DeepPeptide predicts cleaved peptides in proteins using conditional random fields.

MOTIVATION: Peptides are ubiquitous throughout life and involved in a wide range of biological processes, ranging from neural signaling in higher organisms to antimicrobial peptides in bacteria. Many peptides are generated post-translationally by cleavage of precursor proteins and can thus not be detected directly from genomics data, as the specificities of the responsible proteases are often not completely understood. RESULTS: We present DeepPeptide, a deep learning model that predicts cleaved peptides directly from the amino acid sequence. DeepPeptide shows both improved precision and recall for peptide detection compared to previous methodology. We show that the model is capable of identifying peptides in underannotated proteomes. AVAILABILITY AND IMPLEMENTATION: DeepPeptide is available online at ku.biolib.com/DeepPeptide.

Deorphanizing Peptides Using Structure Prediction.

Many endogenous peptides rely on signaling pathways to exert their function, but identifying their cognate receptors remains a challenging problem. We investigate the use of AlphaFold-Multimer complex structure prediction together with transmembrane topology prediction for peptide deorphanization. We find that AlphaFold's confidence metrics have strong performance for prioritizing true peptide-receptor interactions. In a library of 1112 human receptors, the method ranks true receptors in the top percentile on average for 11 benchmark peptide-receptor pairs.

Semantic text mining in early drug discovery for type 2 diabetes.

BACKGROUND: Surveying the scientific literature is an important part of early drug discovery; and with the ever-increasing amount of biomedical publications it is imperative to focus on the most interesting articles. Here we present a project that highlights new understanding (e.g. recently discovered modes of action) and identifies potential drug targets, via a novel, data-driven text mining approach to score type 2 diabetes (T2D) relevance. We focused on monitoring trends and jumps in T2D relevance to help us be timely informed of important breakthroughs. METHODS: We extracted over 7 million n-grams from PubMed abstracts and then clustered around 240,000 linked to T2D into almost 50,000 T2D relevant 'semantic concepts'. To score papers, we weighted the concepts based on co-mentioning with core T2D proteins. A protein's T2D relevance was determined by combining the scores of the papers mentioning it in the five preceding years. Each week all proteins were ranked according to their T2D relevance. Furthermore, the historical distribution of changes in rank from one week to the next was used to calculate the significance of a change in rank by T2D relevance for each protein. RESULTS: We show that T2D relevant papers, even those not mentioning T2D explicitly, were prioritised by relevant semantic concepts. Well known T2D proteins were therefore enriched among the top scoring proteins. Our 'high jumpers' identified important past developments in the apprehension of how certain key proteins relate to T2D, indicating that our method will make us aware of future breakthroughs. In summary, this project facilitated keeping up with current T2D research by repeatedly providing short lists of potential novel targets into our early drug discovery pipeline.

The EndoC-ßH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.

OBJECTIVE: To characterize the EndoC-ßH1 cell line as a model for human beta cells and evaluate its beta cell functionality, focusing on insulin secretion, proliferation, apoptosis and ER stress, with the objective to assess its potential as a screening platform for identification of novel anti-diabetic drug candidates. METHODS: EndoC-ßH1 was transplanted into mice for validation of in vivo functionality. Insulin secretion was evaluated in cells cultured as monolayer and as pseudoislets, as well as in diabetic mice. Cytokine induced apoptosis, glucolipotoxicity, and ER stress responses were assessed. Beta cell relevant mRNA and protein expression were investigated by qPCR and antibody staining. Hundreds of proteins or peptides were tested for their effect on insulin secretion and proliferation. RESULTS: Transplantation of EndoC-ßH1 cells restored normoglycemia in streptozotocin induced diabetic mice. Both in vitro and in vivo, we observed a clear insulin response to glucose, and, in vitro, we found a significant increase in insulin secretion from EndoC-ßH1 pseudoislets compared to monolayer cultures for both glucose and incretins. Apoptosis and ER stress were inducible in the cells and caspase 3/7 activity was elevated in response to cytokines, but not affected by the saturated fatty acid palmitate. By screening of various proteins and peptides, we found Bombesin (BB) receptor agonists and Pituitary Adenylate Cyclase-Activating Polypeptides (PACAP) to significantly induce insulin secretion and the proteins SerpinA6, STC1, and APOH to significantly stimulate proliferation. ER stress was readily induced by Tunicamycin and resulted in a reduction of insulin mRNA. Somatostatin (SST) was found to be expressed by 1% of the cells and manipulation of the SST receptors was found to significantly affect insulin secretion. CONCLUSIONS: Overall, the EndoC-ßH1 cells strongly resemble human islet beta cells in terms of glucose and incretin stimulated insulin secretion capabilities. The cell line has an active cytokine induced caspase 3/7 apoptotic pathway and is responsive to ER stress initiation factors. The cells' ability to proliferate can be further increased by already known compounds as well as by novel peptides and proteins. Based on its robust performance during the functionality assessment assays, the EndoC-ßH1 cell line was successfully used as a screening platform for identification of novel anti-diabetic drug candidates.

Construction of a versatile expression library for all human single-pass transmembrane proteins for receptor pairings by high throughput screening.

Interactions between protein ligands and receptors play crucial roles in cell-cell signalling. Most of the human cell surface receptors have been identified in the post-Human Genome Project era but many of their corresponding ligands remain unknown. To facilitate the pairing of orphan receptors, 2762 sequences encoding all human single-pass transmembrane proteins were selected for inclusion into a mammalian-cell expression library. This expression library, consisting of all the individual extracellular domains (ECDs), was constructed as a Fab fusion for each protein. In this format, individual ECD can be produced as a soluble protein or displayed on cell surface, depending on the applied heavy-chain Fab configuration. The unique design of the Fab fusion concept used in the library led to not only superior success rate of protein production, but also versatile applications in various high-throughput screening paradigms including protein-protein binding assays as well as cell binding assays, which were not possible for any other existing expression libraries. The protein library was screened against human coagulation factor VIIa (FVIIa), an approved therapeutic for the treatment of hemophilia, for binding partners by AlphaScreen and ForteBio assays. Two previously known physiological ligands of FVIIa, tissue factor (TF) and endothelial protein C receptor (EPCR) were identified by both assays. The cell surface displayed library was screened against V-domain Ig suppressor of T-cell activation (VISTA), an important immune-checkpoint regulator. Immunoglobulin superfamily member 11 (IgSF11), a potential target for cancer immunotherapy, was identified as a new and previously undescribed binding partner for VISTA. The specificity of the binding was confirmed and validated by both fluorescence-activated cell sorting (FACS) and surface plasmon resonance (SPR) assays in different experimental setups.

GFRAL is the receptor for GDF15 and is required for the anti-obesity effects of the ligand.

Growth differentiation factor 15 (GDF15; also known as MIC-1) is a divergent member of the TGF-ß superfamily and is associated with body-weight regulation in humans and rodents. However, the cognate receptor of GDF15 is unknown. Here we show that GDF15 binds specifically to GDNF family receptor α-like (GFRAL) with high affinity, and that GFRAL requires association with the coreceptor RET to elicit intracellular signaling in response to GDF15 stimulation. We also found that GDF15-mediated reductions in food intake and body weight of mice with obesity were abolished in GFRAL-knockout mice. We further found that GFRAL expression was limited to hindbrain neurons and not present in peripheral tissues, which suggests that GDF15-GFRAL-mediated regulation of food intake is by a central mechanism. Lastly, given that GDF15 did not increase energy expenditure in treated mice with obesity, the anti-obesity actions of the cytokine are likely driven primarily by a reduction in food intake.

The existence of multiple conformers of interleukin-21 directs engineering of a superpotent analogue.

The high resolution three-dimensional structure of human interleukin (hIL)-21 has been resolved by heteronuclear NMR spectroscopy. Overall, the hIL-21 structure is dominated by a well defined central four-helical bundle, arranged in an up-up-down-down topology, as observed for other cytokines. A segment of the hIL-21 molecule that includes the third helical segment, helix C, is observed to exist in two distinct and interchangeable states. In one conformer, the helix C segment is presented in a regular, alpha-helical conformation, whereas in the other conformer, this segment is largely disordered. A structure-based sequence alignment of hIL-21 with receptor complexes of the related cytokines, interleukin-2 and -4, implied that this particular segment is involved in receptor binding. An hIL-21 analog was designed to stabilize the region around helix C through the introduction of a segment grafted from hIL-4. This novel hIL-21 analog was demonstrated to exhibit a 10-fold increase in potency in a cellular assay.

The effect of neurogenin3 deficiency on pancreatic gene expression in embryonic mice.

To understand the molecular mechanisms regulating pancreatic endocrine development and function, pancreatic gene expression was compared between Ngn3-deficient mice and littermate controls on embryonic days 13 and 15. Microarray analysis identified 504 genes with significant differences in expression. Fifty-two of these showed at least twofold reduction in Ngn3 knockouts compared to controls. Many of them were previously described to be involved in endocrine development and function. Among the genes not previously characterized were Rhomboid veinlet-like 4, genes involved in tetrahydrobiopterin biosynthesis and the Iroquois-type homeobox gene Irx1, the latter was selected for further investigation. In situ hybridisation demonstrated that two Iroquois genes, Irx1 and Irx2, were expressed in pancreatic endoderm of wild-type, but not Ngn3 mutant embryos. Furthermore, ectopic Ngn3 induced prominent Irx2 expression in chicken endoderm. Co-labelling established that Irx1 and Irx2 mRNA is located to glucagon-, but not insulin- or somatostatin-producing cells in mice and chicken. These data suggest that Irx1 and Irx2 serve an evolutionary conserved role in the regulation of alpha-cell-specific gene expression.

Using evolutionary information and ancestral sequences to understand the sequence-function relationship in GLP-1 agonists.

Glucagon-like peptide-1 (GLP-1) is an incretin hormone with therapeutic potential for type 2 diabetes. A variety of GLP-1 sequences are known from amphibian species, and some of these have been tested here and found to be able to bind and activate the human GLP-1 receptor. While little difference was observed for the in vitro potency for the human GLP-1 receptor, larger differences were found in the enzymatic stability of these peptides. Two peptides showed increased enzymatic stability, and they group together phylogenetically, though they originate from Amphibia and Reptilia. We have used ancestral sequence reconstruction to analyze the evolution of these GLP-1 molecules, including the synthesis of new peptides. We find that the increased stability could not be observed in the resurrected peptides from the common ancestor of frogs, even though they maintain the ability to activate the human GLP-1 receptor. Another method, using residue mapping on evolutionary branches yielded peptides that had maintained potency towards the receptor and also showed increased stability. This represents a new approach using evolutionary data in protein engineering.

New crystal structures of human glutathione transferase A1-1 shed light on glutathione binding and the conformation of the C-terminal helix.

Human glutathione transferase A1-1 is a well studied enzyme, but despite a wealth of structural and biochemical data a number of aspects of its catalytic function are still poorly understood. Here, five new crystal structures of this enzyme are described that provide several insights. Firstly, the structure of a complex of the wild-type human enzyme with glutathione was determined for the first time at 2.0 angstroms resolution. This reveals that glutathione binds in the G site in a very similar fashion as the glutathione portion of substrate analogues in other structures and also that glutathione binding alone is sufficient to stabilize the C-terminal helix of the protein. Secondly, we have studied the complex with a decarboxylated glutathione conjugate that is known to dramatically decrease the activity of the enzyme. The T68E mutant of human glutathione transferase A1-1 recovers some of the activity that is lost with the decarboxylated glutathione, but our structures of this mutant show that none of the earlier explanations of this phenomenon are likely to be correct. Thirdly, and serendipitously, the apo structures also reveal the conformation of the crucial C-terminal region that is disordered in all previous apo structures. The C-terminal region can adopt an ordered helix-like structure even in the apo state, but shows a strong tendency to unwind. Different conformations of the C-terminal regions were observed in the apo states of the two monomers, which suggests that cooperativity could play a role in the activity of the enzyme.

The charge density of urea from synchrotron diffraction data.

The charge density of urea is studied using very high precision single-crystal synchrotron-radiation diffraction data collected at the Swiss-Norwegian Beam Lines at ESRF. An unprecedented resolution of 1.44 A(-1) in sin theta;/lambda is obtained at 123 K. The optimization of the experiment for charge-density studies is discussed. The high precision of the data allowed the refinement of a multipole model extending to hexadecapoles and quadrupoles on the heavy and H atoms, respectively, as well as a liberal treatment of radial functions. The topological properties of the resulting electron density are analysed and compared with earlier experimental results as well as with periodic Hartree-Fock calculations. The properties of the strongly polarized C-O bond agree with trends derived from previous experimental results while the ab initio calculations differ significantly. The results indicate that the description of the C-O bond requires more flexible basis sets in the theoretical calculations. The calculated integrated atomic charges are much larger than the observed ones. It is suggested that the present experimental results provide new target values for validation of future ab initio calculations. The molecular dipole moment derived from the integrated atomic properties is the same as the one obtained from the multipole model even though the individual atomic contributions differ. Comparison with literature data for urea in solution and the gas phase yields a dipole enhancement in the solid of about 1.5 D. The thermal expansion of urea is determined using synchrotron powder diffraction data. With decreasing temperature, an increasing anisotropic strain is observed.

Structures of three salts of phthalic acid; variation in crystal packing and geometry of the hydrogen phthalate ion.

The hydrogen-bond patterns in hydrogen phthalates (HP) have been elucidated by the determination of the crystal structures of hydrogen phthalate salts of [Mg(H(2)O)(4)-(CH(3)OH)(2)](2+), 2-(2'-pyridylamine)pyridinium and diethylammonium. The stoichiometry of the latter salt corresponds to a hydrogen phthalate salt; however, it contains the phthalic acid in its three possible acid-base forms. The hydrogen phthalate ions in the three salts display the two common hydrogen-bonding motifs. One has a very short intramolecular O-H-O hydrogen bond (Intra-H), as seen in the magnesium salt [O...O 2.397 (2) A]. In the other two salts the hydrogen phthalate ions are connected by hydrogen bonds into infinite chains (Chain-H) with O...O distances of 2.460 (1) A in the diethylamine salt and 2.610 (1) A in the 2,2'-dipyridylamine salt. This difference in hydrogen-bond pattern was further elucidated by a search in the Cambridge Structural Database for other hydrogen phthalate salts. Intra-H possesses the shortest O...O distances and an almost planar hydrogen phthalate ion, whereas the hydrogen phthalate ions show significantly more variation in the Chain-H salts. The two hydrogen-bonding motifs are reflected in the molecular geometry of the hydrogen phthalate ion.

Evaluation of protein fold comparison servers.

When a new protein structure has been determined, comparison with the database of known structures enables classification of its fold as new or belonging to a known class of proteins. This in turn may provide clues about the function of the protein. A large number of fold comparison programs have been developed, but they have never been subjected to a comprehensive and critical comparative analysis. Here we describe an evaluation of 11 publicly available, Web-based servers for automatic fold comparison. Both their functionality (e.g., user interface, presentation, and annotation of results) and their performance (i.e., how well established structural similarities are recognized) were assessed. The servers were subjected to a battery of performance tests covering a broad spectrum of folds as well as special cases, such as multidomain proteins, Calpha-only models, new folds, and NMR-based models. The CATH structural classification system was used as a reference. These tests revealed the strong and weak sides of each server. On the whole, CE, DALI, MATRAS, and VAST showed the best performance, but none of the servers achieved a 100% success rate. Where no structurally similar proteins are found by any individual server, it is recommended to try one or two other servers before any conclusions concerning the novelty of a fold are put on paper.