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.

Epilogue to the Gerald Maggiora Festschrift: a tribute to an exemplary mentor, colleague, collaborator, and innovator.

In May 2022, JCAMD published a Special Issue in honor of Gerald (Gerry) Maggiora, whose scientific leadership over many decades advanced the fields of computational chemistry and chemoinformatics for drug discovery. Along the way, he has impacted many researchers in both academia and the pharmaceutical industry. In this Epilogue, we explain the origins of the Festschrift and present a series of first-hand vignettes, in approximate chronological sequence, that together paint a picture of this remarkable man.

Machine Learning to Identify Flexibility Signatures of Class A GPCR Inhibition.

We show that machine learning can pinpoint features distinguishing inactive from active states in proteins, in particular identifying key ligand binding site flexibility transitions in GPCRs that are triggered by biologically active ligands. Our analysis was performed on the helical segments and loops in 18 inactive and 9 active class A G protein-coupled receptors (GPCRs). These three-dimensional (3D) structures were determined in complex with ligands.

Deciphering the three-domain architecture in schlafens and the structures and roles of human schlafen12 and serpinB12 in transcriptional regulation.

Schlafen proteins are important in cell differentiation and defense against viruses, and yet this family of vertebrate proteins is just beginning to be understood at the molecular level. Here, the three-dimensional architecture and molecular interfaces of human schlafen12 (hSLFN12), which promotes intestinal stem cell differentiation, are analyzed by sequence conservation and structural modeling in light of the functions of its homologs and binding partners. Our analysis shows that the schlafen or divergent AAA ATPase domain described in the N-terminal region of schlafens in databases and the literature is a misannotation.

Schlafen 12 Interaction with SerpinB12 and Deubiquitylases Drives Human Enterocyte Differentiation.

Background/aims: Human enterocytic differentiation is altered during development, fasting, adaptation, and bariatric surgery, but its intracellular control remains unclear. We hypothesized that Schlafen 12 (SLFN12) regulates enterocyte differentiation.

Methods: We used laser capture dissection of epithelium, qRT-PCR, and immunohistochemistry to evaluate SLFN12 expression in biopsies of control and fasting human duodenal mucosa, and viral overexpression and siRNA to trace the SLFN12 pathway in human Caco-2 and HIEC6 intestinal epithelial cells.

Automated Inference of Chemical Discriminants of Biological Activity.

Ligand-based virtual screening has become a standard technique for the efficient discovery of bioactive small molecules. Following assays to determine the activity of compounds selected by virtual screening, or other approaches in which dozens to thousands of molecules have been tested, machine learning techniques make it straightforward to discover the patterns of chemical groups that correlate with the desired biological activity. Defining the chemical features that generate activity can be used to guide the selection of molecules for subsequent rounds of screening and assaying, as well as help design new, more active molecules for organic synthesis.

Protein-ligand interfaces are polarized: discovery of a strong trend for intermolecular hydrogen bonds to favor donors on the protein side with implications for predicting and designing ligand complexes.

Understanding how proteins encode ligand specificity is fascinating and similar in importance to deciphering the genetic code. For protein-ligand recognition, the combination of an almost infinite variety of interfacial shapes and patterns of chemical groups makes the problem especially challenging. Here we analyze data across non-homologous proteins in complex with small biological ligands to address observations made in our inhibitor discovery projects: that proteins favor donating H-bonds to ligands and avoid using groups with both H-bond donor and acceptor capacity.

Enabling the hypothesis-driven prioritization of ligand candidates in big databases: Screenlamp and its application to GPCR inhibitor discovery for invasive species control.

While the advantage of screening vast databases of molecules to cover greater molecular diversity is often mentioned, in reality, only a few studies have been published demonstrating inhibitor discovery by screening more than a million compounds for features that mimic a known three-dimensional (3D) ligand. Two factors contribute: the general difficulty of discovering potent inhibitors, and the lack of free, user-friendly software to incorporate project-specific knowledge and user hypotheses into 3D ligand-based screening. The Screenlamp modular toolkit presented here was developed with these needs in mind.

Inhibition of pressure-activated cancer cell adhesion by FAK-derived peptides.

Forces within the surgical milieu or circulation activate cancer cell adhesion and potentiate metastasis through signaling requiring FAK-Akt1 interaction. Impeding FAK-Akt1 interaction might inhibit perioperative tumor dissemination, facilitating curative cancer surgery without global FAK or AKT inhibitor toxicity. Serial truncation and structurally designed mutants of FAK identified a seven amino acid, short helical structure within FAK that effectively competes with Akt1-FAK interaction.

Detecting the native ligand orientation by interfacial rigidity: SiteInterlock.

Understanding the physical attributes of protein-ligand interfaces, the source of most biological activity, is a fundamental problem in biophysics. Knowing the characteristic features of interfaces also enables the design of molecules with potent and selective interactions. Prediction of native protein-ligand interactions has traditionally focused on the development of physics-based potential energy functions, empirical scoring functions that are fit to binding data, and knowledge-based potentials that assess the likelihood of pairwise interactions.

Translocator Protein 18 kDa (TSPO): An Old Protein with New Functions?

Translocator protein 18 kDa (TSPO) was previously known as the peripheral benzodiazepine receptor (PBR) in eukaryotes, where it is mainly localized to the mitochondrial outer membrane. Considerable evidence indicates that it plays regulatory roles in steroidogenesis and apoptosis and is involved in various human diseases, such as metastatic cancer, Alzheimer's and Parkinson's disease, inflammation, and anxiety disorders. Ligands of TSPO are widely used as diagnostic tools and treatment options, despite there being no clear understanding of the function of TSPO.

CholMine: Determinants and Prediction of Cholesterol and Cholate Binding Across Nonhomologous Protein Structures.

Identifying physiological ligands is necessary for annotating new protein structures, yet this presents a significant challenge to biologists and pharmaceutical chemists. Here we develop a predictor of cholesterol and cholate binding that works across diverse protein families, extending beyond sequence motif-based prediction. This approach combines SimSite3D site comparison with the detection of conserved interactions in cholesterol/cholate bound crystal structures to define three-dimensional interaction motifs.

Structural and functional assessment of perilipin 2 lipid binding domain(s).

Although perilipin 2 (Plin2) has been shown to bind lipids with high affinity, the Plin2 lipid binding site has yet to be defined. This is of interest since Plin2's affinity for lipids has been suggested to be important for lipid droplet biogenesis and intracellular triacylglycerol accumulation. To define these regions, mouse Plin2 and several deletion mutants expressed as recombinant proteins and in mammalian cells were assessed by molecular modeling, fluorescence binding, circular dichroic, and fluorescence resonance energy transfer techniques to identify the structural and functional requirements for lipid binding.

The P-loop region of Schlafen 3 acts within the cytosol to induce differentiation of human Caco-2 intestinal epithelial cells.

Schlafen 3 (Slfn3) mediates rodent enterocyte differentiation in vitro and in vivo, required for intestinal function. Little is known about Schlafen protein structure-function relationships. To define the Slfn3 domain that promotes differentiation, we studied villin and sucrase isomaltase (SI) promoter activity in Slfn3-null human Caco-2BBE cells transfected with full-length rat Slfn3 DNA or truncated constructs.

Structural determinants allowing transferase activity in SENSITIVE TO FREEZING 2, classified as a family I glycosyl hydrolase.

SENSITIVE TO FREEZING 2 (SFR2) is classified as a family I glycosyl hydrolase but has recently been shown to have galactosyltransferase activity in Arabidopsis thaliana. Natural occurrences of apparent glycosyl hydrolases acting as transferases are interesting from a biocatalysis standpoint, and knowledge about the interconversion can assist in engineering SFR2 in crop plants to resist freezing. To understand how SFR2 evolved into a transferase, the relationship between its structure and function are investigated by activity assay, molecular modeling, and site-directed mutagenesis.

Computational prediction and in vitro analysis of potential physiological ligands of the bile acid binding site in cytochrome c oxidase.

A conserved bile acid site has been crystallographically defined in the membrane domain of mammalian and Rhodobacter sphaeroides cytochrome c oxidase (RsCcO). Diverse amphipathic ligands were shown previously to bind to this site and affect the electron transfer equilibrium between heme a and a3 cofactors by blocking the K proton uptake path. Current studies identify physiologically relevant ligands for the bile acid site using a novel three-pronged computational approach: ROCS comparison of ligand shape and electrostatics, SimSite3D comparison of ligand binding site features, and SLIDE screening of potential ligands by docking.

From static structure to living protein: computational analysis of cytochrome c oxidase main-chain flexibility.

Crystallographic structure and deuterium accessibility comparisons of CcO in different redox states have suggested conformational changes of mechanistic significance. To predict the intrinsic flexibility and low energy motions in CcO, this work has analyzed available high-resolution crystallographic structures with ProFlex and elNémo computational methods. The results identify flexible regions and potential conformational changes in CcO that correlate well with published structural and biochemical data and provide mechanistic insights.

Interleukin-8-like activity in a filarial asparaginyl-tRNA synthetase.

A wide range of secondary biological functions have been documented for eukaryotic aminoacyl-tRNA synthetases including roles in transcriptional regulation, mitochondrial RNA splicing, cell growth, and chemokine-like activities. The asparaginyl-tRNA synthetase (AsnRS) of the filarial nematode, Brugia malayi, is a highly expressed excretory-secretory molecule which activates interleukin 8 (IL-8) receptors via extracellular domains that are different from those used by IL-8. Recent success in determining the complete atomic structure of the B.

DNA binding by the Arabidopsis CBF1 transcription factor requires the PKKP/RAGRxKFxETRHP signature sequence.

The CBF/DREB1 transcriptional activators are key regulators of plant freezing tolerance. They are members of the AP2/ERF multi-gene family, which in Arabidopsis comprises about 145 members. Common to these proteins is the AP2/ERF DNA-binding domain, a 60-amino-acid fold composed of a three-stranded beta-sheet followed by a C-terminal alpha-helix.

Scoring ligand similarity in structure-based virtual screening.

Scoring to identify high-affinity compounds remains a challenge in virtual screening. On one hand, protein-ligand scoring focuses on weighting favorable and unfavorable interactions between the two molecules. Ligand-based scoring, on the other hand, focuses on how well the shape and chemistry of each ligand candidate overlay on a three-dimensional reference ligand.

StoneHinge: hinge prediction by network analysis of individual protein structures.

Hinge motions are important for molecular recognition, and knowledge of their location can guide the sampling of protein conformations for docking. Predicting domains and intervening hinges is also important for identifying structurally self-determinate units and anticipating the influence of mutations on protein flexibility and stability. Here we present StoneHinge, a novel approach for predicting hinges between domains using input from two complementary analyses of noncovalent bond networks: StoneHingeP, which identifies domain-hinge-domain signatures in ProFlex constraint counting results, and StoneHingeD, which does the same for DomDecomp Gaussian network analyses.

The structure of urease activation complexes examined by flexibility analysis, mutagenesis, and small-angle X-ray scattering.

Conformational changes of Klebsiella aerogenes urease apoprotein (UreABC)(3) induced upon binding of the UreD and UreF accessory proteins were examined by a combination of flexibility analysis, mutagenesis, and small-angle X-ray scattering (SAXS). ProFlex analysis of urease provided evidence that the major domain of UreB can move in a hinge-like motion to account for prior chemical cross-linking results. Rigidification of the UreB hinge region, accomplished through a G11P mutation, reduced the extent of urease activation, in part by decreasing the nickel content of the mutant enzyme, and by sequestering a portion of the urease apoprotein in a novel activation complex that includes all of the accessory proteins.

HingeMaster: normal mode hinge prediction approach and integration of complementary predictors.

Protein motion is often the link between structure and function and a substantial fraction of proteins move through a domain hinge bending mechanism. Predicting the location of the hinge from a single structure is thus a logical first step towards predicting motion. Here, we describe ways to predict the hinge location by grouping residues with correlated normal-mode motions.