What is the computational chemistry approach in industrial enzyme research?
The paper introduces the main computational chemistry methods used in industrial enzyme research, especially for guiding enzyme engineering, including molecular mechanics force-field and molecular dynamics simulations, quantum mechanics and combined quantum mechanics/molecular mechanics modeling, continuum electrostatic modeling, and molecular docking. These methods are summarized in the text from two perspectives, respectively: first, the basic concepts of the methods themselves, the results of the original calculations, the conditions of applicability, and the advantages and disadvantages; and second, the valuable information obtained from the calculations to guide the design of mutants and mutant libraries.
The industrial application of enzymes has a hundred-year history, and enzyme catalysis is widely used in different industries such as food, agriculture, medicine, and chemical industry because of its high efficiency, high specificity and selectivity, and environmental friendliness [1-2]. Since the industrial application environment is far different from the environment in which enzymes live in nature, the nature and catalytic function of natural enzymes and their application environment usually do not match or are not optimal. In this case, it is necessary to modify the natural amino acid sequence of the enzyme with the help of enzyme engineering to improve its performance [3]. The most commonly used enzyme engineering strategy is to construct mutation libraries for screening, i.e., laboratory directed evolution [4]. One of the necessary prerequisites for effective directed evolution is that the library capacity (i.e., the number of mutants contained in the library) of the mutant library undergoing screening is sufficiently large relative to the proportion of potentially beneficial mutants in the library. The size of a mutant library is often limited by the screening method, available resources, and other objective conditions. The key issue is how to increase the percentage of effective mutants in the mutant library. In-depth understanding of the relationship between enzyme sequence, structure and important properties can help to identify mutation hotspots, limit the scope of mutation and realize quality mutant library design. Computational chemistry methods are an important means of gaining this understanding. It has been shown that computationally designed protein mutation libraries based on computational design can increase the percentage of effective mutants by several orders of magnitude relative to random mutation libraries [5]. For some difficult enzyme engineering or protein engineering topics, the substantial improvement that computation can bring may be enough to determine the ultimate success or failure of the topic, which is no longer limited to the efficiency improvement. In fact, computational chemistry and computational biology methods have successfully enabled the design of artificial enzymes from scratch with catalytic functions that natural enzymes do not possess. Since other reviews in this album have been devoted to methods for automated optimal design of amino acid sequences, this paper will focus on computational methods for simulating and analyzing enzymes with a given amino acid sequence. Of course, researchers can use these methods to study wild type and mutants separately and then compare the results.
The study of proteins, especially enzymes, has long been an important frontier of computational chemistry research [6-8]. The main methods include molecular dynamics simulations (classical MD) based on classical molecular mechanics force fields [9], quantum mechanics (QM) [10] and combined quantum mechanics/molecular mechanics (QM/MM) methods [8,11-12], intermolecular complex prediction, i.e., molecular docking (Docking) [13], and polarizable continuum model (PCM) to quantify electrostatic and solvent effects [ 14]. ) such as the Poisson-Boltzmann model (PB) [14], and some models based on geometrical properties. In this paper, we will give an overview of each of these methods from two perspectives: first, about the methods themselves, including the basic principles, original computational results, applicability conditions, and (potential) advantages and disadvantages, etc.; and second, about how to use these methods to obtain engineering-relevant information, such as a deeper understanding of catalytically relevant mechanisms, theoretical predictions or explanations of changes in the properties or functions of different mutants with respect to the wild-type, which can guide the design of high-quality mutation libraries for directed evolution, or the design of high quality mutation libraries, or the design of mutation banks. design of high-quality mutation libraries, or suggesting specific mutation sites and mutation types based on the analysis of raw computational results, etc.
1
Molecular dynamics simulation (MD) based on classical molecular mechanics force fields (MM)
1.1
Introduction to the method
We do not consider the chemical changes involved in enzyme catalysis for the time being, but only the processes of enzyme conformational changes, formation and dissociation of non-covalent complexes between the enzyme and the reactants (or products) due to molecular thermal motion. During these processes, the electronic state of the molecule does not change (e.g., no covalent bonds are broken or created), and the molecular mechanics force field model applies. The so-called molecular mechanics force field is an empirical mathematical function that expresses the dependence of the potential energy of a molecular system on the geometrical configuration (i.e., the spatial coordinates of all the atoms that make up the molecular system) (Fig. 1A). In other words, if we use X to represent the spatial coordinates of all atoms and VMM (X ) to represent the molecular force field potential energy, the potential energy changes as the molecule changes from one conformation X1 to another X2:
∆VMM = VMM ( X2 ) – VMM ( X1 ).
According to thermodynamic theory, the atoms in a molecule are always in thermal motion, i.e., X is constantly changing with time; moreover, when we make experimental observations, the sample always consists of a large number of molecules (with the exception of single-molecule experiments), with different molecules in different conformational states. Therefore, from a kinetic point of view, we need to consider the conformational changes over time, and from a thermodynamic point of view, we need to consider the probability distribution of molecules with different conformations. Molecular dynamics simulation (MD) is the most straightforward model to examine these two aspects of properties (Fig. 1B). In the MD simulation, we start from an initial conformation, calculate the force acting on each atom at each time point based on the current conformation and the potential energy function (the force is the negative derivative of the potential energy function with respect to the coordinates of the atoms), numerically integrate Newton’s equations of motion to obtain the conformation at the next time point, and repeat the process to obtain the trajectory of the conformation evolution over time.
In between, special algorithms can be used to simulate the effect of environmental factors (e.g., temperature, pressure, etc.) on molecular motion. According to the thermodynamic principle, when the time interval is long enough, the probability distribution of the conformation of the same molecule at different time points and the conformation of different molecules in the thermodynamic equilibrium state are the same (i.e., time averaging is equivalent to system averaging). Therefore, if the MD simulation is performed for a long enough period of time, the set of conformations obtained from the simulation can be used as a sample of the distribution of molecular conformations at a particular thermodynamic equilibrium state. Based on this principle, we can analyze arbitrary observable properties of a system in its thermodynamic equilibrium state based on the time trajectories obtained by MD.
MD provides a powerful computational tool to comprehensively analyze the kinetic change process of conformational changes and important microscopic quantum thermodynamic distributions at atomic resolution, which is particularly important for elucidating the design principles and working mechanisms of complex biomacromolecular machines such as enzymes. Since the current experimental methods for macromolecular structure analysis can only provide spatio-temporally averaged static structures, MD simulations have an irreplaceable function in related research. Under this premise, the MD tool itself is still in the process of continuous improvement and refinement. Methodologically, the main limitations of MD come from two aspects: first, the accuracy of the molecular force field model; and second, the limited simulation time makes it difficult to realize the full sampling of the conformational space. For the first problem, the molecular force field has been greatly improved in recent years, and the accuracy of the thermodynamic description of the conformational equilibrium of biological macromolecules, especially protein systems, has been increased, successfully simulating a variety of protein molecules from the process of head-folding to the natural structure [15-16]. In terms of simulation time, due to the development of computer hardware and software, it is now possible to accomplish simulations on the order of microseconds for systems of usual size (e.g., enzyme molecules of hundreds of residues in aqueous solution) using conventional computing hardware (e.g., multi-core servers used by the group). At this time scale, processes such as opening and closing of structural domains or ring regions can be observed. If more computational resources are available, direct simulations of processes such as substrate binding/dissociation can also be realized. To study processes at time scales beyond the reach of the simulation (e.g., large-scale functional changes of allosteric proteins, etc.), enhanced sampling methods can be used [17], provided that the user has a more in-depth understanding of MD theory.
Currently, most MD simulation applications cover time scales from nanoseconds to microseconds, and sampling of the conformational space is mostly restricted to the vicinity of the initial structure (for single-domain proteins, it is usually the structural rise and fall of the root-mean-square (RMS) displacement in the 3-4 Å range). Therefore, it is necessary to use a reasonable initial structure as input to the MD for the simulation results to be meaningful. In most cases, experimentally determined crystal structures or structures based on comparative modeling of homologous proteins are used as initial structures for MD. When simulating enzyme-substrate complexes, it is often necessary to model the initial structure of the complex based on the structure of the empty enzyme or of the enzyme in complex with other molecules, either by using molecular docking or by directly replacing other small molecules (e.g., inhibitors) in the crystal structure with the substrate. MD simulations also require the construction of a molecular force field that portrays all chemical units in the system. When the system to be simulated includes a small molecule as a substrate, it is often the case that the standard molecular force field provided in the MD software package does not cover the small molecule as a substrate. In this case, tool software that can automatically generate force fields for small molecules can be used [18-19]. The force field files should be manually checked and used for short simulation trials before using the automatically generated force fields for long MD simulations.
Fig. 1 Molecular mechanics force field (A) and molecular dynamics simulation (B)
Fig. 1 Molecular mechanics force field (A) and molecular dynamics simulation (B)
1.2
Application to enzyme research
Information obtained from MD simulations can be applied in different ways to guide enzyme engineering modifications [20]. For example, by comparing room temperature and high temperature MD simulations, it is possible to predict which regions of the enzyme molecule are likely to have the most sensitive structural stability to ambient temperature. Introducing proline point mutations, disulfide bonds, etc. in these regions could potentially enhance the heat resistance of the enzyme [21-24]. Another strategy to improve stability is to design mutants that form more surface hydrogen and salt bonds [25-26]. Before experimentally validating such mutants, wild-type and mutants can be simulated in parallel to theoretically assess whether the mutation may achieve the desired effect [27-28]. In addition to temperature, MD can be used to analyze the effect of changes in environmental pH, solvents, etc. on protein conformation and its stability [29-30].
In addition to stability, MD has been applied to predict hotspot residues that have the potential to significantly affect conformational dynamics associated with substrate binding/product release, providing a basis for designing mutations or mutant libraries that can alter substrate selectivity, reaction selectivity, product release rate, etc [31-32]. One of the ways to study substrate/reaction selectivity with MD is to compare the simulation results of enzyme-substrate complexes with different (initial structures) and predict substrate or structural states with higher affinity (or higher reactivity). A rigorous quantitative method for calculating affinity (or reactivity) is free energy calculations [33-34]. Due to the computationally intensive nature of free energy calculations, most current applications use qualitative methods for prediction: qualitative discrimination of relative affinity can be based on the stability of the structure of the small-molecule-macromolecule complexes, the average intermolecular interaction energy, etc., while qualitative discrimination of reactivity is based on the relative geometric configuration distributions of the catalytic and reactive functional groups, etc. [35]. The results of such qualitative discrimination can be used as a basis for designing libraries of directed evolutionary sequences. In addition, MD simulations can also be used to analyze hotspot residues around the substrate binding/product dissociation pore [36-37]. This type of application involves the simulation of dissociation pathways for the dissociation of small molecules from proteins, and if there are difficulties with insufficient simulation time scales, these can be overcome using enhanced sampling techniques [38-39].
2
Quantum mechanics (QM) and combined quantum mechanics/molecular mechanics (QM/MM) models
2.1
Introduction to the method
To simulate chemical steps in enzyme catalysis, such as covalent bond creation and breaking, electron transfer, and leaps between different electronic states, quantum mechanical (QM) models are required. Currently, QM models commonly used in computational chemistry are classified into several types such as ab initio ( ab initio ), density-functional theory (DFT), and semi-empirical methods [40]. Among them, semi-empirical methods are the least computationally expensive. However, they are non-first-principles methods and the reliability of the computational results is highly dependent on the particular system and problem. Both ab initio and DFT methods are first-principle methods and are universal. Practical DFT models may contain more empirical theoretical approximations than ab initio calculations, but DFT can handle electron correlation energies with very high computational efficiency. In addition, for many chemical reaction problems, the computational errors of the best DFT models for key parameters such as the energy changes during the reaction process are already as small as about 1 kcal/mol or so, and the results are sufficient to be used as a basis for determining the chemical soundness of a particular catalytic mechanism or reaction pathway.
Given the geometrical configuration of a molecule, its energy can be calculated using QM. This is known as a single-point calculation (i.e., it deals with only one point in the geometrical configuration space.) The QM model is more often used for the optimization of molecular geometries, i.e., to find a locally stable structure (with a lower energy than the neighboring structures) after successive changes from an initial configuration, or to find the lowest-energy paths connecting the reactants to the products, and the transition states along the paths. These calculations are computationally intensive as different geometrical configurations have to be considered and compared, and typically tens to thousands of single-point calculations have to be performed. A common strategy to save computational effort is to first optimize a wide range of reaction path searches using efficient QM models with limited accuracy, and then complete the configuration optimization using higher accuracy models near the lowest energy configuration/path searched for, or to perform single-point computations.
Currently, the application of first-principle QM methods to the whole enzyme molecule is computationally intensive, basically limited to single-point calculations, and still lacks practicality. The QM/MM model (Fig. 2) is commonly used for large molecules [11]. In this model, the molecular system is divided into at least two parts: the part directly involved in the chemical reaction is treated with the QM model, and the rest is treated with molecular mechanics (MM). There are several different strategies to deal with QM-MM boundaries and interactions [41]. In first-principles QM/MM models, QM calculations are far more costly than MM. therefore, conformational optimization methods are mostly used for the QM region to predict or simulate its geometry, and molecular dynamics simulations can be used to sample the MM part [42]. This means that the computational results may be more sensitive to the initial structure of the QM region of the system. In this case, calculations with different initial structure models are required to obtain reliable results. If semi-empirical methods [43] or empirical valence bond theories [44-45] are used for the QM part , it may be possible to explore the conformational space more fully and reduce the effect of the initial structure by longer QM/MM MD sampling.
Fig. 2 Quantum mechanical (QM)/Molecular Mechanics (MM) models
Fig. 2 Quantum mechanical (QM)/molecular mechanical (MM) model.
2.2
Application of the method
Both the QM model [10] and the QM/MM model [41] have been widely used for theoretical prediction and testing of the chemical mechanisms of enzyme-catalyzed reactions. Their results can help us to discern which key residues participate in the chemical reaction process, find the rate-limiting step of the reaction, model the structure of reaction intermediates and transition states, analyze how they interact with the enzyme environment, etc. Compared with the QM cluster model, the QM/MM model can more realistically simulate the enzyme environment in which the chemical reaction centers are located.QM/MM has been widely used to theoretically predict/test the chemical mechanism of enzyme catalysis and to analyze and predict the possible effects of environmental amino acid residues on the catalytic process [46]. In principle, these results can be used to guide the design of directed evolution mutation libraries with the goal of enhancing catalytic activity and altering specificity or selectivity. A more challenging study would be to obtain entirely new artificial enzymes based on the design of new active centers from scratch from QM or QM/MM predicted transition state structure models [47].
3
Electrostatic continuum media modeling
3.1
Principle of the method
Enzyme catalysis is almost always accomplished in a specific solution environment. Solvent effects have a crucial influence on enzyme properties. Models for calculating the solvent effects of chemical treatments fall into two categories: explicit solvent models, e.g., in molecular mechanics force fields or QM models, where each solvent molecule and each atom in it is explicitly included in the model; and implicit solvent or continuum medium models [48], where the solvent molecules and atoms are not included in the model and the so-called “solvent mean fields” are used to deal with solvent effects. which does not include solvent molecules and atoms, but treats solvent effects with a so-called “solvent mean field”. The advantage of the explicit solvent model is that it is able to treat solute and solvent in a fully consistent manner, realistically modeling specific interactions such as solute-solvent hydrogen bonding, salt bonding, and so on. The disadvantage is that the number of solvent molecules is large and computationally intensive. In addition, the solvent stochastic rise contributes significantly to the total energy of the system, and long simulation sample averaging must be performed to eliminate the effect of the rise. The hidden solvent model portrays the averaging effect of the solvent, and the thermodynamic rise and fall of the solvent has been averaged.
For simplicity of treatment, we usually separate the non-polar solvent effect (hydrophobic effect) from the polar solvent effect in the hidden solvent model. Experience has shown that the free energy of solvation of a nonpolar solute is proportional to its solvent accessible surface area (SASA). Therefore, the SASA solvation model is often used for this component. Parameters in this model include the atomic radii required to calculate the SASA, the radius of the solvent molecule (1.4 Å for water molecules), and a proportionality constant for the solvation free energy proportional to the SASA. These parameters are generally determined by fitting experimental values of the solvation free energy of small molecules.
The most commonly used models to consider polar solvent effects treat the region occupied by the solvent as a continuous medium with a specific dielectric constant (78.4 for water), and the solute region as being occupied by a medium with a low dielectric constant (commonly valued at 2-8) or a vacuum (dielectric constant of 1) (Fig. 3A). The continuous medium is polarized by the electrostatic field generated by the charge distribution in the solute region, and the resulting polarized charge distribution in turn generates an electrostatic field in the solute region that acts on the solute charge. The electric field generated by the polarized charges is called the reaction field. Therefore, the electrostatic continuous medium model is also known as the reaction field model. In the continuous medium model with no free ions in the solvent region, the relationship between the space electrostatic potential and the space charge distribution satisfies the Poisson equation. For a solution environment containing free ions, the spatial distribution of ions is affected by the spatial electrostatic potential. Considering this factor, the relationship between the spatial electrostatic potential and the spatial charge distribution satisfies the Poisson-Boltzmann equation (PB equation).The PB equation is a partial differential equation concerning the relationship between the distribution of the electrostatic potential and the distribution of the charge and dielectric in three-dimensional space, and can be solved numerically. The most common numerical method for solving the PB equation for macromolecular systems such as enzymes is the finite difference method (FD), collectively referred to as the FDPB model (Figure 3B) [14]. With FDPB it is possible to calculate the electrostatic potential in three dimensions based on the space charge distribution of the solute, which in turn allows the calculation of other properties such as the electrostatic free energy. In QM calculations for small molecule systems, the reaction field is often equivalently replaced by the electric field generated by the surface charge distribution on the molecular surface, and the corresponding model is called the polarizable continuous medium (PCM) model.
The most commonly used model to account for polar solvent effects treats the region occupied by the solvent as a continuous medium with a specific dielectric constant (78.4 for water), while the solute region is treated as if it were occupied by a medium with a low dielectric constant (commonly valued at 2-8) or a vacuum (dielectric constant of 1) (Fig. 3A). The continuous medium is polarized by the electrostatic field generated by the charge distribution in the solute region, and the resulting polarized charge distribution in turn generates an electrostatic field in the solute region that acts on the solute charge. The electric field generated by the polarized charges is called the reaction field. Therefore, the electrostatic continuous medium model is also known as the reaction field model. In the continuous medium model with no free ions in the solvent region, the relationship between the space electrostatic potential and the space charge distribution satisfies the Poisson equation. For a solution environment containing free ions, the spatial distribution of ions is affected by the spatial electrostatic potential. Considering this factor, the relationship between the spatial electrostatic potential and the spatial charge distribution satisfies the Poisson-Boltzmann equation (PB equation).The PB equation is a partial differential equation concerning the relationship between the distribution of the electrostatic potential and the distribution of the charge and dielectric in three-dimensional space, and can be solved numerically. The most common numerical method for solving the PB equation for macromolecular systems such as enzymes is the finite difference method (FD), collectively referred to as the FDPB model (Figure 3B) [14]. With FDPB it is possible to calculate the electrostatic potential in three dimensions based on the space charge distribution of the solute, which in turn allows the calculation of other properties such as the electrostatic free energy. In QM calculations of small molecular systems, the reaction field is often equivalently replaced by the electric field generated by the surface charge distribution on the surface of the molecule, and the corresponding model is called the polarizable continuous medium (PCM) model.
Fig. 3 Electrostatic continuum model (A) and finite-difference Poisson-Boltzmann (FDPB) method (B)
Fig. 3 Electrostatic continuum model (A) and the finite difference Poisson-Boltzmann (FDPB) method (B).
3.2
Application of the method
One of the important applications of the continuum model is to study the protonation state of charged amino acid side chain groups in enzyme molecules. The software PROPKA predicts the pKa of each dissociable group by solving the PB equation to calculate the electrostatic free energy for different protonation states [50]. The surface electrostatic potential distribution of an enzyme molecule is an important factor affecting the substrate selectivity of the enzyme. Given the spatial structure and protonation state of an enzyme molecule, the surface electrostatic potential distribution of an enzyme molecule can be calculated by the FDPB method, which also predicts the effect of mutations in amino acid residues or changes in the environmental pH, changes in ionic concentration, etc., on the surface electrostatic potential [14].
When studying the chemical steps of enzyme catalysis with the QM cluster model, it is often necessary to use the PCM model to simulate the electrostatic influence of the environment on the reaction zone. If the reaction process involves significant changes in charge distribution, the results of vacuum QM calculations without the use of a continuous medium are not reasonable and may even lead to erroneous qualitative conclusions. In the QM/MM model, the reaction center is generally surrounded by soluble molecules that are treated in the MM manner, and there is generally no need to consider the continuous medium reaction field. However, if the net charge of the system changes before and after the reaction (e.g., redox potential calculations), it is likely that the contribution of the solution environment outside the system boundary to the free energy of the reaction needs to be taken into account, and in this case the QM/MM results can be corrected using the continuum medium model.
As a method that balances efficiency and accuracy, MM/PBSA can be used to analyze the affinity of protein-protein and protein-small molecule complexes [49]. To achieve error cancellation, it is conventional to perform explicit solvent molecular dynamics simulations of the complexes to obtain a collection of conformations; for each complex conformation, calculate the MM/PBSA energy of the complex as a whole, and of each monomer that makes up the complex, respectively; and approximate the free energy of binding by using the average of the difference between the MM/PBSA energy of the whole and that of the monomers for the entire conformation. This method can be used to analyze hotspot residues that affect substrate affinity and can also be used to predict substrate selectivity changes in mutants.
4
Other methods
4.1
Molecular Docking
Docking refers to the computational process of predicting the structure (and affinity) of a complex based on the structure of a monomer. Small molecule-protein docking is a central tool for structure-based virtual screening of drugs, for which several algorithms have been developed [13]. These algorithms and models can also be applied to the docking of substrate-enzyme complexes. Virtual drug screening requires the consideration of a large number of different small molecules, and for reasons of computational efficiency, the structural changes of the receptor (or just the side chain) are often not taken into account in molecular docking calculations. In contrast to virtual screening, in substrate-enzyme docking studies only one or several different substrates are often considered, and structural changes in the enzyme can in principle be more fully taken into account. The most straightforward way to accomplish this is to obtain diverse enzyme structures by conformational sampling methods such as MD, which are docked separately to the substrate. In substrate-enzyme docking, it is often also possible to use the relative spatial arrangement of substrate and catalytic functional groups to screen/evaluate docking results.
4.2
Prediction of small molecule pores based on geometry
Numerous experimental studies have revealed that some mutations far from the active center can have a great impact on the catalytic performance of enzymes. Some of these sites may act by altering the substrate binding/product release pore, and the pore size, physicochemical properties of the residues around the pore, etc. can change the substrate/product passage rate and affect the substrate selectivity. Pore prediction methods can be used to find relevant hot residues and provide a basis for the design of directed evolution libraries. Several geometric structure-based methods are available to predict protein surface pits, internal cavities, pores connecting different regions, etc [51-53]. These methods use static spatial structures as inputs, and mostly employ geometric and graph-theoretic methods to realize the prediction with high computational efficiency.
4.3
Active Center Comparison Methods
Currently, a large amount of 3D structure data of enzymes with different structure types and families have been accumulated in the Protein 3D Structure Database (PDB). If we compare different enzymes, we will find that some of them have a high degree of similarity in active centers (typical examples are the catalytic triad active centers shared by serine proteases), even though the overall structural sequences are not similar. The active center structure comparison method [54-55] can be used to automatically retrieve the active centers of other enzymes that are similar to the active center of the current enzyme. Stacking multiple similar active centers together in three-dimensional space and analyzing the similarities and differences between different active centers can provide valuable information for mutation site selection.
5
Summary
For the sake of clarity of presentation, our introduction of the above methods is categorized. In practice, different types of methods are not mutually exclusive. They can be used in combination in many ways to better answer our questions of interest. For example, in enzyme-substrate complex simulations, molecular docking can be used to obtain the initial conformation of the simulation; the set of conformations obtained from MD simulations can be used for pore prediction analysis, molecular docking, QM/MM simulations, etc.; MM models inscribing the transition states obtained from QM or QM/MM models can be built and used for long time classical MD simulations to analyze the effects of conformational rise and fall on chemical processes, or for the simulate a large number of mutants to realize virtual screening of mutants based on MD simulations; the MM/PBSA approach, which we have already mentioned, is a combination of MD and continuum medium modeling, and so on.
The study of biomolecule systems such as proteins by computational chemistry methods has a history of more than 40 years. These methods have been used more and more widely in industrial enzyme research while developing themselves continuously. China’s research teams in both computational chemistry and industrial enzyme engineering are expanding and their research capabilities are rapidly improving. The application of computational chemistry in enzyme engineering will be broadened and deepened as the cross-combination of these two disciplines becomes closer and closer. Protein engineering, directed evolution and other techniques have had a great impact on industrial enzyme research. The future development of computational methods, especially the breakthrough of new enzyme design methods, is expected to bring new technological breakthroughs for industrial enzyme research in the era of synthetic biology.
Contact Us Now!
If you need Price, please fill in your contact information in the form below, we will usually contact you within 24 hours. You could also email me info@longchangchemical.com during working hours ( 8:30 am to 6:00 pm UTC+8 Mon.~Sat. ) or use the website live chat to get prompt reply.
| Compound Glucoamylase | 9032-08-0 |
| Pullulanase | 9075-68-7 |
| Xylanase | 37278-89-0 |
| Cellulase | 9012-54-8 |
| Naringinase | 9068-31-9 |
| β-Amylase | 9000-91-3 |
| Glucose oxidase | 9001-37-0 |
| alpha-Amylase | 9000-90-2 |
| Pectinase | 9032-75-1 |
| Peroxidase | 9003-99-0 |
| Lipase | 9001-62-1 |
| Catalase | 9001-05-2 |
| TANNASE | 9025-71-2 |
| Elastase | 39445-21-1 |
| Urease | 9002-13-5 |
| DEXTRANASE | 9025-70-1 |
| L-Lactic dehydrogenase | 9001-60-9 |
| Dehydrogenase malate | 9001-64-3 |
| Cholesterol oxidase | 9028-76-6 |