which aligns with previous studies that several psychrophilic enzymes employ local rigidity/flexibility to regulate temperature adaptation (Isaksen et al., 2016; Sang et al., 2017). The local flexibility/rigidity mechanism indicated that optimizing enzyme function at different temperatures requires an appropriate balance between structural rigidity (maintaining a specific three‐dimensional conformation at physiological temperature) and flexibility (allowing protein structure to perform catalysis and substrate recognition) (Åqvist et al., 2017; Socan et al., 2020). To express the difference in structural fluctuation of WT and H5 intuitively, the backbone flexibility was represented by the average structure derived from MD simulations, where the thickness of the tube represents different degrees of fluctuations (Figure 3c,d).

FIGURE 4.

FIGURE 4

Open in a new tab

The average RMSF values of catalytic triad residues of WT and H5 in three MD simulations. MD, molecular dynamics; RMSF, root mean square fluctuation; WT, wild type.

3.2. Comparison of distance between two substrate‐binding loops

The average distance between the two substrate‐binding loops of WT and H5 was calculated in three MD simulations, and the individual values of the three MD simulations were provided in Figure S1. The distance was defined as the distance between the center of mass of residue G100‐S104 and S125‐S129 domains. As the results depicted in Figure 5, the distance between loop G100‐S104 and loop S125‐S129 in H5 was higher than that of WT, and a representative conformation derived from the MD trajectories indicated that the distance was 12.83 Å and 9.17 Å, respectively. The coordination of these two loop regions away from each other assist the substrate enter the catalytic pocket easily, and the accessibility of active sites may be a contributor to improved catalytic activity at low temperatures. Similarly, MD simulations of acetyl xylan esterase from the Arctic marine bacterium showed that the enzyme increases low‐temperature activity by regulating the distance between the catalytic nucleophilic residue Ser32 and the catalytic residue His203 (Zhang, Ding, et al., 2021).

FIGURE 5.

FIGURE 5

Open in a new tab

The average distance between the substrate‐binding loops G100‐S104 and S125‐S129 in three MD simulations. (a) The plot of average distance as a function of time for the WT and H5 in three MD simulations; a representative snapshot for the 3D structure of WT (b) and H5 (c) taken from the MD simulations, the dotted line indicates the distance between the αC atom of Gly102 residue and the αC atom of Gly127 residue (Å). Enzymes are represented as cartoons and loops G100‐S104 and S125‐S129 segments are shown in sticks. MD, molecular dynamics; WT, wild type.

3.3. Binding free energy of the substrate with enzyme

The binding energy of substrate with enzyme was calculated utilizing g_mmpbsa. As shown in Table 1, the binding free energy of AAPF with WT was higher than that of H5, which was consistent with the experimentally determined higher K m value of H5. The weaker substrate affinity of H5 pointed out that mutant H5 improves the catalytic activity at low temperatures by sacrificing the affinity for the substrate. There are hydrogen bond interactions and hydrophobic interactions between the two loop regions and the substrate. A representative frame is selected to represent the interaction between substrate and protein. In WT, there are hydrogen bond interactions between the substrate and residues G100, S125, and G127, along with hydrophobic interactions with nonpolar residues L124 and L126 (provided in Figure S3). In H5, the hydrogen bond interaction with G100 and S125 is disrupted, so the loss of hydrogen bond interaction may lead to the weakening of binding energy. Furthermore, the binding energy was decomposed to each residue, and the results (provided in Figure S4) revealed that certain amino acid residues such as G100 and S125 showed reduced binding energy.

TABLE 1.

The binding energy between substrate and enzymes.

Protein WT (kJ/mol) H5 (kJ/mol)
van der Waals energy −246.32 ± 10.39 −205.98 ± 16.32
Electrostatic energy −150.29 ± 14.87 −118.66 ± 10.58
Polar solvation energy 255.14 ± 15.14 219.57 ± 17.68
SASA energy −28.25 ± 5.65 −19.68 ± 3.64
Binding energy −169.72 ± 13.92 −124.75 ± 15.95
Open in a new tab

Abbreviation: WT, wild type.

3.4. Hydrogen bonds calculation

The average hydrogen bonds between protein–protein and protein‐solvent were calculated in three MD simulations (Figure 6). In addition, the hydrogen bonds formed in each of the three simulations are provided in Figure S2. The analysis of hydrogen bonding patterns within WT and mutant H5 revealed that there were no obvious differences in the number of intra‐protein hydrogen bonds. However, when examining the hydrogen bonds between the protein and water molecules, it was observed that the H5 protein had more interactions with the solvent compared to the WT protein. Previous studies have also reported that cold‐adapted serine hydroxymethyl transferase enhanced the interaction with solvents to improve their conformational flexibility, which resulted in high catalytic activity at low temperatures (Zhang, Xia, et al., 2021). In addition, statistical analyses of amino acids of the psychrophilic and mesophilic proteins showed that there are more Ser and Thr amino acids in the psychrophilic proteins, which lead to the formation of more interactions between the proteins and water molecules, thus increasing the flexibility of protein structure (Huang et al., 2023; Nath & Subbiah, 2014).

FIGURE 6.

FIGURE 6

Open in a new tab

The number of hydrogen bonds formed between protein–protein (a), protein‐water (b) of WT and mutant H5 in three MD simulations. WT, wild type; MD, molecular dynamics.

3.5. Salt bridges calculation

Salt bridges are also one of the most important factors in stabilizing protein conformation, and the salt bridges in three MD runs of WT and H5 were calculated using VMD software. The results showed that the total number of salt bridges in H5 was less than in WT (Table 2), that is, Asp197‐Arg247 and Asp41‐His39. Especially considering the frequency of the salt bridges, Asp197‐Arg247 salt bridge in WT was very stable, whereas this interaction was absent in H5. Furthermore, the residue Asp197 also formed a salt bridge with Lys251 with a duration of 14% in WT, while in H5 the persistence was only 2%. Asp197 residue played a vital role in the formation of the salt bridge network Lys251‐Asp197‐Arg247, and the absence of this salt bridge network may contribute to the conformational flexibility of H5. The formation of a stable salt bridge between the catalytic residues Asp32 and His64 promoted the stabilization of protonated nitrogen atoms on the histidine imidazole ring by aspartic acid, which in turn helped the process of catalytic reaction.

TABLE 2.

The formed salt bridges and persistency of WT and H5 in MD simulations.

WT Persistency (%) H5 Persistency (%)
Asp32‐His64 100 Asp32‐His64 97
Asp41‐His39 6 0 0
Asp197‐Arg247 89 0 0
Asp197‐Lys251 16 Asp197‐Lys251 3
Glu54‐Lys94 13 Glu54‐Lys94 14
Glu89‐Arg45 15 Glu89‐Arg45 10
Glu89‐Lys27 19 Glu89‐Lys27 9
Glu112‐Arg145 1 Glu112‐Arg145 1
Glu271‐Arg19 1 Glu271‐Arg19 2
Glu271‐Arg275 3 Glu271‐Arg275 6
Open in a new tab

Abbreviation: WT, wild type.

Psychrophilic proteins have evolved different strategies to increase molecular flexibility, one of which is to reduce intramolecular salt bridges. For example, A comparison of homology zinc metalloproteases revealed that the psychrophilic Vibriolysin E495 had 8 fewer salt bridges than the mesophilic Pseudolysin (Xie et al., 2009). As H5 is mutated only by eight amino acid residues from WT, the difference in salt bridges between WT and H5 should not be so significant. But the subtle decreased trend in salt bridges can provide clues about the protein's structural stability.

3.6. Essential dynamics analyses

The last 200 ns of equilibrium trajectories from three MD simulations are joined to perform the ED analyses. The eigenvalues corresponding to the first 30 eigenvectors of WT and H5 were calculated, and from Figure 7 we can see that the first few eigenvectors described the main movement patterns of serine proteases, especially the first five eigenvectors. In particular, the eigenvalue of the first five eigenvectors in H5 was higher than in WT, reflecting larger particle fluctuations along the first few eigenvectors.

FIGURE 7.

FIGURE 7

Open in a new tab

The eigenvalues of the first 30 eigenvectors of WT and H5. WT, wild type.

To visualize the correlated motions throughout the protein, the concatenated MD trajectories were projected onto the first two eigenvectors. Two‐dimensional free energy landscapes were obtained, and PC1 and PC2 derived from the ED analysis were used as collective coordinates to detect the profile of structural characteristics (Figure 8). At 300 K, the landscape of the H5 had several minima, whereas the WT tended to have two local minimum. Engineered H5 possessed larger, more “rugged” and metastable states, but the WT was trapped in the main basin. The landscape distribution of the H5 and WT were resembling the homologous psychrophilic and mesophilic elastase and uracil DNA glycosylases, which revealed that cold‐adapted enzymes are characterized by a large population of substates to promote structural flexibility (Mereghetti et al., 2010). Through the loop region derived from the psychrophilic protease transferred to the mesophilic protease, the modification of several amino acids made the mutant protease H5 undergo richer conformational states. The ED and FEL analysis revealed that H5 possessed larger and more metastable states, which contributed to lowering the energy barrier of conformation conversion.

FIGURE 8.

FIGURE 8

Open in a new tab

The free energy landscape of WT (A) and H5 (B) of joined equilibrium trajectories in three MD simulations, the free energy is represented in kcal/mol. MD, molecular dynamics; WT, wild type.

3.7. Residue interaction network analysis

The RIN has been utilized to explore the protein mutation effects, protein functional residues and active site prediction, protein–protein interaction, and interdomain communications. To investigate the difference of residue interactions between WT and H5, seven network topology parameters were computed and the results were listed in Table 3. As we expected, through modification of several residues, there are differences in internal protein residue‐residue interactions between WT and H5. Compared with H5, WT had a larger number of residue–residue interaction, and the higher value of the average degree and average cluster coefficient in WT reflected the closely packed protein structure. The average path length indicates the connection strength between nodes in the network, and a smaller value means stronger interaction between two nodes. The lower value of the average path length of WT also implied the increased rigidity in protein. The systematical comparison of network topology parameters for thermophilic proteins and mesophilic proteins indicated that there are certain differences between two categories of enzymes. For example, in thermophilic proteins, amino acids are more closely linked and the average degree and average cluster coefficient are higher, whereas the average path length is lower than in mesophilic proteins (Gao & Ding, 2017; Vijayabaskar & Vishveshwara, 2010). Besides, HisF proteins from Thermotoga maritima based on 12 single alanine substitution mutants were investigated using RIN to underlying the role of high centrality residue in protein thermal stability (Medeiros Almeida et al., 2021). Similarly, the complex network approach was used to elucidate the structural basis of enhanced thermostability in the mutant lipase A from Bacillus subtilis (Kandhari & Sinha, 2017). The results revealed that despite the overall structure of the protein not changed, there were subtle changes in the topological parameters of WT and mutants. It also pointed out that the distribution of network parameters of WT and mutants lies in a small range, such as the distribution range of the average degree is 8.291–8.413, and the shortest path length is 4.173–4.239. Consistent with the results of the lipase A, the topological parameters of WT and H5 varied in a relatively narrow range. Nevertheless, the fluctuation of topological parameters could still indicate that the protein structure of H5 became looser than that of WT, and the residue–residue interactions are weakened. These data suggested that the residue interaction network provides an additional technique for describing the differences and the analysis of network parameters is also helpful for understanding the mechanism of temperature adaptation for different enzymes and in engineering specific properties of proteins.

TABLE 3.

The residue interaction network topology parameters of WT and H5.

Enzyme Average degree Average weighted degree Graph density Average cluster coefficient Average path length
WT 1.654 1.882 0.014 0.137 6.169
H5 1.586 1.811 0.013 0.091 6.703
Open in a new tab

Abbreviation: WT, wild type.

4. CONCLUSIONS

In this study, the mechanism of engineered H5 with promoted low‐temperature activity was investigated using MD simulations and RIN. The MD simulations revealed that the more flexible substrate binding loops (G100‐S104 and S125‐S129) and the enlarged distance between these two loop regions made the increment of accessibility for substrate. And the enhanced conformational flexibility for mutant H5 was mainly by decreased salt bridge especially the absence of salt bridge network Lys251‐Asp197‐Arg247, and increased interaction with water molecules. The factors that lead to structural fluctuations are not single, but a combination optimization of multiple strategies. Moreover, ED analysis and FEL indicated that H5 possessed rugged landscape and richer conformational space with low barriers, and this type of energy landscape would facilitate different conformations interconvertibility. Increased conformational flexibility also resulted in lower substrate binding affinity, which is supported by a higher K m value and weaker binding free energy between substrate and variant H5. Additionally, the RIN exploration provides another technique to study the effects of mutation on protein structure. The observed difference between the seven topology parameters among WT and H5 indicated that H5 is more loosely packed. MD simulations coupled with RIN enhanced our understanding of the basis of enzyme cold‐adaption, which aided in the rational design of more efficient cold‐adapted protease.

AUTHOR CONTRIBUTIONS

Fufeng Liu: Writing – review and editing; funding acquisition; supervision. Ailan Huang: Methodology; writing – original draft; data curation; investigation; conceptualization. Fuping Lu: Funding acquisition; formal analysis.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing financial interest.

Supporting information

DATA S1: Supporting Information.

Click here for additional data file. (2MB, docx)

ACKNOWLEDGMENTS

This work was supported by the National Key R&D Program of China (2021YFC2102700) and National Natural Science Foundation of China (No. 32272269).

Huang A, Lu F, Liu F. Exploring the molecular mechanism of cold‐adaption of an alkaline protease mutant by molecular dynamics simulations and residue interaction network. Protein Science. 2023;32(12):e4837. 10.1002/pro.4837

Review Editor: Nir Ben‐Tal

REFERENCES

  1. Almog O, Gonzalez A, Godin N, De Leeuw M, Mekel MJ, Klein D. The crystal structures of the psychrophilic subtilisin S41 and the mesophilic subtilisin Sph reveal the same calcium‐loaded state. Proteins. 2009;74(2):489–496. [DOI] [PubMed] [Google Scholar]
  2. Åqvist J, Isaksen GV, Brandsdal BO. Computation of enzyme cold adaptation. Nat Rev Chem. 2017;1(7):0051. [Google Scholar]
  3. Arnorsdottir J, Kristjansson MM, Ficner R. Crystal structure of a subtilisin‐like serine proteinase from a psychrotrophic Vibrio species reveals structural aspects of cold adaptation. FEBS J. 2005;272(3):832–845. [DOI] [PubMed] [Google Scholar]
  4. Barzkar N. Marine microbial alkaline protease: an efficient and essential tool for various industrial applications. Int J Biol Macromol. 2020;161:1216–1229. [DOI] [PubMed] [Google Scholar]
  5. Bastian M, Heymann S, Jacomy M. Gephi: an open source software for exploring and manipulating networks. Proceedings of the international AAAI conference on web and social media. 2009.
  6. Bentahir M, Feller G, Aittaleb M, Lamotte‐Brasseur J, Himri T, Chessa JP, et al. Structural, kinetic, and calorimetric characterization of the cold‐active phosphoglycerate kinase from the Antarctic pseudomonas sp TACII18. J Biol Chem. 2000;275(15):11147–11153. [DOI] [PubMed] [Google Scholar]
  7. Betzel C, Klupsch S, Papendorf G, Hastrup S, Branner S, Wilson KS. Crystal structure of the alkaline proteinase Savinase™ from Bacillus lentus at 1.4 Å resolution. J Mol Biol. 1992;223(2):427–445. [DOI] [PubMed] [Google Scholar]
  8. Bhattacharyya M, Vishveshwara S. Probing the allosteric mechanism in pyrrolysyl‐tRNA synthetase using energy‐weighted network formalism. Biochemistry. 2011;50(28):6225–6236. [DOI] [PubMed] [Google Scholar]
  9. Brinda KV, Vishveshwara S. A network representation of protein structures: implications for protein stability. Biophys J. 2005;89(6):4159–4170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chea E, Livesay DR. How accurate and statistically robust are catalytic site predictions based on closeness centrality? BMC Bioinform. 2007;8:153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen K, Mo Q, Liu H, Yuan F, Chai H, Lu F, et al. Identification and characterization of a novel cold‐tolerant extracellular protease from Planococcus sp. CGMCC 8088. Extremophiles. 2018;22(3):473–484. [DOI] [PubMed] [Google Scholar]
  12. Collins T, Roulling F, Piette F, Marx JC, Feller G, Gerday C, et al. Fundamentals of cold‐adapted enzymes. Psychrophiles: From Biodiversity to Biotechnology. Berlin, Heidelberg: Springer; 2008. p. 211–227. [Google Scholar]
  13. Di Paola L, De Ruvo M, Paci P, Santoni D, Giuliani A. Protein contact networks: an emerging paradigm in chemistry. Chem Rev. 2013;113(3):1598–1613. [DOI] [PubMed] [Google Scholar]
  14. Dill KA, Ozkan SB, Shell MS, Weikl TR. The protein folding problem. Annu Rev Biophys. 2008;37:289–316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Doncheva NT, Assenov Y, Domingues FS, Albrecht M. Topological analysis and interactive visualization of biological networks and protein structures. Nat Protoc. 2012;7(4):670–685. [DOI] [PubMed] [Google Scholar]
  16. East NJ, Clifton BE, Jackson CJ, Kaczmarski JA. The role of oligomerization in the optimization of cyclohexadienyl dehydratase conformational dynamics and catalytic activity. Protein Sci. 2022;31(12):e4510. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Ellaiah P, Srinivasulu B, Adinarayana K. A review on microbial alkaline proteases. J Sci Ind Res. 2002;61(9):690–704. [Google Scholar]
  18. Feller G. Psychrophilic enzymes: from folding to function and biotechnology. Scientifica (Cairo). 2013;2013:512840. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Feller G, Gerday C. Psychrophilic enzymes: hot topics in cold adaptation. Nat Rev Microbiol. 2003;1(3):200–208. [DOI] [PubMed] [Google Scholar]
  20. Feng SS, Pumroy RA, Protopopova AD, Moiseenkova‐Bell VY, Im W. Modulation of TRPV2 by endogenous and exogenous ligands: a computational study. Protein Sci. 2023;32(1):e4490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gao XM, Ding YR. Using the residue interaction network improve the classification of thermophilic and mesophilic proteins. Curr Bioinforma. 2017;12(3):249–257. [Google Scholar]
  22. Ghosh A, Vishveshwara S. Variations in clique and community patterns in protein structures during allosteric communication: investigation of dynamically equilibrated structures of methionyl tRNA synthetase complexes. Biochemistry. 2008;47(44):11398–11407. [DOI] [PubMed] [Google Scholar]
  23. Grewal RK, Roy S. Modeling proteins as residue interaction networks. Protein Pept Lett. 2015;22(10):923–933. [DOI] [PubMed] [Google Scholar]
  24. Gromiha MM. Multiple contact network is a key determinant to protein folding rates. J Chem Inf Model. 2009;49(4):1130–1135. [DOI] [PubMed] [Google Scholar]
  25. Guo J, Pang X, Zhou HX. Two pathways mediate interdomain allosteric regulation in pin1. Structure. 2015;23(1):237–247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Gupta R, Beg QK, Khan S, Chauhan B. An overview on fermentation, downstream processing and properties of microbial alkaline proteases. Appl Microbiol Biotechnol. 2002;60(4):381–395. [DOI] [PubMed] [Google Scholar]
  27. Gupta R, Beg QK, Lorenz P. Bacterial alkaline proteases: molecular approaches and industrial applications. Appl Microbiol Biotechnol. 2002;59(1):15–32. [DOI] [PubMed] [Google Scholar]
  28. Hedstrom L. Serine protease mechanism and specificity. Chem Rev. 2002;102(12):4501–4524. [DOI] [PubMed] [Google Scholar]
  29. Hess B, Bekker H, Berendsen HJC, Fraaije JGEM. LINCS: a linear constraint solver for molecular simulations. J Comput Chem. 1997;18(12):1463–1472. [Google Scholar]
  30. Hu C, Ye BJ, Huang ZD, Chen F. Development of an engineered ketoreductase with improved activity, stereoselectivity and relieved substrate inhibition for enantioselective synthesis of a key (R)‐alpha‐lipoic acid precursor. Mol Catal. 2022;522:522. [Google Scholar]
  31. Huang A, Lu F, Liu F. Discrimination of psychrophilic enzymes using machine learning algorithms with amino acid composition descriptor. Front Microbiol. 2023;14:1130594. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Huang AL, Chai CC, Zhang JY, Zhao L, Lu FP, Liu FF. Engineered N57P variant of Ulvan Lyase with improvement of catalytic efficiency and Thermostability via reducing loop flexibility and anchoring substrate. ACS Sustain Chem Eng. 2021;9(48):16415–16423. [Google Scholar]
  33. Isaksen GV, Åqvist J, Brandsdal BO. Enzyme surface rigidity tunes the temperature dependence of catalytic rates. Proc Natl Acad Sci U S A. 2016;113(28):7822–7827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kandhari N, Sinha S. Complex network analysis of thermostable mutants of Bacillus subtilis lipase a. Appl Netw Sci. 2017;2(1):18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kumari R, Kumar R, Open Source Drug Discovery C , Lynn A. g_mmpbsa—a GROMACS tool for high‐throughput MM‐PBSA calculations. J Chem Inf Model. 2014;54(7):1951–1962. [DOI] [PubMed] [Google Scholar]
  36. Lilina AV, Leekens S, Hashim HM, Vermeire PJ, Harvey JN, Strelkov SV. Stability profile of vimentin rod domain. Protein Sci. 2022;31(12):e4450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Liu YH, Zhang T, Zhang ZM, Sun TY, Wang JL, Lu FP. Improvement of cold adaptation of Bacillus alcalophilus alkaline protease by directed evolution. J Mol Catal B‐Enzym. 2014;106:117–123. [Google Scholar]
  38. Medeiros Almeida V, Chaudhuri A, Cangussu Cardoso MV, Matsuyama BY, Monteiro Ferreira G, Goulart Trossini GH, et al. Role of a high centrality residue in protein dynamics and thermal stability. J Struct Biol. 2021;213(3):107773. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mereghetti P, Riccardi L, Brandsdal BO, Fantucci P, De Gioia L, Papaleo E. Near native‐state conformational landscape of psychrophilic and mesophilic enzymes: probing the folding funnel model. J Phys Chem B. 2010;114(22):7609–7619. [DOI] [PubMed] [Google Scholar]
  40. Nath A, Subbiah K. Inferring biological basis about psychrophilicity by interpreting the rules generated from the correctly classified input instances by a classifier. Comput Biol Chem. 2014;53:198–203. [DOI] [PubMed] [Google Scholar]
  41. Olufsen M, Smalas AO, Moe E, Brandsdal BO. Increased flexibility as a strategy for cold adaptation: a comparative molecular dynamics study of cold‐ and warm‐active uracil DNA glycosylase. J Biol Chem. 2005;280(18):18042–18048. [DOI] [PubMed] [Google Scholar]
  42. Papaleo E, Tiberti M, Invernizzi G, Pasi M, Ranzani V. Molecular determinants of enzyme cold adaptation: comparative structural and computational studies of cold‐ and warm‐adapted enzymes. Curr Protein Pept Sci. 2011;12(7):657–683. [DOI] [PubMed] [Google Scholar]
  43. Pathak AP, Rathod MG. A review on alkaline protease producers and their biotechnological perspectives. Indian J Geo‐Mar Sci. 2018;47(6):1113–1119. [Google Scholar]
  44. Piovesan D, Minervini G, Tosatto SC. The RING 2.0 web server for high quality residue interaction networks. Nucleic Acids Res. 2016;44(W1):W367–W374. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sang P, Du X, Yang LQ, Meng ZH, Liu SQ. Molecular motions and free‐energy landscape of serine proteinase K in relation to its cold‐adaptation: a comparative molecular dynamics simulation study and the underlying mechanisms. RSC Adv. 2017;7(46):28580–28590. [Google Scholar]
  46. Sethi A, Eargle J, Black AA, Luthey‐Schulten Z. Dynamical networks in tRNA:protein complexes. Proc Natl Acad Sci U S A. 2009;106(16):6620–6625. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sharma M, Gat Y, Arya S, Kumar V, Panghal A, Kumar A. A review on microbial alkaline protease: an essential tool for various industrial approaches. Ind Biotechnol. 2019;15(2):69–78. [Google Scholar]
  48. Siddiqui KS, Cavicchioli R. Cold‐adapted enzymes. Annu Rev Biochem. 2006;75:403–433. [DOI] [PubMed] [Google Scholar]
  49. Siezen RJ, Leunissen JA. Subtilases: the superfamily of subtilisin‐like serine proteases. Protein Sci. 1997;6(3):501–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Socan J, Purg M, Aqvist J. Computer simulations explain the anomalous temperature optimum in a cold‐adapted enzyme. Nat Commun. 2020;11(1):2644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Stroet M, Caron B, Visscher KM, Geerke DP, Malde AK, Mark AE. Automated topology builder version 3.0: prediction of solvation free enthalpies in water and hexane. J Chem Theory Comput. 2018;14(11):5834–5845. [DOI] [PubMed] [Google Scholar]
  52. Tang YR, Sheng ZY, Chen YZ, Zhang Z. An improved prediction of catalytic residues in enzyme structures. Protein Eng Des Sel. 2008;21(5):295–302. [DOI] [PubMed] [Google Scholar]
  53. Tao Y, Rao ZH, Liu SQ. Insight derived from molecular dynamics simulation into substrate‐induced changes in protein motions of proteinase K. J Biomol Struct Dyn. 2010;28(2):143–158. [DOI] [PubMed] [Google Scholar]
  54. Tiberti M, Papaleo E. Dynamic properties of extremophilic subtilisin‐like serine‐proteases. J Struct Biol. 2011;174(1):69–83. [DOI] [PubMed] [Google Scholar]
  55. Tindbaek N, Svendsen A, Oestergaard PR, Draborg H. Engineering a substrate‐specific cold‐adapted subtilisin. Protein Eng Des Sel. 2004;17(2):149–156. [DOI] [PubMed] [Google Scholar]
  56. Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Tsigos I, Mavromatis K, Tzanodaskalaki M, Pozidis C, Kokkinidis M, Bouriotis V. Engineering the properties of a cold active enzyme through rational redesign of the active site. Eur J Biochem. 2001;268(19):5074–5080. [DOI] [PubMed] [Google Scholar]
  58. Vijayabaskar MS, Vishveshwara S. Comparative analysis of thermophilic and mesophilic proteins using protein energy networks. BMC Bioinform. 2010;11(Suppl 1):S49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Waterhouse A, Bertoni M, Bienert S, Studer G, Tauriello G, Gumienny R, et al. SWISS‐MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res. 2018;46(W1):W296–W303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Wintrode PL, Miyazaki K, Arnold FH. Cold adaptation of a mesophilic subtilisin‐like protease by laboratory evolution. J Biol Chem. 2000;275(41):31635–31640. [DOI] [PubMed] [Google Scholar]
  61. Wu XW, Jiang YY, Wang ZY, Yu XB, Sun ZT, Luo W. Enhanced thermostability of formate dehydrogenase via semi‐rational design. Mol Catal. 2022;530:112628. [Google Scholar]
  62. Xie BB, Bian F, Chen XL, He HL, Guo J, Gao X, et al. Cold adaptation of zinc metalloproteases in the thermolysin family from deep sea and arctic sea ice bacteria revealed by catalytic and structural properties and molecular dynamics: new insights into relationship between conformational flexibility and hydrogen bonding. J Biol Chem. 2009;284(14):9257–9269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Xu C, Suo HB, Xue Y, Qin J, Chen HY, Hu Y. Experimental and theoretical evidence of enhanced catalytic performance of lipase B from Candida Antarctica acquired by the chemical modification with amino acid ionic liquids. Mol Catal. 2021;501:111355. [Google Scholar]
  64. Yan Y, Zhang S, Wu FX. Applications of graph theory in protein structure identification. Proteome Sci. 2011;9(Suppl 1):S17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Zhang Y, Ding HT, Jiang WX, Zhang X, Cao HY, Wang JP, et al. Active site architecture of an acetyl xylan esterase indicates a novel cold adaptation strategy. J Biol Chem. 2021;297(1):100841. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Zhang ZB, Xia YL, Dong GH, Fu YX, Liu SQ. Exploring the cold‐adaptation mechanism of serine hydroxymethyltransferase by comparative molecular dynamics simulations. Int J Mol Sci. 2021;22(4):1781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Zhao HY, Feng H. Engineering Bacillus pumilus alkaline serine protease to increase its low‐temperature proteolytic activity by directed evolution. BMC Biotechnol. 2018;18(1):34. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

DATA S1: Supporting Information.

Click here for additional data file. (2MB, docx)

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES