Prediction of drug side effects with transductive matrix co-completion.

MOTIVATION: Side effects of drugs could cause severe health problems and the failure of drug development. Drug-target interactions are the basis for side effect production and are important for side effect prediction. However, the information on the known targets of drugs is incomplete. Furthermore, there could be also some missing data in the existing side effect profile of drugs. As a result, new methods are needed to deal with the missing features and missing labels in the problem of side effect prediction. RESULTS: We propose a novel computational method based on transductive matrix co-completion and leverage the low-rank structure in the side effects and drug-target data. Positive-unlabelled learning is incorporated into the model to handle the impact of unobserved data. We also introduce graph regularization to integrate the drug chemical information for side effect prediction. We collect the data on side effects, drug targets, drug-associated proteins and drug chemical structures to train our model and test its performance for side effect prediction. The experiment results show that our method outperforms several other state-of-the-art methods under different scenarios. The case study and additional analysis illustrate that the proposed method could not only predict the side effects of drugs but also could infer the missing targets of drugs. AVAILABILITY AND IMPLEMENTATION: The data and the code for the proposed method are available at https://github.com/LiangXujun/GTMCC. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Mechanism of forkhead transcription factors binding to a novel palindromic DNA site.

Forkhead transcription factors bind a canonical consensus DNA motif, RYAAAYA (R = A/G, Y = C/T), as a monomer. However, the molecular mechanisms by which forkhead transcription factors bind DNA as a dimer are not well understood. In this study, we show that FOXO1 recognizes a palindromic DNA element DIV2, and mediates transcriptional regulation. The crystal structure of FOXO1/DIV2 reveals that the FOXO1 DNA binding domain (DBD) binds the DIV2 site as a homodimer. The wing1 region of FOXO1 mediates the dimerization, which enhances FOXO1 DNA binding affinity and complex stability. Further biochemical assays show that FOXO3, FOXM1 and FOXI1 also bind the DIV2 site as homodimer, while FOXC2 can only bind this site as a monomer. Our structural, biochemical and bioinformatics analyses not only provide a novel mechanism by which FOXO1 binds DNA as a homodimer, but also shed light on the target selection of forkhead transcription factors.

IMF: Interpretable Multi-Hop Forecasting on Temporal Knowledge Graphs.

Temporal knowledge graphs (KGs) have recently attracted increasing attention. The temporal KG forecasting task, which plays a crucial role in such applications as event prediction, predicts future links based on historical facts. However, current studies pay scant attention to the following two aspects. First, the interpretability of current models is manifested in providing reasoning paths, which is an essential property of path-based models. However, the comparison of reasoning paths in these models is operated in a black-box fashion. Moreover, contemporary models utilize separate networks to evaluate paths at different hops. Although the network for each hop has the same architecture, each network achieves different parameters for better performance. Different parameters cause identical semantics to have different scores, so models cannot measure identical semantics at different hops equally. Inspired by the observation that reasoning based on multi-hop paths is akin to answering questions step by step, this paper designs an Interpretable Multi-Hop Reasoning (IMR) framework based on consistent basic models for temporal KG forecasting. IMR transforms reasoning based on path searching into stepwise question answering. In addition, IMR develops three indicators according to the characteristics of temporal KGs and reasoning paths: the question matching degree, answer completion level, and path confidence. IMR can uniformly integrate paths of different hops according to the same criteria; IMR can provide the reasoning paths similarly to other interpretable models and further explain the basis for path comparison. We instantiate the framework based on common embedding models such as TransE, RotatE, and ComplEx. While being more explainable, these instantiated models achieve state-of-the-art performance against previous models on four baseline datasets.

Structural insight into the molecular mechanism of cilofexor binding to the farnesoid X receptor.

Farnesoid X receptor (FXR) is a bile acid-related nuclear receptor and is considered a promising target to treat several liver disorders. Cilofexor is a selective FXR agonist and has already entered phase III trials in primary sclerosing cholangitis (PSC) patients. Pruritis caused by cilofexor treatment is dose dependent. The binding characteristics of cilofexor with FXR and its pruritogenic mechanism remain unclear. In our research, the affinity of cilofexor bound to FXR was detected using an isothermal titration calorimetry (ITC) assay. The binding mechanism between cilofexor and FXR-LBD is explained by the cocrystal structure of the FXR/cilofexor complex. Structural models indicate the possibility that cilofexor activates Mas-related G protein-coupled receptor X4 (MRGPRX4) or G protein-coupled bile acid receptor 1 (GPBAR1), leading to pruritus. In summary, our analyses provide a molecular mechanism of cilofexor binding to FXR and provide a possible explanation for the dose-dependent pruritis of cilofexor.

Characterization of the modification of Kelch-like ECH-associated protein 1 by different fumarates.

Fumarates (fumaric acid esters), primarily dimethyl fumarate (DMF) and monoethyl fumarate (MEF) and its salts, are orally administered systemic agents used for the treatment of psoriasis and multiple sclerosis. It is widely believed that the pharmaceutical activities of fumarates are exerted through the Keap1-Nrf2 pathway. Although it has been revealed that DMF and MEF differentially modify specific Keap1 cysteine residues and result in the differential activation of Nrf2, how the modification of DMF and MEF impacts the biochemical properties of Keap1 has not been well characterized. Here, we found that both DMF and MEF can only modify the BTB domain of Keap1 and that only C151 is accessible for covalent binding in vitro. Dynamic fluorescence scanning (DSF) assays showed that the modification of DMF to Keap1 BTB increased its thermal stability, while the modification of MEF dramatically decreased its thermal stability. Further crystal structures revealed no significant conformational variation between the DMF-modified and MEF-modified BTBs. Overall, our biochemical and structural study provides a better understanding of the covalent modification of fumarates to Keap1 and may suggest fundamentally different mechanisms adopted by fumarates in regulating the Keap1-Nrf2 pathway.

Structural study of ponatinib in inhibiting SRC kinase.

Ponatinib is a multi-target tyrosine kinase inhibitor that targets ABL, SRC, FGFR, and so on. It was designed to overcome the resistance of BCR-ABL mutation to imatinib, especially the gatekeeper mutation ABLT315I. The molecular mechanism by which ponatinib overcomes mutations of BCR-ABL and some other targets has been explained, but little information is known about the characteristics of ponatinib binding to SRC. Here, we showed that ponatinib inhibited wild type SRC kinase but failed to inhibit SRC gatekeeper mutants in both biochemical and cellular assays. We determined the crystal structure of ponatinib in complex with the SRC kinase domain. In addition, by structural analysis, we provided a possible explanation for why ponatinib showed different effects on SRC and other kinases with gatekeeper mutations. The resistance mechanism of SRC gatekeeper mutations to ponatinib may provide meaningful information for designing inhibitors against SRC family kinases in the future.

A novel machine learning model based on sparse structure learning with adaptive graph regularization for predicting drug side effects.

Drug side effects are closely related to the success and failure of drug development. Here we present a novel machine learning method for side effect prediction. The proposed method treats side effect prediction as a multi-label learning problem and uses sparse structure learning to model the relationships between side effects. Additionally, the proposed method adopts the adaptive graph regularization strategy to explore the local structure in drug data and fuse multiple types of drug features. An alternating optimization algorithm is proposed to solve the optimization problem. We collected chemical structures and biological pathway features of drugs as the inputs of our method to predict drug side effects. The results of the cross-validation experiment showed that our method could significantly improve the prediction performance compared to the other state-of-the-art methods. Besides, our model is highly interpretable. It could learn the drug neighbourhood relationships, side effect relationships, and drug features related to side effects. We systematically validated the information extracted by the model with independent data. Some prediction results could also be supported by literature reports. The proposed method could be applied to integrate both chemical and biological data to predict side effects and helps improve drug safety.

Structural basis of tropifexor as a potent and selective agonist of farnesoid X receptor.

Farnesoid X receptor (FXR) is considered as a potential target for the treatment of several liver disorders such as primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC). Tropifexor is a highly potent and non-steroidal FXR agonist that has progressed into phase II clinical trials in patients with PBC. The clinical trials demonstrate that tropifexor improved serum markers of patients with liver diseases and lower side effect such as pruritus that might be implicated with TGR5 activation. However, the molecular mechanism of the potency and selectivity of tropifexor remains unclear. In this study, the binding affinity of FXR and tropifexor is measured by isothermal titration calorimetry (ITC) assays. The crystal structure of the FXR/tropifexor complex is determined at 2.7 Å resolution to explain the molecular mechanism of tropifexor bound to FXR-LBD. Structural comparison with other FXR/agonists structures reveals the conformational change in the FXR/tropifexor structure. Moreover the structural superposition of TGR5/tropifexor indicates that the steric hindrance between tropifexor and TGR5 might be a possible explanation for the impotency arises of tropifexor to TGR5. Overall, our analyses might provide an insight into the molecular mechanism of tropifexor binding to FXR-LBD and account for the high selectivity of tropifexor for FXR versus TGR5.

Characterization of ibrutinib as a non-covalent inhibitor of SRC-family kinases.

Ibrutinib is a BTK-targeted irreversible inhibitor. In this study, we demonstrate that ibrutinib potently inhibits SRC activity in a non-covalent manner via mass spectrometry and crystallography. The S345C mutation renders SRC to bind covalently with ibrutinib, and restores the potency of ibrutinib against the gatekeeper mutant. The co-crystal structure of ibrutinib/SRC shows Ser345 of SRC did not form covalent bond with ibrutinib, leading to a decrease of potency and loss of the ability to overcome the gatekeeper mutation of SRC. The X-ray crystallographic studies also provide structural insight into why ibrutinib behaves differently against gatekeeper mutants of different kinases.

Structural basis for DNA recognition by FOXC2.

The FOXC family of transcription factors (FOXC1 and FOXC2) plays essential roles in the regulation of embryonic, ocular, and cardiac development. Mutations and abnormal expression of FOXC proteins are implicated in genetic diseases as well as cancer. In this study, we determined two crystal structures of the DNA-binding domain (DBD) of human FOXC2 protein, in complex with different DNA sites. The FOXC2-DBD adopts the winged-helix fold with helix H3 contributing to all the base specific contacts, while the N-terminus, wing 1, and the C-terminus of FOXC2-DBD all make additional contacts with the phosphate groups of DNA. Our structural, biochemical, and bioinformatics analyses allow us to revise the previously proposed DNA recognition mechanism and provide a model of DNA binding for the FOXC proteins. In addition, our structural analysis and accompanying biochemical assays provide a molecular basis for understanding disease-causing mutations in FOXC1 and FOXC2.

Structural basis of binding of homodimers of the nuclear receptor NR4A2 to selective Nur-responsive DNA elements.

Proteins of nuclear receptor subfamily 4 group A (NR4A), including NR4A1/NGFI-B, NR4A2/Nurr1, and NR4A3/NOR-1, are nuclear transcription factors that play important roles in metabolism, apoptosis, and proliferation. NR4A proteins recognize DNA response elements as monomers or dimers to regulate the transcription of a variety of genes involved in multiple biological processes. In this study, we determined two crystal structures of the NR4A2 DNA-binding domain (NR4A2-DBD) bound to two Nur-responsive elements: an inverted repeat and an everted repeat at 2.6-2.8 Å resolution. The structures revealed that two NR4A2-DBD molecules bind independently to the everted repeat, whereas two other NR4A2-DBD molecules form a novel dimer interface on the inverted repeat. Moreover, substitution of the interfacial residue valine 298 to lysine as well as mutation of DNA bases involved in the interactions abolished the dimerization. Overall, our structural, biochemical, and bioinformatics analyses provide a molecular basis for the binding of the NR4A2 protein dimers to NurREs and advance our understanding of the dimerization specificity of nuclear receptors.

Structural basis of NR4A1 bound to the human pituitary proopiomelanocortin gene promoter.

The nuclear receptor NR4A subfamily (NR4A1/NGFI-B, NR4A2/Nurr1 and NR4A3/NOR-1) can recognize different classes of DNA response elements either as a monomer, homodimer, or heterodimer. In this study, we determined the structure of the NR4A1 DNA-binding domain (NR4A1-DBD) bound to natural Nur-responsive elements (NurREs) in the promoter region of the pituitary proopiomelanocortin (POMC) gene (NurREPOMC) at 3.12 Å resolution. The NR4A1-DBD molecules bound independently to this element in our structure. The N-terminal helix H1 forms specific contacts with major groove, and C-terminal extension interact extensively with minor groove. Moreover our modelling structure of NR4A1 large fragment complexed with NurREPOMC indicated that ligand binding domain of NR4A might form dimer interactions to help recognize DNA. Overall, our analyses provide a molecular basis for DNA binding of NR4A proteins as a homodimer and novel insight into the molecular functions of NR4A modulation of gene expression.

LRSSL: predict and interpret drug-disease associations based on data integration using sparse subspace learning.

Motivation: : Exploring the potential curative effects of drugs is crucial for effective drug development. Previous studies have indicated that integration of multiple types of information could be conducive to discovering novel indications of drugs. However, how to efficiently identify the mechanism behind drug-disease associations while integrating data from different sources remains a challenging problem. Results: : In this research, we present a novel method for indication prediction of both new drugs and approved drugs. This method is based on Laplacian regularized sparse subspace learning (LRSSL), which integrates drug chemical information, drug target domain information and target annotation information. Experimental results show that the proposed method outperforms several recent approaches for predicting drug-disease associations. Some drug therapeutic effects predicted by the method could be validated by database records or literatures. Moreover, with L1-norm constraint, important drug features have been extracted from multiple drug feature profiles. Case studies suggest that the extracted drug features could be beneficial to interpretation of the predicted results. Availability and Implementation: https://github.com/LiangXujun/LRSSL. Contact: proteomics@csu.edu.cn. Supplementary information: Supplementary data are available at Bioinformatics online.

Structure of the Forkhead Domain of FOXA2 Bound to a Complete DNA Consensus Site.

FOXA2, a member of the forkhead family of transcription factors, plays essential roles in liver development and bile acid homeostasis. In this study, we report a 2.8 Å co-crystal structure of the FOXA2 DNA-binding domain (FOXA2-DBD) bound to a DNA duplex containing a forkhead consensus binding site (GTAAACA). The FOXA2-DBD adopts the canonical winged-helix fold, with helix H3 and wing 1 regions mainly mediating the DNA recognition. Although the wing 2 region was not defined in the structure, isothermal titration calorimetry assays suggested that this region was required for optimal DNA binding. Structure comparison with the FOXA3-DBD bound to DNA revealed more major groove contacts and fewer minor groove contacts in the FOXA2 structure than in the FOXA3 structure. Structure comparison with the FOXO1-DBD bound to DNA showed that different forkhead proteins could induce different DNA conformations upon binding to identical DNA sequences. Our findings provide the structural basis for FOXA2 protein binding to a consensus forkhead site and elucidate how members of the forkhead protein family bind different DNA sites.

DNA-binding properties of FOXP3 transcription factor.

FOXP3, a lineage-specific forkhead (FKH) transcription factor, plays essential roles in the development and function of regulatory T cells. However, the DNA-binding properties of FOXP3 are not well understood. In this study, FOXP3 fragments containing different domains were purified, and their DNA-binding properties were investigated using electrophoretic mobility shift assay and isothermal titration calorimetry (ITC). Both the FKH and leucine-zipper domains were required for optimal DNA binding for FOXP3. FOXP3 protein not only binds with DNA sequences containing one FKH consensus sequence, but also binds with DNA sequences with two direct repeats of consensus sequences separated by three-nucleotides (DRE3). Our results shed lights on the mechanisms by which FOXP3 recognizes cognate DNA elements, and would facilitate further structural and functional studies of FOXP3.

Expression and purification of the kinase domain of PINK1 in Pichia pastoris.

PTEN-induced putative kinase 1 (PINK1) is a Ser/Thr kinase that specifically localizes on the mitochondrial membrane. It cooperates with Parkin to regulate mitochondrial quality control. Mutations in PINK1 protein which account for 8-15% of Parkinson's disease (PD), are the second most common cause of early-onset Autosomal Recessive Parkinson's disease (AR-PD). The lack of methods for PINK1 heterologous expression and purification has slowed progress in the AR-PD research field. To pave the way for direct structural study of this important protein, in this study, we developed an efficient expression system of recombinant PINK1 kinase domain (rPINK1) using Pichia pastoris (P. pastoris). Our results showed that rPINK1 is best expressed in P. pastoris at 25 °C induction. Additionally, we determined that the optimal induction time was 72 h and the optimal induction methanol concentration was 1% for the expression of rPINK1 in P. pastoris. Subsequent purification by Ni affinity chromatography (Ni-NTA) and cation-exchange chromatography (Mono S) produced the protein with purity higher than 95%. The pure rPINK1 was active to phosphorylate ubiquitin in a substrate phosphorylation assay. Overall, these studies provide the first effective method for heterologous expression and purification of the rPINK1 with a high purity. These findings can help contribute to further researches on the interactions study and biochemical characterization of PINK1.

Structural basis and selectivity of sulfatinib binding to FGFR and CSF-1R.

Acquired drug resistance poses a challenge for single-target FGFR inhibitors, leading to the development of dual- or multi-target FGFR inhibitors. Sulfatinib is a multi-target kinase inhibitor for treating neuroendocrine tumors, selectively targeting FGFR1/CSF-1R. To elucidate the molecular mechanisms behind its binding and kinase selectivity, we determined the crystal structures of sulfatinib with FGFR1/CSF-1R. The results reveal common structural features and distinct conformational adaptability of sulfatinib in response to FGFR1/CSF-1R binding. Further biochemical and structural analyses disclose sensitivity of sulfatinib to FGFR/CSF-1R gatekeeper mutations. The insensitivity of sulfatinib to FGFR gatekeeper mutations highlights the indispensable interactions with the hydrophobic pocket for FGFR selectivity, whereas the rotatory flexibility may enable sulfatinib to overcome CSF-1RT663I. This study not only sheds light on the structural basis governing sulfatinib's FGFR/CSF-1R inhibition, but also provides valuable insights into the rational design of dual- or multi-target FGFR inhibitors with selectivity for CSF-1R and sensitivity to gatekeeper mutations.

Design, synthesis, and biological evaluation of selective covalent inhibitors of FGFR4.

Aberrant signaling via fibroblast growth factor 19 (FGF19)/fibroblast growth factor receptor 4 (FGFR4) has been identified as a driver of tumorigenesis and the development of many solid tumors, making FGFR4 is a promising target for anticancer therapy. Herein, we designed and synthesized a series of bis-acrylamide covalent FGFR4 inhibitors and evaluated their inhibitory activity against FGFRs, FGFR4 mutants, and their antitumor activity. CXF-007, verified by mass spectrometry and crystal structures to form covalent bonds with Cys552 of FGFR4 and Cys488 of FGFR1, exhibited stronger selectivity and potent inhibitory activity for FGFR4 and FGFR4 cysteine mutants. Moreover, CXF-007 exhibited significant antitumor activity in hepatocellular carcinoma cell lines and breast cancer cell lines through sustained inhibition of the FGFR4 signaling pathway. In summary, our study highlights a novel covalent FGFR4 inhibitor, CXF-007, which has the potential to overcome drug-induced FGFR4 mutations and might provide a new strategy for future anticancer drug discovery.

Versatile self-assembly of water-soluble thiol-capped CdTe quantum dots: external destabilization and internal stability of colloidal QDs.

In this paper, we report on the versatile self-assembly of water-soluble thiol-capped CdTe quantum dots (QDs), nanoparticles (NPs), or nanocrystals induced by L-cysteine (L-Cys). Major efforts are focused on the control of the self-organization of QDs into nanosheets (NSs), for example, by altering the solution pH and the QD size. The as-prepared nanosheets exhibit bright photoluminescence (PL) and retain the size-quantized properties of initial CdTe QDs, since they are actually formed by a 2D network of assembled QDs. By optical techniques, TEM, EDX, powder XRD, etc., it is found that the unique L-Cys-induced external destabilization is responsible for the template-free self-organization process, with the further assistance of the specific NP-NP interactions. And the internal chemical stability of initial CdTe QDs also is proven for the first time to play an important role. These results help to enhance the current understanding about the mechanism for the destabilization of colloidal NPs and their self-assembly behavior.