Jiaxing Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jiaxing Chen has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- Nikolay V. Dokholyan
Department of Neurology, University of Virginia, Charlottesville, VA 22901, USA; Departments of Neuroscience, Pharmacology, Biomedical Engineering, Microbiology, Immunology, & Cancer Biology, University of Virginia, Charlottesville, VA 22901, USA
4 shared publications - Leslie A. Kuhn
Michigan State University
3 shared publications - Dong-Hui Yang
State Key Laboratory of Natural and Biomimetic Drugs, School of Pharmaceutical Sciences, Peking University, Beijing 100191, China
2 shared publications - Erdem D. Tabdanov
Penn State College of Medicine
2 shared publications Huijun Jiang
2 shared publications- Sebastian Raschka
Department of Statistics University of Wisconsin-Madison
2 shared publications
Latest Publications
- 8 September 2023
- Published by Springer Nature
- Vol. 2714, 235-268
Techniques for Developing Reliable Machine Learning Classifiers Applied to Understanding and Predicting Protein:Protein Interaction Hot Spots
With machine learning now transforming the sciences, successful prediction of biological structure or activity is mainly limited by the extent and quality of data available for training, the astute choice of features for prediction, and thorough assessment of the robustness of prediction on a variety of new cases. In this chapter, we address these issues while developing and sharing protocols to build a robust dataset and rigorously compare several predictive classifiers using the open-source Python machine learning library, scikit-learn. We show how to evaluate whether enough data has been used for training and whether the classifier has been overfit to training data. The most telling experiment is 500-fold repartitioning of the training and test sets, followed by prediction, which gives a good indication of whether a classifier performs consistently well on different datasets. An intuitive method is used to quantify which features are most important for correct prediction. The resulting well-trained classifier, hotspotter, can robustly predict the small subset of amino acid residues on the surface of a protein that are energetically most important for binding a protein partner: the interaction hot spots. Hotspotter has been trained and tested here on a curated dataset assembled from 1046 non-redundant alanine scanning mutation sites with experimentally measured change in binding free energy values from 97 different protein complexes; this dataset is available to download. The accessible surface area of the wild-type residue at a given site and its degree of evolutionary conservation proved the most important features to identify hot spots. A variant classifier was trained and validated for proteins where only the amino acid sequence is available, augmented by secondary structure assignment. This version of hotspotter requiring fewer features is almost as robust as the structure-based classifier. Application to the ACE2 (angiotensin converting enzyme 2) receptor, which mediates COVID-19 virus entry into human cells, identified the critical hot spot triad of ACE2 residues at the center of the small interface with the CoV-2 spike protein. Hotspotter results can be used to guide the strategic design of protein interfaces and ligands and also to identify likely interfacial residues for protein:protein docking. Key words Classifier training and validation Data science Protein design Protein docking Ligand design Feature importance Overfitting Scikit-learn SARS CoV-2 spike protein ACE2 receptor
- 8 September 2023
- Published by Springer Nature
- Vol. 2714, 235-268
Techniques for Developing Reliable Machine Learning Classifiers Applied to Understanding and Predicting Protein:Protein Interaction Hot Spots
With machine learning now transforming the sciences, successful prediction of biological structure or activity is mainly limited by the extent and quality of data available for training, the astute choice of features for prediction, and thorough assessment of the robustness of prediction on a variety of new cases. In this chapter, we address these issues while developing and sharing protocols to build a robust dataset and rigorously compare several predictive classifiers using the open-source Python machine learning library, scikit-learn. We show how to evaluate whether enough data has been used for training and whether the classifier has been overfit to training data. The most telling experiment is 500-fold repartitioning of the training and test sets, followed by prediction, which gives a good indication of whether a classifier performs consistently well on different datasets. An intuitive method is used to quantify which features are most important for correct prediction. The resulting well-trained classifier, hotspotter, can robustly predict the small subset of amino acid residues on the surface of a protein that are energetically most important for binding a protein partner: the interaction hot spots. Hotspotter has been trained and tested here on a curated dataset assembled from 1046 non-redundant alanine scanning mutation sites with experimentally measured change in binding free energy values from 97 different protein complexes; this dataset is available to download. The accessible surface area of the wild-type residue at a given site and its degree of evolutionary conservation proved the most important features to identify hot spots. A variant classifier was trained and validated for proteins where only the amino acid sequence is available, augmented by secondary structure assignment. This version of hotspotter requiring fewer features is almost as robust as the structure-based classifier. Application to the ACE2 (angiotensin converting enzyme 2) receptor, which mediates COVID-19 virus entry into human cells, identified the critical hot spot triad of ACE2 residues at the center of the small interface with the CoV-2 spike protein. Hotspotter results can be used to guide the strategic design of protein interfaces and ligands and also to identify likely interfacial residues for protein:protein docking. Key words Classifier training and validation Data science Protein design Protein docking Ligand design Feature importance Overfitting Scikit-learn SARS CoV-2 spike protein ACE2 receptor
- Open Access
- 16 June 2023
- Published by Wiley in Protein Science
- Vol. 32 (7) , e4686
Big versus small: The impact of aggregate size in disease
Protein aggregation results in an array of different size soluble oligomers and larger insoluble fibrils. Insoluble fibrils were originally thought to cause neuronal cell deaths in neurodegenerative diseases due to their prevalence in tissue samples and disease models. Despite recent studies demonstrating the toxicity associated with soluble oligomers, many therapeutic strategies still focus on fibrils or consider all types of aggregates as one group. Oligomers and fibrils require different modeling and therapeutic strategies, targeting the toxic species is crucial for successful study and therapeutic development. Here, we review the role of different‐size aggregates in disease, and how factors contributing to aggregation (mutations, metals, post‐translational modifications, and lipid interactions) may promote oligomers opposed to fibrils. We review two different computational modeling strategies (molecular dynamics and kinetic modeling) and how they are used to model both oligomers and fibrils. Finally, we outline the current therapeutic strategies targeting aggregating proteins and their strengths and weaknesses for targeting oligomers versus fibrils. Altogether, we aim to highlight the importance of distinguishing the difference between oligomers and fibrils and determining which species is toxic when modeling and creating therapeutics for protein aggregation in disease.
- 16 June 2023
- Published by Wiley in Protein Science
- Vol. 32 (7) , e4686
Open AccessBig versus small: The impact of aggregate size in disease
Protein aggregation results in an array of different size soluble oligomers and larger insoluble fibrils. Insoluble fibrils were originally thought to cause neuronal cell deaths in neurodegenerative diseases due to their prevalence in tissue samples and disease models. Despite recent studies demonstrating the toxicity associated with soluble oligomers, many therapeutic strategies still focus on fibrils or consider all types of aggregates as one group. Oligomers and fibrils require different modeling and therapeutic strategies, targeting the toxic species is crucial for successful study and therapeutic development. Here, we review the role of different‐size aggregates in disease, and how factors contributing to aggregation (mutations, metals, post‐translational modifications, and lipid interactions) may promote oligomers opposed to fibrils. We review two different computational modeling strategies (molecular dynamics and kinetic modeling) and how they are used to model both oligomers and fibrils. Finally, we outline the current therapeutic strategies targeting aggregating proteins and their strengths and weaknesses for targeting oligomers versus fibrils. Altogether, we aim to highlight the importance of distinguishing the difference between oligomers and fibrils and determining which species is toxic when modeling and creating therapeutics for protein aggregation in disease.
- 26 May 2023
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 9 (21) , eadg1062
A noncommutative combinatorial protein logic circuit controls cell orientation in nanoenvironments
Single-protein–based devices that integrate signal sensing with logical operations to generate functional outputs offer exceptional promise for monitoring and modulating biological systems. Engineering such intelligent nanoscale computing agents is challenging, as it requires the integration of sensor domains into a functional protein via intricate allosteric networks. We incorporate a rapamycin-sensitive sensor (uniRapR) and a blue light-responsive LOV2 domain into human Src kinase, creating a protein device that functions as a noncommutative combinatorial logic circuit. In our design, rapamycin activates Src kinase, causing protein localization to focal adhesions, whereas blue light exerts the reverse effect that inactivates Src translocation. Focal adhesion maturation induced by Src activation reduces cell migration dynamics and shifts cell orientation to align along collagen nanolane fibers. Using this protein device, we reversibly control cell orientation by applying the appropriate input signals, a framework that may be useful in tissue engineering and regenerative medicine.
- 26 May 2023
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 9 (21) , eadg1062
A noncommutative combinatorial protein logic circuit controls cell orientation in nanoenvironments
Single-protein–based devices that integrate signal sensing with logical operations to generate functional outputs offer exceptional promise for monitoring and modulating biological systems. Engineering such intelligent nanoscale computing agents is challenging, as it requires the integration of sensor domains into a functional protein via intricate allosteric networks. We incorporate a rapamycin-sensitive sensor (uniRapR) and a blue light-responsive LOV2 domain into human Src kinase, creating a protein device that functions as a noncommutative combinatorial logic circuit. In our design, rapamycin activates Src kinase, causing protein localization to focal adhesions, whereas blue light exerts the reverse effect that inactivates Src translocation. Focal adhesion maturation induced by Src activation reduces cell migration dynamics and shifts cell orientation to align along collagen nanolane fibers. Using this protein device, we reversibly control cell orientation by applying the appropriate input signals, a framework that may be useful in tissue engineering and regenerative medicine.
- Open Access
Techniques for Developing Reliable Machine Learning Classifiers Applied to Understanding and Predicting Protein:Protein Interaction Hot Spots
With machine learning now transforming the sciences, successful prediction of biological structure or activity is mainly limited by the extent and quality of data available for training, the astute choice of features for prediction, and thorough assessment of the robustness of prediction on a variety of new cases. Here we address these issues while developing and sharing protocols to build a robust dataset and rigorously compare several predictive classifiers using the opensource Python machine learning library, scikit-learn. We show how to evaluate whether enough data has been used for training and whether the classifier has been overfit to training data. The most telling experiment is 500-fold repartitioning of the training and test sets, followed by prediction, which gives a good indication of whether a classifier performs consistently well on different datasets. An intuitive method is used to quantify which features are most important for correct prediction. The resulting well-trained classifier, hotspotter , can robustly predict the small subset of amino acid residues on the surface of a protein that are energetically most important for binding a protein partner: the interaction hot spots. Hotspotter has been trained and tested here on a curated dataset assembled from 1,046 non-redundant alanine scanning mutation sites with experimentally measured change in binding free energy values from 97 different protein complexes; this dataset is available to download. The accessible surface area of the wild-type residue at a given site and its degree of evolutionary conservation proved the most important features to identify hot spots. A variant classifier was trained and validated for proteins where only the amino acid sequence is available, augmented by secondary structure assignment. This version of hotspotter requiring fewer features is almost as robust as the structure-based classifier. Application to the ACE2 receptor, which mediates COVID-19 virus entry into human cells, identified the critical hot spot triad of ACE2 residues at the center of the small interface with the CoV-2 spike protein. Hotspotter results can be used to guide the strategic design of protein interfaces and ligands and also to identify likely interfacial residues for protein:protein docking.
Open AccessTechniques for Developing Reliable Machine Learning Classifiers Applied to Understanding and Predicting Protein:Protein Interaction Hot Spots
With machine learning now transforming the sciences, successful prediction of biological structure or activity is mainly limited by the extent and quality of data available for training, the astute choice of features for prediction, and thorough assessment of the robustness of prediction on a variety of new cases. Here we address these issues while developing and sharing protocols to build a robust dataset and rigorously compare several predictive classifiers using the opensource Python machine learning library, scikit-learn. We show how to evaluate whether enough data has been used for training and whether the classifier has been overfit to training data. The most telling experiment is 500-fold repartitioning of the training and test sets, followed by prediction, which gives a good indication of whether a classifier performs consistently well on different datasets. An intuitive method is used to quantify which features are most important for correct prediction. The resulting well-trained classifier, hotspotter , can robustly predict the small subset of amino acid residues on the surface of a protein that are energetically most important for binding a protein partner: the interaction hot spots. Hotspotter has been trained and tested here on a curated dataset assembled from 1,046 non-redundant alanine scanning mutation sites with experimentally measured change in binding free energy values from 97 different protein complexes; this dataset is available to download. The accessible surface area of the wild-type residue at a given site and its degree of evolutionary conservation proved the most important features to identify hot spots. A variant classifier was trained and validated for proteins where only the amino acid sequence is available, augmented by secondary structure assignment. This version of hotspotter requiring fewer features is almost as robust as the structure-based classifier. Application to the ACE2 receptor, which mediates COVID-19 virus entry into human cells, identified the critical hot spot triad of ACE2 residues at the center of the small interface with the CoV-2 spike protein. Hotspotter results can be used to guide the strategic design of protein interfaces and ligands and also to identify likely interfacial residues for protein:protein docking.
- Open Access
- 16 November 2021
- Published by Springer Nature in Nature Communications
- Vol. 12 (1) , 1-12
Two-input protein logic gate for computation in living cells
Advances in protein design have brought us within reach of developing a nanoscale programming language, in which molecules serve as operands and their conformational states function as logic gates with precise input and output behaviors. Combining these nanoscale computing agents into larger molecules and molecular complexes will allow us to write and execute “code”. Here, in an important step toward this goal, we report an engineered, single protein design that is allosterically regulated to function as a ‘two-input logic OR gate’. Our system is based on chemo- and optogenetic regulation of focal adhesion kinase. In the engineered FAK, all of FAK domain architecture is retained and key intramolecular interactions between the kinase and the FERM domains are externally controlled through a rapamycin-inducible uniRapR module in the kinase domain and a light-inducible LOV2 module in the FERM domain. Orthogonal regulation of protein function was possible using the chemo- and optogenetic switches. We demonstrate that dynamic FAK activation profoundly increased cell multiaxial complexity in the fibrous extracellular matrix microenvironment and decreased cell motility. This work provides proof-of-principle for fine multimodal control of protein function and paves the way for construction of complex nanoscale computing agents.
- 16 November 2021
- Published by Springer Nature in Nature Communications
- Vol. 12 (1) , 1-12
Open AccessTwo-input protein logic gate for computation in living cells
Advances in protein design have brought us within reach of developing a nanoscale programming language, in which molecules serve as operands and their conformational states function as logic gates with precise input and output behaviors. Combining these nanoscale computing agents into larger molecules and molecular complexes will allow us to write and execute “code”. Here, in an important step toward this goal, we report an engineered, single protein design that is allosterically regulated to function as a ‘two-input logic OR gate’. Our system is based on chemo- and optogenetic regulation of focal adhesion kinase. In the engineered FAK, all of FAK domain architecture is retained and key intramolecular interactions between the kinase and the FERM domains are externally controlled through a rapamycin-inducible uniRapR module in the kinase domain and a light-inducible LOV2 module in the FERM domain. Orthogonal regulation of protein function was possible using the chemo- and optogenetic switches. We demonstrate that dynamic FAK activation profoundly increased cell multiaxial complexity in the fibrous extracellular matrix microenvironment and decreased cell motility. This work provides proof-of-principle for fine multimodal control of protein function and paves the way for construction of complex nanoscale computing agents.
- Open Access
Structural and Dynamic Insights Into α-Synuclein Dimer Conformations
Parkinson’s disease is associated with the aggregation of the protein α-synuclein. While α-synuclein can exist in multiple oligomeric states, the dimer has been a subject of extensive debates. Here, using an array of biophysical approaches, we demonstrate that α-synuclein in vitro exhibits primarily a monomer-dimer equilibrium in nanomolar concentrations and up to a few micromolars. We then use spatial information from hetero-isotopic cross-linking mass spectrometry experiments as restrains in discrete molecular dynamics simulations to obtain the ensemble structure of dimeric species. Out of eight structural sub-populations of dimers, we identify one that is compact, stable, abundant, and exhibits partially exposed β-sheet structures. This compact dimer is the only one where the hydroxyls of tyrosine 39 are in proximity that may promote dityrosine covalent linkage upon hydroxyl radicalization, which is implicated in α-synuclein amyloid fibrils. We propose that this α-synuclein dimer features etiological relevance to Parkinson’s disease.
Open AccessStructural and Dynamic Insights Into α-Synuclein Dimer Conformations
Parkinson’s disease is associated with the aggregation of the protein α-synuclein. While α-synuclein can exist in multiple oligomeric states, the dimer has been a subject of extensive debates. Here, using an array of biophysical approaches, we demonstrate that α-synuclein in vitro exhibits primarily a monomer-dimer equilibrium in nanomolar concentrations and up to a few micromolars. We then use spatial information from hetero-isotopic cross-linking mass spectrometry experiments as restrains in discrete molecular dynamics simulations to obtain the ensemble structure of dimeric species. Out of eight structural sub-populations of dimers, we identify one that is compact, stable, abundant, and exhibits partially exposed β-sheet structures. This compact dimer is the only one where the hydroxyls of tyrosine 39 are in proximity that may promote dityrosine covalent linkage upon hydroxyl radicalization, which is implicated in α-synuclein amyloid fibrils. We propose that this α-synuclein dimer features etiological relevance to Parkinson’s disease.
- Open Access
- 20 February 2019
- Published by Springer Nature in Nature Communications
- Vol. 10 (1) , 1-12
Cytosolic lipid droplets as engineered organelles for production and accumulation of terpenoid biomaterials in leaves
Cytosolic lipid droplets are endoplasmic reticulum-derived organelles typically found in seeds as reservoirs for physiological energy and carbon to fuel germination. Here, we report synthetic biology approaches to co-produce high-value sesqui- or diterpenoids together with lipid droplets in plant leaves. The formation of cytosolic lipid droplets is enhanced in the transient Nicotiana benthamiana system through ectopic production of WRINKLED1, a key regulator of plastid fatty acid biosynthesis, and a microalgal lipid droplet surface protein. Engineering of the pathways providing the universal C5-building blocks for terpenoids and installation of terpenoid biosynthetic pathways through direction of the enzymes to native and non-native compartments boost the production of target terpenoids. We show that anchoring of distinct biosynthetic steps onto the surface of lipid droplets leads to efficient production of terpenoid scaffolds and functionalized terpenoids. The co-produced lipid droplets “trap” the terpenoids in the cells.
- 20 February 2019
- Published by Springer Nature in Nature Communications
- Vol. 10 (1) , 1-12
Open AccessCytosolic lipid droplets as engineered organelles for production and accumulation of terpenoid biomaterials in leaves
Cytosolic lipid droplets are endoplasmic reticulum-derived organelles typically found in seeds as reservoirs for physiological energy and carbon to fuel germination. Here, we report synthetic biology approaches to co-produce high-value sesqui- or diterpenoids together with lipid droplets in plant leaves. The formation of cytosolic lipid droplets is enhanced in the transient Nicotiana benthamiana system through ectopic production of WRINKLED1, a key regulator of plastid fatty acid biosynthesis, and a microalgal lipid droplet surface protein. Engineering of the pathways providing the universal C5-building blocks for terpenoids and installation of terpenoid biosynthetic pathways through direction of the enzymes to native and non-native compartments boost the production of target terpenoids. We show that anchoring of distinct biosynthetic steps onto the surface of lipid droplets leads to efficient production of terpenoid scaffolds and functionalized terpenoids. The co-produced lipid droplets “trap” the terpenoids in the cells.
- 26 April 2018
- Published by Wiley in Biomedical Chromatography
- Vol. 32 (8) , e4236
In vivo metabolite profiles of isoimperatorin and phellopterin in rats analyzed using HPLC coupled with diode array detector and electrospray ionization ion trap time‐of‐flight mass spectrometry technique
Isoimperatorin (IP) and phellopterin (PP) are two furocoumarins existing in Angelicae Dahuricae Radix. There is an isopentenyloxyl substituted at C‐5 in IP, and an isopentenyloxyl and a methoxyl substituted at C‐8 and C‐5, respectively, in PP. To elucidate the in vivo metabolic characteristics of PP and IP, HPLC coupled with diode array detector and electrospray ionization ion trap time‐of‐flight mass spectrometry technique was used. In total, 111 metabolites, including 53 new ones, were identified from the urine and plasma samples of rats after oral administration of IP and PP, respectively. The metabolites were formed through eight reactions on IP and PP: oxidation, hydroxylation–hydrogenation, carboxylation on the isopentenyloxyl, O‐dealkylation, hydroxylation on the furocoumarin nucleus, ring‐opening reaction on the furan ring and reduction or ring‐opening reaction on the lactone ring. Among these, hydroxylation on the furocoumarin nucleus was found for the first time for in vivo metabolites of PP and IP, and the ring‐opening reaction on the furan ring or lactone ring was found for the first time for in vivo metabolites of isopentenyloxyl furocoumarins. The research gave us a new insight into the in vivo metabolic profiles of IP and PP, which could help us better understand their important roles as two active constituents of Angelicae Dahuricae Radix.
- 26 April 2018
- Published by Wiley in Biomedical Chromatography
- Vol. 32 (8) , e4236
In vivo metabolite profiles of isoimperatorin and phellopterin in rats analyzed using HPLC coupled with diode array detector and electrospray ionization ion trap time‐of‐flight mass spectrometry technique
Isoimperatorin (IP) and phellopterin (PP) are two furocoumarins existing in Angelicae Dahuricae Radix. There is an isopentenyloxyl substituted at C‐5 in IP, and an isopentenyloxyl and a methoxyl substituted at C‐8 and C‐5, respectively, in PP. To elucidate the in vivo metabolic characteristics of PP and IP, HPLC coupled with diode array detector and electrospray ionization ion trap time‐of‐flight mass spectrometry technique was used. In total, 111 metabolites, including 53 new ones, were identified from the urine and plasma samples of rats after oral administration of IP and PP, respectively. The metabolites were formed through eight reactions on IP and PP: oxidation, hydroxylation–hydrogenation, carboxylation on the isopentenyloxyl, O‐dealkylation, hydroxylation on the furocoumarin nucleus, ring‐opening reaction on the furan ring and reduction or ring‐opening reaction on the lactone ring. Among these, hydroxylation on the furocoumarin nucleus was found for the first time for in vivo metabolites of PP and IP, and the ring‐opening reaction on the furan ring or lactone ring was found for the first time for in vivo metabolites of isopentenyloxyl furocoumarins. The research gave us a new insight into the in vivo metabolic profiles of IP and PP, which could help us better understand their important roles as two active constituents of Angelicae Dahuricae Radix.
- 29 May 2016
- Published by Springer Nature in Applied Microbiology and Biotechnology
- Vol. 100 (21) , 9111-9124
Cloning, expression, and characterization of a four-component O-demethylase from human intestinal bacterium Eubacterium limosum ZL-II
Eubacterium limosum ZL-II was described to convert secoisolariciresinol (SECO) to its demethylating product 4,4′-dihydroxyenterodiol (DHEND) under anoxic conditions. However, the reaction cascade remains unclear. Here, the O-demethylase being responsible for the conversion was identified and characterized. Nine genes encoding two methyltransferase-Is (MT-I), two corrinoid proteins (CP), two methyltransferase-IIs (MT-II), and three activating enzymes (AE) were screened, cloned, and expressed in Escherichia coli. Four of the nine predicted enzymes, including ELI_2003 (MT-I), ELI_2004 (CP), ELI_2005 (MT-II), and ELI_0370 (AE), were confirmed to constitute the O-demethylase in E. limosum ZL-II. The complete O-demethylase (combining the four components) reaction system was reconstructed in vitro. As expected, the demethylating products 3-demethyl-SECO and DHEND were both produced. During the reaction process, ELI_2003 (MT-I) initially catalyzed the transfer of methyl group from SECO to the corrinoid of ELI_2004 ([CoI]-CP), yielding demethylating products and [CH3-CoIII]-CP; then ELI_2005 (MT-II) mediated the transfer of methyl group from [CH3-CoIII]-CP to tetrahydrofolate, forming methyltetrahydrofolate and [CoI]-CP. Due to the low redox potential of [CoII]/[CoI], [CoI]-CP was oxidized to [CoII]-CP immediately in vitro, and ELI_0370 (AE) was responsible for catalyzing the reduction of [CoII]-CP to its active form [CoI]-CP. The active-site residues in ELI_2003, ELI_2005, and ELI_0370 were subsequently determined using molecular modeling combined with site-directed mutagenesis. To our knowledge, this is the first study on the identification and characterization of a four-component O-demethylase from E. limosum ZL-II, which will facilitate the development of method to artificial synthesis of related bioactive chemicals.
- 29 May 2016
- Published by Springer Nature in Applied Microbiology and Biotechnology
- Vol. 100 (21) , 9111-9124
Cloning, expression, and characterization of a four-component O-demethylase from human intestinal bacterium Eubacterium limosum ZL-II
Eubacterium limosum ZL-II was described to convert secoisolariciresinol (SECO) to its demethylating product 4,4′-dihydroxyenterodiol (DHEND) under anoxic conditions. However, the reaction cascade remains unclear. Here, the O-demethylase being responsible for the conversion was identified and characterized. Nine genes encoding two methyltransferase-Is (MT-I), two corrinoid proteins (CP), two methyltransferase-IIs (MT-II), and three activating enzymes (AE) were screened, cloned, and expressed in Escherichia coli. Four of the nine predicted enzymes, including ELI_2003 (MT-I), ELI_2004 (CP), ELI_2005 (MT-II), and ELI_0370 (AE), were confirmed to constitute the O-demethylase in E. limosum ZL-II. The complete O-demethylase (combining the four components) reaction system was reconstructed in vitro. As expected, the demethylating products 3-demethyl-SECO and DHEND were both produced. During the reaction process, ELI_2003 (MT-I) initially catalyzed the transfer of methyl group from SECO to the corrinoid of ELI_2004 ([CoI]-CP), yielding demethylating products and [CH3-CoIII]-CP; then ELI_2005 (MT-II) mediated the transfer of methyl group from [CH3-CoIII]-CP to tetrahydrofolate, forming methyltetrahydrofolate and [CoI]-CP. Due to the low redox potential of [CoII]/[CoI], [CoI]-CP was oxidized to [CoII]-CP immediately in vitro, and ELI_0370 (AE) was responsible for catalyzing the reduction of [CoII]-CP to its active form [CoI]-CP. The active-site residues in ELI_2003, ELI_2005, and ELI_0370 were subsequently determined using molecular modeling combined with site-directed mutagenesis. To our knowledge, this is the first study on the identification and characterization of a four-component O-demethylase from E. limosum ZL-II, which will facilitate the development of method to artificial synthesis of related bioactive chemicals.